1 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3 * config/i386/i386.md (isa): Remove fma_avx512f.
4 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5 <avx512>_fmadd_<mode>_mask3<round_name>,
6 <avx512>_fmsub_<mode>_mask<round_name>,
7 <avx512>_fmsub_<mode>_mask3<round_name>,
8 <avx512>_fnmadd_<mode>_mask<round_name>,
9 <avx512>_fnmadd_<mode>_mask3<round_name>,
10 <avx512>_fnmsub_<mode>_mask<round_name>,
11 <avx512>_fnmsub_<mode>_mask3<round_name>,
12 <avx512>_fmaddsub_<mode>_mask<round_name>,
13 <avx512>_fmaddsub_<mode>_mask3<round_name>,
14 <avx512>_fmsubadd_<mode>_mask<round_name>,
15 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
16 (*vec_widen_umult_even_v16si<mask_name>,
17 *vec_widen_smult_even_v16si<mask_name>): Likewise.
18 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
20 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22 * extend.texi: Add 'nocf_check' documentation.
23 * gimple.texi: Add second parameter to
24 gimple_build_call_from_tree.
25 * invoke.texi: Add -fcf-protection documentation.
26 * rtl.texi: Add REG_CALL_NOTRACK documenation.
28 2017-10-20 Richard Biener <rguenther@suse.de>
30 PR tree-optimization/82473
31 * tree-vect-loop.c (vectorizable_reduction): Properly get at
32 the largest input type.
34 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
36 * c-attribs.c (handle_nocf_check_attribute): New function.
37 (c_common_attribute_table): Add 'nocf_check' handling.
38 * gimple-parser.c: Add second argument NULL to
39 gimple_build_call_from_tree.
40 * attrib.c (comp_type_attributes): Check nocf_check attribute.
41 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
43 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
44 * common.opt: Add fcf-protection flag.
45 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
46 * flag-types.h: Add enum cf_protection_level.
47 * gimple.c (gimple_build_call_from_tree): Add second parameter.
48 Add 'nocf_check' attribute propagation to gimple call.
49 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
50 (gimple_build_call_from_tree): Update prototype.
51 (gimple_call_nocf_check_p): New function.
52 (gimple_call_set_nocf_check): Likewise.
53 * gimplify.c: Add second argument to gimple_build_call_from_tree.
54 * ipa-icf.c: Add nocf_check attribute in statement hash.
55 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
56 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
57 * toplev.c (process_options): Add flag_cf_protection handling.
59 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
61 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
63 2017-10-20 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/82603
66 * tree-if-conv.c (predicate_mem_writes): Make sure to only
67 remove false predicated stores.
69 2017-10-20 Richard Biener <rguenther@suse.de>
71 * graphite-isl-ast-to-gimple.c
72 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
73 Remove return value and simplify, dump copied stmt after lhs
75 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
76 Reduce dump verbosity.
77 (gsi_insert_earliest): Likewise.
78 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
79 * graphite.c (print_global_statistics): Adjust dumping.
80 (print_graphite_scop_statistics): Likewise.
81 (print_graphite_statistics): Do not dump loops here.
82 (graphite_transform_loops): But here.
84 2017-10-20 Nicolas Roche <roche@adacore.com>
86 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to the include dir.
87 * configure: Regenerate.
89 2017-10-20 Jakub Jelinek <jakub@redhat.com>
92 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
93 functions when optimizing replace GIMPLE_RETURN stmts with
94 calls to __builtin_unreachable ().
97 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
98 for -fsanitize=thread link of executables.
99 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
103 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
105 (<shift_insn><mode>3<mask_name>): Change the last of the 3
106 define_insns for logical vector shifts to use VI248_AVX512BW
107 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
108 condition, useless isa and prefix attributes. Change the first
109 2 of these define_insns to ...
110 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
111 define_insn for avx512vl.
112 (<shift_insn><mode>3): ... and this, new define_insn without
113 masking for non-avx512vl.
116 * config/i386/sse.md (*andnot<mode>3,
117 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
118 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
119 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
120 not applied use empty suffix even for TARGET_AVX512VL.
121 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
122 is applied, supply evex,evex or evex,evex,evex instead of just
125 2017-10-20 Julia Koval <julia.koval@intel.com>
127 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
128 (OPTION_MASK_ISA_GFNI_UNSET): New.
129 (ix86_handle_option): Handle OPT_mgfni.
130 * config/i386/cpuid.h (bit_GFNI): New.
131 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
132 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
133 * config/i386/i386.c (ix86_target_string): Add -mgfni.
134 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
135 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
136 * config/i386/i386.opt: Add mgfni.
138 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
140 * config/msp430/msp430.c (msp430_option_override): Disable
141 -fdelete-null-pointer-checks.
142 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
144 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
146 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
147 of x87 and SSE instructions.
149 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
151 * asan.c (create_cond_insert_point): Do not update edge count.
152 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
153 (afdo_propagate_circuit): Likewise.
154 (afdo_calculate_branch_prob): Likewise.
155 (afdo_annotate_cfg): Likewise.
156 * basic-block.h (struct edge_def): Remove count.
157 (edge_def::count): New accessor.
158 * bb-reorder.c (rotate_loop): Update.
159 (find_traces_1_round): Update.
160 (connect_traces): Update.
161 (sanitize_hot_paths): Update.
162 * cfg.c (unchecked_make_edge): Update.
163 (make_single_succ_edge): Update.
164 (check_bb_profile): Update.
165 (dump_edge_info): Update.
166 (update_bb_profile_for_threading): Update.
167 (scale_bbs_frequencies_int): Update.
168 (scale_bbs_frequencies_gcov_type): Update.
169 (scale_bbs_frequencies_profile_count): Update.
170 (scale_bbs_frequencies): Update.
171 * cfganal.c (connect_infinite_loops_to_exit): Update.
172 * cfgbuild.c (compute_outgoing_frequencies): Update.
173 (find_many_sub_basic_blocks): Update.
174 * cfgcleanup.c (try_forward_edges): Update.
175 (try_crossjump_to_edge): Update
176 * cfgexpand.c (expand_gimple_cond): Update
177 (expand_gimple_tailcall): Update
178 (construct_exit_block): Update
179 * cfghooks.c (verify_flow_info): Update
180 (redirect_edge_succ_nodup): Update
182 (make_forwarder_block): Update
183 (duplicate_block): Update
184 (account_profile_record): Update
185 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
186 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
187 * cfgloopmanip.c (scale_loop_profile): Update.
189 (lv_adjust_loop_entry_edge): Update.
190 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
191 (force_nonfallthru_and_redirect): Update.
192 (purge_dead_edges): Update.
193 (rtl_flow_call_edges_add): Update.
194 * cgraphunit.c (init_lowered_empty_function): Update.
195 (cgraph_node::expand_thunk): Update.
196 * gimple-pretty-print.c (dump_probability): Update.
197 (dump_edge_probability): Update.
198 * gimple-ssa-isolate-paths.c (isolate_path): Update.
199 * haifa-sched.c (sched_create_recovery_edges): Update.
200 * hsa-gen.c (convert_switch_statements): Update.
201 * ifcvt.c (dead_or_predicable): Update.
202 * ipa-inline-transform.c (inline_transform): Update.
203 * ipa-split.c (split_function): Update.
204 * ipa-utils.c (ipa_merge_profiles): Update.
205 * loop-doloop.c (add_test): Update.
206 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
207 * lto-streamer-in.c (input_cfg): Update.
208 (input_function): Update.
209 * lto-streamer-out.c (output_cfg): Update.
210 * modulo-sched.c (sms_schedule): Update.
211 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
212 * predict.c (maybe_hot_edge_p): Update.
213 (unlikely_executed_edge_p): Update.
214 (probably_never_executed_edge_p): Update.
215 (dump_prediction): Update.
216 (drop_profile): Update.
217 (propagate_unlikely_bbs_forward): Update.
218 (determine_unlikely_bbs): Update.
219 (force_edge_cold): Update.
220 * profile.c (compute_branch_probabilities): Update.
221 * reg-stack.c (better_edge): Update.
222 * shrink-wrap.c (handle_simple_exit): Update.
223 * tracer.c (better_p): Update.
224 * trans-mem.c (expand_transaction): Update.
225 (split_bb_make_tm_edge): Update.
226 * tree-call-cdce.c: Update.
227 * tree-cfg.c (gimple_find_sub_bbs): Update.
228 (gimple_split_edge): Update.
229 (gimple_duplicate_sese_region): Update.
230 (gimple_duplicate_sese_tail): Update.
231 (gimple_flow_call_edges_add): Update.
232 (insert_cond_bb): Update.
233 (execute_fixup_cfg): Update.
234 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
235 * tree-complex.c (expand_complex_div_wide): Update.
236 * tree-eh.c (lower_resx): Update.
237 (unsplit_eh): Update.
238 (cleanup_empty_eh_move_lp): Update.
239 * tree-inline.c (copy_edges_for_bb): Update.
240 (freqs_to_counts): Update.
241 (copy_cfg_body): Update.
242 * tree-ssa-dce.c (remove_dead_stmt): Update.
243 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
244 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
245 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
246 (unloop_loops): Update.
247 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
248 * tree-ssa-loop-split.c (connect_loops): Update.
249 (split_loop): Update.
250 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
251 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
252 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
253 * tree-ssa-reassoc.c (branch_fixup): Update.
254 * tree-ssa-tail-merge.c (replace_block_by): Update.
255 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
256 (compute_path_counts): Update.
257 (update_profile): Update.
258 (recompute_probabilities): Update.
259 (update_joiner_offpath_counts): Update.
260 (estimated_freqs_path): Update.
261 (freqs_to_counts_path): Update.
262 (clear_counts_path): Update.
263 (ssa_fix_duplicate_block_edges): Update.
264 (duplicate_thread_path): Update.
265 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
266 (case_bit_test_cmp): Update.
267 (collect_switch_conv_info): Update.
268 (gen_inbound_check): Update.
269 (do_jump_if_equal): Update.
270 (emit_cmp_and_jump_insns): Update.
271 * tree-tailcall.c (decrease_profile): Update.
272 (eliminate_tail_call): Update.
273 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
274 (vect_do_peeling): Update.
275 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
276 * ubsan.c (ubsan_expand_null_ifn): Update.
277 (ubsan_expand_ptr_ifn): Update.
278 * value-prof.c (gimple_divmod_fixed_value): Update.
279 (gimple_mod_pow2): Update.
280 (gimple_mod_subtract): Update.
282 (gimple_stringop_fixed_value): Update.
284 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
287 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
289 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
291 PR rtl-optimization/82395
292 * ira-color.c (allocno_priority_compare_func): Fix comparison step
293 based on non_spilled_static_chain_regno_p.
295 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
297 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
298 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
299 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
301 2017-10-19 Martin Sebor <msebor@redhat.com>
303 PR tree-optimization/82596
304 * tree.c (array_at_struct_end_p): Handle STRING_CST.
306 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
308 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
309 (get_mem_refs_of_builtin_call): Likewise.
310 * builtins.c (expand_builtin_apply): Adjust call to
311 allocate_dynamic_stack_space.
312 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
313 the third argument to allocate_dynamic_stack_space, otherwise -1.
314 (expand_builtin): Deal with all alloca variants.
315 (is_inexpensive_builtin): Likewise.
316 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
317 * calls.c (special_function_p): Deal with all alloca variants.
318 (initialize_argument_information): Adjust call to
319 allocate_dynamic_stack_space.
320 (expand_call): Likewise.
321 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
322 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
323 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
324 use it for the stack usage computation.
325 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
326 * function.c (gimplify_parameters): Call build_alloca_call_expr.
327 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
328 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
329 (in_loop_p): Remove first argument and useless check.
330 (pass_walloca::execute): Remove useless test and adjust call to above.
331 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
332 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
333 (gimplify_call_expr): Deal with all alloca variants.
334 * hsa-gen.c (gen_hsa_alloca): Likewise.
335 (gen_hsa_insns_for_call): Likewise.
336 * ipa-pure-const.c (special_builtin_state): Likewise.
337 * tree-chkp.c (chkp_build_returned_bound): Likewise.
338 * tree-object-size.c (alloc_object_size): Likewise.
339 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
340 (call_may_clobber_ref_p_1): Likewise.
341 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
342 (ccp_fold_stmt): Likewise.
343 (optimize_stack_restore): Likewise.
344 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
345 (mark_all_reaching_defs_necessary_1): Likewise.
346 (propagate_necessity): Likewise.
347 (eliminate_unnecessary_stmts): Likewise.
348 * tree.c (build_common_builtin_nodes): Build
349 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
350 (build_alloca_call_expr): New function.
351 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
352 (CASE_BUILT_IN_ALLOCA): Likewise.
353 (build_alloca_call_expr): Declare.
354 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
356 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
359 * dwarf2out.c (new_die_raw): New static inline function.
360 (new_die): Use it to create the DIE.
361 (add_AT_external_die_ref): Likewise.
362 (clone_die): Likewise.
363 (clone_as_declaration): Likewise.
364 (dwarf2out_vms_debug_main_pointer): Likewise.
365 (base_type_die): Likewise. Remove early return for corner cases.
366 Do not call add_pubtype on the DIE here.
367 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
368 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
369 typedefs for base types with DW_AT_endianity. Make sure a DIE with
370 native order exists for base types, attach the DIE manually and call
371 add_pubtype on it. Do not equate a reverse order DIE to the type.
373 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
375 * config/arm/arm.c (align_ok_ldrd_strd): New function.
376 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
378 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
380 2017-10-19 Jakub Jelinek <jakub@redhat.com>
382 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
383 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
384 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
385 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
386 * opts.c (sanitizer_opts): Add builtin.
387 * ubsan.c (instrument_builtin): New function.
388 (pass_ubsan::execute): Call it.
389 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
390 * doc/invoke.texi: Document -fsanitize=builtin.
392 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
393 builtins, store max (log2 (align), 0) into uchar field instead of
394 align into uptr field.
395 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
396 store uchar 0 field instead of uptr 0 field.
397 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
398 instead of passing one address of struct with 2 locations pass
399 two addresses of structs with 1 location each.
400 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
401 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
402 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
403 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
404 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
405 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
406 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
407 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
409 2017-10-19 Martin Liska <mliska@suse.cz>
412 * file-find.c (remove_prefix): Remove.
413 * file-find.h (remove_prefix): Likewise.
414 * gcc-ar.c: Remove smartness of lookup.
416 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
418 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
419 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
420 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
422 2017-10-19 Jakub Jelinek <jakub@redhat.com>
425 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
426 (setcc + and to xor + setcc): New peephole2.
428 2017-10-19 Tom de Vries <tom@codesourcery.com>
430 * doc/sourcebuild.texi (Test Directives, Variants of
431 dg-require-support): Add dg-require-stack-size.
433 2017-10-19 Martin Liska <mliska@suse.cz>
436 * gimplify.c (gimplify_decl_expr): Do not instrument variables
437 that have a large alignment.
438 (gimplify_target_expr): Likewise.
440 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
442 PR rtl-optimization/82602
443 * ira.c (rtx_moveable_p): Return false for volatile asm.
445 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
448 * config/i386/i386-modes.def (CCGZ): New CC mode.
449 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
450 * config/i386/predicates.md (ix86_comparison_operator):
452 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
453 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
454 with double-word subtraction.
455 (put_condition_code): Handle CCGZmode.
457 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
459 * wide-int.cc (debug (const wide_int &)): New.
460 (debug (const wide_int *)): New.
461 (debug (const widest_int &)): New.
462 (debug (const widest_int *)): New.
464 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
467 * lra-constraints.c (curr_insn_transform): Use non-input operand
468 instead of output one for matched reload.
470 2017-10-18 Bin Cheng <bin.cheng@arm.com>
472 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
473 (tree-ssa-loop-ivopts.h): New header file.
474 (struct builtin_info): New fields.
475 (classify_builtin_1): Compute and record base and offset parts for
476 memset builtin partition by calling strip_offset.
477 (offset_cmp, fuse_memset_builtins): New functions.
478 (finalize_partitions): Fuse adjacent memset partitions by calling
480 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
481 Expose the interface.
482 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
484 2017-10-18 Bin Cheng <bin.cheng@arm.com>
486 PR tree-optimization/82574
487 * tree-loop-distribution.c (find_single_drs): New parameter. Check
488 that data reference must be executed exactly once per iteration
489 against the outermost loop in nest.
490 (classify_partition): Update call to above function.
492 2017-10-18 Richard Biener <rguenther@suse.de>
494 PR tree-optimization/82591
495 * graphite.c (graphite_transform_loops): Move code gen message
497 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
498 Here. Handle scop_to_isl_ast failing.
499 (scop_to_isl_ast): Limit the number of ISL operations.
501 2017-10-18 Richard Biener <rguenther@suse.de>
503 * graphite-isl-ast-to-gimple.c
504 (translate_isl_ast_to_gimple::set_rename): Simplify.
505 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
506 (graphite_copy_stmts_from_block): ... here.
507 (copy_bb_and_scalar_dependences): Simplify.
508 (add_parameters_to_ivs_params): Canonicalize.
509 (generate_entry_out_of_ssa_copies): Simplify.
510 * graphite-sese-to-poly.c (extract_affine_name): Simplify
511 by passing in ISL dimension.
512 (parameter_index_in_region_1): Rename to ...
513 (parameter_index_in_region): ... this.
514 (extract_affine): Adjust assert, pass down parameter index.
515 (add_param_constraints): Use range-info when available.
516 (build_scop_context): Adjust.
517 * sese.c (new_sese_info): Adjust.
518 (free_sese_info): Likewise.
519 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
520 Remove unused typedefs.
521 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
523 2017-10-18 Martin Liska <mliska@suse.cz>
525 * combine.c (simplify_compare_const): Add gcc_fallthrough.
527 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
529 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
530 (s390_sched_init): Do not reset s390_sched_state if we entered the
531 current basic block via a fallthru edge and all others are unlikely.
533 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
535 * config/s390/s390.c (NUM_SIDES): New variable.
536 (LONGRUNNING_THRESHOLD): New variable.
537 (LATENCY_FACTOR): New variable.
538 (s390_sched_score): Decrease score for long-running instructions on
540 (s390_sched_variable_issue): Perform bookkeeping for long-running
543 2017-10-18 Richard Biener <rguenther@suse.de>
545 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
546 Simplify with removal of the parameter rename map.
547 (set_rename): Likewise.
548 (should_copy_to_new_region): Likewise.
549 (graphite_copy_stmts_from_block): Likewise.
550 (copy_bb_and_scalar_dependences): Remove initialization of
551 unused copied_bb_map.
553 (copy_internal_parameters): Likewise.
554 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
555 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
557 (parameter_index_in_region_1): Rename to ...
558 (assign_parameter_index_in_region): ... this. Assert we have
559 a parameter we handle.
560 (scan_tree_for_params): Adjust.
561 * sese.h (parameter_rename_map_t): Remove.
562 (struct sese_info_t): Remove unused parameter_rename_map and
563 copied_bb_map members.
564 * sese.c (new_sese_info): Adjust.
565 (free_sese_info): Likewise.
567 2017-10-18 Martin Liska <mliska@suse.cz>
570 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
573 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
575 * doc/invoke.texi (ffunction-sections and fdata-sections):
578 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
580 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
581 the use statement can throw internally.
583 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
585 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
586 any RTX present on the RHS of a SET.
587 * compare-elim.c (try_eliminate_compare): Restore comment.
589 2017-10-17 Jakub Jelinek <jakub@redhat.com>
591 * langhooks.h (struct lang_hooks): Document that tree_size langhook
592 may be also called on tcc_type nodes.
593 * langhooks.c (lhd_tree_size): Likewise.
595 2017-10-17 David Malcolm <dmalcolm@redhat.com>
597 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
598 format_warning_at_substring.
599 (maybe_warn): Convert source_range * param to a location_t. Pass
600 UNKNOWN_LOCATION rather than NULL to fmtwarn.
601 (format_directive): Remove code to extract source_ranges and
602 source_range * in favor of just a location_t.
603 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
605 * substring-locations.c (format_warning_va): Convert
606 source_range * param to a location_t.
607 (format_warning_at_substring): Likewise.
608 * substring-locations.h (format_warning_va): Likewise.
609 (format_warning_at_substring): Likewise.
611 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
613 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
615 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
616 and scatter/gather ops.
618 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
619 vec_gather_load and vec_scatter_store.
620 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
621 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
622 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
623 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
624 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
625 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
627 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
629 * reg-stack.c (compare_for_stack_reg): Add bool argument.
630 Detect FTST instruction and handle its register pops. Only pop
631 second operand if can_pop_second_op is true.
632 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
633 set can_pop_second_op to false in the compare_for_stack_reg call.
635 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
636 output_fp_compare for stack register operands.
637 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
638 instructions here. Do not emit stack register pops here. Assert
639 that FCOMPP pops next to top stack register. Rewrite function.
641 2017-10-17 Nathan Sidwell <nathan@acm.org>
644 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
645 use DECL_ASSEMBLER_NAME_RAW.
648 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
651 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
652 Wilco Dijkstra <wilco.dijkstra@arm.com>
654 * builtins.c (expand_builtin_update_setjmp_buf): Add a
655 converstion to Pmode from the buf_addr.
657 2017-10-17 Richard Biener <rguenther@suse.de>
659 * graphite-dependences.c (scop_get_reads_and_writes): Change
660 output parameters to references.
662 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
664 PR 71026/tree-optimization
665 * fold-const.c (distribute_real_division): Removed.
666 (fold_binary_loc): Remove calls to distribute_real_divison.
668 2017-10-17 Richard Biener <rguenther@suse.de>
670 * graphite-scop-detection.c
671 (scop_detection::stmt_has_simple_data_refs_p): Always use
672 the full nest as region.
673 (try_generate_gimple_bb): Likewise.
674 * sese.c (scalar_evolution_in_region): Simplify now that
675 SCEV can handle instantiation in regions.
676 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
677 in the non-loop part of a function if requested.
679 2017-10-17 Richard Biener <rguenther@suse.de>
681 PR tree-optimization/82563
682 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
684 (graphite_regenerate_ast_isl): Call it.
685 * graphite-scop-detection.c (build_scops): Remove entry edge split.
687 2017-10-17 Jakub Jelinek <jakub@redhat.com>
689 PR tree-optimization/82549
690 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
691 Formatting fixes. Instead of calling make_bit_field_ref with negative
694 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
696 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
697 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
698 _mm_maskz_reduce_ss): New.
699 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
700 __builtin_ia32_reducess_mask): Ditto..
701 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
702 * config/i386/sse.md (reduces<mode>): Renamed to ...
703 (reduces<mode><mask_scalar_name>): ... this.
704 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
706 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
707 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
709 2017-10-16 David Malcolm <dmalcolm@redhat.com>
711 * Makefile.in (OBJS): Add unique-ptr-tests.o.
712 * selftest-run-tests.c (selftest::run_tests): Call
713 selftest::unique_ptr_tests_cc_tests.
714 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
715 * unique-ptr-tests.cc: New file.
717 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
720 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
722 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
723 (make_hard_regno_born, make_hard_regno_dead): Update
724 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
726 2017-10-16 Jeff Law <law@redhat.com>
728 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
730 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
732 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
734 2017-10-16 Olivier Hainque <hainque@adacore.com>
736 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
737 with_cpu if we were configured for an e500v2 target cpu name.
739 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
741 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
742 * doc/invoke.texi: Document +nodsp as a valid extension for
745 2017-10-16 Martin Liska <mliska@suse.cz>
747 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
748 (test_set_range): Likewise.
749 (test_range_functions): Rename to ...
750 (test_bit_in_range): ... this.
751 (sbitmap_c_tests): Add new test.
753 2017-10-16 Tamar Christina <tamar.christina@arm.com>
755 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
756 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
757 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
759 2017-10-16 Tamar Christina <tamar.christina@arm.com>
761 * config/aarch64/aarch64-builtins.c
762 (aarch64_types_quadopu_lane_qualifiers): New.
763 (TYPES_QUADOPU_LANE): New.
764 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
765 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
766 (aarch64_<sur>dot_laneq<vsi2qi>): New.
767 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
768 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
769 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
770 (Vdottype, DOTPROD): New.
771 (sur): Add SDOT and UDOT.
773 2017-10-16 Tamar Christina <tamar.christina@arm.com>
775 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
776 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
777 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
778 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
779 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
780 (cortex-a75.cortex-a55): Likewise.
781 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
783 2017-10-16 Tamar Christina <tamar.christina@arm.com>
785 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
786 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
787 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
788 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
789 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
790 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
791 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
792 * config/arm/types.md (neon_dot, neon_dot_q): New.
793 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
795 2017-10-16 Tamar Christina <tamar.christina@arm.com>
797 * config/arm/arm.h (TARGET_DOTPROD): New.
798 * config/arm/arm.c (arm_arch_dotprod): New.
799 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
800 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
801 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
802 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
803 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
804 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
805 * doc/invoke.texi (armv8.2-a): Document dotprod
807 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
809 * i386.c (ix86_vec_cost): New function.
810 (ix86_rtx_costs): Handle vector operations better.
811 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
812 * x86-tune-costs.h: Add new costs to all tables.
814 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
816 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
818 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
819 divsd, sqrtss and sqrtsd
820 * x86-tune-costs.h: Add new entries to all costs.
821 (znver1_cost): Fix to match real instruction latencies.
823 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
824 Michael Collison <michael.collison@arm.com>
826 * compare-elim.c: Include emit-rtl.h.
827 (can_merge_compare_into_arith): New function.
828 (try_validate_parallel): Likewise.
829 (try_merge_compare): Likewise.
830 (try_eliminate_compare): Call the above when no previous clobber
832 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
835 2017-10-14 Jakub Jelinek <jakub@redhat.com>
839 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
840 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
844 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
845 to be any operand_equal_p operands. For & (B - 1) require
846 B to be power of 2. Recognize
847 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
849 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
852 * optabs.c (expand_memory_blockage): Fix call of
853 targetm.have_memory_blockage.
855 2017-10-14 Jakub Jelinek <jakub@redhat.com>
858 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
859 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
860 objects to extra_objs instead of overwriting it.
862 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
864 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
865 Use any_fp_register_operand as operand[3] predicate. Simplify
866 equality test for operands[2] and operands[4] memory location.
867 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
868 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
869 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
870 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
871 any_fp_register_operand as operand[1] predicate. Simplify
872 equality test for operands[0] and operands[3] memory location.
873 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
874 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
875 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
877 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
879 * target-insns.def: Add memory_blockage.
880 * optabs.c (expand_memory_blockage): New function.
881 (expand_asm_memory_barrier): Rename ...
882 (expand_asm_memory_blockage): ... to this.
883 (expand_mem_thread_fence): Call expand_memory_blockage
884 instead of expand_asm_memory_barrier.
885 (expand_mem_singnal_fence): Ditto.
886 (expand_atomic_load): Ditto.
887 (expand_atomic_store): Ditto.
888 * doc/md.texi (Standard Pattern Names For Generation):
889 Document memory_blockage instruction pattern.
891 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
893 * config/rl78/rl78.c (rl78_emit_libcall): New function.
894 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
895 * config/rl78/rl78.md: New define_expand "adddi3".
897 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
899 * cfghooks.c (verify_flow_info): Disable check that all probabilities
902 2017-10-13 Jeff Law <law@redhat.com>
904 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
906 2017-10-13 Jakub Jelinek <jakub@redhat.com>
909 * internal-fn.c (expand_mul_overflow): If both operands have
910 the same highpart of -1 or 0 and the topmost bit of lowpart
911 is different, overflow is if res <= 0 rather than res < 0.
913 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
915 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
916 TARGET_P9_VECTOR code for unaligned_load case.
918 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
920 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
922 2017-10-13 Nathan Sidwell <nathan@acm.org>
924 * tree-core.h (tree_contains_struct): Make bool.
925 * tree.c (tree_contains_struct): Likewise.
926 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
927 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
928 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
929 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
930 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
932 2017-10-13 Richard Biener <rguenther@suse.de>
934 * graphite-isl-ast-to-gimple.c
935 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
936 parameters and dominance check.
937 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
938 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
939 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
940 Do not update SSA form here or do intermediate IL verification.
941 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
942 (graphite_initialize): Remove check on the number of loops in
943 the function and inline into graphite_transform_loops.
944 (graphite_finalize): Inline into graphite_transform_loops.
945 (graphite_transform_loops): Perform SSA update and IL verification
947 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
949 2017-10-13 Richard Biener <rguenther@suse.de>
951 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
952 graphite_expression_type_precision): Avoid global constructor
954 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
955 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
956 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
957 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
958 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
959 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
961 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
964 * config/i386/i386.h (ix86_red_zone_size): New.
965 * config/i386/i386.md (push peephole2s): Replace
966 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
968 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
969 Alan Hayward <alan.hayward@arm.com>
970 David Sherwood <david.sherwood@arm.com>
972 * combine.c (can_change_dest_mode): Reject changes in
973 REGMODE_NATURAL_SIZE.
975 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
976 Alan Hayward <alan.hayward@arm.com>
977 David Sherwood <david.sherwood@arm.com>
979 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
980 (expand_debug_source_expr): Likewise.
981 * combine.c (combine_simplify_rtx): Likewise.
982 * cse.c (fold_rtx): Likewise.
983 * fwprop.c (canonicalize_address): Likewise.
984 * targhooks.c (default_shift_truncation_mask): Likewise.
986 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
987 Alan Hayward <alan.hayward@arm.com>
988 David Sherwood <david.sherwood@arm.com>
990 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
991 (widened_mode): Likewise.
992 (expand_unop): Likewise.
993 * ree.c (transform_ifelse): Likewise.
994 (merge_def_and_ext): Likewise.
995 (combine_reaching_defs): Likewise.
996 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
998 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
999 Alan Hayward <alan.hayward@arm.com>
1000 David Sherwood <david.sherwood@arm.com>
1002 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1003 * combine.c (gen_lowpart_for_combine): Likewise.
1004 * dwarf2out.c (rtl_for_decl_location): Likewise.
1005 * final.c (alter_subreg): Likewise.
1006 * rtlhooks.c (gen_lowpart_general): Likewise.
1007 (gen_lowpart_if_possible): Likewise.
1009 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1010 Alan Hayward <alan.hayward@arm.com>
1011 David Sherwood <david.sherwood@arm.com>
1013 * calls.c (expand_call): Use subreg_lowpart_offset.
1014 * cse.c (cse_insn): Likewise.
1015 * regcprop.c (copy_value): Likewise.
1016 (copyprop_hardreg_forward_1): Likewise.
1018 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1021 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1022 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1023 =Q constraints to +Q and into insn condition add check
1024 that operands[0] and operands[1] are equal.
1025 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1026 =Q constraints to +Q and into insn condition add check
1027 that operands[0] is equal to either operands[1] or operands[2].
1030 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1031 instead of handling MINUS_EXPR twice (once for each argument),
1032 canonicalize operand order and handle just once, use rtype where
1033 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1036 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1037 any values of __C while still being pattern recognizable as a simple
1040 2017-10-13 Richard Biener <rguenther@suse.de>
1042 PR tree-optimization/82451
1044 2017-10-02 Richard Biener <rguenther@suse.de>
1046 PR tree-optimization/82355
1047 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1048 a mapping for the enclosing loop but avoid generating one for
1050 (copy_bb_and_scalar_dependences): Remove premature codegen
1051 error on PHIs in blocks duplicated into multiple places.
1052 * graphite-scop-detection.c
1053 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1054 in the region use it as loop and nest to analyze the DR in.
1055 (try_generate_gimple_bb): Likewise.
1056 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1057 (add_loop_constraints): For blocks in a loop not in the region
1058 create a dimension with a single iteration.
1059 * sese.h (gbb_loop_at_index): Remove assert.
1061 * cfgloop.c (loop_preheader_edge): For the loop tree root
1062 return the single successor of the entry block.
1063 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1064 Reset the SCEV hashtable and niters.
1065 * graphite-scop-detection.c
1066 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1067 assert that we only have POLYNOMIAL_CHREC that vary in loops
1068 contained in the region.
1069 (scop_detection::graphite_can_represent_expr): Adjust.
1070 (scop_detection::stmt_has_simple_data_refs_p): For loops
1071 not in the region set loop to NULL. The nest is now the
1072 entry edge to the region.
1073 (try_generate_gimple_bb): Likewise.
1074 * sese.c (scalar_evolution_in_region): Adjust for
1075 instantiate_scev change.
1076 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1077 Make nest parameter the edge into the region.
1078 (create_data_ref): Likewise.
1079 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1080 entry edge into a region and adjust instantiate_scev calls.
1081 (create_data_ref): Likewise.
1082 (graphite_find_data_references_in_stmt): Likewise.
1083 (find_data_references_in_stmt): Pass the loop preheader edge
1084 from the nest argument.
1085 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1086 parameter the edge into the region.
1087 (instantiate_parameters): Use the loop preheader edge as entry.
1088 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1090 (get_instantiated_value_entry): Make instantiate_below parameter
1091 the edge into the region.
1092 (instantiate_scev_name): Likewise. Adjust dominance checks,
1093 when we cannot use loop-based instantiation instantiate by
1094 walking use-def chains.
1095 (instantiate_scev_poly): Adjust.
1096 (instantiate_scev_binary): Likewise.
1097 (instantiate_scev_convert): Likewise.
1098 (instantiate_scev_not): Likewise.
1099 (instantiate_array_ref): Remove.
1100 (instantiate_scev_3): Likewise.
1101 (instantiate_scev_2): Likewise.
1102 (instantiate_scev_1): Likewise.
1103 (instantiate_scev_r): Do not blindly handle N-operand trees.
1104 Do not instantiate array-refs. Handle all constants and invariants.
1105 (instantiate_scev): Make instantiate_below parameter
1106 the edge into the region.
1107 (resolve_mixers): Use the loop preheader edge for the region
1108 parameter to instantiate_scev_r.
1109 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1111 2017-10-13 Richard Biener <rguenther@suse.de>
1113 PR tree-optimization/82525
1114 * graphite-isl-ast-to-gimple.c
1115 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1117 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1118 Fail code generation when we cannot represent the isl integer.
1119 (binary_op_to_tree): Elide modulo operations that are no-ops
1120 in the type we code generate. Remove now superfluous code
1123 2017-10-13 Richard Biener <rguenther@suse.de>
1125 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1126 (scop_detection::harmful_loop_in_region): Remove premature
1127 IV type restriction.
1128 (scop_detection::graphite_can_represent_scev): We can handle
1129 pointer IVs just fine.
1131 2017-10-13 Alan Modra <amodra@gmail.com>
1133 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1134 "Clobbers and Scratch Registers". Add paragraph on
1135 alternative to clobbers for scratch registers and OpenBLAS
1138 2017-10-13 Alan Modra <amodra@gmail.com>
1140 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1141 example of a memory input for a string of known length. Move
1142 commentary out of table. Add a number of new examples
1143 covering array memory inputs.
1145 2017-10-12 Martin Liska <mliska@suse.cz>
1147 PR tree-optimization/82493
1148 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1149 (test_range_functions): New function.
1150 (sbitmap_c_tests): Likewise.
1151 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1152 * selftest.h (sbitmap_c_tests): New function.
1154 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1156 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1158 * config/rs6000/amo.h: Fix spacing issue.
1160 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1163 * config/i386/i386.md (*ashl<mode>3_mask_1,
1164 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1165 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1168 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1170 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1171 (profile_probability): Set max_probability
1172 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1174 * profile-count.c (profile_probability::differs_from_p): Do not
1175 rely on max_probaiblity == 10000
1177 2017-10-12 Jeff Law <law@redhat.com>
1179 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1182 2017-10-12 Martin Sebor <msebor@redhat.com>
1186 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1187 (handle_alias_pairs): Call it.
1188 * common.opt (-Wattribute-alias): New option.
1189 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1190 * doc/invoke.texi (-Wattribute-alias): Document.
1192 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1195 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1197 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1199 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1200 (make_hard_regno_born, make_hard_regno_dead): Update
1201 bb_killed_pseudos and bb_gen_pseudos.
1203 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1205 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1207 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1209 * config/alpha/alpha.c (alpha_split_conditional_move):
1210 Use std::swap instead of manually swapping.
1211 (alpha_stdarg_optimize_hook): Ditto.
1212 (alpha_canonicalize_comparison): Ditto.
1214 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1216 * tree-loop-distribution.c (struct builtin_info): New struct.
1217 (struct partition): Refactor fields into struct builtin_info.
1218 (partition_free): Free struct builtin_info.
1219 (build_size_arg_loc, build_addr_arg_loc): Delete.
1220 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1221 information from struct builtin_info.
1222 (find_single_drs): New function refactored from classify_partition.
1223 Also moved builtin validity checks to this function.
1224 (compute_access_range, alloc_builtin): New functions.
1225 (classify_builtin_st, classify_builtin_ldst): New functions.
1226 (classify_partition): Refactor code into functions find_single_drs,
1227 classify_builtin_st and classify_builtin_ldst.
1228 (distribute_loop): Don't do runtime alias check when distributing
1230 (find_seed_stmts_for_distribution): New function.
1231 (pass_loop_distribution::execute): Refactor code finding seed
1232 stmts into above function. Support distribution for the innermost
1233 two-level loop nest. Adjust dump information.
1235 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1237 * tree-loop-distribution.c: Adjust the general comment.
1238 (NUM_PARTITION_THRESHOLD): New macro.
1239 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1240 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1241 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1242 in call to build_partition_graph.
1243 (finalize_partitions): New parameter. Make loop distribution more
1244 conservative by fusing more partitions.
1245 (distribute_loop): Don't do runtime alias check in case of loop nest
1247 (find_seed_stmts_for_distribution): New function.
1248 (prepare_perfect_loop_nest): New function.
1249 (pass_loop_distribution::execute): Refactor code finding seed stmts
1250 and loop nest into above functions. Support loop nest distribution.
1251 Adjust dump information accordingly.
1253 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1255 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1256 and set PTYPE_SEQUENTIAL for merged partition.
1258 2017-10-12 Richard Biener <rguenther@suse.de>
1260 PR tree-optimization/69728
1262 2017-09-19 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/69728
1265 * graphite-sese-to-poly.c (schedule_error): New global.
1266 (add_loop_schedule): Handle empty domain by failing the
1268 (build_original_schedule): Handle schedule_error.
1270 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1271 domain by returning an unchanged schedule.
1273 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1275 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1276 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1278 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1280 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1283 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1286 * expr.c (store_field): Don't optimize away bitsize == 0 store
1287 from CALL_EXPR with addressable return type.
1289 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1291 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1292 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1293 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1294 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1295 TARGET_ISEL instead of TARGET_ISEL<sel>.
1297 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1299 * config/rs6000/rs6000.c
1300 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1302 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1304 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1306 (reg_or_cint_operand): Fix comment.
1307 (reg_or_zero_operand): New predicate.
1308 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1309 * config/rs6000/rs6000.c (output_isel): Delete.
1310 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1311 instead of reg_or_cint_operand. Output instruction directly (not via
1313 (isel_unsigned_<mode>): Ditto.
1314 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1315 gpc_reg_operand. Add an instruction alternative for this. Output
1316 instruction directly.
1317 (*isel_reversed_unsigned_<mode>): Ditto.
1319 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1321 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1322 (TARGET_CANONICALIZE_COMPARISON): Define.
1324 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1327 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1328 Check whether the dest is REG before adding REG_EQUIV note.
1330 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1333 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1335 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1336 (make_hard_regno_born, make_hard_regno_dead): Update
1337 bb_killed_pseudos and bb_gen_pseudos.
1339 2017-10-11 Nathan Sidwell <nathan@acm.org>
1341 * incpath.h (enum incpath_kind): Name enum, prefix values.
1342 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1343 * incpath.c (heads, tails): Use INC_MAX.
1344 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1345 (merge_include_chains, split_quote_chain,
1346 register_include_chains): Update incpath_kind names.
1347 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1348 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1350 (add_framework_path, darwin_register_objc_includes): Likewise.
1351 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1353 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1355 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1356 Do not use float_operator operator predicate.
1357 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1358 * config/i386/predicates.md (float_operator): Remove predicate.
1360 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1362 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1363 (*jccxf_i387): Ditto.
1364 (*jcc<mode>_i387): Ditto.
1365 (*jccu<mode>_i387): Ditto.
1366 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1367 (*jcc_*_i387 splitters): Remove.
1368 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1369 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1370 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1373 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1375 * profile-count.h (slow_safe_scale_64bit): New function.
1376 (safe_scale_64bit): New inline.
1377 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1378 * profile-count.c: Include wide-int.h
1379 (slow_safe_scale_64bit): New.
1381 2017-10-11 Nathan Sidwell <nathan@acm.org>
1383 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1384 HAS_DECL_ASSEMBLER_NAME_P.
1385 * gimple-expr.c (gimple_decl_printable_name: Check
1386 HAS_DECL_ASSEMBLER_NAME_P too.
1387 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1388 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1389 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1390 * passes.c (rest_of_decl_compilation): Check
1391 HAS_DECL_ASSEMBLER_NAME_P too.
1392 * recog.c (verify_changes): Likewise.
1393 * tree-pretty-print.c (dump_decl_name): Likewise.
1394 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1396 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1397 (SET_DECL_ASSEMBLER_NAME): Use it.
1398 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1399 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1400 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1402 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1404 * config.gcc (i386, x86_64): Add extra objects.
1405 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1406 (ix86_min_insn_size): Declare.
1407 (ix86_issue_rate): Declare.
1408 (ix86_adjust_cost): Declare.
1409 (ia32_multipass_dfa_lookahead): Declare.
1410 (ix86_macro_fusion_p): Declare.
1411 (ix86_macro_fusion_pair_p): Declare.
1412 (ix86_bd_has_dispatch): Declare.
1413 (ix86_bd_do_dispatch): Declare.
1414 (ix86_core2i7_init_hooks): Declare.
1415 (ix86_atom_sched_reorder): Declare.
1416 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1417 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1418 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1419 (rip_relative_addr_p): Rename to ...
1420 (ix86_rip_relative_addr_p): ... this one; export.
1421 (memory_address_length): Update.
1422 (ix86_issue_rate): Move to x86-tune-sched.c.
1423 (ix86_flags_dependent): Move to x86-tune-sched.c.
1424 (ix86_agi_dependent): Move to x86-tune-sched.c.
1425 (exact_dependency_1): Move to x86-tune-sched.c.
1426 (exact_store_load_dependency): Move to x86-tune-sched.c.
1427 (ix86_adjust_cost): Move to x86-tune-sched.c.
1428 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1429 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1430 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1431 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1432 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1433 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1434 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1435 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1436 (min_insn_size): Rename to ...
1437 (ix86_min_insn_size): ... this one; export.
1438 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1439 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1440 (core2i7_first_cycle_multipass_backtrack): Move to
1441 x86-tune-sched-core.c.
1442 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1443 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1444 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1445 (ix86_avoid_jump_mispredicts): Update.
1446 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1447 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1448 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1449 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1450 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1451 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1452 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1453 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1454 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1455 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1456 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1457 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1458 (BIG): Move to ix86-tune-sched-bd.c.
1459 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1460 (enum insn_path): Move to ix86-tune-sched-bd.c.
1461 (get_mem_group): Move to ix86-tune-sched-bd.c.
1462 (is_cmp): Move to ix86-tune-sched-bd.c.
1463 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1464 (is_branch): Move to ix86-tune-sched-bd.c.
1465 (is_prefetch): Move to ix86-tune-sched-bd.c.
1466 (init_window): Move to ix86-tune-sched-bd.c.
1467 (allocate_window): Move to ix86-tune-sched-bd.c.
1468 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1469 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1470 (process_end_window): Move to ix86-tune-sched-bd.c.
1471 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1472 (find_constant): Move to ix86-tune-sched-bd.c.
1473 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1474 (has_immediate): Move to ix86-tune-sched-bd.c.
1475 (get_insn_path): Move to ix86-tune-sched-bd.c.
1476 (get_insn_group): Move to ix86-tune-sched-bd.c.
1477 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1478 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1479 (add_insn_window): Move to ix86-tune-sched-bd.c.
1480 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1481 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1482 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1483 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1484 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1485 (do_dispatch): Move to ix86-tune-sched-bd.c.
1486 (has_dispatch): Move to ix86-tune-sched-bd.c.
1487 * config/i386/t-i386: Add new object files.
1488 * config/i386/x86-tune-costs.h: New file.
1489 * config/i386/x86-tune-sched-atom.c: New file.
1490 * config/i386/x86-tune-sched-bd.c: New file.
1491 * config/i386/x86-tune-sched-core.c: New file.
1492 * config/i386/x86-tune-sched.c: New file.
1494 2017-10-11 Liu Hao <lh_mouse@126.com>
1496 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1497 the generic version below instead.
1498 (should_colorize): Recognize Windows consoles as terminals
1500 * pretty-print.c [__MINGW32__] (write_all): New function.
1501 [__MINGW32__] (find_esc_head): Likewise.
1502 [__MINGW32__] (find_esc_terminator): Likewise.
1503 [__MINGW32__] (eat_esc_sequence): Likewise.
1504 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1506 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1509 2017-10-11 Richard Biener <rguenther@suse.de>
1511 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1512 Properly call analyze_scalar_evolution with the loop of the stmt.
1514 2017-10-11 Richard Biener <rguenther@suse.de>
1516 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1517 * tree-core.h (tree_base): Add chrec_var union member.
1518 * tree.h (CHREC_VAR): Remove.
1519 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1520 * tree-chrec.h (build_polynomial_chrec): Adjust.
1521 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1522 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1524 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1526 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1527 * match.pd: ... here.
1528 ((T) X == (T) Y): Relax condition.
1530 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1532 PR tree-optimization/82472
1533 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1535 (break_alias_scc_partitions): Update postorder number.
1537 2017-10-11 Martin Liska <mliska@suse.cz>
1540 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1542 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1544 2017-10-11 Martin Liska <mliska@suse.cz>
1546 * print-rtl.c (print_insn): Move declaration of idbuf
1547 to same scope as name.
1549 2017-10-11 Martin Liska <mliska@suse.cz>
1554 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1557 2017-10-11 Martin Liska <mliska@suse.cz>
1560 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1563 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1565 * config.gcc (default_gnu_indirect_function): Default to yes for
1566 arm*-*-linux* with glibc.
1568 2017-10-11 Richard Biener <rguenther@suse.de>
1570 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1571 default-defs and types we do not want to analyze.
1572 (interpret_loop_phi): Replace unreachable code with an assert.
1573 (compute_scalar_evolution_in_loop): Remove and inline ...
1574 (analyze_scalar_evolution_1): ... here, replacing condition with
1575 what makes the intent clearer. Remove handling of cases
1576 get_scalar_evolution now handles.
1578 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1580 PR rtl-optimization/81434
1581 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1582 comment for main loop. In sched_group_found if, also add checks for
1583 pass and min_cost_group.
1585 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1587 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1588 (rs6000_insn_cost): New function.
1589 * config/rs6000/rs6000.md (cost): New attribute.
1591 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1592 H.J. Lu <hongjiu.lu@intel.com>
1596 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1597 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1598 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1599 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1600 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1601 OPTION_MASK_ISA_MMX and some other bit set require both
1602 mmx and the ISAs without the mmx bit.
1603 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1604 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1605 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1606 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1607 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1608 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1609 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1610 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1611 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1612 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1613 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1614 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1615 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1616 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1617 Add OPTION_MASK_ISA_MMX.
1619 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1621 * config.gcc (armv7*-*-freebsd*): New target.
1622 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1624 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1626 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1627 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1630 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1632 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1633 a template parameter.
1634 (WIDE_INT_REF_FOR): Update accordingly.
1635 * tree.h (wi::int_traits <const_tree>): Delete.
1636 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1637 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1638 (wi::tree_to_wide_ref): New typedef.
1639 (wi::to_wide): New function.
1640 * calls.c (get_size_range): Use wi::to_wide when operating on
1642 * cgraph.c (cgraph_node::create_thunk): Likewise.
1643 * config/i386/i386.c (ix86_data_alignment): Likewise.
1644 (ix86_local_alignment): Likewise.
1645 * dbxout.c (stabstr_O): Likewise.
1646 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1647 * expr.c (const_vector_from_tree): Likewise.
1648 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1649 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1650 (fold_negate_expr_1, int_const_binop_1, const_binop)
1651 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1652 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1653 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1654 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1655 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1656 (fold_not_const, round_up_loc): Likewise.
1657 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1658 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1659 (alloca_call_type): Likewise.
1660 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1661 * godump.c (go_output_typedef): Likewise.
1662 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1663 * internal-fn.c (get_min_precision): Likewise.
1664 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1665 * ipa-polymorphic-call.c
1666 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1667 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1668 (ipa_modify_call_arguments): Likewise.
1669 * match.pd: Likewise.
1670 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1671 * print-tree.c (print_node_brief, print_node): Likewise.
1672 * stmt.c (expand_case): Likewise.
1673 * stor-layout.c (layout_type): Likewise.
1674 * tree-affine.c (tree_to_aff_combination): Likewise.
1675 * tree-cfg.c (group_case_labels_stmt): Likewise.
1676 * tree-data-ref.c (dr_analyze_indices): Likewise.
1677 (prune_runtime_alias_test_list): Likewise.
1678 * tree-dump.c (dequeue_and_dump): Likewise.
1679 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1680 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1681 * tree-pretty-print.c (dump_generic_node): Likewise.
1682 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1683 (simple_iv_with_niters): Likewise.
1684 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1685 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1686 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1687 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1688 (refine_value_range_using_guard, number_of_iterations_ne_max)
1689 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1690 (get_cst_init_from_scev, record_nonwrapping_iv)
1691 (scev_var_range_cant_overflow): Likewise.
1692 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1693 * tree-ssa-pre.c (compute_avail): Likewise.
1694 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1695 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1696 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1697 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1698 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1699 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1700 (dump_case_nodes, try_switch_expansion): Likewise.
1701 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1702 (vect_do_peeling): Likewise.
1703 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1704 * tree-vect-stmts.c (vectorizable_load): Likewise.
1705 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1706 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1707 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1708 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1709 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1710 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1711 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1712 (evrp_dom_walker::before_dom_children): Likewise.
1713 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1714 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1715 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1716 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1717 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1718 (get_range_pos_neg): Likewise.
1719 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1720 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1721 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1722 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1723 * config/avr/avr.c (avr_fold_builtin): Likewise.
1724 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1725 * config/msp430/msp430.c (msp430_attr): Likewise.
1726 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1727 * config/powerpcspe/powerpcspe-c.c
1728 (altivec_resolve_overloaded_builtin): Likewise.
1729 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1730 (rs6000_expand_ternop_builtin): Likewise.
1731 * config/rs6000/rs6000-c.c
1732 (altivec_resolve_overloaded_builtin): Likewise.
1733 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1734 (rs6000_expand_ternop_builtin): Likewise.
1735 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1737 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1739 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1740 when copying loop nest with only one inner loop.
1742 2017-10-10 Richard Biener <rguenther@suse.de>
1744 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1745 blocks if SCEV is active.
1746 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1748 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1749 (scev_initialize): Assert we are not yet initialized.
1751 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1753 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1754 inner loop's exit stmt by making it always exit the loop, otherwise
1755 we would generate an infinite empty loop.
1757 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1759 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1760 renaming variables in new preheader if it's deleted.
1762 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1764 * tree-loop-distribution.c (struct partition): Remove unused field
1765 loops of the structure.
1766 (partition_alloc, partition_free): Ditto.
1767 (build_rdg_partition_for_vertex): Ditto.
1769 2017-10-09 Jeff Law <law@redhat.com>
1771 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1772 return type to match prototype and documentation.
1774 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1776 * config/rs6000/rs6000.c (processor_costs): Move to ...
1777 * config/rs6000/rs6000.h: ... here.
1778 (rs6000_cost): Declare.
1780 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1782 * except.c (setjmp_fn): New global variable.
1783 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1784 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1785 if DONT_USE_BUILTIN_SETJMP is defined.
1787 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1789 * target.def (insn_cost): New hook.
1790 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1791 * doc/tm.texi: Regenerate.
1792 * rtlanal.c (insn_cost): Use the new hook.
1794 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1796 * combine.c (combine_validate_cost): Compute the new insn_cost,
1797 not just pattern_cost.
1798 (try_combine): Adjust comment.
1800 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1802 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1804 * combine.c (uid_insn_cost): Adjust comment.
1805 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1807 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1808 * dse.c (find_shift_sequence): Ditto.
1809 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1810 (bb_valid_for_noce_process_p): Use pattern_cost.
1811 * rtl.h (insn_rtx_cost): Delete.
1812 (pattern_cost): New prototype.
1813 (insn_cost): New prototype.
1814 * rtlanal.c (insn_rtx_cost): Rename to...
1815 (pattern_cost): ... this.
1818 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1820 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1821 (*jcc<mode>_0_r_i387): Ditto.
1822 (*jccxf_r_i387): Ditto.
1823 (*jcc<mode>_r_i387): Ditto.
1824 (*jccu<mode>_r_i387): Ditto.
1825 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1826 (*jcc): Rename from *jcc_1.
1828 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1830 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1831 deferred rescans after the lvx/stvx recombination pre-pass.
1833 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1835 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1836 memory operation instruction support.
1837 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1838 (rs6000-ibm-aix[789]*): Likewise.
1839 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1840 Document new functions.
1842 2017-10-09 Richard Biener <rguenther@suse.de>
1844 PR tree-optimization/82397
1845 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1846 equality only for semantically equal trees.
1848 2017-10-09 Richard Biener <rguenther@suse.de>
1850 PR tree-optimization/82449
1851 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1852 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1853 allow constant addresses.
1854 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1857 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1859 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1862 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1865 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1868 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1871 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1873 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1876 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1877 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1879 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1881 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1882 (WI_BINARY_PREDICATE_RESULT): Likewise.
1883 (wi::binary_traits::operator_result): New type.
1884 (wi::binary_traits::predicate_result): Likewise.
1885 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1886 (generic_wide_int::operator==, generic_wide_int::operator!=)
1887 (generic_wide_int::operator&, generic_wide_int::and_not)
1888 (generic_wide_int::operator|, generic_wide_int::or_not)
1889 (generic_wide_int::operator^, generic_wide_int::operator+
1890 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1892 (operator~, unary operator-, operator==, operator!=, operator&)
1893 (operator|, operator^, operator+, binary operator-, operator*): New
1895 * expr.c (get_inner_reference): Use wi::bit_and_not.
1896 * fold-const.c (fold_binary_loc): Likewise.
1897 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1898 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1899 (bit_value_binop): Likewise.
1900 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1901 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1902 (extract_range_from_binary_expr_1): Likewise.
1903 (masked_increment): Likewise.
1904 (simplify_bit_ops_using_ranges): Likewise.
1906 2017-10-09 Martin Jambor <mjambor@suse.cz>
1909 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1910 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1911 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1913 (hsa_fixup_mov_insn_type): New function.
1914 (hsa_op_with_type::get_in_type): Use it.
1915 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1916 immediates in an assert.
1917 (hsa_op_with_type::extend_int_to_32bit): New method.
1918 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1919 types. Convert to dest type if necessary.
1920 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1921 (reg_for_gimple_ssa): Pass false as min32int to
1922 hsa_type_for_scalar_tree_type.
1923 (gen_hsa_addr): Fixup type when creating addresable temporary.
1924 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1925 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1926 necessary. Call hsa_fixup_mov_insn_type.
1927 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1928 extend operands and convert to dest type if necessary.
1929 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1930 to dest type if necessary.
1931 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1932 if conversion nt necessary and size matches.
1933 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1934 to dest type if necessary.
1935 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1936 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1938 (gen_hsa_clrsb): Likewise.
1939 (gen_hsa_ffs): Likewise.
1940 (gen_hsa_divmod): Extend operands and convert to dest type if
1942 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1944 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1946 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1947 Remove empty default arguments. Use a brace block as output
1949 (conditional return): Ditto.
1951 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1952 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1954 (group_ending_nop): Ditto.
1955 (doloop_end): Ditto.
1956 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1957 (splitters for those): Ditto.
1959 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1961 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1962 a conditional jump (and the compare for it) so that pc_rtx is the
1964 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1965 for the deleted and renamed ctr<mode>_internal[234] patterns.
1966 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1967 Delete second conditional return pattern.
1968 (ctr<mode>_internal2): Delete this second bdnz pattern.
1969 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1970 (ctr<mode>_internal4): Delete this second bdz pattern.
1972 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1974 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1975 (always_initialized_rtx_for_ssa_name_p): New predicate.
1976 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1977 (finish_out_of_ssa): Free new field of SA.
1978 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1979 * tree-ssa-coalesce.c: Include tree-ssa.h.
1980 (get_parm_default_def_partitions): Remove extern keyword.
1981 (get_undefined_value_partitions): New function.
1982 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1983 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1985 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1987 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1989 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1991 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1993 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1994 for targets that preffer 128bit.
1996 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1998 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2000 2017-10-08 Olivier Hainque <hainque@adacore.com>
2002 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2003 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2005 (thumb_set_return_address): Likewise.
2007 2017-10-08 Olivier Hainque <hainque@adacore.com>
2009 * common/config/arm/arm-common.c (arm_except_unwind_info):
2010 Handle DWARF2_UNWIND_INFO.
2012 2017-10-07 Michael Collison <michael.collison@arm.com>
2014 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2017 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2019 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2020 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2021 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2022 defined, force the creation of a new block for a dispatch label.
2024 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2026 * invoke.texi (Wsuggest-attribute=cold): Document.
2027 * common.opt (Wsuggest-attribute=cold): New
2028 * ipa-pure-const.c (warn_function_cold): New function.
2029 * predict.c (compute_function_frequency): Use it.
2030 * predict.h (warn_function_cold): Declare.
2032 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2034 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2037 2017-10-06 Martin Liska <mliska@suse.cz>
2039 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2040 keyword for member functions.
2041 (struct sanopt_tree_couple): New struct.
2042 (struct sanopt_tree_couple_hash): New function.
2043 (struct sanopt_ctx): Add new hash_map.
2044 (has_dominating_ubsan_ptr_check): New function.
2045 (record_ubsan_ptr_check_stmt): Likewise.
2046 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2047 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2048 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2050 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2053 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2054 aarch64_simd_valid_immediate on CONST_VECTORs.
2055 (aarch64_reg_or_bic_imm): Likewise.
2057 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2059 PR rtl-optimization/82396
2060 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2062 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2064 * graphite-dependences.c (scop_get_reads): Move code to...
2065 (scop_get_must_writes): Move code to...
2066 (scop_get_may_writes): Move code to...
2067 (scop_get_reads_and_writes): ... here.
2068 (scop_get_dependences): Call scop_get_reads_and_writes.
2070 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2072 PR tree-optimization/82434
2073 * fold-const.h (can_native_encode_type_p,
2074 can_native_encode_string_p): Remove.
2075 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2076 don't encode anything, just return what would be otherwise returned.
2077 (native_encode_fixed, native_encode_complex, native_encode_vector):
2079 (native_encode_string): Likewise. Inline by hand
2080 can_native_encode_string_p.
2081 (can_native_encode_type_p): Remove.
2082 (can_native_encode_string_p): Remove.
2083 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2084 STRING_CSTs using can_native_encode_string_p, test all
2085 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2086 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2087 argument from native_encode_expr.
2088 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2089 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2092 2017-10-06 Richard Biener <rguenther@suse.de>
2094 PR tree-optimization/82397
2095 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2096 operand_equal_p but rely on data_ref_compare_tree for detecting
2098 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2099 to match up with dr_group_sort_cmp.
2101 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2104 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2106 * config/s390/s390-builtin-types.def: Regenerate.
2108 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2111 * config/s390/s390-builtin-types.def: Regenerate.
2112 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2113 Change flag from B_VXE to B_VX.
2114 (s390_vec_min_dbl): Remove B_VXE flag.
2116 2017-10-06 Richard Biener <rguenther@suse.de>
2118 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2119 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2120 translate_isl_ast_to_gimple::is_valid_rename,
2121 translate_isl_ast_to_gimple::get_rename,
2122 translate_isl_ast_to_gimple::get_def_bb_for_const,
2123 translate_isl_ast_to_gimple::get_new_name,
2124 translate_isl_ast_to_gimple::collect_all_ssa_names,
2125 translate_isl_ast_to_gimple::copy_loop_phi_args,
2126 translate_isl_ast_to_gimple::collect_all_ssa_names,
2127 translate_isl_ast_to_gimple::copy_loop_phi_args,
2128 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2129 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2130 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2131 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2132 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2133 translate_isl_ast_to_gimple::copy_cond_phi_args,
2134 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2135 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2136 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2137 translate_isl_ast_to_gimple::rename_uses,
2138 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2139 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2140 (set_rename_for_each_def): Likewise.
2141 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2142 here. Handle rewriting SCEV analyzable uses here.
2143 (copy_bb_and_scalar_dependences): Generate code for PHI
2145 (graphite_regenerate_ast_isl): Adjust.
2146 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2147 (add_write, add_read): New functions.
2148 (build_cross_bb_scalars_def): Use it and simplify.
2149 (build_cross_bb_scalars_use): Likewise.
2150 (graphite_find_cross_bb_scalar_vars): Inline into...
2151 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2152 simulating out-of-SSA. Compute liveout and add dependencies.
2153 (build_scops): Force an empty entry block.
2154 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2156 (sese_build_liveouts): Declare.
2157 (sese_trivially_empty_bb_p): Likewise.
2158 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2159 compute liveout and debug_liveout.
2160 (sese_bad_liveouts_use): Remove.
2161 (sese_reset_debug_liveouts_bb): Likewise.
2162 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2163 (sese_build_liveouts): Build liveout and debug_liveout and store
2165 (new_sese_info): Adjust.
2166 (free_sese_info): Likewise.
2167 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2168 do not build liveout here.
2169 (move_sese_in_condition): Adjust region entry.
2170 (scev_analyzable_p): Match up with chrec_apply requirements.
2171 (sese_trivially_empty_bb_p): New.
2172 * tree-into-ssa.c (get_reaching_def): Properly support generating
2173 default-defs for incremental rewrite of anonymous names.
2175 2017-10-06 Richard Biener <rguenther@suse.de>
2177 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2178 precision do not perform modulo reduction.
2180 2017-10-06 Richard Biener <rguenther@suse.de>
2182 PR tree-optimization/82436
2183 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2184 conservatively choose the vectorization factor when checking
2185 whether we can perform the required load permutation.
2186 (vect_transform_slp_perm_load): Assert when we may not fail.
2188 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2190 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2191 message for incompatible -msdata=* and -mcall-* options.
2193 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2195 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2196 rate for post-reload scheduling.
2198 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2200 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2202 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2204 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2205 to improve monte carlo in scimark.
2207 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2209 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2210 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2211 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2212 pentium4_cost, nocona_cost): Set reassociation width to 1.
2213 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2214 width to 2 for fp operations and 1 otherwise.
2215 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2217 (atom_cost): Set reassociation width to 2.
2218 (slm_cost, generic_cost): Set fp reassociation width
2219 to 2 and 1 otherwise.
2220 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2221 (core_cost): Set fp reassociation width to 4 and vector to 2.
2222 (ix86_reassociation_width): Rewrite using cost table; special case
2223 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2224 and TARGET_AVX128_OPTIMAL.
2225 * config/i386/i386.h (processor_costs): Add
2226 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2227 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2228 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2229 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2230 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2231 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2233 2017-10-05 Nathan Sidwell <nathan@acm.org>
2235 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2237 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2239 * config/arm/arm.c (arm_test_fpu_data): New.
2240 (arm_run_selftests): Call arm_test_fpu_data.
2242 2017-10-04 Nathan Sidwell <nathan@acm.org>
2244 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2246 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2248 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2250 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2251 check type for aarch64_simd_valid_immediate.
2252 (aarch64_output_simd_mov_immediate): Update prototype.
2253 (aarch64_simd_valid_immediate): Update prototype.
2254 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2255 support for ORR-immediate.
2256 (and<mode>3): modified pattern to add support for BIC-immediate.
2257 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2258 now checks for valid immediate for BIC and ORR based on new enum
2260 (aarch64_output_simd_mov_immediate): Function now used to output
2261 BIC/ORR imm as well based on new enum argument.
2262 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2264 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2265 (aarch64_reg_or_bic_imm): Likewise.
2267 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2269 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2270 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2272 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2276 PR rtl-optimization/82396
2277 * haifa-sched.c (autopref_multipass_init): Simplify
2279 (autopref_rank_data): Simplify sort order.
2280 * sched-int.h (autopref_multipass_data_): Remove
2281 multi_mem_insn_p, min_offset and max_offset.
2283 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2285 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2287 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2289 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2292 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2294 * doc/sourcebuild.texi: Document vect_long_mult.
2296 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2298 PR tree-optimization/82413
2299 * fold-const.c (build_range_check): Use widest_int when comparing
2300 the maximum ETYPE value with HIGH.
2302 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2304 PR rtl-optimization/82396
2305 * haifa-sched.c (autopref_multipass_init): Simplify
2307 (autopref_rank_data): Simplify sort order.
2308 * sched-int.h (autopref_multipass_data_): Remove
2309 multi_mem_insn_p, min_offset and max_offset.
2311 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2313 PR tree-optimization/82381
2314 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2315 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2318 PR tree-optimization/82374
2319 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2320 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2321 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2322 current_function_decl to the new decl.
2324 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2326 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2327 helper macro for IEEE float128 hardware built-in functions.
2328 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2330 (TRUNCF128_ODD): Likewise.
2331 (ADDF128_ODD): Likewise.
2332 (SUBF128_ODD): Likewise.
2333 (MULF128_ODD): Likewise.
2334 (DIVF128_ODD): Likewise.
2335 (FMAF128_ODD): Likewise.
2336 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2337 UNSPEC_TRUNC_ROUND_TO_ODD.
2338 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2339 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2340 floating point round to odd instructions.
2341 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2342 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2343 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2344 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2345 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2346 (trunc<mode>sf2_hw): Change the truncate with round to odd
2347 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2348 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2349 to odd hardware instructions.
2350 (sub<mode>3_odd): Likewise.
2351 (mul<mode>3_odd): Likewise.
2352 (div<mode>3_odd): Likewise.
2353 (sqrt<mode>2_odd): Likewise.
2354 (fma<mode>4_odd): Likewise.
2355 (fms<mode>4_odd): Likewise.
2356 (nfma<mode>4_odd): Likewise.
2357 (nfms<mode>4_odd): Likewise.
2358 (trunc<mode>df2_odd): Change the truncate with round to odd
2359 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2361 * doc/extend.texi (PowerPC built-in functions): Update documentation
2362 for existing IEEE float128-bit built-in functions. Add built-in
2363 functions that generate the IEEE 128-bit floating point round to
2366 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2368 PR rtl-optimization/77729
2369 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2370 to (X&(C1&~C2))|C2 transformations.
2372 2017-10-03 Martin Jambor <mjambor@suse.cz>
2374 PR tree-optimization/82363
2375 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2376 mismatch, mark lacc written regardless of racc.
2378 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2380 PR tree-optimization/82381
2381 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2382 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2383 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2384 fallthrough into reassoc_stmt_dominates_stmt_p.
2387 * combine.c (combine_instructions): Don't combine in unreachable
2390 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2393 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2394 function to not use the have_cpu variable. Do not set cpu_index,
2395 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2397 (rs6000_valid_attribute_p): Remove duplicate initializations of
2398 old_optimize and func_optimize.
2399 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2400 (rs6000_activate_target_options): Make global.
2401 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2404 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2406 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2407 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2408 Set *poffset to 0 and *psize and *pmax_size to -1 if
2409 *poffset + *psize overflows in HOST_WIDE_INT.
2411 PR tree-optimization/82387
2412 PR tree-optimization/82388
2413 PR tree-optimization/82389
2414 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2415 instead of live_bytes non-NULL.
2417 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2420 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2423 2017-10-02 Richard Biener <rguenther@suse.de>
2425 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2426 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2427 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2429 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2431 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2432 requested precision matches the type's.
2433 * calls.c (alloc_max_size): Calculate the new candidate size as
2434 a widest_int and use wi::to_widest when comparing it with the
2435 current candidate size.
2436 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2437 zero rather than integer_zero_node.
2438 * match.pd: Check for a no-op conversion before using wi::add
2439 rather than after. Use tree_to_uhwi when summing small shift
2440 counts into an unsigned int.
2442 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2443 Alan Hayward <alan.hayward@arm.com>
2444 David Sherwood <david.sherwood@arm.com>
2447 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2448 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2449 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2450 POINTER_AND_FP_REGS.
2452 2017-10-02 Richard Biener <rguenther@suse.de>
2454 PR tree-optimization/82355
2455 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2456 a mapping for the enclosing loop but avoid generating one for
2458 (copy_bb_and_scalar_dependences): Remove premature codegen
2459 error on PHIs in blocks duplicated into multiple places.
2460 * graphite-scop-detection.c
2461 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2462 in the region use it as loop and nest to analyze the DR in.
2463 (try_generate_gimple_bb): Likewise.
2464 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2465 (add_loop_constraints): For blocks in a loop not in the region
2466 create a dimension with a single iteration.
2467 * sese.h (gbb_loop_at_index): Remove assert.
2469 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2471 * omp-expand.c (adjust_context_scope): New function.
2472 (expand_parallel_call): Call adjust_context_scope.
2474 2017-10-01 Jeff Law <law@redhat.com>
2476 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2477 dom_opt_dom_walker class with direct access to private members.
2478 Add comments. Call test_for_singularity.
2479 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2480 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2481 m_dummy_cond anymore.
2482 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2484 (pass_dominator:execute): Build the dummy_cond here and pass it
2485 to the dom_opt_dom_walker ctor.
2486 (test_for_singularity): New function.
2488 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2489 Maya Rashish <coypu@sdf.org>
2491 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2492 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2493 (alpha*-*-netbsd*) Use nbsd_tm_file.
2494 (arm*-*-netbsdelf*) Likewise.
2495 (i[34567]86-*-netbsdelf*) Likewise.
2496 (x86_64-*-netbsd*) Likewise.
2497 (mips*-*-netbsd*) Likewise.
2498 (powerpc-*-netbsd*) Likewise.
2499 (sh*-*-netbsd*) Likewise.
2500 (sparc-*-netbsdelf*) Likewise.
2501 (sparc64-*-netbsd*) Likewise.
2502 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2504 (vax-*-netbsdelf*) Likewise.
2505 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2506 (UINT_FAST8_TYPE) Likewise.
2507 (INT_FAST16_TYPE) Check CHAR_FAST16.
2508 (UINT_FAST16_TYPE) Likewise.
2510 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2513 * config/i386/i386.md
2514 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2515 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2516 *divmodsi4_zext_2): New define_insn_and_split.
2517 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2518 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2519 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2520 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2521 New define_insn_and_split.
2522 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2523 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2524 operands[1] having DImode when mode is SImode.
2526 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2527 always SImode for DIV and MOD in REG_EQUAL notes.
2529 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2532 * match.pd: Fix handling of NaNs in pattern.
2534 2017-09-29 Jeff Law <law@redhat.com>
2536 * sbitmap.c (bitmap_bit_in_range_p): New function.
2537 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2538 * tree-ssa-dse.c (live_bytes_read): New function.
2539 (dse_classify_store): Ignore reads of dead bytes.
2541 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2542 typos and whitespace errors.
2543 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2544 * config/s390/s390.c (s390_emit_prologue): Likewise.
2546 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2549 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2550 with a symbol for LRA.
2552 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2554 PR rtl-optimization/82338
2555 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2557 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2559 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2560 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2561 (qsort_chk): Declare.
2562 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2563 (qsort_chk): New function.
2565 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2567 PR tree-optimization/82337
2568 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2569 phi definition if the PHI result appears in an abnormal PHI.
2570 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2571 basis appears in an abnormal PHI.
2573 2017-09-29 Richard Biener <rguenther@suse.de>
2575 * graphite-isl-ast-to-gimple.c
2576 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2577 (binary_op_to_tree): Use it.
2578 (get_rename_from_scev): Likewise.
2579 (copy_loop_phi_nodes): Likewise.
2580 (copy_bb_and_scalar_dependences): Likewise.
2581 (translate_pending_phi_nodes): Likewise.
2583 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2586 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2587 for movabsq $(i32 << shift), r64.
2589 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2591 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2592 index when encoding %esp as %rsp to avoid 0x67 prefix.
2594 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2596 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2597 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2599 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2601 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2602 Extensions with more than 16 double VFP registers.
2603 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2604 to_clear_mask and all code related to it. Replace the remaining
2605 entry by a sbitmap and adapt code accordingly.
2607 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2609 * brig-builtins.def: Change pure attributes to const.
2611 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2613 * config.gcc (default_gnu_indirect_function): Default to yes for
2614 sparc*-*-linux* with glibc.
2616 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2618 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2619 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2620 when creating .init_array and .fini_array sections with priority
2623 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2626 * config/aarch64/aarch64.c
2627 (aarch64_builtin_support_vector_misalignment): Always return false
2628 when misalignment is unknown.
2630 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2632 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2633 this function to return false if the definition used by the swap
2634 instruction is artificial, or if the memory address from which the
2635 constant value is loaded is not represented by a base address held
2636 in a register or if the base address register is a frame or stack
2637 pointer. Additionally, return false if the base address of the
2638 loaded constant is a SYMBOL_REF but is not considered to be a
2640 (replace_swapped_load_constant): New function.
2641 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2642 loaded constant vector with a load of a swapped constant vector.
2644 2017-09-27 Carl Love <cel@us.ibm.com>
2646 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2647 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2648 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2651 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2653 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2654 first, always call autopref_rank_data otherwise.
2656 2017-09-27 Richard Biener <rguenther@suse.de>
2658 * graphite-scop-detection.c (find_scop_parameters): Move
2659 loop bound handling ...
2660 (gather_bbs::before_dom_children): ... here, avoiding the need
2661 to build scop_info->loop_nest.
2662 (record_loop_in_sese): Remove.
2663 * sese.h (sese_info_t::loop_nest): Remove.
2664 * sese.c (new_sese_info): Do not allocate loop_nest.
2665 (free_sese_info): Do not free loop_nest.
2667 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2670 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2671 lhs from calls if the lhs has addressable type.
2673 2017-09-27 Richard Biener <rguenther@suse.de>
2675 * graphite.h (scop::max_alias_set): New member.
2676 * graphite-scop-detection.c: Remove references to non-existing
2677 --param in comments.
2678 (build_alias_sets): Record the maximum alias set used for drs.
2679 (build_scops): Support zero as unlimited for
2680 --param graphite-max-arrays-per-scop.
2681 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2683 (build_poly_sr_1): ... here. Compute alias set based on the
2684 maximum alias set used for drs rather than
2685 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2687 2017-09-27 Richard Biener <rguenther@suse.de>
2689 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2690 --param loop-block-tile-size=0 to disable tiling.
2692 2017-09-27 Richard Biener <rguenther@suse.de>
2694 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2695 (graphite-max-nb-scop-params): Document special value zero.
2696 * domwalk.h (dom_walker::STOP): New symbolical constant.
2697 (dom_walker::dom_walker): Add optional parameter for bb to
2699 (dom_walker::~dom_walker): Declare.
2700 (dom_walker::before_dom_children): Document STOP return value.
2701 (dom_walker::m_user_bb_to_rpo): New member.
2702 (dom_walker::m_bb_to_rpo): Likewise.
2703 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2704 mapping here if not provided by the user.
2705 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2706 provided by the user.
2707 (dom_walker::STOP): Define.
2708 (dom_walker::walk): Do not compute bb to RPO mapping here.
2709 Support STOP return value from before_dom_children to stop
2711 * graphite-optimize-isl.c (optimize_isl): If the schedule
2712 is the same still generate code if -fgraphite-identity
2713 or -floop-parallelize-all are given.
2714 * graphite-scop-detection.c: Include cfganal.h.
2715 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2717 (gather_bbs::before_dom_children): Return STOP for BBs
2719 (build_scops): Compute bb to RPO mapping and pass it to
2720 the domwalk. Treat --param graphite-max-nb-scop-params=0
2721 as not limiting the number of params.
2722 * graphite.c (graphite_initialize): Remove limit on the
2723 number of basic-blocks in a function.
2724 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2725 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2726 default value of 10.
2728 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2730 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2731 Adjust code to eliminate needing to do the shift right 32-bits
2732 operation after XSCVDPSPN.
2734 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2736 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2737 ((X / Y) != 0 -> X >= Y): Likewise.
2739 2017-09-26 Carl Love <cel@us.ibm.com>
2741 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2742 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2743 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2744 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2745 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2746 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2747 definitions and overloading.
2748 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2749 statement for P9V_BUILTIN_XST_LEN_R.
2750 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2751 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2752 define_expand and define_insn for the instructions and builtins.
2753 * doc/extend.texi: Update the built-in documentation file for the new
2755 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2756 define_insn for the instructions
2758 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2761 * gcc/config/netbsd-protos.h: New file.
2762 * gcc/config/netbsd.c: New file.
2763 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2764 * gcc/config/t-netbsd: New file.
2765 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2766 (tmake_file) Add t-netbsd.
2767 (extra_objs) Add netbsd.o.
2769 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2773 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2775 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2777 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2778 sign extension from a vector register to a GPR by doing a 32-bit
2779 direct move and then an EXTSW.
2780 (extendsi<mode>2 splitter): Likewise.
2781 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2782 right or vector extract after doing XSCVDPSPN. Use
2783 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2785 (movdi_from_sf_zero_ext): Likewise.
2786 (reload_gpr_from_vsxsf): Likewise.
2787 (p8_mfvsrd_4_disf): Delete, no longer used.
2788 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2789 then needing to move the SFmode to a GPR to use the XSCVDPSP
2790 instruction instead of FRSP and XSCVDPSPN.
2791 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2792 it is adjacent to the other XSCVSPDP insns.
2793 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2794 SFmode to be in traditional Altivec registers.
2795 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2796 (vsx_xscvspdpn): Likewise.
2797 (vsx_xscvspdpn_scalar): Likewise.
2799 2017-09-26 Martin Jambor <mjambor@suse.cz>
2801 * tree-sra.c (compare_access_positions): Put integral types first,
2802 stabilize sorting of integral types, remove conditions putting
2803 non-full-precision integers last.
2804 (sort_and_splice_var_accesses): Disable scalarization if a
2805 non-integert would be represented by a non-full-precision integer.
2807 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2809 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2810 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2811 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2812 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2813 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2814 conditionals inside the function instead of around it. Call
2815 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2816 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2818 2017-09-26 Richard Biener <rguenther@suse.de>
2820 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2822 (scop_detection::build_scop_breadth): ... this. Removed.
2823 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2824 (scop_detection::harmful_stmt_in_bb): Likewise.
2825 (scop_detection::graphite_can_represent_stmt): Likewise.
2826 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2827 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2828 (scop_detection::can_represent_loop_1): ... this. Removed.
2829 (scop_detection::harmful_loop_in_region): Simplify after inlining
2830 the above and remove more quadraticness.
2831 (build_scops): Adjust.
2832 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2835 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2838 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2839 REGNO (base) == SP_REG if base is a REG.
2842 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2843 if it is different SSA_NAME.
2844 (optimize_range_tests_cmp_bitwise): New function.
2845 (optimize_range_tests): Call it.
2847 2017-09-26 Richard Biener <rguenther@suse.de>
2849 PR tree-optimization/82321
2850 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2851 for the def being inside the loop.
2853 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2855 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2857 * config/s390/s390-builtins.def: Fix constraint on op4.
2859 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2861 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2862 independent expanders.
2863 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2864 ("vec_ordered", "vec_unordered"): New expanders.
2866 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2868 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2871 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2873 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2874 vec_unpacks_lo_v16qi.
2875 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2877 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2879 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2880 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2881 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2883 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2885 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2887 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2889 ("*vec_slb<mode>"): New insn pattern.
2890 ("vec_shr_<mode>"): New expander.
2891 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2892 and force the shift count operand to V16QImode.
2893 ("vec_srb<mode>"): Set shift count mode to V16QI.
2895 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2897 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2898 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2899 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2901 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2904 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2906 2017-09-26 Richard Biener <rguenther@suse.de>
2908 PR tree-optimization/82320
2909 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2912 2017-09-25 Jeff Law <law@redhat.com>
2914 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2915 prototype for new argument.
2916 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2917 mostly extracted from rs6000_emit_allocate_stack.
2918 (rs6000_emit_probe_stack_range_stack_clash): New function.
2919 (rs6000_emit_allocate_stack): Call
2920 rs6000_emit_probe_stack_range_stack_clash as needed.
2921 (rs6000_emit_probe_stack_range): Add additional argument
2922 to call to gen_probe_stack_range{si,di}.
2923 (output_probe_stack_range): New.
2924 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2925 (output_probe_stack_range_stack_clash): New.
2926 (rs6000_emit_prologue): Emit notes into dump file as requested.
2927 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2928 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2929 Add additional operand and pass it to output_probe_stack_range.
2931 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2933 PR tree-optimization/82163
2934 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2935 (checking_verify_loop_closed_ssa): New parameter.
2936 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2937 (check_loop_closed_ssa_stmt): Delete.
2938 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2939 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2940 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2943 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2945 * brig-builtins.def: Treat HSAIL barrier builtins as
2946 setjmp/longjump style functions.
2948 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2950 * target.def (constant_alignment): New hook.
2951 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2952 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2953 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2954 * doc/tm.texi: Regenerate.
2955 * targhooks.h (default_constant_alignment): Declare.
2956 (constant_alignment_word_strings): Likewise.
2957 * targhooks.c (default_constant_alignment): New function.
2958 (constant_alignment_word_strings): Likewise.
2959 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2960 instead of CONSTANT_ALIGNMENT.
2961 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2962 (force_const_mem): Likewise.
2963 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2964 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2965 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2966 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2967 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2969 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2970 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2971 constant_alignment_word_strings.
2972 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2973 (CONSTANT_ALIGNMENT): Likewise.
2974 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2975 (arm_constant_alignment): New function.
2976 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2977 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2978 constant_alignment_word_strings.
2979 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2980 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2981 constant_alignment_word_strings.
2982 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2983 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2984 (cris_constant_alignment): New function.
2985 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2986 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2987 (epiphany_constant_alignment): New function.
2988 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2989 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2990 constant_alignment_word_strings.
2991 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2992 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2993 constant_alignment_word_strings.
2994 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2995 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2996 constant_alignment_word_strings.
2997 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2998 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2999 * config/i386/i386.c (ix86_constant_alignment): Make static.
3000 Use the same interface as the target hook.
3001 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3002 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3003 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3004 constant_alignment_word_strings.
3005 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3006 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3007 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3008 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3009 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3010 constant_alignment_word_strings.
3011 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3012 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3013 constant_alignment_word_strings.
3014 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3015 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3016 constant_alignment_word_strings.
3017 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3018 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3020 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3021 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3022 * config/mips/mips.c (mips_constant_alignment): New function.
3023 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3024 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3025 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3026 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3027 (mmix_constant_alignment): Make static. Use the same interface
3029 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3030 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3031 constant_alignment_word_strings.
3032 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3033 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3034 constant_alignment_word_strings.
3035 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3036 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3037 constant_alignment_word_strings.
3038 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3039 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3040 (rs6000_constant_alignment): New function.
3041 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3042 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3043 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3044 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3045 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3046 (rs6000_constant_alignment): New function.
3047 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3048 * config/s390/s390.c (s390_constant_alignment): New function.
3049 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3050 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3051 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3052 constant_alignment_word_strings.
3053 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3054 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3055 (sparc_constant_alignment): New function.
3056 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3057 * config/spu/spu.c (spu_constant_alignment): New function.
3058 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3059 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3060 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3061 constant_alignment_word_strings.
3062 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3063 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3064 constant_alignment_word_strings.
3065 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3066 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3067 constant_alignment_word_strings.
3068 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3069 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3070 (visium_constant_alignment): New function.
3071 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3072 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3073 (xtensa_constant_alignment): New function.
3074 * system.h (CONSTANT_ALIGNMENT): Poison.
3076 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3079 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3080 (rs6000_builtin_valid_without_lhs): New helper function.
3081 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3082 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3084 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3086 * target.h (vec_perm_indices): Use unsigned short rather than
3088 (auto_vec_perm_indices): Likewise.
3089 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3090 Use unsigned int rather than unsigned char.
3091 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3093 2017-09-25 Richard Biener <rguenther@suse.de>
3095 * cfgloop.h (sort_sibling_loops): Declare.
3096 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3097 (sort_sibling_loops): New function sorting the sibling loop list
3099 * graphite.c (graphite_transform_loops): Sort sibling loops.
3101 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3103 * target.def (vec_perm_const_ok): Change sel parameter to
3105 * optabs-query.c (can_vec_perm_p): Update accordingly.
3106 * doc/tm.texi: Regenerate.
3107 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3108 auto_vec_perm_indices and remove separate nelt field.
3109 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3110 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3111 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3112 (aarch64_expand_vec_perm_const): Update accordingly.
3113 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3114 to vec_perm_indices.
3115 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3116 auto_vec_perm_indices and remove separate nelt field.
3117 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3118 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3119 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3121 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3122 to vec_perm_indices.
3123 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3124 sel to vec_perm_indices.
3125 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3126 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3127 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3129 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3132 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3135 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3136 on the FUNCTION_DECL function context if it has a DIE that is a
3139 2017-09-25 Richard Biener <rguenther@suse.de>
3141 PR tree-optimization/82285
3142 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3145 2017-09-25 Tom de Vries <tom@codesourcery.com>
3149 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3152 2017-09-25 Richard Biener <rguenther@suse.de>
3154 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3155 ISL errors other than isl_error_quota happen. Dump if the
3156 schedule is the same.
3157 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3158 errors instead of aborting inside ISL.
3160 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3163 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3164 of libgcc_eh for m64.
3165 * config/i386/darwin64.h: Likewise.
3167 2017-09-25 Richard Biener <rguenther@suse.de>
3170 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3171 attribute for incomplete types nor twice for complete ones.
3173 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3176 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3177 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3180 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3183 * config/i386/i386.opt (mprefer-avx256): Use
3184 ix86_target_flags variable.
3185 * config/i386/i386.c (ix86_target_string): Move
3186 -mprefer-avx256 to flag2_opts.
3188 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3191 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3192 and x != -1 | y != -1 into (x & y) != -1.
3194 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3196 * config.gcc: Add new case statement to set
3197 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3198 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3199 s390x-*-linux* case statements. Added aarch64 to the list of
3200 supported architectures.
3202 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3204 PR tree-optimization/82289
3205 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3206 STMT_VINFO_RELEVANT_P.
3208 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3209 Alan Hayward <alan.hayward@arm.com>
3210 David Sherwood <david.sherwood@arm.com>
3212 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3213 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3214 (extract_range_from_binary_expr_1): Don't call
3215 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3217 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3218 Alan Hayward <alan.hayward@arm.com>
3219 David Sherwood <david.sherwood@arm.com>
3221 * target.def (preferred_vector_alignment): New hook.
3222 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3224 * doc/tm.texi: Regenerate.
3225 * targhooks.h (default_preferred_vector_alignment): Declare.
3226 * targhooks.c (default_preferred_vector_alignment): New function.
3227 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3229 (DR_TARGET_ALIGNMENT): New macro.
3230 (aligned_access_p): Update commentary.
3231 (vect_known_alignment_in_bytes): New function.
3232 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3234 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3235 Calculate the misalignment based on the target alignment rather than
3237 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3238 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3239 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3240 the target alignment, rather than masking the element misalignment
3241 with the number of elements in a vector. Also use the target
3242 alignment when calculating the maximum number of peels.
3243 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3244 instead of TYPE_ALIGN_UNIT.
3245 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3246 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3247 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3248 (vect_create_data_ref_ptr): Likewise.
3249 (vect_setup_realignment): Realign by ANDing with
3250 -DR_TARGET_MISALIGNMENT.
3251 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3252 the number of peels based on DR_TARGET_ALIGNMENT.
3253 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3254 with the guaranteed alignment boundary when deciding whether
3256 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3257 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3258 (ensure_base_align): Remove stmt_info parameter. Get the
3259 target base alignment from DR_TARGET_ALIGNMENT.
3260 (vectorizable_store): Update call accordingly. Interpret
3261 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3263 (vectorizable_load): Likewise.
3265 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3266 Alan Hayward <alan.hayward@arm.com>
3267 David Sherwood <david.sherwood@arm.com>
3269 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3270 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3271 (vect_enhance_data_refs_alignment): Likewise.
3273 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3275 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3276 error. Only quit immediately if parsing is complete.
3277 (BEGIN): Initialize fatal_err and parse_done.
3278 (begin fpu, end fpu): Check number of arguments.
3279 (begin arch, end arch): Likewise.
3280 (begin cpu, end cpu): Likewise.
3281 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3282 (optalias): Likewise.
3284 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3286 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3287 * config/arm/arm-isa.h: Delete. Move definitions to ...
3288 * arm-cpus.in: ... here. Use new feature and fgroup values.
3289 * config/arm/arm.c (arm_option_override): Use lower case for feature
3291 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3292 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3293 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3295 (print_isa_bits_for): New function.
3296 (gen_isa): New function.
3297 (gen_comm_data): Use print_isa_bits_for.
3298 (define feature): New keyword.
3299 (define fgroup): New keyword.
3300 * config/arm/t-arm (TM_H): Remove.
3301 (GTM_H): Add arm-isa.h.
3302 (arm-isa.h): Add rule to generate file.
3303 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3304 case for feature bit names.
3306 2017-09-22 Richard Biener <rguenther@suse.de>
3308 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3310 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3311 print of no dependency loops ...
3312 * graphite.c (graphite_transform_loops): ... here.
3313 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3315 (same_close_phi_node, remove_duplicate_close_phi,
3316 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3317 (canonicalize_loop_closed_ssa): ... here and simplify.
3318 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3319 (optimize_isl): Use dump_printf_loc to tell when we stopped
3320 optimizing because of an ISL timeout.
3322 2017-09-22 Richard Biener <rguenther@suse.de>
3324 PR tree-optimization/82291
3325 * tree-if-conv.c (predicate_mem_writes): Make sure to
3326 remove writes in blocks predicated with false.
3328 2017-09-22 Richard Biener <rguenther@suse.de>
3330 * sese.c: Include cfganal.h.
3331 (if_region_set_false_region): Remove.
3332 (create_if_region_on_edge): Likewise.
3333 (move_sese_in_condition): Re-implement without destroying
3336 2017-09-22 Richard Biener <rguenther@suse.de>
3338 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3339 Verify both BBs contain loop PHI nodes before dispatching to
3341 (graphite_regenerate_ast_isl): Do not recompute dominators,
3342 do not verify three times. Restructure for clarity.
3343 * graphite-scop-detection.c (same_close_phi_node,
3344 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3345 defined_in_loop_p, canonicalize_loop_closed_ssa,
3346 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3347 checking and SSA rewrite, move to ...
3348 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3349 (graphite_initialize): Do not pass in ctx, do not reset the
3350 SCEV cache, compute only dominators.
3351 (graphite_transform_loops): Allocate ISL ctx after
3352 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3353 Maintain post-dominators only around build_scops.
3354 * sese.c (if_region_set_false_region): Make static. Free
3355 and recompute dominators.
3356 (move_sese_in_condition): Assert we don't get called with
3357 post-dominators computed.
3358 * sese.h (if_region_set_false_region): Remove.
3360 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3362 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3363 mode attribute for TARGET_AVX512VL.
3365 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3367 * config/i386/i386.opt (mprefer-avx256): New option.
3368 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3370 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3371 for TARGET_PREFER_AVX256.
3372 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3374 2017-09-21 Jeff Law <law@redhat.com>
3376 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3377 Fix dump output if the only stack space is for pushed registers.
3379 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3381 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3384 2017-09-21 Martin Sebor <msebor@redhat.com>
3387 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3388 code (in C++) or code that triggers warnings.
3390 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3392 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3394 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3396 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3397 * sched-rgn.c: Ditto.
3398 * sel-sched-ir.c: Ditto.
3400 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3402 * toplev.h (set_random_seed): Adjust return type.
3403 * toplev.c (init_local_tick): Move eager initialization of random_seed
3404 to get_random_seed. Adjust comment.
3405 (init_random_seed): Inline to get_random_seed, delete.
3406 (get_random_seed): Initialize random_seed lazily.
3407 (set_random_seed): Do not return previous value.
3408 (print_switch_value): Do not call get_random_seed.
3410 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3412 * cgraph.c (delete_function_version): New, broken out from...
3413 (cgraph_node::delete_function_version): ...here. Rename to
3414 cgraph_node::delete_function_version_by_decl. Update all uses.
3415 (cgraph_node::remove): Call delete_function_version.
3417 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3420 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3421 VAR_DECLs to which addressable non-volatile parameters are mapped
3422 and for id->retvar after the return value assignment. Clear
3423 id->retval and id->retbnd after inlining.
3425 2017-09-21 Richard Biener <rguenther@suse.de>
3427 PR tree-optimization/82276
3428 PR tree-optimization/82244
3429 * tree-vrp.c (build_assert_expr_for): Set
3430 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3432 (remove_range_assertions): Revert earlier change.
3434 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3437 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3439 2017-09-21 Richard Biener <rguenther@suse.de>
3441 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3442 Restore valid IL after code generation errors.
3443 * graphite.c (graphite_transform_loops): Diagnose code
3444 generation issues as MSG_MISSED_OPTIMIZATION and continue
3445 with processing SCOPs.
3447 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3448 Alan Hayward <alan.hayward@arm.com>
3449 David Sherwood <david.sherwood@arm.com>
3451 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3452 rather than choosing between plus_constant and gen_rtx_<CODE>.
3453 * expr.c (emit_push_insn): Likewise.
3454 (expand_expr_real_2): Likewise.
3456 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3457 Alan Hayward <alan.hayward@arm.com>
3458 David Sherwood <david.sherwood@arm.com>
3460 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3462 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3463 Alan Hayward <alan.hayward@arm.com>
3464 David Sherwood <david.sherwood@arm.com>
3466 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3467 calling tree_to_uhwi.
3469 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3470 Alan Hayward <alan.hayward@arm.com>
3471 David Sherwood <david.sherwood@arm.com>
3473 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3474 INTEGER_CST rather than a negative test for ADDR_EXPR.
3476 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3477 Alan Hayward <alan.hayward@arm.com>
3478 David Sherwood <david.sherwood@arm.com>
3480 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3481 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3483 2017-09-21 Richard Biener <rguenther@suse.de>
3485 PR tree-optimization/71351
3486 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3487 graphite_create_new_loop_guard): Remove, fold remaining parts
3489 (translate_isl_ast_node_for): ... here and simplify.
3491 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3494 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3495 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3496 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3497 alternative always use QI mode, for -Os imov (=R,R) alternative
3498 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3501 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3502 Jeff Law <law@redhat.com>
3504 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3505 (allocate_stack_space): New function, partially extracted from
3507 (s390_emit_prologue): Track offset to most recent stack probe.
3508 Code to allocate space moved into allocate_stack_space.
3509 Dump actions when no stack is allocated.
3510 (s390_prologue_plus_offset): New function.
3511 (s390_emit_stack_probe): Likewise.
3513 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3515 * common.opt (Wa, Wl, Wp, g, gz=): Add
3517 (gno-column-info): Remove.
3518 (gcolumn-info): Drop RejectNegative.
3520 (gno-record-gcc-switches): Remove.
3521 (grecord-gcc-switches): Drop RejectNegative.
3522 (gno-split-dwarf): Remove.
3523 (gsplit-dwarf): Drop RejectNegative.
3524 (gno-strict-dwarf): Remove.
3525 (gstrict-dwarf): Drop RejectNegative.
3526 * config/darwin.opt (gfull, gused): Add RejectNegative.
3527 * dwarf2out.c (gen_producer_string): Drop
3528 gno-record-gcc-switches handler.
3529 * optc-gen.awk: Add g to prefixes with negative forms.
3530 * opts-common.c (remapping_prefix_p): New.
3531 (find_opt): Check it.
3532 (generate_canonical_option): Test g prefix.
3533 (option_map): Add -gno- mapping.
3534 (add_misspelling_candidates): Check remapping_prefix_p.
3536 2017-09-20 Jeff Law <law@redhat.com>
3538 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3539 thinko in stack clash protection support.
3541 * explow.c (compute_stack_clash_protection_loop_data): Use
3542 CONST_INT_P instead of explicit test. Verify object is a
3543 CONST_INT_P before looking at INTVAL.
3544 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3545 instead of explicit test.
3547 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3550 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3551 address instead of to r1 and r11.
3553 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3555 * config.gcc: Support "knm".
3556 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3557 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3559 * config/i386/i386.c (m_KNM): Define.
3560 (processor_target_table): Add "knm".
3562 (ix86_option_override_internal): Add "knm".
3563 (ix86_issue_rate): Add PROCESSOR_KNM.
3564 (ix86_adjust_cost): Ditto.
3565 (ia32_multipass_dfa_lookahead): Ditto.
3566 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3567 (fold_builtin_cpu): Add M_INTEL_KNM.
3568 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3569 (processor_type): Add PROCESSOR_KNM.
3570 * config/i386/x86-tune.def: Add m_KNM.
3571 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3573 2017-09-20 Richard Biener <rguenther@suse.de>
3575 PR tree-optimization/80213
3576 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3577 are allowed in empty BBs as well.
3578 (canonicalize_loop_closed_ssa): Also look for other complex
3580 (scop_detection::get_sese): Include the loop-closed PHI block
3582 (scop_detection::merge_sese): Remove code adding extra blocks.
3583 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3584 (build_scops): Assert the final returned scop is invalid.
3586 2017-09-20 Richard Biener <rguenther@suse.de>
3588 PR tree-optimization/82264
3589 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3591 (vn_phi_lookup): Likewise.
3592 (vn_phi_insert): Likewise.
3594 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3596 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3597 that fits into uhwi or shwi, add DW_AT_const_value regardless
3598 of early_dwarf without going through RTL, using add_AT_unsigned
3601 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3602 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3603 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3604 (DEBUG_MACRO_SECTION): Likewise.
3605 (DEBUG_LTO_MACRO_SECTION): Likewise.
3606 (DEBUG_STR_DWO_SECTION): Likewise.
3607 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3608 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3609 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3610 (DEBUG_LTO_LINE_STR_SECTION): Define.
3611 (init_sections_and_labels): Initialize debug_line_str_section
3612 variable. Initialize debug_loc_section for -gdwarf-5 to
3613 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3615 2017-09-20 Richard Biener <rguenther@suse.de>
3617 * graphite-sese-to-poly.c (extract_affine): Properly handle
3618 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3620 2017-09-20 Richard Biener <rguenther@suse.de>
3622 PR tree-optimization/81373
3623 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3624 Force SESE live-out defs to be handled even if they are
3627 2017-09-19 Jeff Law <law@redhat.com>
3629 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3630 nothing for stack adjustments with REG_STACK_CHECK.
3631 * sched-deps.c (parse_add_or_inc): Reject insns with
3632 REG_STACK_CHECK from dependency breaking.
3633 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3634 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3635 * reg-notes.def (STACK_CHECK): New note.
3637 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3638 (ix86_expand_prologue): Dump stack clash info as needed.
3639 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3641 * function.c (dump_stack_clash_frame_info): New function.
3642 * function.h (dump_stack_clash_frame_info): Prototype.
3643 (enum stack_clash_probes): New enum.
3645 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3646 flag_stack_clash_protection.
3647 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3648 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3649 (arm_frame_pointer_required): Likewise.
3650 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3651 (ia64_expand_prologue): Likewise.
3652 * config/mips/mips.c (mips_expand_prologue): Likewise.
3653 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3654 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3655 (sparc_flat_expand_prologue): Likewise.
3656 * config/spu/spu.c (spu_expand_prologue): Likewise.
3658 * explow.c: Include "params.h".
3659 (anti_adjust_stack_and_probe_stack_clash): New function.
3660 (get_stack_check_protect): Likewise.
3661 (compute_stack_clash_protection_loop_data): Likewise.
3662 (emit_stack_clash_protection_loop_start): Likewise.
3663 (emit_stack_clash_protection_loop_end): Likewise.
3664 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3665 Use anti_adjust_stack_and_probe_stack_clash.
3666 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3667 (emit_stack_clash_protection_loop_start): Likewise.
3668 (emit_stack_clash_protection_loop_end): Likewise.
3669 * rtl.h (get_stack_check_protect): Prototype.
3670 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3671 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3672 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3674 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3676 * doc/tm.texi: Rebuilt.
3677 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3678 get_stack_check_protect.
3679 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3680 * config/arm/arm.c (arm_expand_prologue): Likewise.
3681 (arm_frame_pointer_required): Likewise.
3682 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3683 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3684 * config/mips/mips.c (mips_expand_prologue): Likewise.
3685 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3686 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3687 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3688 (sparc_flat_expand_prologue): Likewise.
3690 * common.opt (-fstack-clash-protection): New option.
3691 * flag-types.h (enum stack_check_type): Note difference between
3692 -fstack-check= and -fstack-clash-protection.
3693 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3694 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3695 * toplev.c (process_options): Issue warnings/errors for cases
3696 not handled with -fstack-clash-protection.
3697 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3698 (-fstack-check): Note additional problem with -fstack-check=generic.
3699 Note that -fstack-check is primarily for Ada and refer users
3700 to -fstack-clash-protection for stack-clash-protection.
3701 Document new params for stack clash protection.
3703 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3705 * config/i386/i386.c (ix86_split_long_move): Do not handle
3706 address used for LEA in a special way.
3708 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3710 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3712 2017-09-19 Martin Sebor <msebor@redhat.com>
3715 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3716 of incompatible types.
3718 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3721 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3722 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3723 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3725 2017-09-19 Richard Biener <rguenther@suse.de>
3727 PR tree-optimization/82244
3728 * tree-vrp.c (remove_range_assertions): Do not propagate
3729 a constant to abnormals but replace the assert with a copy.
3731 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3733 PR rtl-optimization/57878
3734 PR rtl-optimization/68988
3735 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3736 avoidance test involving non_reload_pseudos. Move frequency test
3737 below the general fragmentation avoidance test.
3739 2017-09-19 Richard Biener <rguenther@suse.de>
3741 PR tree-optimization/69728
3742 * graphite-sese-to-poly.c (schedule_error): New global.
3743 (add_loop_schedule): Handle empty domain by failing the
3745 (build_original_schedule): Handle schedule_error.
3747 2017-09-19 Richard Biener <rguenther@suse.de>
3749 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3750 Do not iterate to sibling loops but only to siblings of inner
3753 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3756 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3757 that the registers are different.
3759 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3761 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3762 to processor_model and "amdfam17h" to arch_names_table.
3763 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3765 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3768 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3770 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3771 Alan Hayward <alan.hayward@arm.com>
3772 David Sherwood <david.sherwood@arm.com>
3774 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3776 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3778 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3779 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3780 vect_schedule_slp_instance.
3781 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3782 Update call to vect_slp_analyze_node_operations. Simplify return
3784 (vect_slp_analyze_bb_1): Update call accordingly.
3785 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3786 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3787 (vect_schedule_slp): Update call accordingly.
3789 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3790 Alan Hayward <alan.hayward@arm.com>
3791 David Sherwood <david.sherwood@arm.com>
3793 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3794 with types that aren't in fact scalar.
3796 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3798 * tree-vect-slp.c (vect_record_max_nunits): New function,
3800 (vect_build_slp_tree_1): ...here.
3801 (vect_build_slp_tree_2): Call it for phis too.
3803 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3805 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3806 to vect_get_vec_def_for_operand when getting the mask operand.
3808 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3809 Alan Hayward <alan.hayward@arm.com>
3810 David Sherwood <david.sherwood@arm.com>
3812 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3815 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3816 Alan Hayward <alan.hayward@arm.com>
3817 David Sherwood <david.sherwood@arm.com>
3819 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3820 calculation for vector booleans.
3822 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3823 Alan Hayward <alan.hayward@arm.com>
3824 David Sherwood <david.sherwood@arm.com>
3826 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3828 (vect_transform_stmt): ...here.
3829 (vect_analyze_stmt): Use it instead of calling
3830 vectorizable_live_operation directly.
3832 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3834 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3835 non-SIMT targets in acc vector loops.
3837 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3839 * configure.ac: Add arc and check if assembler supports gdwarf2.
3840 * configure: Regenerate.
3842 2017-09-18 Richard Biener <rguenther@suse.de>
3844 PR tree-optimization/82220
3845 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3846 epilogue niters from the min_profitable_iters compute.
3848 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3851 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3852 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3853 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3855 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3858 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3859 switching to a new text section.
3861 2017-09-18 Richard Biener <rguenther@suse.de>
3863 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3865 (build_alias_set): Reject aliases with no access function.
3867 2017-09-18 Richard Biener <rguenther@suse.de>
3869 PR tree-optimization/79622
3870 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3872 (build_cross_bb_scalars_use): Likewise.
3874 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3876 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3878 2017-09-18 Alan Modra <amodra@gmail.com>
3881 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3882 stack_pointer_rtx for count 0. Update comments. Break up
3883 large rtl expression.
3885 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3887 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3888 Increase to 20 bytes.
3889 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3890 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3891 or avx version of the stub.
3893 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3896 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3897 compute the minimum stack alignment. Also update preferred stack
3898 boundary for leaf functions.
3900 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3902 PR tree-optimization/82228
3903 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3906 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3908 * common/config/nds32/nds32-common.c
3909 (nds32_option_optimization_table): Refine formatting.
3910 (nds32_option_optimization_table): Use -fsched-pressure and
3911 -fomit-frame-pointer for specific optimization level.
3913 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3915 * config/nds32/nds32.c: Refine formatting and comments.
3916 * config/nds32/nds32.h: Likewise.
3917 * config/nds32/nds32.md: Likewise.
3918 * config/nds32/nds32-cost.c: Likewise.
3919 * config/nds32/nds32-isr.c: Likewise.
3920 * config/nds32/nds32-md-auxiliary.c: Likewise.
3921 * config/nds32/nds32-multiple.md: Likewise.
3922 * config/nds32/nds32-predicates.c: Likewise.
3924 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3925 Jakub Jelinek <jakub@redhat.com>
3927 Add support for -std=c++2a.
3928 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3930 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3932 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3935 * doc/extend.texi (Common Function Attributes): Add
3936 references to ARM, AArch64, and S/390 specific attributes.
3937 (Function Specific Option Pragmas): Add AArch64 and S/390
3938 to list of back ends that support the target pragma.
3940 2017-09-15 Nathan Sidwell <nathan@acm.org>
3942 * doc/standards.texi: Fix C++17 description. Update URLs for
3945 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3947 * common.opt (Wcast-align=strict): New warning option.
3948 * doc/invoke.texi: Document -Wcast-align=strict.
3950 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3952 * cgraph.h (cgraph_thunk_info): Add comments.
3953 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3954 assert for VIRTUAL_* arguments stricter.
3956 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3958 PR tree-optimization/71026
3959 * match.pd: Move RDIV patterns from fold-const.c
3960 * fold-const.c (distribute_real_division): Removed.
3961 (fold_binary_loc): Remove calls to distribute_real_divison.
3963 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3965 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3966 c++1z and gnu++1z as deprecated. Change other references to
3967 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3968 Change -Wc++1z-compat to -Wc++17-compat.
3969 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3970 * dwarf2out.c (highest_c_language): Handle C++17.
3971 (gen_compile_unit_die): Likewise.
3973 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3975 PR rtl-optimization/82192
3976 * combine.c (make_extraction): Don't look through non-paradoxical
3977 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3980 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3981 Alan Hayward <alan.hayward@arm.com>
3982 David Sherwood <david.sherwood@arm.com>
3984 * target.def (function_arg_offset): New hook.
3985 * targhooks.h (default_function_arg_offset): Declare.
3986 * targhooks.c (default_function_arg_offset): New function.
3987 * function.c (locate_and_pad_parm): Use
3988 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3989 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3990 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3991 * doc/tm.texi: Regenerate.
3992 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3993 * config/spu/spu.c (spu_function_arg_offset): New function.
3994 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3995 * system.h (FUNCTION_ARG_OFFSET): Poison.
3997 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3998 Alan Hayard <alan.hayward@arm.com>
3999 David Sherwood <david.sherwood@arm.com>
4001 * target.def (truly_noop_truncation): New hook.
4002 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4003 than TRULY_NOOP_TRUNCATION.
4004 * hooks.h (hook_bool_uint_uint_true): Declare.
4005 * hooks.c (hook_bool_uint_uint_true): New function.
4006 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4007 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4008 * doc/tm.texi: Regenerate.
4009 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4010 rather than TRULY_NOOP_TRUNCATION in comments.
4011 (simplify_comparison): Likewise.
4012 (record_truncated_value): Likewise.
4013 * expmed.c (extract_bit_field_1): Likewise.
4014 (extract_split_bit_field): Likewise.
4015 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4016 instead of TRULY_NOOP_TRUNCATION.
4017 * function.c (assign_parm_setup_block): Likewise.
4018 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4019 * rtlhooks.c: Include target.h.
4020 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4021 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4022 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4023 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4024 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4025 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4026 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4027 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4028 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4029 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4030 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4031 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4032 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4033 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4034 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4035 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4036 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4037 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4038 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4039 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4040 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4041 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4042 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4043 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4044 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4045 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4046 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4047 rather than TRULY_NOOP_TRUNCATION in comments.
4048 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4049 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4050 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4051 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4052 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4053 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4054 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4055 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4056 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4057 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4058 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4059 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4060 rather than TRULY_NOOP_TRUNCATION in comments.
4061 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4062 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4063 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4064 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4065 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4066 TRULY_NOOP_TRUNCATION condition.
4067 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4068 (TRULY_NOOP_TRUNCATION): Delete.
4069 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4070 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4071 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4072 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4073 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4074 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4075 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4076 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4077 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4078 rather than TRULY_NOOP_TRUNCATION in comments.
4079 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4080 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4081 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4082 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4083 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4084 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4086 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4089 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4090 (*cmp_ior): Likewise.
4091 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4092 (*ior_scc_scc_cmp): Likewise.
4093 (*and_scc_scc): Likewise.
4094 (*and_scc_scc_cmp): Likewise.
4096 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4097 Alan Hayard <alan.hayward@arm.com>
4098 David Sherwood <david.sherwood@arm.com>
4100 * target.def (can_change_mode_class): New hook.
4101 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4102 (hard_regno_nregs): Likewise.
4103 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4104 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4105 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4106 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4107 (LOAD_EXTEND_OP): Update accordingly.
4108 * doc/tm.texi: Regenerate.
4109 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4110 CANNOT_CHANGE_MODE_CLASS.
4111 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4112 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4113 * combine.c (simplify_set): Update accordingly.
4114 * emit-rtl.c (validate_subreg): Likewise.
4115 * recog.c (general_operand): Likewise.
4116 * regcprop.c (mode_change_ok): Likewise.
4117 * reload1.c (choose_reload_regs): Likewise.
4118 (inherit_piecemeal_p): Likewise.
4119 * rtlanal.c (simplify_subreg_regno): Likewise.
4120 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4121 instead of CANNOT_CHANGE_MODE_CLASS.
4122 (reload_cse_simplify_operands): Likewise.
4123 * reload.c (push_reload): Use targetm.can_change_mode_class
4124 instead of CANNOT_CHANGE_MODE_CLASS.
4125 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4126 REG_CANNOT_CHANGE_MODE_P.
4127 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4128 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4129 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4130 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4131 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4132 (arm_can_change_mode_class): New function.
4133 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4134 than CANNOT_CHANGE_MODE_CLASS in comments.
4135 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4136 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4137 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4138 (ix86_can_change_mode_class): ...this new function, inverting the
4139 sense of the return value.
4140 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4141 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4142 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4143 (ia64_can_change_mode_class): New function.
4144 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4145 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4146 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4147 (m32c_can_change_mode_class): ...this new function, inverting the
4148 sense of the return value.
4149 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4150 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4151 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4152 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4153 (mips_can_change_mode_class): ...this new function, inverting the
4154 sense of the return value.
4155 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4156 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4157 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4158 (msp430_can_change_mode_class): New function.
4159 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4160 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4161 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4162 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4163 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4164 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4165 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4166 (pa_cannot_change_mode_class): Replace with...
4167 (pa_can_change_mode_class): ...this new function, inverting the
4168 sense of the return value.
4169 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4170 than CANNOT_CHANGE_MODE_CLASS in comments.
4171 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4172 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4173 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4174 (pdp11_cannot_change_mode_class): Replace with...
4175 (pdp11_can_change_mode_class): ...this new function, inverting the
4176 sense of the return value.
4177 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4178 * config/powerpcspe/powerpcspe-protos.h
4179 (rs6000_cannot_change_mode_class_ptr): Delete.
4180 * config/powerpcspe/powerpcspe.c
4181 (rs6000_cannot_change_mode_class_ptr): Delete.
4182 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4183 (rs6000_option_override_internal): Assign to
4184 targetm.can_change_mode_class instead of
4185 rs6000_cannot_change_mode_class_ptr.
4186 (rs6000_cannot_change_mode_class): Replace with...
4187 (rs6000_can_change_mode_class): ...this new function, inverting the
4188 sense of the return value.
4189 (rs6000_debug_cannot_change_mode_class): Replace with...
4190 (rs6000_debug_can_change_mode_class): ...this new function.
4191 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4192 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4193 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4194 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4195 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4197 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4198 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4199 (rs6000_option_override_internal): Assign to
4200 targetm.can_change_mode_class instead of
4201 rs6000_cannot_change_mode_class_ptr.
4202 (rs6000_cannot_change_mode_class): Replace with...
4203 (rs6000_can_change_mode_class): ...this new function, inverting the
4204 sense of the return value.
4205 (rs6000_debug_cannot_change_mode_class): Replace with...
4206 (rs6000_debug_can_change_mode_class): ...this new function.
4207 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4208 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4209 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4210 (s390_can_change_mode_class): ...this new function, inverting the
4211 sense of the return value.
4212 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4213 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4214 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4215 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4216 (sh_cannot_change_mode_class): Replace with...
4217 (sh_can_change_mode_class): ...this new function, inverting the
4218 sense of the return value.
4219 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4220 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4221 (sparc_can_change_mode_class): New function.
4222 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4223 * config/spu/spu.c (spu_can_change_mode_class): New function.
4224 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4225 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4226 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4227 (visium_can_change_mode_class): New function.
4228 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4230 2017-09-15 Richard Biener <rguenther@suse.de>
4232 PR tree-optimization/82217
4233 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4234 but not undefined case.
4236 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4239 * postreload.c (reload_cse_simplify_operands): Skip
4240 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4242 2017-09-15 Richard Biener <rguenther@suse.de>
4244 PR tree-optimization/68823
4245 * graphite-scop-detection.c (build_alias_set): If we have a
4246 possible dependence check whether we can handle them by just
4247 looking at the DRs DR_ACCESS_FNs.
4248 (build_scops): If build_alias_set fails, fail the SCOP.
4250 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4252 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4253 to support float128 built-in functions that require the ISA 3.0
4255 (BU_FLOAT128_3_HW): Likewise.
4256 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4258 (FMAF128): Likewise.
4260 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4261 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4262 floating point instructions.
4263 (rs6000_invalid_builtin): Likewise.
4264 (rs6000_builtin_mask_names): Likewise.
4265 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4266 (RS6000_BTM_FLOAT128_HW): Likewise.
4267 (RS6000_BTM_COMMON): Likewise.
4268 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4270 * doc/extend.texi (RS/6000 built-in functions): Document the
4271 IEEE 128-bit floating point square root and fused multiply-add
4274 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4276 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4277 reg (r2) isn't in the set of registers defined in the prologue.
4279 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4280 Alan Hayward <alan.hayward@arm.com>
4281 David Sherwood <david.sherwood@arm.com>
4283 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4284 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4285 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4286 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4287 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4289 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4290 max_vectorization_factor.
4291 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4293 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4294 Alan Hayward <alan.hayward@arm.com>
4295 David Sherwood <david.sherwood@arm.com>
4297 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4298 (vect_worthwhile_without_simd_p): Declare.
4299 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4300 (vectorizable_reduction): Use it.
4301 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4302 (vectorizable_operation): Likewise.
4304 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4305 Alan Hayward <alan.hayward@arm.com>
4306 David Sherwood <david.sherwood@arm.com>
4308 * tree-vectorizer.h (vect_get_num_copies): New function.
4309 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4310 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4311 (vectorizable_induction): Likewise.
4312 (vectorizable_live_operation): Likewise.
4313 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4314 (vectorizable_bswap): Likewise.
4315 (vectorizable_call): Likewise.
4316 (vectorizable_conversion): Likewise.
4317 (vectorizable_assignment): Likewise.
4318 (vectorizable_shift): Likewise.
4319 (vectorizable_operation): Likewise.
4320 (vectorizable_store): Likewise.
4321 (vectorizable_load): Likewise.
4322 (vectorizable_condition): Likewise.
4323 (vectorizable_comparison): Likewise.
4324 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4326 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4327 Alan Hayward <alan.hayward@arm.com>
4328 David Sherwood <david.sherwood@arm.com>
4330 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4331 of vect_init_vector.
4333 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4334 Alan Hayward <alan.hayward@arm.com>
4335 David Sherwood <david.sherwood@arm.com>
4337 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4338 an inline wrapper that provides a location.
4339 (gimple_build_vector): Likewise.
4340 * gimple-fold.c (gimple_build_vector_from_val): New function.
4341 (gimple_build_vector): Likewise.
4342 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4343 functions to build the initial value. Always return a gimple value.
4344 (get_initial_defs_for_reduction): Likewise. Only compute
4346 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4347 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4348 (vectorizable_induction): Use gimple_build_vector rather than
4351 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4352 Alan Hayward <alan.hayward@arm.com>
4353 David Sherwood <david.sherwood@arm.com>
4355 * target.h (vec_perm_indices): New typedef.
4356 (auto_vec_perm_indices): Likewise.
4357 * optabs-query.h: Include target.h
4358 (can_vec_perm_p): Take a vec_perm_indices *.
4359 * optabs-query.c (can_vec_perm_p): Likewise.
4360 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4361 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4362 * tree-vect-generic.c (lower_vec_perm): Likewise.
4363 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4364 (vect_grouped_load_supported): Likewise.
4365 (vect_shift_permute_load_chain): Likewise.
4366 (vect_permute_store_chain): Use auto_vec_perm_indices.
4367 (vect_permute_load_chain): Likewise.
4368 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4369 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4370 Update uses of can_vec_perm_p.
4371 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4372 mode with a number of elements. Take a vec_perm_indices *.
4373 (vect_create_epilog_for_reduction): Update accordingly.
4374 Use auto_vec_perm_indices.
4375 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4376 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4377 (vect_transform_slp_perm_load): Likewise.
4378 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4379 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4380 (vect_gen_perm_mask_checked): Likewise.
4381 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4382 (vect_gen_perm_mask_checked): Likewise.
4383 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4384 (vectorizable_store): Likewise.
4385 (vectorizable_load): Likewise.
4386 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4387 (vectorizable_bswap): Likewise.
4389 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4390 Alan Hayward <alan.hayward@arm.com>
4391 David Sherwood <david.sherwood@arm.com>
4393 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4394 * tree.c (build_vector): Likewise.
4395 (build_vector_from_ctor): Update accordingly.
4396 (build_vector_from_val): Likewise.
4397 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4398 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4399 * tree-vect-generic.c (add_rshift): Likewise.
4400 (expand_vector_divmod): Likewise.
4401 (optimize_vector_constructor): Likewise.
4402 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4403 (vect_transform_slp_perm_load): Likewise.
4404 (vect_schedule_slp_instance): Likewise.
4405 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4406 (vectorizable_call): Likewise.
4407 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4408 * expmed.c (make_tree): Likewise.
4409 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4410 a vector passed to build_vector.
4411 (fold_convert_const): Likewise.
4412 (exact_inverse): Likewise.
4413 (fold_ternary_loc): Likewise.
4414 (fold_relational_const): Likewise.
4415 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4416 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4417 (const_unop): Likewise. Store the reduction accumulator in a
4418 variable rather than an array.
4419 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4420 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4421 the new vector, rather than constructing it after the input arrays.
4422 (native_interpret_vector): Use auto_vec<tree> when building
4423 a vector passed to build_vector. Add elements in order.
4424 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4425 auto_vec<tree> when building a vector passed to build_vector.
4426 (vect_create_epilog_for_reduction): Likewise.
4427 (vectorizable_induction): Likewise.
4428 (get_initial_def_for_reduction): Likewise. Fix indentation of
4430 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4432 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4433 passed to build_vector.
4435 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4436 Alan Hayward <alan.hayward@arm.com>
4437 David Sherwood <david.sherwood@arm.com>
4439 * tree-core.h (tree_base::u): Add an "nelts" field.
4440 (tree_vector): Use VECTOR_CST_NELTS as the length.
4441 * tree.c (tree_size): Likewise.
4442 (make_vector): Initialize VECTOR_CST_NELTS.
4443 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4444 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4445 TYPE_VECTOR_SUBPARTS.
4446 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4447 as the number of units, setting it from VECTOR_CST_NELTS.
4448 (const_vector_from_tree): Likewise.
4449 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4450 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4451 (fold_negate_expr_1): Likewise.
4452 (fold_convert_const): Likewise.
4453 (const_binop): Likewise. Differentiate the number of output and
4455 (const_unop): Likewise.
4456 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4457 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4458 in cases that did the opposite.
4460 2017-09-14 Richard Biener <rguenther@suse.de>
4462 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4465 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4467 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4469 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4472 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4473 if and where to split a bb, except for splitting before debug insn
4474 sequences followed by non-label real insn. Delete debug insns
4475 in between basic blocks.
4477 * combine.c (make_compound_operation_int): Formatting fixes.
4479 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4480 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4481 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4482 * config/sol2.h (LINK_EH_SPEC): Likewise.
4483 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4484 * config/s390/linux.h (LINK_SPEC): Likewise.
4485 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4486 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4487 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4488 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4489 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4490 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4491 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4492 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4494 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4496 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4498 (ENDFILE_LINUX_SPEC): Likewise.
4499 (LINK_EH_SPEC): Likewise.
4500 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4501 (LINK_OS_LINUX_SPEC32): Likewise.
4502 (LINK_OS_LINUX_SPEC64): Likewise.
4503 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4504 (LINK_OS_LINUX_SPEC): Likewise.
4506 2017-09-13 Martin Liska <mliska@suse.cz>
4509 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4510 CASE_HIGH is NULL_TREE.
4512 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4513 Alan Hayward <alan.hayward@arm.com>
4514 David Sherwood <david.sherwood@arm.com>
4516 * target.def (secondary_memory_needed): New hook.
4517 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4518 instead of SECONDARY_MEMORY_NEEDED.
4519 (secondary_memory_needed_mode): Likewise.
4520 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4521 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4522 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4523 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4524 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4525 * doc/tm.texi: Regenerate.
4526 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4527 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4528 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4529 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4530 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4531 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4532 mode argument first and change the reg_class arguments to reg_class_t.
4533 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4534 Make static. Update the call to inline_secondary_memory_needed.
4535 (ix86_register_move_cost): Update the call to
4536 inline_secondary_memory_needed.
4537 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4538 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4540 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4541 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4543 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4544 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4545 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4546 and match hook interface. Add comment from mips.h.
4547 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4548 * config/mmix/mmix.md (truncdfsf2): Refer to
4549 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4551 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4552 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4553 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4554 (pa_secondary_memory_needed): New function.
4555 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4556 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4557 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4558 (pdp11_secondary_memory_needed): Make static and match hook interface.
4559 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4560 * config/powerpcspe/powerpcspe-protos.h
4561 (rs6000_secondary_memory_needed_ptr): Delete.
4562 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4564 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4565 (rs6000_option_override_internal): Assign to
4566 targetm.secondary_memory_needed rather than
4567 rs6000_secondary_memory_needed_ptr.
4568 (rs6000_secondary_memory_needed): Match hook interface.
4569 (rs6000_debug_secondary_memory_needed): Likewise.
4570 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4571 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4572 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4573 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4574 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4575 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4577 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4578 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4579 (rs6000_option_override_internal): Assign to
4580 targetm.secondary_memory_needed rather than
4581 rs6000_secondary_memory_needed_ptr.
4582 (rs6000_secondary_memory_needed): Match hook interface.
4583 (rs6000_debug_secondary_memory_needed): Likewise.
4584 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4585 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4586 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4587 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4588 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4589 (sparc_secondary_memory_needed): New function.
4590 * lra-constraints.c (check_and_process_move): Refer to
4591 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4593 (curr_insn_transform): Likewise.
4594 (process_alt_operands): Use targetm.secondary_memory_needed
4595 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4596 (check_secondary_memory_needed_p): Likewise.
4597 (choose_split_class): Likewise.
4598 * reload.c: Unconditionally include code that was previously
4599 conditional on SECONDARY_MEMORY_NEEDED.
4600 (push_secondary_reload): Use targetm.secondary_memory_needed
4601 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4602 (push_reload): Likewise.
4603 * reload1.c: Unconditionally include code that was previously
4604 conditional on SECONDARY_MEMORY_NEEDED.
4605 (choose_reload_regs): Use targetm.secondary_memory_needed
4606 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4607 (gen_reload): Likewise.
4608 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4610 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4611 Alan Hayward <alan.hayward@arm.com>
4612 David Sherwood <david.sherwood@arm.com>
4614 * target.def (secondary_memory_needed_mode): New hook:
4615 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4616 * targhooks.h (default_secondary_memory_needed_mode): New function.
4617 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4618 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4619 * doc/tm.texi: Regenerate.
4620 * lra-constraints.c (check_and_process_move): Use
4621 targetm.secondary_memory_needed_mode instead of
4622 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4623 (curr_insn_transform): Likewise.
4624 * reload.c (get_secondary_mem): Likewise.
4625 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4626 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4628 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4629 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4630 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4631 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4632 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4634 * config/powerpcspe/powerpcspe-protos.h
4635 (rs6000_secondary_memory_needed_mode): Delete.
4636 * config/powerpcspe/powerpcspe.c
4637 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4638 (rs6000_secondary_memory_needed_mode): Make static.
4639 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4640 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4642 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4644 (rs6000_secondary_memory_needed_mode): Make static.
4645 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4646 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4647 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4648 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4649 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4651 (sparc_secondary_memory_needed_mode): New function.
4652 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4654 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4656 * config/aarch64/constraints.md (Umq): New constraint.
4657 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4659 (mov<mode>): Update condition.
4661 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4663 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4664 when bitposition is the same.
4666 2017-09-13 Richard Biener <rguenther@suse.de>
4668 * dwarf2out.c (output_die_symbol): Remove.
4669 (output_die): Do not output a DIEs symbol.
4671 2017-09-13 Richard Biener <rguenther@suse.de>
4674 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4675 default-def to avoid breaking iterator update with the weird
4676 interaction with cgraph_update_edges_for_call_stmt_node.
4678 2017-09-13 Richard Biener <rguenther@suse.de>
4680 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4681 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4682 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4683 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4685 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4687 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4688 Disable pc relative literal load irrespective of
4689 TARGET_FIX_ERR_A53_84341 for default.
4691 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4693 * config/sparc/sparc.c (output_return): Output the source location of
4694 the insn in the delay slot, if any.
4695 (output_sibcall): Likewise.
4697 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4700 * common.opt (-static-pie): New alias.
4701 (shared): Negate static-pie.
4702 (-no-pie): Update help text.
4704 (static-pie): New option.
4705 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4706 -static-pie support.
4707 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4708 (LINK_EH_SPEC): Likewise.
4709 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4710 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4711 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4712 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4713 (init_gcc_specs): Likewise.
4714 (init_spec): Likewise.
4715 (display_help): Update help message for -pie.
4716 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4719 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4721 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4722 (movdi_aarch64): Likewise.
4723 (movti_aarch64): Likewise.
4725 2017-09-12 Simon Wright <simon@pushface.org>
4728 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4729 calculation of the minor version, always output as 0.
4731 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4734 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4735 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4736 on it early, rather than manual conversion late. For
4737 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4738 instead of performing manual conversion.
4740 2017-09-12 Carl Love <cel@us.ibm.com>
4742 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4743 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4745 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4746 VMULOSW): Add definitions.
4747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4748 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4749 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4751 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4753 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4755 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4757 (movti_aarch64): Likewise.
4758 (movdf_aarch64): Likewise.
4759 (movtf_aarch64): Likewise.
4760 (load_pairdi): Likewise.
4761 (store_pairdi): Likewise.
4762 (load_pairdf): Likewise.
4763 (store_pairdf): Likewise.
4764 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4765 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4766 (ldr_got_small_<mode>): Likewise.
4767 (ldr_got_small_28k_<mode>): Likewise.
4768 (ldr_got_tiny): Likewise.
4769 * config/aarch64/iterators.md (ldst_sz): New.
4770 (ldpstp_sz): Likewise.
4771 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4773 (thunderx_load): Split load_8 to load_16.
4774 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4776 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4777 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4778 (xgene1_store_pair): Split store_8 to store_16.
4779 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4780 (falkor_st_0_st_sd): Split store_8 to store_16.
4782 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4784 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4785 and store1/2/3/4 to store_4/8/12/16.
4786 * config/aarch64/aarch64.md: Update for rename.
4787 * config/arm/arm.md: Likewise.: Likewise.
4788 * config/arm/arm.c: Likewise.
4789 * config/arm/thumb1.md: Likewise.
4790 * config/arm/thumb2.md: Likewise.
4791 * config/arm/vfp.md: Likewise.
4792 * config/arm/arm-generic.md: Likewise.
4793 * config/arm/arm1020e.md: Likewise.
4794 * config/arm/arm1026ejs.md: Likewise.
4795 * config/arm/arm1136jfs.md: Likewise.
4796 * config/arm/arm926ejs.md: Likewise.
4797 * config/arm/cortex-a15.md: Likewise.
4798 * config/arm/cortex-a17.md: Likewise.
4799 * config/arm/cortex-a5.md: Likewise.
4800 * config/arm/cortex-a53.md: Likewise.
4801 * config/arm/cortex-a57.md: Likewise.
4802 * config/arm/cortex-a7.md: Likewise.
4803 * config/arm/cortex-a8.md: Likewise.
4804 * config/arm/cortex-a9.md: Likewise.
4805 * config/arm/cortex-m4.md: Likewise.
4806 * config/arm/cortex-m7.md: Likewise.
4807 * config/arm/cortex-r4.md: Likewise.
4808 * config/arm/exynos-m1.md: Likewise.
4809 * config/arm/fa526.md: Likewise.
4810 * config/arm/fa606te.md: Likewise.
4811 * config/arm/fa626te.md: Likewise.
4812 * config/arm/fa726te.md: Likewise.
4813 * config/arm/fmp626.md: Likewise.
4814 * config/arm/iwmmxt.md: Likewise.
4815 * config/arm/ldmstm.md: Likewise.
4816 * config/arm/marvell-pj4.md: Likewise.
4817 * config/arm/xgene1.md: Likewise.
4818 * config/aarch64/thunderx.md: Likewise.
4819 * config/aarch64/thunderx2t99.md: Likewise.
4820 * config/aarch64/falkor.md: Likewise.
4822 2017-09-12 Martin Liska <mliska@suse.cz>
4824 * attribs.c (private_lookup_attribute): New function.
4825 * attribs.h (private_lookup_attribute): Declared here.
4826 (lookup_attribute): Called from this place.
4828 2017-09-12 Richard Biener <rguenther@suse.de>
4830 PR tree-optimization/82157
4831 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4832 stmts with side-effects.
4834 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4835 Alan Hayward <alan.hayward@arm.com>
4836 David Sherwood <david.sherwood@arm.com>
4838 * target.def (hard_regno_nregs): New hook.
4839 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4840 * targhooks.h (default_hard_regno_nregs): Declare.
4841 * targhooks.c (default_hard_regno_nregs): New function.
4842 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4843 (TARGET_HARD_REGNO_NREGS): ...this hook.
4844 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4845 (CLASS_MAX_NREGS): Likewise.
4846 * doc/tm.texi: Regenerate.
4847 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4848 instead of HARD_REGNO_NREGS.
4849 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4850 HARD_REGNO_NREGS in the comment.
4851 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4852 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4853 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4854 Return an unsigned int.
4855 (TARGET_HARD_REGNO_NREGS): Redefine.
4856 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4857 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4858 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4859 (arc_hard_regno_nregs): New function.
4860 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4861 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4862 (arm_hard_regno_nregs): New function.
4863 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4864 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4865 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4866 (TARGET_HARD_REGNO_NREGS): Redefine.
4867 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4868 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4869 (HARD_REGNO_NREGS): Delete.
4870 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4871 (cr16_hard_regno_nregs): New function.
4872 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4873 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4874 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4875 (cris_hard_regno_nregs): New function.
4876 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4877 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4878 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4879 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4880 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4881 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4882 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4883 (frv_hard_regno_nregs): Make static. Take and return an
4885 (frv_class_max_nregs): Remove outdated copy of documentation.
4886 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4887 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4888 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4889 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4890 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4891 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4892 (TARGET_HARD_REGNO_NREGS): Redefine.
4893 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4894 (CLASS_MAX_NREGS): Update comment.
4895 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4896 (ia64_hard_regno_nregs): New function.
4897 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4898 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4899 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4900 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4901 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4903 (m32c_hard_regno_nregs): Likewise. Make static.
4904 (TARGET_HARD_REGNO_NREGS): Redefine.
4905 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4906 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4907 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4908 (m68k_hard_regno_nregs): New function.
4909 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4910 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4911 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4912 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4913 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4914 Take and return an unsigned int.
4915 (TARGET_HARD_REGNO_NREGS): Redefine.
4916 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4917 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4918 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4919 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4920 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4921 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4922 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4923 (msp430_hard_regno_nregs): Make static. Take and return an
4925 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4926 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4927 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4928 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4929 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4930 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4931 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4932 (TARGET_HARD_REGNO_NREGS): Redefine.
4933 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4934 (PA_HARD_REGNO_NREGS): ...this.
4935 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4936 (PA_HARD_REGNO_NREGS): ...this.
4937 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4938 (pa_hard_regno_nregs): New function.
4939 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4940 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4941 (pdp11_hard_regno_nregs): New function.
4942 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4943 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4944 (rs6000_hard_regno_nregs_hook): New function.
4945 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4946 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4947 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4948 Take and return an unsigned int. Move earlier in file.
4949 (TARGET_HARD_REGNO_NREGS): Redefine.
4950 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4951 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4952 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4953 (rl78_hard_regno_nregs): Make static. Take and return an
4955 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4956 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4957 (rs6000_hard_regno_nregs_hook): New function.
4958 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4959 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4960 (TARGET_HARD_REGNO_NREGS): Redefine.
4961 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4962 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4963 instead of HARD_REGNO_NREGS.
4964 (s390_hard_regno_nregs): New function.
4965 (s390_hard_regno_mode_ok): Add comment from s390.h.
4966 (TARGET_HARD_REGNO_NREGS): Redefine.
4967 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4968 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4969 (sh_hard_regno_nregs): New function.
4970 (sh_pass_in_reg_p): Use it.
4971 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4972 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4973 (sparc_hard_regno_nregs): New function.
4974 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4975 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4976 (spu_function_arg_advance): Use it, supplying a valid register number.
4977 (TARGET_HARD_REGNO_NREGS): Redefine.
4978 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4979 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4980 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4981 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4982 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4983 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4984 (CLASS_MAX_NREGS): Remove copy of old documentation.
4985 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4986 (visium_hard_regno_nregs): New function.
4987 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4988 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4989 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4990 xtensa_hard_regno_nregs): New function.
4991 * system.h (HARD_REGNO_NREGS): Poison.
4993 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4995 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4996 hard_regno_nregs instead of HARD_REGNO_NREGS.
4997 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4998 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4999 (c6x_expand_epilogue): Likewise.
5000 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5001 (frv_read_iacc_argument): Likewise.
5002 * config/sh/sh.c: Include regs.h.
5003 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5004 (regs_used): Likewise.
5005 (output_stack_adjust): Likewise.
5006 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5007 * expmed.c: Include regs.h.
5008 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5009 * ree.c: Include regs.h.
5010 (combine_reaching_defs): Use hard_regno_nregs instead of
5012 (add_removable_extension): Likewise.
5014 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5016 * regs.h (hard_regno_nregs): Turn into a function.
5017 (end_hard_regno): Update accordingly.
5018 * caller-save.c (setup_save_areas): Likewise.
5019 (save_call_clobbered_regs): Likewise.
5020 (replace_reg_with_saved_mem): Likewise.
5021 (insert_restore): Likewise.
5022 (insert_save): Likewise.
5023 * combine.c (can_change_dest_mode): Likewise.
5024 (move_deaths): Likewise.
5025 (distribute_notes): Likewise.
5026 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5027 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5028 (rs6000_split_multireg_move): Likewise.
5029 (rs6000_register_move_cost): Likewise.
5030 (rs6000_memory_move_cost): Likewise.
5031 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5032 (rs6000_split_multireg_move): Likewise.
5033 (rs6000_register_move_cost): Likewise.
5034 (rs6000_memory_move_cost): Likewise.
5035 * cselib.c (cselib_reset_table): Likewise.
5036 (cselib_lookup_1): Likewise.
5037 * emit-rtl.c (set_mode_and_regno): Likewise.
5038 * function.c (aggregate_value_p): Likewise.
5039 * ira-color.c (setup_profitable_hard_regs): Likewise.
5040 (check_hard_reg_p): Likewise.
5041 (calculate_saved_nregs): Likewise.
5042 (assign_hard_reg): Likewise.
5043 (improve_allocation): Likewise.
5044 (calculate_spill_cost): Likewise.
5045 * ira-emit.c (modify_move_list): Likewise.
5046 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5047 (ira_hard_reg_in_set_p): Likewise.
5048 * ira.c (setup_reg_mode_hard_regset): Likewise.
5049 (clarify_prohibited_class_mode_regs): Likewise.
5050 (check_allocation): Likewise.
5051 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5052 (lra_setup_reg_renumber): Likewise.
5053 (setup_try_hard_regno_pseudos): Likewise.
5054 (spill_for): Likewise.
5055 (assign_hard_regno): Likewise.
5056 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5057 * lra-constraints.c (in_class_p): Likewise.
5058 (lra_constraint_offset): Likewise.
5059 (simplify_operand_subreg): Likewise.
5060 (lra_constraints): Likewise.
5061 (split_reg): Likewise.
5062 (split_if_necessary): Likewise.
5063 (invariant_p): Likewise.
5064 (inherit_in_ebb): Likewise.
5065 * lra-lives.c (process_bb_lives): Likewise.
5066 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5067 (get_hard_regs): Likewise.
5068 (do_remat): Likewise.
5069 * lra-spills.c (assign_spill_hard_regs): Likewise.
5070 * mode-switching.c (create_pre_exit): Likewise.
5071 * postreload.c (reload_combine_recognize_pattern): Likewise.
5072 * recog.c (peep2_find_free_register): Likewise.
5073 * regcprop.c (kill_value_regno): Likewise.
5074 (set_value_regno): Likewise.
5075 (copy_value): Likewise.
5076 (maybe_mode_change): Likewise.
5077 (find_oldest_value_reg): Likewise.
5078 (copyprop_hardreg_forward_1): Likewise.
5079 * regrename.c (check_new_reg_p): Likewise.
5080 (regrename_do_replace): Likewise.
5081 * reload.c (push_reload): Likewise.
5082 (combine_reloads): Likewise.
5083 (find_dummy_reload): Likewise.
5084 (operands_match_p): Likewise.
5085 (find_reloads): Likewise.
5086 (find_equiv_reg): Likewise.
5087 (reload_adjust_reg_for_mode): Likewise.
5088 * reload1.c (count_pseudo): Likewise.
5089 (count_spilled_pseudo): Likewise.
5090 (find_reg): Likewise.
5091 (clear_reload_reg_in_use): Likewise.
5092 (free_for_value_p): Likewise.
5093 (allocate_reload_reg): Likewise.
5094 (choose_reload_regs): Likewise.
5095 (reload_adjust_reg_for_temp): Likewise.
5096 (emit_reload_insns): Likewise.
5097 (delete_output_reload): Likewise.
5098 * rtlanal.c (subreg_get_info): Likewise.
5099 * sched-deps.c (sched_analyze_reg): Likewise.
5100 * sel-sched.c (init_regs_for_mode): Likewise.
5101 (mark_unavailable_hard_regs): Likewise.
5102 (choose_best_reg_1): Likewise.
5103 (verify_target_availability): Likewise.
5104 * valtrack.c (dead_debug_insert_temp): Likewise.
5105 * var-tracking.c (track_loc_p): Likewise.
5106 (emit_note_insn_var_location): Likewise.
5107 * varasm.c (make_decl_rtl): Likewise.
5108 * reginfo.c (choose_hard_reg_mode): Likewise.
5109 (init_reg_modes_target): Refer directly to
5110 this_target_regs->x_hard_regno_nregs.
5112 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5114 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5115 instead of hard_regno_nregs.
5117 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5119 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5120 end_hard_regno instead of hard_regno_nregs.
5121 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5122 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5123 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5124 * ira-color.c (improve_allocation): Likewise.
5125 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5126 * lra-lives.c (mark_regno_live): Likewise.
5127 (mark_regno_dead): Likewise.
5128 * lra-remat.c (operand_to_remat): Likewise.
5129 * lra.c (collect_non_operand_hard_regs): Likewise.
5130 * postreload.c (reload_combine_note_store): Likewise.
5131 (move2add_valid_value_p): Likewise.
5132 * reload.c (regno_clobbered_p): Likewise.
5134 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5136 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5138 * config/v850/v850.c (v850_reorg): Likewise.
5139 * reload.c (refers_to_regno_for_reload_p): Likewise.
5140 (find_equiv_reg): Likewise.
5141 * reload1.c (reload_reg_reaches_end_p): Likewise.
5143 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5145 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5147 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5148 * config/arm/arm.c (output_move_neon): Likewise.
5149 (arm_attr_length_move_neon): Likewise.
5150 (neon_split_vcombine): Likewise.
5151 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5152 (c6x_mark_reg_written): Likewise.
5153 (c6x_dwarf_register_span): Likewise.
5154 * config/i386/i386.c (ix86_save_reg): Likewise.
5155 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5156 (rws_access_reg): Likewise.
5157 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5158 * mode-switching.c (create_pre_exit): Likewise.
5159 * ree.c (combine_reaching_defs): Likewise.
5160 (add_removable_extension): Likewise.
5161 * regcprop.c (find_oldest_value_reg): Likewise.
5162 (copyprop_hardreg_forward_1): Likewise.
5163 * reload.c (reload_inner_reg_of_subreg): Likewise.
5164 (push_reload): Likewise.
5165 (combine_reloads): Likewise.
5166 (find_dummy_reload): Likewise.
5167 (reload_adjust_reg_for_mode): Likewise.
5168 * reload1.c (find_reload_regs): Likewise.
5169 (forget_old_reloads_1): Likewise.
5170 (reload_reg_free_for_value_p): Likewise.
5171 (reload_adjust_reg_for_temp): Likewise.
5172 (emit_reload_insns): Likewise.
5173 (delete_output_reload): Likewise.
5174 * sel-sched.c (choose_best_reg_1): Likewise.
5175 (choose_best_pseudo_reg): Likewise.
5177 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5178 Alan Hayward <alan.hayward@arm.com>
5179 David Sherwood <david.sherwood@arm.com>
5181 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5182 * target.def (slow_unaligned_access): New hook.
5183 * targhooks.h (default_slow_unaligned_access): Declare.
5184 * targhooks.c (default_slow_unaligned_access): New function.
5185 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5186 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5187 * doc/tm.texi: Regenerate.
5188 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5189 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5190 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5192 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5193 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5195 (rs6000_slow_unaligned_access): New function.
5196 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5197 (expand_block_compare): Likewise.
5198 (expand_strn_compare): Likewise.
5199 (rs6000_rtx_costs): Likewise.
5200 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5201 (riscv_slow_unaligned_access): Likewise.
5202 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5203 (riscv_slow_unaligned_access_p): ...this and make static.
5204 (riscv_option_override): Update accordingly.
5205 (riscv_slow_unaligned_access): New function.
5206 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5207 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5208 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5209 (rs6000_slow_unaligned_access): New function.
5210 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5211 (rs6000_rtx_costs): Likewise.
5212 * config/rs6000/rs6000-string.c (expand_block_compare)
5213 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5214 of SLOW_UNALIGNED_ACCESS.
5215 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5216 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5217 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5218 of SLOW_UNALIGNED_ACCESS.
5219 * expmed.c (simple_mem_bitfield_p): Likewise.
5220 * expr.c (alignment_for_piecewise_move): Likewise.
5221 (emit_group_load_1): Likewise.
5222 (emit_group_store): Likewise.
5223 (copy_blkmode_from_reg): Likewise.
5224 (emit_push_insn): Likewise.
5225 (expand_assignment): Likewise.
5226 (store_field): Likewise.
5227 (expand_expr_real_1): Likewise.
5228 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5229 * lra-constraints.c (simplify_operand_subreg): Likewise.
5230 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5231 * gimple-ssa-store-merging.c: Likewise in block comment at start
5233 * tree-ssa-strlen.c: Include target.h.
5234 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5235 of SLOW_UNALIGNED_ACCESS.
5236 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5238 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5240 PR rtl-optimization/82185
5241 * expmed.c (emit_store_flag_int): Only test tem if it has been
5244 2017-09-12 Richard Biener <rguenther@suse.de>
5247 * match.pd ((FTYPE) N CMP CST): Fix typo.
5249 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5251 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5253 (mips_near_type_p): Add 'short_call' attribute as a synonym
5255 * doc/extend.texi (short_call): Document new function attribute.
5257 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5260 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5261 assertion check that in the condition.
5262 (get_atomic_generic_size): Likewise. Before testing if parameter
5263 has pointer type, if it has array type, call for C++
5264 default_conversion to perform array-to-pointer conversion.
5266 2017-09-12 Richard Biener <rguenther@suse.de>
5268 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5269 for operations we cannot scalarize.
5271 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5273 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5274 vectors heap vectors. Clean up comments.
5275 Make visited_bbs a reference.
5276 (profitable_jump_thread_path): Make GC
5277 vectors heap vectors. Clean up comments.
5279 (convert_and_register_jump_thread_path): Make GC vectors heap
5281 (check_subpath_and_update_thread_path): Same. Clean up comments.
5282 Make visited_bbs a reference.
5283 (handle_phi): Abstract common code to to
5284 register_jump_thread_path_if_profitable.
5285 Rename VAR_BB to DEF_BB.
5287 Make GC vectors heap vectors.
5288 Make visited_bbs a reference.
5289 (handle_assignment): Same.
5290 (register_jump_thread_path_if_profitable): New.
5291 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5293 Make GC vectors heap vectors. Clean up comments.
5294 Make visited_bbs a reference.
5295 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5296 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5299 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5302 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5303 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5305 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5307 Revert r251800 and r251799.
5309 2017-09-11 Martin Jambor <mjambor@suse.cz>
5312 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5313 arguments in advance.
5314 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5315 use it to find predecessor edges.
5316 (naive_outof_ssa): Collect vector of predecessors.
5318 2017-09-08 Jason Merrill <jason@redhat.com>
5320 PR c++/70029 - ICE with ref-qualifier and -flto
5321 * langhooks.h (struct lang_hooks_for_types): Add
5322 copy_lang_qualifiers.
5323 * attribs.c (build_type_attribute_qual_variant): Use it.
5324 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5326 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5327 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5329 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5332 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5333 (*mulsi3_sp64): New instruction.
5334 (mulsi3): New expander.
5336 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5338 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5340 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5342 * sancov.c: Include memmodel.h.
5344 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5347 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5350 2017-09-07 Carl Love <cel@us.ibm.com>
5352 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5353 the sldi instruction.
5355 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5357 * sancov.c: Include tm_p.h.
5359 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5362 * output.h (switch_to_other_text_partition): New declaration.
5363 * varasm.c (switch_to_other_text_partition): New function.
5364 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5365 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5366 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5367 to the other text partition before emitting LCL label and switch back
5368 after emitting the word after it.
5370 2017-09-07 Richard Biener <rguenther@suse.de>
5372 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5373 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5374 critical edges here, after loop init.
5375 (pass_data_pre): Remove PROP_no_crit_edges flags.
5376 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5377 for valueization of call args to avoid leaking VN_TOP.
5378 (visit_use): Assert we do not visit default defs.
5379 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5380 Use error_mark_node to more easily detect leaking VN_TOP.
5381 All default-defs are varying, not VN_TOP. Mark them visited.
5382 (run_scc_vn): Make code match comment.
5384 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5386 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5387 OPTION_MASK_FLOAT128_KEYWORD.
5388 (POWERPC_MASKS): Likewise.
5389 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5390 support for the -mfloat128-type option, and make -mfloat128
5391 default on PowerPC Linux systems. Define or undefine
5392 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5393 Define __float128 to be __ieee128 if IEEE 128-bit support is
5394 enabled, or undefine it.
5395 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5396 Delete defining __FLOAT128_TYPE__.
5397 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5398 -mfloat128-type option and make -mfloat128 default on PowerPC
5400 (TARGET_FLOAT128_TYPE): Likewise.
5401 (-mfloat128-type): Likewise.
5402 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5403 Delete the -mfloat128-type option and make -mfloat128 default on
5404 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5405 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5406 128-bit floating point is enabled. Change tests from using
5407 -mfloat128-type to -mfloat128.
5408 (rs6000_mangle_type): Use the correct mangling for the __float128
5409 type even if normal long double is restricted to 64-bits.
5410 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5412 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5413 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5414 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5416 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5417 documentation for -mfloat128.
5419 2017-09-06 Olivier Hainque <hainque@adacore.com>
5421 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5423 2017-09-06 Wish Wu <wishwu007@gmail.com>
5424 Jakub Jelinek <jakub@redhat.com>
5426 * asan.c (initialize_sanitizer_builtins): Add
5427 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5428 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5429 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5430 BT_FN_VOID_UINT64_PTR variables.
5431 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5432 (BT_FN_VOID_UINT16_UINT16): Likewise.
5433 (BT_FN_VOID_UINT32_UINT32): Likewise.
5434 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5435 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5436 (BT_FN_VOID_UINT64_PTR): Likewise.
5437 * common.opt (flag_sanitize_coverage): New variable.
5438 (fsanitize-coverage=trace-pc): Remove.
5439 (fsanitize-coverage=): Add.
5440 * flag-types.h (enum sanitize_coverage_code): New enum.
5441 * fold-const.c (fold_range_test): Disable non-short-circuit
5442 optimization if flag_sanitize_coverage.
5443 (fold_truth_andor): Likewise.
5444 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5445 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5446 (coverage_sanitizer_opts): New array.
5447 (get_closest_sanitizer_option): Add OPTS argument, handle also
5448 OPT_fsanitize_coverage_.
5449 (parse_sanitizer_options): Adjusted to also handle
5450 OPT_fsanitize_coverage_.
5451 (common_handle_option): Add OPT_fsanitize_coverage_.
5452 * sancov.c (instrument_comparison, instrument_switch): New function.
5453 (sancov_pass): Add trace-cmp support.
5454 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5455 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5456 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5457 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5458 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5459 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5460 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5461 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5462 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5463 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5465 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5467 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5468 error. Only quit immediately if parsing is complete.
5469 (BEGIN): Initialize fatal_err and parse_done.
5470 (begin fpu, end fpu): Check number of arguments.
5471 (begin arch, end arch): Likewise.
5472 (begin cpu, end cpu): Likewise.
5473 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5474 (optalias): Likewise.
5476 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5478 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5479 * config/arm/arm-isa.h: Delete. Move definitions to ...
5480 * arm-cpus.in: ... here. Use new feature and fgroup values.
5481 * config/arm/arm.c (arm_option_override): Use lower case for feature
5483 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5484 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5485 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5487 (print_isa_bits_for): New function.
5488 (gen_isa): New function.
5489 (gen_comm_data): Use print_isa_bits_for.
5490 (define feature): New keyword.
5491 (define fgroup): New keyword.
5492 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5493 (arm-isa.h): Add rule to generate file.
5494 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5495 case for feature bit names.
5497 2017-09-06 Richard Biener <rguenther@suse.de>
5499 * tree-ssa-pre.c (NECESSARY): Remove.
5500 (create_expression_by_pieces): Do not touch pass-local flags.
5501 (insert_into_preds_of_block): Likewise.
5502 (do_pre_regular_insertion): Likewise.
5503 (eliminate_insert): Likewise.
5504 (eliminate_dom_walker::before_dom_children): Likewise.
5505 (fini_eliminate): Do not look at inserted_exprs.
5506 (mark_operand_necessary): Remove.
5507 (remove_dead_inserted_code): Replace with simple work-list
5508 algorithm based on inserted_exprs and SSA uses.
5509 (pass_pre::execute): Re-order fini_eliminate and
5510 remove_dead_inserted_code.
5512 2017-09-06 Olivier Hainque <hainque@adacore.com>
5514 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5515 for VxWorks 7. Adjust surrounding comments.
5517 2017-09-06 Richard Biener <rguenther@suse.de>
5519 * gimple-ssa-strength-reduction.c
5520 (find_candidates_dom_walker::before_dom_children): Also allow
5523 2017-09-06 Richard Biener <rguenther@suse.de>
5525 PR tree-optimization/82108
5526 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5527 for gap in the non-permutation SLP case.
5529 2017-09-06 Martin Jambor <mjambor@suse.cz>
5531 PR tree-optimization/82078
5532 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5533 add_access_to_work_queue...
5534 (build_accesses_from_assign): ...here.
5535 (propagate_all_subaccesses): Make sure racc is the group
5536 representative, if there is one.
5538 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5541 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5544 2017-09-06 Richard Biener <rguenther@suse.de>
5546 * gimple-ssa-strength-reduction.c
5547 (find_candidates_dom_walker::before_doom_children): Use a
5548 type and not a mode check.
5550 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5553 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5555 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5557 2017-09-05 Jeff Law <law@redhat.com>
5559 PR tree-optimization/64910
5560 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5561 cases where we have 3 or more operands.
5563 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5566 * omp-low.c (lower_omp_for): Recompute tree invariant if
5567 gimple_omp_for_initial/final is ADDR_EXPR.
5570 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5571 into gimple val before gimplification fo the COND_EXPR.
5573 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5575 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5576 REGION_COPY argument.
5577 (thread_through_all_blocks): Remove unused argument to
5578 duplicate_thread_path.
5580 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5581 Alan Hayward <alan.hayward@arm.com>
5582 David Sherwood <david.sherwood@arm.com>
5584 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5585 Take a scalar_mode rather than a machine_mode.
5586 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5587 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5588 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5589 (aarch64_gen_adjusted_ldpstp): Likewise.
5590 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5592 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5593 Alan Hayward <alan.hayward@arm.com>
5594 David Sherwood <david.sherwood@arm.com>
5596 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5597 Take a scalar_int_mode instead of a machine_mode.
5598 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5599 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5600 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5601 (aarch64_simd_attr_length_rglist): Delete.
5602 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5603 a scalar_int_mode instead of a machine_mode.
5604 (aarch64_add_offset): Likewise.
5605 (aarch64_internal_mov_immediate): Likewise
5606 (aarch64_add_constant_internal): Likewise.
5607 (aarch64_add_constant): Likewise.
5608 (aarch64_movw_imm): Likewise.
5609 (aarch64_rtx_arith_op_extract_p): Likewise.
5610 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5611 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5612 Remove assert that the mode isn't a vector.
5613 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5614 (aarch64_expand_mov_immediate): Update calls after above changes.
5615 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5616 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5617 (aarch64_move_imm): Likewise.
5618 (aarch64_can_const_movi_rtx_p): Likewise.
5619 (aarch64_strip_extend): Likewise.
5620 (aarch64_extr_rtx_p): Likewise.
5621 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5622 a CONST_INT when the mode parameter is VOIDmode.
5623 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5625 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5627 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5628 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5629 (bitwise_type_for_mode): Update accordingly.
5631 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5633 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5634 * stor-layout.c (mode_for_size_tree): Likewise.
5635 (mode_for_array): Update accordingly.
5636 (layout_decl): Likewise.
5637 (compute_record_mode): Likewise. Only set the mode once.
5639 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5641 * target.def (get_mask_mode): Change return type to opt_mode.
5643 * doc/tm.texi: Regenerate.
5644 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5645 * targhooks.c (default_get_mask_mode): Likewise.
5646 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5647 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5648 targetm.get_mask_mode.
5649 * tree.c (build_truth_vector_type): Likewise.
5651 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5653 * machmode.h (mode_for_vector): Return an opt_mode.
5654 * stor-layout.c (mode_for_vector): Likewise.
5655 (mode_for_int_vector): Update accordingly.
5656 (layout_type): Likewise.
5657 * config/i386/i386.c (emit_memmov): Likewise.
5658 (ix86_expand_set_or_movmem): Likewise.
5659 (ix86_expand_vector_init): Likewise.
5660 (ix86_get_mask_mode): Likewise.
5661 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5663 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5664 * expmed.c (extract_bit_field_1): Likewise.
5665 * expr.c (expand_expr_real_2): Likewise.
5666 * optabs-query.c (can_vec_perm_p): Likewise.
5667 (can_vec_mask_load_store_p): Likewise.
5668 * optabs.c (expand_vec_perm): Likewise.
5669 * targhooks.c (default_get_mask_mode): Likewise.
5670 * tree-vect-stmts.c (vectorizable_store): Likewise.
5671 (vectorizable_load): Likewise.
5672 (get_vectype_for_scalar_type_and_size): Likewise.
5674 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5676 * machmode.h (mode_for_int_vector): New function.
5677 * stor-layout.c (mode_for_int_vector): Likewise.
5678 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5679 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5680 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5681 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5682 (s390_expand_vcond): Likewise.
5684 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5686 * machmode.h (opt_machine_mode): New type.
5687 (opt_mode<T>): Allow construction from anything that can be
5689 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5690 (mode_for_size): Return an opt_machine_mode.
5691 * stor-layout.c (mode_for_size): Likewise.
5692 (mode_for_size_tree): Update call accordingly.
5693 (bitwise_mode_for_mode): Likewise.
5694 (make_fract_type): Likewise.
5695 (make_accum_type): Likewise.
5696 * caller-save.c (replace_reg_with_saved_mem): Update call
5698 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5699 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5700 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5701 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5702 * expmed.c (extract_bit_field_1): Likewise.
5703 * reload.c (get_secondary_mem): Likewise.
5704 * varasm.c (assemble_integer): Likewise.
5705 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5708 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5710 * machmode.h (decimal_float_mode_for_size): New function.
5711 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5712 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5713 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5714 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5715 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5716 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5718 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5720 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5721 (get_builtin_sync_mode): Likewise.
5722 (expand_ifn_atomic_compare_exchange): Likewise.
5723 (expand_builtin_atomic_clear): Likewise.
5724 (expand_builtin_atomic_test_and_set): Likewise.
5725 (fold_builtin_atomic_always_lock_free): Likewise.
5726 * calls.c (compute_argument_addresses): Likewise.
5727 (emit_library_call_value_1): Likewise.
5728 (store_one_arg): Likewise.
5729 * combine.c (combine_instructions): Likewise.
5730 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5731 * config/arm/arm.c (arm_function_value): Likewise.
5732 (aapcs_allocate_return_reg): Likewise.
5733 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5734 * config/i386/i386.c (construct_container): Likewise.
5735 (ix86_gimplify_va_arg): Likewise.
5736 (ix86_expand_sse_cmp): Likewise.
5737 (emit_memmov): Likewise.
5738 (emit_memset): Likewise.
5739 (expand_small_movmem_or_setmem): Likewise.
5740 (ix86_expand_pextr): Likewise.
5741 (ix86_expand_pinsr): Likewise.
5742 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5743 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5745 * config/mips/mips.c (mips_function_value_1) Likewise.
5746 (mips_block_move_straight): Likewise.
5747 (mips_expand_ins_as_unaligned_store): Likewise.
5748 * config/powerpcspe/powerpcspe.c
5749 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5750 (rs6000_darwin64_record_arg_flush): Likewise.
5751 * config/rs6000/rs6000.c
5752 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5753 (rs6000_darwin64_record_arg_flush): Likewise.
5754 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5755 (sparc_function_value_1): Likewise.
5756 * config/spu/spu.c (adjust_operand): Likewise.
5757 (spu_emit_branch_or_set): Likewise.
5758 (arith_immediate_p): Likewise.
5759 * emit-rtl.c (gen_lowpart_common): Likewise.
5760 * expr.c (expand_expr_real_1): Likewise.
5761 * function.c (assign_parm_setup_block): Likewise.
5762 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5763 * reload1.c (alter_reg): Likewise.
5764 * stor-layout.c (mode_for_vector): Likewise.
5765 (layout_type): Likewise.
5767 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5769 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5770 (spu_convert_move): Likewise.
5771 * lower-subreg.c (resolve_simple_move): Likewise.
5773 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5776 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5777 define_insn to a define_expand.
5778 (altivec_vsum2sws_direct): New define_insn.
5779 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5781 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5783 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5786 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5791 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5792 realign stack if stack alignment needed is less than incoming
5795 2017-09-05 Marek Polacek <polacek@redhat.com>
5798 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5801 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5803 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5805 2017-09-05 Richard Biener <rguenther@suse.de>
5807 PR tree-optimization/82084
5808 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5810 2017-09-05 Richard Biener <rguenther@suse.de>
5812 PR tree-optimization/82102
5813 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5815 2017-09-05 Martin Liska <mliska@suse.cz>
5817 PR tree-optimization/82032
5818 * tree-cfg.c (generate_range_test): New function.
5819 * tree-cfg.h (generate_range_test): Declared here.
5820 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5821 (cleanup_control_expr_graph): Use it.
5822 * tree-switch-conversion.c (try_switch_expansion): Remove
5824 (emit_case_nodes): Use generate_range_test.
5826 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5829 * config/i386/i386.md (*<btsc><mode>_mask): Add
5830 TARGET_USE_BT to insn constraint.
5831 (*btr<mode>_mask): Ditto.
5833 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5835 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5836 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5838 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5841 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5842 TARGET_NEON and TARGET_IWMMXT.
5843 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5844 TARGET_NEON and TARGET_IWMMXT.
5845 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5847 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5849 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5850 (ix86_rewrite_tls_address): Ditto.
5851 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5852 (ix86_rewrite_tls_address_1): Ditto.
5853 (ix86_rewrite_tls_address): Ditto.
5854 * config/i386/predicates.md (tls_address_pattern): New predicate.
5855 * config/i386/i386.md (TLS address splitter): New splitter.
5857 2017-09-04 Richard Biener <rguenther@suse.de>
5859 PR tree-optimization/82084
5860 * fold-const.h (can_native_encode_string_p): Declare.
5861 * fold-const.c (can_native_encode_string_p): Factor out from ...
5862 (native_encode_string): ... here.
5863 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5864 vectorizing stores from constants we later cannot handle.
5866 2017-09-04 Marek Polacek <polacek@redhat.com>
5869 * doc/invoke.texi: Update -Wtautological-compare documentation.
5871 2017-09-04 Jeff Law <law@redhat.com>
5873 PR tree-optimization/64910
5874 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5875 swap the first and last operand if the last is a constant.
5877 2017-09-04 Marek Polacek <polacek@redhat.com>
5880 * convert.c (do_narrow): When sanitizing signed integer overflows,
5881 bail out for signed types.
5882 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5884 2017-09-04 Richard Biener <rguenther@suse.de>
5886 PR tree-optimization/82060
5887 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5888 Move devirtualization after stmt folding and before EH/AB/noreturn
5889 cleanup to get the stmt refs canonicalized. Use a bool instead
5890 of gimple_modified_p since that doesn't work for NOPs. Schedule
5891 NOPs generated by folding for removal.
5893 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5894 Alan Hayward <alan.hayward@arm.com>
5895 David Sherwood <david.sherwood@arm.com>
5897 * coretypes.h (pad_direction): New enum.
5898 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5899 (FUNCTION_ARG_PADDING): Likewise.
5900 * target.def (function_arg_padding): New hook.
5901 * targhooks.h (default_function_arg_padding): Declare.
5902 * targhooks.c (default_function_arg_padding): New function.
5903 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5904 (TARGET_FUNCTION_ARG_PADDING): ...this.
5905 * doc/tm.texi: Regenerate.
5906 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5907 instead of direction.
5908 (compute_argument_addresses): Likewise.
5909 (load_register_parameters): Likewise.
5910 (emit_library_call_value_1): Likewise.
5911 (store_one_arg): Use targetm.calls.function_arg_padding instead
5912 of FUNCTION_ARG_PADDING.
5913 (must_pass_in_stack_var_size_or_pad): Likewise.
5914 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5915 (emit_group_store): Likewise.
5916 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5917 instead of FUNCTION_ARG_PADDING.
5918 (emit_push_insn): Likewise, and propagate enum change throughout
5920 * function.h (direction): Delete.
5921 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5923 * function.c (assign_parm_find_stack_rtl): Likewise.
5924 (assign_parm_setup_block_p): Likewise.
5925 (assign_parm_setup_block): Likewise.
5926 (gimplify_parameters): Likewise.
5927 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5928 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5930 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5931 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5932 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5933 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5934 (aarch64_function_arg_padding): ...this new function.
5935 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5936 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5937 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5938 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5939 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5940 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5941 (arm_pad_arg_upward): Replace with...
5942 (arm_function_arg_padding): ...this new function.
5943 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5945 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5946 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5947 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5948 (ia64_hpux_function_arg_padding): Replace with...
5949 (ia64_function_arg_padding): ...this new function. Use pad_direction
5950 instead of direction. Check for TARGET_HPUX.
5951 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5952 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5953 (iq2000_function_arg_padding): New function.
5954 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5955 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5956 (mips_function_arg_padding): ...this new function.
5957 (mips_pad_reg_upward): Update accordingly.
5958 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5959 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5960 targetm.calls.function_arg_padding.
5961 (FUNCTION_ARG_PADDING): Delete.
5962 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5963 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5964 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5965 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5966 (nios2_block_reg_padding): Return pad_direction instead of direction.
5967 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5968 instead of direction.
5969 (nios2_function_arg_padding): Likewise. Make static.
5970 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5971 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5972 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5973 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5974 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5975 (pa_function_arg_padding): Make static. Return pad_direction instead
5977 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5978 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5979 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5980 instead of direction. Use targetm.calls.function_arg_padding.
5981 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5982 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5983 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5984 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5985 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5987 (function_arg_padding): Rename to...
5988 (rs6000_function_arg_padding): ...this. Make static. Return
5989 pad_direction instead of direction.
5990 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5991 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5992 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5993 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5994 instead of direction. Use targetm.calls.function_arg_padding.
5995 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5996 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5997 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5998 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5999 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6000 (function_arg_padding): Rename to...
6001 (rs6000_function_arg_padding): ...this. Make static. Return
6002 pad_direction instead of direction.
6003 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6004 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6005 * config/s390/s390.c (s390_function_arg_padding): New function.
6006 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6007 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6008 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6009 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6010 (function_arg_padding): Rename to...
6011 (sparc_function_arg_padding): ...this. Make static. Return
6012 pad_direction instead of direction.
6013 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6014 * config/spu/spu.c (spu_function_arg_padding): New function.
6015 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6016 * system.h (FUNCTION_ARG_PADDING): Poison.
6018 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6019 Alan Hayward <alan.hayward@arm.com>
6020 David Sherwood <david.sherwood@arm.com>
6022 * target.def (modes_tieable_p): New hook.
6023 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6024 (TARGET_MODES_TIEABLE_P): ...this.
6025 * doc/tm.texi.in: Regenerate.
6026 * hooks.h (hook_bool_mode_mode_true): Declare.
6027 * hooks.c (hook_bool_mode_mode_true): New function.
6028 * combine.c (subst): Use targetm.modes_tieable_p instead of
6030 * dse.c (find_shift_sequence): Likewise.
6031 * expmed.c (extract_low_bits): Likewise.
6032 * lower-subreg.c: Include target.h.
6033 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6035 * rtlanal.c (rtx_cost): Likewise.
6036 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6037 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6038 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6039 (TARGET_MODES_TIEABLE_P): Redefine.
6040 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6041 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6042 (TARGET_MODES_TIEABLE_P): Redefine.
6043 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6044 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6045 (arc_modes_tieable_p): New function.
6046 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6047 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6048 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6049 (arm_modes_tieable_p): Make static.
6050 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6051 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6052 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6053 (TARGET_MODES_TIEABLE_P): Redefine.
6054 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6055 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6056 (TARGET_MODES_TIEABLE_P): Redefine.
6057 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6058 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6059 (cr16_modes_tieable_p): New function.
6060 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6061 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6062 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6063 (TRULY_NOOP_TRUNCATION): Update comment.
6064 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6065 (TRULY_NOOP_TRUNCATION): Update comment.
6066 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6067 (frv_modes_tieable_p): New function.
6068 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6069 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6070 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6071 (TARGET_MODES_TIEABLE_P): Redefine.
6072 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6073 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6074 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6075 (TARGET_MODES_TIEABLE_P): Redefine.
6076 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6077 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6078 (ia64_modes_tieable_p): New function.
6079 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6080 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6081 (iq2000_modes_tieable_p): New function.
6082 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6083 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6084 (lm32_modes_tieable_p): New function.
6085 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6086 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6087 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6088 (TARGET_MODES_TIEABLE_P): Redefine.
6089 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6090 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6091 (m32r_modes_tieable_p): New function.
6092 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6093 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6094 (m68k_modes_tieable_p): New function.
6095 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6096 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6097 (mcore_modes_tieable_p): New function.
6098 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6099 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6101 (TARGET_MODES_TIEABLE_P): Redefine.
6102 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6103 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6104 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6105 (TARGET_MODES_TIEABLE_P): Redefine.
6106 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6107 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6108 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6109 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6110 (mn10300_modes_tieable_p): ...this and make static.
6111 (TARGET_MODES_TIEABLE_P): Redefine.
6112 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6113 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6114 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6115 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6116 (msp430_modes_tieable_p): Make static.
6117 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6118 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6119 (TARGET_MODES_TIEABLE_P): Redefine.
6120 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6121 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6122 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6123 (TARGET_MODES_TIEABLE_P): Redefine.
6124 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6125 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6126 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6127 (TARGET_MODES_TIEABLE_P): Redefine.
6128 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6129 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6130 (pdp11_modes_tieable_p): New function.
6131 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6132 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6133 (rs6000_modes_tieable_p): New function.
6134 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6135 * config/powerpcspe/powerpcspe.md: Update comment.
6136 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6137 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6138 (TARGET_MODES_TIEABLE_P): Redefine.
6139 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6140 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6141 (rl78_modes_tieable_p): New function.
6142 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6143 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6144 (rs6000_modes_tieable_p): New function.
6145 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6146 * config/rs6000/rs6000.md: Update comment.
6147 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6148 * config/rx/rx.c (rx_modes_tieable_p): New function.
6149 (TARGET_MODES_TIEABLE_P): Redefine.
6150 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6151 * config/s390/s390.c (s390_modes_tieable_p): New function.
6152 (TARGET_MODES_TIEABLE_P): Redefine.
6153 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6154 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6155 (sh_modes_tieable_p): New function.
6156 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6157 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6158 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6159 (sparc_modes_tieable_p): Make static.
6160 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6161 * config/spu/spu.c (spu_modes_tieable_p): New function.
6162 (TARGET_MODES_TIEABLE_P): Redefine.
6163 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6164 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6165 (TARGET_MODES_TIEABLE_P): Redefine.
6166 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6167 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6168 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6169 * config/v850/v850.c (v850_modes_tieable_p): New function.
6170 (TARGET_MODES_TIEABLE_P): Redefine.
6171 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6172 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6173 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6174 (visium_modes_tieable_p): New function.
6175 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6176 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6177 (xtensa_modes_tieable_p): New function.
6178 * system.h (MODES_TIEABLE_P): Poison.
6180 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6181 Alan Hayward <alan.hayward@arm.com>
6182 David Sherwood <david.sherwood@arm.com>
6184 * target.def (hard_regno_mode_ok): New hook.
6185 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6186 (TARGET_HARD_REGNO_MODE_OK): ...this.
6187 * doc/tm.texi.in: Regenerate.
6188 * hooks.h (hook_bool_uint_mode_true): Declare.
6189 * hooks.c (hook_bool_uint_mode_true): New function.
6190 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6192 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6193 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6194 instead of HARD_REGNO_MODE_OK.
6195 * caller-save.c: Include target.h.
6196 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6198 * combine.c (can_combine_p): Likewise.
6199 (combinable_i3pat): Likewise.
6200 (can_change_dest_mode): Likewise.
6201 * expr.c (init_expr_target): Likewise.
6202 (convert_move): Likewise.
6203 (convert_modes): Likewise.
6204 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6205 (setup_prohibited_mode_move_regs): Likewise.
6206 * ira.h (target_ira): Likewise.
6207 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6208 * lra-constraints.c (process_alt_operands): Likewise.
6209 (split_reg): Likewise.
6210 * recog.c (peep2_find_free_register): Likewise.
6211 * ree.c (combine_reaching_defs): Likewise.
6212 * regcprop.c (maybe_mode_change): Likewise.
6213 * reginfo.c (init_reg_sets_1): Likewise.
6214 (choose_hard_reg_mode): Likewise.
6215 (simplifiable_subregs): Likewise.
6216 * regrename.c (check_new_reg_p): Likewise.
6217 * reload.c (find_valid_class): Likewise.
6218 (find_valid_class_1): Likewise.
6219 (reload_inner_reg_of_subreg): Likewise.
6220 (push_reload): Likewise.
6221 (combine_reloads): Likewise.
6222 (find_dummy_reload): Likewise.
6223 (find_reloads): Likewise.
6224 * reload1.c (find_reg): Likewise.
6225 (set_reload_reg): Likewise.
6226 (allocate_reload_reg): Likewise.
6227 (choose_reload_regs): Likewise.
6228 (reload_adjust_reg_for_temp): Likewise.
6229 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6230 (simplify_subreg_regno): Likewise.
6231 * sel-sched.c (init_regs_for_mode): Likewise.
6232 * varasm.c (make_decl_rtl): Likewise.
6233 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6234 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6236 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6237 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6238 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6239 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6240 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6242 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6243 (arc_mode_class): Delete.
6244 (HARD_REGNO_MODE_OK): Delete.
6245 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6246 (arc_hard_regno_mode_ok): Rename old array to...
6247 (arc_hard_regno_mode_ok_modes): ...this.
6248 (arc_conditional_register_usage): Update accordingly.
6249 (arc_mode_class): Make static.
6250 (arc_hard_regno_mode_ok): New function.
6251 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6252 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6253 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6254 (arm_hard_regno_mode_ok): Make static.
6255 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6257 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6258 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6259 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6261 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6262 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6263 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6264 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6265 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6266 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6267 * config/bfin/predicates.md (valid_reg_operand): Use
6268 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6269 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6270 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6271 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6272 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6273 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6274 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6275 (cr16_hard_regno_mode_ok): Make static and return a bool.
6276 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6277 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6278 (cris_hard_regno_mode_ok): New function.
6279 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6280 (epiphany_mode_class): Delete.
6281 (HARD_REGNO_MODE_OK): Delete.
6282 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6283 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6284 (hard_regno_mode_ok): Rename to...
6285 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6286 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6287 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6289 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6290 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6291 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6292 (frv_hard_regno_mode_ok): Make static and return a bool.
6293 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6295 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6296 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6297 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6298 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6300 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6301 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6302 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6303 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6305 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6306 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6307 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6308 (ia64_hard_regno_mode_ok): New function.
6309 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6310 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6311 (iq2000_hard_regno_mode_ok): New function.
6312 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6313 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6314 (lm32_hard_regno_mode_ok): New function.
6315 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6316 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6317 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6318 instead of HARD_REGNO_MODE_OK.
6319 (m32c_hard_regno_ok): Rename to...
6320 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6321 (m32c_cannot_change_mode_class): Update accordingly.
6322 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6323 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6324 (m32r_mode_class): Delete.
6325 (HARD_REGNO_MODE_OK): Delete.
6326 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6327 (m32r_hard_regno_mode_ok): Rename to...
6328 (m32r_hard_regno_modes): ...this.
6329 (m32r_mode_class): Make static.
6330 (m32r_hard_regno_mode_ok): New function.
6331 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6332 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6333 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6334 (m68k_hard_regno_mode_ok): Make static.
6335 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6336 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6337 (mcore_hard_regno_mode_ok): New function.
6338 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6339 (HARD_REGNO_MODE_OK): Delete.
6340 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6342 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6343 (microblaze_hard_regno_mode_ok): New function.
6344 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6345 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6346 (mips_hard_regno_mode_ok): Delete.
6347 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6348 (mips_hard_regno_mode_ok_p): ...this and make static.
6349 (mips_hard_regno_mode_ok_p): Rename to...
6350 (mips_hard_regno_mode_ok_uncached): ...this.
6351 (mips_hard_regno_mode_ok): New function.
6352 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6353 of HARD_REGNO_MODE_OK.
6354 (mips_option_override): Update after above name changes.
6355 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6356 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6357 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6358 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6359 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6360 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6361 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6362 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6363 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6364 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6365 (msp430_hard_regno_mode_ok): Make static and return a bool.
6366 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6367 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6368 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6370 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6371 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6372 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6373 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6374 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6375 (PA_HARD_REGNO_MODE_OK): ...this
6376 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6377 (PA_HARD_REGNO_MODE_OK): ...this.
6378 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6379 (pa_hard_regno_mode_ok): New function.
6380 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6381 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6382 (pdp11_hard_regno_mode_ok): New function.
6383 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6384 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6386 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6389 (rs6000_hard_regno_mode_ok): Rename to...
6390 (rs6000_hard_regno_mode_ok_uncached): ...this.
6391 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6392 (rs6000_hard_regno_mode_ok): New function.
6393 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6394 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6395 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6396 (riscv_hard_regno_mode_ok): ...this and make static.
6397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6398 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6399 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6400 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6401 (rl78_hard_regno_mode_ok): Make static and return bool.
6402 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6403 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6405 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6406 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6407 (rs6000_hard_regno_mode_ok): Rename to...
6408 (rs6000_hard_regno_mode_ok_uncached): ...this.
6409 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6410 (rs6000_hard_regno_mode_ok): New function.
6411 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6412 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6414 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6415 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6416 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6418 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6419 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6420 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6421 (sh_hard_regno_mode_ok): Make static.
6422 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6423 instead of HARD_REGNO_MODE_OK.
6424 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6425 (sparc_mode_class): Delete.
6426 (HARD_REGNO_MODE_OK): Delete.
6427 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6428 (hard_regno_mode_classes): Make static.
6429 (sparc_mode_class): Likewise.
6430 (sparc_hard_regno_mode_ok): New function.
6431 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6432 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6433 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6436 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6437 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6438 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6439 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6441 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6442 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6443 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6444 (visium_hard_regno_mode_ok): New function.
6445 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6446 instead of HARD_REGNO_MODE_OK.
6447 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6448 (HARD_REGNO_MODE_OK): Delete.
6449 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6450 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6451 (xtensa_option_override): Update accordingly.
6452 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6453 (xtensa_hard_regno_mode_ok): New function.
6454 * system.h (HARD_REGNO_MODE_OK): Poison.
6456 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6457 Alan Hayward <alan.hayward@arm.com>
6458 David Sherwood <david.sherwood@arm.com>
6460 * target.def (hard_regno_call_part_clobbered): New hook.
6461 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6462 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6463 * doc/tm.texi: Regenerate.
6464 * hooks.h (hook_bool_uint_mode_false): Declare.
6465 * hooks.c (hook_bool_uint_mode_false): New function.
6466 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6467 * cselib.c (cselib_process_insn): Use
6468 targetm.hard_regno_call_part_clobbered instead of
6469 HARD_REGNO_CALL_PART_CLOBBERED.
6470 * ira-conflicts.c (ira_build_conflicts): Likewise.
6471 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6472 * lra-constraints.c (need_for_call_save_p): Likewise.
6473 * lra-lives.c: Include target.h.
6474 (check_pseudos_live_through_calls): Use
6475 targetm.hard_regno_call_part_clobbered instead of
6476 HARD_REGNO_CALL_PART_CLOBBERED.
6477 * regcprop.c: Include target.h.
6478 (copyprop_hardreg_forward_1): Use
6479 targetm.hard_regno_call_part_clobbered instead of
6480 HARD_REGNO_CALL_PART_CLOBBERED.
6481 * reginfo.c (choose_hard_reg_mode): Likewise.
6482 * regrename.c (check_new_reg_p): Likewise.
6483 * reload.c (find_equiv_reg): Likewise.
6484 * reload1.c (emit_reload_insns): Likewise.
6485 * sched-deps.c (deps_analyze_insn): Likewise.
6486 * sel-sched.c (init_regs_for_mode): Likewise.
6487 (mark_unavailable_hard_regs): Likewise.
6488 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6489 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6490 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6492 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6493 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6494 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6496 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6498 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6499 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6500 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6503 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6504 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6507 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6509 * config/powerpcspe/powerpcspe.c
6510 (rs6000_hard_regno_call_part_clobbered): New function.
6511 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6512 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6513 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6516 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6517 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6519 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6520 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6521 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6523 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6524 Alan Hayward <alan.hayward@arm.com>
6525 David Sherwood <david.sherwood@arm.com>
6527 * rtl.h (subreg_memory_offset): Declare.
6528 * emit-rtl.c (subreg_memory_offset): New function.
6529 * expmed.c (store_bit_field_1): Use it.
6530 * expr.c (undefined_operand_subword_p): Likewise.
6531 * simplify-rtx.c (simplify_subreg): Likewise.
6533 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6535 PR rtl-optimization/57448
6538 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6539 using atomic_load pattern.
6540 (expand_atomic_store): Likewise.
6542 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6545 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6546 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6547 handling. Use replace_call_with_value with NULL instead of
6548 gsi_replace, unlink_stmt_vdef and release_defs.
6550 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6553 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6555 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6558 * rtl.h (emit_library_call_value_1): Declare.
6559 (emit_library_call): Replace declaration with a series of overloads.
6560 Remove the parameter count argument.
6561 (emit_library_call_value): Likewise.
6562 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6563 with an "rtx_mode_t *".
6564 (emit_library_call_value): Delete.
6565 (emit_library_call): Likewise.
6566 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6567 (asan_emit_allocas_unpoison): Likewise.
6568 * builtins.c (expand_builtin_powi): Likewise.
6569 (expand_asan_emit_allocas_unpoison): Likewise.
6570 * cfgexpand.c (expand_main_function): Likewise.
6571 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6572 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6573 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6574 * config/arm/arm.c (arm_trampoline_init): Likewise.
6575 (arm_call_tls_get_addr): Likewise.
6576 (arm_expand_divmod_libfunc): Likewise.
6577 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6578 (smulsi3_highpart): Likewise.
6579 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6580 (c6x_expand_compare): Likewise.
6581 (c6x_expand_movmem): Likewise.
6582 * config/frv/frv.c (frv_trampoline_init): Likewise.
6583 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6584 (ix86_expand_divmod_libfunc): Likewise.
6585 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6586 (ia64_expand_compare): Likewise.
6587 (ia64_profile_hook): Likewise.
6588 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6589 (nonlocal_goto): Likewise.
6590 (restore_stack_nonlocal): Likewise.
6591 * config/m32r/m32r.c (block_move_call): Likewise.
6592 (m32r_trampoline_init): Likewise.
6593 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6594 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6595 (m68k_call_m68k_read_tp): Likewise.
6596 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6597 (microblaze_expand_divide): Likewise.
6598 * config/mips/mips.h (mips_args): Likewise.
6599 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6600 (MIPS_ICACHE_SYNC): Likewise.
6601 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6602 (nios2_trampoline_init): Likewise.
6603 * config/pa/pa.c (hppa_tls_call): Likewise.
6604 (pa_trampoline_init): Likewise.
6605 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6606 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6607 (expand_strn_compare): Likewise.
6608 (rs6000_generate_compare): Likewise.
6609 (rs6000_expand_float128_convert): Likewise.
6610 (output_profile_hook): Likewise.
6611 (rs6000_trampoline_init): Likewise.
6612 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6613 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6614 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6615 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6616 (rs6000_generate_compare): Likewise.
6617 (rs6000_expand_float128_convert): Likewise.
6618 (output_profile_hook): Likewise.
6619 (rs6000_trampoline_init): Likewise.
6620 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6621 * config/sh/sh.c (sh_trampoline_init): Likewise.
6622 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6623 (sparc_emit_float_lib_cmp): Likewise.
6624 (sparc32_initialize_trampoline): Likewise.
6625 (sparc64_initialize_trampoline): Likewise.
6626 (sparc_profile_hook): Likewise.
6627 * config/spu/spu.c (ea_load_store): Likewise.
6628 * config/spu/spu.md (floatunssidf2): Likewise.
6629 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6630 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6631 * config/visium/visium.c (expand_block_move_4): Likewise.
6632 (expand_block_move_2): Likewise.
6633 (expand_block_move_1): Likewise.
6634 (expand_block_set_4): Likewise.
6635 (expand_block_set_2): Likewise.
6636 (expand_block_set_1): Likewise.
6637 (visium_trampoline_init): Likewise.
6638 (visium_profile_hook): Likewise.
6639 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6640 (xtensa_setup_frame_addresses): Likewise.
6641 (xtensa_trampoline_init): Likewise.
6642 * except.c (sjlj_emit_function_enter): Likewise.
6643 (sjlj_emit_function_exit): Likewise.
6644 * explow.c (allocate_dynamic_stack_space): Likewise.
6645 (probe_stack_range): Likewise.
6646 * expr.c (convert_mode_scalar): Likewise.
6647 * optabs.c (expand_binop): Likewise.
6648 (expand_twoval_binop_libfunc): Likewise.
6649 (expand_unop): Likewise.
6650 (prepare_cmp_insn): Likewise.
6651 (prepare_float_lib_cmp): Likewise.
6652 (expand_float): Likewise.
6653 (expand_fix): Likewise.
6654 (expand_fixed_convert): Likewise.
6655 (maybe_emit_sync_lock_test_and_set): Likewise.
6656 (expand_atomic_compare_and_swap): Likewise.
6657 (expand_mem_thread_fence): Likewise.
6658 (expand_atomic_fetch_op): Likewise.
6660 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6662 * doc/generic.texi (OpenACC): Adjust URL.
6663 * doc/invoke.texi (C Dialect Options): Ditto.
6665 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6667 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6668 predicate for operand 1. Add (m,<S>) constraint.
6669 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6670 Prevent memory operand 1 with register operand 2.
6672 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6674 PR rtl-optimization/82024
6675 * combine.c (try_combine): If the combination result is a PARALLEL,
6676 and we only need to retain the SET in there that would be placed
6677 at I2, check that we can place that at I3 instead, before doing so.
6679 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6682 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6684 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6685 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6688 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6689 Jeff Law <law@redhat.com>
6691 * varasm.c (bss_initializer_p): Do not put constants into .bss
6692 (categorize_decl_for_section): Handle bss_initializer_p returning
6693 false when DECL_INITIAL is NULL.
6695 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6698 * config/s390/s390.c (s390_can_inline_p): New function.
6700 2017-09-01 Jeff Law <law@redhat.com>
6702 PR tree-optimization/82052
6703 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6704 Always initialize the returned slot after a hash table miss
6705 when INSERT is true.
6707 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6709 * config/s390/s390.md (mem_signal_fence): Remove.
6710 * doc/md.texi (mem_signal_fence): Remove.
6711 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6713 * target-insns.def (mem_signal_fence): Remove.
6715 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6718 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6721 * asan.c (create_odr_indicator): Strip name encoding from assembler
6722 name before appending it after __odr_asan_.
6724 2017-09-01 Martin Liska <mliska@suse.cz>
6726 PR tree-optimization/82059
6727 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6728 frequency only when an edge is redirected.
6730 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6732 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6733 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6734 (arc_conditional_register_usage): Remove ARC600 lp_count
6736 (arc_file_start): Emit Tag_ARC_CPU_variation.
6737 (arc_can_use_doloop_p): New conditions to use ZOLs.
6738 (hwloop_fail): New function.
6739 (hwloop_optimize): Likewise.
6740 (hwloop_pattern_reg): Likewise.
6741 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6742 (arc_reorg_loops): New function, calls reorg_loops.
6743 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6744 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6746 (arc_loop_hazard): Remove function, functionality moved into
6748 (arc_hazard): Remove arc_loop_hazard call.
6749 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6750 into hwloop_optimize.
6751 (arc_label_align): Remove ZOL handling.
6752 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6753 * config/arc/arc.md (doloop_begin): Remove pattern.
6754 (doloop_begin_i): Likewise.
6755 (doloop_end_i): Likewise.
6756 (doloop_fallback): Likewise.
6757 (doloop_fallback_m): Likewise.
6758 (doloop_end): Reimplement expand.
6759 (arc_lp): New pattern for LP instruction.
6760 (loop_end): New pattern.
6761 (loop_fail): Likewise.
6762 (decrement_and_branch_until_zero): Likewise.
6763 * config/arc/arc.opt (mlpc-width): New option.
6764 * doc/invoke.texi (mlpc-width): Document option.
6766 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6768 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6769 (arc_ccfsm_advance): Fix checking for delay slots.
6770 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6772 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6774 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6776 (movhi_insn): Update store instruction constraint which are saving
6777 6-bit short immediates.
6778 (movsi_insn): Consider also short scaled load operations.
6779 (zero_extendhisi2_i): Use Usd constraint instead of T.
6780 (extendhisi2_i): Add q constraint.
6781 (arc_clzsi2): Add type and length attributes.
6782 (arc_ctzsi2): Likewise.
6783 * config/arc/constraints.md (Usc): Update constraint, the
6784 assembler can parse two relocations for a single instruction.
6786 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6788 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6789 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6791 2017-08-31 Olivier Hainque <hainque@adacore.com>
6793 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6794 match as powerpc-wrs-vxworks*.
6796 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6798 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6799 register constraint for by-element operand.
6800 (aarch64_mls_elt_merge<mode>): Likewise.
6802 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6804 * config/arc/arc.c (arc_can_follow_jump): Check for short
6807 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6809 * config.gcc: Use g.opt for arc.
6810 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6811 functionality moved to ...
6812 (legitimate_scaled_address_p): New function, ...here.
6813 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6814 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6815 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6817 (arc_override_options): Handle G option.
6818 (arc_output_pic_addr_const): Correct function definition.
6819 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6820 (arc_decl_anon_ns_mem_p): Delete.
6821 (arc_in_small_data_p): Overhaul this function to take into
6822 consideration the value given via G option.
6823 (arc_rewrite_small_data_1): Renamed and corrected old
6824 arc_rewrite_small_data function.
6825 (arc_rewrite_small_data): New function.
6826 (small_data_pattern): Don't use pic_offset_table_rtx.
6827 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6828 * config/arc/simdext.md (movmisalignv2hi): Use
6829 prepare_move_operands function.
6831 (movmisalign*): Likewise.
6832 * doc/invoke.texi (ARC options): Document -G option.
6834 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6836 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6838 * config/arc/arc.c (arc_print_operand): Output scalled address for
6839 sdata whenever is possible.
6840 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6841 load/stores are available.
6842 (compact_sda_memory_operand): Check for the alignment required by
6843 code density instructions.
6844 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6846 * config/arc/constraints.md (Usd): Update constraint.
6847 (Us0): New constraint.
6848 (Usc): Update constraint.
6850 2017-08-31 Richard Biener <rguenther@suse.de>
6853 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6856 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6858 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6859 Resize type_signature.
6861 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6862 Alan Hayward <alan.hayward@arm.com>
6863 David Sherwood <david.sherwood@arm.com>
6865 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6866 subregs whose inner modes can be stored in GPRs.
6867 (aarch64_classify_index): Likewise.
6869 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6870 Alan Hayward <alan.hayward@arm.com>
6871 David Sherwood <david.sherwood@arm.com>
6873 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6874 (V_INT_EQUIV): ...this.
6875 (v_cmp_result): Rename to...
6876 (v_int_equiv): ...this.
6877 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6878 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6879 (copysign<mode>3): Likewise.
6880 (aarch64_simd_bsl<mode>_internal): Likewise.
6881 (aarch64_simd_bsl<mode>): Likewise.
6882 (vec_cmp<mode><mode>): Likewise.
6883 (vcond<mode><mode>): Likewise.
6884 (vcond<v_cmp_mixed><mode>): Likewise.
6885 (vcondu<mode><v_cmp_mixed>): Likewise.
6886 (aarch64_cm<optab><mode>): Likewise.
6887 (aarch64_cmtst<mode>): Likewise.
6888 (aarch64_fac<optab><mode>): Likewise.
6889 (vec_perm_const<mode>): Likewise.
6890 (vcond_mask_<mode><v_cmp_result>): Rename to...
6891 (vcond_mask_<mode><v_int_equiv>): ...this.
6892 (vec_cmp<mode><v_cmp_result>): Rename to...
6893 (vec_cmp<mode><v_int_equiv>): ...this.
6895 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6896 Alan Hayward <alan.hayward@arm.com>
6897 David Sherwood <david.sherwood@arm.com>
6899 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6901 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6902 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6903 (UNSPEC_LD4_DREG): New unspecs.
6904 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6905 (aarch64_ld2<mode>_dreg_be): Replace with...
6906 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6908 (aarch64_ld3<mode>_dreg_le)
6909 (aarch64_ld3<mode>_dreg_be): Replace with...
6910 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6912 (aarch64_ld4<mode>_dreg_le)
6913 (aarch64_ld4<mode>_dreg_be): Replace with...
6914 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6917 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6919 PR tree-optimization/81987
6920 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6921 insert an initializer in a location not dominated by the stride
6924 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6926 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6927 on the entire header of the finally block in the fallthru case.
6929 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6931 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6933 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6935 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6936 rs6000_emit_move_from_cr and call renamed function.
6937 (rs6000_emit_prologue): Call renamed functions.
6938 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6939 movesi_from_cr, remove volatile CRs.
6941 2017-08-30 Jon Beniston <jon@beniston.com>
6942 Richard Biener <rguenther@suse.de>
6944 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6945 of VECTOR_MODE_P check.
6946 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6947 element vector types.
6949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6951 * df.h (df_read_modify_subreg_p): Remove in favor of...
6952 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6953 const_rtx instead of an rtx.
6954 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6955 * df-problems.c (df_word_lr_mark_ref): Likewise.
6956 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6957 (mark_pseudo_reg_dead): Likewise.
6958 (mark_ref_dead): Likewise.
6959 * reginfo.c (init_subregs_of_mode): Likewise.
6960 * sched-deps.c (sched_analyze_1): Likewise.
6961 * df-scan.c (df_def_record_1): Likewise.
6962 (df_uses_record): Likewise.
6963 (df_read_modify_subreg_p): Remove in favor of...
6964 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6965 const_rtx instead of an rtx.
6967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6968 Alan Hayward <alan.hayward@arm.com>
6969 David Sherwood <david.sherwood@arm.com>
6971 * rtl.h (partial_subreg_p): New function.
6972 * caller-save.c (save_call_clobbered_regs): Use it.
6973 * calls.c (expand_call): Likewise.
6974 * combine.c (combinable_i3pat): Likewise.
6975 (simplify_set): Likewise.
6976 (make_extraction): Likewise.
6977 (make_compound_operation_int): Likewise.
6978 (gen_lowpart_or_truncate): Likewise.
6979 (force_to_mode): Likewise.
6980 (make_field_assignment): Likewise.
6981 (reg_truncated_to_mode): Likewise.
6982 (record_truncated_value): Likewise.
6983 (move_deaths): Likewise.
6984 * cse.c (record_jump_cond): Likewise.
6985 (cse_insn): Likewise.
6986 * cselib.c (cselib_lookup_1): Likewise.
6987 * expmed.c (extract_bit_field_using_extv): Likewise.
6988 * function.c (assign_parm_setup_reg): Likewise.
6989 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6990 * ira-build.c (create_insn_allocnos): Likewise.
6991 * lra-coalesce.c (merge_pseudos): Likewise.
6992 * lra-constraints.c (match_reload): Likewise.
6993 (simplify_operand_subreg): Likewise.
6994 (curr_insn_transform): Likewise.
6995 * lra-lives.c (process_bb_lives): Likewise.
6996 * lra.c (new_insn_reg): Likewise.
6997 (lra_substitute_pseudo): Likewise.
6998 * regcprop.c (mode_change_ok): Likewise.
6999 (maybe_mode_change): Likewise.
7000 (copyprop_hardreg_forward_1): Likewise.
7001 * reload.c (push_reload): Likewise.
7002 (find_reloads): Likewise.
7003 (find_reloads_subreg_address): Likewise.
7004 * reload1.c (alter_reg): Likewise.
7005 (eliminate_regs_1): Likewise.
7006 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7008 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7010 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7011 back to if statements, including unpack.
7013 2017-08-30 Martin Liska <mliska@suse.cz>
7016 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7018 (func_checker::compare_asm_inputs_outputs): ... this function.
7019 (func_checker::compare_gimple_asm): Use the function to compare
7020 also ASM constrains.
7021 * ipa-icf-gimple.h: Rename the function.
7023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7024 Alan Hayward <alan.hayward@arm.com>
7025 David Sherwood <david.sherwood@arm.com>
7027 * coretypes.h (complex_mode): New type.
7028 * gdbhooks.py (build_pretty_printer): Handle it.
7029 * machmode.h (complex_mode): New class.
7030 (complex_mode::includes_p): New function.
7031 (is_complex_int_mode): Likewise.
7032 (is_complex_float_mode): Likewise.
7033 * genmodes.c (get_mode_class): Handle complex mode classes.
7034 * function.c (expand_function_end): Use is_complex_int_mode.
7036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7037 Alan Hayward <alan.hayward@arm.com>
7038 David Sherwood <david.sherwood@arm.com>
7040 * coretypes.h (scalar_mode_pod): New typedef.
7041 * gdbhooks.py (build_pretty_printer): Handle it.
7042 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7043 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7044 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7045 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7049 Alan Hayward <alan.hayward@arm.com>
7050 David Sherwood <david.sherwood@arm.com>
7052 * machmode.h (mode_for_vector): Take a scalar_mode instead
7054 * stor-layout.c (mode_for_vector): Likewise.
7055 * explow.c (promote_mode): Use as_a <scalar_mode>.
7056 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7059 Alan Hayward <alan.hayward@arm.com>
7060 David Sherwood <david.sherwood@arm.com>
7062 * target.def (preferred_simd_mode): Take a scalar_mode
7063 instead of a machine_mode.
7064 * targhooks.h (default_preferred_simd_mode): Likewise.
7065 * targhooks.c (default_preferred_simd_mode): Likewise.
7066 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7067 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7068 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7069 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7070 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7071 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7072 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7073 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7075 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7076 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7077 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7078 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7079 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7080 * doc/tm.texi: Regenerate.
7081 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7085 Alan Hayward <alan.hayward@arm.com>
7086 David Sherwood <david.sherwood@arm.com>
7088 * target.def (scalar_mode_supported_p): Take a scalar_mode
7089 instead of a machine_mode.
7090 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7091 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7092 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7093 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7094 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7095 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7096 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7097 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7098 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7099 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7100 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7101 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7102 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7103 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7105 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7106 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7107 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7108 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7109 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7111 * doc/tm.texi: Regenerate.
7113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7114 Alan Hayward <alan.hayward@arm.com>
7115 David Sherwood <david.sherwood@arm.com>
7117 * coretypes.h (opt_scalar_mode): New typedef.
7118 * gdbhooks.py (build_pretty_printers): Handle it.
7119 * machmode.h (mode_iterator::get_2xwider): Add overload for
7121 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7123 * expr.c (convert_mode_scalar): Likewise.
7124 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7125 * optabs.c (expand_float): Likewise.
7126 (expand_fix): Likewise.
7127 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7130 Alan Hayward <alan.hayward@arm.com>
7131 David Sherwood <david.sherwood@arm.com>
7133 * optabs.c (expand_float): Explicitly check for scalars before
7134 using a branching expansion.
7135 (expand_fix): Likewise.
7137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7138 Alan Hayward <alan.hayward@arm.com>
7139 David Sherwood <david.sherwood@arm.com>
7141 * expr.c (convert_mode): Split scalar handling out into...
7142 (convert_mode_scalar): ...this new function. Treat the modes
7145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7146 Alan Hayward <alan.hayward@arm.com>
7147 David Sherwood <david.sherwood@arm.com>
7149 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7151 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7154 Alan Hayward <alan.hayward@arm.com>
7155 David Sherwood <david.sherwood@arm.com>
7157 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7158 rather than a machine_mode.
7159 (fixed_from_string): Likewise.
7160 (fixed_convert): Likewise.
7161 (fixed_convert_from_int): Likewise.
7162 (fixed_convert_from_real): Likewise.
7163 (real_convert_from_fixed): Likewise.
7164 * fixed-value.c (fixed_from_double_int): Likewise.
7165 (fixed_from_string): Likewise.
7166 (fixed_convert): Likewise.
7167 (fixed_convert_from_int): Likewise.
7168 (fixed_convert_from_real): Likewise.
7169 (real_convert_from_fixed): Likewise.
7170 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7173 Alan Hayward <alan.hayward@arm.com>
7174 David Sherwood <david.sherwood@arm.com>
7176 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7177 of separate mode class checks. Do not allow vector modes here.
7178 (immed_wide_int_const): Use as_a <scalar_mode>.
7179 * explow.c (trunc_int_for_mode): Likewise.
7180 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7181 (wi::shwi): Likewise.
7182 (wi::min_value): Likewise.
7183 (wi::max_value): Likewise.
7184 * dwarf2out.c (loc_descriptor): Likewise.
7185 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7189 Alan Hayward <alan.hayward@arm.com>
7190 David Sherwood <david.sherwood@arm.com>
7192 * tree.h (SCALAR_TYPE_MODE): New macro.
7193 * expr.c (expand_expr_addr_expr_1): Use it.
7194 (expand_expr_real_2): Likewise.
7195 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7196 (fold_convert_const_fixed_from_int): Likewise.
7197 (fold_convert_const_fixed_from_real): Likewise.
7198 (native_encode_fixed): Likewise
7199 (native_encode_complex): Likewise
7200 (native_encode_vector): Likewise.
7201 (native_interpret_fixed): Likewise.
7202 (native_interpret_real): Likewise.
7203 (native_interpret_complex): Likewise.
7204 (native_interpret_vector): Likewise.
7205 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7206 (simd_clone_adjust_argument_types): Likewise.
7207 (simd_clone_init_simd_arrays): Likewise.
7208 (simd_clone_adjust): Likewise.
7209 * stor-layout.c (layout_type): Likewise.
7210 * tree.c (build_minus_one_cst): Likewise.
7211 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7212 * tree-inline.c (estimate_move_cost): Likewise.
7213 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7214 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7215 (vectorizable_reduction): Likewise.
7216 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7217 (vect_recog_mixed_size_cond_pattern): Likewise.
7218 (check_bool_pattern): Likewise.
7219 (adjust_bool_pattern): Likewise.
7220 (search_type_for_mask_1): Likewise.
7221 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7222 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7223 (vectorizable_load): Likewise.
7224 (vectorizable_store): Likewise.
7225 * ubsan.c (ubsan_encode_value): Likewise.
7226 * varasm.c (output_constant): Likewise.
7228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7229 Alan Hayward <alan.hayward@arm.com>
7230 David Sherwood <david.sherwood@arm.com>
7232 * coretypes.h (scalar_mode): New class.
7233 * machmode.h (scalar_mode): Likewise.
7234 (scalar_mode::includes_p): New function.
7235 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7236 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7237 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7238 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7239 * expmed.c (store_bit_field_1): Likewise.
7240 (extract_bit_field_1): Likewise.
7241 * expr.c (write_complex_part): Likewise.
7242 (read_complex_part): Likewise.
7243 (emit_move_complex_push): Likewise.
7244 (expand_expr_real_2): Likewise.
7245 * function.c (assign_parm_setup_reg): Likewise.
7246 (assign_parms_unsplit_complex): Likewise.
7247 * optabs.c (expand_binop): Likewise.
7248 * rtlanal.c (subreg_get_info): Likewise.
7249 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7250 * varasm.c (output_constant_pool_2): Likewise.
7252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7253 Alan Hayward <alan.hayward@arm.com>
7254 David Sherwood <david.sherwood@arm.com>
7256 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7258 (expmed_mult_highpart_optab): Likewise.
7259 (expmed_mult_highpart): Likewise.
7261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7262 Alan Hayward <alan.hayward@arm.com>
7263 David Sherwood <david.sherwood@arm.com>
7265 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7266 instead of a machine_mode.
7267 (builtin_memset_read_str): Likewise.
7268 * builtins.c (c_readstr): Likewise.
7269 (builtin_memcpy_read_str): Likewise.
7270 (builtin_strncpy_read_str): Likewise.
7271 (builtin_memset_read_str): Likewise.
7272 (builtin_memset_gen_str): Likewise.
7273 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7274 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7275 instead of a machine_mode.
7276 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7278 (make_extraction): Likewise.
7279 (try_widen_shift_mode): Take and return scalar_int_modes instead
7281 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7282 a scalar_int_mode instead of a machine_mode.
7283 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7284 (avr_addr_space_pointer_mode): Likewise.
7285 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7286 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7287 (msp430_unwind_word_mode): Likewise.
7288 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7289 (spu_addr_space_pointer_mode): Likewise.
7290 (spu_addr_space_address_mode): Likewise.
7291 (spu_libgcc_cmp_return_mode): Likewise.
7292 (spu_libgcc_shift_count_mode): Likewise.
7293 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7294 (rl78_addr_space_pointer_mode): Likewise.
7295 (fl78_unwind_word_mode): Likewise.
7296 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7298 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7299 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7300 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7301 (mips_valid_pointer_mode): Likewise.
7302 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7303 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7304 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7306 (ft32_addr_space_address_mode): Likewise.
7307 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7308 scalar_int_mode instead of a machine_mode.
7309 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7311 (m32c_addr_space_address_mode): Likewise.
7312 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7313 (rs6000_eh_return_filter_mode): Likewise.
7314 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7315 (rs6000_eh_return_filter_mode): Likewise.
7316 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7317 (s390_libgcc_shift_count_mode): Likewise.
7318 (s390_unwind_word_mode): Likewise.
7319 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7321 * target.def (mode_rep_extended): Likewise.
7322 (valid_pointer_mode): Likewise.
7323 (addr_space.valid_pointer_mode): Likewise.
7324 (eh_return_filter_mode): Return a scalar_int_mode rather than
7326 (libgcc_cmp_return_mode): Likewise.
7327 (libgcc_shift_count_mode): Likewise.
7328 (unwind_word_mode): Likewise.
7329 (addr_space.pointer_mode): Likewise.
7330 (addr_space.address_mode): Likewise.
7331 * doc/tm.texi: Regenerate.
7332 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7334 (do_jump): Use scalar_int_mode for local variables.
7335 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7336 rather than a machine_mode.
7337 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7338 (scompare_loc_descriptor_wide): Likewise.
7339 (scompare_loc_descriptor_narrow): Likewise.
7340 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7342 * except.c (sjlj_emit_dispatch_table): Likewise.
7343 (expand_builtin_eh_copy_values): Likewise.
7344 * explow.c (convert_memory_address_addr_space_1): Likewise.
7345 Take a scalar_int_mode rather than a machine_mode.
7346 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7347 than a machine_mode.
7348 (memory_address_addr_space): Use scalar_int_mode for local variables.
7349 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7350 rather than a machine_mode.
7351 * expmed.c (mask_rtx): Likewise.
7352 (init_expmed_one_conv): Likewise.
7353 (expand_mult_highpart_adjust): Likewise.
7354 (extract_high_half): Likewise.
7355 (expmed_mult_highpart_optab): Likewise.
7356 (expmed_mult_highpart): Likewise.
7357 (expand_smod_pow2): Likewise.
7358 (expand_sdiv_pow2): Likewise.
7359 (emit_store_flag_int): Likewise.
7360 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7362 (extract_low_bits): Likewise.
7363 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7365 * expr.c (pieces_addr::adjust): Likewise.
7366 (can_store_by_pieces): Likewise.
7367 (store_by_pieces): Likewise.
7368 (clear_by_pieces_1): Likewise.
7369 (expand_expr_addr_expr_1): Likewise.
7370 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7371 (expand_expr_real_1): Likewise.
7372 (try_casesi): Likewise.
7373 * final.c (shorten_branches): Likewise.
7374 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7375 type of "mode" to machine_mode.
7376 * internal-fn.c (expand_arith_overflow_result_store): Take a
7377 scalar_int_mode rather than a machine_mode.
7378 (expand_mul_overflow): Use scalar_int_mode for local variables.
7379 * loop-doloop.c (doloop_modify): Likewise.
7380 (doloop_optimize): Likewise.
7381 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7382 than a machine_mode.
7383 (expand_doubleword_shift_condmove): Likewise.
7384 (expand_doubleword_shift): Likewise.
7385 (expand_doubleword_clz): Likewise.
7386 (expand_doubleword_popcount): Likewise.
7387 (expand_doubleword_parity): Likewise.
7388 (expand_absneg_bit): Use scalar_int_mode for local variables.
7389 (prepare_float_lib_cmp): Likewise.
7390 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7391 rather than a machine_mode.
7392 (convert_memory_address_addr_space): Likewise.
7393 (get_mode_bounds): Likewise.
7394 (get_address_mode): Return a scalar_int_mode rather than a
7396 * rtlanal.c (get_address_mode): Likewise.
7397 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7398 than a machine_mode.
7399 * targhooks.c (default_mode_rep_extended): Likewise.
7400 (default_valid_pointer_mode): Likewise.
7401 (default_addr_space_valid_pointer_mode): Likewise.
7402 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7403 than a machine_mode.
7404 (default_libgcc_cmp_return_mode): Likewise.
7405 (default_libgcc_shift_count_mode): Likewise.
7406 (default_unwind_word_mode): Likewise.
7407 (default_addr_space_pointer_mode): Likewise.
7408 (default_addr_space_address_mode): Likewise.
7409 * targhooks.h (default_eh_return_filter_mode): Likewise.
7410 (default_libgcc_cmp_return_mode): Likewise.
7411 (default_libgcc_shift_count_mode): Likewise.
7412 (default_unwind_word_mode): Likewise.
7413 (default_addr_space_pointer_mode): Likewise.
7414 (default_addr_space_address_mode): Likewise.
7415 (default_mode_rep_extended): Take a scalar_int_mode rather than
7417 (default_valid_pointer_mode): Likewise.
7418 (default_addr_space_valid_pointer_mode): Likewise.
7419 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7421 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7422 rather than a machine_mode.
7423 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7424 for local variables.
7425 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7426 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7427 than a machine_mode.
7429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7430 Alan Hayward <alan.hayward@arm.com>
7431 David Sherwood <david.sherwood@arm.com>
7433 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7434 the mode argument to scalar_int_mode.
7435 (do_jump_by_parts_zero_rtx): Likewise.
7436 (do_jump_by_parts_equality_rtx): Likewise.
7437 (do_jump_by_parts_greater): Take a mode argument.
7438 (do_jump_by_parts_equality): Likewise.
7439 (do_jump_1): Update calls accordingly.
7441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7442 Alan Hayward <alan.hayward@arm.com>
7443 David Sherwood <david.sherwood@arm.com>
7445 * is-a.h (safe_dyn_cast): New function.
7446 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7447 (jump_table_for_label): Likewise.
7448 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7449 instead of an rtx_insn *.
7450 (shorten_branches): Use dyn_cast instead of LABEL_P and
7451 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7452 rtx_jump_table_data::get_data_mode.
7453 (final_scan_insn): Likewise.
7455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7456 Alan Hayward <alan.hayward@arm.com>
7457 David Sherwood <david.sherwood@arm.com>
7459 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7460 trying to combine a full-register integer set with a subreg
7463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7464 Alan Hayward <alan.hayward@arm.com>
7465 David Sherwood <david.sherwood@arm.com>
7467 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7468 that is always either address_mode or pointer_mode.
7470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7471 Alan Hayward <alan.hayward@arm.com>
7472 David Sherwood <david.sherwood@arm.com>
7474 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7475 when the two are known to be equal.
7477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7478 Alan Hayward <alan.hayward@arm.com>
7479 David Sherwood <david.sherwood@arm.com>
7481 * simplify-rtx.c (simplify_const_unary_operation): Use
7482 is_a <scalar_int_mode> instead of checking for a nonzero
7483 precision. Forcibly convert op_mode to a scalar_int_mode
7484 in that case. More clearly differentiate the operand and
7485 result modes and use the former when deciding what the value
7486 of a count-bits operation should be. Use is_int_mode instead
7487 of checking for a MODE_INT. Remove redundant check for whether
7488 this mode has a zero precision.
7490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7491 Alan Hayward <alan.hayward@arm.com>
7492 David Sherwood <david.sherwood@arm.com>
7494 * optabs.c (widen_leading): Change the type of the mode argument
7495 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7496 (widen_bswap): Likewise.
7497 (expand_parity): Likewise.
7498 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7499 (expand_ffs): Likewise.
7500 (epand_unop): Check for scalar integer modes before calling the
7503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7504 Alan Hayward <alan.hayward@arm.com>
7505 David Sherwood <david.sherwood@arm.com>
7507 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7509 (expand_expr_real_1): Update call accordingly.
7511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7512 Alan Hayward <alan.hayward@arm.com>
7513 David Sherwood <david.sherwood@arm.com>
7515 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7516 value_mode arguments. Use scalar_int_mode internally.
7517 (store_bit_field_1): Rename the new integer mode from imode
7518 to op0_mode and use it instead of GET_MODE (op0). Update calls
7519 to store_split_bit_field, store_bit_field_using_insv and
7520 store_fixed_bit_field.
7521 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7522 Use scalar_int_mode internally. Use a bit count rather than a mode
7523 when calculating the largest bit size for get_best_mode.
7524 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7525 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7526 Remove assertion that OP0 has a scalar integer mode.
7527 (store_split_bit_field): Add op0_mode and value_mode arguments.
7528 Update calls to extract_fixed_bit_field.
7529 (extract_bit_field_using_extv): Add an op0_mode argument.
7530 Use scalar_int_mode internally.
7531 (extract_bit_field_1): Rename the new integer mode from imode to
7532 op0_mode and use it instead of GET_MODE (op0). Update calls to
7533 extract_split_bit_field, extract_bit_field_using_extv and
7534 extract_fixed_bit_field.
7535 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7536 to extract_split_bit_field and extract_fixed_bit_field_1.
7537 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7538 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7540 (extract_split_bit_field): Add an op0_mode argument. Update call
7541 to extract_fixed_bit_field.
7543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7548 * explow.c (hard_function_value): Likewise.
7549 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7550 convert_to_mode call outside the loop.
7551 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7552 for the mode iterator. Require the mode specified by max_pieces
7554 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7556 (copy_blkmode_to_reg): Likewise.
7557 (set_storage_via_setmem): Likewise.
7558 * optabs.c (prepare_cmp_insn): Likewise.
7559 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7560 * stor-layout.c (finish_bitfield_representative): Likewise.
7562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7563 Alan Hayward <alan.hayward@arm.com>
7564 David Sherwood <david.sherwood@arm.com>
7566 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7567 * expr.c (convert_move): Use them.
7568 (convert_modes): Likewise.
7569 (store_expr_with_bounds): Likewise.
7571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7572 Alan Hayward <alan.hayward@arm.com>
7573 David Sherwood <david.sherwood@arm.com>
7575 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7576 parameter for the mode of "x". Remove the "known_x", "known_mode"
7577 and "known_ret" arguments. Change the type of the mode argument
7579 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7580 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7581 (reg_num_sign_bit_copies_for_combine): Likewise.
7582 * rtlanal.c (nonzero_bits1): Likewise.
7583 (num_sign_bit_copies1): Likewise.
7584 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7585 (reg_num_sign_bit_copies_general): Likewise.
7586 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7587 (reg_nonzero_bits_general): Likewise.
7589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7590 Alan Hayward <alan.hayward@arm.com>
7591 David Sherwood <david.sherwood@arm.com>
7593 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7594 than in subroutines. Return 1 for non-integer modes.
7595 (cached_num_sign_bit_copies): Change the type of the mode parameter
7597 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7598 classes. Handle CONST_INT_P first and then check whether X also
7599 has a scalar integer mode. Check the same thing for inner registers
7600 of a SUBREG and for values that are being extended or truncated.
7602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7603 Alan Hayward <alan.hayward@arm.com>
7604 David Sherwood <david.sherwood@arm.com>
7606 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7607 in subroutines. Return the mode mask for non-integer modes.
7608 (cached_nonzero_bits): Change the type of the mode parameter
7610 (nonzero_bits1): Likewise. Remove early exit for other mode
7611 classes. Handle CONST_INT_P first and then check whether X
7612 also has a scalar integer mode.
7614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7615 Alan Hayward <alan.hayward@arm.com>
7616 David Sherwood <david.sherwood@arm.com>
7618 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7619 Return a scalar_int_mode and assert that the size is greater than
7621 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7622 (op_by_pieces_d::op_by_pieces_d): Likewise.
7623 (op_by_pieces_d::run): Likewise.
7624 (can_store_by_pieces): Likewise.
7626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7627 Alan Hayward <alan.hayward@arm.com>
7628 David Sherwood <david.sherwood@arm.com>
7630 * combine.c (extract_left_shift): Add a mode argument and update
7632 (make_compound_operation_int): Change the type of the mode parameter
7633 to scalar_int_mode and update the call to extract_left_shift.
7635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7636 Alan Hayward <alan.hayward@arm.com>
7637 David Sherwood <david.sherwood@arm.com>
7639 * combine.c (simplify_and_const_int): Change the type of the mode
7640 parameter to scalar_int_mode.
7641 (simplify_and_const_int_1): Likewise. Update recursive call.
7643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7644 Alan Hayward <alan.hayward@arm.com>
7645 David Sherwood <david.sherwood@arm.com>
7647 * combine.c (simplify_compare_const): Check that the mode is a
7648 scalar_int_mode (rather than VOIDmode) before testing its
7650 (simplify_comparison): Move COMPARISON_P handling out of the
7651 loop and restrict the latter part of the loop to scalar_int_modes.
7652 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7653 and when considering SUBREG_REGs. Use is_int_mode instead of
7654 checking GET_MODE_CLASS against MODE_INT.
7656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7657 Alan Hayward <alan.hayward@arm.com>
7658 David Sherwood <david.sherwood@arm.com>
7660 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7661 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7662 shift_unit_mode and for modes involved in scalar shifts.
7664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7665 Alan Hayward <alan.hayward@arm.com>
7666 David Sherwood <david.sherwood@arm.com>
7668 * combine.c (force_int_to_mode): New function, split out from...
7669 (force_to_mode): ...here. Keep xmode up-to-date and use it
7670 instead of GET_MODE (x).
7672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7673 Alan Hayward <alan.hayward@arm.com>
7674 David Sherwood <david.sherwood@arm.com>
7676 * optabs-query.h (extraction_insn::struct_mode): Change type to
7677 opt_scalar_int_mode and update comment.
7678 (extraction_insn::field_mode): Change type to scalar_int_mode.
7679 (extraction_insn::pos_mode): Likewise.
7680 * combine.c (make_extraction): Update accordingly.
7681 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7682 (get_optab_extraction_insn): Likewise.
7683 * recog.c (simplify_while_replacing): Likewise.
7684 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7685 parameter to opt_scalar_int_mode.
7687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7688 Alan Hayward <alan.hayward@arm.com>
7689 David Sherwood <david.sherwood@arm.com>
7691 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7692 to a scalar_int_mode instead of a machine_mode.
7693 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7694 (get_best_mode): Return a boolean and use a pointer argument to store
7695 the selected mode. Replace the limit mode parameter with a bit limit.
7696 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7697 for the values returned by bit_field_mode_iterator::next_mode.
7698 (store_bit_field): Update call to get_best_mode.
7699 (store_fixed_bit_field): Likewise.
7700 (extract_fixed_bit_field): Likewise.
7701 * expr.c (optimize_bitfield_assignment_op): Likewise.
7702 * fold-const.c (optimize_bit_field_compare): Likewise.
7703 (fold_truth_andor_1): Likewise.
7704 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7705 Update for new type of m_mode.
7706 (get_best_mode): As above.
7708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7709 Alan Hayward <alan.hayward@arm.com>
7710 David Sherwood <david.sherwood@arm.com>
7712 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7713 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7714 (store_bit_field): Check is_a <scalar_int_mode> before calling
7715 strict_volatile_bitfield_p.
7716 (extract_bit_field): Likewise.
7718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7719 Alan Hayward <alan.hayward@arm.com>
7720 David Sherwood <david.sherwood@arm.com>
7722 * target.def (cstore_mode): Return a scalar_int_mode.
7723 * doc/tm.texi: Regenerate.
7724 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7725 * targhooks.h (default_cstore_mode): Likewise.
7726 * targhooks.c (default_cstore_mode): Likewise, using a forced
7728 * expmed.c (emit_cstore): Expect the target of the cstore to be
7731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7732 Alan Hayward <alan.hayward@arm.com>
7733 David Sherwood <david.sherwood@arm.com>
7735 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7737 (niter_desc): Likewise mode.
7738 (iv_analyze): Add a mode parameter.
7740 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7741 and change its type to scalar_int_mode.
7742 * loop-iv.c: Update commentary at head of file.
7743 (iv_constant): Pass the mode paraeter before the rtx it describes
7744 and change its type to scalar_int_mode. Remove VOIDmode handling.
7745 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7746 (iv_extend): Likewise.
7747 (shorten_into_mode): Likewise.
7748 (iv_add): Use scalar_int_mode.
7749 (iv_mult): Likewise.
7750 (iv_shift): Likewise.
7751 (canonicalize_iv_subregs): Likewise.
7752 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7753 it describes and change its mode to scalar_int_mode. Also change
7754 the type of the returned inner_mode to scalar_int_mode.
7755 (get_biv_step): Likewise, turning outer_mode from a pointer
7756 into a direct parameter. Update call to get_biv_step_1.
7757 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7758 iv_constant and get_biv_step.
7759 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7760 and change its type to scalar_int_mode. Don't initialise iv->mode
7761 to VOIDmode and remove later checks for its still being VOIDmode.
7762 Update calls to iv_analyze_op and iv_analyze_expr. Check
7763 is_a <scalar_int_mode> when changing the mode under consideration.
7764 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7765 Update call to iv_analyze_expr.
7766 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7767 inner register is not also a scalar_int_mode. Update call to
7769 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7770 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7771 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7772 separate mode class checks. Update calls to iv_analyze. Remove
7773 fix-up of VOIDmodes after iv_analyze_biv.
7774 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7775 don't have a scalar_int_mode. Update call to biv_p.
7777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7778 Alan Hayward <alan.hayward@arm.com>
7779 David Sherwood <david.sherwood@arm.com>
7781 * cfgexpand.c (convert_debug_memory_address): Use
7782 as_a <scalar_int_mode>.
7783 * combine.c (expand_compound_operation): Likewise.
7784 (make_extraction): Likewise.
7785 (change_zero_ext): Likewise.
7786 (simplify_comparison): Likewise.
7787 * cse.c (cse_insn): Likewise.
7788 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7789 (mem_loc_descriptor): Likewise.
7790 (loc_descriptor): Likewise.
7791 * expmed.c (init_expmed_one_mode): Likewise.
7792 (synth_mult): Likewise.
7793 (emit_store_flag_1): Likewise.
7794 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7795 of a comparison with size.
7796 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7797 (reduce_to_bit_field_precision): Likewise.
7798 * function.c (expand_function_end): Likewise.
7799 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7800 * loop-doloop.c (doloop_modify): Likewise.
7801 * optabs.c (expand_binop): Likewise.
7802 (expand_unop): Likewise.
7803 (expand_copysign_absneg): Likewise.
7804 (prepare_cmp_insn): Likewise.
7805 (maybe_legitimize_operand): Likewise.
7806 * recog.c (const_scalar_int_operand): Likewise.
7807 * rtlanal.c (get_address_mode): Likewise.
7808 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7809 (simplify_cond_clz_ctz): Likewise.
7810 * tree-nested.c (get_nl_goto_field): Likewise.
7811 * tree.c (build_vector_type_for_mode): Likewise.
7812 * var-tracking.c (use_narrower_mode): Likewise.
7814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7815 Alan Hayward <alan.hayward@arm.com>
7816 David Sherwood <david.sherwood@arm.com>
7818 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7819 * builtins.c (expand_builtin_signbit): Use it.
7820 * cfgexpand.c (expand_debug_expr): Likewise.
7821 * dojump.c (do_jump): Likewise.
7822 (do_compare_and_jump): Likewise.
7823 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7824 * expmed.c (make_tree): Likewise.
7825 * expr.c (expand_expr_real_2): Likewise.
7826 (expand_expr_real_1): Likewise.
7827 (try_casesi): Likewise.
7828 * fold-const-call.c (fold_const_call_ss): Likewise.
7829 * fold-const.c (unextend): Likewise.
7830 (extract_muldiv_1): Likewise.
7831 (fold_single_bit_test): Likewise.
7832 (native_encode_int): Likewise.
7833 (native_encode_string): Likewise.
7834 (native_interpret_int): Likewise.
7835 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7836 * internal-fn.c (expand_addsub_overflow): Likewise.
7837 (expand_neg_overflow): Likewise.
7838 (expand_mul_overflow): Likewise.
7839 (expand_arith_overflow): Likewise.
7840 * match.pd: Likewise.
7841 * stor-layout.c (layout_type): Likewise.
7842 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7843 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7844 * tree-ssanames.c (get_range_info): Likewise.
7845 * tree-switch-conversion.c (array_value_type) Likewise.
7846 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7847 (vect_recog_divmod_pattern): Likewise.
7848 (vect_recog_mixed_size_cond_pattern): Likewise.
7849 * tree-vrp.c (extract_range_basic): Likewise.
7850 (simplify_float_conversion_using_ranges): Likewise.
7851 * tree.c (int_fits_type_p): Likewise.
7852 * ubsan.c (instrument_bool_enum_load): Likewise.
7853 * varasm.c (mergeable_string_section): Likewise.
7854 (narrowing_initializer_constant_valid_p): Likewise.
7855 (output_constant): Likewise.
7857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7858 Alan Hayward <alan.hayward@arm.com>
7859 David Sherwood <david.sherwood@arm.com>
7861 * machmode.h (NARROWEST_INT_MODE): New macro.
7862 * expr.c (alignment_for_piecewise_move): Use it instead of
7863 GET_CLASS_NARROWEST_MODE (MODE_INT).
7864 (push_block): Likewise.
7865 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7867 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7870 Alan Hayward <alan.hayward@arm.com>
7871 David Sherwood <david.sherwood@arm.com>
7873 * postreload.c (move2add_valid_value_p): Change the type of the
7874 mode parameter to scalar_int_mode.
7875 (move2add_use_add2_insn): Add a mode parameter and use it instead
7877 (move2add_use_add3_insn): Likewise.
7878 (reload_cse_move2add): Update accordingly.
7880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7881 Alan Hayward <alan.hayward@arm.com>
7882 David Sherwood <david.sherwood@arm.com>
7884 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7886 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7887 * optabs.c (expand_unop): Likewise.
7889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7890 Alan Hayward <alan.hayward@arm.com>
7891 David Sherwood <david.sherwood@arm.com>
7893 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7894 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7895 (popcount_loc_descriptor): Likewise.
7896 (bswap_loc_descriptor): Likewise.
7897 (rotate_loc_descriptor): Likewise.
7898 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7899 calling the functions above.
7901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7902 Alan Hayward <alan.hayward@arm.com>
7903 David Sherwood <david.sherwood@arm.com>
7905 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7907 (try_combine): Likewise.
7908 (simplify_if_then_else): Likewise.
7909 * cse.c (cse_insn): Likewise.
7910 * dwarf2out.c (mem_loc_descriptor): Likewise.
7911 * emit-rtl.c (gen_lowpart_common): Likewise.
7912 * simplify-rtx.c (simplify_truncation): Likewise.
7913 (simplify_binary_operation_1): Likewise.
7914 (simplify_const_relational_operation): Likewise.
7915 (simplify_ternary_operation): Likewise.
7916 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7919 Alan Hayward <alan.hayward@arm.com>
7920 David Sherwood <david.sherwood@arm.com>
7922 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7923 * reload.c (push_reload): Likewise.
7924 (find_reloads): Likewise.
7926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7927 Alan Hayward <alan.hayward@arm.com>
7928 David Sherwood <david.sherwood@arm.com>
7930 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7931 (make_compound_operation_int): Likewise.
7932 (change_zero_ext): Likewise.
7933 * expr.c (convert_move): Likewise.
7934 (convert_modes): Likewise.
7935 * fwprop.c (forward_propagate_subreg): Likewise.
7936 * loop-iv.c (get_biv_step_1): Likewise.
7937 * optabs.c (widen_operand): Likewise.
7938 * postreload.c (move2add_valid_value_p): Likewise.
7939 * recog.c (simplify_while_replacing): Likewise.
7940 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7941 (simplify_binary_operation_1): Likewise. Remove redundant
7942 mode equality check.
7944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7945 Alan Hayward <alan.hayward@arm.com>
7946 David Sherwood <david.sherwood@arm.com>
7948 * combine.c (combine_simplify_rtx): Add checks for
7949 is_a <scalar_int_mode>.
7950 (simplify_if_then_else): Likewise.
7951 (make_field_assignment): Likewise.
7952 (simplify_comparison): Likewise.
7953 * ifcvt.c (noce_try_bitop): Likewise.
7954 * loop-invariant.c (canonicalize_address_mult): Likewise.
7955 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7958 Alan Hayward <alan.hayward@arm.com>
7959 David Sherwood <david.sherwood@arm.com>
7961 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7962 is_a <scalar_int_mode> instead of != BLKmode.
7964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7965 Alan Hayward <alan.hayward@arm.com>
7966 David Sherwood <david.sherwood@arm.com>
7968 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7969 instead of != VOIDmode.
7970 * combine.c (if_then_else_cond): Likewise.
7971 (change_zero_ext): Likewise.
7972 * dwarf2out.c (mem_loc_descriptor): Likewise.
7973 (loc_descriptor): Likewise.
7974 * rtlanal.c (canonicalize_condition): Likewise.
7975 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7978 Alan Hayward <alan.hayward@arm.com>
7979 David Sherwood <david.sherwood@arm.com>
7981 * simplify-rtx.c (simplify_binary_operation_1): Use
7982 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7985 Alan Hayward <alan.hayward@arm.com>
7986 David Sherwood <david.sherwood@arm.com>
7988 * wide-int.h (int_traits<unsigned char>) New class.
7989 (int_traits<unsigned short>) Likewise.
7990 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7991 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7993 * combine.c (set_nonzero_bits_and_sign_copies): Use
7994 is_a <scalar_int_mode>.
7995 (find_split_point): Likewise.
7996 (combine_simplify_rtx): Likewise.
7997 (simplify_logical): Likewise.
7998 (expand_compound_operation): Likewise.
7999 (expand_field_assignment): Likewise.
8000 (make_compound_operation): Likewise.
8001 (extended_count): Likewise.
8002 (change_zero_ext): Likewise.
8003 (simplify_comparison): Likewise.
8004 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8005 (ucompare_loc_descriptor): Likewise.
8006 (minmax_loc_descriptor): Likewise.
8007 (mem_loc_descriptor): Likewise.
8008 (loc_descriptor): Likewise.
8009 * expmed.c (init_expmed_one_mode): Likewise.
8010 * lra-constraints.c (lra_constraint_offset): Likewise.
8011 * optabs.c (prepare_libcall_arg): Likewise.
8012 * postreload.c (move2add_note_store): Likewise.
8013 * reload.c (operands_match_p): Likewise.
8014 * rtl.h (load_extend_op): Likewise.
8015 * rtlhooks.c (gen_lowpart_general): Likewise.
8016 * simplify-rtx.c (simplify_truncation): Likewise.
8017 (simplify_unary_operation_1): Likewise.
8018 (simplify_binary_operation_1): Likewise.
8019 (simplify_const_binary_operation): Likewise.
8020 (simplify_const_relational_operation): Likewise.
8021 (simplify_subreg): Likewise.
8022 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8023 * var-tracking.c (adjust_mems): Likewise.
8024 (prepare_call_arguments): Likewise.
8026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8027 Alan Hayward <alan.hayward@arm.com>
8028 David Sherwood <david.sherwood@arm.com>
8030 * machmode.h (is_int_mode): New fuction.
8031 * combine.c (find_split_point): Use it.
8032 (combine_simplify_rtx): Likewise.
8033 (simplify_if_then_else): Likewise.
8034 (simplify_set): Likewise.
8035 (simplify_shift_const_1): Likewise.
8036 (simplify_comparison): Likewise.
8037 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8038 * cse.c (notreg_cost): Likewise.
8039 (cse_insn): Likewise.
8040 * cselib.c (cselib_lookup_1): Likewise.
8041 * dojump.c (do_jump_1): Likewise.
8042 (do_compare_rtx_and_jump): Likewise.
8043 * dse.c (get_call_args): Likewise.
8044 * dwarf2out.c (rtl_for_decl_init): Likewise.
8045 (native_encode_initializer): Likewise.
8046 * expmed.c (emit_store_flag_1): Likewise.
8047 (emit_store_flag): Likewise.
8048 * expr.c (convert_modes): Likewise.
8049 (store_field): Likewise.
8050 (expand_expr_real_1): Likewise.
8051 * fold-const.c (fold_read_from_constant_string): Likewise.
8052 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8053 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8054 * optabs.c (expand_binop): Likewise.
8055 (expand_unop): Likewise.
8056 (expand_abs_nojump): Likewise.
8057 (expand_one_cmpl_abs_nojump): Likewise.
8058 * simplify-rtx.c (mode_signbit_p): Likewise.
8059 (val_signbit_p): Likewise.
8060 (val_signbit_known_set_p): Likewise.
8061 (val_signbit_known_clear_p): Likewise.
8062 (simplify_relational_operation_1): Likewise.
8063 * tree.c (vector_type_mode): Likewise.
8065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8066 Alan Hayward <alan.hayward@arm.com>
8067 David Sherwood <david.sherwood@arm.com>
8069 * machmode.h (smallest_mode_for_size): Fix formatting.
8070 (smallest_int_mode_for_size): New function.
8071 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8072 instead of smallest_mode_for_size.
8073 * combine.c (make_extraction): Likewise.
8074 * config/arc/arc.c (arc_expand_movmem): Likewise.
8075 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8076 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8077 * config/s390/s390.c (s390_expand_insv): Likewise.
8078 * config/sparc/sparc.c (assign_int_registers): Likewise.
8079 * config/spu/spu.c (spu_function_value): Likewise.
8080 (spu_function_arg): Likewise.
8081 * coverage.c (get_gcov_type): Likewise.
8082 (get_gcov_unsigned_t): Likewise.
8083 * dse.c (find_shift_sequence): Likewise.
8084 * expmed.c (store_bit_field_1): Likewise.
8085 * expr.c (convert_move): Likewise.
8086 (store_field): Likewise.
8087 * internal-fn.c (expand_arith_overflow): Likewise.
8088 * optabs-query.c (get_best_extraction_insn): Likewise.
8089 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8090 * stor-layout.c (layout_type): Likewise.
8091 (initialize_sizetypes): Likewise.
8092 * targhooks.c (default_get_mask_mode): Likewise.
8093 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8096 Alan Hayward <alan.hayward@arm.com>
8097 David Sherwood <david.sherwood@arm.com>
8099 * machmode.h (opt_mode::else_blk): New function.
8100 (int_mode_for_mode): Declare.
8101 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8102 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8104 * cfgexpand.c (expand_debug_expr): Likewise.
8105 * combine.c (gen_lowpart_or_truncate): Likewise.
8106 (gen_lowpart_for_combine): Likewise.
8107 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8108 * config/avr/avr.c (avr_to_int_mode): Likewise.
8109 (avr_out_plus_1): Likewise.
8110 (avr_out_plus): Likewise.
8111 (avr_out_round): Likewise.
8112 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8113 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8114 (s390_expand_vcond): Likewise.
8115 * config/spu/spu.c (spu_split_immediate): Likewise.
8116 (spu_expand_mov): Likewise.
8117 * dse.c (get_stored_val): Likewise.
8118 * expmed.c (store_bit_field_1): Likewise.
8119 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8121 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8122 (extract_low_bits): Likewise.
8123 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8124 handling rather than repeating the check.
8125 (emit_group_store): Likewise.
8126 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8127 * optabs.c (expand_absneg_bit): Likewise.
8128 (expand_copysign_absneg): Likewise.
8129 (expand_copysign_bit): Likewise.
8130 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8131 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8132 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8133 * var-tracking.c (prepare_call_arguments): Likewise.
8134 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8135 int_mode_for_mode instead of mode_for_size.
8136 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8139 Alan Hayward <alan.hayward@arm.com>
8140 David Sherwood <david.sherwood@arm.com>
8142 * machmode.h (int_mode_for_size): New function.
8143 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8144 instead of mode_for_size.
8145 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8147 * combine.c (expand_field_assignment): Use int_mode_for_size
8148 instead of mode_for_size.
8149 (make_extraction): Likewise.
8150 (simplify_shift_const_1): Likewise.
8151 (simplify_comparison): Likewise.
8152 * dojump.c (do_jump): Likewise.
8153 * dwarf2out.c (mem_loc_descriptor): Likewise.
8154 * emit-rtl.c (init_derived_machine_modes): Likewise.
8155 * expmed.c (flip_storage_order): Likewise.
8156 (convert_extracted_bit_field): Likewise.
8157 * expr.c (copy_blkmode_from_reg): Likewise.
8158 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8159 * internal-fn.c (expand_mul_overflow): Likewise.
8160 * lower-subreg.c (simple_move): Likewise.
8161 * optabs-libfuncs.c (init_optabs): Likewise.
8162 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8163 * tree.c (vector_type_mode): Likewise.
8164 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8165 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8166 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8167 * tree-vect-stmts.c (vectorizable_load): Likewise.
8168 (vectorizable_store): Likewise.
8170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8171 Alan Hayward <alan.hayward@arm.com>
8172 David Sherwood <david.sherwood@arm.com>
8174 * coretypes.h (pod_mode): New type.
8175 (scalar_int_mode_pod): New typedef.
8176 * machmode.h (pod_mode): New class.
8177 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8178 * genmodes.c (emit_mode_int_n): Update accordingly.
8179 * lower-subreg.h (target_lower_subreg): Change type to
8180 scalar_int_mode_pod.
8181 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8182 scalar_int_mode_pod.
8184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8185 Alan Hayward <alan.hayward@arm.com>
8186 David Sherwood <david.sherwood@arm.com>
8188 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8189 machine_mode to scalar_int_mode.
8190 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8191 (rs6000_option_override_internal): Remove cast to int.
8192 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8193 machine_mode to scalar_int_mode.
8194 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8195 (rs6000_option_override_internal): Remove cast to int.
8196 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8197 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8199 * config/s390/s390.c (s390_expand_builtin): Likewise.
8200 * coretypes.h (scalar_int_mode): New type.
8201 (opt_scalar_int_mode): New typedef.
8202 * machmode.h (scalar_int_mode): New class.
8203 (scalar_int_mode::includes_p): New function.
8204 (byte_mode): Change type to scalar_int_mode.
8205 (word_mode): Likewise.
8206 (ptr_mode): Likewise.
8207 * emit-rtl.c (byte_mode): Likewise.
8208 (word_mode): Likewise.
8209 (ptr_mode): Likewise.
8210 (init_derived_machine_modes): Update accordingly.
8211 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8212 and MODE_PARTIAL_INT.
8213 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8214 opt_scalar_int_mode.
8216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8217 Alan Hayward <alan.hayward@arm.com>
8218 David Sherwood <david.sherwood@arm.com>
8220 * target.def (libgcc_floating_mode_supported_p): Take a
8222 * doc/tm.texi: Regenerate.
8223 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8225 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8226 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8230 Alan Hayward <alan.hayward@arm.com>
8231 David Sherwood <david.sherwood@arm.com>
8233 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8234 * doc/tm.texi: Regenerate.
8235 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8236 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8237 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8238 * targhooks.h (default_floatn_mode): Likewise.
8239 * targhooks.c (default_floatn_mode): Likewise.
8240 * tree.c (build_common_tree_nodes): Update accordingly.
8242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8243 Alan Hayward <alan.hayward@arm.com>
8244 David Sherwood <david.sherwood@arm.com>
8246 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8247 (mode_iterator::iterate_p): Likewise.
8248 (mode_iterator::get_wider): Likewise.
8249 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * coretypes.h (opt_scalar_float_mode): New typedef.
8256 * machmode.h (float_mode_for_size): New function.
8257 * emit-rtl.c (double_mode): Delete.
8258 (init_emit_once): Use float_mode_for_size.
8259 * stor-layout.c (layout_type): Likewise.
8260 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8263 Alan Hayward <alan.hayward@arm.com>
8264 David Sherwood <david.sherwood@arm.com>
8266 * output.h (assemble_real): Take a scalar_float_mode.
8267 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8268 * config/arm/arm.md (consttable_4): Likewise.
8269 (consttable_8): Likewise.
8270 (consttable_16): Likewise.
8271 * config/mips/mips.md (consttable_float): Likewise.
8272 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8273 * varasm.c (assemble_real): Take a scalar_float_mode.
8274 (output_constant_pool_2): Update accordingly.
8275 (output_constant): Likewise.
8277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8278 Alan Hayward <alan.hayward@arm.com>
8279 David Sherwood <david.sherwood@arm.com>
8281 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8282 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8283 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8284 (native_encode_real): Likewise.
8285 (native_interpret_real): Likewise.
8286 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8287 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8290 Alan Hayward <alan.hayward@arm.com>
8291 David Sherwood <david.sherwood@arm.com>
8293 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8294 <scalar_float_mode>. Simplify.
8295 (gen_extend_conv_libfunc): Likewise.
8297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8298 Alan Hayward <alan.hayward@arm.com>
8299 David Sherwood <david.sherwood@arm.com>
8301 * coretypes.h (scalar_float_mode): New type.
8302 * machmode.h (mode_traits::from_int): Use machine_mode if
8303 USE_ENUM_MODES is defined.
8304 (is_a): New function.
8306 (dyn_cast): Likewise.
8307 (scalar_float_mode): New class.
8308 (scalar_float_mode::includes_p): New function.
8309 (is_float_mode): Likewise.
8310 * gdbhooks.py (MachineModePrinter): New class.
8311 (build_pretty_printer): Use it for scalar_float_mode.
8312 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8313 (format_helper::format_helper): Turn into a template.
8314 * genmodes.c (get_mode_class): New function.
8315 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8316 or machine_mode if none.
8317 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8318 as_a <scalar_float_mode>.
8319 * dwarf2out.c (mem_loc_descriptor): Likewise.
8320 (insert_float): Likewise.
8321 (add_const_value_attribute): Likewise.
8322 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8323 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8324 (expand_unop): Update accordingly.
8325 (expand_abs_nojump): Likewise.
8326 (expand_copysign_absneg): Take a scalar_float_mode.
8327 (expand_copysign_bit): Likewise.
8328 (expand_copysign): Update accordingly.
8330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * coretypes.h (opt_mode): New class.
8335 * machmode.h (opt_mode): Likewise.
8336 (opt_mode::else_void): New function.
8337 (opt_mode::require): Likewise.
8338 (opt_mode::exists): Likewise.
8339 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8340 (GET_MODE_2XWIDER_MODE): Likewise.
8341 (mode_iterator::get_wider): Update accordingly.
8342 (mode_iterator::get_2xwider): Likewise.
8343 (mode_iterator::get_known_wider): Likewise, turning into a template.
8344 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8345 forcing a wider mode to exist.
8346 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8347 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8348 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8349 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8350 * lower-subreg.c (init_lower_subreg): Likewise.
8351 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8352 on the final iteration.
8353 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8354 a wider mode exists before asking for a move pattern.
8355 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8356 forcing a wider mode to exist.
8357 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8358 returning false if no such mode exists.
8359 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8360 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8361 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8362 Avoid checking for a MODE_INT if we already know the mode is not a
8364 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8365 forcing a wider mode to exist.
8366 (expmed_mult_highpart_optab): Likewise.
8367 (expmed_mult_highpart): Likewise.
8368 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8370 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8371 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8372 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8373 * internal-fn.c (expand_mul_overflow): Update use of
8374 GET_MODE_2XWIDER_MODE.
8375 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8376 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8377 GET_MODE_WIDER_MODE.
8378 (convert_plusminus_to_widen): Likewise.
8379 * tree-switch-conversion.c (array_value_type): Likewise.
8380 * var-tracking.c (emit_note_insn_var_location): Likewise.
8381 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8382 Return false inside rather than outside the loop if no wider mode
8384 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8385 and GET_MODE_2XWIDER_MODE
8386 (can_compare_p): Use else_void.
8387 * gdbhooks.py (OptMachineModePrinter): New class.
8388 (build_pretty_printer): Use it for opt_mode.
8390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8391 Alan Hayward <alan.hayward@arm.com>
8392 David Sherwood <david.sherwood@arm.com>
8394 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8395 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8398 Alan Hayward <alan.hayward@arm.com>
8399 David Sherwood <david.sherwood@arm.com>
8401 * machmode.h (mode_traits): New structure.
8402 (get_narrowest_mode): New function.
8403 (mode_iterator::start): Likewise.
8404 (mode_iterator::iterate_p): Likewise.
8405 (mode_iterator::get_wider): Likewise.
8406 (mode_iterator::get_known_wider): Likewise.
8407 (mode_iterator::get_2xwider): Likewise.
8408 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8409 (FOR_EACH_MODE): Likewise.
8410 (FOR_EACH_MODE_FROM): Likewise.
8411 (FOR_EACH_MODE_UNTIL): Likewise.
8412 (FOR_EACH_WIDER_MODE): Likewise.
8413 (FOR_EACH_2XWIDER_MODE): Likewise.
8414 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8415 * combine.c (simplify_comparison): Likewise
8416 * config/i386/i386.c (type_natural_mode): Likewise.
8417 * cse.c (cse_insn): Likewise.
8418 * dse.c (find_shift_sequence): Likewise.
8419 * emit-rtl.c (init_derived_machine_modes): Likewise.
8420 (init_emit_once): Likewise.
8421 * explow.c (hard_function_value): Likewise.
8422 * expmed.c (extract_fixed_bit_field_1): Likewise.
8423 (extract_bit_field_1): Likewise.
8424 (expand_divmod): Likewise.
8425 (emit_store_flag_1): Likewise.
8426 * expr.c (init_expr_target): Likewise.
8427 (convert_move): Likewise.
8428 (alignment_for_piecewise_move): Likewise.
8429 (widest_int_mode_for_size): Likewise.
8430 (emit_block_move_via_movmem): Likewise.
8431 (copy_blkmode_to_reg): Likewise.
8432 (set_storage_via_setmem): Likewise.
8433 (compress_float_constant): Likewise.
8434 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8435 * optabs-query.c (get_best_extraction_insn): Likewise.
8436 * optabs.c (expand_binop): Likewise.
8437 (expand_twoval_unop): Likewise.
8438 (expand_twoval_binop): Likewise.
8439 (widen_leading): Likewise.
8440 (widen_bswap): Likewise.
8441 (expand_parity): Likewise.
8442 (expand_unop): Likewise.
8443 (prepare_cmp_insn): Likewise.
8444 (prepare_float_lib_cmp): Likewise.
8445 (expand_float): Likewise.
8446 (expand_fix): Likewise.
8447 (expand_sfix_optab): Likewise.
8448 * postreload.c (move2add_use_add2_insn): Likewise.
8449 * reg-stack.c (reg_to_stack): Likewise.
8450 * reginfo.c (choose_hard_reg_mode): Likewise.
8451 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8452 * stor-layout.c (mode_for_size): Likewise.
8453 (smallest_mode_for_size): Likewise.
8454 (mode_for_vector): Likewise.
8455 (finish_bitfield_representative): Likewise.
8456 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8457 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8458 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8459 * var-tracking.c (prepare_call_arguments): Likewise.
8461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8462 Alan Hayward <alan.hayward@arm.com>
8463 David Sherwood <david.sherwood@arm.com>
8465 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8466 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8467 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8468 * machmode.h (mode_size): Move earlier in file.
8469 (mode_precision): Likewise.
8470 (mode_inner): Likewise.
8471 (mode_nunits): Likewise.
8472 (mode_unit_size): Likewise.
8473 (unit_unit_precision): Likewise.
8474 (mode_wider): Likewise.
8475 (mode_2xwider): Likewise.
8476 (machine_mode): New class.
8477 (mode_to_bytes): New function.
8478 (mode_to_bits): Likewise.
8479 (mode_to_precision): Likewise.
8480 (mode_to_inner): Likewise.
8481 (mode_to_unit_size): Likewise.
8482 (mode_to_unit_precision): Likewise.
8483 (mode_to_nunits): Likewise.
8484 (GET_MODE_SIZE): Use mode_to_bytes.
8485 (GET_MODE_BITSIZE): Use mode_to_bits.
8486 (GET_MODE_PRECISION): Use mode_to_precision.
8487 (GET_MODE_INNER): Use mode_to_inner.
8488 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8489 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8490 (GET_MODE_NUNITS): Use mode_to_nunits.
8491 * system.h (ALWAYS_INLINE): New macro.
8492 * config/powerpcspe/powerpcspe-c.c
8493 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8494 int for arg1_mode and arg2_mode.
8496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8497 Alan Hayward <alan.hayward@arm.com>
8498 David Sherwood <david.sherwood@arm.com>
8500 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8501 Prefix mode names with E_ in case statements.
8502 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8503 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8504 (aarch64_split_simd_move): Likewise.
8505 (aarch64_gen_storewb_pair): Likewise.
8506 (aarch64_gen_loadwb_pair): Likewise.
8507 (aarch64_gen_store_pair): Likewise.
8508 (aarch64_gen_load_pair): Likewise.
8509 (aarch64_get_condition_code_1): Likewise.
8510 (aarch64_constant_pool_reload_icode): Likewise.
8511 (get_rsqrte_type): Likewise.
8512 (get_rsqrts_type): Likewise.
8513 (get_recpe_type): Likewise.
8514 (get_recps_type): Likewise.
8515 (aarch64_gimplify_va_arg_expr): Likewise.
8516 (aarch64_simd_container_mode): Likewise.
8517 (aarch64_emit_load_exclusive): Likewise.
8518 (aarch64_emit_store_exclusive): Likewise.
8519 (aarch64_expand_compare_and_swap): Likewise.
8520 (aarch64_gen_atomic_cas): Likewise.
8521 (aarch64_emit_bic): Likewise.
8522 (aarch64_emit_atomic_swap): Likewise.
8523 (aarch64_emit_atomic_load_op): Likewise.
8524 (aarch64_evpc_trn): Likewise.
8525 (aarch64_evpc_uzp): Likewise.
8526 (aarch64_evpc_zip): Likewise.
8527 (aarch64_evpc_ext): Likewise.
8528 (aarch64_evpc_rev): Likewise.
8529 (aarch64_evpc_dup): Likewise.
8530 (aarch64_gen_ccmp_first): Likewise.
8531 (aarch64_gen_ccmp_next): Likewise.
8532 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8533 (alpha_emit_xfloating_libcall): Likewise.
8534 (emit_insxl): Likewise.
8535 (alpha_arg_type): Likewise.
8536 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8537 (arc_preferred_simd_mode): Likewise.
8538 (arc_secondary_reload): Likewise.
8539 (get_arc_condition_code): Likewise.
8540 (arc_print_operand): Likewise.
8541 (arc_legitimate_constant_p): Likewise.
8542 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8543 * config/arc/arc.md (casesi_load): Likewise.
8544 (casesi_compact_jump): Likewise.
8545 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8546 (cc_use_register): Likewise.
8547 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8548 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8549 (arm_init_iwmmxt_builtins): Likewise.
8550 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8551 (neon_expand_vector_init): Likewise.
8552 (arm_attr_length_move_neon): Likewise.
8553 (maybe_get_arm_condition_code): Likewise.
8554 (arm_emit_vector_const): Likewise.
8555 (arm_preferred_simd_mode): Likewise.
8556 (arm_output_iwmmxt_tinsr): Likewise.
8557 (thumb1_output_casesi): Likewise.
8558 (thumb2_output_casesi): Likewise.
8559 (arm_emit_load_exclusive): Likewise.
8560 (arm_emit_store_exclusive): Likewise.
8561 (arm_expand_compare_and_swap): Likewise.
8562 (arm_evpc_neon_vuzp): Likewise.
8563 (arm_evpc_neon_vzip): Likewise.
8564 (arm_evpc_neon_vrev): Likewise.
8565 (arm_evpc_neon_vtrn): Likewise.
8566 (arm_evpc_neon_vext): Likewise.
8567 (arm_validize_comparison): Likewise.
8568 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8569 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8570 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8571 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8572 (c6x_preferred_simd_mode): Likewise.
8573 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8574 (epiphany_rtx_costs): Likewise.
8575 * config/epiphany/predicates.md (proper_comparison_operator):
8577 * config/frv/frv.c (condexec_memory_operand): Likewise.
8578 (frv_emit_move): Likewise.
8579 (output_move_single): Likewise.
8580 (output_condmove_single): Likewise.
8581 (frv_hard_regno_mode_ok): Likewise.
8582 (frv_matching_accg_mode): Likewise.
8583 * config/h8300/h8300.c (split_adds_subs): Likewise.
8584 (h8300_rtx_costs): Likewise.
8585 (h8300_print_operand): Likewise.
8586 (compute_mov_length): Likewise.
8587 (output_logical_op): Likewise.
8588 (compute_logical_op_length): Likewise.
8589 (compute_logical_op_cc): Likewise.
8590 (h8300_shift_needs_scratch_p): Likewise.
8591 (output_a_shift): Likewise.
8592 (compute_a_shift_length): Likewise.
8593 (compute_a_shift_cc): Likewise.
8594 (expand_a_rotate): Likewise.
8595 (output_a_rotate): Likewise.
8596 * config/i386/i386.c (classify_argument): Likewise.
8597 (function_arg_advance_32): Likewise.
8598 (function_arg_32): Likewise.
8599 (function_arg_64): Likewise.
8600 (function_value_64): Likewise.
8601 (ix86_gimplify_va_arg): Likewise.
8602 (ix86_legitimate_constant_p): Likewise.
8603 (put_condition_code): Likewise.
8604 (split_double_mode): Likewise.
8605 (ix86_avx256_split_vector_move_misalign): Likewise.
8606 (ix86_expand_vector_logical_operator): Likewise.
8607 (ix86_split_idivmod): Likewise.
8608 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8609 (ix86_build_const_vector): Likewise.
8610 (ix86_build_signbit_mask): Likewise.
8611 (ix86_match_ccmode): Likewise.
8612 (ix86_cc_modes_compatible): Likewise.
8613 (ix86_expand_branch): Likewise.
8614 (ix86_expand_sse_cmp): Likewise.
8615 (ix86_expand_sse_movcc): Likewise.
8616 (ix86_expand_int_sse_cmp): Likewise.
8617 (ix86_expand_vec_perm_vpermi2): Likewise.
8618 (ix86_expand_vec_perm): Likewise.
8619 (ix86_expand_sse_unpack): Likewise.
8620 (ix86_expand_int_addcc): Likewise.
8621 (ix86_split_to_parts): Likewise.
8622 (ix86_vectorize_builtin_gather): Likewise.
8623 (ix86_vectorize_builtin_scatter): Likewise.
8624 (avx_vpermilp_parallel): Likewise.
8625 (inline_memory_move_cost): Likewise.
8626 (ix86_tieable_integer_mode_p): Likewise.
8627 (x86_maybe_negate_const_int): Likewise.
8628 (ix86_expand_vector_init_duplicate): Likewise.
8629 (ix86_expand_vector_init_one_nonzero): Likewise.
8630 (ix86_expand_vector_init_one_var): Likewise.
8631 (ix86_expand_vector_init_concat): Likewise.
8632 (ix86_expand_vector_init_interleave): Likewise.
8633 (ix86_expand_vector_init_general): Likewise.
8634 (ix86_expand_vector_set): Likewise.
8635 (ix86_expand_vector_extract): Likewise.
8636 (emit_reduc_half): Likewise.
8637 (ix86_emit_i387_round): Likewise.
8638 (ix86_mangle_type): Likewise.
8639 (ix86_expand_round_sse4): Likewise.
8640 (expand_vec_perm_blend): Likewise.
8641 (canonicalize_vector_int_perm): Likewise.
8642 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8643 (expand_vec_perm_1): Likewise.
8644 (expand_vec_perm_interleave3): Likewise.
8645 (expand_vec_perm_even_odd_pack): Likewise.
8646 (expand_vec_perm_even_odd_1): Likewise.
8647 (expand_vec_perm_broadcast_1): Likewise.
8648 (ix86_vectorize_vec_perm_const_ok): Likewise.
8649 (ix86_expand_vecop_qihi): Likewise.
8650 (ix86_expand_mul_widen_hilo): Likewise.
8651 (ix86_expand_sse2_abs): Likewise.
8652 (ix86_expand_pextr): Likewise.
8653 (ix86_expand_pinsr): Likewise.
8654 (ix86_preferred_simd_mode): Likewise.
8655 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8656 * config/i386/sse.md (*andnot<mode>3): Likewise.
8657 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8658 (*<code><mode>3): Likewise.
8659 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8660 (ia64_expand_atomic_op): Likewise.
8661 (ia64_arg_type): Likewise.
8662 (ia64_mode_to_int): Likewise.
8663 (ia64_scalar_mode_supported_p): Likewise.
8664 (ia64_vector_mode_supported_p): Likewise.
8665 (expand_vec_perm_broadcast): Likewise.
8666 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8667 (iq2000_function_arg_advance): Likewise.
8668 (iq2000_function_arg): Likewise.
8669 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8670 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8671 (m68k_libcall_value): Likewise.
8672 (m68k_function_value): Likewise.
8673 (sched_attr_op_type): Likewise.
8674 * config/mcore/mcore.c (mcore_output_move): Likewise.
8675 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8677 (microblaze_function_arg): Likewise.
8678 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8679 (mips_print_operand): Likewise.
8680 (mips_mode_ok_for_mov_fmt_p): Likewise.
8681 (mips_vector_mode_supported_p): Likewise.
8682 (mips_preferred_simd_mode): Likewise.
8683 (mips_expand_vpc_loongson_even_odd): Likewise.
8684 (mips_expand_vec_unpack): Likewise.
8685 (mips_expand_vi_broadcast): Likewise.
8686 (mips_expand_vector_init): Likewise.
8687 (mips_expand_vec_reduc): Likewise.
8688 (mips_expand_msa_cmp): Likewise.
8689 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8690 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8691 (cc_flags_for_mode): Likewise.
8692 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8693 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8694 (nds32_output_casesi_pc_relative): Likewise.
8695 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8696 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8697 (nvptx_gen_unpack): Likewise.
8698 (nvptx_gen_pack): Likewise.
8699 (nvptx_gen_shuffle): Likewise.
8700 (nvptx_gen_wcast): Likewise.
8701 (nvptx_preferred_simd_mode): Likewise.
8702 * config/pa/pa.c (pa_secondary_reload): Likewise.
8703 * config/pa/predicates.md (base14_operand): Likewise.
8704 * config/powerpcspe/powerpcspe-c.c
8705 (altivec_resolve_overloaded_builtin): Likewise.
8706 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8708 (rs6000_preferred_simd_mode): Likewise.
8709 (output_vec_const_move): Likewise.
8710 (rs6000_expand_vector_extract): Likewise.
8711 (rs6000_split_vec_extract_var): Likewise.
8712 (reg_offset_addressing_ok_p): Likewise.
8713 (rs6000_legitimate_offset_address_p): Likewise.
8714 (rs6000_legitimize_address): Likewise.
8715 (rs6000_emit_set_const): Likewise.
8716 (rs6000_const_vec): Likewise.
8717 (rs6000_emit_move): Likewise.
8718 (spe_build_register_parallel): Likewise.
8719 (rs6000_darwin64_record_arg_recurse): Likewise.
8720 (swap_selector_for_mode): Likewise.
8721 (spe_init_builtins): Likewise.
8722 (paired_init_builtins): Likewise.
8723 (altivec_init_builtins): Likewise.
8724 (do_load_for_compare): Likewise.
8725 (rs6000_generate_compare): Likewise.
8726 (rs6000_expand_float128_convert): Likewise.
8727 (emit_load_locked): Likewise.
8728 (emit_store_conditional): Likewise.
8729 (rs6000_output_function_epilogue): Likewise.
8730 (rs6000_handle_altivec_attribute): Likewise.
8731 (rs6000_function_value): Likewise.
8732 (emit_fusion_gpr_load): Likewise.
8733 (emit_fusion_p9_load): Likewise.
8734 (emit_fusion_p9_store): Likewise.
8735 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8736 (fusion_gpr_mem_load): Likewise.
8737 (fusion_addis_mem_combo_load): Likewise.
8738 (fusion_addis_mem_combo_store): Likewise.
8739 * config/rs6000/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/rs6000-c.c (altivec_resolve_overloaded_builtin):
8745 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8746 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8747 (rs6000_preferred_simd_mode): Likewise.
8748 (output_vec_const_move): Likewise.
8749 (rs6000_expand_vector_extract): Likewise.
8750 (rs6000_split_vec_extract_var): Likewise.
8751 (reg_offset_addressing_ok_p): Likewise.
8752 (rs6000_legitimate_offset_address_p): Likewise.
8753 (rs6000_legitimize_address): Likewise.
8754 (rs6000_emit_set_const): Likewise.
8755 (rs6000_const_vec): Likewise.
8756 (rs6000_emit_move): Likewise.
8757 (rs6000_darwin64_record_arg_recurse): Likewise.
8758 (swap_selector_for_mode): Likewise.
8759 (paired_init_builtins): Likewise.
8760 (altivec_init_builtins): Likewise.
8761 (rs6000_expand_float128_convert): Likewise.
8762 (emit_load_locked): Likewise.
8763 (emit_store_conditional): Likewise.
8764 (rs6000_output_function_epilogue): Likewise.
8765 (rs6000_handle_altivec_attribute): Likewise.
8766 (rs6000_function_value): Likewise.
8767 (emit_fusion_gpr_load): Likewise.
8768 (emit_fusion_p9_load): Likewise.
8769 (emit_fusion_p9_store): Likewise.
8770 * config/rx/rx.c (rx_gen_move_template): Likewise.
8771 (flags_from_mode): Likewise.
8772 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8773 (s390_slb_comparison): Likewise.
8774 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8775 (s390_vector_mode_supported_p): Likewise.
8776 (s390_cc_modes_compatible): Likewise.
8777 (s390_match_ccmode_set): Likewise.
8778 (s390_canonicalize_comparison): Likewise.
8779 (s390_emit_compare_and_swap): Likewise.
8780 (s390_branch_condition_mask): Likewise.
8781 (s390_rtx_costs): Likewise.
8782 (s390_secondary_reload): Likewise.
8783 (__SECONDARY_RELOAD_CASE): Likewise.
8784 (s390_expand_cs): Likewise.
8785 (s390_preferred_simd_mode): Likewise.
8786 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8787 * config/sh/sh.c (sh_print_operand): Likewise.
8788 (dump_table): Likewise.
8789 (sh_secondary_reload): Likewise.
8790 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8791 * config/sh/sh.md (casesi_worker_1): Likewise.
8792 (casesi_worker_2): Likewise.
8793 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8794 (fcc_comparison_operator): Likewise.
8795 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8796 (emit_soft_tfmode_cvt): Likewise.
8797 (sparc_preferred_simd_mode): Likewise.
8798 (output_cbranch): Likewise.
8799 (sparc_print_operand): Likewise.
8800 (sparc_expand_vec_perm_bmask): Likewise.
8801 (vector_init_bshuffle): Likewise.
8802 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8803 (spu_vector_mode_supported_p): Likewise.
8804 (spu_expand_insv): Likewise.
8805 (spu_emit_branch_or_set): Likewise.
8806 (spu_handle_vector_attribute): Likewise.
8807 (spu_builtin_splats): Likewise.
8808 (spu_builtin_extract): Likewise.
8809 (spu_builtin_promote): Likewise.
8810 (spu_expand_sign_extend): Likewise.
8811 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8812 (tilegx_simd_int): Likewise.
8813 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8814 (tilepro_simd_int): Likewise.
8815 * config/v850/v850.c (const_double_split): Likewise.
8816 (v850_print_operand): Likewise.
8817 (ep_memory_offset): Likewise.
8818 * config/vax/vax.c (vax_rtx_costs): Likewise.
8819 (vax_output_int_move): Likewise.
8820 (vax_output_int_add): Likewise.
8821 (vax_output_int_subtract): Likewise.
8822 * config/visium/predicates.md (visium_branch_operator): Likewise.
8823 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8824 (visium_print_operand_address): Likewise.
8825 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8826 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8827 (xtensa_expand_conditional_branch): Likewise.
8828 (xtensa_copy_incoming_a7): Likewise.
8829 (xtensa_output_literal): Likewise.
8830 * dfp.c (decimal_real_maxval): Likewise.
8831 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8834 Alan Hayward <alan.hayward@arm.com>
8835 David Sherwood <david.sherwood@arm.com>
8837 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8838 (mode_nunits_inline): Likewise.
8839 (mode_inner_inline): Likewise.
8840 (mode_unit_size_inline): Likewise.
8841 (mode_unit_precision_inline): Likewise.
8842 (emit_insn_modes_h): Likewise. Also emit a #define of the
8844 (emit_mode_wider): Add an E_ prefix to mode names.
8845 (emit_mode_complex): Likewise.
8846 (emit_mode_inner): Likewise.
8847 (emit_mode_adjustments): Likewise.
8848 (emit_mode_int_n): Likewise.
8849 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8850 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8851 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8852 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8853 (CRC32_BUILTIN, ENTRY): Likewise.
8854 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8855 (aarch64_pop_regs): Likewise.
8856 (aarch64_process_components): Likewise.
8857 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8858 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8859 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8860 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8861 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8862 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8864 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8865 (pcmpistr): Likewise.
8866 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8867 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8868 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8870 * config/rl78/rl78.c (mduc_regs): Likewise.
8871 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8872 (htm_expand_builtin): Likewise.
8873 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8874 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8875 * config/xtensa/xtensa.c (print_operand): Likewise.
8876 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8877 (NUM_MODE_VECTOR_INT): Likewise.
8878 * genoutput.c (null_operand): Likewise.
8879 (output_operand_data): Likewise.
8880 * genrecog.c (print_parameter_value): Likewise.
8881 * lra.c (debug_operand_data): Likewise.
8883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8885 * dwarf2out.c (scompare_loc_descriptor_wide)
8886 (scompare_loc_descriptor_narrow): New functions, split out from...
8887 (scompare_loc_descriptor): ...here.
8888 * expmed.c (emit_store_flag_int): New function, split out from...
8889 (emit_store_flag): ...here.
8891 2017-08-30 Richard Biener <rguenther@suse.de>
8893 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8894 (dwarf2out_early_finish): Move setting of AT_pubnames from
8895 early debug output to early finish.
8897 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8899 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8900 and -mdata-region to the assembler.
8902 2017-08-30 Richard Biener <rguenther@suse.de>
8904 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8906 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8907 (dwarf2out_early_global_decl): Only generate a DIE for the
8908 abstract origin if it doesn't already exist or is a declaration DIE.
8909 (resolve_addr): Do not add the linkage name twice when
8910 generating a stub DIE for the DW_TAG_GNU_call_site target.
8912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8914 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8915 Use machine_mode rather than int for arg1_mode.
8917 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8920 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8921 that the second argument of the built-in functions to unpack
8922 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8923 switch statement instead a lot of if statements.
8924 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8925 Allow 64-bit values to be in Altivec registers as well as
8926 traditional floating point registers.
8927 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8929 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8931 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8932 MAX_REGS_PER_ADDRESS == 1.
8934 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8936 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8937 * config/i386/i386.c (ix86_option_override_internal): Simplify
8938 setting of opts->x_flag_entry.
8940 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8941 Jakub Jelinek <jakub@redhat.com>
8942 Richard Biener <rguenther@suse.de>
8944 PR tree-optimization/81503
8945 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8946 folded constant fits in the target type; reorder tests for clarity.
8948 2017-08-29 Martin Liska <mliska@suse.cz>
8950 * passes.def: Include pass_lower_switch.
8951 * stmt.c (dump_case_nodes): Remove and move to
8952 tree-switch-conversion.
8953 (case_values_threshold): Likewise.
8954 (expand_switch_as_decision_tree_p): Likewise.
8955 (emit_case_decision_tree): Likewise.
8956 (expand_case): Likewise.
8957 (balance_case_nodes): Likewise.
8958 (node_has_low_bound): Likewise.
8959 (node_has_high_bound): Likewise.
8960 (node_is_bounded): Likewise.
8961 (emit_case_nodes): Likewise.
8962 (struct simple_case_node): New struct.
8963 (add_case_node): Remove.
8964 (emit_case_dispatch_table): Use vector instead of case_list.
8965 (reset_out_edges_aux): Remove.
8966 (compute_cases_per_edge): Likewise.
8967 (expand_case): Build list of simple_case_node.
8968 (expand_sjlj_dispatch_table): Use it.
8969 * tree-switch-conversion.c (struct case_node): Moved from
8970 stmt.c and adjusted.
8971 (emit_case_nodes): Likewise.
8972 (node_has_low_bound): Likewise.
8973 (node_has_high_bound): Likewise.
8974 (node_is_bounded): Likewise.
8975 (case_values_threshold): Likewise.
8976 (reset_out_edges_aux): Likewise.
8977 (compute_cases_per_edge): Likewise.
8978 (add_case_node): Likewise.
8979 (dump_case_nodes): Likewise.
8980 (balance_case_nodes): Likewise.
8981 (expand_switch_as_decision_tree_p): Likewise.
8982 (emit_jump): Likewise.
8983 (emit_case_decision_tree): Likewise.
8984 (try_switch_expansion): Likewise.
8985 (do_jump_if_equal): Likewise.
8986 (emit_cmp_and_jump_insns): Likewise.
8987 (fix_phi_operands_for_edge): New function.
8988 (record_phi_operand_mapping): Likewise.
8989 (class pass_lower_switch): New pass.
8990 (pass_lower_switch::execute): New function.
8991 (make_pass_lower_switch): Likewise.
8992 (conditional_probability):
8993 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8994 * tree-pass.h: Add make_pass_lower_switch.
8996 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8999 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9002 2017-08-29 Richard Biener <rguenther@suse.de>
9004 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9005 we do not add a DW_AT_inline attribute twice.
9006 (gen_subprogram_die): Remove code setting DW_AT_inline on
9007 DECL_ABSTRACT_P nodes.
9009 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9011 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9012 calls to internal functions.
9013 (gimplify_modify_expr): Likewise.
9014 * tree-call-cdce.c (use_internal_fn): Likewise.
9015 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9016 (convert_to_divmod): Set the nothrow flag.
9017 * tree-if-conv.c (predicate_mem_writes): Likewise.
9018 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9019 (vectorizable_call): Likewise.
9020 (vectorizable_store): Likewise.
9021 (vectorizable_load): Likewise.
9022 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9023 (vect_recog_mask_conversion_pattern): Likewise.
9025 2017-08-29 Martin Liska <mliska@suse.cz>
9028 * gcc.c (driver_handle_option): Add new argument.
9029 * opts-common.c (handle_option): Pass
9030 target_option_override_hook.
9031 * opts-global.c (lang_handle_option): Add new option.
9032 (set_default_handlers): Add new argument.
9033 (decode_options): Likewise.
9034 * opts.c (target_handle_option): Likewise.
9035 (common_handle_option): Call target_option_override_hook.
9036 * opts.h (struct cl_option_handler_func): Add hook for
9037 target option override.
9038 (struct cl_option_handlers): Likewise.
9039 (set_default_handlers): Add new argument.
9040 (decode_options): Likewise.
9041 (common_handle_option): Likewise.
9042 (target_handle_option): Likewise.
9043 * toplev.c (toplev::main): Pass targetm.target_option.override
9046 2017-08-29 Richard Biener <rguenther@suse.de>
9047 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9049 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9050 life to the active subtree.
9052 2017-08-28 Jeff Law <law@redhat.com>
9054 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9055 derive_equivalences.
9056 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9058 (edge_info::derive_equivalences): New private member function
9060 * tree-ssa-dom.c (class edge_info): Changed from a struct
9061 to a class. Add ctor/dtor, methods and data members.
9062 (edge_info::edge_info): Renamed from allocate_edge_info.
9063 Initialize additional members.
9064 (edge_info::~edge_info): New.
9065 (free_dom_edge_info): Delete the edge info.
9066 (record_edge_info): Use new class & associated member functions.
9067 Tighten forms for testing for edge equivalences.
9068 (record_temporary_equivalences): Iterate over the simple
9069 equivalences rather than assuming there's only one per edge.
9070 (cprop_into_successor_phis): Iterate over the simple
9071 equivalences rather than assuming there's only one per edge.
9072 (optimize_stmt): Use operand_equal_p rather than pointer
9073 equality for mini-DSE code.
9075 2017-08-28 Nathan Sidwell <nathan@acm.org>
9077 * gcc.c (execute): Fold SIGPIPE handling into switch
9078 statement. Adjust internal error message.
9080 2017-08-28 Richard Biener <rguenther@suse.de>
9083 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9084 Do nothing for removed DIEs.
9086 2017-08-28 Richard Biener <rguenther@suse.de>
9088 PR tree-optimization/81977
9089 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9092 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9095 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9096 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9097 using targetm.gen_mem_thread_fence.
9099 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9102 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9103 predicate to register_operand. Reorder operands.
9104 (*btr<mode>): Ditto.
9105 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9106 (*btr<mode>_mask): Ditto.
9108 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9110 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9111 * config/rs6000/xmmintrin.h: New file.
9112 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9114 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9117 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9118 parameter and_insn and return it.
9119 (recombine_lvx_pattern): Insert a copy to ensure availability of
9120 the base register of the copied masking operation at the point of
9121 the instruction replacement.
9122 (recombine_stvx_pattern): Likewise.
9124 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9126 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9127 undocumented switches.
9128 (-mpower9-dform-vector): Likewise.
9129 (-mpower9-dform): Likewise.
9130 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9131 comments to delete references to -mpower9-dform* switches.
9132 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9133 Delete reference to -mpower9-dform* switches, test for
9134 -mpower9-vector instead.
9135 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9136 (OTHER_P9_VECTOR_MASKS): Likewise.
9137 (POWERPC_MASKS): Likewise.
9138 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9139 tests against -mpower9-dform* to -mpower9-vector. Delete code
9140 that checked for -mpower9-dform* consistancy with other options.
9141 Add test for -mpower9-misc to enable other power9 switches.
9142 (rs6000_init_hard_regno_mode_ok): Likewise.
9143 (rs6000_option_override_internal): Likewise.
9144 (rs6000_emit_prologue): Likewise.
9145 (rs6000_emit_epilogue): Likewise.
9146 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9147 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9148 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9150 (emit_fusion_p9_store): Likewise.
9151 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9152 resetting these macros if the assembler does not support ISA 3.0
9154 (TARGET_P9_DFORM_VECTOR): Likewise.
9155 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9156 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9158 2017-08-25 Alan Modra <amodra@gmail.com>
9160 PR rtl-optimization/81747
9161 * cse.c (cse_extended_basic_block): Don't attempt to record
9162 equivalences for degenerate conditional branches that branch
9163 to their fall-through.
9165 2017-08-24 Martin Sebor <msebor@redhat.com>
9168 * gimple-fold.c (size_must_be_zero_p): New function.
9169 (gimple_fold_builtin_memory_op): Call it.
9171 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9173 * config/rs6000/mm_malloc.h: New file.
9175 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9177 PR tree-optimization/81913
9178 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9179 analysis when either IVs in condition can wrap.
9181 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9183 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9184 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9186 2017-08-24 Richard Biener <rguenther@suse.de>
9189 * targhooks.c (default_target_can_inline_p): Properly
9190 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9191 is present and always compare.
9192 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9193 infer -mfpmath=sse from TARGET_SSE_P.
9194 (ix86_can_inline_p): Properly use target_option_default_node when
9195 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9197 2017-08-24 Richard Biener <rguenther@suse.de>
9200 * dwarf2out.c (output_die): Handle flag_generate_offload like
9202 (output_comp_unit): Likewise.
9203 (gen_array_type_die): Likewise.
9204 (dwarf2out_early_finish): Likewise.
9205 (note_variable_value_in_expr): Likewise.
9206 (dwarf2out_finish): Likewise. Adjust assert.
9207 * cgraphunit.c (symbol_table::compile): Move setting of
9208 flag_generate_offload earlier ...
9209 (symbol_table::finalize_compilation_unit): ... here, before
9210 early debug finalization.
9212 2017-08-24 Richard Biener <rguenther@suse.de>
9214 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9215 and ipa-fnsummary.h.
9216 (ix86_can_inline_p): When ix86_fpmath flags do not match
9217 check whether the callee uses FP math at all.
9219 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9222 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9223 instead of TYPE_PRECISION.
9225 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9226 Alan Hayward <alan.hayward@arm.com>
9227 David Sherwood <david.sherwood@arm.com>
9229 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9231 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9233 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9235 (ix86_frame::stack_realign_allocate): New field.
9236 (struct machine_frame_state): Modify comments.
9237 (machine_frame_state::sp_realigned_fp_end): New field.
9238 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9240 (sp_valid_at): Add assertion to assure no attempt to access invalid
9241 offset of a realigned stack.
9242 (fp_valid_at): Likewise.
9243 (choose_baseaddr): Modify comments.
9244 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9245 ix86_expand_prologue.
9246 (ix86_expand_prologue): Modify stack realignment and allocation.
9247 (ix86_expand_epilogue): Modify comments.
9248 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9249 avx2_runtime, avx512f, and avx512f_runtime.
9251 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9253 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9254 (mstackrealign): Do not init to -1.
9255 * config/i386/i386.c (ix86_option_override_internal):
9256 Check opts_set, not opts when setting default value of
9257 opts->x_ix86_force_align_arg_pointer.
9259 2017-08-23 Richard Biener <rguenther@suse.de>
9261 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9262 lang_hooks.decl_printable_name.
9263 * print-rtl-function.c (print_rtx_function): Likewise.
9264 * tree-pretty-print.c (dump_function_header): Likewise.
9266 2017-08-23 Richard Biener <rguenther@suse.de>
9269 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9270 -g0 at compile-time.
9272 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9275 * doc/sourcebuild.texi (Other hardware attributes):
9278 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9281 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9282 Fix single-use check.
9284 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9286 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9288 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9290 * doc/install.texi: Modify to add more details on running selected
9293 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9295 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9296 is combined with -mabi=ms.
9297 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9300 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9302 PR tree-optimization/81488
9303 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9304 and cached_basis fields.
9305 (MAX_SPREAD): New constant.
9306 (alloc_cand_and_find_basis): Initialize new fields.
9307 (clear_visited): New function.
9308 (create_phi_basis_1): Rename from create_phi_basis, set visited
9309 and cached_basis fields.
9310 (create_phi_basis): New wrapper function.
9311 (phi_add_costs_1): Rename from phi_add_costs, add spread
9312 parameter, set visited field, short-circuit when limits reached.
9313 (phi_add_costs): New wrapper function.
9314 (record_phi_increments_1): Rename from record_phi_increments, set
9316 (record_phi_increments): New wrapper function.
9317 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9318 (phi_incr_cost): New wrapper function.
9319 (all_phi_incrs_profitable_1): Rename from
9320 all_phi_incrs_profitable, set visited field.
9321 (all_phi_incrs_profitable): New wrapper function.
9323 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9324 Alan Hayward <alan.hayward@arm.com>
9325 David Sherwood <david.sherwood@arm.com>
9327 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9328 that takes the outer and inner modes.
9329 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9330 comparison as the canonical test for a paradoxical subreg.
9331 * combine.c (simplify_set): Use paradoxical_subreg_p.
9332 (make_extraction): Likewise.
9333 (force_to_mode): Likewise.
9334 (rtx_equal_for_field_assignment_p): Likewise.
9335 (gen_lowpart_for_combine): Likewise.
9336 (simplify_comparison): Likewise.
9337 * cse.c (equiv_constant): Likewise.
9338 * expmed.c (store_bit_field_1): Likewise.
9339 * final.c (alter_subreg): Likewise.
9340 * fwprop.c (propagate_rtx): Likewise.
9341 (forward_propagate_subreg): Likewise.
9342 * ira-conflicts.c (ira_build_conflicts): Likewise.
9343 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9344 * lra-constraints.c (curr_insn_transform): Likewise.
9345 (split_reg): Likewise.
9346 * lra-eliminations.c (move_plus_up): Likewise.
9347 (lra_eliminate_regs_1): Likewise.
9348 * recog.c (general_operand): Likewise.
9349 * ree.c (combine_reaching_defs): Likewise.
9350 * reload.c (push_reload): Likewise.
9351 (find_reloads): Likewise.
9352 * reload1.c (elimination_effects): Likewise.
9353 (compute_reload_subreg_offset): Likewise.
9354 (choose_reload_regs): Likewise.
9355 * rtlanal.c (subreg_lsb_1): Likewise.
9356 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9357 (simplify_subreg): Likewise.
9358 * var-tracking.c (track_loc_p): Likewise.
9359 * emit-rtl.c (byte_lowpart_offset): Likewise.
9360 (paradoxical_subreg_p): Delete out-of-line definition.
9362 2017-08-22 Jeff Law <law@redhat.com>
9364 PR tree-optimization/81741
9365 PR tree-optimization/71947
9366 * tree-ssa-dom.c: Include tree-inline.h.
9367 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9368 equivalences if one is more expensive to compute than the other.
9369 * tree-ssa-scopedtables.h (class const_or_copies): Make
9370 record_const_or_copy_raw method private.
9371 (class avail_exprs_stack): New method simplify_binary_operation.
9372 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9373 avail_exprs_stack::simplify_binary_operation as needed.
9374 (avail_exprs_stack::simplify_binary_operation): New function.
9376 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9378 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9379 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9380 (DOT_SYMBOLS): Likewise.
9381 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9382 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9383 (RS6000_ABI_NAME): Likewise.
9384 (TARGET_CMODEL): Likewise.
9385 (TOC_SECTION_ASM_OP): Likewise.
9386 (SET_CMODEL): New macro.
9387 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9389 2017-08-22 Richard Biener <rguenther@suse.de>
9391 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9392 to allow for free-lang-data replacements similar to verify_type_variant.
9394 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9395 Alan Hayward <alan.hayward@arm.com>
9396 David Sherwood <david.sherwood@arm.com>
9398 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9399 VOIDmode for the LABEL_REF.
9401 2017-08-22 Richard Biener <rguenther@suse.de>
9403 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9405 2017-08-22 Richard Biener <rguenther@suse.de>
9407 * common.opt (feliminate-dwarf2-dups): Ignore.
9408 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9409 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9410 same_die_p_wrap, compute_section_prefix,
9411 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9412 (comdat_symbol_id, comdat_symbol_number): Likewise.
9413 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9415 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9416 (output_die): Mark unreachable path unreachable.
9417 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9418 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9419 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9420 (dwarf2out_early_finish): Likewise.
9422 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9424 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9426 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9429 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9430 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9431 (avr_attribute_table) <io, io_low, address>: Initialize
9432 .decl_required with true.
9434 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9436 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9437 undocumented debugging options.
9438 (-mvsx-scalar-double): Likewise.
9439 (-mallow-df-permute): Likewise.
9440 (-mvectorize-builtins): Likewise.
9441 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9442 (rs6000_builtin_vectorized_function): Likewise.
9443 (rs6000_builtin_md_vectorized_function): Likewise.
9444 (rs6000_opt_vars): Likewise.
9446 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9449 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9450 (*btrq_imm): Rename from *btrq.
9451 (*btcq_imm): Rename from *btcq.
9452 (btsc): New code attribute.
9453 (*<btsc><mode>): New insn pattern.
9454 (*btr<mode>): Ditto.
9455 (*<btsc><mode>_mask): New insn_and_split pattern.
9456 (*btr<mode>_mask): Ditto.
9458 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9459 Alan Hayward <alan.hayward@arm.com>
9460 David Sherwood <david.sherwood@arm.com>
9462 * function.c (pad_below): Simplify padding calculation.
9464 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * target.def (function_prologue): Remove frame size argument.
9469 (function_epilogue): Likewise.
9470 * doc/tm.texi: Regenerate.
9471 * final.c (final_start_function): Update call to function_prologue.
9472 (final_end_function): Update call to function_epilogue.
9473 (default_function_pro_epilogue): Remove frame size argument.
9474 * output.h (default_function_pro_epilogue): Likewise.
9475 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9476 (arm_output_function_prologue): Likewise.
9477 * config/frv/frv.c (frv_function_prologue): Likewise.
9478 (frv_function_epilogue): Likewise.
9479 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9480 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9481 (ia64_output_function_epilogue): Likewise.
9482 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9483 (m32r_output_function_epilogue): Likewise.
9484 * config/microblaze/microblaze.c (microblaze_function_prologue)
9485 (microblaze_function_epilogue): Likewise.
9486 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9487 (mips_output_function_epilogue): Likewise.
9488 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9489 (mmix_target_asm_function_epilogue): Likewise.
9490 * config/msp430/msp430.c (msp430_start_function): Likewise.
9491 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9492 (nds32_asm_function_epilogue): Likewise.
9493 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9494 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9495 (pa_output_function_epilogue): Likewise.
9496 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9497 (rs6000_output_function_epilogue): Likewise.
9498 * config/rl78/rl78.c (rl78_start_function): Likewise.
9499 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9500 (rs6000_output_function_epilogue): Likewise.
9501 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9502 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9503 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9504 (sparc_asm_function_epilogue): Likewise.
9506 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9508 * tree.h (type_has_mode_precision_p): New function.
9509 * convert.c (convert_to_integer_1): Use it.
9510 * expr.c (expand_expr_real_2): Likewise.
9511 (expand_expr_real_1): Likewise.
9512 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9513 * match.pd: Likewise.
9514 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9515 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9516 * tree-tailcall.c (process_assignment): Likewise.
9517 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9518 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9519 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9520 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9521 (vectorizable_assignment): Likewise.
9522 (vectorizable_shift): Likewise.
9523 (vectorizable_operation): Likewise.
9524 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9526 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9528 * match.pd: Add pow (C, x) simplification.
9530 2017-08-21 Richard Biener <rguenther@suse.de>
9532 PR tree-optimization/81900
9533 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9534 for blocks with abnormal predecessors.
9535 (compute_antic): Do not set visited flag prematurely.
9537 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9540 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9542 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9544 * stor-layout.h (vector_type_mode): Move to...
9545 * tree.h (vector_type_mode): ...here.
9546 * stor-layout.c (vector_type_mode): Move to...
9547 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9549 2017-08-21 Richard Biener <rguenther@suse.de>
9551 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9552 register_external_die hooks.
9553 (debug_false_tree_charstarstar_uhwistar): Declare.
9554 (debug_nothing_tree_charstar_uhwi): Likewise.
9555 * debug.c (do_nothing_debug_hooks): Adjust.
9556 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9557 (debug_nothing_tree_charstar_uhwi): Likewise.
9558 * dbxout.c (dbx_debug_hooks): Adjust.
9559 (xcoff_debug_hooks): Likewise.
9560 * sdbout.c (sdb_debug_hooks): Likewise.
9561 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9562 * dwarf2out.c (macinfo_label_base): New global.
9563 (dwarf2out_register_external_die): New function for the
9564 register_external_die hook.
9565 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9566 (dwarf2_debug_hooks): Use them.
9567 (dwarf2_lineno_debug_hooks): Adjust.
9568 (struct die_struct): Add with_offset flag.
9569 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9570 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9571 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9572 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9573 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9574 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9575 defining section names for the early LTO debug variants.
9576 (reset_indirect_string): New helper.
9577 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9578 (print_dw_val): Add support for offsetted symbol references.
9579 (get_ultimate_context): Split out from is_cxx.
9580 (is_cxx): Use get_ultimate_context.
9581 (is_fortran): Add decl overload.
9582 (compute_comp_unit_symbol): Split out worker from
9583 compute_section_prefix.
9584 (compute_section_prefix): Call compute_comp_unit_symbol and
9585 set comdat_type_p here.
9586 (output_die): Skip DIE symbol output for the LTO added one.
9587 Handle DIE symbol references with offset.
9588 (output_comp_unit): Guard section name mangling properly.
9589 For LTO debug sections emit a symbol at the section beginning
9590 which we use to refer to its DIEs.
9591 (add_abstract_origin_attribute): For DIEs registered via
9592 dwarf2out_register_external_die directly refer to the early
9593 DIE rather than indirectly through the shadow one we created.
9594 Remove obsolete call to dwarf2out_abstract_function for
9595 non-function/block origins.
9596 (gen_array_type_die): When generating early LTO debug do
9597 not emit DW_AT_string_length.
9598 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9599 late when in LTO. As suggested place a gcc_unreachable for
9600 the DECL_ABSTRACT_P case.
9601 (gen_subprogram_die): Avoid another specification DIE
9602 for early built declarations/definitions for the late LTO case.
9603 (gen_variable_die): Add type references for late duplicated VLA dies
9605 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9606 we have the abstract instance already.
9607 (process_scope_var): Adjust decl DIE contexts in LTO which
9608 first puts them in limbo.
9609 (gen_decl_die): Do not generate type DIEs late apart from
9610 types for VLAs or for decls we do not yet have a DIE. Do not
9611 call dwarf2out_abstract_function late.
9612 (dwarf2out_early_global_decl): Make sure to create DIEs
9613 for abstract instances of a decl first.
9614 (dwarf2out_late_global_decl): Adjust comment.
9615 (output_macinfo_op): With multiple macro sections use
9616 macinfo_label_base to distinguish labels.
9617 (output_macinfo): Likewise. Update macinfo_label_base.
9618 Pass in the line info label.
9619 (note_variable_value_in_expr): When generating LTO resolve
9620 all variable values here by generating DIEs as needed.
9621 (init_sections_and_labels): Add early LTO debug flag parameter
9622 and generate different sections and names if set. Add generation
9623 counter for the labels so we can have multiple of them.
9624 (reset_dies): Helper to allow DIEs to be output multiple times.
9625 (dwarf2out_finish): When outputting DIEs to the fat part of an
9626 LTO object first reset DIEs.
9627 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9628 (modified_type_die): Check for decl_ultimate_origin being self
9630 (gen_type_die_with_usage): Likewise.
9631 (gen_typedef_die): Allow decl_ultimate_origin being self.
9632 (set_decl_abstract_flags): Remove.
9633 (set_block_abstract_flags): Likewise.
9634 (dwarf2out_abstract_function): Treat the early generated DIEs
9635 as the abstract copy and only add DW_AT_inline and
9636 DW_AT_artificial here and call set_decl_origin_self.
9637 If the DIE has an abstract origin don't do anything.
9638 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9639 if we have none yet (Go fails to build one, PR78628).
9640 (variably_modified_type_p): Prevent endless recursion for Ada
9641 cyclic pointer types.
9642 * lto-streamer-in.c: Include debug.h.
9643 (dref_queue): New global.
9644 (lto_read_tree_1): Stream in DIE references.
9645 (lto_input_tree): Register DIE references.
9646 (input_function): Stream DECL_DEBUG_ARGS.
9647 * lto-streamer-out.c: Include debug.h.
9648 (lto_write_tree_1): Output DIE references.
9649 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9650 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9651 (output_function): Stream DECL_DEBUG_ARGS.
9652 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9653 Stream DECL_ABSTRACT_ORIGIN.
9654 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9655 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9656 DECL_CONTEXT for file-scope decls.
9657 * lto-streamer.h (struct dref_entry): Declare.
9658 (dref_queue): Likewise.
9659 * cfgexpand.c (pass_expand::execute): Do not call the
9660 outlining_inline_function hook here.
9661 * lto-wrapper.c (debug_obj): New global.
9662 (tool_cleanup): Unlink it if required.
9663 (debug_objcopy): New function.
9664 (run_gcc): Handle early debug sections in the IL files by
9665 extracting them to separate files, partially linkin them and
9666 feeding the result back as result to the linker.
9667 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9668 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9669 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9670 sections into a separate segment.
9671 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9673 (darwin_asm_dwarf_section): Likewise.
9674 (darwin_asm_output_dwarf_offset): Likewise.
9675 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9677 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9678 Alan Hayward <alan.hayward@arm.com>
9679 David Sherwood <david.sherwood@arm.com>
9681 * read-md.h (md_reader::record_potential_iterator_use): Replace
9682 pointer argument with an rtx and an index.
9683 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9684 (apply_mode_iterator): Likewise.
9685 (apply_code_iterator): Likewise.
9686 (apply_int_iterator): Likewise.
9687 (apply_subst_iterator): Likewise.
9688 (record_iterator_use): Likewise.
9689 (record_attribute_use): Likewise.
9690 (md_reader::record_potential_iterator_use): Likewise. Update calls
9691 to record_iterator_use and apply_iterator.
9692 (iterator_use): Replace ptr with x and index.
9693 (attribute_use): Likewise.
9694 (apply_attribute_uses): Update calls to apply_iterator.
9695 (apply_iterators): Likewise. Update initialization of iterator_use.
9696 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9697 and record_potential_iterator_use.
9698 (rtx_reader::read_rtx_operand): Likewise.
9700 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9701 Alan Hayward <alan.hayward@arm.com>
9702 David Sherwood <david.sherwood@arm.com>
9704 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9707 2017-08-21 Richard Biener <rguenther@suse.de>
9710 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9711 at struct end conservatively when comparing common bases.
9713 2017-08-21 Richard Biener <rguenther@suse.de>
9715 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9716 (mem_ref_in_stmt): Remove.
9717 (determine_max_movement): Use ref index to get at the reference.
9718 (invariantness_dom_walker::before_dom_children): Deal with
9719 lim data already initialized.
9720 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9722 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9724 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9725 (TARGET_ISA_ROUND): Ditto.
9726 (TARGET_ROUND): Ditto.
9727 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9728 * config/i386/i386.md: Ditto.
9729 * config/i386/sse.md: Ditto.
9730 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9731 with OPTION_MASK_ISA_SSE4_1.
9733 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9736 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9737 __builtin_ia32_lzcnt_u16.
9739 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9742 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9743 (rs6000_set_current_function): Rewrite function to use it.
9745 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9748 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9749 and TYPE_WARN_IF_NOT_ALIGN.
9750 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9751 (handle_warn_if_not_align): New.
9752 (place_union_field): Call handle_warn_if_not_align.
9753 (place_field): Call handle_warn_if_not_align.
9754 Copy TYPE_WARN_IF_NOT_ALIGN.
9755 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9756 (layout_type): Likewise.
9757 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9759 (tree_decl_common): Add warn_if_not_align.
9760 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9761 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9762 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9763 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9764 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9765 * doc/extend.texi: Document warn_if_not_aligned attribute.
9766 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9768 2017-08-17 Martin Liska <mliska@suse.cz>
9771 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9772 (get_data_dependence): Use it as pointer type.
9773 (distribute_loop): Likewise.
9775 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9777 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9778 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9779 (p8_vmrgow_<mode>_direct): New define_insn.
9780 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9781 handle endianness for vmrgew and vmrgow permute patterns.
9783 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9785 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9786 * config/rs6000/rs6000-cpus.def: Remove comment.
9787 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9788 (POWERPC_MASKS): Likewise.
9789 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9790 use of TARGET_VSX_TIMODE.
9791 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9792 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9793 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9794 (rs6000_option_override_internal): Remove dead code.
9795 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9796 (rs6000_legitimize_reload_address): Likewise.
9797 (rs6000_legitimate_address_p): Likewise.
9798 (rs6000_opt_masks): Delete "vsx-timode".
9799 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9800 from function comment.
9801 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9802 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9803 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9805 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9806 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9807 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9810 (define_peephole2): Likewise.
9812 2017-08-17 Martin Sebor <msebor@redhat.com>
9815 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9816 past the end of an array.
9817 (test_pp_format): Add test cases.
9819 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9821 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9822 missing ECF_NOTHROW flags.
9824 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9827 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9828 operands residing in integer registers.
9829 (*vsx_le_perm_load_<mode>): Likewise.
9830 (*vsx_le_perm_store_<mode>): Likewise.
9831 (define_peephole2): Add peepholes to optimize the above.
9833 2017-08-17 Marek Polacek <polacek@redhat.com>
9836 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9837 to mimic what shorten_compare did. Change the return type to bool.
9838 (fold_cond_expr_with_comparison): Update call to
9839 operand_equal_for_comparison_p.
9840 (fold_ternary_loc): Likewise.
9842 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9844 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9846 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9848 2017-08-17 Richard Biener <rguenther@suse.de>
9850 * tree-ssa-structalias.c (solve_graph): When propagating
9851 to successors update the graphs succ edges and avoid duplicate work.
9853 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9856 * config/i386/i386.c (ix86_option_override_internal): Save target
9857 specific options after ix86_stack_protector_guard_reg was changed.
9859 2017-08-17 Richard Biener <rguenther@suse.de>
9861 PR tree-optimization/81827
9862 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9863 (new_var_info): Initialize it conservatively.
9864 (get_call_vi): Mark register vars.
9865 (new_scalar_tmp_constraint_exp): Likewise.
9866 (handle_rhs_call): Likewise.
9867 (handle_const_call): Likewise.
9868 (create_function_info_for): Likewise.
9869 (solve_constraints): Sort varinfos to separate register from
9870 non-register vars to pack points-to solution bitmaps during
9873 2017-08-17 Marek Polacek <polacek@redhat.com>
9875 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9877 2017-08-17 Richard Biener <rguenther@suse.de>
9879 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9880 to true when overflow is undefined and we saturated the result.
9882 2017-08-17 Alan Modra <amodra@gmail.com>
9885 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9886 Don't use store multiple if only one reg needs saving.
9887 (interesting_frame_related_regno): New function.
9888 (rs6000_frame_related): Don't emit frame info for regs that
9890 (rs6000_emit_epilogue): Likewise.
9892 2017-08-16 Nathan Sidwell <nathan@acm.org>
9894 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9895 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9896 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9897 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9898 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9899 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9900 (verify_type): Adjust for TYPE_BINFO move.
9901 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9902 process TYPE_BINFO directly.
9903 (hash_tree): Likewise.
9904 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9906 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9909 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9911 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9913 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9916 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9917 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9918 Add ix86_binary_operator_ok to insn constraint.
9919 (*iordi_1_bts): Ditto.
9920 (*xordi_1_btc): Ditto.
9921 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9922 Update corresponding peephole2 pattern.
9926 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9928 PR tree-optimization/81832
9929 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9930 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9932 2017-08-16 Marek Polacek <polacek@redhat.com>
9935 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9936 regarding size_zero_node.
9938 2017-08-16 Martin Liska <mliska@suse.cz>
9941 * config.gcc: Respect previously set extra_objs in case
9944 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9946 PR tree-optimization/81835
9947 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9948 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9949 not depend on the phi.
9951 2017-08-16 Alan Modra <amodra@gmail.com>
9953 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9956 2017-08-16 Alan Modra <amodra@gmail.com>
9958 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9959 (save_reg_p): ..into this. Update all callers.
9960 (first_reg_to_save): Simplify.
9962 2017-08-16 Alan Modra <amodra@gmail.com>
9964 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9967 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9971 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9972 constant count if that count is less than 32.
9974 2017-08-15 Nathan Sidwell <nathan@acm.org>
9976 * gcc.c (execute): Emit friendlier message if inferior is killed
9977 by an external cause.
9979 2017-08-15 Richard Biener <rguenther@suse.de>
9981 PR tree-optimization/81790
9982 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9983 CONSTRUCTORs from simplifying and VN.
9985 2017-08-14 Martin Sebor <msebor@redhat.com>
9987 * builtin-attrs.def: Add comments.
9989 2017-08-14 Martin Sebor <msebor@redhat.com>
9992 * doc/extend.texi (attribute nonstring): Document new attribute.
9994 2017-08-14 Martin Sebor <msebor@redhat.com>
9997 * tree-diagnostic.c (default_tree_printer): Handle %G.
9998 * gimple-pretty-print.h (percent_G_format): Declare new function.
9999 * gimple-pretty-print.c (percent_G_format): Define.
10000 * tree-pretty-print.c (percent_K_format): Add argument.
10002 2017-08-14 Martin Sebor <msebor@redhat.com>
10004 PR translation/79998
10005 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10006 Remove a stray space.
10008 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10011 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10012 (*iordi_1_bts): Ditto.
10013 (*xordi_1_btc): Ditto.
10015 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10018 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10019 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10021 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10022 quoted strings, and make more translator-friendly.
10023 (darwin_rs6000_override_options): Likewise.
10024 (rs6000_option_override_internal): Likewise.
10025 (rs6000_return_in_memory): Fix overlong line.
10026 (init_cmulative_args): Use quoted strings, and make more
10027 translator-friendly.
10028 (rs6000_pass_by_reference): Fix overlong line.
10029 (def_builtin): Use quoted strings.
10030 (altivec_expand_predicate_builtin): Use quoted strings, and make
10031 more translator-friendly.
10032 (htm_expand_builtin): Use quoted strings.
10033 (cpu_expand_builtin): Use quoted strings, and make more
10034 translator-friendly.
10035 (altivec_expand_builtin): Likewise.
10036 (paired_expand_predicate_builtin): Likewise.
10037 (rs6000_invalid_builtin): Likewise.
10038 (builtin_function_type): Use quoted strings.
10039 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10040 more translator-friendly.
10041 (rs6000_trampoline_init): Likewise.
10042 (rs6000_handle_altivec_attribute): Likewise.
10043 (rs6000_inner_target_options): Use quoted strings.
10044 (rs6000_disable_incompatible_switches): Likewise.
10045 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10046 strings, and make more translator-friendly.
10047 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10049 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10051 PR tree-optimization/81799
10052 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10053 cond_expr to simple gimple operand.
10055 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10057 PR middle-end/46932
10058 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10060 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10064 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10065 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10066 TARGET_GASISR_PROLOGUES.
10067 * config/avr/avr.c (avr_option_override): Same.
10068 (avr_pass_pre_proep::execute): Same.
10070 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10073 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10074 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10076 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10078 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10079 define_insn_and_split. Split to a memory load from 0 in
10080 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10081 using PTR mode iterator.
10082 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10083 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10084 (*add_tp_<mode>): Redefine as define_insn_and_split.
10085 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10086 address space. Merge with *add_tp_x32 using PTR mode iterator.
10087 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10088 Split to an add with a memory load from 0 in
10089 DEFAULT_TLS_SEG_REG address space.
10091 2017-08-12 Andrew Pinski <apinski@cavium.com>
10093 * config/aarch64/aarch64-option-extensions.def (rdma):
10094 Fix feature string to what Linux prints out in /proc/cpuinfo.
10096 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10099 * dwarf2out.c (modified_type_die): For C typedef types that have
10100 an ultimate origin, process the ultimate origin instead of the
10102 (gen_typedef_die): Assert that input DECLs have no ultimate
10104 (gen_type_die_with_usage): For typedef variants that have an
10105 ultimate origin, just call gen_decl_die on the original DECL.
10106 (process_scope_var): Avoid creating DIEs for local typedefs and
10107 concrete static variables.
10109 2017-08-12 Alan Modra <amodra@gmail.com>
10113 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10114 match gnu-user.h startfile.
10115 (ENDFILE_LINUX_SPEC): Similarly.
10117 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10120 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10122 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10124 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10125 * config/aarch64/aarch64.md (mov<mode>): Change.
10126 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10127 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10128 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10130 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10132 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10133 for storage order barriers.
10135 2017-08-11 Martin Liska <mliska@suse.cz>
10138 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10139 variables of void type.
10141 2017-08-11 Martin Liska <mliska@suse.cz>
10143 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10144 TARGET_SUPPORTS_ALIASES.
10145 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10146 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10147 (optimize_weakref): Likewise.
10148 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10149 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10150 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10152 2017-08-11 Martin Liska <mliska@suse.cz>
10155 * config/i386/i386.c (make_resolver_func): Do complete
10156 refactoring of the function.
10158 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10161 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10162 * config/i386/i386.c (ix86_stack_protect_guard): Use
10163 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10164 * doc/invoke.texi (x86 Options): Document
10165 -mstack-protector-guard-symbol= option.
10167 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10169 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10170 * config/i386/i386.c (ix86_split_stack_guard): New function.
10171 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10172 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10173 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10174 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10175 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10176 (split_stack_space_check): Call ix86_split_stack_guard.
10178 2017-08-10 Martin Sebor <msebor@redhat.com>
10180 * print-tree.c (print_node): Print location using the established
10182 Replace spaces with colons.
10183 (debug_raw, debug): Ditto.
10185 2017-08-10 Martin Sebor <msebor@redhat.com>
10188 * pretty-print.c (pp_format): Correct the handling of %s precision.
10190 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10193 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10195 (ix86_finalize_stack_frame_flags): This. Also clear
10196 frame_pointer_needed if -fno-omit-frame-pointer is used without
10198 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10199 with ix86_finalize_stack_frame_flags.
10200 (ix86_expand_epilogue): Likewise.
10201 (ix86_expand_split_stack_prologue): Likewise.
10202 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10204 2017-08-10 Martin Liska <mliska@suse.cz>
10207 * c-attribs.c (handle_target_attribute):
10208 Report warning for an empty string argument of target attribute.
10210 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10213 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10215 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10216 or DECL_NONLOCAL labels.
10217 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10218 or DECL_NONLOCAL labels here.
10220 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10223 to indicate when early gimple folding has been disabled.
10224 (rs6000_gimple_fold_builtin): Add debug content.
10225 (rs6000_invalid_builtin): Fix whitespace.
10226 (rs6000_expand_builtin): Fix whitespace.
10227 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10229 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10232 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10233 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10235 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10237 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10238 (qdf24xx): Likewise.
10239 * config/aarch64/aarch64.md: Include falkor.md.
10240 * config/aarch64/falkor.md: New.
10242 2017-08-09 Marek Polacek <polacek@redhat.com>
10245 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10246 * diagnostic.c (emit_diagnostic): Add a comment.
10247 (emit_diagnostic_valist): New function.
10249 2017-08-09 Marek Polacek <polacek@redhat.com>
10252 * input.c (make_location): New overload.
10253 * input.h (make_location): Declare.
10255 2017-08-08 Alan Modra <amodra@gmail.com>
10256 H.J. Lu <hongjiu.lu@intel.com>
10259 * gcc.c (NO_PIE_SPEC): Delete.
10260 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10262 (LINK_PIE_SPEC): ..to here.
10263 (LINK_COMMAND_SPEC): Support -no-pie.
10264 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10265 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10267 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10268 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10269 (ENDFILE_CRTEND_SPEC): Similarly.
10271 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10274 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10275 (mstack-protector-guard-offset=): Ditto.
10276 * config/i386/i386.c (ix86_option_override): Handle
10277 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10279 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10280 ix86_stack_protect_guard_offset variables.
10281 (TARGET_STACK_PROTECT_GUARD): Always define.
10282 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10283 and -mstack-protector-guard-offset= options.
10285 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10287 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10288 boundary case for the last candidate.
10290 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10292 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10293 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10295 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10297 PR middle-end/19706
10298 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10299 * config/aarch64/aarch64-builtins.c
10300 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10301 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10302 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10304 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10305 Andrew Pinski <pinskia@gmail.com>
10307 PR middle-end/19706
10308 * internal-fn.def (XORSIGN): New.
10309 * optabs.def (xorsign_optab): New.
10310 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10311 (convert_expand_mult_copysign): New.
10312 (pass_optimize_widening_mul::execute): Call
10313 convert_expand_mult_copysign.
10315 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10317 PR tree-optimization/81354
10318 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10319 Insert on edges rather than explicitly creating landing pads.
10320 (analyze_candidates_and_replace): Commit edge inserts.
10322 2017-08-08 Richard Biener <rguenther@suse.de>
10324 PR middle-end/81719
10325 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10326 (expand_simple_operations): Also look through ADDR_EXPRs with
10327 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10329 2017-08-08 Richard Biener <rguenther@suse.de>
10331 PR tree-optimization/81723
10332 * tree-vect-slp.c (struct bst_traits): New hash traits.
10333 (bst_fail): New global.
10334 (vect_build_slp_tree_2): New worker, split out from ...
10335 (vect_build_slp_tree): ... this now wrapping it with using
10336 bst_fail set to cache SLP tree build fails. Properly handle
10338 (vect_analyze_slp_instance): Allocate and free bst_fail.
10340 2017-08-08 Martin Liska <mliska@suse.cz>
10343 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10344 LABEL_DECLs that can be from a different function.
10346 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10348 PR tree-optimization/81744
10349 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10350 loop's number of iterations.
10352 2017-08-08 Martin Liska <mliska@suse.cz>
10354 * asan.c: Include header files.
10355 * attribs.c (build_decl_attribute_variant): New function moved
10357 (build_type_attribute_qual_variant): Likewise.
10358 (cmp_attrib_identifiers): Likewise.
10359 (simple_cst_list_equal): Likewise.
10360 (omp_declare_simd_clauses_equal): Likewise.
10361 (attribute_value_equal): Likewise.
10362 (comp_type_attributes): Likewise.
10363 (build_type_attribute_variant): Likewise.
10364 (lookup_ident_attribute): Likewise.
10365 (remove_attribute): Likewise.
10366 (merge_attributes): Likewise.
10367 (merge_type_attributes): Likewise.
10368 (merge_decl_attributes): Likewise.
10369 (merge_dllimport_decl_attributes): Likewise.
10370 (handle_dll_attribute): Likewise.
10371 (attribute_list_equal): Likewise.
10372 (attribute_list_contained): Likewise.
10373 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10374 (lookup_attribute_by_prefix): Likewise.
10375 * bb-reorder.c: Include header files.
10376 * builtins.c: Likewise.
10377 * calls.c: Likewise.
10378 * cfgexpand.c: Likewise.
10379 * cgraph.c: Likewise.
10380 * cgraphunit.c: Likewise.
10381 * convert.c: Likewise.
10382 * dwarf2out.c: Likewise.
10383 * final.c: Likewise.
10384 * fold-const.c: Likewise.
10385 * function.c: Likewise.
10386 * gimple-expr.c: Likewise.
10387 * gimple-fold.c: Likewise.
10388 * gimple-pretty-print.c: Likewise.
10389 * gimple.c: Likewise.
10390 * gimplify.c: Likewise.
10391 * hsa-common.c: Likewise.
10392 * hsa-gen.c: Likewise.
10393 * internal-fn.c: Likewise.
10394 * ipa-chkp.c: Likewise.
10395 * ipa-cp.c: Likewise.
10396 * ipa-devirt.c: Likewise.
10397 * ipa-fnsummary.c: Likewise.
10398 * ipa-inline.c: Likewise.
10399 * ipa-visibility.c: Likewise.
10401 * lto-cgraph.c: Likewise.
10402 * omp-expand.c: Likewise.
10403 * omp-general.c: Likewise.
10404 * omp-low.c: Likewise.
10405 * omp-offload.c: Likewise.
10406 * omp-simd-clone.c: Likewise.
10407 * opts-global.c: Likewise.
10408 * passes.c: Likewise.
10409 * predict.c: Likewise.
10410 * sancov.c: Likewise.
10411 * sanopt.c: Likewise.
10412 * symtab.c: Likewise.
10413 * toplev.c: Likewise.
10414 * trans-mem.c: Likewise.
10415 * tree-chkp.c: Likewise.
10416 * tree-eh.c: Likewise.
10417 * tree-into-ssa.c: Likewise.
10418 * tree-object-size.c: Likewise.
10419 * tree-parloops.c: Likewise.
10420 * tree-profile.c: Likewise.
10421 * tree-ssa-ccp.c: Likewise.
10422 * tree-ssa-live.c: Likewise.
10423 * tree-ssa-loop.c: Likewise.
10424 * tree-ssa-sccvn.c: Likewise.
10425 * tree-ssa-structalias.c: Likewise.
10426 * tree-ssa.c: Likewise.
10427 * tree-streamer-in.c: Likewise.
10428 * tree-vectorizer.c: Likewise.
10429 * tree-vrp.c: Likewise.
10430 * tsan.c: Likewise.
10431 * ubsan.c: Likewise.
10432 * varasm.c: Likewise.
10433 * varpool.c: Likewise.
10434 * tree.c: Remove functions moved to attribs.[ch].
10435 * tree.h: Likewise.
10436 * config/aarch64/aarch64.c: Add attrs.h header file.
10437 * config/alpha/alpha.c: Likewise.
10438 * config/arc/arc.c: Likewise.
10439 * config/arm/arm.c: Likewise.
10440 * config/avr/avr.c: Likewise.
10441 * config/bfin/bfin.c: Likewise.
10442 * config/c6x/c6x.c: Likewise.
10443 * config/cr16/cr16.c: Likewise.
10444 * config/cris/cris.c: Likewise.
10445 * config/darwin.c: Likewise.
10446 * config/epiphany/epiphany.c: Likewise.
10447 * config/fr30/fr30.c: Likewise.
10448 * config/frv/frv.c: Likewise.
10449 * config/ft32/ft32.c: Likewise.
10450 * config/h8300/h8300.c: Likewise.
10451 * config/i386/winnt.c: Likewise.
10452 * config/ia64/ia64.c: Likewise.
10453 * config/iq2000/iq2000.c: Likewise.
10454 * config/lm32/lm32.c: Likewise.
10455 * config/m32c/m32c.c: Likewise.
10456 * config/m32r/m32r.c: Likewise.
10457 * config/m68k/m68k.c: Likewise.
10458 * config/mcore/mcore.c: Likewise.
10459 * config/microblaze/microblaze.c: Likewise.
10460 * config/mips/mips.c: Likewise.
10461 * config/mmix/mmix.c: Likewise.
10462 * config/mn10300/mn10300.c: Likewise.
10463 * config/moxie/moxie.c: Likewise.
10464 * config/msp430/msp430.c: Likewise.
10465 * config/nds32/nds32-isr.c: Likewise.
10466 * config/nds32/nds32.c: Likewise.
10467 * config/nios2/nios2.c: Likewise.
10468 * config/nvptx/nvptx.c: Likewise.
10469 * config/pa/pa.c: Likewise.
10470 * config/pdp11/pdp11.c: Likewise.
10471 * config/powerpcspe/powerpcspe.c: Likewise.
10472 * config/riscv/riscv.c: Likewise.
10473 * config/rl78/rl78.c: Likewise.
10474 * config/rx/rx.c: Likewise.
10475 * config/s390/s390.c: Likewise.
10476 * config/sh/sh.c: Likewise.
10477 * config/sol2.c: Likewise.
10478 * config/sparc/sparc.c: Likewise.
10479 * config/spu/spu.c: Likewise.
10480 * config/stormy16/stormy16.c: Likewise.
10481 * config/tilegx/tilegx.c: Likewise.
10482 * config/tilepro/tilepro.c: Likewise.
10483 * config/v850/v850.c: Likewise.
10484 * config/vax/vax.c: Likewise.
10485 * config/visium/visium.c: Likewise.
10486 * config/xtensa/xtensa.c: Likewise.
10488 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10491 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10492 constraints since the -mupper-regs-* switches have been
10494 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10495 into a vector from a double word element that was extracted from
10496 another vector, and eliminate extra XXPERMDI instructions.
10497 (vsx_concat_<mode>_2): Likewise.
10498 (vsx_concat_<mode>_3): Likewise.
10499 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10500 concat to allow optimizing inserts from previous extracts.
10502 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10504 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10505 memory reference to a SSP offset in TLS address space.
10506 (ix86_print_operand) <case '@'>: Remove.
10507 (ix86_print_operand_punct_valid_p): Remove '@' code.
10508 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10509 UNSPEC_SP_TLS_TEST.
10510 (stack_tls_protect_set_<mode>): Remove.
10511 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10512 (stack_tls_protect_test_<mode>): Remove.
10513 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10515 2017-08-07 Olivier Hainque <hainque@adacore.com>
10518 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10520 2017-08-07 Douglas Rupp <rupp@adacore.com>
10522 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10523 variable was referenced as multidir in command.
10525 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10528 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10531 2017-08-07 Martin Liska <mliska@suse.cz>
10533 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10534 * config/rl78/rl78.c: Add include of attribs.h.
10535 * config/sh/sh.c: Likewise.
10536 * config/v850/v850.c: Likewise.
10538 2017-08-07 Tom de Vries <tom@codesourcery.com>
10540 PR middle-end/78266
10541 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10543 2017-08-07 Martin Liska <mliska@suse.cz>
10545 * config/mips/mips.c: Include attribs.h.
10547 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10550 * doc/invoke.texi: Document change in behvaior for -Ofast for
10553 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10555 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10557 (aarch64_pop_regs): Likewise.
10558 (aarch64_gen_load_pair): Likewise.
10559 (aarch64_save_callee_saves): Likewise.
10560 (aarch64_restore_callee_saves): Likewise.
10562 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10564 * config/i386/i386.c: Revert the last change.
10566 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10569 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10571 (ix86_finalize_stack_frame_flags): This. Also clear
10572 frame_pointer_needed if -fno-omit-frame-pointer is used without
10574 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10575 with ix86_finalize_stack_frame_flags.
10576 (ix86_expand_epilogue): Likewise.
10577 (ix86_expand_split_stack_prologue): Likewise.
10579 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10582 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10583 to P_AES for Westmere.
10585 2017-08-07 Jonathan Yong <10walls@gmail.com>
10587 * config/i386/mingw.opt (fset-stack-executable): Removed.
10588 * config/i386/cygming.opt (fset-stack-executable): Moved
10590 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10592 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10594 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10596 2017-08-07 Marek Polacek <polacek@redhat.com>
10598 PR middle-end/81737
10599 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10601 2017-08-07 Martin Liska <mliska@suse.cz>
10603 * attribs.h (canonicalize_attr_name): New function.
10604 (cmp_attribs): Move from c-format.c and adjusted.
10605 (is_attribute_p): Moved from tree.h.
10606 * tree-inline.c: Add new includes.
10607 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10608 (private_is_attribute_p): Remove.
10609 (private_lookup_attribute): Likewise.
10610 (private_lookup_attribute_by_prefix): Simplify.
10611 (remove_attribute): Use is_attribute_p.
10612 * tree.h: Remove removed declarations.
10614 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10616 PR middle-end/81698
10617 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10618 instead of computing it in the function. Formatting fix.
10619 (expand_case): Don't rely on default_edge being the first edge,
10620 clear it if removing it, pass default_edge to
10621 emit_case_dispatch_table.
10622 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10625 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10627 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10628 insn in the function, emit NOP after the insn.
10630 2017-08-06 Tom de Vries <tom@codesourcery.com>
10632 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10635 2017-08-06 Tom de Vries <tom@codesourcery.com>
10637 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10640 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10642 PR tree-optimization/57371
10643 * match.pd: New pattern.
10645 2017-08-04 Marek Polacek <polacek@redhat.com>
10647 PR middle-end/81695
10648 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10649 perform the computation in offset_int.
10651 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10653 PR tree-optimization/81136
10654 * tree-vectorizer.h: Include tree-hash-traits.h.
10655 (vec_base_alignments): New typedef.
10656 (vec_info): Add a base_alignments field.
10657 (vect_record_base_alignments): Declare.
10658 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10660 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10661 (create_data_ref): Add an is_conditional_in_stmt argument.
10662 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10663 the is_conditional_in_stmt field.
10664 (data_ref_loc): Add an is_conditional_in_stmt field.
10665 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10666 (find_data_references_in_stmt): Update call to create_data_ref.
10667 (graphite_find_data_references_in_stmt): Likewise.
10668 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10669 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10670 (vect_record_base_alignment): New function.
10671 (vect_record_base_alignments): Likewise.
10672 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10673 for nested statements even if we fail to compute a misalignment.
10674 Use pooled base alignments for unconditional references.
10675 (vect_find_same_alignment_drs): Compare base addresses instead
10677 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10678 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10680 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10682 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10683 Add an explicit name for the enum. Use auto_vec for slp_instances
10684 and grouped_stores.
10685 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10687 (_bb_vec_info): Add a constructor and destructor.
10688 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10689 (destroy_loop_vec_info): Delete.
10690 (vect_destroy_datarefs): Likewise.
10691 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10692 (vec_info::vec_info): New function.
10693 (vec_info::~vec_info): Likewise.
10694 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10695 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10696 destroy_loop_vec_info.
10697 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10698 (_loop_vec_info::_loop_vec_info): ...this.
10699 (destroy_loop_vec_info): Replace with...
10700 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10701 the stmt_vec_infos. Leave handling of vec_info information to its
10702 destructor. Remove explicit vector releases.
10703 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10704 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10705 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10706 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10707 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10708 (destroy_bb_vec_info): Replace with...
10709 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10710 information to its destructor.
10711 (vect_slp_analyze_bb_1): Use new and delete instead of
10712 new_bb_vec_info and destroy_bb_vec_info.
10713 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10716 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10718 * tree-data-ref.h (subscript): Add access_fn field.
10719 (data_dependence_relation): Add could_be_independent_p.
10720 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10721 (same_access_functions): Move to tree-data-ref.c.
10722 * tree-data-ref.c (ref_contains_union_access_p): New function.
10723 (access_fn_component_p): Likewise.
10724 (access_fn_components_comparable_p): Likewise.
10725 (dr_analyze_indices): Add a reference to access_fn_component_p.
10726 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10728 (constant_access_functions): Likewise.
10729 (add_other_self_distances): Likewise.
10730 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10731 (initialize_data_dependence_relation): Use XCNEW and remove
10732 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10733 of access functions that have the same type. Allow the
10734 subsequence to end with different bases in some circumstances.
10735 Record the chosen access functions in SUB_ACCESS_FN.
10736 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10737 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10738 (subscript_dependence_tester_1): Likewise dra and drb.
10739 (build_classic_dist_vector): Update calls accordingly.
10740 (subscript_dependence_tester): Likewise.
10741 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10742 DDR_COULD_BE_INDEPENDENT_P.
10743 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10744 comp_alias_ddrs instead of may_alias_ddrs.
10745 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10747 (vect_analyze_data_ref_dependence): Use it if
10748 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10749 distance vectors if that fails.
10750 (dependence_distance_ge_vf): New function.
10751 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10752 LOOP_VINFO_MAY_ALIAS_DDRS.
10754 2017-08-04 Richard Biener <rguenther@suse.de>
10756 PR middle-end/81705
10757 * fold-const.c (fold_binary_loc): Properly restrict
10758 minus_var0 && minus_var1 case when associating undefined overflow
10761 2017-08-04 Tom de Vries <tom@codesourcery.com>
10763 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10765 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10767 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10768 Don't start diagnostic messages with a capital letter.
10769 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10771 (rs6000_invalid_builtin): Likewise.
10772 (rs6000_trampoline_init): Likewise.
10774 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10777 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10778 after setting changeable df flags.
10780 2017-08-03 Richard Biener <rguenther@suse.de>
10782 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10783 up if the use is in USE - X.
10785 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10787 * toplev.c (dumpfile.h): New include.
10788 (internal_error_reentered): New static function. Use it...
10789 (internal_error_function): ...here to handle reentered internal_error.
10791 2017-08-03 Richard Biener <rguenther@suse.de>
10793 PR middle-end/81148
10794 * fold-const.c (split_tree): Add minus_var and minus_con
10795 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10796 here but always use minus_*.
10797 (associate_trees): Assert we never associate with MINUS_EXPR
10798 and NULL first operand. Do not recurse for PLUS_EXPR operands
10799 when associating as MINUS_EXPR either.
10800 (fold_binary_loc): Track minus_var and minus_con.
10802 2017-08-03 Tom de Vries <tom@codesourcery.com>
10805 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10806 ACCEL_COMPILER, apply finish_options on
10807 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10809 2017-08-03 Tom de Vries <tom@codesourcery.com>
10812 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10813 function_entry_patch_area_size > 0.
10815 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10818 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10819 instead of 10??LU, perform unit multiplication in wide_int,
10820 don't change alloc_object_size_limit if the limit is larger
10823 PR tree-optimization/81655
10824 PR tree-optimization/81588
10825 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10826 the case when ranges[i].low and high are 1 for unsigned type with
10829 PR middle-end/81052
10830 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10831 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10833 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10835 * tree-vrp.h: Add include guard.
10837 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10840 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10841 (ud2): New insn pattern.
10842 * config/i386/i386.c (ix86_expand_epilogue):
10843 For naked functions, generate ud2 instead of trap insn.
10845 2017-08-02 Marek Polacek <polacek@redhat.com>
10848 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10850 2017-08-02 Tom de Vries <tom@codesourcery.com>
10851 Cesar Philippidis <cesar@codesourcery.com>
10853 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10854 Add missing edge probabilities.
10856 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10858 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10859 Correct endianness.
10861 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10863 PR middle-end/79499
10864 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10865 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10866 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10867 prologue_seq sequences - if any.
10869 2017-08-02 Richard Biener <rguenther@suse.de>
10871 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10872 via vectors if supported, integer extracts via punning if supported
10873 or otherwise vector extracts.
10875 2017-08-02 Richard Biener <rguenther@suse.de>
10877 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10879 (bitmap_insert_into_set): ... this.
10881 2017-08-02 Richard Biener <rguenther@suse.de>
10883 PR tree-optimization/81633
10885 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10887 PR tree-optimization/71752
10888 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10890 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10892 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10893 (machine_function::call_ms2sysv_pad_out): Remove field.
10894 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10895 (ix86_compute_frame_layout): Likewise.
10897 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10900 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10901 attribute with interrupt attribute.
10903 2017-08-01 Andrew Pinski <apinski@cavium.com>
10905 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10906 BIT_INSERT_EXPR's operand 1
10907 to see if the types precision matches.
10909 2017-08-01 Martin Liska <mliska@suse.cz>
10911 PR middle-end/70140
10912 * builtins.c (expand_builtin_memcpy_args): Remove.
10913 (expand_builtin_memcpy): Call newly added function
10914 expand_builtin_memory_copy_args.
10915 (expand_builtin_memcpy_with_bounds): Likewise.
10916 (expand_builtin_mempcpy): Remove last argument.
10917 (expand_builtin_mempcpy_with_bounds): Likewise.
10918 (expand_builtin_memory_copy_args): New function created from
10919 expand_builtin_mempcpy_args with small modifications.
10920 (expand_builtin_mempcpy_args): Remove.
10921 (expand_builtin_stpcpy): Remove unused argument.
10922 (expand_builtin): Likewise.
10923 (expand_builtin_with_bounds): Likewise.
10925 2017-08-01 Martin Liska <mliska@suse.cz>
10928 Make mempcpy more optimal (PR middle-end/70140).
10930 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10933 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10934 __builtin_vec_cmpne verify both arguments are compatible vectors
10935 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10936 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10937 move computation of aligned to after checking the argument types.
10941 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10942 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10945 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10946 Doug Rupp <rupp@adacore.com>
10947 Olivier Hainque <hainque@adacore.com>
10949 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10950 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10952 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10953 for TARGET_OS_CPP_BUILTIN.
10954 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10955 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10957 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10958 passing required abi options to the assembler for EABI configurations.
10959 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10960 of .text.hot and .text.unlikely sections for kernel modules when
10961 using ARM style exceptions.
10962 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10963 options. Add EXTRA_CC1_SPEC.
10964 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10966 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10968 (ARM_TARGET2_DWARF_FORMAT): Define.
10969 * config/arm/t-vxworks: Adjust multilib control to removal of the
10970 Diab command line options.
10972 2017-08-01 Martin Liska <mliska@suse.cz>
10974 PR gcov-profile/81561
10975 * gcov.c (unblock): Make unblocking safe as we need to preserve
10976 index correspondence of blocks and block_lists.
10978 2017-08-01 Richard Biener <rguenther@suse.de>
10980 PR tree-optimization/81181
10981 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10982 (compute_antic): ... end of iteration here.
10984 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10986 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10987 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10988 (ftree-slp-vectorize): Likewise.
10989 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10990 can no longer be set independent of flag_tree_loop_vectorize.
10991 * omp-general.c (emp_max_vf): Likewise.
10992 * opts.c (enable_fdo_optimizations): Remove references to
10993 flag_tree_vectorize, these are now implicit.
10994 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10995 and leave it for the options machinery.
10997 2017-08-01 Martin Liska <mliska@suse.cz>
10999 PR middle-end/70140
11000 * builtins.c (expand_builtin_memcpy_args): Remove.
11001 (expand_builtin_memcpy): Call newly added function
11002 expand_builtin_memory_copy_args.
11003 (expand_builtin_memcpy_with_bounds): Likewise.
11004 (expand_builtin_mempcpy): Remove last argument.
11005 (expand_builtin_mempcpy_with_bounds): Likewise.
11006 (expand_builtin_memory_copy_args): New function created from
11007 expand_builtin_mempcpy_args with small modifications.
11008 (expand_builtin_mempcpy_args): Remove.
11009 (expand_builtin_stpcpy): Remove unused argument.
11010 (expand_builtin): Likewise.
11011 (expand_builtin_with_bounds): Likewise.
11013 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11016 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11017 print "ds:" only for immediates in generic address space.
11019 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11022 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11023 (ix86_function_ok_for_sibcall): Return false for naked functions.
11025 2017-08-01 Richard Biener <rguenther@suse.de>
11027 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11028 (compute_antic): Seed worklist with exit block predecessors.
11029 * cfganal.c (dfs_find_deadend): For a cycle return the source
11030 of the edge closing it.
11032 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11034 * config/aarch64/aarch64.c
11035 (aarch64_can_const_movi_rtx_p): Move 0 check.
11037 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11039 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11040 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11042 * match.pd: Ditto in address comparison pattern.
11044 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11046 PR tree-optimization/81627
11047 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11048 closed ssa form for store-store chain.
11050 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11052 PR tree-optimization/81620
11053 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11054 for store-store chain.
11056 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11058 PR tree-optimization/81588
11059 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11060 ranges[i].in_p, invert comparison code ccode. For >/>=,
11061 swap rhs1 and rhs2 and comparison code unconditionally,
11062 for </<= don't do that. Don't swap rhs1/rhs2 again if
11063 ranges[i].in_p, instead invert comparison code ccode if
11064 opcode or oe->rank is BIT_IOR_EXPR.
11067 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11068 a direct optab to conversion optab.
11069 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11070 with GET_MODE_INNER as last argument instead of optab_handler.
11071 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11072 vector extraction if possible and optab is available.
11073 * expr.c (store_constructor): Use convert_optab_handler instead
11074 of optab_handler. Use vector initialization from smaller
11075 vectors if possible and optab is available.
11076 * tree-vect-stmts.c (vectorizable_load): Likewise.
11077 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11078 now have two modes.
11079 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11080 of vec_init from half-sized vectors with the same element mode.
11081 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11082 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11083 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11084 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11085 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11086 after mode in gen_vec_extract* calls.
11087 (vec_extract<mode>): Renamed to ...
11088 (vec_extract<mode><ssescalarmodelower>): ... this.
11089 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11090 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11091 element mode after mode in gen_vec_init* calls.
11092 (VEC_INIT_HALF_MODE): New mode iterator.
11093 (vec_init<mode>): Renamed to ...
11094 (vec_init<mode><ssescalarmodelower>): ... this.
11095 (vec_init<mode><ssehalfvecmodelower>): New expander.
11096 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11097 (vec_extractv2sfsf): ... this.
11098 (vec_initv2sf): Renamed to ...
11099 (vec_initv2sfsf): ... this.
11100 (vec_extractv2si): Renamed to ...
11101 (vec_extractv2sisi): ... this.
11102 (vec_initv2si): Renamed to ...
11103 (vec_initv2sisi): ... this.
11104 (vec_extractv4hi): Renamed to ...
11105 (vec_extractv4hihi): ... this.
11106 (vec_initv4hi): Renamed to ...
11107 (vec_initv4hihi): ... this.
11108 (vec_extractv8qi): Renamed to ...
11109 (vec_extractv8qiqi): ... this.
11110 (vec_initv8qi): Renamed to ...
11111 (vec_initv8qiqi): ... this.
11112 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11113 (vec_init<mode>): Renamed to ...
11114 (vec_init<mode><VEC_base_l>): ... this.
11115 (vec_extract<mode>): Renamed to ...
11116 (vec_extract<mode><VEC_base_l>): ... this.
11117 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11118 (vec_initv2sfsf): ... this.
11119 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11120 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11121 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11122 element mode after mode in gen_vec_init* calls.
11123 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11124 (vec_init<mode><Vel>): ... this.
11125 (vec_extract<mode>): Renamed to ...
11126 (vec_extract<mode><Vel>): ... this.
11127 * config/aarch64/iterators.md (Vel): New mode attribute.
11128 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11129 Add element mode after mode in gen_vec_extract* calls.
11130 * config/s390/vector.md (non_vec_l): New mode attribute.
11131 (vec_extract<mode>): Renamed to ...
11132 (vec_extract<mode><non_vec_l>): ... this.
11133 (vec_init<mode>): Renamed to ...
11134 (vec_init<mode><non_vec_l>): ... this.
11135 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11136 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11138 * config/arm/iterators.md (V_elem_l): New mode attribute.
11139 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11140 (vec_extract<mode><V_elem_l>): ... this.
11141 (vec_extractv2di): Renamed to ...
11142 (vec_extractv2didi): ... this.
11143 (vec_init<mode>): Renamed to ...
11144 (vec_init<mode><V_elem_l>): ... this.
11145 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11146 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11147 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11148 Add element mode after gen_vec_extract* calls.
11149 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11150 (vec_init<mode><unitmode>): ... this.
11151 (vec_extract<mode>): Renamed to ...
11152 (vec_extract<mode><unitmode>): ... this.
11153 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11154 (vec_init<mode><unitmode>): ... this.
11155 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11156 (vec_initv2sfsf): ... this.
11157 (vec_extractv2sf): Renamed to ...
11158 (vec_extractv2sfsf): ... this.
11159 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11160 Add element mode after gen_vec_extract* calls.
11161 * config/mips/mips.md (unitmode): New mode iterator.
11162 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11163 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11164 * config/spu/spu.md (inner_l): New mode attribute.
11165 (vec_init<mode>): Renamed to ...
11166 (vec_init<mode><inner_l>): ... this.
11167 (vec_extract<mode>): Renamed to ...
11168 (vec_extract<mode><inner_l>): ... this.
11169 * config/sparc/sparc.md (veltmode): New mode iterator.
11170 (vec_init<VMALL:mode>): Renamed to ...
11171 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11172 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11173 (vec_initv2sisi): ... this.
11174 (vec_initv2sf): Renamed to ...
11175 (vec_initv2sfsf): ... this.
11176 (vec_extractv2sf): Renamed to ...
11177 (vec_extractv2sfsf): ... this.
11178 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11179 (vec_init<mode>): Renamed to ...
11180 (vec_init<mode><VEC_base_l>): ... this.
11181 (vec_extract<mode>): Renamed to ...
11182 (vec_extract<mode><VEC_base_l>): ... this.
11183 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11184 (vec_initv2sfsf): ... this.
11185 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11186 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11187 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11188 gen_vec_init* calls.
11190 2017-08-01 Richard Biener <rguenther@suse.de>
11192 PR tree-optimization/81297
11193 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11194 TREE_OVERFLOW from INTEGER_CSTs.
11196 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11198 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11200 2017-07-31 Carl Love <cel@us.ibm.com>
11202 * config/rs6000/rs6000-c: Add support for built-in functions
11203 vector signed char vec_xl_be (signed long long, signed char *);
11204 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11205 vector signed int vec_xl_be (signed long long, signed int *);
11206 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11207 vector signed long long vec_xl_be (signed long long, signed long long *);
11208 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11209 vector signed short vec_xl_be (signed long long, signed short *);
11210 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11211 vector double vec_xl_be (signed long long, double *);
11212 vector float vec_xl_be (signed long long, float *);
11213 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11214 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11215 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11217 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11218 (altivec_expand_builtin): Add switch statement to call
11219 altivec_expand_xl_be for each builtin.
11220 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11221 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11222 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11223 __builtin_vsx_le_be_v16qi.
11224 * doc/extend.texi: Update the built-in documentation file for the
11225 new built-in functions.
11227 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11230 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11232 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11234 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11236 * config.gcc: Add z14.
11237 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11238 CPU model numbers for z13s and z14.
11239 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11241 * config/s390/s390-opts.h (enum processor_type): Rename
11242 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11243 * config/s390/s390.c (processor_table): Add field for CPU name to
11244 be passed to Binutils.
11245 (s390_asm_output_machine_for_arch): Use the new field in
11246 processor_table for Binutils.
11247 (s390_expand_builtin): Replace arch12 with z14.
11248 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11249 (s390_get_sched_attrmask): Likewise.
11250 (s390_get_unit_mask): Likewise.
11251 * config/s390/s390.opt: Add z14 to processor_type enum.
11253 2017-07-31 Martin Jambor <mjambor@suse.cz>
11256 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11257 regardless of optimization level.
11259 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11260 Martin Liska <mliska@suse.cz>
11262 * predict.def: Remove old comment and adjust probability.
11263 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11264 PREDICT statements.
11266 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11269 * config/i386/i386.c (ix86_function_naked): New function.
11270 (ix86_can_use_return_insn_p): Return false for naked functions.
11271 (ix86_expand_prologue): Skip prologue for naked functions.
11272 (ix86_expand_epilogue): Skip epilogue for naked functions
11273 and emit trap instruction.
11274 (ix86_warn_func_return): New function.
11275 (ix86_attribute_table): Add "naked" attribute specification.
11276 (TARGET_WARN_FUNC_RETURN): Define.
11277 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11279 2017-07-31 Martin Liska <mliska@suse.cz>
11281 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11282 (dump_gimple_bb_header): Always dump BB info.
11283 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11285 2017-07-31 Martin Liska <mliska@suse.cz>
11288 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11289 also with current_function_decl non-null equality.
11291 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11294 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11295 change type to the element type, instead add eltype variable and
11296 use it where we are interested in the element type.
11298 PR tree-optimization/81603
11299 * ipa-polymorphic-call.c
11300 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11301 offset arithmetic in offset_int, bail out if the resulting bit offset
11302 doesn't fit into shwi.
11304 2017-07-31 Martin Liska <mliska@suse.cz>
11306 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11307 (gimplify_save_expr): Fix comment.
11309 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11312 * config/i386/i386.c (ix86_function_arg): Update arguments for
11314 (ix86_compute_frame_layout): Set the initial stack offset to
11315 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11316 INCOMING_FRAME_SP_OFFSET.
11317 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11318 stack before exception handler returns.
11319 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11320 the 'ERROR_CODE' for exception handler.
11322 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11324 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11325 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11326 (ASM_OUTPUT_REG_POP): Ditto.
11327 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11328 instead of asm_fprintf to output pure string.
11330 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11332 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11333 to imported_module_or_decl hook.
11334 (debug_nothing_tree_tree_tree_bool): Remove.
11335 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11336 * debug.c (do_nothing_debug_hooks): Use
11337 debug_nothing_tree_tree_tree_bool_bool instead of
11338 debug_nothing_tree_tree_tree_bool.
11339 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11340 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11341 * sdbout.c (sdb_debug_hooks): Likewise.
11342 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11343 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11345 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11346 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11347 attribute, don't add anything.
11349 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11351 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11352 (fold_build2_stat_loc): Likewise.
11353 (fold_build3_stat_loc): Likewise.
11354 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11355 (fold_build1_loc): Remove macro.
11356 (fold_build2_loc): Likewise.
11357 (fold_build3_loc): Likewise.
11359 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11361 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11362 (gimple_build_debug_bind_source_stat): Likewise.
11363 * gimple.h (gimple_build_debug_bind): Remove macro.
11364 (gimple_build_debug_bind_source): Likewise.
11366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11368 * bitmap.c (bitmap_alloc): Adjust.
11369 (bitmap_gc_alloc): Likewise.
11370 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11374 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11375 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11376 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11377 (bitmap_gc_alloc_stat): Likewise.
11378 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11380 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11382 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11383 * rtl.h (shallow_copy_rtx): Remove macro.
11385 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11387 * emit-rtl.c (gen_raw_REG): Adjust.
11388 * gengenrtl.c (gendef): Likewise.
11389 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11390 * rtl.h (rtx_alloc): Remove macro.
11392 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11394 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11395 (build_tree_list_stat): Likewise.
11396 * tree.h (build_tree_list): Remove macro.
11397 (build_tree_list_vec): Likewise.
11399 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11401 * tree.c (make_vector_stat): Remove _stat from name.
11402 (build_vector_stat): Likewise.
11403 * tree.h (make_vector_stat): Remove macro.
11404 (build_vector_stat): Likewise.
11406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11408 * tree.h (build_var_debug_value): Remove prototype.
11410 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11412 * tree.c (tree_cons_stat): Remove _stat from name.
11413 * tree.h (tree_cons): Remove macro.
11415 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11417 * tree.c (build_vl_exp_stat): Remove _stat from name.
11418 * tree.h (build_vl_exp): Remove macro.
11420 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11422 * tree.c (build_decl_stat): Remove _stat from name.
11423 * tree.h (build_decl): Remove macro.
11425 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11427 * gimple.c (gimple_build_with_ops_stat): Adjust.
11428 (gimple_alloc_stat): Remove _stat from name.
11429 * gimple.h (gimple_alloc): Remove macro.
11431 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11433 * tree.c (make_tree_vec_stat): Remove _stat from name.
11434 (grow_tree_vec_stat): Likewise.
11435 * tree.h (make_tree_vec_stat): Adjust prototype.
11436 (grow_tree_vec_stat): Likewise.
11437 (make_tree_vec): Remove macro.
11438 (grow_tree_vec): Likewise.
11440 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11442 * fold-const.c (fold_build1_stat_loc): Adjust.
11443 (fold_build2_stat_loc): Likewise.
11444 (fold_build3_stat_loc): Likewise.
11445 * tree.c (build0_stat): Remove _stat from name.
11446 (build1_stat): Likewise.
11447 (build2_stat): Likewise.
11448 (build3_stat): Likewise.
11449 (build4_stat): Likewise.
11450 (build5_stat): Likewise.
11451 * tree.h (build1_loc): Remove macro, and rename _stat function
11453 (build2_loc): Likewise.
11454 (build3_loc): Likewise.
11455 (build4_loc): Likewise.
11456 (build5_loc): Likewise.
11458 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11460 * tree.c (make_int_cst_stat): Remove _stat from name.
11461 * tree.h (make_int_cst_stat): Adjust prototype.
11462 (make_int_cst): Remove macro.
11464 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11466 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11467 * tree.h (make_tree_binfo_stat): Adjust prototype.
11468 (make_tree_binfo): Remove.
11470 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11472 * tree.c (copy_node_stat): Rename to copy_node.
11473 (build_distinct_type_copy): Adjust.
11474 * tree.h (copy_node_stat): Adjust prototype.
11475 (copy_node): Remove macro.
11477 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11479 * tree.c (make_node_stat): rename to make_node.
11480 (build_tree_list_stat): Adjust.
11481 (build0_stat): Likewise.
11482 (build2_stat): Likewise.
11483 (build3_stat): Likewise.
11484 (build4_stat): Likewise.
11485 (build5_stat): Likewise.
11486 (build_decl_stat): Likewise.
11487 * tree.h (make_node_stat): Adjust prototype.
11488 (make_node): remove macro.
11490 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11492 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11493 (PPC_FEATURE2_SCV): Likewise.
11494 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11496 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11498 * config/aarch64/aarch64.c
11499 (aarch64_internal_mov_immediate): Add new special pattern.
11500 * config/aarch64/aarch64.md (*movdi_aarch64):
11501 Add reg/32bit const mov case.
11503 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11504 Richard Sandiford <richard.sandiford@linaro.org>
11506 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11507 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11508 Add integer and movi cases.
11509 (movi-split-hf-df-sf split, fp16): New.
11510 (enabled): Added TARGET_FP_F16INST.
11511 * config/aarch64/iterators.md (GPF_HF): New.
11512 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11514 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11516 * config/aarch64/aarch64.c
11517 (aarch64_simd_container_mode): Add prototype.
11518 (aarch64_expand_mov_immediate): Add HI support.
11519 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11520 (aarch64_can_const_movi_rtx_p): New.
11521 (aarch64_preferred_reload_class):
11522 Remove restrictions of using FP registers for certain SIMD operations.
11523 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11524 (aarch64_valid_floating_const): Add integer move validation.
11525 (aarch64_simd_imm_scalar_p): Remove.
11526 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11527 (aarch64_legitimate_constant_p): Expand list of supported cases.
11528 * config/aarch64/aarch64-protos.h
11529 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11530 (aarch64_reinterpret_float_as_int): New.
11531 (aarch64_simd_imm_scalar_p): Remove.
11532 * config/aarch64/constraints.md (Uvi): New.
11533 (Dd): Split into Ds and new Dd.
11534 * config/aarch64/aarch64.md (*movsi_aarch64):
11536 (*movdi_aarch64): Add SIMD mov case.
11538 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11540 * tree-predcom.c: (struct chain): Handle store-store chain in which
11541 stores for elimination only store loop invariant values.
11542 (execute_pred_commoning_chain): Ditto.
11543 (prepare_initializers_chain_store_elim): Ditto.
11544 (prepare_finalizers): Ditto.
11545 (is_inv_store_elimination_chain): New function.
11546 (initialize_root_vars_store_elim_1): New function.
11548 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11550 * tree-predcom.c: Revise general description of the pass.
11551 (enum chain_type): New enum type for store elimination.
11552 (struct chain): New field supporting store elimination.
11553 (struct component): Ditto.
11554 (dump_chain): Dump store-stores chain.
11555 (release_chain): Release resources.
11556 (split_data_refs_to_components): Compute and create component
11557 contains only stores for elimination.
11558 (get_chain_last_ref_at): New function.
11559 (make_invariant_chain): Initialization.
11560 (make_rooted_chain): Specify chain type in parameter and record it.
11561 (add_looparound_copies): Skip for store-stores chain.
11562 (determine_roots_comp): Compute type of chain and pass it to
11564 (initialize_root_vars_store_elim_2): New function.
11565 (finalize_eliminated_stores): New function.
11566 (remove_stmt): Handle store for elimination.
11567 (execute_pred_commoning_chain): Execute predictive commoning on
11568 store-store chains.
11569 (determine_unroll_factor): Skip unroll for store-stores chain.
11570 (prepare_initializers_chain_store_elim): New function.
11571 (prepare_initializers_chain): Hanlde store-store chain.
11572 (prepare_finalizers_chain, prepare_finalizers): New function.
11573 (tree_predictive_commoning_loop): Return integer value indicating
11574 if loop is unrolled or lcssa form is corrupted.
11575 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11577 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11579 * tree-predcom.c (initialize_root): Delete.
11580 (execute_pred_commoning_chain): Initialize root vars and replace
11581 reference of non-combined chain directly, rather than call above
11584 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11586 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11587 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11589 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11591 * tree-predcom.c (struct chain): New field init_seq.
11592 (release_chain): Release init_seq.
11593 (prepare_initializers_chain): Record intialization stmts in above
11595 (insert_init_seqs): New function.
11596 (tree_predictive_commoning_loop): Call insert_init_seqs.
11598 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11600 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11602 2017-07-28 Richard Biener <rguenther@suse.de>
11604 * match.pd: Remove superfluous :c.
11605 * genmatch.c (simplify::id): Add member.
11606 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11608 (current_id): New global.
11609 (dt_node::parent): Move from ...
11610 (dt_operand::parent): ... here. Add for_id member.
11611 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11612 (decision_tree::find_node): Relax order requirement when
11613 merging DT_TRUE nodes to ones inbetween the current simplify
11614 and the one we try to merge with. Add diagnostic whenever
11615 we need to enforce pattern order by not merging.
11616 (decision_tree::insert): Set current_id.
11617 (decision_tree::print_node): Dump parent node and for_id.
11618 (parser::last_id): Add member.
11619 (parser::push_simplify): Assign unique id.
11620 (parser::parser): Initialize last_id.
11622 2017-07-28 Martin Liska <mliska@suse.cz>
11625 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11626 gimple_build_debug_bind.
11628 2017-07-28 Richard Biener <rguenther@suse.de>
11630 PR tree-optimization/81502
11631 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11633 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11635 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11636 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11637 for BIT_FIELD_REF args.
11638 * fold-const.c (make_bit_field_ref): Likewise.
11639 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11641 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11644 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11645 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11646 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11647 Or it into SANITIZER_UNDEFINED.
11648 * ubsan.c: Include gimple-fold.h and varasm.h.
11649 (ubsan_expand_ptr_ifn): New function.
11650 (instrument_pointer_overflow): New function.
11651 (maybe_instrument_pointer_overflow): New function.
11652 (instrument_object_size): Formatting fix.
11653 (pass_ubsan::execute): Call instrument_pointer_overflow
11654 and maybe_instrument_pointer_overflow.
11655 * internal-fn.c (expand_UBSAN_PTR): New function.
11656 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11657 * sanitizer.def (__ubsan_handle_pointer_overflow,
11658 __ubsan_handle_pointer_overflow_abort): New builtins.
11659 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11660 * internal-fn.def (UBSAN_PTR): New internal function.
11661 * opts.c (sanitizer_opts): Add pointer-overflow.
11662 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11663 * fold-const.c (build_range_check): Compute pointer range check in
11664 integral type if pointer arithmetics would be needed. Formatting
11667 2017-07-28 Martin Liska <mliska@suse.cz>
11670 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11671 parameters that are of a variable-length.
11673 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11675 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11677 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11678 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11679 (CRT_CALL_STATIC_FUNCTION): Likewise.
11680 (ASM_DEFAULT_SPEC): New define.
11681 (ASM_SPEC32): Likewise.
11682 (ASM_SPEC64): Likewise.
11683 (ASM_SPEC_COMMON): Likewise.
11684 (ASM_SPEC): Likewise.
11685 (INVALID_64BIT): Likewise.
11686 (LINK_OS_DEFAULT_SPEC): Likewise.
11687 (LINK_OS_SPEC32): Likewise.
11688 (LINK_OS_SPEC64): Likewise.
11689 (POWERPC_LINUX): Likewise.
11690 (PTRDIFF_TYPE): Likewise.
11691 (RESTORE_FP_PREFIX): Likewise.
11692 (RESTORE_FP_SUFFIX): Likewise.
11693 (SAVE_FP_PREFIX): Likewise.
11694 (SAVE_FP_SUFFIX): Likewise.
11695 (SIZE_TYPE): Likewise.
11696 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11697 (TARGET_64BIT): Likewise.
11698 (TARGET_64BIT): Likewise.
11699 (TARGET_AIX): Likewise.
11700 (WCHAR_TYPE_SIZE): Likewise.
11701 (WCHAR_TYPE): Undefine.
11702 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11703 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11704 (CPP_OS_RTEMS_SPEC): Delete.
11705 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11706 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11708 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11710 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11712 PR tree-optimization/81578
11713 * tree-parloops.c (build_new_reduction): Bail out if
11714 reduction_code isn't one of the standard OpenMP reductions.
11715 Move the details printing after that decision.
11717 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11719 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11720 related to reload_in_progress.
11721 (splat_input_operand): Likewise.
11722 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11724 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11726 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11727 (TARGET_INSTANTIATE_DECLS): Likewise.
11728 (legitimate_indexed_address_p): Delete reload_in_progress code.
11729 (rs6000_debug_legitimate_address_p): Likewise.
11730 (rs6000_eliminate_indexed_memrefs): Likewise.
11731 (rs6000_emit_le_vsx_store): Likewise.
11732 (rs6000_emit_move_si_sf_subreg): Likewise.
11733 (rs6000_emit_move): Likewise.
11734 (register_to_reg_type): Likewise.
11735 (rs6000_pre_atomic_barrier): Likewise.
11736 (rs6000_machopic_legitimize_pic_address): Likewise.
11737 (rs6000_allocate_stack_temp): Likewise.
11738 (rs6000_address_for_fpconvert): Likewise.
11739 (rs6000_address_for_altivec): Likewise.
11740 (rs6000_secondary_memory_needed_rtx): Delete function.
11741 (rs6000_check_sdmode): Likewise.
11742 (rs6000_alloc_sdmode_stack_slot): Likewise.
11743 (rs6000_instantiate_decls): Likewise.
11744 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11745 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11746 Delete reload_in_progress.
11747 (*vec_reload_and_plus_<mptrsize>): Likewise.
11748 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11749 (vsx_div_v2di): Likewise.
11750 (vsx_udiv_v2di): Likewise.
11752 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11754 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11755 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11756 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11757 (rs6000_debug_reg_global): Delete print of LRA status.
11758 (rs6000_option_override_internal): Delete dead LRA related code.
11759 (rs6000_lra_p): Delete function.
11760 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11762 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11764 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11765 * config/riscv/rtems.h: New file.
11767 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11768 Sudakshina Das <sudi.das@arm.com>
11770 * config/aarch64/aarch64.md
11771 (define_split for and<mode>3nr_compare): Move
11772 non aarch64_logical_operand to a register.
11773 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11774 register immediate operand to a register.
11775 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11777 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11779 PR middle-end/81564
11780 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11782 2017-07-27 Richard Biener <rguenther@suse.de>
11784 PR tree-optimization/81573
11785 PR tree-optimization/81494
11786 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11787 multi defuse cycle case.
11789 2017-07-27 Richard Biener <rguenther@suse.de>
11791 PR tree-optimization/81571
11792 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11795 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11797 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11798 earlier and only if MASK_FPU is set. Adjust formatting.
11800 2017-07-27 Martin Liska <mliska@suse.cz>
11802 * opt-functions.awk: Add validation of value of Init.
11803 * optc-gen.awk: Pass new argument.
11805 2017-07-27 Martin Liska <mliska@suse.cz>
11807 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11808 Fix wrong condition.
11810 2017-07-27 Martin Liska <mliska@suse.cz>
11812 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11813 BBs and edges seen by autoFDO.
11815 2017-07-27 Richard Biener <rguenther@suse.de>
11817 PR tree-optimization/81502
11818 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11819 with incompatible but same sized type.
11820 (execute_update_addresses_taken): Likewise.
11822 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11824 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11825 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11827 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11830 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11831 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11832 Change s_operand to memory_operand.
11834 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11836 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11837 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11838 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11839 Emit instructions rather than returning an expression. Handle TFmode
11840 and KFmode by casting to TImode.
11841 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11842 (rs6000_emit_le_vsx_store): Likewise.
11843 * config/rs6000/vsx.md (VSX_TI): New iterator.
11844 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11845 (*vsx_le_undo_permute_<mode>): Likewise.
11846 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11847 emit the split sequence.
11848 (*vsx_le_perm_store_<mode>): Likewise.
11850 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11852 PR tree-optimization/81555
11853 PR tree-optimization/81556
11854 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11855 if true, force CHANGED for the recursive invocation.
11856 (reassociate_bb): Remember original length of ops array, pass
11857 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11859 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11860 attributes for noipa attribute. For naked attribute use
11861 lookup_attribute first before lookup_attribute_spec.
11862 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11864 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11865 for functions with noipa attribute.
11866 (cgraph_externally_visible_p): Return true for functions with noipa
11868 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11869 for functions with noipa attribute.
11870 * doc/extend.texi: Document noipa function attribute.
11871 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11872 also for functions with noipa attribute.
11873 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11875 2017-07-26 Andrew Pinski <apinski@cavium.com>
11877 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11878 vec_unalign_load_cost and vec_unalign_store_cost.
11880 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11882 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11883 -mvsx-small-integer option.
11884 (ISA_3_0_MASKS_IEEE): Likewise.
11885 (OTHER_VSX_VECTOR_MASKS): Likewise.
11886 (POWERPC_MASKS): Likewise.
11887 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11888 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11889 code, only testing for DImode being allowed in non-VSX floating
11891 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11892 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11894 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11895 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11896 TARGET_P8_VECTOR test.
11897 (rs6000_secondary_reload_simple_move): Likewise.
11898 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11899 since TARGET_P9_VECTOR was already tested.
11900 (rs6000_opt_masks): Remove -mvsx-small-integer.
11901 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11902 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11904 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11905 test for TARGET_VEXTRACTUB was used, and that uses
11907 (p9 extract splitter): Likewise.
11908 (vsx_extract_<mode>_di_p9): Likewise.
11909 (vsx_extract_<mode>_store_p9): Likewise.
11910 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11911 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11912 the elimination of TARGET_VSX_SMALL_INTEGER.
11913 (vsx_extract_<mode>_p8): Likewise.
11914 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11915 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11916 (vsx_set_<mode>_p9): Likewise.
11917 (vsx_set_v4sf_p9): Likewise.
11918 (vsx_set_v4sf_p9_zero): Likewise.
11919 (vsx_insert_extract_v4sf_p9): Likewise.
11920 (vsx_insert_extract_v4sf_p9_2): Likewise.
11921 * config/rs6000/rs6000.md (sign extend splitter): Change
11922 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11923 (floatsi<mode>2_lfiwax_mem): Likewise.
11924 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11925 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11926 since a test for TARGET_P9_VECTOR was used.
11927 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11928 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11929 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11930 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11931 TARGET_P8_VECTOR test.
11932 (fix_trunc<mode>si2_stfiwx): Likewise.
11933 (fix_trunc<mode>si2_internal): Likewise.
11934 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11935 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11937 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11938 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11939 TARGET_P8_VECTOR test.
11940 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11941 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11942 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11944 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11945 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11946 since a test for TARGET_P9_VECTOR was used.
11947 (splitter for loading small constants): Likewise.
11949 2017-07-26 Andrew Pinski <apinski@cavium.com>
11951 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11954 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11957 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11958 (fp_valid_at): Likewise.
11960 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11962 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11963 (qdf24xx_addrcost_table): Likewise.
11964 (cortexa57_tunings): Update to use generic_branch_cost.
11965 (cortexa72_tunings): Likewise.
11966 (cortexa73_tunings): Likewise.
11967 (qdf24xx_tunings): Likewise.
11969 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11971 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11972 (thunderx2t99_branch_cost): Likewise.
11973 (cortexa35_tunings): Update to use generic_branch_cost.
11974 (cortexa53_tunings): Likewise.
11975 (cortexa57_tunings): Likewise.
11976 (cortexa72_tunings): Likewise.
11977 (cortexa73_tunings): Likewise.
11978 (thunderx2t99_tunings): Likewise.
11980 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11982 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11983 (sparc_option_override): Honour MASK_FSMULD.
11984 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11985 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11986 * config/sparc/sparc.opt (mfsmuld): New option.
11987 * doc/invoke.texi (mfsmuld): Document option.
11989 2017-07-26 Marek Polacek <polacek@redhat.com>
11991 PR middle-end/70992
11992 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11994 2017-07-26 Richard Biener <rguenther@suse.de>
11996 * gimple-match-head.c (do_valueize): Return OP if valueize
11998 (get_def): New helper to get at the def stmt of a SSA name
11999 if valueize allows.
12000 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12001 do_valueize to get at the def stmt.
12002 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12004 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12006 PR middle-end/46932
12007 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12009 2017-07-26 Martin Liska <mliska@suse.cz>
12012 * function.c (expand_function_start): Make expansion of
12013 nonlocal_goto_save_area after parm_birth_insn.
12015 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12017 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12018 from all CPU target flags enable members.
12020 2017-07-26 Richard Biener <rguenther@suse.de>
12022 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12023 (decision_tree::gen): Make 'type' const.
12024 (write_predicate): Likewise.
12026 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12028 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12029 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12030 (rs6000_option_override_internal): Likewise.
12031 (rs6000_expand_vector_set): Likewise.
12032 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12033 (TARGET_UPPER_REGS_SF): Likewise.
12034 (TARGET_UPPER_REGS_DI): Likewise.
12035 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12036 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12037 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12038 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12039 (Splitters for DI constants in Altivec registers): Likewise.
12040 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12041 (vsx_set_v4sf_p9): Likewise.
12042 (vsx_set_v4sf_p9_zero): Likewise.
12043 (vsx_insert_extract_v4sf_p9): Likewise.
12044 (vsx_insert_extract_v4sf_p9_2): Likewise.
12046 2017-07-25 Carl Love <cel@us.ibm.com>
12048 * doc/extend.texi: Update the built-in documentation file for the
12049 existing built-in functions
12050 vector signed char vec_cnttz (vector signed char);
12051 vector unsigned char vec_cnttz (vector unsigned char);
12052 vector signed short vec_cnttz (vector signed short);
12053 vector unsigned short vec_cnttz (vector unsigned short);
12054 vector signed int vec_cnttz (vector signed int);
12055 vector unsigned int vec_cnttz (vector unsigned int);
12056 vector signed long long vec_cnttz (vector signed long long);
12057 vector unsigned long long vec_cnttz (vector unsigned long long);
12059 2017-07-25 Andrew Pinski <apinski@cavium.com>
12061 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12062 accesses where the use is for the first operand of a BIT_INSERT.
12064 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12067 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12068 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12070 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12072 * config/i386/gstabs.h: Delete.
12073 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12075 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12077 * config/i386/i386.c (ix86_decompose_address): Do not check for
12078 register RTX when looking at index_reg or base_reg.
12079 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12081 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12083 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12084 to update EH info here.
12086 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12088 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12090 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12092 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12094 2017-07-25 Torsten Duwe <duwe@suse.de>
12096 * common.opt: Introduce -fpatchable-function-entry
12097 command line option, and its variables function_entry_patch_area_size
12098 and function_entry_patch_area_start.
12099 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12100 including a two-value parser.
12101 * target.def (print_patchable_function_entry): New target hook.
12102 * targhooks.h (default_print_patchable_function_entry): New function.
12103 * targhooks.c (default_print_patchable_function_entry): Likewise.
12104 * toplev.c (process_options): Switch off IPA-RA if
12105 patchable function entries are being generated.
12106 * varasm.c (assemble_start_function): Look at the
12107 patchable-function-entry command line switch and current
12108 function attributes and maybe generate NOP instructions by
12109 calling the print_patchable_function_entry hook.
12110 * doc/extend.texi: Document patchable_function_entry attribute.
12111 * doc/invoke.texi: Document -fpatchable_function_entry
12112 command line option.
12113 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12115 * doc/tm.texi: Re-generate.
12117 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12120 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12121 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12123 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12125 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12127 2017-07-25 Richard Biener <rguenther@suse.de>
12129 PR tree-optimization/81455
12130 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12131 not walk in cycles when looking for guards.
12133 2017-07-25 Richard Biener <rguenther@suse.de>
12135 PR tree-optimization/81529
12136 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12137 when optimizing backedge uses.
12139 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12141 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12143 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12144 to dl_section_ref. On AIX, append an expression to subtract
12145 the size of the section length to dl_section_ref.
12147 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12149 * configure.ac: If any of the config.* scripts fail, exit 1.
12150 * configure: Regenerate.
12152 2017-07-25 Richard Biener <rguenther@suse.de>
12154 PR middle-end/81546
12155 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12156 of immediate uses, be more verbose on errors.
12158 2017-07-25 Richard Biener <rguenther@suse.de>
12160 PR tree-optimization/81510
12161 * tree-vect-loop.c (vect_is_simple_reduction): When the
12162 reduction stmt is not inside the loop bail out.
12164 2017-07-25 Richard Biener <rguenther@suse.de>
12166 PR tree-optimization/81303
12167 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12168 profitability check against LOOP_VINFO_NITERSM1.
12170 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12172 * domwalk.c (cmp_bb_postorder): Simplify.
12173 (sort_bbs_postorder): New function. Use it...
12174 (dom_walker::walk): ...here to optimize common cases.
12176 2017-07-25 Martin Liska <mliska@suse.cz>
12179 * ipa-visibility.c (function_and_variable_visibility): Make the
12180 redirection just on target that supports aliasing.
12181 Fix GNU coding style.
12183 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12186 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12188 * config/arm/rtems.h: Likewise.
12189 * config/bfin/rtems.h: Likewise.
12190 * config/i386/rtemself.h: Likewise.
12191 * config/lm32/rtems.h: Likewise.
12192 * config/m32c/rtems.h: Likewise.
12193 * config/m68k/rtemself.h: Likewise.
12194 * config/microblaze/rtems.h: Likewise.
12195 * config/mips/rtems.h: Likewise.
12196 * config/moxie/rtems.h: Likewise.
12197 * config/nios2/rtems.h: Likewise.
12198 * config/powerpcspe/rtems.h: Likewise.
12199 * config/rs6000/rtems.h: Likewise.
12200 * config/rtems.h: Likewise.
12201 * config/sh/rtems.h: Likewise.
12202 * config/sh/rtemself.h: Likewise.
12203 * config/sparc/rtemself.h: Likewise.
12205 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12208 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12209 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12210 * tree-ssa-structalias.c (alias_get_name): Same.
12212 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12215 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12216 instructions if no du chain is found.
12218 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12220 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12222 2017-07-25 Richard Biener <rguenther@suse.de>
12224 PR middle-end/81505
12225 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12228 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12230 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12231 upper-regs options.
12232 (ISA_2_7_MASKS_SERVER): Likewise.
12233 (ISA_3_0_MASKS_IEEE): Likewise.
12234 (OTHER_P8_VECTOR_MASKS): Likewise.
12235 (OTHER_VSX_VECTOR_MASKS): Likewise.
12236 (POWERPC_MASKS): Likewise.
12237 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12238 duplicate list of options.
12239 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12240 explicit -mupper-regs options.
12241 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12242 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12243 alias for -mupper-regs-df.
12244 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12245 (rs6000_init_hard_regno_mode_ok): Likewise.
12246 (rs6000_option_override_internal): Likewise.
12247 (rs6000_opt_masks): Likewise.
12248 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12249 options in terms of whether -mvsx or -mpower8-vector was used.
12250 (TARGET_UPPER_REGS_DI): Likewise.
12251 (TARGET_UPPER_REGS_SF): Likewise.
12252 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12253 -mupper-regs-* options.
12255 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12257 * passes.c (emergency_dump_function): Print some empty lines and a
12258 header before the RTL dump.
12260 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12262 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12264 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12267 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12268 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12270 2017-07-24 Carl Love <cel@us.ibm.com>
12272 * config/rs6000/rs6000-c.c: Add support for built-in functions
12273 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12274 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12275 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12276 vec_extract_fp_from_shortl): Add defines for the two builtins.
12277 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12278 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12280 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12281 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12282 * doc/extend.texi: Update the built-in documentation file for the
12283 new built-in function.
12285 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12288 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12290 * doc/generic.texi: Likewise.
12291 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12292 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12294 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12296 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12297 (aarch64_mls_elt_merge<mode>): Likewise.
12299 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12301 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12302 having __cxa_atexit.
12304 2017-07-23 Michael Collison <michael.collison@arm.com>
12306 * config/arm/arm.c (arm_option_override): Deprecate
12307 use of -mstructure-size-boundary.
12308 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12309 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12311 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12314 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12315 Reduce cost estimate for direct moves.
12317 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12320 * config/i386/i386.c (ix86_option_override_internal): Disable
12321 BMI, BMI2 and TBM instructions for -m16.
12323 2017-07-21 Carl Love <cel@us.ibm.com>
12325 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12326 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12327 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12328 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12329 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12330 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12331 VMULOSW): New enum "unspec" values.
12332 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12333 altivec_vmulosw): New patterns.
12334 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12335 VMULOSW): Add definitions.
12337 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12339 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12340 (qdf24xx): Likewise.
12341 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12342 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12343 (AARCH64_FL_V8_1): Renumber.
12344 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12345 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12346 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12347 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12348 rdma to feature modifiers list.
12350 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12352 PR middle-end/56727
12353 * ipa-visibility (function_and_variable_visibility): Convert
12354 recursive PLT call to direct call if appropriate.
12356 2017-07-21 Andrew Pinski <apinski@cavium.com>
12358 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12359 operand 1 to see if the types precision matches.
12360 * fold-const.c (operand_equal_p): Likewise.
12362 2017-07-21 Richard Biener <rguenther@suse.de>
12364 PR tree-optimization/81303
12365 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12366 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12367 (vect_peeling_hash_get_lowest_cost): Adjust.
12368 (vect_enhance_data_refs_alignment): Likewise. Use
12369 vect_get_peeling_costs_all_drs to compute the penalty for no
12370 peeling to match up costs.
12372 2017-07-21 Richard Biener <rguenther@suse.de>
12374 PR tree-optimization/81500
12375 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12376 we didn't identify a reduction path.
12378 2017-07-21 Tom de Vries <tom@codesourcery.com>
12379 Cesar Philippidis <cesar@codesourcery.com>
12381 PR gcov-profile/81442
12382 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12385 2017-07-21 Tom de Vries <tom@codesourcery.com>
12388 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12390 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12391 nvptx_override_options_after_change.
12393 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12395 * dwarf2out.c (output_file_names): Avoid double testing for
12396 dwarf_version >= 5.
12398 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12400 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12402 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12404 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12406 (try_crossjump_to_edge): Do not punt on partitioned functions.
12408 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12410 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12411 Put all BBs reachable only via paths crossing cold region to cold
12413 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12415 2016-07-21 Richard Biener <rguenther@suse.de>
12417 PR tree-optimization/81303
12418 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12419 into account prologue and epilogue iterations when raising
12420 min_profitable_iters to sth at least covering one vector iteration.
12422 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12424 * config/arm/arm.c (arm_test_cpu_arch_dat):
12427 2017-07-20 Nathan Sidwell <nathan@acm.org>
12429 Remove TYPE_METHODS.
12430 * tree.h (TYPE_METHODS): Delete.
12431 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12432 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12433 (dbxout_type_methods): Scan TYPE_FIELDS.
12434 (dbxout_type): Don't check TYPE_METHODS here.
12435 * function.c (use_register_for_decl): Always ignore register for
12436 class types when not optimizing.
12437 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12438 * tree.c (free_lang_data_in_type): Stitch out member functions and
12439 templates from TYPE_FIELDS.
12440 (build_distinct_type_copy, verify_type_variant,
12441 verify_type): Member fns are on TYPE_FIELDS.
12442 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12443 * tree-pretty-print.c (dump_generic_node): Likewise.
12445 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12448 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12449 V2TImode and V4TImode.
12450 (ix86_expand_vector_extract): Likewise.
12451 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12452 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12453 (ssescalarmode): Handle V4TImode and V2TImode.
12454 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12455 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12456 (VEXTRACTI128_MODE): New mode iterator.
12457 (splitter for *vec_extractv?ti first element): New.
12458 (VEC_INIT_MODE): New mode iterator.
12459 (vec_init<mode>): Consolidate 3 expanders into one using
12460 VEC_INIT_MODE mode iterator.
12462 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12464 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12465 non_spilled_static_chain_regno_p.
12467 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12469 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12471 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12473 * bb-reorder.c (connect_traces): Allow copying of blocks within
12476 2017-07-20 Richard Biener <rguenther@suse.de>
12478 * gimple.h (gimple_phi_result): Add gphi * overload.
12479 (gimple_phi_result_ptr): Likewise.
12480 (gimple_phi_arg): Likewise. Adjust index assert to only
12481 allow actual argument accesses rather than all slots available
12483 (gimple_phi_arg_def): Add gphi * overload.
12484 * tree-phinodes.c (make_phi_node): Initialize only actual
12486 (resize_phi_node): Clear memory not covered by old node,
12487 do not initialize excess argument slots.
12488 (reserve_phi_args_for_new_edge): Initialize new argument slot
12491 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12493 PR tree-optimization/81388
12495 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12497 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12498 by removing computation of may_be_zero.
12500 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12501 Tom de Vries <tom@codesourcery.com>
12503 PR middle-end/81030
12504 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12505 when gimple level profile disagrees with what RTL expander did.
12507 2017-07-20 Richard Biener <rguenther@suse.de>
12509 PR tree-optimization/61171
12510 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12511 (vect_analyze_stmt): Add slp instance parameter.
12512 (vectorizable_reduction): Likewise.
12513 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12514 (vect_is_simple_reduction): Deal with chains not detected
12515 as SLP reduction chain, specifically not properly associated
12516 chains containing a mix of plus/minus.
12517 (get_reduction_op): Remove.
12518 (get_initial_defs_for_reduction): Simplify, pass in whether
12519 this is a reduction chain, pass in the SLP node for the PHIs.
12520 (vect_create_epilog_for_reduction): Get the SLP instance as
12522 (vectorizable_reduction): Get the SLP instance as arg.
12523 During analysis remember the SLP node with the PHIs in the
12524 instance. Simplify getting at the vectorized reduction PHIs.
12525 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12526 through SLP instance.
12527 (vect_slp_analyze_operations): Likewise.
12528 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12529 (vect_transform_stmt): Likewise.
12531 2017-07-20 Tom de Vries <tom@codesourcery.com>
12533 PR tree-optimization/81489
12534 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12535 read of phi arg location to before loop that modifies phi.
12537 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12539 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12542 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12544 PR middle-end/81331
12545 * except.c (execute): Fix ordering issue.
12547 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12549 PR rtl-optimization/81423
12550 * combine.c (make_compound_operation_int): Don't try to optimize
12551 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12553 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12555 PR rtl-optimization/81423
12556 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12557 with a constant that is -1 in the truncated to mode.
12559 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12561 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12562 (determine_unlikely_bbs): ... here.
12563 * predict.h (propagate_unlikely_bbs_forward): Declare.
12564 * cfgexpand.c (pass_expand::execute): Use it.
12565 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12567 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12568 propagate_unlikely_bbs_forward.
12570 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12572 PR middle-end/81331
12573 * except.c (maybe_add_nop_after_section_switch): New function.
12576 2017-07-19 Tom de Vries <tom@codesourcery.com>
12578 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12580 2017-07-19 Tom de Vries <tom@codesourcery.com>
12582 * gimple.h (gimple_phi_arg): Make assert more strict.
12584 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12586 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12587 * config/rs6000/mmintrin.h: New file.
12588 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12590 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12592 PR tree-optimization/81346
12593 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12595 2017-07-19 Tom de Vries <tom@codesourcery.com>
12597 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12599 2017-07-19 Tom de Vries <tom@codesourcery.com>
12601 * config/nvptx/nvptx-modes.def: Add V2DImode.
12602 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12603 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12604 (nvptx_output_mov_insn): Handle lack of mov.b128.
12605 (nvptx_print_operand): Handle 'H' and 'L' codes.
12606 (nvptx_vector_mode_supported): Allow V2DImode.
12607 (nvptx_preferred_simd_mode): New function.
12608 (nvptx_data_alignment): New function.
12609 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12610 nvptx_preferred_simd_mode.
12611 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12613 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12615 2017-07-19 Tom de Vries <tom@codesourcery.com>
12617 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12618 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12619 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12620 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12621 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12622 (mov<VECIM>_insn): New define_insn.
12623 (define_expand "mov<VECIM>): New define_expand.
12625 2017-07-19 Tom de Vries <tom@codesourcery.com>
12627 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12629 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12631 PR tree-optimization/81346
12632 * fold-const.h (fold_div_compare, range_check_type): Declare.
12633 * fold-const.c (range_check_type): New function.
12634 (build_range_check): Use range_check_type.
12635 (fold_div_compare): No longer static, rewritten into
12636 a match.pd helper function.
12637 (fold_comparison): Don't call fold_div_compare here.
12638 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12639 as helper function.
12641 2017-07-19 Nathan Sidwell <nathan@acm.org>
12643 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12644 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12645 * tree.c (find_decls_types_r, verify_type): Use
12646 TYPE_{MIN,MAX}_VALUE_RAW.
12647 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12648 (hash_tree): Likewise.
12649 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12651 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12654 2017-07-18 Tom de Vries <tom@codesourcery.com>
12656 PR middle-end/81464
12657 * omp-expand.c (expand_omp_for_static_chunk): Handle
12658 equal-argument loop exit phi.
12660 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12663 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12664 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12665 operand 2 predicate.
12666 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12667 operand 2 predicate.
12668 (ror,rol -> rorx splitters): Use const_int_operand as
12669 operand 2 predicate.
12671 2017-06-18 Richard Biener <rguenther@suse.de>
12673 PR tree-optimization/81410
12674 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12675 the gap in the ! slp_perm SLP case after each group.
12677 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12679 PR middle-end/81463
12680 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12683 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12685 PR middle-end/81462
12686 * predict.c (set_even_probabilities): Cleanup; do not affect
12687 probabilities that are already known.
12688 (combine_predictions_for_bb): Call even when count is set.
12690 2017-07-18 Nathan Sidwell <nathan@acm.org>
12692 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12695 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12698 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12699 optimization for loop niter analysis.
12701 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12704 * config/avr/avr.c (avr_optimize_casesi): Don't use
12705 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12707 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12709 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12710 body_cost_vec from _vect_peel_extended_info.
12711 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12712 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12715 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12717 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12719 2017-07-18 Richard Biener <rguenther@suse.de>
12721 PR tree-optimization/80620
12722 PR tree-optimization/81403
12723 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12724 info when re-using a VN table entry.
12726 2017-07-18 Richard Biener <rguenther@suse.de>
12728 PR tree-optimization/81418
12729 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12730 vectype_in. Verify that with lane-reducing reduction operations
12731 we have a single def-use cycle.
12733 2017-07-17 Carl Love <cel@us.ibm.com>
12735 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12737 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12738 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12739 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12740 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12741 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12742 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12743 VMULOSW): New enum "unspec" values.
12744 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12745 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12746 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12747 altivec_vmulosw): New patterns.
12748 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12749 VMULOSW): Add definitions.
12750 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12752 * config/alpha/alpha.c: Include predict.h.
12754 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12756 * tree-vrp.c (compare_assert_loc): Fix comparison function
12757 to return predictable results.
12759 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12761 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12763 (subdi3): Likewise.
12764 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12765 * doc/invoke.texi (mexpand-adddi): Update text.
12767 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12769 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12770 that also clobbers the CC register. The old expand code is moved
12772 (*arc_clzsi2): ... here.
12773 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12774 the CC register. The old expand code is moved to ...
12775 (arc_ctzsi2): ... here.
12777 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12779 * config/arc/arc.opt (mindexed-loads): Use initial value
12780 TARGET_INDEXED_LOADS_DEFAULT.
12781 (mauto-modify-reg): Use initial value
12782 TARGET_AUTO_MODIFY_REG_DEFAULT.
12783 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12784 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12785 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12786 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12788 2017-07-17 Martin Liska <mliska@suse.cz>
12791 * opts.c (finish_options): Do not allow -fgnu-tm
12792 w/ -fsanitize={kernel-,}address. Say sorry.
12794 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12797 * tree-loop-distribution.c (classify_partition): Only assert on
12798 numer of iterations.
12799 (merge_dep_scc_partitions): Delete prameter. Update function call.
12800 (distribute_loop): Remove code handling loop with unknown niters.
12801 (pass_loop_distribution::execute): Skip loop with unknown niters.
12803 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12806 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12807 function sort_partitions_by_post_order.
12809 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12811 PR tree-optimization/81374
12812 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12813 the max index of basic blocks, rather than number of basic blocks.
12815 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12817 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12819 (arc_legitimate_pic_operand_p): Likewise.
12820 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12822 (arc_needs_pcl_p): Likewise.
12823 (arc_legitimate_pc_offset_p): Likewise.
12824 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12825 function is also used in constrains.md.
12826 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12827 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12828 PLUS. Only return true/false in known cases, otherwise assert.
12829 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12830 is already called in arc_legitimate_constant_p.
12831 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12833 (LEGITIMATE_PIC_OPERAND_P): Use
12834 arc_raw_symbolic_reference_mentioned_p function.
12835 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12840 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12841 Andrew Burgess <andrew.burgess@embecosm.com>
12843 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12844 (arc_return_address_register): New function.
12845 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12846 (arc_handle_fndecl_attribute): Add naked attribute.
12847 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12848 (TARGET_WARN_FUNC_RETURN): Likewise.
12849 (arc_allocate_stack_slots_for_args): New function.
12850 (arc_warn_func_return): Likewise.
12851 (machine_function): Change type fn_type.
12852 (arc_compute_function_type): Consider new naked function type,
12853 change function return type.
12854 (arc_must_save_register): Adapt to handle new
12855 arc_compute_function_type's return type.
12856 (arc_expand_prologue): Likewise.
12857 (arc_expand_epilogue): Likewise.
12858 (arc_return_address_regs): Delete.
12859 (arc_return_address_register): New function.
12860 (arc_epilogue_uses): Use above function.
12861 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12862 (arc_function_type): Change encoding, add naked type.
12863 (ARC_INTERRUPT_P): Change to handle the new encoding.
12864 (ARC_FAST_INTERRUPT_P): Likewise.
12865 (ARC_NORMAL_P): Define.
12866 (ARC_NAKED_P): Likewise.
12867 (arc_compute_function_type): Delete prototype.
12868 * config/arc/arc.md (in_ret_delay_slot): Use
12869 arc_return_address_register function.
12870 (simple_return): Likewise.
12871 (p_return_i): Likewise.
12873 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12875 PR tree-optimization/81428
12876 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12877 can't be built for those types.
12879 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12881 Remove stuff dead since r239246.
12883 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12884 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12885 (avr_inform_devices): Remove dead stuff.
12887 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12889 * config/arm/arm_neon.h: Fix softp typo.
12891 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12893 PR tree-optimization/81365
12894 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12895 aggregate moves onto bb predecessor edges, make sure there are no
12896 loads that could alias the lhs in between the start of bb and the
12899 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12902 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12903 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12904 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12906 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12908 PR tree-optimization/81396
12909 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12910 (init_symbolic_number): Initialize it to 1.
12911 (perform_symbolic_merge): Add n_ops from both operands into the new
12913 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12914 without base_addr as useless if they need more than one operation.
12915 (bswap_replace): Handle !bswap case for NULL base_addr.
12917 2017-07-17 Tom de Vries <tom@codesourcery.com>
12920 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12923 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12925 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12926 conditional builtin define __FIX_LEON3FT_B2BST.
12928 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12930 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12931 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12934 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12936 PR rtl-optimization/81424
12937 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12938 to remove potential trapping from operands if -fnon-call-exceptions.
12940 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12942 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12943 profile_proability for scalling.
12944 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12946 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12948 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12950 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12952 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12953 fixpoint arithmetics.
12955 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12957 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12958 fixpoint arithmetics.
12960 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12962 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12963 fixpoint arithmetics.
12965 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12967 * profile-count.h (profile_probability::from_reg_br_prob_note,
12968 profile_probability::to_reg_br_prob_note): New functions.
12969 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12970 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12971 * predict.c (probability_reliable_p): Update.
12972 (edge_probability_reliable_p): Update.
12973 (br_prob_note_reliable_p): Update.
12974 (invert_br_probabilities): Update.
12975 (add_reg_br_prob_note): New function.
12976 (combine_predictions_for_insn): Update.
12977 * asan.c (asan_clear_shadow): Update.
12978 * cfgbuild.c (compute_outgoing_frequencies): Update.
12979 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12980 (update_br_prob_note): Update.
12981 (rtl_verify_edges): Update.
12982 (purge_dead_edges): Update.
12983 (fixup_reorder_chain): Update.
12984 * emit-rtl.c (try_split): Update.
12985 * ifcvt.c (cond_exec_process_insns): Update.
12986 (cond_exec_process_if_block): Update.
12987 (dead_or_predicable): Update.
12988 * internal-fn.c (expand_addsub_overflow): Update.
12989 (expand_neg_overflow): Update.
12990 (expand_mul_overflow): Update.
12991 * loop-doloop.c (doloop_modify): Update.
12992 * loop-unroll.c (compare_and_jump_seq): Update.
12993 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12994 * predict.h: Update.
12995 * reorg.c (mostly_true_jump): Update.
12997 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12998 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12999 * config/arc/arc.c: (emit_unlikely_jump): Update.
13000 * config/arm/arm.c: (emit_unlikely_jump): Update.
13001 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13002 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13003 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13004 (ix86_print_operand): Update.
13005 (ix86_split_fp_branch): Update.
13006 (predict_jump): Update.
13007 * config/ia64/ia64.c (ia64_print_operand): Update.
13008 * config/mmix/mmix.c (mmix_print_operand): Update.
13009 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13010 (rs6000_expand_split_stack_prologue): Update.
13011 * config/rs6000/rs6000.c: Update.
13012 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13013 (s390_expand_vec_movstr): Update.
13014 (s390_expand_cs_tdsi): Update.
13015 (s390_expand_split_stack_prologue): Update.
13016 * config/sh/sh.c (sh_print_operand): Update.
13017 (expand_cbranchsi4): Update.
13018 (expand_cbranchdi4): Update.
13019 * config/sparc/sparc.c (output_v9branch): Update.
13020 * config/spu/spu.c (get_branch_target): Update.
13021 (ea_load_store_inline): Update.
13022 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13023 * config/tilepro/tilepro.c: Update.
13025 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13027 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13028 (gimplify_save_expr): Likewise.
13030 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13032 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13034 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13036 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13038 * timevar.def (TV_IPA_FNSUMMARY): Define.
13040 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13042 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13043 to back store errata sensitive sequence from being generated.
13044 (sqrtdf2_fix): Likewise.
13046 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13048 * tree-ssa-threadupdate.c (compute_path_counts,
13049 update_joiner_offpath_counts): Use profile_probability.
13051 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13054 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13056 * config/arm/arm-c.c (arm_cpu_builtins): Define
13057 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13059 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13061 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13062 array entries to represent __ieee128 versions of the
13063 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13064 scalar_extract_sig, and scalar_insert_exp built-in functions.
13065 (altivec_resolve_overloaded_builtin): Add special case handling
13066 for the __builtin_scalar_insert_exp function, as represented by
13067 the P9V_BUILTIN_VEC_VSIEDP constant.
13068 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13069 exponent support for __ieee128 argument.
13070 (VSESQP): Add scalar extract signature support for __ieee128
13072 (VSTDCNQP): Add scalar test negative support for __ieee128
13074 (VSIEQP): Add scalar insert exponent support for __int128 argument
13075 with __ieee128 result.
13076 (VSIEQPF): Add scalar insert exponent support for __ieee128
13077 argument with __ieee128 result.
13078 (VSTDCQP): Add scalar test data class support for __ieee128
13080 (VSTDCNQP): Add overload support for scalar test negative with
13081 __ieee128 argument.
13082 (VSTDCQP): Add overload support for scalar test data class
13083 __ieee128 argument.
13084 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13085 UNSPEC_VSX_SXSIGDP.
13086 (UNSPEC_VSX_SIEXPQP): New constant.
13087 (xsxexpqp): New insn for VSX scalar extract exponent quad
13089 (xsxsigqp): New insn for VSX scalar extract significand quad
13091 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13092 precision with floating point argument.
13093 (xststdcqp): New expand for VSX scalar test data class quad
13095 (xststdcnegqp): New expand for VSX scalar test negative quad
13097 (xststdcqp): New insn to match expansions for VSX scalar test data
13098 class quad precision and VSX scalar test negative quad precision.
13099 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13100 special case operand checking to enforce that second operand of
13101 VSX scalar test data class with quad precision argument is a 7-bit
13103 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13104 prototypes and descriptions of __ieee128 versions of
13105 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13106 scalar_test_data_class, and scalar_test_neg built-in functions.
13108 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13110 PR tree-optimization/81162
13111 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13112 replace a negate with an add.
13114 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13116 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13118 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13120 * config/arm/arm-c.c (arm_cpu_builtins): Define
13121 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13123 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13125 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13126 (armv8-r): Set ARM Cortex-R52 as default CPU.
13127 * config/arm/arm-tables.opt: Regenerate.
13128 * config/arm/arm-tune.md: Regenerate.
13129 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13131 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13132 extension for -mcpu=cortex-r52.
13134 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13136 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13137 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13138 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13139 (fp-armv8): Define it as FP_ARMv8 only.
13140 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13141 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13143 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13144 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13145 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13146 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13147 than TARGET_FPU_ARMV8.
13148 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13149 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13150 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13152 * config/arm/neon.md (neon_vrint): Likewise.
13153 (neon_vcvt): Likewise.
13154 (neon_<fmaxmin_op><mode>): Likewise.
13155 (<fmaxmin><mode>3): Likewise.
13156 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13157 * config/arm/predicates.md (arm_cond_move_operator): Check against
13158 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13160 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13162 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13163 to top of function.
13165 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13167 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13168 loop in comment with memset.
13170 2017-07-14 Martin Liska <mliska@suse.cz>
13172 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13173 * dwarf2out.c (is_java): Remove the function.
13174 (output_pubname): Remove usage of the function.
13175 (lower_bound_default): Remove usage of DW_LANG_Java.
13176 (gen_compile_unit_die): Likewise.
13177 * gcc.c: Remove compiler defaults for .java and .zip files.
13178 * gimple-expr.c (remove_suffix): Change as there's no longer
13179 extension than 4-letter one.
13180 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13181 (gimplify_save_expr): Likewise.
13182 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13183 as it's possible even for other languages than Java.
13184 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13185 * lto-opts.c (lto_write_options): Remove reference to Java.
13186 * opts.c (strip_off_ending): Update file extension handling.
13187 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13188 * tree-eh.c (lower_resx): Likewise.
13189 * tree.c (free_lang_data_in_type): Remove dead code.
13190 (find_decls_types_r): Likewise.
13191 (build_common_builtin_nodes): Remove Java from a comment.
13192 (verify_type): Remove dead code.
13193 * varasm.c (assemble_external): Remove Java from a comment.
13195 2017-07-14 Martin Liska <mliska@suse.cz>
13197 * opts.c (finish_options): Add quotes.
13198 (common_handle_option): Likewise.
13200 2017-07-14 Martin Liska <mliska@suse.cz>
13202 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13203 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13204 Remove N_SO_PASCAL.
13205 * dwarf2out.c (lower_bound_default): Do not handle
13207 (gen_compile_unit_die): Likewise.
13208 * gcc.c: Remove default extension binding for GNU Pascal.
13209 * stmt.c: Remove Pascal language from a comment.
13210 * xcoffout.c: Likewise.
13212 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13215 * diagnostic-show-locus.c (fixit_cmp): New function.
13216 (layout::layout): Sort m_fixit_hints.
13217 (column_range::column_range): Assert that the values are valid.
13218 (struct char_span): New struct.
13219 (correction::overwrite): New method.
13220 (struct source_line): New struct.
13221 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13222 calls in terms of classes source_line and char_span, and
13223 correction::overwrite.
13224 (selftest::test_overlapped_fixit_printing_2): New function.
13225 (selftest::diagnostic_show_locus_c_tests): Call it.
13227 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13229 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13230 early if there is no lhs.
13232 2017-07-13 Martin Liska <mliska@suse.cz>
13234 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13235 (gen_reference_type_die): Likewise.
13236 * stor-layout.c: Remove Pascal-related comment.
13238 2017-07-13 Martin Liska <mliska@suse.cz>
13240 * opts.c (finish_options): Add quotes to error messages.
13241 (parse_sanitizer_options): Likewise.
13243 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13245 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13247 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13249 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13251 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13253 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13255 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13257 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13260 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13261 provides the hardware capability bits, define the macro
13262 __BUILTIN_CPU_SUPPORTS__.
13263 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13264 if GLIBC does not provide the hardware capability bits. Add a
13265 gcc_unreachable call if the built-in cpu function is neither
13266 __builtin_cpu_is nor __builtin_cpu_supports.
13267 (rs6000_get_function_versions_dispatcher): Change the warning
13268 that an old GLIBC is used which does not export the capability
13269 bits to be an error.
13270 * doc/extend.texi (target_clones attribute): Document the
13271 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13272 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13273 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13274 the macros defined by GCC if the newer GLIBC is available.
13276 2017-07-12 Jeff Law <law@redhat.com>
13278 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13279 remaining includes slightly.
13280 * config/riscv/riscv-builtins.c: Include profile-count.h.
13282 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13285 * config/avr/avr.c (avr_set_current_function): In diagnostic
13286 messages: Quote keywords and (parts of) identifiers.
13287 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13290 2017-07-12 Carl Love <cel@us.ibm.com>
13292 * config/rs6000/rs6000-c.c: Add support for built-in functions
13293 vector bool char vec_revb (vector bool char);
13294 vector bool short vec_revb (vector short char);
13295 vector bool int vec_revb (vector bool int);
13296 vector bool long long vec_revb (vector bool long long);
13297 * doc/extend.texi: Update the built-in documentation file for the
13298 new built-in functions.
13300 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13302 * config/s390/s390.md: Remove movcc splitter.
13304 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13306 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13307 load/store on condition.
13309 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13312 * config/avr/avr.c (avr_encode_section_info)
13313 [progmem && !TREE_READONLY]: Error if progmem object needs
13316 2017-07-11 Michael Collison <michael.collison@arm.com>
13318 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13321 2017-07-11 Carl Love <cel@us.ibm.com>
13323 * config/rs6000/rs6000-c.c: Add support for builtins
13324 vector unsigned int vec_parity_lsbb (vector signed int);
13325 vector unsigned int vec_parity_lsbb (vector unsigned int);
13326 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13327 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13328 vector unsigned long long vec_parity_lsbb (vector signed long long);
13329 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13330 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13331 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13332 * doc/extend.texi: Update the built-in documentation file for the
13333 new built-in functions.
13335 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13337 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13338 (layout::m_primary_loc): New field.
13339 (layout::layout): Initialize new field. Move location filtering
13340 logic from here to...
13341 (layout::maybe_add_location_range): ...this new method. Add
13342 support for filtering to just the lines already specified by other
13344 (layout::will_show_line_p): New method.
13345 (gcc_rich_location::add_location_if_nearby): New method.
13346 (selftest::test_add_location_if_nearby): New test function.
13347 (selftest::diagnostic_show_locus_c_tests): Call it.
13348 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13351 2017-07-11 Tom de Vries <tom@codesourcery.com>
13353 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13354 (bb_first_real_insn): New function.
13355 (nvptx_single): Add extra initialization of broadcasted condition
13358 2017-07-11 Nathan Sidwell <nathan@acm.org>
13360 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13362 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13364 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13365 Binutils doc as TEXI will mess them up.
13366 * doc/invoke.texi (AVR Options): Same here.
13368 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13370 * config/sparc/sparc.opt (mfix-ut700): New option.
13371 (mfix-gr712rc): Likewise.
13372 (sparc_fix_b2bst): New variable.
13373 * doc/invoke.texi (SPARC options): Document them.
13374 (ARM options): Fix warnings.
13375 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13376 instructions to prevent sequences that can trigger the store-store
13377 errata for certain LEON3FT processors.
13378 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13379 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13380 * config/sparc/sparc.md (fix_b2bst): New attribute.
13381 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13383 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13386 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13388 (*rsqrtsf2_sse): Ditto.
13390 (div<mode>3): Macroize insn from divdf3 and divsf3
13391 using MODEF mode iterator.
13393 2017-07-10 Martin Sebor <msebor@redhat.com>
13395 PR tree-optimization/80397
13396 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13397 instead of testing for equality to INTEGER_TYPE.
13399 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13401 * config.gcc: Remove uclibc from arc target spec.
13403 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13405 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13407 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13410 * lto-wrapper.c (remove_option): New function.
13411 (merge_and_complain): Merge PIC/PIE options more realistically.
13413 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13415 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13419 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13420 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13421 * config.in: Regenerate.
13422 * configure: Regenerate.
13423 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13424 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13425 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13426 (TARGET_GASISR_PROLOGUES): ...target mask.
13427 * common/config/avr/avr-common.c
13428 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13429 Set -mgas-isr-prologues.
13430 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13431 INSERT_PASS_BEFORE for it.
13432 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13433 * config/avr/avr.c (avr_option_override)
13434 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13435 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13436 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13437 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13438 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13440 (make_avr_pass_pre_proep): New function.
13441 (emit_push_sfr) <treg>: Add argument to function and use it
13442 instead of TMP_REG.
13443 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13444 and set machine->gasisr.yes.
13445 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13446 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13447 __gcc_isr.n_pushed to .L__stack_usage.
13448 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13449 (avr_asm_final_postscan_insn): ...this new static function.
13450 * config/avr/avr.h (machine_function)
13451 <is_no_gccisr, use_L__stack_usage>: New fields.
13452 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13453 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13454 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13455 (gasisr, *gasisr): New expander and insn.
13456 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13457 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13458 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13460 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13462 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13465 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13467 Move jump-tables out of .text again.
13470 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13471 (ASM_OUTPUT_ADDR_VEC): New function.
13472 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13473 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13474 INSN_ADDRESSes as asm comment.
13475 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13476 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13477 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13478 * config/avr/avr.md (*tablejump): Adjust comment.
13479 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13480 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13482 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13483 (avr_output_addr_vec): New proto.
13484 (avr_log_t) <insn_addresses>: New field.
13486 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13489 * config/i386/i386.c (ix86_function_arg_advance): Set
13490 outgoing_args_on_stack to true if there are outgoing arguments
13492 (ix86_function_arg): Likewise.
13493 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13494 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13495 * config/i386/i386.h (machine_function): Add
13496 outgoing_args_on_stack.
13498 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13500 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13501 supporting pthreds.
13502 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13504 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13506 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13507 (REAL_H): Remove $(MACHMODE_H).
13508 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13510 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13511 $(MACHMODE_H) and double-int.h.
13512 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13514 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13517 2017-07-07 Andrew Pinski <apinski@cavium.com>
13519 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13520 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13522 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13524 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13525 Add warning if GCC was not configured to link against a GLIBC that
13526 exports the hardware capability bits.
13527 (make_resolver_func): Make resolver function private and not a
13528 COMDAT function. Create the name with clone_function_name instead
13529 of make_unique_name.
13532 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13533 correct operand in doing the split.
13535 2017-07-07 Carl Love <cel@us.ibm.com>
13537 * config/rs6000/rs6000-c: Add support for built-in function
13538 vector unsigned short vec_pack_to_short_fp32 (vector float,
13540 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13541 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13542 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13543 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13544 (convert_4f32_8i16): Add define_expand.
13545 * doc/extend.texi: Update the built-in documentation file for the
13546 new built-in function.
13548 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13550 * config/sparc/m8.md: New file.
13551 * config/sparc/sparc.md: Include m8.md.
13553 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13555 * config/sparc/sparc.opt: New option -mvis4b.
13556 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13557 (sparc_option_override): Handle VIS4B.
13558 (enum sparc_builtins): Define
13559 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13560 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13561 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13562 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13563 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13564 (check_constant_argument): New function.
13565 (sparc_vis_init_builtins): Define builtins
13566 __builtin_vis_dictunpack{8,16,32},
13567 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13568 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13569 __builtin_vis_fpcmpde{8,16,32}shl and
13570 __builtin_vis_fpcmpur{8,16,32}shl.
13571 (sparc_expand_builtin): Check that the constant operands to
13572 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13573 constant and in range.
13574 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13576 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13577 (SPARC_IMM5_P): Likewise.
13578 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13579 (enabled): Handle vis4b.
13580 (UNSPEC_DICTUNPACK): New unspec.
13581 (UNSPEC_FPCMPSHL): Likewise.
13582 (UNSPEC_FPUCMPSHL): Likewise.
13583 (UNSPEC_FPCMPDESHL): Likewise.
13584 (UNSPEC_FPCMPURSHL): Likewise.
13585 (cpu_feature): New CPU feature `vis4b'.
13586 (dictunpack{8,16,32}): New insns.
13587 (FPCSMODE): New mode iterator.
13588 (fpcscond): New code iterator.
13589 (fpcsucond): Likewise.
13590 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13591 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13592 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13593 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13594 * config/sparc/constraints.md: Define constraints `q' for unsigned
13595 2-bit integer constants and `t' for unsigned 5-bit integer
13597 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13599 (imm5_operand_dictunpack16): Likewise.
13600 (imm5_operand_dictunpack32): Likewise.
13601 (imm2_operand): Likewise.
13602 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13603 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13604 ditunpack* and fpcmp*shl builtins.
13606 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13608 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13609 * config.in: Add HAVE_AS_SPARC6 define.
13610 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13612 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13614 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13615 (CPP_CPU_SPEC): Handle m8.
13616 (ASM_CPU_SPEC): Likewise.
13617 * config/sparc/sparc-opts.h (enum processor_type): Add
13619 * config/sparc/sparc.c (m8_costs): New struct.
13620 (sparc_option_override): Handle TARGET_CPU_m8.
13621 (sparc32_initialize_trampoline): Likewise.
13622 (sparc64_initialize_trampoline): Likewise.
13623 (sparc_issue_rate): Likewise.
13624 (sparc_register_move_cost): Likewise.
13625 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13626 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13627 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13628 (CPP_CPU_SPEC): Handle M8.
13629 (ASM_CPU_SPEC): Likewise.
13630 (AS_M8_FLAG): Define.
13631 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13632 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13633 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13635 * configure: Regenerate.
13636 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13639 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13641 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13643 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13644 ("*movdi_insn_sp32"): Do not set v3pipe.
13645 ("*movsi_insn"): Likewise.
13646 ("*movdi_insn_sp64"): Likewise.
13647 ("*movsf_insn"): Likewise.
13648 ("*movdf_insn_sp32"): Likewise.
13649 ("*movdf_insn_sp64"): Likewise.
13650 ("*zero_extendsidi2_insn_sp64"): Likewise.
13651 ("*sign_extendsidi2_insn"): Likewise.
13652 ("*mov<VM32:mode>_insn"): Likewise.
13653 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13654 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13655 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13656 ("<vlop:code><VL:mode>3"): Likewise.
13657 ("*not_<vlop:code><VL:mode>3"): Likewise.
13658 ("*nand<VL:mode>_vis"): Likewise.
13659 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13660 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13661 ("one_cmpl<VL:mode>2"): Likewise.
13662 ("faligndata<VM64:mode>_vis"): Likewise.
13663 ("alignaddrsi_vis"): Likewise.
13664 ("alignaddrdi_vis"): Likweise.
13665 ("alignaddrlsi_vis"): Likewise.
13666 ("alignaddrldi_vis"): Likewise.
13667 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13668 ("bmaskdi_vis"): Likewise.
13669 ("bmasksi_vis"): Likewise.
13670 ("bshuffle<VM64:mode>_vis"): Likewise.
13671 ("cmask8<P:mode>_vis"): Likewise.
13672 ("cmask16<P:mode>_vis"): Likewise.
13673 ("cmask32<P:mode>_vis"): Likewise.
13674 ("pdistn<P:mode>_vis"): Likewise.
13675 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13677 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13679 * config/sparc/sparc.md ("subtype"): New insn attribute.
13680 ("*wrgsr_sp64"): Set insn subtype.
13681 ("*rdgsr_sp64"): Likewise.
13682 ("alignaddrsi_vis"): Likewise.
13683 ("alignaddrdi_vis"): Likewise.
13684 ("alignaddrlsi_vis"): Likewise.
13685 ("alignaddrldi_vis"): Likewise.
13686 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13687 ("fexpand_vis"): Likewise.
13688 ("fpmerge_vis"): Likewise.
13689 ("faligndata<VM64:mode>_vis"): Likewise.
13690 ("bshuffle<VM64:mode>_vis"): Likewise.
13691 ("cmask8<P:mode>_vis"): Likewise.
13692 ("cmask16<P:mode>_vis"): Likewise.
13693 ("cmask32<P:mode>_vis"): Likewise.
13694 ("fchksm16_vis"): Likewise.
13695 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13696 ("fmean16_vis"): Likewise.
13697 ("fp<plusminus_insn>64_vis"): Likewise.
13698 ("<plusminus_insn>v8qi3"): Likewise.
13699 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13700 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13701 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13702 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13703 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13704 ("*movqi_insn"): Likewise.
13705 ("*movhi_insn"): Likewise.
13706 ("*movsi_insn"): Likewise.
13707 ("movsi_pic_gotdata_op"): Likewise.
13708 ("*movdi_insn_sp32"): Likewise.
13709 ("*movdi_insn_sp64"): Likewise.
13710 ("movdi_pic_gotdata_op"): Likewise.
13711 ("*movsf_insn"): Likewise.
13712 ("*movdf_insn_sp32"): Likewise.
13713 ("*movdf_insn_sp64"): Likewise.
13714 ("*zero_extendhisi2_insn"): Likewise.
13715 ("*zero_extendqihi2_insn"): Likewise.
13716 ("*zero_extendqisi2_insn"): Likewise.
13717 ("*zero_extendqidi2_insn"): Likewise.
13718 ("*zero_extendhidi2_insn"): Likewise.
13719 ("*zero_extendsidi2_insn_sp64"): Likewise.
13720 ("ldfsr"): Likewise.
13721 ("prefetch_64"): Likewise.
13722 ("prefetch_32"): Likewise.
13723 ("tie_ld32"): Likewise.
13724 ("tie_ld64"): Likewise.
13725 ("*tldo_ldub_sp32"): Likewise.
13726 ("*tldo_ldub1_sp32"): Likewise.
13727 ("*tldo_ldub2_sp32"): Likewise.
13728 ("*tldo_ldub_sp64"): Likewise.
13729 ("*tldo_ldub1_sp64"): Likewise.
13730 ("*tldo_ldub2_sp64"): Likewise.
13731 ("*tldo_ldub3_sp64"): Likewise.
13732 ("*tldo_lduh_sp32"): Likewise.
13733 ("*tldo_lduh1_sp32"): Likewise.
13734 ("*tldo_lduh_sp64"): Likewise.
13735 ("*tldo_lduh1_sp64"): Likewise.
13736 ("*tldo_lduh2_sp64"): Likewise.
13737 ("*tldo_lduw_sp32"): Likewise.
13738 ("*tldo_lduw_sp64"): Likewise.
13739 ("*tldo_lduw1_sp64"): Likewise.
13740 ("*tldo_ldx_sp64"): Likewise.
13741 ("*mov<VM32:mode>_insn"): Likewise.
13742 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13743 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13745 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13747 * config/sparc/sparc.md ("type"): New insn type viscmp.
13748 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13750 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13751 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13752 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13753 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13755 ("n7_vis_logical_11cycle"): Likewise.
13756 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13757 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13758 * config/sparc/niagara.md ("niag_vis"): Likewise.
13759 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13760 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13762 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13764 * config/sparc/sparc.md: New instruction type `bmask'.
13765 (bmaskdi_vis): Use the `bmask' type.
13766 (bmasksi_vis): Likewise.
13767 * config/sparc/ultra3.md (us3_array): Likewise.
13768 * config/sparc/niagara7.md (n7_array): Likewise.
13769 * config/sparc/niagara4.md (n4_array): Likewise.
13770 * config/sparc/niagara2.md (niag2_vis): Likewise.
13771 (niag3_vis): Likewise.
13772 * config/sparc/niagara.md (niag_vis): Likewise.
13774 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13776 * ipa-comdats.c: Remove optimize check from gate.
13777 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13778 for functions not optimized.
13779 (ipa_fn_summary_read): Skip optimize check.
13780 (ipa_fn_summary_write): Likewise.
13781 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13783 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13785 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13786 (check_callers): Give up on caller which is not optimized.
13787 (inline_small_functions): Likewise.
13788 (ipa_inline): Do not give up when not optimizing.
13789 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13790 away unoptimizes cdtors.
13791 (whole_program_function_and_variable_visibility): Do
13792 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13793 * ipa.c (process_references): Do not check optimize.
13794 (symbol_table::remove_unreachable_nodes): Update optimize check.
13795 (set_writeonly_bit): Update optimize check.
13796 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13797 (pass_ipa_single_use::gate): Remove.
13799 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13801 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13802 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13803 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13804 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13805 permute_load, permute_store, adjust_extract, adjust_splat,
13806 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13807 replace_swap_with_copy, dump_swap_insn_table,
13808 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13809 recombine_lvx_pattern, recombine_stvx_pattern,
13810 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13811 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13812 to file rs6000-p8swap.c.
13813 * config/rs6000/rs6000-p8swap.c: New file.
13814 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13815 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13816 and rs6000*-*-* targets.
13818 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13820 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13822 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13824 * lto-wrapper.c (merge_and_complain): Do not merge
13825 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13826 fsigned_zeros, ftrapping_math, fwrapv.
13827 (append_compiler_options): Do not track these options.
13828 (append_linker_options): Likewie
13830 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13832 * cgraphunit.c (cgraph_node::finalize_function): When
13833 !flag_toplevel_reorde set no_reorder flag.
13834 (varpool_node::finalize_decl): Likewise.
13835 (symbol_table::compile): Drop no toplevel reorder path.
13837 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13839 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13840 edges; zero probability is not better than uninitialized.
13842 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13844 * asan.h (asan_sanitize_allocas_p): Declare.
13845 * asan.c (asan_sanitize_allocas_p): New function.
13846 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13847 (handle_builtin_alloca): Likewise.
13848 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13849 if !asan_sanitize_allocas_p.
13850 * params.def (asan-instrument-allocas): Add new option.
13851 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13852 * opts.c (common_handle_option): Disable allocas sanitization for
13855 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13857 * asan.c: Include gimple-fold.h.
13858 (get_last_alloca_addr): New function.
13859 (handle_builtin_stackrestore): Likewise.
13860 (handle_builtin_alloca): Likewise.
13861 (asan_emit_allocas_unpoison): Likewise.
13862 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13863 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13864 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13865 (instrument_builtin_call): Pass gimple iterator to
13866 get_mem_refs_of_builtin_call.
13867 (last_alloca_addr): New global.
13868 * asan.h (asan_emit_allocas_unpoison): Declare.
13869 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13870 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13871 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13872 if function calls alloca.
13873 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13874 * gimple-fold.h (replace_call_with_value): Declare.
13875 * internal-fn.c: Include asan.h.
13876 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13877 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13879 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13881 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13882 (C_SELFTEST_FLAGS): New.
13883 (CPP_SELFTEST_FLAGS): New.
13884 (SELFTEST_DEPS): New, from deps of s-selftest.
13885 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13886 (CPP_SELFTEST_DEPS): New.
13887 (selftest): Add dependency on s-selftest-c++.
13888 (s-selftest): Rename to...
13889 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13890 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13891 than SELFTEST_FLAGS.
13892 (selftest-gdb): Rename to...
13893 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13895 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13896 (selftest-valgrind): Rename to...
13897 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13899 (selftest-valgrind): Reintroduce as an alias for
13900 selftest-c-valgrind.
13901 (s-selftest-c++): New.
13902 (selftest-c++-gdb): New.
13903 (selftest-c++-valgrind): New.
13905 2017-07-06 Olivier Hainque <hainque@adacore.com>
13907 * gcc.c (process_command): When deciding if undefined variables
13908 should be ignored when processing specs, accept "gcc -v" as well.
13910 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13912 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13913 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13915 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13917 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13918 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13919 * config/arm/arm-tables.opt: Regenerate.
13920 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13922 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13924 2017-07-06 Carl Love <cel@us.ibm.com>
13926 * ChangeLog: Clean up from mid air collision
13928 2017-07-06 Carl Love <cel@us.ibm.com>
13930 * config/rs6000/rs6000-c.c: Add support for built-in functions
13931 vector signed int vec_subc (vector signed int, vector signed int);
13932 vector signed __int128 vec_subc (vector signed __int128,
13933 vector signed __int128);
13934 vector unsigned __int128 vec_subc (vector unsigned __int128,
13935 vector unsigned __int128);
13936 vector signed int vec_sube (vector signed int, vector signed int,
13937 vector signed int);
13938 vector unsigned int vec_sube (vector unsigned int,
13939 vector unsigned int,
13940 vector unsigned int);
13941 vector signed __int128 vec_sube (vector signed __int128,
13942 vector signed __int128,
13943 vector signed__int128);
13944 vector unsigned __int128 vec_sube (vector unsigned __int128,
13945 vector unsigned __int128,
13946 vector unsigned __int128);
13947 vector signed int vec_subec (vector signed int, vector signed int,
13948 vector signed int);
13949 vector unsigned int vec_subec (vector unsigned int,
13950 vector unsigned int,
13951 vector unsigned int);
13952 vector signed __int128 vec_subec (vector signed __int128,
13953 vector signed __int128,
13954 vector signed__int128);
13955 vector unsigned __int128 vec_subec (vector unsigned __int128,
13956 vector unsigned __int128,
13957 vector unsigned __int128);
13958 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13959 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13960 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13961 BU_ALTIVEC_OVERLOAD_X definitions.
13962 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13963 * doc/extend.texi: Update the built-in documentation file for the new
13964 built-in functions.
13966 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13969 * diagnostic-show-locus.c (layout::layout): Use start and finish
13970 spelling location for the start and finish of each range.
13971 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13972 Add unused aspect param.
13973 * input.c (expand_location_1): Add "aspect" param, and use it
13974 to access the correct part of the location.
13975 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13977 (expand_location_to_spelling_point): Likewise.
13978 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13979 param, and pass it to expand_location_1.
13981 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13983 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13984 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13985 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13986 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13987 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13988 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13989 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13990 _mm_maskz_getmant_ss): New intrinsics.
13991 (__builtin_ia32_getexpss128_mask): Changed to ...
13992 __builtin_ia32_getexpss128_round ... this.
13993 (__builtin_ia32_getexpsd128_mask): Changed to ...
13994 __builtin_ia32_getexpsd128_round ... this.
13995 * config/i386/i386-builtin-types.def
13996 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13997 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13998 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13999 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14000 __builtin_ia32_getmantss_mask_round): New builtins.
14001 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14002 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14003 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14004 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14005 * config/i386/sse.md
14006 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14007 avx512f_sgetexp<mode><mask_scalar_name>
14008 <round_saeonly_scalar_name> ... this.
14009 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14010 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14011 vgetexp<ssescalarmodesuffix>
14012 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14013 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14014 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14015 avx512f_vgetmant<mode><mask_scalar_name>
14016 <round_saeonly_scalar_name> ... this.
14017 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14018 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14019 vgetmant<ssescalarmodesuffix>
14020 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14021 %0<mask_scalar_operand4>, %1, %2
14022 <round_saeonly_scalar_mask_op4>, %3} ... this.
14023 * config/i386/subst.md (mask_scalar_operand4,
14024 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14025 round_saeonly_scalar_nimm_predicate): New subst attributes.
14027 2017-07-06 Julia Koval <julia.koval@intel.com>
14029 * config/i386/i386.c (ix86_erase_embedded_rounding):
14030 Remove code for old rounding pattern.
14032 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14034 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14036 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14038 * doc/sourcebuild.texi (Test Directives, Variants of
14039 dg-require-support): Add documentation for dg-require-stack-check.
14041 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14043 * config/i386/subst.md (mask_scalar, round_scalar,
14044 round_saeonly_scalar): New meta-templates.
14045 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14046 round_scalar_mask_operand3, round_scalar_mask_op3,
14047 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14048 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14049 round_saeonly_scalar_constraint,
14050 round_saeonly_scalar_prefix): New subst attribute.
14051 * config/i386/sse.md
14052 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14053 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14054 <round_scalar_name> ... this.
14055 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14056 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14057 <round_scalar_name> ... this.
14058 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14059 <sse>_vm<code><mode>3<mask_scalar_name>
14060 <round_saeonly_scalar_name> ... this.
14061 (v<plusminus_mnemonic><ssescalarmodesuffix>
14062 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14063 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14064 v<plusminus_mnemonic><ssescalarmodesuffix>
14065 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14066 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14067 (v<multdiv_mnemonic><ssescalarmodesuffix>
14068 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14069 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14070 v<multdiv_mnemonic><ssescalarmodesuffix>
14071 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14072 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14073 (v<maxmin_float><ssescalarmodesuffix>
14074 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14075 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14076 v<maxmin_float><ssescalarmodesuffix>
14077 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14078 %0<mask_scalar_operand3>, %1, %<iptr>2
14079 <round_saeonly_scalar_mask_op3>} ... this.
14081 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14083 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14084 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14086 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14087 Alan Hayward <alan.hayward@arm.com>
14088 David Sherwood <david.sherwood@arm.com>
14090 * combine.c (simplify_if_then_else): Remove "enum" before
14092 * compare-elim.c (can_eliminate_compare): Likewise.
14093 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14095 (aarch64_lookup_simd_builtin_type): Likewise.
14096 (aarch64_simd_builtin_type): Likewise.
14097 (aarch64_init_simd_builtin_types): Likewise.
14098 (aarch64_simd_expand_args): Likewise.
14099 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14101 (aarch64_reverse_mask): Likewise.
14102 (aarch64_simd_emit_reg_reg_move): Likewise.
14103 (aarch64_gen_adjusted_ldpstp): Likewise.
14104 (aarch64_ccmp_mode_to_code): Likewise.
14105 (aarch64_operands_ok_for_ldpstp): Likewise.
14106 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14107 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14109 (aarch64_min_divisions_for_recip_mul): Likewise.
14110 (aarch64_reassociation_width): Likewise.
14111 (aarch64_get_condition_code_1): Likewise.
14112 (aarch64_simd_emit_reg_reg_move): Likewise.
14113 (aarch64_simd_attr_length_rglist): Likewise.
14114 (aarch64_reverse_mask): Likewise.
14115 (aarch64_operands_ok_for_ldpstp): Likewise.
14116 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14117 (aarch64_gen_adjusted_ldpstp): Likewise.
14118 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14120 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14121 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14122 (arm_lookup_simd_builtin_type): Likewise.
14123 (arm_simd_builtin_type): Likewise.
14124 (arm_init_simd_builtin_types): Likewise.
14125 (arm_expand_builtin_args): Likewise.
14126 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14127 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14128 (ft32_setup_incoming_varargs): Likewise.
14129 (ft32_function_arg): Likewise.
14130 (ft32_function_arg_advance): Likewise.
14131 (ft32_pass_by_reference): Likewise.
14132 (ft32_arg_partial_bytes): Likewise.
14133 (ft32_valid_pointer_mode): Likewise.
14134 (ft32_addr_space_pointer_mode): Likewise.
14135 (ft32_addr_space_legitimate_address_p): Likewise.
14136 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14138 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14139 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14140 (iamcu_alignment): Likewise.
14141 (canonicalize_vector_int_perm): Likewise.
14142 (ix86_noce_conversion_profitable_p): Likewise.
14143 (ix86_mpx_bound_mode): Likewise.
14144 (ix86_operands_ok_for_move_multiple): Likewise.
14145 * config/microblaze/microblaze-protos.h
14146 (microblaze_expand_conditional_branch_reg): Likewise.
14147 * config/microblaze/microblaze.c
14148 (microblaze_expand_conditional_branch_reg): Likewise.
14149 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14151 (rs6000_reassociation_width): Likewise.
14152 (rs6000_invalid_binary_op): Likewise.
14153 (fusion_p9_p): Likewise.
14154 (emit_fusion_p9_load): Likewise.
14155 (emit_fusion_p9_store): Likewise.
14156 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14158 (riscv_hard_regno_mode_ok_p): Likewise.
14159 (riscv_address_insns): Likewise.
14160 (riscv_split_symbol): Likewise.
14161 (riscv_legitimize_move): Likewise.
14162 (riscv_function_value): Likewise.
14163 (riscv_hard_regno_nregs): Likewise.
14164 (riscv_expand_builtin): Likewise.
14165 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14166 (riscv_build_integer): Likewise.
14167 (riscv_split_integer): Likewise.
14168 (riscv_legitimate_constant_p): Likewise.
14169 (riscv_cannot_force_const_mem): Likewise.
14170 (riscv_regno_mode_ok_for_base_p): Likewise.
14171 (riscv_valid_base_register_p): Likewise.
14172 (riscv_valid_offset_p): Likewise.
14173 (riscv_valid_lo_sum_p): Likewise.
14174 (riscv_classify_address): Likewise.
14175 (riscv_legitimate_address_p): Likewise.
14176 (riscv_address_insns): Likewise.
14177 (riscv_load_store_insns): Likewise.
14178 (riscv_force_binary): Likewise.
14179 (riscv_split_symbol): Likewise.
14180 (riscv_force_address): Likewise.
14181 (riscv_legitimize_address): Likewise.
14182 (riscv_move_integer): Likewise.
14183 (riscv_legitimize_const_move): Likewise.
14184 (riscv_legitimize_move): Likewise.
14185 (riscv_address_cost): Likewise.
14186 (riscv_subword): Likewise.
14187 (riscv_output_move): Likewise.
14188 (riscv_canonicalize_int_order_test): Likewise.
14189 (riscv_emit_int_order_test): Likewise.
14190 (riscv_function_arg_boundary): Likewise.
14191 (riscv_pass_mode_in_fpr_p): Likewise.
14192 (riscv_pass_fpr_single): Likewise.
14193 (riscv_pass_fpr_pair): Likewise.
14194 (riscv_get_arg_info): Likewise.
14195 (riscv_function_arg): Likewise.
14196 (riscv_function_arg_advance): Likewise.
14197 (riscv_arg_partial_bytes): Likewise.
14198 (riscv_function_value): Likewise.
14199 (riscv_pass_by_reference): Likewise.
14200 (riscv_setup_incoming_varargs): Likewise.
14201 (riscv_print_operand): Likewise.
14202 (riscv_elf_select_rtx_section): Likewise.
14203 (riscv_save_restore_reg): Likewise.
14204 (riscv_for_each_saved_reg): Likewise.
14205 (riscv_register_move_cost): Likewise.
14206 (riscv_hard_regno_mode_ok_p): Likewise.
14207 (riscv_hard_regno_nregs): Likewise.
14208 (riscv_class_max_nregs): Likewise.
14209 (riscv_memory_move_cost): Likewise.
14210 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14211 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14212 (rl78_addr_space_address_mode): Likewise.
14213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14216 (rs6000_reassociation_width): Likewise.
14217 (rs6000_invalid_binary_op): Likewise.
14218 (fusion_p9_p): Likewise.
14219 (emit_fusion_p9_load): Likewise.
14220 (emit_fusion_p9_store): Likewise.
14221 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14222 (ok_for_simple_move_operands): Likewise.
14223 (ok_for_simple_move_strict_operands): Likewise.
14224 (ok_for_simple_arith_logic_operands): Likewise.
14225 (visium_legitimize_reload_address): Likewise.
14226 (visium_select_cc_mode): Likewise.
14227 (output_cbranch): Likewise.
14228 (visium_split_double_move): Likewise.
14229 (visium_expand_copysign): Likewise.
14230 (visium_expand_int_cstore): Likewise.
14231 (visium_expand_fp_cstore): Likewise.
14232 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14233 (visium_function_arg): Likewise.
14234 (visium_function_arg_advance): Likewise.
14235 (visium_libcall_value): Likewise.
14236 (visium_setup_incoming_varargs): Likewise.
14237 (visium_legitimate_constant_p): Likewise.
14238 (visium_legitimate_address_p): Likewise.
14239 (visium_legitimize_address): Likewise.
14240 (visium_secondary_reload): Likewise.
14241 (visium_register_move_cost): Likewise.
14242 (visium_memory_move_cost): Likewise.
14243 (prepare_move_operands): Likewise.
14244 (ok_for_simple_move_operands): Likewise.
14245 (ok_for_simple_move_strict_operands): Likewise.
14246 (ok_for_simple_arith_logic_operands): Likewise.
14247 (visium_function_value_1): Likewise.
14248 (rtx_ok_for_offset_p): Likewise.
14249 (visium_legitimize_reload_address): Likewise.
14250 (visium_split_double_move): Likewise.
14251 (visium_expand_copysign): Likewise.
14252 (visium_expand_int_cstore): Likewise.
14253 (visium_expand_fp_cstore): Likewise.
14254 (visium_split_cstore): Likewise.
14255 (visium_select_cc_mode): Likewise.
14256 (visium_split_cbranch): Likewise.
14257 (output_cbranch): Likewise.
14258 (visium_print_operand_address): Likewise.
14259 * expmed.c (flip_storage_order): Likewise.
14260 * expmed.h (emit_cstore): Likewise.
14261 (flip_storage_order): Likewise.
14262 * genrecog.c (validate_pattern): Likewise.
14263 * hsa-gen.c (gen_hsa_addr): Likewise.
14264 * internal-fn.c (expand_arith_overflow): Likewise.
14265 * ira-color.c (allocno_copy_cost_saving): Likewise.
14266 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14267 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14268 (process_invariant_for_inheritance): Likewise.
14269 * lra-eliminations.c (move_plus_up): Likewise.
14270 * omp-low.c (lower_oacc_reductions): Likewise.
14271 * simplify-rtx.c (simplify_subreg): Likewise.
14272 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14273 (TARGET_CHKP_BOUND_MODE): Likewise..
14274 * targhooks.c (default_chkp_bound_mode): Likewise.
14275 (default_setup_incoming_vararg_bounds): Likewise.
14276 * targhooks.h (default_chkp_bound_mode): Likewise.
14277 (default_setup_incoming_vararg_bounds): Likewise.
14278 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14279 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14280 (have_whole_vector_shift): Likewise.
14281 * tree-vect-stmts.c (vectorizable_load): Likewise.
14282 * doc/tm.texi: Regenerate.
14284 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14286 Graceful degrade if Binutils PR21472 is not available.
14289 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14290 .rodata in flash test fails.
14291 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14292 * confgure: Regenerate.
14293 * config.in: Regenerate.
14294 * config/avr/avr.c (avr_asm_named_section)
14295 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14296 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14297 (avr_asm_init_sections): Same.
14299 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14301 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14302 (fma<VH:mode>4_intrinsic): Likewise.
14303 (*fmsub<VCVTF:mode>4): Likewise.
14304 (*fmsub<VH:mode>4_intrinsic): Likewise.
14306 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14309 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14310 Don't depend on "optimize > 0".
14311 (out_movhi_r_mr, out_movqi_mr_r): Same.
14312 (out_movhi_mr_r, out_movqi_r_mr): Same.
14313 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14314 io_address_operand on "optimize > 0".
14316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14318 * tree-loop-distribution.c: Add general explanantion on the pass.
14319 (generate_loops_for_partition): Mark distributed loop.
14320 (pg_add_dependence_edges): New parameter. Handle alias data
14321 dependence specially and record it in the parameter if asked.
14322 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14323 (init_partition_graph_vertices, add_partition_graph_edge): New.
14324 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14325 (free_partition_graph_vdata, build_partition_graph): New.
14326 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14327 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14328 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14329 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14330 (version_for_distribution_p, finalize_partitions): New.
14331 (distribute_loop): Handle alias data dependence specially. Factor
14332 out loop fusion code as functions and call these functions.
14334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14336 * tree-loop-distribution.c (classify_partition): New parameter and
14337 better handle reduction statement.
14338 (rdg_build_partitions): Revise comment.
14339 (distribute_loop): Compute statements in all partitions and pass it
14340 to classify_partition.
14342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14344 * tree-loop-distribution.c (enum partition_type): New.
14345 (struct partition): New field type.
14346 (partition_merge_into): Add parameter. Update partition type.
14347 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14348 (build_rdg_partition_for_vertex): Compute partition type.
14349 (rdg_build_partitions): Dump partition type.
14350 (distribute_loop): Update calls to partition_merge_into.
14352 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14354 * tree-loop-distribution.c (struct ddr_hasher): New.
14355 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14357 (classify_partition): Call get_data_dependence.
14358 (pg_add_dependence_edges): Ditto.
14359 (distribute_loop): Release data dependence hash table.
14361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14363 * tree-loop-distribution.c (ref_base_address): Delete.
14364 (similar_memory_accesses): Rename ...
14365 (share_memory_accesses): ... to this. Check if partitions access
14366 the same memory reference.
14367 (distribute_loop): Call share_memory_accesses.
14369 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14371 * tree-loop-distribution.c (struct partition): New field recording
14372 its data reference.
14373 (partition_alloc, partition_free): Init and release data refs.
14374 (partition_merge_into): Merge data refs.
14375 (build_rdg_partition_for_vertex): Collect data refs for partition.
14376 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14378 (distribute_loop): Remve data refs from vertice data of partition
14381 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14383 * tree-loop-distribution.c (params.h): Include header file.
14384 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14385 (datarefs_vec): New global var.
14386 (create_rdg_vertices): Use datarefs_vec directly.
14387 (free_rdg): Don't free data references.
14388 (build_rdg): Update use. Don't free data references.
14389 (distribute_loop): Compute global variable for data references.
14390 Bail out if there are too many data references.
14392 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14394 * tree-loop-distribution.c (loop_nest): New global var.
14395 (build_rdg): Use loop directly, rather than loop nest.
14396 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14398 (distribute_loop): Compute global variable loop nest. Update use.
14400 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14402 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14403 (partition_merge_into): New parameter. Dump reason for fusion.
14404 (distribute_loop): Update use of partition_merge_into.
14406 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14408 * tree-loop-distribution.c (bb_top_order_index): New.
14409 (bb_top_order_index_size, bb_top_order_cmp): New.
14410 (stmts_from_loop): Use topological order.
14411 (pass_loop_distribution::execute): Compute and release topological
14412 order for basic blocks.
14414 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14416 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14421 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14422 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14423 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14424 * internal-fn.def (LOOP_DIST_ALIAS): New.
14425 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14426 (fold_loop_internal_call): ... this.
14427 (vect_loop_dist_alias_call): New function.
14428 (set_uid_loop_bbs): Call fold_loop_internal_call.
14429 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14432 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14435 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14436 Require dead FLAGS_REG at the beginning of a peephole.
14438 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14441 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14442 arguments in the call to __builtin_ia32_sbb_u32.
14443 (_subborrow_u64): Swap _X and _Y arguments in the call to
14444 __builtin_ia32_sbb_u64.
14446 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14449 * tree-vrp.c (compare_assert_loc): Turn into a function template
14450 with stable template parameter. Only test if a->e is NULL,
14451 !a->e == !b->e has been verified already. Use e == NULL or
14452 e != NULL instead of e or ! e tests. If stable is true, don't use
14453 iterative_hash_expr, on the other side allow a or b or both NULL
14454 and sort the NULLs last.
14455 (process_assert_insertions): Sort using compare_assert_loc<false>
14456 instead of compare_assert_loc, later sort using
14457 compare_assert_loc<true> before calling process_assert_insertions_for
14458 in a loop. Use break instead of continue once seen NULL pointer.
14460 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14462 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14463 Cortex-R7 and Cortex-R8 processors.
14465 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14467 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14468 uninitialized while src is not.
14470 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14472 * common/config/arm/arm-common.c: Adjust include path for
14474 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14475 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14476 (arm-cpu-data.h): Likewise.
14477 (arm-cpu-cdata.h): Likewise.
14478 * config/arm/arm-cpu.h: Delete.
14479 * config/arm/arm-cpu-cdata.h: Delete.
14480 * config/arm/arm-cpu-data.h: Delete.
14482 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14484 * config/arm/arm-cpus.in (cortex-a55): New.
14485 (cortex-a75): Likewise.
14486 (cortex-a75.cortex-a55): Likewise.
14487 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14489 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14490 * config/arm/arm-cpu-cdata.h: Regenerate.
14491 * config/arm/arm-cpu-data.h: Regenerate.
14492 * config/arm/arm-cpu.h: Regenerate.
14493 * config/arm/arm-tables.opt: Regenerate.
14494 * config/arm/arm-tune.md: Regenerate.
14496 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14498 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14500 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14502 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14505 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14507 PR tree-optimization/81292
14508 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14509 full_string_p, also call adjust_related_strinfos if the adjustment
14510 is simple, otherwise invalidate related strinfos.
14512 2017-07-04 Martin Liska <mliska@suse.cz>
14515 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14516 newly created variable as DECL_IGNORED_P.
14518 2017-07-04 Martin Liska <mliska@suse.cz>
14521 * ipa-inline.c (inline_small_functions):
14522 Use xstrdup_for_dump.
14524 2017-07-04 Tom de Vries <tom@codesourcery.com>
14526 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14528 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14531 * config/darwin.c (darwin_function_switched_text_sections):
14534 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14536 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14538 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14540 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14542 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14544 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14545 min_profitable_iters, and th as inclusive lower bounds.
14546 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14547 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14548 for min_profitable_iters and min_profitable_estimate.
14549 (vect_transform_loop): Treat th as an inclusive lower bound.
14550 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14552 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14555 * config/darwin.c (darwin_function_switched_text_sections):
14556 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14557 in two pieces, and suppress the use of buf.
14559 2017-07-03 Nathan Sidwell <nathan@acm.org>
14561 * hash-table.h (hash_table_mod1): Fix indentation.
14563 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14565 PR middle-end/81290
14566 * predict.c (force_edge_cold): Be more careful about propagation
14568 * profile-count.h (profile_probability::guessed,
14569 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14571 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14573 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14575 * doc/invoke.texi (rcpc architecture extension): Document it.
14577 2017-07-03 Richard Biener <rguenther@suse.de>
14579 PR tree-optimization/60510
14580 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14581 the scalar reduction PHI and use it.
14582 (vectorizable_reduction): Properly guard the single_defuse_cycle
14583 path for non-SLP reduction chains where we cannot use it.
14584 Rework reduc_def/index and vector type deduction. Rework
14585 vector operand gathering during reduction op code-gen.
14586 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14587 chains dissolve the chain and leave it to non-SLP reduction
14590 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14592 * tree-data-ref.h (dr_alignment): Declare.
14593 * tree-data-ref.c (dr_alignment): New function.
14594 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14595 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14597 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14599 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14601 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14602 and base_misalignment fields.
14603 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14604 * tree-data-ref.c: Include builtins.h.
14605 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14606 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14607 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14608 * tree-vect-data-refs.c: Include tree-cfg.h.
14609 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14610 fields instead of calculating an alignment here.
14611 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14612 innermost_loop_behavior fields.
14614 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14616 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14618 (DR_STEP_ALIGNMENT): New macro.
14619 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14620 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14621 (create_data_ref): Print it.
14622 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14623 to tell whether the step preserves vector (mis)alignment.
14624 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14625 Move the check for an integer step and generalise to all INTEGER_CST.
14626 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14627 Print the outer step alignment.
14629 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14631 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14632 with offset_alignment.
14633 (DR_ALIGNED_TO): Delete.
14634 (DR_OFFSET_ALIGNMENT): New macro.
14635 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14636 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14637 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14638 (create_data_ref): Likewise.
14639 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14640 (vect_analyze_data_refs): Likewise.
14641 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14642 creating dummy innermost behavior.
14644 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14646 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14647 with a "innermost_loop_behavior *" and refeence tree.
14648 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14649 (create_data_ref): Update call accordingly.
14650 * tree-predcom.c (find_looparound_phi): Likewise.
14652 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14654 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14655 fields with dr_wrt_vec_loop.
14656 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14657 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14658 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14659 (vect_dr_behavior): New function.
14660 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14661 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14662 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14663 track whether the step preserves the misalignment.
14664 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14665 Use vect_dr_behavior.
14666 (vect_setup_realignment): Update call accordingly.
14667 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14668 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14669 call to vect_create_addr_base_for_vector_ref.
14670 (vect_create_cond_for_align_checks): Likewise.
14671 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14672 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14673 (vect_recog_mask_conversion_pattern): Likewise.
14674 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14675 (new_stmt_vec_info): Remove redundant zeroing.
14677 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14679 * common/config/arm/arm-common.c (arm_be8_option): New function.
14680 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14681 (ISA_ARMv6): Add isa_bit_be8.
14682 * config/arm/arm.h (arm_be8_option): Add prototype.
14683 (BE8_SPEC_FUNCTION): New define.
14684 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14685 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14686 (mlittle-endian): Similarly.
14687 (mbe8, mbe32): New options.
14688 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14689 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14691 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14693 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14697 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14698 (cleanup_tree_cfg_bb): Use it.
14699 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14701 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14703 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14706 * loop-doloop.c (add_test): Update profile.
14708 2017-07-03 Martin Liska <mliska@suse.cz>
14711 * sanopt.c (rewrite_usage_of_param): New function.
14712 (sanitize_rewrite_addressable_params): Likewise.
14713 (pass_sanopt::execute): Call rewrite_usage_of_param.
14715 2017-07-03 Richard Biener <rguenther@suse.de>
14717 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14718 back to using VIEW_CONVERT_EXPR.
14720 2017-07-03 Martin Liska <mliska@suse.cz>
14723 * doc/extend.texi: Document when a resolver function is
14724 generated for target_clones.
14726 2017-07-03 Martin Liska <mliska@suse.cz>
14728 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14729 and shadow memory of auto variables which are subject of
14730 use-after-scope sanitization.
14731 (asan_expand_mark_ifn): Add do set only when is_poison.
14733 2016-07-03 Richard Biener <rguenther@suse.de>
14735 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14737 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14738 (vectorizable_reduction): Perform reduction PHI creation when
14739 visiting a reduction PHI and adjust and simplify code generation
14740 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14741 (vect_transform_loop): Visit reduction PHIs.
14742 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14743 defs into the SLP tree.
14744 (vect_build_slp_tree): Reduction defs terminate the recursion.
14745 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14747 (vect_get_vec_defs_for_stmt_copy): Export.
14748 (vect_get_vec_defs): Likewise.
14749 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14751 (vect_get_vec_defs_for_stmt_copy): Declare.
14752 (vect_get_vec_defs): Likewise.
14754 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14756 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14757 parameter with a "loop" parameter and use it instead of the
14758 loop containing DR_STMT. Don't check simple_iv when doing
14759 BB analysis. Describe the two analysis modes in the comment.
14761 2017-07-03 Tom de Vries <tom@codesourcery.com>
14763 PR tree-optimization/69468
14764 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14765 (find_same_succ_bb): Handle ignore_edge_flags.
14767 2017-07-03 Tom de Vries <tom@codesourcery.com>
14769 PR tree-optimization/81192
14770 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14772 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14774 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14776 2017-07-03 Tom de Vries <tom@codesourcery.com>
14778 PR tree-optimization/81192
14779 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14780 BB_SAME_SUCC (bb) == NULL.
14782 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14784 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14787 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14789 * dumpfile.c: Include profile-count.h
14790 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14792 (insert_cond_bb): Update profile.
14793 * tree-cfg.h (insert_cond_bb): Update prototype.
14794 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14795 * tree-dump.c: Do not include tree-cfg.
14797 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14799 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14803 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14806 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14808 * tree-complex.c (expand_complex_div_wide): update profile.
14810 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14811 Alan Hayward <alan.hayward@arm.com>
14812 David Sherwood <david.sherwood@arm.com>
14814 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14815 (CORETYPES_H): New define.
14816 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14817 (insn-modes-inline.h, s-modes-inline-h): New rules.
14818 (generated_files): Add insn-modes-inline.h.
14819 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14820 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14821 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14822 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14823 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14824 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14825 (build/gencodes.o, build/genconditions.o): Likewise.
14826 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14827 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14828 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14829 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14830 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14831 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14832 * coretypes.h: Include everything up to real.h for generators.
14833 Include insn-modes.h first. Include wide-int-print.h after
14834 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14835 * machmode.h: Don't include insn-modes.h here.
14836 * function-tests.c: Remove includes of signop.h, machmode.h,
14837 double-int.h and wide-int.h.
14839 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14841 * optc-save-gen.awk: Likewise.
14842 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14843 * godump.c: Remove include of wide-int-print.h.
14844 * pretty-print.h: Likewise.
14845 * wide-int-print.cc: Likewise.
14846 * wide-int.cc: Likewise.
14847 * hash-map-tests.c: Remove include of signop.h.
14848 * hash-set-tests.c: Likewise.
14849 * rtl-tests.c: Likewise.
14850 * mkconfig.sh: Remove include of machmode.h.
14851 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14853 (emit_insn_modes_inline_h): ...this new function. Emit the code
14854 into an insn-modes-inline.h header file, adding appropriate
14855 include guards and end comments.
14856 (emit_insn_modes_c_header): Remove include of machmode.h.
14857 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14859 (main): Handle -i flag and call emit_insn_modes_inline_h when
14862 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14864 * tree-ssa-strlen.c (strinfo): Rename the length field to
14865 nonzero_chars. Add a full_string_p field.
14866 (compare_nonzero_chars, zero_length_string_p): New functions.
14867 (get_addr_stridx): Add an offset_out parameter.
14868 Use compare_nonzero_chars.
14869 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14870 (new_strinfo): Update after above changes to strinfo.
14871 (set_endptr_and_length): Set full_string_p.
14872 (get_string_length): Update after above changes to strinfo.
14873 (unshare_strinfo): Update call to new_strinfo.
14874 (maybe_invalidate): Likewise.
14875 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14876 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14877 as a uhwi instead of an shwi. Update after above changes to
14878 strinfo and new_strinfo.
14879 (zero_length_string): Assert that chainsi contains full strings.
14880 Use zero_length_string_p. Update call to new_strinfo.
14881 (adjust_related_strinfos): Update after above changes to strinfo.
14882 Copy full_string_p from origsi.
14883 (adjust_last_stmt): Use zero_length_string_p.
14884 (handle_builtin_strlen): Update after above changes to strinfo and
14885 new_strinfo. Install the lhs as the string length if the previous
14886 entry didn't describe a full string.
14887 (handle_builtin_strchr): Update after above changes to strinfo
14889 (handle_builtin_strcpy): Likewise.
14890 (handle_builtin_strcat): Likewise.
14891 (handle_builtin_malloc): Likewise.
14892 (handle_pointer_plus): Likewise.
14893 (handle_builtin_memcpy): Likewise. Track nonzero characters
14894 that aren't necessarily followed by a nul terminator.
14895 (handle_char_store): Likewise.
14897 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14899 PR tree-optimization/80769
14900 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14901 for malloc and calloc. Document the new invariant that all related
14902 strinfos have delayed lengths or none do.
14903 (verify_related_strinfos): Move earlier in file.
14904 (set_endptr_and_length): New function, split out from...
14905 (get_string_length): ...here. Also set the lengths of related
14907 (zero_length_string): Assert that chainsi has known (rather than
14909 (adjust_related_strinfos): Likewise.
14911 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14913 PR tree-optimization/81136
14914 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14915 assert that two references with the same misalignment have the same
14916 compile-time misalignment if those compile-time misalignments
14919 2017-07-01 Andi Kleen <ak@linux.intel.com>
14921 * print-tree.c (print_node): Print all attributes.
14923 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14925 * cfg.c (scale_bbs_frequencies): New function.
14926 * cfg.h (scale_bbs_frequencies): Declare it.
14927 * cfgloopanal.c (single_likely_exit): Cleanup.
14928 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14930 (scale_loop_profile): Likewise.
14931 (loop_version): Likewise.
14932 (create_empty_loop_on_edge): Update.
14933 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14934 scale_loop_frequencies, scale_loop_profile, loopify,
14935 loop_version): Update prototypes.
14936 * modulo-sched.c (sms_schedule): Update.
14937 * predict.c (unlikely_executed_edge_p): Also check probability.
14938 (probably_never_executed_edge_p): Fix typo.
14939 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14940 * tree-parloops.c (gen_parallel_loop): Update.
14941 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14942 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14943 * tree-ssa-loop-split.c (split_loop): Update.
14944 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14945 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14946 (vect_loop_versioning): Update.
14947 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14949 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14951 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14953 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14955 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14956 to keep profile consistent.
14958 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14960 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14961 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14962 * profile-count.h (max_safe_multiplier): Make unsigned.
14963 (profile_count::guessed_zero): New.
14965 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14967 * bb-reorder.c (fix_up_crossing_landing_pad,
14968 fix_crossing_conditional_branches): Use make_single_succ_edge
14969 to keep profile consistent.
14971 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14973 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14976 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14979 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14980 the right scopes, make sure cond_jump isn't preserved between multiple
14981 iterations. Search for fallthru edge whenever there are 3+ edges and
14982 use find_fallthru_edge for it.
14984 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14986 Patch by Alexander Monakov <amonakov@ispras.ru>
14987 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14988 probabilities consistently.
14990 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14992 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14993 emit_cmp_and_jump_insns.
14995 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14998 * tree-inline.c (expand_call_inline): Combine profile statuses.
15000 2017-06-30 Andrew Pinski <apinski@cavium.com>
15002 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15003 fold_stmt returned true.
15005 2017-06-30 Nathan Sidwell <nathan@acm.org>
15007 * ggc.h (empty_string): Delete.
15008 * cfgexpand.c (expand_asm_stmt): Use plain "".
15009 * optabs.c (expand_asm_memory_barrier): Likewise.
15010 * stringpool.c (empty_string): Delete.
15011 (digit_vector, digit_string): Delete.
15012 (ggc_alloc_string): Use plain "", don't optimize single digit
15013 strings. Use ggc_alloc_atomic.
15015 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15017 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15018 comparison set and one other set, use the cost of the non-comparison
15021 2017-06-30 Nathan Sidwell <nathan@acm.org>
15023 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15026 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15028 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15029 loops. Remove now unneeded calls to gimple_switch_set_label() that
15030 just set removed labels to NULL_TREE.
15032 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15034 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15035 (set_range_info): ...here. Only call set_range_info_raw if domain
15037 (set_nonzero_bits): Call set_range_info_raw.
15038 * tree-ssanames.h (set_range_info_raw): New.
15040 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15043 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15044 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15045 of nonimmediate_operand and <store_mask_constraint> instead of m
15046 for the input operand. For V8FI iterator, always split if input
15047 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15048 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15049 <store_mask_predicate> instead of register_operand and
15050 <store_mask_constraint> instead of v for the input operand. Make
15051 sure both operands aren't MEMs for if not <mask_applied>.
15053 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15055 * lto-wrapper.c (copy_file) Close both file descriptors before
15058 2017-06-30 Martin Liska <mliska@suse.cz>
15061 * multiple_target.c (create_dispatcher_calls): Make ifunc
15062 also for function that don't have calls or are not referenced.
15064 2017-06-30 Richard Biener <rguenther@suse.de>
15066 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15067 analyze the first scalar stmt. Move vector type computation
15068 for the BB case here from ...
15069 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15070 live operation processing in the SLP case properly.
15072 2017-06-30 Richard Biener <rguenther@suse.de>
15074 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15076 2017-06-30 Martin Liska <mliska@suse.cz>
15079 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15080 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15082 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15084 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15087 2017-06-30 Richard Biener <rguenther@suse.de>
15089 PR tree-optimization/81249
15090 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15091 condition reduction result to original scalar type.
15093 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15095 * profile-count.h (enum profile_quality): Fix typos and whitespace
15098 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15100 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15101 type for branch probabilities.
15103 2017-06-29 Julian Brown <julian@codesourcery.com>
15104 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15106 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15107 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15108 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15109 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15111 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15113 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15114 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15115 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15116 CC usage from generic code to here.
15117 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15118 CC usage into the target macros.
15120 2017-06-29 Maya Rashish <coypu@sdf.org>
15122 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15125 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15127 * arm/arm-builtins.c: Include profile-count.h
15128 * except.c (sjlj_emit_function_enter): Use
15129 profile_probability::unlikely.
15131 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15133 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15134 and tocrel_offset be pointer args rather than implicitly using
15136 (legitimate_constant_pool_address_p, rs6000_emit_move,
15137 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15138 tocrel_offset and use in toc_relative_expr_p call.
15139 (print_operand, print_operand_address): Use static tocrel_base_oac
15140 and tocrel_offset_oac.
15141 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15144 2017-06-29 Maya Rashish <coypu@sdf.org>
15146 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15148 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15150 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15151 objects, take into account only the alignment of 'op0' and 'mode1' if
15154 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15156 * ccmp.c (ccmp_tree_comparison_p): New function.
15157 (ccmp_candidate_p): Update to use above function.
15158 (get_compare_parts): New function.
15159 (expand_ccmp_next): Update to use new functions.
15160 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15162 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15163 take mode as argument.
15164 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15165 * expr.c (expand_expr_real_1): Pass mode as argument.
15167 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15169 * combine.c (combine_instructions): Print insns to dump_file, together
15172 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15174 * asan.c (asan_emit_stack_protection): Update.
15175 (create_cond_insert_point): Update.
15176 * auto-profile.c (afdo_propagate_circuit): Update.
15177 * basic-block.h (struct edge_def): Turn probability to
15178 profile_probability.
15179 (EDGE_FREQUENCY): Update.
15180 * bb-reorder.c (find_traces_1_round): Update.
15181 (better_edge_p): Update.
15182 (sanitize_hot_paths): Update.
15183 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15184 (make_single_succ_edge): Update.
15185 (check_bb_profile): Update.
15186 (dump_edge_info): Update.
15187 (update_bb_profile_for_threading): Update.
15188 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15189 probabilitycount to 0.
15190 * cfgbuild.c (compute_outgoing_frequencies): Update.
15191 * cfgcleanup.c (try_forward_edges): Update.
15192 (outgoing_edges_match): Update.
15193 (try_crossjump_to_edge): Update.
15194 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15195 (expand_gimple_tailcall): Update.
15196 (construct_init_block): Use make_single_succ_edge.
15197 (construct_exit_block): Use make_single_succ_edge.
15198 * cfghooks.c (verify_flow_info): Update.
15199 (redirect_edge_succ_nodup): Update.
15200 (split_edge): Update.
15201 (account_profile_record): Update.
15202 * cfgloopanal.c (single_likely_exit): Update.
15203 * cfgloopmanip.c (scale_loop_profile): Update.
15204 (set_zero_probability): Remove.
15205 (duplicate_loop_to_header_edge): Update.
15206 * cfgloopmanip.h (loop_version): Update prototype.
15207 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15208 (force_nonfallthru_and_redirect): Update.
15209 (update_br_prob_note): Update.
15210 (rtl_verify_edges): Update.
15211 (purge_dead_edges): Update.
15212 (rtl_lv_add_condition_to_bb): Update.
15213 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15214 * cgraphunit.c (init_lowered_empty_function): Update.
15215 (cgraph_node::expand_thunk): Update.
15216 * cilk-common.c: Include profile-count.h
15217 * dojump.c (inv): Remove.
15218 (jumpifnot): Update.
15219 (jumpifnot_1): Update.
15220 (do_jump_1): Update.
15222 (do_jump_by_parts_greater_rtx): Update.
15223 (do_compare_rtx_and_jump): Update.
15224 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15225 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15226 * dwarf2cfi.c: Include profile-count.h
15227 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15228 (sjlj_emit_dispatch_table): Likewise.
15229 * explow.c: Include profile-count.h
15230 * expmed.c (emit_store_flag_force): Update.
15231 (do_cmp_and_jump): Update.
15232 * expr.c (compare_by_pieces_d::generate): Update.
15233 (compare_by_pieces_d::finish_mode): Update.
15234 (emit_block_move_via_loop): Update.
15235 (store_expr_with_bounds): Update.
15236 (store_constructor): Update.
15237 (expand_expr_real_2): Update.
15238 (expand_expr_real_1): Update.
15239 * expr.h (try_casesi, try_tablejump): Update prototypes.
15240 * gimple-pretty-print.c (dump_probability): Update.
15241 (dump_profile): New.
15242 (dump_gimple_label): Update.
15243 (dump_gimple_bb_header): Update.
15244 * graph.c (draw_cfg_node_succ_edges): Update.
15245 * hsa-gen.c (convert_switch_statements): Update.
15246 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15247 (find_if_case_1): Update.
15248 (find_if_case_2): Update.
15249 * internal-fn.c (expand_arith_overflow_result_store): Update.
15250 (expand_addsub_overflow): Update.
15251 (expand_neg_overflow): Update.
15252 (expand_mul_overflow): Update.
15253 (expand_vector_ubsan_overflow): Update.
15254 * ipa-cp.c (good_cloning_opportunity_p): Update.
15255 * ipa-split.c (split_function): Use make_single_succ_edge.
15256 * ipa-utils.c (ipa_merge_profiles): Update.
15257 * loop-doloop.c (add_test): Update.
15258 (doloop_modify): Update.
15259 * loop-unroll.c (compare_and_jump_seq): Update.
15260 (unroll_loop_runtime_iterations): Update.
15261 * lra-constraints.c (lra_inheritance): Update.
15262 * lto-streamer-in.c (input_cfg): Update.
15263 * lto-streamer-out.c (output_cfg): Update.
15264 * mcf.c (adjust_cfg_counts): Update.
15265 * modulo-sched.c (sms_schedule): Update.
15266 * omp-expand.c (expand_omp_for_init_counts): Update.
15267 (extract_omp_for_update_vars): Update.
15268 (expand_omp_ordered_sink): Update.
15269 (expand_omp_for_ordered_loops): Update.
15270 (expand_omp_for_generic): Update.
15271 (expand_omp_for_static_nochunk): Update.
15272 (expand_omp_for_static_chunk): Update.
15273 (expand_cilk_for): Update.
15274 (expand_omp_simd): Update.
15275 (expand_omp_taskloop_for_outer): Update.
15276 (expand_omp_taskloop_for_inner): Update.
15277 * omp-simd-clone.c (simd_clone_adjust): Update.
15278 * optabs.c (expand_doubleword_shift): Update.
15279 (expand_abs): Update.
15280 (emit_cmp_and_jump_insn_1): Update.
15281 (expand_compare_and_swap_loop): Update.
15282 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15283 * predict.c (predictable_edge_p): Update.
15284 (edge_probability_reliable_p): Update.
15285 (set_even_probabilities): Update.
15286 (combine_predictions_for_insn): Update.
15287 (combine_predictions_for_bb): Update.
15288 (propagate_freq): Update.
15289 (estimate_bb_frequencies): Update.
15290 (force_edge_cold): Update.
15291 * profile-count.c (profile_count::dump): Add missing space into dump.
15292 (profile_count::debug): Add newline.
15293 (profile_count::differs_from_p): Explicitly convert to unsigned.
15294 (profile_count::stream_in): Update.
15295 (profile_probability::dump): New member function.
15296 (profile_probability::debug): New member function.
15297 (profile_probability::differs_from_p): New member function.
15298 (profile_probability::differs_lot_from_p): New member function.
15299 (profile_probability::stream_in): New member function.
15300 (profile_probability::stream_out): New member function.
15301 * profile-count.h (profile_count_quality): Rename to ...
15302 (profile_quality): ... this one.
15303 (profile_probability): New.
15304 (profile_count): Update.
15305 * profile.c (compute_branch_probabilities): Update.
15306 * recog.c (peep2_attempt): Update.
15307 * sched-ebb.c (schedule_ebbs): Update.
15308 * sched-rgn.c (find_single_block_region): Update.
15309 (compute_dom_prob_ps): Update.
15310 (schedule_region): Update.
15311 * sel-sched-ir.c (compute_succs_info): Update.
15312 * stmt.c (struct case_node): Update.
15313 (do_jump_if_equal): Update.
15314 (get_outgoing_edge_probs): Update.
15315 (conditional_probability): Update.
15316 (emit_case_dispatch_table): Update.
15317 (expand_case): Update.
15318 (expand_sjlj_dispatch_table): Update.
15319 (emit_case_nodes): Update.
15320 * targhooks.c: Update.
15321 * tracer.c (better_p): Update.
15322 (find_best_successor): Update.
15323 * trans-mem.c (expand_transaction): Update.
15324 * tree-call-cdce.c: Update.
15325 * tree-cfg.c (gimple_split_edge): Upate.
15326 (move_sese_region_to_fn): Upate.
15327 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15328 * tree-eh.c (lower_resx): Upate.
15329 (cleanup_empty_eh_move_lp): Upate.
15330 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15331 * tree-inline.c (copy_edges_for_bb): Update.
15332 (copy_cfg_body): Update.
15333 * tree-parloops.c (gen_parallel_loop): Update.
15334 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15335 (gimple_gen_time_profiler): Update.
15336 * tree-ssa-dce.c (remove_dead_stmt): Update.
15337 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15338 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15339 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15340 (unloop_loops): Update.
15341 (try_peel_loop): Update.
15342 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15343 * tree-ssa-loop-split.c (connect_loops): Update.
15344 (split_loop): Update.
15345 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15346 (hoist_guard): Update.
15347 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15348 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15349 (value_replacement): Update.
15350 * tree-ssa-reassoc.c (branch_fixup): Update.
15351 * tree-ssa-tail-merge.c (replace_block_by): Update.
15352 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15353 (create_edge_and_update_destination_phis): Update.
15354 (compute_path_counts): Update.
15355 (recompute_probabilities): Update.
15356 (update_joiner_offpath_counts): Update.
15357 (freqs_to_counts_path): Update.
15358 (duplicate_thread_path): Update.
15359 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15360 (struct switch_conv_info): Update.
15361 (gen_inbound_check): Update.
15362 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15363 (vect_do_peeling): Update.
15364 (vect_loop_versioning): Update.
15365 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15366 (optimize_mask_stores): Update.
15367 * ubsan.c (ubsan_expand_null_ifn): Update.
15368 * value-prof.c (gimple_divmod_fixed_value): Update.
15369 (gimple_divmod_fixed_value_transform): Update.
15370 (gimple_mod_pow2): Update.
15371 (gimple_mod_pow2_value_transform): Update.
15372 (gimple_mod_subtract): Update.
15373 (gimple_mod_subtract_transform): Update.
15374 (gimple_ic): Update.
15375 (gimple_stringop_fixed_value): Update.
15376 (gimple_stringops_transform): Update.
15377 * value-prof.h: Update.
15379 2017-06-29 Carl Love <cel@us.ibm.com>
15381 * config/rs6000/rs6000-c.c: Add support for built-in functions
15382 vector signed int vec_signed (vector float);
15383 vector signed long long vec_signed (vector double);
15384 vector signed int vec_signed2 (vector double, vector double);
15385 vector signed int vec_signede (vector double);
15386 vector signed int vec_signedo (vector double);
15387 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15388 instruction generator.
15389 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15390 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15391 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15393 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15394 vunsignede_v2df): Add define_expands.
15395 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15396 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15397 VEC_UNSIGNEDO): Add definitions.
15398 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15399 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15400 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15401 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15402 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15403 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15404 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15405 vec_unsignede, vec_unsignedo): Add builtin defines.
15406 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15408 * doc/extend.texi: Update the built-in documentation file for the
15409 new built-in functions.
15411 2017-06-29 Richard Biener <rguenther@suse.de>
15413 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15414 reduction chains to LOOP_VINFO_REDUCTIONS.
15415 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15416 SLP reductions after processing reduction chains.
15418 2017-06-29 Nathan Sidwell <nathan@acm.org>
15420 * builtins.c (fold_builtin_FUNCTION): Use
15421 lang_hooks.decl_printable_name.
15423 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15425 PR middle-end/81194
15426 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15427 with only one label.
15428 * stmt.c (expand_case): Assert NCASES is greater than one.
15430 2017-06-29 Richard Biener <rguenther@suse.de>
15432 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15434 (group_case_labels): Likewise.
15435 (find_taken_edge): Push sanity checking on val to workers...
15436 (find_taken_edge_cond_expr): ... here
15437 (find_taken_edge_switch_expr): ... and here, handle cases
15438 with just a default label.
15439 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15440 (group_case_labels): Likewise.
15441 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15442 group_case_labels does anything cleanup the CFG again.
15444 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15446 PR tree-optimization/81196
15447 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15448 exit condition comparing two IVs.
15450 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15452 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15453 profile to the dummy entry at the end of the list of architectures.
15454 * config/arm/arm-cpu-cdata.h: Regenerated.
15456 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15457 Michael Collison <michael.collison@arm.com>
15460 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15462 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15463 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15464 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15465 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15466 of shift when the shift amount is masked with constant equal to
15467 the size of the mode.
15468 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15471 2017-06-29 Martin Liska <mliska@suse.cz>
15473 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15475 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15478 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15481 (edge_growth_cache_entry::edge_growth_cache_entry): New
15483 (reset_edge_growth_cache): Update to use constructor.
15485 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15487 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15488 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15489 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15491 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15493 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15494 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15496 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15498 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15499 (*-linux-uclibc*): Add t-uclibc tmake_file.
15500 * config/t-musl: New.
15501 * config/t-uclibc: New.
15503 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15505 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15507 (gen_comm_data): Emit architectural setting of arch_prof.
15508 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15510 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15511 (armv8-m.base, armv8-m.main): Likewise.
15512 * arm-protos.h (arm_build_target): Add profile field.
15513 (arch_option): Likewise.
15514 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15516 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15517 arm_active_target.profile.
15519 2017-06-28 Richard Biener <rguenther@suse.de>
15521 PR middle-end/81227
15522 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15523 TYPE_OVERFLOW_WRAPS.
15524 * match.pd (negate_expr_p): Likewise.
15525 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15526 fold_build2, not fold_binary.
15528 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15530 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15531 Convert memory address to Pmode.
15532 (aarch64_print_operand): Assert MEM operands are always Pmode.
15534 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15537 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15538 Remove redundant if.
15539 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15540 * config/arm/aarch-common-protos.h
15541 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15542 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15544 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15547 * multiple_target.c (create_dispatcher_calls): Set the default
15548 clone to be static, not public.
15550 2017-06-28 Richard Biener <rguenther@suse.de>
15552 * tree-vect-loop.c (vectorizable_reduction): Move special
15553 cond reduction IV var creation ...
15554 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15555 parameter. Use STMT_VINFO_VECTYPE.
15556 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15559 2017-06-28 Martin Liska <mliska@suse.cz>
15562 * ipa-visibility.c (non_local_p): Handle visibility.
15564 2017-06-28 Martin Liska <mliska@suse.cz>
15567 * common.opt: Add IntegerRange to various options.
15568 * opt-functions.awk (integer_range_info): New function.
15569 * optc-gen.awk: Add integer_range_info to cl_options struct.
15570 * opts-common.c (decode_cmdline_option): Handle
15571 CL_ERR_INT_RANGE_ARG.
15572 (cmdline_handle_error): Likewise.
15573 * opts.c (print_filtered_help): Show valid interval in
15574 when --help is provided.
15575 * opts.h (struct cl_option): Add range_min and range_max fields.
15576 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15578 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15580 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15581 (x * C EQ/NE y * C): New transformation.
15583 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15585 * genmultilib (combination_space): Accept '+' in option names.
15587 2017-06-28 Martin Liska <mliska@suse.cz>
15590 * asan.c (instrument_derefs): Bail out inner references
15591 that are hard register variables.
15593 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15596 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15597 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15599 2017-06-28 Richard Biener <rguenther@suse.de>
15601 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15602 (vect_get_slp_defs): Adjust.
15603 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15605 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15607 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15608 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15609 (vectorizable_reduction): Adjust.
15610 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15612 (vect_get_slp_defs): Likewise.
15613 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15614 (vectorizable_bswap): Adjust.
15615 (vectorizable_call): Likewise.
15616 (vectorizable_conversion): Likewise.
15617 (vectorizable_assignment): Likewise.
15618 (vectorizable_shift): Likewise.
15619 (vectorizable_operation): Likewise.
15620 (vectorizable_store): Likewise.
15621 (vectorizable_condition): Likewise.
15622 (vectorizable_comparison): Likewise.
15624 2017-06-28 Michael Collison <michael.collison@arm.com>
15627 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15629 (arm_gen_movmemqi): Removed unused variable 'i'.
15630 Convert 'for' loop into 'while' loop.
15631 (arm_expand_prologue): Remove last unnecessary set of insn.
15632 (thumb_pop): Remove unused variable 'pushed_words'.
15633 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15635 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15637 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15638 * config/s390/s390-protos.h: Add prototype of
15639 s390_rel_address_ok_p.
15640 * config/s390/s390.c (s390_got_symbol): New function.
15641 (s390_rel_address_ok_p): New function.
15642 (legitimize_pic_address): Use s390_rel_address_ok_p.
15643 (s390_load_got): Use s390_got_symbol.
15644 (s390_option_override): Issue error if
15645 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15646 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15648 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15650 2017-06-27 Andrew Pinski <apinski@cavium.com>
15652 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15653 (X * copysign (1.0, X)): New pattern.
15654 (X * copysign (1.0, -X)): New pattern.
15655 (copysign (-1.0, CST)): New pattern.
15657 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15659 * genmultilib (combination_space): Remove variable.
15660 Validate reuse rules against regular expression for any sequence
15661 of multilib options in any order.
15663 2017-06-27 Michael Collison <michael.collison@arm.com>
15665 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15666 call aarch64_split_simd_combine.
15667 * (aarch64_combine_internal<mode>): Delete pattern.
15668 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15669 Allow register and subreg operands.
15671 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15673 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15674 specific need, just fallback on defaults.
15675 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15677 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15678 Olivier Hainque <hainque@adacore.com>
15680 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15682 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15683 targets. Pick a default if no particular attempt applied.
15684 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15687 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15689 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15690 (x86_64-wrs-vxworks7): Likewise.
15692 2017-06-27 Marek Polacek <polacek@redhat.com>
15695 * ubsan.c (instrument_null): Check get_base_address's result for null.
15697 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15699 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15701 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15703 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15704 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15705 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15706 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15707 New function types.
15708 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15709 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15710 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15711 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15712 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15713 BUILT_IN_FEUPDATEENV): New builtins.
15714 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15715 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15716 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15717 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15719 (builtin_structptr_types): Adjust size.
15720 * tree.c (builtin_structptr_types): Add four entries.
15722 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15723 Olivier Hainque <hainque@adacore.com>
15725 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15726 (TLS_SYM): New local macro, forcing reference to __tls__ on
15727 link command lines for VxWorks 7 RTPs, triggering initialization
15729 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15730 OS features TLS support, true for RTPs on VxWorks 7.
15731 * config/vxworks.c (vxworks_override_options): Setup emutls
15734 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15736 * predict.c (test_prediction_value_range): Use -1U instead of -1
15737 to avoid narrowing conversion warning.
15738 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15739 to avoid narrowing conversion warning.
15740 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15742 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15744 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15746 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15747 64bit configurations.
15748 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15749 (SIZE_TYPE): Likewise.
15750 * config/vxworks.c (vxworks_emutls_var_fields): Use
15751 long_unsigned_type_node instead of unsigned_type_node as the offset
15752 field type, which is "pointer" mode in emutls.c.
15754 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15757 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15759 PR middle-end/81207
15760 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15761 gimple_vuse copying separately from gimple_vdef copying.
15763 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15765 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15768 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15769 Olivier Hainque <hainque@adacore.com>
15771 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15772 for all vxworks7 targets.
15773 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15774 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15775 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15776 variations for VX6/VX7 and 32/64bits later on in ...
15777 (VXWORKS_LIB_SPEC): Leverage new macros.
15778 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15779 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15781 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15783 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15784 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15786 2017-06-26 Carl Love <cel@us.ibm.com>
15788 * config/rs6000/rs6000-c.c: Add support for built-in functions
15789 vector bool char vec_reve (vector bool char);
15790 vector signed char vec_reve (vector signed char);
15791 vector unsigned char vec_reve (vector unsigned char);
15792 vector bool int vec_reve (vector bool int);
15793 vector signed int vec_reve (vector signed int);
15794 vector unsigned int vec_reve (vector unsigned int);
15795 vector bool long long vec_reve (vector bool long long);
15796 vector signed long long vec_reve (vector signed long long);
15797 vector unsigned long long vec_reve (vector unsigned long long);
15798 vector bool short vec_reve (vector bool short);
15799 vector signed short vec_reve (vector signed short);
15800 vector double vec_reve (vector double);
15801 vector float vec_reve (vector float);
15802 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15803 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15804 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15805 (altivec_vreve): New pattern.
15806 * config/rs6000/altivec.h (vec_reve): New define.
15807 * doc/extend.texi (vec_rev): Update the built-in documentation file
15808 for the new built-in functions.
15810 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15812 PR tree-optimization/71815
15813 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15815 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15817 (slsr_process_phi): Likewise.
15818 (replace_uncond_cands_and_profitable_phis): Don't replace a
15819 multiply candidate with a stride of 1 (copy or cast).
15820 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15822 (lowest_cost_path): Likewise.
15823 (total_savings): Likewise.
15825 2017-06-26 Richard Biener <rguenther@suse.de>
15828 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15829 Use def_builtin_pure for all gather builtins.
15831 2017-06-26 Richard Biener <rguenther@suse.de>
15833 PR tree-optimization/81203
15834 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15835 non-dominating BBs.
15837 2017-06-26 Marek Polacek <polacek@redhat.com>
15840 * doc/invoke.texi: Document -Wmultistatement-macros.
15842 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15844 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15845 arm_neon_ok_no_float_abi effective target.
15847 2017-06-26 Richard Biener <rguenther@suse.de>
15849 PR tree-optimization/80928
15850 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15851 (copy_bbs): Set BB_DUPLICATED flag early.
15852 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15854 (execute_on_shrinking_pred): Likewise.
15855 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15856 BB_DUPLICATED blocks.
15857 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15858 iterate over all PHIs considering removal of *gsi.
15860 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15862 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15865 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15867 * config/rs6000/rs6000-string.c: (expand_block_clear,
15868 do_load_for_compare, select_block_compare_mode,
15869 compute_current_alignment, expand_block_compare,
15870 expand_strncmp_align_check, expand_strn_compare,
15871 expand_block_move, rs6000_output_load_multiple)
15872 Move functions related to string/block move/compare
15873 to a separate file.
15874 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15875 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15876 for this function which is now used in two files.
15877 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15878 * config.gcc: Add rs6000-string.o to extra_objs for
15879 targets powerpc*-*-* and rs6000*-*-*.
15881 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15884 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15885 32-bit, since indexed is not valid for DImode.
15886 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15887 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15888 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15889 (define_peephole2 for Altivec d-form store): Likewise.
15892 * multiple_target.c (create_dispatcher_calls): Only create the
15893 dispatcher call if the function is the default clone of a
15894 versioned function.
15896 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15898 PR middle-end/80902
15899 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15900 a call, force the call to not be a tail call.
15902 2017-06-23 Jeff Law <law@redhat.com>
15904 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15907 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15909 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15910 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15911 handling for early expansion of vector shifts (sl,sr,sra,rl).
15912 (builtin_function_type): Add vector shift right instructions
15913 to the unsigned argument list.
15915 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15917 rtl-optimizatoin/79286
15918 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15919 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15920 trap. PIC register plus a const unspec without offset can never trap.
15922 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15924 * tree.h (builtin_structptr_type): New type.
15925 (builtin_structptr_types): Declare new array.
15926 * tree.c (builtin_structptr_types): New array.
15927 (free_lang_data, build_common_tree_nodes): Use it.
15929 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15932 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15935 2017-06-22 Matt Turner <mattst88@gmail.com>
15937 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15938 Lake models to skylake case. Assume skylake for unknown
15939 models with clflushopt.
15941 2017-06-22 Jeff Law <law@redhat.com>
15943 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15944 frame sizes that do not satisfy aarch64_uimm12_shift.
15946 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15948 * profile-count.h (apply_probability,
15949 apply_scale, probability_in): Fix checks for zero.
15951 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15953 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15954 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15956 2016-06-22 Richard Biener <rguenther@suse.de>
15958 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15959 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15960 REDUC_MAX_EXPR support.
15961 (vectorizable_reduction): Likewise.
15962 (vect_create_epilog_for_reduction): Likewise.
15964 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15966 * match.pd (A / (1 << B) -> A >> B): New.
15967 * generic-match-head.c: Include optabs-tree.h.
15968 * gimple-match-head.c: Likewise.
15969 * optabs-tree.h (target_supports_op_p): New.
15970 * optabs-tree.c (target_supports_op_p): New.
15972 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15974 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15975 $gcc_cv_ld --help output.
15976 (gcc_cv_ld_demangle): Likewise.
15977 (gcc_cv_ld_eh_frame_hdr): Likewise.
15978 (gcc_cv_ld_pie): Likewise.
15979 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15980 (gcc_cv_ld_buildid): Likewise.
15981 (gcc_cv_ld_sysroot): Likewise.
15982 (ld_bndplt_support): Likewise.
15983 (ld_pushpopstate_support): Likewise.
15984 * configure: Regenerate.
15985 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15987 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15990 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15991 operands indexes to avoid gap between operands and match_dups.
15993 2017-06-21 Andrew Pinski <apinski@cavium.com>
15995 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15996 Increment Arith_shift and Arith_shift_reg by 1.
15997 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15999 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16000 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16001 (aarch64_strip_extend): Add new argument and test for it.
16002 (aarch64_cheap_mult_shift_p): New function.
16003 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16004 add a cost if it is true.
16005 Update calls to aarch64_strip_extend.
16006 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16008 2017-06-21 Andrew Pinski <apinski@cavium.com>
16010 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16012 (thunderxt88): Likewise.
16013 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16014 (thunderx_prefetch_tune): New variable.
16015 (thunderx2t99_prefetch_tune): Update for the correct values.
16016 (thunderxt88_tunings): New variable.
16017 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16018 generic_prefetch_tune.
16019 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16021 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16023 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16024 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16025 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16026 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16027 (aarch64_atomic_cas<mode>, GPI): Likewise.
16029 2017-06-21 Martin Liska <mliska@suse.cz>
16031 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16032 statements on cold and hot labels.
16033 * predict.c (tree_estimate_probability_bb): Remove the
16034 prediction from this place.
16036 2017-06-21 Martin Liska <mliska@suse.cz>
16038 PR tree-optimization/79489
16039 * gimplify.c (maybe_add_early_return_predict_stmt): New
16041 (gimplify_return_expr): Call the function.
16042 * predict.c (tree_estimate_probability_bb): Remove handling
16044 * predict.def: Update comment about early return predictor.
16045 * gimple-predict.h (is_gimple_predict): New function.
16046 * predict.def: Change default value of early return to 66.
16047 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16049 * passes.def: Put pass_strip_predict_hints to the beginning of
16052 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16054 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16055 FUNCTION_DECL declarations.
16056 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16058 (dwaf2out_decl): Likewise.
16059 * godump.c (go_early_global_decl): Skip call to the real debug hook
16060 for FUNCTION_DECL declarations.
16061 * passes.c (rest_of_decl_compilation): Skip call to the
16062 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16063 -fdump-go-spec is passed.
16065 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16067 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16068 Reorder for compactness.
16069 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16070 (def_builtin_pure, def_builtin_pure2): New functions.
16071 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16073 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16075 * match.pd (nop_convert): New predicate.
16076 ((A +- CST1) +- CST2): Allow some NOP conversions.
16078 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16081 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16082 with ctors/dtors if GOVD_SHARED is set.
16084 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16086 * config/aarch64/aarch64.md (movti_aarch64):
16087 Emit mov rather than orr.
16088 (movtf_aarch64): Likewise.
16089 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16090 Emit mov rather than orr.
16092 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16094 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16095 Swap alternatives, make integer dup more expensive.
16097 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16099 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16100 Return true for non-tls symbols.
16102 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16104 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16105 (cortex-a75): Likewise.
16106 (cortex-a75.cortex-a55): Likewise.
16107 * config/aarch64/aarch64-tune.md: Regenerate.
16108 * doc/invoke.texi (-mtune): Document new values for -mtune.
16110 2017-06-21 Tom de Vries <tom@codesourcery.com>
16112 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16113 stack_size feature.
16114 (Effective-Target Keywords, Other attributes): Suggest using
16115 dg-add-options stack_size feature to get stack limit in stack_size
16116 effective target documentation.
16118 2017-06-21 Julian Brown <julian@codesourcery.com>
16119 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16121 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16122 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16123 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16125 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16126 attribute type list for neon_multiply.
16127 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16128 attribute type list for neon_multiply.
16129 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16130 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16131 attribute type list for neon_multiply.
16132 * config/arm/types.md (crypto_pmull): Add.
16133 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16134 attribute type list.
16136 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16138 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16141 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16143 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16144 to make sure not to dereference a NULL cost_classes_ptr pointer.
16146 2017-06-20 Carl Love <cel@us.ibm.com>
16148 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16149 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16150 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16151 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16152 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16153 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16154 VMULOSW): New enum "unspec" values.
16155 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16156 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16157 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16158 altivec_vmulosw): New patterns.
16159 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16160 VMULOSW): Add definitions.
16162 2017-06-20 Julia Koval <julia.koval@intel.com>
16164 * config/i386/i386.c: Fix rounding expand for new pattern.
16165 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16167 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16169 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16170 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16172 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16174 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16177 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16179 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16180 architecture, then by implementer ID.
16181 * config/aarch64/aarch64-tune.md: Regenerate.
16183 2017-06-20 Richard Biener <rguenther@suse.de>
16185 PR middle-end/81097
16186 * fold-const.c (split_tree): Fold to type before negating.
16188 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16190 * diagnostic-show-locus.c
16191 (selftest::test_fixit_deletion_affecting_newline): New function.
16192 (selftest::diagnostic_show_locus_c_tests): Call it.
16194 2017-06-20 Andreas Schwab <schwab@suse.de>
16197 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16200 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16202 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16203 __ARM_FEATURE_COPROC according to support.
16205 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16207 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16208 Rewritten to avoid overflow for > 32-bit pointers.
16211 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16212 by removing enum keyword.
16213 (ubsan_type_descriptor): Likewise. Formatting fix.
16216 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16217 splitter): Require TARGET_SSE2 in the condition.
16219 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16222 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16223 for doing vector set of SFmode on ISA 3.0.
16224 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16225 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16227 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16228 SFmode value into a V4SF variable that was extracted from another
16229 V4SF variable without converting the element to double precision
16230 and back to single precision vector format.
16231 (vsx_insert_extract_v4sf_p9_2): Likewise.
16233 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16235 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16236 in UWHI to avoid undefined overflow.
16239 * ubsan.h (enum ubsan_encode_value_phase): New.
16240 (ubsan_encode_value): Change second argument to
16241 enum ubsan_encode_value_phase with default value of
16242 UBSAN_ENCODE_VALUE_GENERIC.
16243 * ubsan.c (ubsan_encode_value): Change second argument to
16244 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16245 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16246 create_tmp_var_raw instead of create_tmp_var and use a
16248 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16249 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16250 ubsan_encode_value callers.
16253 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16254 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16255 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16257 2017-06-19 Richard Biener <rguenther@suse.de>
16259 PR middle-end/81118
16260 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16261 estimates if we changed anything.
16263 2017-06-19 Richard Biener <rguenther@suse.de>
16265 PR tree-optimization/80887
16266 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16267 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16268 simplified lookups, then reset mprts_hook.
16269 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16271 (try_to_simplify): Likewise.
16273 2017-06-19 Martin Liska <mliska@suse.cz>
16276 * gimplify.c (gimplify_switch_expr):
16277 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16279 2017-06-19 Martin Liska <mliska@suse.cz>
16281 * doc/install.texi: Document that PGO runs in 4 stages.
16283 2017-06-19 Martin Liska <mliska@suse.cz>
16286 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16287 to dispatcher function name.
16288 * multiple_target.c (replace_function_decl): New function.
16289 (create_dispatcher_calls): Redirect both edges and references.
16291 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16293 * profile-count.c (profile_count::dump): Dump quality.
16294 (profile_count::differs_from_p): Update for unsigned val.
16295 * profile-count.h (profile_count_quality): New enum.
16296 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16298 2017-06-19 Richard Biener <rguenther@suse.de>
16300 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16301 struct function as arg.
16302 (estimate_numbers_of_iterations): Export overload with loop arg.
16303 (free_numbers_of_iterations_estimates_loop): Use an overload of
16304 free_numbers_of_iterations_estimates instead.
16305 * tree-cfg.c (remove_bb): Adjust.
16306 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16307 * tree-parloops.c (gen_parallel_loop): Likewise.
16308 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16310 (tree_unroll_loops_completely): Likewise.
16311 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16312 Use an overload instead and export.
16313 (estimated_loop_iterations): Adjust.
16314 (max_loop_iterations): Likewise.
16315 (likely_max_loop_iterations): Likewise.
16316 (estimate_numbers_of_iterations): Take struct function as arg
16318 (loop_exits_before_overflow): Adjust.
16319 (free_numbers_of_iterations_estimates_loop): Use an overload.
16320 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16321 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16323 2017-06-19 Richard Biener <rguenther@suse.de>
16326 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16327 RANGE_EXPR conservatively.
16329 2017-06-16 Carl Love <cel@us.ibm.com>
16331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16332 definitions for vec_float, vec_float2, vec_floato,
16333 vec_floate built-ins.
16334 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16335 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16337 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16338 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16339 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16340 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16341 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16342 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16343 vec_floato): Add builtin defines.
16344 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16345 Update the built-in documentation file for the new built-in
16348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16350 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16351 (mthumb): Mark as the negative of -marm.
16353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16355 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16357 (ARM Options, -mtune): Document that this accepts the same
16358 extension options as -mcpu.
16359 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16361 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16363 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16364 permitted extensions.
16366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16368 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16369 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16370 (armv8-m.main): Add option +nodsp.
16371 * config/arm/arm-cpu-cdata.h: Regenerated.
16373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16375 * config/arm/t-fuchsia: New file.
16376 * config.gcc (arm*-*-fuchsia*): Use it.
16378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16380 * config/arm/t-symbian: Rewrite for new option infrastructure.
16382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16384 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16385 (MULTILIB_REQUIRED): Likewise.
16387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16389 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16390 (MULTILIB_RESUE): Likewise.
16391 (MULTILIB_MATCHES): Likewise.
16392 (MULTLIB_REQUIRED): Likewise.
16394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16396 * config/arm/t-rtems: Rewrite for new option framework.
16398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16400 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16401 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16402 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16403 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16404 * config/arm/t-multilib: ... here.
16405 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16406 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16407 armv7-a and armv8*-a when A-profile libraries have not been built.
16408 * config/arm/t-rmprofile: Rewrite.
16410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16412 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16413 with a backslash. Remove the backslash after substituting unescaped
16415 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16419 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16420 the driver to pass through the default setting of -mfloat-abi.
16421 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16423 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16424 (all_feat_combs): New rule.
16425 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16427 * config/arm/t-aprofile: Rewrite.
16429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16431 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16432 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16433 fpu is not specified by the user/command-line.
16434 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16435 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16436 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16437 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16438 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16439 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16441 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16443 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16444 * config/arm/t-arm-elf: Rewritten.
16446 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16448 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16449 have some floating-point instructions.
16450 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16451 (TARGET_MAYBE_HARD_FLOAT): New macro.
16452 * config/arm/arm-builtins.c (arm_init_builtins): Use
16453 TARGET_MAYBE_HARD_FLOAT.
16454 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16458 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16459 (configargs.h): Include it.
16460 (arm_print_hint_for_fpu_option): New function.
16461 (arm_parse_fpu_option): New function.
16462 (candidate_extension): New class.
16463 (arm_canon_for_multilib): New function.
16464 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16465 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16466 (ARCH_CANONICAL_SPECS): New macro.
16467 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16471 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16472 are set after handling multilib fragments. Set target_cpu_default2
16475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16477 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16479 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16480 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16482 (check_arch): Likewise.
16483 * config/arm/arm.c (arm_configure_build_target): Handle
16484 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16485 options in the default.
16487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16489 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16490 when an option is an alias of another.
16491 * config/arm/parsecpu.awk (optalias): New parser token.
16492 (gen_comm_data): Mark non-alias options as such. Emit entries
16493 for extension aliases.
16494 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16495 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16496 (armv6kz, armv6zk, armv6t2): Likewise.
16497 (armv7): Make vfpv3-d16 an alias.
16498 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16500 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16501 Sort in canonical order.
16502 (armv8-a): Sort in canonical order.
16503 (armv8.1-a, armv8.2-a): Likewise.
16504 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16506 (cortex-a9): Sort in canonical order.
16507 * config/arm/arm.c (selftests.h): Include it.
16508 (arm_test_cpu_arch_data): New function.
16509 (arm_run_self_tests): New function.
16510 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16511 (targetm): Move declaration to the end of the file.
16512 * arm-cpu-cdata.h: Regenerated.
16514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16516 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16517 call to target_mode_check describing the type of option passed.
16518 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16519 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16520 arm_parse_cpu_option_name to match parameters against list of
16522 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16523 arm_arch_core_flags data structure.
16524 * config/arm/arm-cpu_cdata.h: Regenerated.
16526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16528 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16530 (arm_print_hint_for_cpu_option): Likewise.
16531 (arm_print_hint_for_arch_option): Likewise.
16532 (arm_parse_cpu_option_name): Likewise.
16533 (arm_parse_arch_option_name): Likewise.
16534 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16535 of entries in the all_fpus list.
16536 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16537 (arm_parse_cpu_option_name): Declare.
16538 (arm_parse_arch_option_name): Declare.
16539 (arm_parse_option_features): Declare.
16540 (arm_intialize_isa): Declare.
16541 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16543 (gen_comm_data): ... here. Make definitions non-static.
16544 * config/arm/arm-cpu-data.h: Regenerated.
16545 * config/arm/arm-cpu-cdata.h: Regenerated.
16547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16549 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16550 (cpu_arch_extension): New structure.
16551 (cpu_arch_option, arch_option, cpu_option): New structures.
16552 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16553 architecture types.
16554 (gen_data): Generate new format data tables.
16555 * config/arm/arm.c (cpu_tune): New structure.
16556 (cpu_option, processors): Delete.
16557 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16558 (arm_print_hint_for_cpu_option): ... this and ...
16559 (arm_print_hint_for_arch_option): ... this.
16560 (arm_parse_arch_cpu_name): Delete. Replace with ...
16561 (arm_parse_cpu_option_name): ... this and ...
16562 (arm_parse_arch_option_name): ... this.
16563 (arm_unrecognized_feature): Change type of target parameter to
16565 (arm_parse_arch_cpu_features): Delete. Replace with ...
16566 (arm_parse_option_features): ... this.
16567 (arm_configure_build_target): Rework to use new configuration data
16569 (arm_print_tune_info): Rework for new configuration data tables.
16570 * config/arm/arm-cpu-data.h: Regenerated.
16571 * config/arm/arm-cpu.h: Regenerated.
16573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16575 * Makefile.in (OBJS): Move sbitmap.o from here ...
16576 (OBJS-libcommon): ... to here.
16578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16580 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16581 (ISA_ALL_CRYPTO): New macro.
16582 (ISA_ALL_SIMD): New macro
16583 (ISA_ALL_FP): New macro.
16584 * config/arm/arm.c (fpu_bitlist): Update initializer.
16585 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16587 (arm9e): Add fpu. Add option for nofp
16588 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16589 (arm926ej-s, arm1026ej-s): Likewise.
16590 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16591 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16592 neon-fp16, neon-vfpv4, nofp and nosimd.
16593 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16594 (cortex-a8): Add fpu. Add option for nofp.
16595 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16596 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16597 (cortex-r4f): Add fpu.
16598 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16599 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16601 (cortex-r8): Likewise.
16602 (cortex-m4): Add fpu. Add option for nofp.
16603 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16604 (cortex-a17.cortex-a7): Likewise.
16605 (cortex-a32): Add fpu. Add options for crypto and nofp.
16606 (cortex-a35, cortex-a53): Likewise.
16607 (cortex-a57): Add fpu. Add option for crypto.
16608 (cortex-a72, cortex-a73): Likewise.
16609 (exynos-m1): Likewise.
16610 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16611 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16612 (cortex-m33): Add fpu. Add option for nofp.
16613 * config/arm/arm-cpu-cdata.h: Regenerated
16614 * config/arm/arm-cpu-data.h: Regenerated.
16616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16618 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16619 (armv5te, armv5tej): Likewise.
16620 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16621 (armv7): Add options fp and vfpv3-d16.
16622 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16623 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16625 (armv7ve): Likewise.
16626 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16627 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16628 (armv8-a): Add nocrypto option.
16629 (armv8.1-a, armv8.2-a): Likewise.
16630 (armv8-m.main): add options fp, fp.dp and nofp.
16632 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16634 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16636 (armv8-a+crc): Delete.
16637 (armv8.1-a): Add options simd, crypto and nofp.
16638 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16639 (armv8.2-a+fp16): Delete.
16640 (armv8-m.main): Add option dsp.
16641 (armv8-m.main+dsp): Delete.
16642 (cortex-a8): Add fpu. Add nofp option.
16643 (cortex-a9): Add fpu. Add nofp and nosimd options.
16644 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16645 link to main cpu and architecture data structures.
16646 (gen_comm_data): Only put isa attributes from the main architecture
16648 (option): New statement for architecture and CPU entries.
16649 * arm.c (struct cpu_option): New structure.
16650 (struct processors): Add entry for options.
16651 (arm_unrecognized_feature): New function.
16652 (arm_parse_arch_cpu_name): Ignore any characters after the first
16654 (arm_parse_arch_cpu_feature): New function.
16655 (arm_configure_build_target): Separate out any CPU and architecture
16656 features and parse separately. Don't error out if -mfpu=auto is
16657 used with only an architecture string.
16658 (arm_print_asm_arch_directives): New function.
16659 (arm_file_start): Call it.
16660 * config/arm/arm-cpu-cdata.h: Regenerated.
16661 * config/arm/arm-cpu-data.h: Likewise.
16662 * config/arm/arm-tables.opt: Likewise.
16664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16666 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16667 assembler when it is not -mfpu=auto.
16669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16671 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16672 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16673 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16674 (ASM_CPU_SPEC): Rewrite.
16675 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16676 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16677 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16678 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16679 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16680 copied string is NUL-terminated. Also strip any characters prefixed
16682 (arm_rewrite_selected_arch): New function.
16683 (arm_rewrite_march): New function.
16685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16687 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16688 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16689 (march, mcpu, mtune): Convert to string-based options.
16690 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16691 (arm_parse_arch_cpu_name): New function.
16692 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16693 identify selected architecture or CPU.
16694 (arm_option_save): New function.
16695 (TARGET_OPTION_SAVE): Redefine.
16696 (arm_option_restore): Restore string options.
16697 (arm_option_print): Print string options.
16699 2017-06-16 Martin Sebor <msebor@redhat.com>
16701 PR tree-optimization/80933
16702 PR tree-optimization/80934
16703 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16704 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16705 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16706 (gimple_fold_builtin): Call them.
16708 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16710 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16711 as unlikely; update profile.
16713 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16715 * predict.c (force_edge_cold): Handle declaring edges impossible
16718 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16720 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16722 (try_unroll_loop_completely): Fix reporting.
16724 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16726 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16728 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16731 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16732 if given a non-constant argument for an intrinsic which requires a
16735 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16737 * profile.c (compare_freqs): New function.
16738 (branch_prob): Sort edge list.
16739 (find_spanning_tree): Assume that the list is priority sorted.
16741 2017-06-16 Richard Biener <rguenther@suse.de>
16743 PR tree-optimization/81090
16744 * passes.def (pass_record_bounds): Remove.
16745 * tree-pass.h (make_pass_record_bounds): Likewise.
16746 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16747 make_pass_record_bounds): Likewise.
16748 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16749 not free niter estimates at the beginning but at the end.
16750 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16752 2017-06-16 Richard Biener <rguenther@suse.de>
16754 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16755 initializer to workaround ICE in host GCC 4.8.
16757 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16759 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16761 (clone_inlined_nodes): Update.
16763 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16765 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16766 prefetch settings, and enable prefetching by default at -O3.
16768 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16770 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16771 Set flag_prefetch_loop_arrays according to tuning data.
16773 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16775 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16776 New tune structure.
16777 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16778 [Unrelated to main purpose of the patch] Place the pointer field last
16779 to enable type checking errors when tune structure are wrongly merged.
16780 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16781 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16782 (thunderx2t99_prefetch_tune): New tune constants.
16783 (tune_params *_tunings): Update all tunings (no functional change).
16784 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16785 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16786 from tunings structures.
16788 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16791 * ubsan.c (instrument_null): Add T argument, use it instead
16792 of computing it based on IS_LHS.
16793 (instrument_object_size): Likewise.
16794 (pass_ubsan::execute): Adjust instrument_null and
16795 instrument_object_size callers to pass gimple_get_lhs or
16796 gimple_assign_rhs1 result to it. Use instrument_null instead of
16797 calling get_base_address and instrument_mem_ref. Handle
16798 aggregate call arguments for object-size sanitization.
16800 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16802 PR tree-optimization/81089
16803 * tree-vrp.c (is_masked_range_test): Validate operands of
16806 2017-06-15 Martin Sebor <msebor@redhat.com>
16809 * dumpfile.c (dump_register): Avoid calling memset to initialize
16810 a class with a default ctor.
16811 * gcc.c (struct compiler): Remove const qualification.
16812 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16813 * hash-table.h: Ditto.
16814 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16816 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16817 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16819 * params.h (struct param_info): Make struct members non-const.
16820 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16821 with default initialization.
16822 * vec.h (vec_copy_construct, vec_default_construct): New helper
16824 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16825 with vec_copy_construct.
16826 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16827 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16828 * doc/invoke.texi (-Wclass-memaccess): Document.
16830 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16832 * emit-rtl.h (is_leaf): Update comment about local
16833 register allocator.
16835 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16838 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16839 for a variable to have a section before checking if the section has a
16841 Set section to.persistent if persistent attribute is set.
16842 Warn if .persistent attribute is used on an automatic variable.
16844 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16846 PR rtl-optimization/80474
16847 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16849 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16851 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16854 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16856 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16857 (casesi): Emit bounds checking as RTL.
16858 (casesi_internal_mips16_<mode>): Remove bounds checking.
16860 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16862 * config/xtensa/xtensa.c (xtensa_option_override): Append
16863 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16864 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16865 xtensa_doloop_hooks): Define unconditionally.
16866 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16868 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16869 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16870 for it in xtensa_option_override.
16871 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16872 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16874 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16876 * doc/cppopts.texi: Document '-' special value to -MF.
16878 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16880 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16881 (cortex_a53_fconst): Likewise.
16882 (cortex_a53_fpmul): Likewise.
16883 (cortex_a53_f_load_64): Likewise.
16884 (cortex_a53_f_load_many): Likewise.
16885 (cortex_a53_advsimd_alu): Likewise.
16886 (cortex_a53_advsimd_alu_q): Likewise.
16887 (cortex_a53_advsimd_mul): Likewise.
16888 (cortex_a53_advsimd_mul_q): Likewise.
16889 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16890 Add missing fmul, r2f_cvt and fconst cases.
16892 2017-06-14 Richard Biener <rguenther@suse.de>
16894 PR middle-end/81088
16895 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16897 (fold_binary_loc): When associating do not treat pre-existing
16898 TREE_OVERFLOW on literal constants as a reason to allow
16899 TREE_OVERFLOW on associated literal constants.
16901 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16903 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16904 (MASK_FEATURES): New macro.
16905 * config/sparc/sparc.c (sparc_option_override): Remove the special
16906 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16908 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16910 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16911 a division of 0 if non-call exceptions are enabled.
16913 2017-06-14 Andrew Pinski <apinski@cavium.com>
16914 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16917 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16918 Improve vector initialization code gen for only variable case.
16920 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16922 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16924 2017-06-14 Richard Biener <rguenther@suse.de>
16926 PR tree-optimization/81083
16927 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16930 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16932 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16933 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16934 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16935 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16936 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16937 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16939 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16941 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16942 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16944 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16946 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16950 * config/rs6000/t-rtems: Don't handle SPE.
16952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16954 * config/rs6000/t-linux: Don't handle SPE.
16956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16958 * config/rs6000/eabispe.h: Delete file.
16960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16962 * config/rs6000/t-spe: Delete file.
16964 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16966 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16967 (rs6000_legitimate_offset_address_p): Return false for anything in
16968 V2SImode or V2SFmode.
16970 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16972 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16973 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16974 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16976 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16977 (rs6000_legitimate_offset_address_p): Ditto.
16978 (rs6000_emit_move): Ditto.
16979 (rs6000_init_builtins): Remove V4HI_type_node.
16981 2017-06-13 Martin Liska <mliska@suse.cz>
16984 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16985 (gate_asan): Likewise.
16986 * asan.h (asan_no_sanitize_address_p): Remove the function.
16987 (sanitize_flags_p): New function.
16988 * builtins.def: Fix coding style.
16989 * common.opt: Use renamed enum value.
16990 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16991 * doc/extend.texi: Document no_sanitize attribute.
16992 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16993 to SANITIZE_UNDEFINED_NONDEFAULT.
16994 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16995 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16996 Use sanitize_flags_p.
16997 * gimplify.c (gimplify_function_tree): Likewise.
16998 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16999 * opts.c (parse_no_sanitize_attribute): New function.
17000 (common_handle_option): Use renamed enum value.
17001 * opts.h (parse_no_sanitize_attribute): Declare.
17002 * tree.c (sanitize_flags_p): New function.
17003 * tree.h: Declared here.
17004 * tsan.c: Use sanitize_flags_p.
17005 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17006 (instrument_mem_ref): Likewise.
17007 (instrument_bool_enum_load): Likewise.
17008 (do_ubsan_in_current_function): Remove the function.
17009 (pass_ubsan::execute): Use sanitize_flags_p.
17010 * ubsan.h: Remove do_ubsan_in_current_function
17011 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17012 (dump_function_to_file): Use it here.
17014 2017-06-13 Martin Jambor <mjambor@suse.cz>
17016 PR tree-optimization/80803
17017 PR tree-optimization/81063
17018 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17019 (propagate_subaccesses_across_link): Enqueue subtree whenever
17020 necessary instead of relying on the caller.
17022 2017-06-13 Martin Jambor <mjambor@suse.cz>
17024 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17025 that have a first_link.
17026 (sort_and_splice_var_accesses): Do not check first_link before
17028 (subtree_mark_written_and_enqueue): Likewise.
17029 (propagate_all_subaccesses): Likewise and do not stop at first
17030 parent with a first_link.
17032 2017-06-13 Martin Jambor <mjambor@suse.cz>
17034 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17037 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17039 * match.pd: New pattern.
17041 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17043 * tree-vrp.c (is_masked_range_test): New function.
17044 (register_edge_assert_for): Determine ranges for
17047 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17049 PR tree-optimization/67328
17050 * fold-const.c (maskable_range_p): New function.
17051 (build_range_check): Generate bittests if possible.
17053 2017-06-13 Martin Liska <mliska@suse.cz>
17055 * gimple-pretty-print.c (dump_probability): Add new argument.
17056 (dump_edge_probability): Dump both probability and count.
17057 (dump_gimple_label): Likewise.
17058 (dump_gimple_bb_header): Likewise.
17060 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17063 * config/avr/avr-devices.c: Fix indentation.
17064 * config/avr/gen-avr-mmcu-specs.c: Dito.
17066 2017-06-13 Richard Biener <rguenther@suse.de>
17068 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17069 instead get vector type from stmt_info.
17070 (vectorizable_reduction): Adjust. Remove dead code.
17072 2017-06-13 Richard Biener <rguenther@suse.de>
17074 PR middle-end/81065
17075 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17076 case of C * (x * C2 + C3).
17077 (fold_addr_of_array_ref_difference): Properly fold index difference.
17079 2017-06-12 David S. Miller <davem@davemloft.net>
17082 * config/sparc/sparc.md (return expander): Emit frame blockage if
17083 function uses alloca.
17085 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17087 * combine.c (make_field_assignment): Check len rather than the mode
17088 precision when calling force_to_mode.
17090 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17092 Support multilibs and devices that see flash in RAM address range.
17095 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17096 (avr_mcu_t) <flash_pm_offset>: New field.
17097 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17098 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17099 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17100 (AVR_TINY_PM_OFFSET): Remove macro.
17101 * config/avr/avr.opt (-mshort-calls): New option.
17102 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17103 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17104 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17105 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17106 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17107 instead of avr_arch->have_jmp_call.
17108 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17109 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17110 avr_arch->flash_pm_offset to define.
17111 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17112 new field flash_pm_offset. Add entry for avrxmega3.
17113 (avr_texinfo): Add entry for avrxmega3.
17114 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17115 attiny212, attiny214,
17116 attiny412, attiny414, attiny416, attiny417,
17117 attiny814, attiny816, attiny817,
17118 attiny1614, attiny1616, attiny1617,
17119 attiny3214, attiny3216, attiny3217.
17120 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17121 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17122 (avr_print_operand_address) [AVR_TINY]: Same.
17123 (avr_asm_init_sections) <readonly_data_section>: Only patch
17124 callback if avr_arch->flash_pm_offset = 0.
17125 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17126 for rodata if avr_arch->flash_pm_offset != 0.
17127 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17128 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17129 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17130 (m_options): Append opt_rcall.
17131 (m_dirnames): Append dir_rcall.
17132 * config/avr/t-multilib: Regenerate.
17134 * configure.ac [target=avr]: Check whether avrxmega3 default
17135 linker description file works as needed.
17136 * configure: Regenerate.
17137 * doc/avr-mmcu.texi: Regenerate.
17138 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17139 <__AVR_ARCH__>: Document avrxmega3 and 103.
17140 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17141 <__AVR_SHORT_CALLS__>: Document it.
17142 <__AVR_PM_BASE_ADDRESS__>: Document it.
17143 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17144 (AVR Variable Attributes) <progmem>: Document this is
17145 not needed for avrxmega3.
17146 (AVR Named Address Spaces) <__flash>: Dito.
17148 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17150 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17152 2017-06-12 Doug Rupp <rupp@adacore.com>
17154 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17155 Append vxworks-stdint.h to the tm_file list.
17156 * config/vxworks-stdint.h: New file.
17158 2017-06-12 Martin Liska <mliska@suse.cz>
17160 PR tree-optimization/81041
17161 * tree-profile.c (gimple_gen_ic_func_profiler):
17162 Create an extra BB in profile-generate
17163 (gimple_gen_time_profiler): Likewise.
17165 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17167 PR tree-optimization/81003
17168 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17169 (update_range_test): Use it instead of force_gimple_operand_gsi.
17171 2017-06-12 Richard Biener <rguenther@suse.de>
17173 PR tree-optimization/81053
17174 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17175 with backedge value not defined in loop. Simplify def stmt
17178 2017-06-11 Tom de Vries <tom@codesourcery.com>
17181 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17183 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17184 nvptx_cannot_force_const_mem.
17186 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17188 * opts.c (finish_options): Move test for flag_split_stack after
17189 it has been initialized.
17191 2017-06-11 Jason Merrill <jason@redhat.com>
17193 * tree.h (id_equal): New.
17194 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17195 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17196 instead of strcmp of IDENTIFIER_POINTER.
17198 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17200 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17201 (mark_all_inlined_calls_cdtor): Fix formating.
17202 (inline_transform): Rescale profile before inlining.
17204 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17206 * cgraph.h (cgraph_edge::clone): Update prototype.
17207 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17208 (cgraph_node::create_clone): Update.
17209 (cgraph_node::create_version_clone): Update.
17210 * tree-inline.c (copy_bb): Update.
17211 (expand_call_inline): Update.
17213 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17215 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17216 factored out from ...
17217 (rs6000_emit_prologue): ... here.
17219 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17221 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17222 factored out from ...
17223 (rs6000_emit_prologue): ... here.
17225 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17227 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17229 (handle_missing_profiles): Fix computation of tp_first_run.
17230 (counts_to_freqs): Do not touch freqs when count is 0.
17232 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17234 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17237 2017-06-10 Tom de Vries <tom@codesourcery.com>
17239 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17240 attributes): Document signal effective target.
17242 2017-06-10 Tom de Vries <tom@codesourcery.com>
17244 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17245 Document effective target stack_size.
17247 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17249 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17250 to the edit_context if they can be auto-applied.
17252 2017-06-9 Ian Lance Taylor <iant@golang.org>
17254 * opts.c (finish_options): If -fsplit-stack, disable implicit
17255 -forder-blocks-and-partition.
17256 * doc/invoke.texi (Optimize Options): Document that when using
17257 -fsplit-stack -forder-blocks-and-partition is not implicitly
17260 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17262 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17263 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17264 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17265 * builtins.def (abort, trap, unreachable): Declare cold.
17266 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17267 * tree-core.h (ECF_COLD): New.
17268 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17269 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17271 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17273 * predict.c (unlikely_executed_stmt_p): Cleanup.
17275 2017-06-09 Richard Biener <rguenther@suse.de>
17277 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17278 model if the ref is always written to.
17280 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17282 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17284 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17286 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17289 2017-06-09 Tom de Vries <tom@codesourcery.com>
17292 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17293 "target cannot support label values" when encountering LABEL_REF.
17295 2017-06-09 Martin Liska <mliska@suse.cz>
17297 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17298 (gimple_gen_ic_func_profiler): Emit direct comparison
17299 of __gcov_indirect_call_callee with NULL.
17300 (gimple_gen_time_profiler): Change probability from
17301 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17303 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17305 * profile.c (edge_gcov_counts): Turn to pointer.
17306 (compute_branch_probabilities, compute_branch_probabilities): Update.
17307 (branch_prob): Do not clear edge_gcov_count.
17308 * profile.h (edge_gcov_counts): Turn to pointer.
17309 (edge_gcov_count): Update.
17311 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17313 * gimple.h (gimple_check_failed): Mark cold.
17315 2017-06-09 Richard Biener <rguenther@suse.de>
17317 PR tree-optimization/66623
17318 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17319 refactor check_reduction into two parts, properly computing
17320 whether we have to check reduction validity for outer loop
17323 2017-06-09 Richard Biener <rguenther@suse.de>
17325 PR tree-optimization/79483
17326 * graphite-scop-detection.c (order): New global.
17327 (get_order): Compute bb to order mapping that satisfies code
17328 generation constraints.
17329 (cmp_pbbs): New helper.
17330 (build_scops): Start domwalk at entry block, sort generated
17333 2017-06-09 Richard Biener <rguenther@suse.de>
17335 PR middle-end/81007
17336 * ipa-polymorphic-call.c
17337 (ipa_polymorphic_call_context::restrict_to_inner_class):
17338 Skip FIELD_DECLs with error_mark_node type.
17339 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17342 2017-06-09 Martin Liska <mliska@suse.cz>
17344 * predict.c (struct branch_predictor): New struct.
17345 (test_prediction_value_range): New test.
17346 (predict_c_tests): New function.
17347 * selftest-run-tests.c (selftest::run_tests): Run the function.
17348 * selftest.h: Declare new tests.
17350 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17353 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17354 gen_add3_insn did not fail.
17355 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17356 r0, construct that number in a temporary reg and add that reg to r0.
17357 If asked to put the result in r0 as well, fail.
17359 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17362 for early expansion of vec_eqv.
17364 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17366 PR middle-end/81005
17367 * ubsan.c (instrument_null): Avoid pointless code temporary.
17368 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17370 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17374 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17377 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17380 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17382 * predict.c (unlikely_executed_edge_p): Move ahead.
17383 (probably_never_executed_edge_p): Use it.
17385 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17387 PR middle-end/79988
17388 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17389 gimple_call_builtin_p call.
17391 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17393 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17394 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17395 rtl_check_failed_type2, rtl_check_failed_code1,
17396 rtl_check_failed_code2, rtl_check_failed_code_mode,
17397 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17398 rtvec_check_failed_bounds, rtl_check_failed_flag,
17399 _fatal_insn_not_found, _fatal_insn): Likewise.
17400 * tree.h (tree_contains_struct_check_failed,
17401 tree_check_failed, tree_not_check_failed,
17402 tree_class_check_failed, tree_range_check_failed,
17403 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17404 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17405 tree_operand_check_failed, omp_clause_check_failed,
17406 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17409 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17411 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17412 flag_branch_probabilities.
17413 * ipa-inline.c (edge_badness): Likewise.
17414 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17415 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17416 * predict.c (maybe_hot_frequency_p): Likewise.
17417 (probably_never_executed): Likewise.
17418 * sched-ebb.c (schedule_ebbs): Likewise.
17419 * sched-rgn.c (find_single_block_region): Likewise.
17420 * tracer.c (tail_duplicate): Likewise.
17422 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17424 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17425 longer requires x_flag_profile_use.
17427 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17429 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17430 instead of flag_reorder_blocks_and_partition.
17431 * dbxout.c (dbxout_function_end): Likewise.
17432 * dwarf2out.c (gen_subprogram_die): Likewise.
17433 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17434 * hw-doloop.c (reorg_loops): Likewise.
17435 * varasm.c (assemble_start_function,
17436 assemble_end_function): Likewise.
17437 (decide_function_section): Do not check for
17438 flag_reorder_blocks_and_partition.
17440 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17442 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17444 (chkp_get_hard_register_fake_addr_expr): Ditto.
17445 (chkp_build_addr_expr): Add check for hard reg case.
17446 (chkp_parse_array_and_component_ref): Ditto.
17447 (chkp_find_bounds_1): Ditto.
17448 (chkp_process_stmt): Don't generate bounds store for
17451 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17453 * predict.c (maybe_hot_bb_p): Do not check profile status.
17454 (maybe_hot_edge_p): Likewise.
17455 (probably_never_executed): Check for zero counts even if profile
17457 (unlikely_executed_edge_p): New function.
17458 (unlikely_executed_stmt_p): New function.
17459 (unlikely_executed_bb_p): New function.
17460 (set_even_probabilities): Use unlikely predicates.
17461 (combine_predictions_for_bb): Likewise.
17462 (predict_paths_for_bb): Likewise.
17463 (predict_paths_leading_to_edge): Likewise.
17464 (determine_unlikely_bbs): New function.
17465 (estimate_bb_frequencies): Use it.
17466 (compute_function_frequency): Use zero counts even if profile is
17468 * profile-count.h: Fix typo.
17470 2017-08-08 Julia Koval <julia.koval@intel.com>
17472 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17473 _mm512_mask_cvtsepi16_storeu_epi8,
17474 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17475 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17476 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17477 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17478 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17479 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17480 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17481 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17482 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17483 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17484 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17485 __builtin_ia32_pmovuswb256mem_mask,
17486 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17487 __builtin_ia32_pmovwb512mem_mask): New builtins.
17489 2017-08-08 Julia Koval <julia.koval@intel.com>
17491 PR target/73350,80862
17492 * config/i386/subst.md (round): Fix round pattern.
17493 * config/i386/i386.c (ix86_erase_embedded_rounding):
17494 Fix erasing rounding for the fixed pattern.
17496 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17498 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17500 2017-06-08 Martin Liska <mliska@suse.cz>
17502 PR gcov-profile/80911
17503 * gcov.c (block_info::block_info): New constructor.
17505 2017-06-07 Carl Love <cel@us.ibm.com>
17507 * config/rs6000/rs6000-c: The return type of the following
17508 built-in functions was implemented as int not long long. Fix sign
17509 of return value for the unsigned version of vec_mulo and vec_mule.
17510 vector unsigned long long vec_bperm (vector unsigned long long,
17511 vector unsigned char)
17512 vector signed long long vec_mule (vector signed int,
17514 vector unsigned long long vec_mule (vector unsigned int,
17515 vector unsigned int)
17516 vector signed long long vec_mulo (vector signed int,
17518 vector unsigned long long vec_mulo (vector unsigned int,
17519 vector unsigned int)
17520 * doc/extend.texi: Fix the documentation for the built-in
17523 2017-06-07 Carl Love <cel@us.ibm.com>
17526 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17529 2017-06-07 Carl Love <cel@us.ibm.com>
17531 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17532 support, Generate doublehv for signed int/float for BE case only.
17534 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17536 * doc/invoke.texi (mcx16): Rewrite.
17538 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17540 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17541 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17542 *mov<mode>_softfloat, and an anonymous splitter): Use
17543 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17545 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17547 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17549 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17550 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17551 (rs6000_debug_reg_global): Adjust.
17552 (rs6000_init_hard_regno_mode_ok): Adjust.
17553 (rs6000_dbx_register_number): Adjust.
17554 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17555 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17556 Remove SPE_ACC and SPEFSCR.
17557 (REG_ALLOC_ORDER): Ditto.
17558 (FRAME_POINTER_REGNUM): Change to 111.
17559 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17560 (REG_CLASS_NAMES): Ditto.
17561 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17562 (REGISTER_NAMES): Ditto.
17563 (ADDITIONAL_REG_NAMES): Ditto.
17564 (rs6000_reg_names): Ditto.
17565 * config/rs6000/rs6000.md: Renumber some register number
17568 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17570 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17572 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17573 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17575 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17576 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17577 Delete the SPE high registers.
17578 (REG_ALLOC_ORDER): Ditto.
17579 (enum reg_class): Remove SPE_HIGH_REGS.
17580 (REG_CLASS_NAMES): Ditto.
17581 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17582 (REGISTER_NAMES): Ditto.
17583 (rs6000_reg_names): Ditto.
17584 * doc/tm.texi.in: Remove SPE as example.
17585 * doc/tm.texi: Regenerate.
17587 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17589 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17590 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17591 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17592 * config/rs6000/rs6000.md (type): Remove "brinc".
17594 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17596 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17597 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17598 * config/rs6000/linuxspe.h: Delete file.
17599 * config/rs6000/rs6000.md: Don't include spe.md.
17600 * config/rs6000/spe.h: Delete file.
17601 * config/rs6000/spe.md: Delete file.
17602 * config/rs6000/t-rs6000: Remove spe.md.
17604 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17606 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17607 (reg_or_none500mem_operand): Delete.
17608 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17609 instead of reg_or_none500mem_operand.
17611 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17614 handling of SPE flags.
17615 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17617 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17619 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17621 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17622 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17623 paired_divv2sf3): Similar.
17624 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17625 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17626 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17628 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17629 Rename the paired_* instruction patterns.
17630 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17632 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17633 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17634 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17635 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17636 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17637 PAIRED_VECTOR_MODE.
17638 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17639 (spe_func_has_64bit_regs_p): Delete.
17640 (spe_expand_predicate_builtin): Delete.
17641 (spe_expand_evsel_builtin): Delete.
17642 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17643 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17644 (invalid_e500_subreg): Delete.
17645 (rs6000_legitimize_address): Always force_reg op2 as well, for
17646 paired single memory accesses.
17647 (rs6000_member_type_forces_blk): Delete.
17648 (rs6000_spe_function_arg): Delete.
17649 (rs6000_expand_unop_builtin): Delete SPE handling.
17650 (rs6000_expand_binop_builtin): Ditto.
17651 (spe_expand_stv_builtin): Delete.
17652 (bdesc_2arg_spe): Delete.
17653 (spe_expand_builtin): Delete.
17654 (spe_expand_predicate_builtin): Delete.
17655 (spe_expand_evsel_builtin): Delete.
17656 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17657 (spe_init_builtins): Delete.
17658 (spe_func_has_64bit_regs_p): Delete.
17659 (savres_routine_name): Delete "info" parameter. Adjust callers.
17660 (rs6000_emit_stack_reset): Ditto.
17661 (rs6000_dwarf_register_span): Delete.
17662 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17663 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17664 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17666 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17668 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17669 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17670 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17671 mulv2sf3, divv2sf3): Delete expanders.
17673 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17675 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17679 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17680 * config/rs6000/rs6000.c: Ditto.
17682 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17684 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17685 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17686 comparison_operator.
17688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17690 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17691 * config/rs6000/rs6000.opt: Ditto.
17692 * config/rs6000/t-rtems: Ditto.
17694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17696 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17697 TARGET_E500_SINGLE by 0, simplify.
17698 * config/rs6000/rs6000.c: Ditto.
17699 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17700 (spe_build_register_parallel): Delete.
17701 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17702 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17703 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17704 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17705 (E500_CONVERT): Delete.
17706 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17708 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17710 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17711 * config/rs6000/dfp.md: Ditto.
17712 (negdd2, *negdd2_fpr): Merge.
17713 (absdd2, *absdd2_fpr): Merge.
17714 (negtd2, *negtd2_fpr): Merge.
17715 (abstd2, *abstd2_fpr): Merge.
17716 * config/rs6000/e500.h: Delete file.
17717 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17718 TARGET_FPRS by 1 and simplify.
17719 * config/rs6000/rs6000-c.c: Ditto.
17720 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17721 TARGET_DF_SPE by 0.
17722 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17724 * config/rs6000/rs6000.md: Ditto.
17725 (floatdidf2, *floatdidf2_fpr): Merge.
17726 (move_from_CR_gt_bit): Delete.
17727 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17728 (E500_CR_IOR_COMPARE): Delete.
17729 (All patterns that require !TARGET_FPRS): Delete.
17730 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17732 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17734 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17738 * graphds.c (add_edge): Intitialize edge's attached data.
17739 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17740 pointer parameter. Call pointed function on each edge during
17741 graph traversing. Skip traversing the edge when the function
17743 (graphds_dfs, graphds_scc): Ditto.
17744 (for_each_edge): New parameter. Pass the new parameter to callback
17746 * graphds.h (skip_edge_callback): New function pointer type.
17747 (graphds_dfs, graphds_scc): New function pointer parameter.
17748 (graphds_edge_callback, for_each_edge): New parameter.
17750 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17752 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17753 out code checking if runtime alias check is possible to below ...
17754 Call the new function.
17755 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17756 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17758 2017-06-07 Marek Polacek <polacek@redhat.com>
17761 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17762 TYPE_OVERFLOW_WRAPS check.
17764 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17766 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17767 if versioning is required.
17768 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17769 peeling with the check for versioning.
17771 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17773 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17774 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17775 Set true to new parameter if new ssa variable is defined.
17776 (vect_gen_vector_loop_niters): Refactor. Set range information
17777 for the new vector loop bound variable.
17778 (vect_do_peeling): Ditto.
17780 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17782 * tree-affine.c (ssa.h): Include header file.
17783 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17784 has wrapping overflow behavior.
17786 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17788 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17792 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17793 (tree_to_aff_combination): ... here.
17795 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17797 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17798 reg_pressure model function.
17799 (ivopts_global_cost_for_size): Delete.
17800 (determine_set_costs, iv_ca_recount_cost): Call new model function
17801 ivopts_estimate_reg_pressure.
17803 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17805 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17806 expensive than udiv. Remove floating point cases from mod.
17808 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17810 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17811 Increase idiv cost.
17813 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17815 * config/aarch64/aarch64.md
17816 (copysignsf3): Fix mask generation.
17818 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17820 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17822 (class dump_manager): Add register_dumps method.
17823 * dumpfile.c: Include langhooks.h.
17824 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17825 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17826 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17827 (dump_manager::dump_register): Start with 512 entries instead of 32.
17828 (dump_manager::register_dumps): New method.
17829 * toplev.c (general_init): Instead of invoking register_dumps
17830 langhook, invoke register_dumps method on the dump manager.
17831 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17834 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17836 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17837 Say that # requires an associated define_split to exist, and that
17838 the define_split must be suitable for use after register allocation.
17840 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17842 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17843 (compute_outgoing_frequencies): Also initialize zero counts.
17844 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17845 around loops; preserve more of profile when nothing changes.
17847 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17849 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17851 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17852 * config/arm/arm-cpu-cdata.h: Regenerate.
17853 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17854 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17855 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17856 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17857 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17859 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17861 * config/arm/t-rmprofile: Likewise.
17862 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17864 2017-06-06 David S. Miller <davem@davemloft.net>
17867 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17868 blockage if function uses alloca.
17870 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17872 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17873 New "uid" fields to hold pretty-print IDs of group and ref.
17874 Memory references are now identified as <group_id>:<ref_id>
17875 instead of using [random] addresses.
17876 (dump_mem_details): Simplify, no functional change.
17877 (dump_mem_ref): Simplify and make output more concise.
17878 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17879 (find_or_create_group): Initialize group uid.
17880 (record_ref): Initialize ref uid. Improve debug output.
17881 (prune_group_by_reuse, should_issue_prefetch_p,)
17882 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17883 (mark_nontemporal_store, determine_loop_nest_reuse):
17884 Improve debug output.
17886 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17888 * dbgcnt.def (prefetch): New debug counter.
17889 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17890 (schedule_prefetches): Stop issueing prefetches if debug counter
17893 2017-06-06 Tom de Vries <tom@codesourcery.com>
17895 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17896 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17898 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17900 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17901 Use aarch64_reg_or_zero predicate for operand 4.
17902 (aarch64_compare_and_swap<mode> define_insn_and_split):
17903 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17904 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17906 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17908 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17909 (arm_compute_save_core_reg_mask): This.
17910 (thumb1_compute_save_reg_mask): Rename into ...
17911 (thumb1_compute_save_core_reg_mask): This.
17912 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17913 (arm_compute_frame_layout): Likewise.
17915 2017-06-06 Richard Biener <rguenther@suse.de>
17917 PR tree-optimization/80974
17918 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17919 keep or clear leaders SSA info.
17921 2017-06-06 Tom de Vries <tom@codesourcery.com>
17923 * config/nvptx/nvptx.c (split_mode_p): New function.
17924 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17926 2017-06-06 Tom de Vries <tom@codesourcery.com>
17928 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17930 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17933 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17936 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17938 * shrink-wrap.c (handle_simple_exit): Update profile.
17939 (try_shrink_wrapping): Upate profile.
17941 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17943 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17944 (tree_guess_outgoing_edge_probabilities): New.
17945 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17946 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17948 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17950 * ipa-split.c (split_function): Initialize return bb profile.
17952 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17954 * profile.c (compute_branch_probabilities): Also initialize
17955 EXIT_BLOCK profile.
17957 2017-06-06 Richard Biener <rguenther@suse.de>
17959 PR tree-optimization/80928
17960 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17961 (vect_analyze_loop_operations): Properly guard analysis for
17963 (vect_transform_loop): Likewise.
17964 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17965 (vect_model_induction_cost): Do not cost for pure SLP.
17966 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17967 of induction in inner loop vectorization.
17968 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17969 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17970 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17972 (vect_analyze_slp_cost_1): Cost induction.
17973 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17974 (vect_get_slp_vect_defs): Likewise.
17975 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17976 (vect_transform_stmt): Handle SLP reductions.
17977 * tree-vectorizer.h (vectorizable_induction): Adjust.
17979 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17981 * config/rs6000/rs6000.c (make_resolver_func): Update
17982 init_lowered_empty_function call.
17984 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17986 * doc/invoke.texi: Document the -fprofile-abs-path option.
17987 * common.opt (fprofile-abs-path): New option.
17988 * gcov-io.h (gcov_write_filename): Declare.
17989 * gcov-io.c (gcov_write_filename): New function.
17990 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17991 * profile.c (output_location): Likewise.
17993 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17995 * shring-wrap.c: Revert accidental commit.
17997 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17999 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18001 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18003 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18005 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18006 profile in callgraph edge.
18007 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18008 (apply_scale): Likewise.
18009 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18010 Also scale profile when inlining function with zero profile.
18011 (initialize_cfun): Update exit block profile even when it is zero.
18012 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18013 when profile is read.
18015 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18017 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18018 (CLONE_*): New constants to define the processors we can generate
18019 code for with the target_clone attribute.
18020 (rs6000_clone_map): New array to identify which clone processors
18021 the current program is running on.
18022 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18023 target_clone attribute.
18024 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18025 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18026 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18027 (cpu_expand_builtin): Add support for target_clone attribute.
18028 (rs6000_valid_attribute_p): Allow "default" attribute.
18029 (get_decl_name): New debug function to simplify printing the
18030 current function name in debugging statements.
18031 (rs6000_clone_priority): New functions to support the target_clone
18032 attribute, and be able to generate code to switch between ISA 2.05
18033 through ISA 3.0 (power6 through power9).
18034 (rs6000_compare_version_priority): Likewise.
18035 (rs6000_get_function_versions_dispatcher): Likewise.
18036 (make_resolver_func): Likewise.
18037 (add_condition_to_bb): Likewise.
18038 (dispatch_function_versions): Likewise.
18039 (rs6000_generate_version_dispatcher_body): Likewise.
18040 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18041 (fusion_gpr_load_p): Fix a spacing issue.
18042 * doc/extend.texi (Common Function Attributes): Document that the
18043 PowerPC supports the target_clone attribute.
18045 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18047 * config/arm/arm.h: explain F symbol found in description of ARM
18048 register allocation in its legend.
18050 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18052 * config/mips/frame-header-opt.c: Include profile-count.h.
18053 * config/riscv/riscv.c: Include profile-count.h
18055 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18057 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18059 (sm_set_flag_if_changed): Add bbs field.
18060 (execute_sm_if_changed_flag_set): Pass BBS.
18061 (execute_sm): Update.
18063 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18065 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18068 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18070 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18071 (peephole2): New peephole2 to emit the above.
18072 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18074 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18076 * config/aarch64/aarch64.c (define_peephole2 above
18077 *sub_<shift>_<mode>): New peephole.
18079 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18081 * config/i386/i386.c (make_resolver_func): Update.
18082 * Makefile.in: Add profile-count.h and profile-count.o
18083 * auto-profile.c (afdo_indirect_call): Update to new API.
18084 (afdo_set_bb_count): Update.
18085 (afdo_propagate_edge): Update.
18086 (afdo_propagate_circuit): Update.
18087 (afdo_calculate_branch_prob): Update.
18088 (afdo_annotate_cfg): Update.
18089 * basic-block.h: Include profile-count.h
18090 (struct edge_def): Turn count to profile_count.
18091 (struct basic_block_def): Likewie.
18092 (REG_BR_PROB_BASE): Move to profile-count.h
18093 (RDIV): Move to profile-count.h
18094 * bb-reorder.c (max_entry_count): Turn to profile_count.
18095 (find_traces): Update.
18096 (rotate_loop):Update.
18097 (connect_traces):Update.
18098 (sanitize_hot_paths):Update.
18099 * bt-load.c (migrate_btr_defs): Update.
18100 * cfg.c (RDIV): Remove.
18101 (init_flow): Use alloc_block.
18102 (alloc_block): Uninitialize count.
18103 (unchecked_make_edge): Uninitialize count.
18104 (check_bb_profile): Update.
18105 (dump_edge_info): Update.
18106 (dump_bb_info): Update.
18107 (update_bb_profile_for_threading): Update.
18108 (scale_bbs_frequencies_int): Update.
18109 (scale_bbs_frequencies_gcov_type): Update.
18110 (scale_bbs_frequencies_profile_count): New.
18111 * cfg.h (update_bb_profile_for_threading): Update.
18112 (scale_bbs_frequencies_profile_count): Declare.
18113 * cfgbuild.c (compute_outgoing_frequencies): Update.
18114 (find_many_sub_basic_blocks): Update.
18115 * cfgcleanup.c (try_forward_edges): Update.
18116 (try_crossjump_to_edge): Update.
18117 * cfgexpand.c (expand_gimple_tailcall): Update.
18118 (construct_exit_block): Update.
18119 * cfghooks.c (verify_flow_info): Update.
18120 (dump_bb_for_graph): Update.
18121 (split_edge): Update.
18122 (make_forwarder_block): Update.
18123 (duplicate_block): Update.
18124 (account_profile_record): Update.
18125 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18126 (get_estimated_loop_iterations): Update.
18127 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18128 (single_likely_exit): Update.
18129 * cfgloopmanip.c (scale_loop_profile): Update.
18131 (set_zero_probability): Update.
18132 (lv_adjust_loop_entry_edge): Update.
18133 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18134 (purge_dead_edges): Update.
18135 (rtl_account_profile_record): Update.
18136 * cgraph.c (cgraph_node::create): Uninitialize count.
18137 (symbol_table::create_edge): Uninitialize count.
18138 (cgraph_update_edges_for_call_stmt_node): Update.
18139 (cgraph_edge::dump_edge_flags): Update.
18140 (cgraph_node::dump): Update.
18141 (cgraph_edge::maybe_hot_p): Update.
18142 * cgraph.h: Include profile-count.h
18143 (create_clone), create_edge, create_indirect_edge): Update.
18144 (cgraph_node): Turn count to profile_count.
18145 (cgraph_edge0: Likewise.
18146 (make_speculative, clone): Update.
18147 (create_edge): Update.
18148 (init_lowered_empty_function): Update.
18149 * cgraphclones.c (cgraph_edge::clone): Update.
18150 (duplicate_thunk_for_node): Update.
18151 (cgraph_node::create_clone): Update.
18152 * cgraphunit.c (cgraph_node::analyze): Update.
18153 (cgraph_node::expand_thunk): Update.
18154 * final.c (dump_basic_block_info): Update.
18155 * gimple-streamer-in.c (input_bb): Update.
18156 * gimple-streamer-out.c (output_bb): Update.
18157 * graphite.c (print_global_statistics): Update.
18158 (print_graphite_scop_statistics): Update.
18159 * hsa-brig.c: Include basic-block.h.
18160 * hsa-dump.c: Include basic-block.h.
18161 * hsa-gen.c (T sum_slice): Update.
18162 (convert_switch_statements):Update.
18163 * hsa-regalloc.c: Include basic-block.h.
18164 * ipa-chkp.c (chkp_produce_thunks): Update.
18165 * ipa-cp.c (struct caller_statistics): Update.
18166 (init_caller_stats): Update.
18167 (gather_caller_stats): Update.
18168 (ipcp_cloning_candidate_p): Update.
18169 (good_cloning_opportunity_p): Update.
18170 (get_info_about_necessary_edges): Update.
18171 (dump_profile_updates): Update.
18172 (update_profiling_info): Update.
18173 (update_specialized_profile): Update.
18174 (perhaps_add_new_callers): Update.
18175 (decide_about_value): Update.
18176 (ipa_cp_c_finalize): Update.
18177 * ipa-devirt.c (struct odr_type_warn_count): Update.
18178 (struct decl_warn_count): Update.
18179 (struct final_warning_record): Update.
18180 (possible_polymorphic_call_targets): Update.
18181 (ipa_devirt): Update.
18182 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18183 * ipa-icf.c (sem_function::merge): Update.
18184 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18185 * ipa-inline.c (compute_uninlined_call_time): Update.
18186 (compute_inlined_call_time): Update.
18187 (want_inline_small_function_p): Update.
18188 (want_inline_self_recursive_call_p): Update.
18189 (edge_badness): Update.
18190 (lookup_recursive_calls): Update.
18191 (recursive_inlining): Update.
18192 (inline_small_functions): Update.
18193 (dump_overall_stats): Update.
18194 (dump_inline_stats): Update.
18195 * ipa-profile.c (ipa_profile_generate_summary): Update.
18196 (ipa_propagate_frequency): Update.
18197 (ipa_profile): Update.
18198 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18199 * ipa-utils.c (ipa_merge_profiles): Update.
18200 * loop-doloop.c (doloop_modify): Update.
18201 * loop-unroll.c (report_unroll): Update.
18202 (unroll_loop_runtime_iterations): Update.
18203 * lto-cgraph.c (lto_output_edge): Update.
18204 (lto_output_node): Update.
18205 (input_node): Update.
18206 (input_edge): Update.
18207 (merge_profile_summaries): Update.
18208 * lto-streamer-in.c (input_cfg): Update.
18209 * lto-streamer-out.c (output_cfg): Update.
18210 * mcf.c (create_fixup_graph): Update.
18211 (adjust_cfg_counts): Update.
18212 (sum_edge_counts): Update.
18213 * modulo-sched.c (sms_schedule): Update.
18214 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18215 * predict.c (maybe_hot_count_p): Update.
18216 (probably_never_executed): Update.
18217 (dump_prediction): Update.
18218 (combine_predictions_for_bb): Update.
18219 (propagate_freq): Update.
18220 (handle_missing_profiles): Update.
18221 (counts_to_freqs): Update.
18222 (rebuild_frequencies): Update.
18223 (force_edge_cold): Update.
18224 * predict.h: Include profile-count.h
18225 (maybe_hot_count_p, counts_to_freqs): UPdate.
18226 * print-rtl-function.c: Do not include cfg.h
18227 * print-rtl.c: Include basic-block.h
18228 * profile-count.c: New file.
18229 * profile-count.h: New file.
18230 * profile.c (is_edge_inconsistent): Update.
18231 (correct_negative_edge_counts): Update.
18232 (is_inconsistent): Update.
18233 (set_bb_counts): Update.
18234 (read_profile_edge_counts): Update.
18235 (compute_frequency_overlap): Update.
18236 (compute_branch_probabilities): Update; Initialize and deinitialize
18238 (branch_prob): Update.
18239 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18240 (edge_gcov_count): New.
18241 (bb_gcov_count): New.
18242 * shrink-wrap.c (try_shrink_wrapping): Update.
18243 * tracer.c (better_p): Update.
18244 * trans-mem.c (expand_transaction): Update.
18245 (ipa_tm_insert_irr_call): Update.
18246 (ipa_tm_insert_gettmclone_call): Update.
18247 * tree-call-cdce.c: Update.
18248 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18249 (gimple_duplicate_sese_tail): Update.
18250 (gimple_account_profile_record): Update.
18251 (execute_fixup_cfg): Update.
18252 * tree-inline.c (copy_bb): Update.
18253 (copy_edges_for_bb): Update.
18254 (initialize_cfun): Update.
18255 (freqs_to_counts): Update.
18256 (copy_cfg_body): Update.
18257 (expand_call_inline): Update.
18258 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18259 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18260 (try_unroll_loop_completely): Update.
18261 (try_peel_loop): Update.
18262 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18263 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18264 * tree-ssa-loop-split.c (connect_loops): Update.
18265 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18266 * tree-ssa-reassoc.c (branch_fixup): Update.
18267 * tree-ssa-tail-merge.c (replace_block_by): Update.
18268 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18269 (compute_path_counts): Update.
18270 (update_profile): Update.
18271 (recompute_probabilities): Update.
18272 (update_joiner_offpath_counts): Update.
18273 (estimated_freqs_path): Update.
18274 (freqs_to_counts_path): Update.
18275 (clear_counts_path): Update.
18276 (ssa_fix_duplicate_block_edges): Update.
18277 (duplicate_thread_path): Update.
18278 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18279 (struct switch_conv_info): Update.
18280 * tree-tailcall.c (decrease_profile): Update.
18281 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18282 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18283 * value-prof.c (check_counter): Update.
18284 (gimple_divmod_fixed_value): Update.
18285 (gimple_mod_pow2): Update.
18286 (gimple_mod_subtract): Update.
18287 (gimple_ic_transform): Update.
18288 (gimple_stringop_fixed_value): Update.
18289 * value-prof.h (gimple_ic): Update.
18291 2017-06-02 Carl Love <cel@us.ibm.com>
18293 * config/rs6000/rs6000-c: Add support for built-in functions
18294 vector double vec_doublee (vector signed int);
18295 vector double vec_doublee (vector unsigned int);
18296 vector double vec_doublee (vector float);
18297 vector double vec_doubleh (vector signed int);
18298 vector double vec_doubleh (vector unsigned int);
18299 vector double vec_doubleh (vector float);
18300 vector double vec_doublel (vector signed int);
18301 vector double vec_doublel (vector unsigned int);
18302 vector double vec_doublel (vector float);
18303 vector double vec_doubleo (vector signed int);
18304 vector double vec_doubleo (vector unsigned int);
18305 vector double vec_doubleo (vector float);.
18306 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18307 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18309 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18310 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18311 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18313 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18314 vec_doublel, vec_doubleh.
18315 * doc/extend.texi: Update the built-in documentation file for the
18316 new built-in functions.
18318 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18321 * ipa-inline-analysis.c (free_growth_caches): Set
18322 edge_removal_hook_holder to NULL after removing it.
18324 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18326 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18327 comparision with zero.
18329 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18331 for early expansion of vec_min and vec_max builtins.
18332 (builtin_function_type): Add min/max unsigned variants to those
18333 identified as having unsigned arguments.
18335 2017-06-02 Olivier Hainque <hainque@adacore.com>
18337 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18339 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18341 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18342 Use VALL_F16 iterator rather than VALL.
18344 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18346 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18347 Emit CBNZ inside loop when doing a strong exchange and comparing
18348 against zero. Generate the CC flags after the loop.
18350 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18352 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18353 (dl_section_ref): New.
18354 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18355 On AIX, append an expression to subtract the size of the
18356 section length to dl_section_ref.
18358 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18360 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18361 for early expansion of vector absolute builtins.
18363 2017-06-02 Richard Biener <rguenther@suse.de>
18365 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18366 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18368 2017-06-02 Richard Biener <rguenther@suse.de>
18370 PR tree-optimization/80948
18371 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18372 stmt order as well.
18374 2017-06-02 Richard Biener <rguenther@suse.de>
18376 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18378 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18379 for inductions as relevant.
18381 2017-06-02 Richard Biener <rguenther@suse.de>
18383 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18384 (vectorizable_induction): ... this. Remove dead code.
18386 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18388 * builtins. (expand_builtin_alloca): Remove second parameter and
18389 infer its value from the first parameter instead.
18390 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18392 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18394 PR rtl-optimization/80903
18395 * loop-doloop.c (add_test): Unshare sequence.
18397 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18399 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18401 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18403 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18405 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18406 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18407 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18408 (xlogue_layout::get_stub_rtx): Make static.
18409 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18410 (xlogue_layout::compute_stub_managed_regs): Rename to...
18411 (xlogue_layout::count_stub_managed_regs): ...this.
18412 (xlogue_layout::is_stub_managed_reg): New function.
18413 (xlogue_layout::m_stub_names): Rename to...
18414 (xlogue_layout::s_stub_names): ...this, make static.
18415 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18416 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18417 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18418 xlogue_layout::s_stub_names): Instantiate statics.
18419 (stub_managed_regs): Remove.
18420 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18421 (disable_call_ms2sysv_xlogues): Rename to...
18422 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18423 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18425 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18426 change after reload_completed.
18427 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18429 (ix86_expand_prologue): Likewise.
18430 (ix86_expand_epilogue): Likewise.
18431 (ix86_expand_split_stack_prologue): Likewise.
18432 (ix86_compute_frame_layout): Remove frame parameter ...
18433 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18434 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18436 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18437 (ix86_frame): Move from here ...
18438 * config/i386/i386.h (ix86_frame): ... to here.
18439 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18440 complete ix86_frame data structure instead. Remove some_ld_name.
18442 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18444 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18445 symbols that hold a DECL_VALUE_EXPR.
18447 2017-06-01 Martin Jambor <mjambor@suse.cz>
18449 PR tree-optimization/80898
18450 * tree-sra.c (process_subtree_disqualification): Removed.
18451 (disqualify_candidate): Do not acll
18452 process_subtree_disqualification.
18453 (subtree_mark_written_and_enqueue): New function.
18454 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18455 RHS has been disqualified and re-queue LHS if necessary. Apart
18456 from that, ignore disqualified RHS.
18458 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18460 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18461 address fetch for z10 or later.
18463 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18465 * config/arc/arc.md (tst_movb): Add guard when splitting.
18467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18469 * config/arc/arc.c (arc_can_eliminate): Test against
18470 arc_frame_pointer_needed.
18472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18474 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18475 to prevent store reordering.
18476 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18477 (type): Add block type.
18478 (stack_tie): Define special instruction to be used in
18481 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18483 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18484 constraint. It is not valid for the pattern.
18485 (noncommutative_binary_comparison): Likewise.
18487 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18489 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18492 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18494 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18495 be used by the reg-alloc.
18497 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18499 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18500 reg-alloc when having mul64 or mul32x16 instructions.
18501 (mulsidi3): Likewise.
18502 (umulsidi3): Likewise.
18503 (mulsi32x16): New pattern.
18504 (mulsi64): Likewise.
18505 (mulsidi64): Likewise.
18506 (umulsidi64): Likewise.
18507 (MUL32x16_REG): Define.
18508 (mul64_600): Use MUL32x16_REG.
18509 (mac64_600): Likewise.
18510 (umul64_600): Likewise.
18511 (umac64_600): Likewise.
18513 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18515 * config/arc/arc.md (mulsi3_700): Make it commutative.
18517 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18519 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18521 (*sign_extendsidi2_insn): Likewise for movstosw.
18523 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18525 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18526 the type of the input discriminant value. Convert the
18527 discriminant value of signedness vary.
18529 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18531 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18532 Add to -Wall section.
18534 2017-06-01 Richard Biener <rguenther@suse.de>
18536 PR middle-end/66313
18537 * fold-const.c (fold_plusminus_mult_expr): If the factored
18538 factor may be zero use a wrapping type for the inner operation.
18539 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18540 and handle moved defs.
18541 (process_assignment): Properly guard the unary op case. Return a
18542 tri-state indicating that moving the stmt before the call may allow
18543 to continue. Pass through to_move.
18544 (find_tail_calls): Handle moving unrelated defs before
18547 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18550 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18551 splitter result in the canonical way.
18553 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18555 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18556 also for 32bit target. Update insn attributes.
18557 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18559 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18561 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18562 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18563 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18564 (_mm_maskz_min_ss): New intrinsics.
18566 2017-05-31 Martin Liska <mliska@suse.cz>
18568 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18569 Change comment style to one we normally use.
18570 (vectorizable_reduction): Likewise.
18571 (vectorizable_induction): Likewise.
18572 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18573 (vectorizable_call): Likewise.
18574 (vectorizable_simd_clone_call): Likewise.
18575 (vectorizable_conversion): Likewise.
18576 (vectorizable_assignment): Likewise.
18577 (vectorizable_shift): Likewise.
18578 (vectorizable_operation): Likewise.
18579 (vectorizable_store): Likewise.
18580 (vectorizable_load): Likewise.
18581 * tree-vectorizer.h: Likewise.
18583 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18585 * passes.c (emergency_dump_function): New.
18586 * tree-pass.h (emergency_dump_function): Declare.
18587 * plugin.c (plugins_internal_error_function): Remove.
18588 * plugin.h (plugins_internal_error_function): Remove declaration.
18589 * toplev.c (internal_error_function): New static function. Use it...
18590 (general_init): ...here.
18592 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18594 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18595 (arc_rtx_costs): Add costs for new patterns.
18596 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18597 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18599 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18601 * tree-ssa-strlen.c (get_next_strinfo): New function.
18602 (get_stridx_plus_constant): Use it.
18603 (zero_length_string): Likewise.
18604 (adjust_related_strinfos): Likewise.
18605 (adjust_last_stmt): Likewise.
18607 2017-05-31 Richard Biener <rguenther@suse.de>
18610 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18611 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18613 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18615 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18616 loop_vinfo argument and use of dependence distance vectors.
18617 Check instead whether the two references differ only in their
18618 initial value and assume that they have the same alignment if the
18619 difference is a multiple of the vector alignment.
18620 (vect_analyze_data_refs_alignment): Update call accordingly.
18622 2017-05-31 Martin Liska <mliska@suse.cz>
18625 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18627 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18629 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18630 (create_intersect_range_checks): Move from ...
18631 * tree-data-ref.c (create_intersect_range_checks_index)
18632 (create_intersect_range_checks): ... to here.
18633 (create_runtime_alias_checks): New function factored from ...
18634 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18635 here. Call above function.
18636 * tree-data-ref.h (create_runtime_alias_checks): New function.
18638 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18640 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18641 segment length for dr_b and compute it in wide_int.
18643 2017-05-31 Richard Biener <rguenther@suse.de>
18645 PR tree-optimization/80906
18646 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18647 and pass through iv_map.
18648 (copy_bb_and_scalar_dependences): Adjust.
18649 (translate_pending_phi_nodes): Likewise.
18650 (copy_loop_close_phi_args): Handle code-generating IVs instead
18653 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18655 * diagnostic-color.c (color_dict): Add "type-diff".
18656 (parse_gcc_colors): Update comment.
18657 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18658 -fdiagnostics-show-template-tree and -fno-elide-type.
18659 (GCC_COLORS): Add type-diff to example.
18661 (-fdiagnostics-show-template-tree): New.
18662 (-fno-elide-type): New.
18663 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18664 the pp_format_decoder callback. Call any m_format_postprocessor's
18666 (pretty_printer::pretty_printer): Initialize
18667 m_format_postprocessor.
18668 (pretty_printer::~pretty_printer): Delete any
18669 m_format_postprocessor.
18670 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18671 (class format_postprocessor): New class.
18672 (struct pretty_printer::format_decoder): Document the new parameters.
18673 (struct pretty_printer::m_format_postprocessor): New field.
18674 * tree-diagnostic.c (default_tree_printer): Update for new
18675 bool and const char ** params.
18676 * tree-diagnostic.h (default_tree_printer): Likewise.
18678 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18680 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18681 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18682 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18683 rs6000_gen_cell_microcode code.
18684 (rs6000_final_prescan_insn): Delete.
18685 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18686 "warn-cell-microcode" entries.
18687 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18688 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18689 throughout. Change cc_reg_not_micro_cr0_operand to
18690 cc_reg_not_cr0_operand throughout.
18691 (*extendhi<mode>2_noload): Delete.
18692 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18693 (mwarn-cell-microcode): Delete.
18694 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18695 -mgen-cell-microcode and -mwarn-cell-microcode.
18697 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18700 * config/i386/constraints.md (Yd): New constraint.
18702 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18703 and (?Yd, r) alternatives. Update insn attributes.
18704 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18705 and (?*Yd, r) alternatives. Update insn attributes.
18706 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18708 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18710 * gimplify.c (gimplify_modify_expr): Don't create a
18711 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18714 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18716 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18718 2017-05-30 Richard Biener <rguenther@suse.de>
18720 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18721 and reduc_def fields.
18722 (STMT_VINFO_REDUC_TYPE): New define.
18723 (STMT_VINFO_REDUC_DEF): Likewise.
18724 (vect_force_simple_reduction): Adjust prototype.
18725 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18726 (vect_is_simple_reduction): Remove check_reduction argument.
18727 (vect_force_simple_reduction): Adjust and set
18728 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18729 (vectorizable_reduction): Do not re-do reduction analysis
18730 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18731 * tree-parloops.c (gather_scalar_reductions): Adjust.
18733 2017-05-30 Richard Biener <rguenther@suse.de>
18735 PR middle-end/80901
18736 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18739 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18741 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18742 Introduce unknown_misalignment parameter and remove vf.
18743 (vect_peeling_hash_get_lowest_cost):
18744 Pass unknown_misalignment parameter.
18745 (vect_enhance_data_refs_alignment):
18746 Fix unsupportable data ref treatment.
18748 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18750 * tree-vect-data-refs.c (vect_get_data_access_cost):
18751 Workaround for SLP handling.
18752 (vect_enhance_data_refs_alignment):
18753 Compute costs for doing no peeling at all, compare to the best
18754 peeling costs so far and avoid peeling if cheaper.
18756 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18758 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18759 Return peeling info and set costs to zero for unlimited cost
18761 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18762 with unknown misalignment. Compute and costs for unknown
18763 misalignment, compare them to the costs for known misalignment
18764 and choose the cheapest for peeling.
18766 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18768 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18769 (vect_get_peeling_costs_all_drs): Create function.
18770 (vect_peeling_hash_get_lowest_cost):
18771 Use vect_get_peeling_costs_all_drs.
18772 (vect_peeling_supportable): Create function.
18773 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18775 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18777 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18778 DR_HAS_NEGATIVE_STEP.
18779 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18780 (vect_enhance_data_refs_alignment): Use.
18781 (vect_duplicate_ssa_name_ptr_info): Use.
18782 * tree-vectorizer.h (dr_misalignment): Use.
18783 (known_alignment_for_access_p): Use.
18785 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18788 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18790 (has_section_name): New function.
18792 2017-05-30 Martin Liska <mliska@suse.cz>
18795 * auto-profile.c (get_function_decl_from_block): Fix
18798 2017-05-30 Richard Biener <rguenther@suse.de>
18800 PR middle-end/80876
18801 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18803 2017-05-30 Martin Liska <mliska@suse.cz>
18805 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18806 * dumpfile.h (struct dump_file_info): Remove ctors.
18808 2017-05-30 Martin Liska <mliska@suse.cz>
18810 * predict.def: Fix GNU coding style.
18812 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18814 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18815 Mark 'to' argument with ATTRIBUTE_UNUSED.
18817 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18819 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18820 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18821 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18824 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18826 * doc/install.texi (Options specification): Restore entry of
18827 --enable-sjlj-exceptions.
18829 2017-05-27 Michael Eager <eager@eagercon.com>
18832 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18834 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18836 * config/microblaze/microblaze.h
18837 (FIXED_REGISTERS): Update in macro.
18838 (CALL_USED_REGISTERS): Update in macro.
18840 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18842 * doc/install.texi: Add links to macOS binary distributions.
18844 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18848 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18850 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18852 2017-05-26 Martin Liska <mliska@suse.cz>
18854 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18856 2017-05-26 Martin Liska <mliska@suse.cz>
18858 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18859 always leading ';; '.
18860 (dump_bb_info): Likewise.
18861 (brief_dump_cfg): Likewise.
18862 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18863 * dumpfile.c: Remove usage of TDF_VERBOSE.
18864 * dumpfile.h (enum dump_kind): Likewise.
18865 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18866 * print-tree.c (debug_verbose): Remove.
18867 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18868 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18869 * tree-diagnostic.c (default_tree_printer): Replace
18870 TDF_DIAGNOSTIC with TDF_SLIM.
18872 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18874 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18875 in parameter loop, rather than loop_vinfo.
18876 (create_intersect_range_checks): Ditto.
18877 (vect_create_cond_for_alias_checks): Update call to above functions.
18879 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18881 PR tree-optimization/80815
18882 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18883 for merging runtime alias checks. Handle negative DR_STEPs.
18885 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18887 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18889 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18890 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18891 out code pruning runtime alias checks.
18892 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18893 factored out from above.
18894 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18896 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18898 (prune_runtime_alias_test_list): New decalaration.
18900 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18902 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18903 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18904 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18905 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18906 (operator==, comp_dr_with_seg_len_pair): Ditto.
18907 (vect_prune_runtime_alias_test_list): Ditto.
18909 2017-05-26 Martin Liska <mliska@suse.cz>
18912 * params.def: Bound partial-inlining-entry-probability param.
18914 2017-05-26 Marek Polacek <polacek@redhat.com>
18917 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18920 2017-05-26 Richard Biener <rguenther@suse.de>
18922 PR tree-optimization/80842
18923 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18926 2017-05-26 Richard Biener <rguenther@suse.de>
18928 PR tree-optimization/80844
18929 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18931 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18933 * doc/md.texi (Machine Constraints): Update x86 family
18934 machine constraints section to match 'config/i386/constraints.md'.
18936 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18938 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18940 2017-05-25 Nathan Sidwell <nathan@acm.org>
18942 * doc/invoke.texi (--enable-languages): Update documentation.
18944 2017-05-25 Martin Liska <mliska@suse.cz>
18946 * dumpfile.c: Add TDF_FOLDING.
18947 * dumpfile.h (enum dump_kind): Likewise.
18948 * genmatch.c (dt_simplify::gen_1): Use it.
18950 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18952 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18954 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18956 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18957 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18959 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18961 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18962 * match.pd (X == C): Rewrite it here.
18963 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18964 with_certain_nonzero_bits2): New predicates.
18965 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18967 2017-05-24 Nathan Sidwell <nathan@acm.org>
18969 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18972 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18975 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18977 * config/powerpcspe: New port. Files are copied from the rs6000
18978 port, with "rs6000" in filenames replaced by "powerpcspe".
18980 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18982 PR rtl-optimization/80754
18983 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18985 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18987 * config/sparc/sparc.md (length): Return the correct value for -mflat
18988 sibcalls to match output_sibcall.
18990 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18994 * config/rs6000/rs6000.c (struct machine_function): Add new field
18996 (rs6000_get_separate_components): Init that field, use it.
18997 (rs6000_components_for_bb): Use the field.
18999 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19001 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19003 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19005 PR middle-end/80823
19006 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19008 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19011 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19012 address against address_operand predicate.
19013 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19015 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19017 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19018 parameters passed indirectly.
19020 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19022 * config/i386/i386.md (*movdi_internal): Remove SSE4
19023 alternative 18 (?r, *v). Update insn attributes.
19024 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19025 Update insn attributes.
19026 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19027 Update insn attributes.
19028 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19029 alternative 1 (r, v). Remove isa attribute.
19030 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19031 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19032 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19034 2017-05-23 Tom de Vries <tom@codesourcery.com>
19036 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19039 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19041 * cgraphunit.c (symbol_table::process_new_functions): Update.
19042 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19043 (inline_generate_summary): Rename to ...
19044 (ipa_fn_summary_generate): ... this one.
19045 (inline_read_summary): Rename to ...
19046 (ipa_fn_summary_read): ... this one.
19047 (inline_write_summary): Rename to ...
19048 (ipa_fn_summary_write): ... this one.
19049 (inline_free_summary): Rename to ...
19050 (ipa_free_fn_summary): ... this one.
19051 (pass_data_local_fn_summary, pass_local_fn_summary,
19052 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19053 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19054 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19055 make_pass_ipa_fn_summary): New.
19056 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19057 inline_write_summary, inline_free_summary): Remove.
19058 (ipa_free_fn_summary) : New.
19059 * ipa-inline.c (ipa_inline): Update.
19060 (pass_ipa_inline): Do not generate summaries.
19061 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19063 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19064 and add pass_ipa_fn_summary.
19065 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19067 (make_pass_inline_parameters): Remove.
19069 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19071 * omp-low.c (struct omp_context): Remove "default_kind" member.
19074 * omp-offload.c (execute_oacc_device_lower): Remove the
19075 parallelism dimensions function attributes for unparallelized
19076 OpenACC kernels constructs.
19078 2017-05-23 Martin Liska <mliska@suse.cz>
19080 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19082 (cgraph_edge::make_speculative): Likewise.
19083 (cgraph_edge::resolve_speculation): Likewise.
19084 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19085 (cgraph_node::dump): Likewise.
19086 * cgraph.h: Likewise.
19087 * cgraphunit.c (analyze_functions): Likewise.
19088 (symbol_table::compile): Likewise.
19089 * ipa-cp.c (print_all_lattices): Likewise.
19090 (determine_versionability): Likewise.
19091 (initialize_node_lattices): Likewise.
19092 (ipcp_verify_propagated_values): Likewise.
19093 (estimate_local_effects): Likewise.
19094 (update_profiling_info): Likewise.
19095 (create_specialized_node): Likewise.
19096 (perhaps_add_new_callers): Likewise.
19097 (decide_about_value): Likewise.
19098 (decide_whether_version_node): Likewise.
19099 (identify_dead_nodes): Likewise.
19100 (ipcp_store_bits_results): Likewise.
19101 * ipa-devirt.c (dump_targets): Likewise.
19102 (ipa_devirt): Likewise.
19103 * ipa-icf.c (sem_item::dump): Likewise.
19104 (sem_function::equals): Likewise.
19105 (sem_variable::equals): Likewise.
19106 (sem_item_optimizer::read_section): Likewise.
19107 (sem_item_optimizer::execute): Likewise.
19108 (congruence_class::dump): Likewise.
19109 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19110 (dump_inline_summary): Likewise.
19111 (estimate_node_size_and_time): Likewise.
19112 (inline_analyze_function): Likewise.
19113 * ipa-inline-transform.c (inline_call): Likewise.
19114 * ipa-inline.c (report_inline_failed_reason): Likewise.
19115 (want_early_inline_function_p): Likewise.
19116 (edge_badness): Likewise.
19117 (update_edge_key): Likewise.
19118 (inline_small_functions): Likewise.
19119 * ipa-profile.c (ipa_profile): Likewise.
19120 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19121 (ipa_make_edge_direct_to_target): Likewise.
19122 (remove_described_reference): Likewise.
19123 (ipa_impossible_devirt_target): Likewise.
19124 (propagate_controlled_uses): Likewise.
19125 (ipa_print_node_params): Likewise.
19126 (ipcp_transform_function): Likewise.
19127 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19128 (propagate_pure_const): Likewise.
19129 * ipa-reference.c (generate_summary): Likewise.
19130 (read_write_all_from_decl): Likewise.
19131 (propagate): Likewise.
19132 (ipa_reference_read_optimization_summary): Likewise.
19133 * ipa-utils.c (ipa_merge_profiles): Likewise.
19134 * ipa.c (walk_polymorphic_call_targets): Likewise.
19135 (symbol_table::remove_unreachable_nodes): Likewise.
19136 (ipa_single_use): Likewise.
19137 * passes.c (execute_todo): Likewise.
19138 * predict.c (drop_profile): Likewise.
19139 * symtab.c (symtab_node::get_dump_name): New function.
19140 (symtab_node::dump_name): Likewise.
19141 (symtab_node::dump_asm_name): Likewise.
19142 (symtab_node::dump_references): Likewise.
19143 (symtab_node::dump_referring): Likewise.
19144 (symtab_node::dump_base): Likewise.
19145 (symtab_node::debug_symtab): Likewise.
19146 * tree-sra.c (convert_callers_for_node): Likewise.
19147 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19148 * value-prof.c (init_node_map): Likewise.
19150 2017-05-23 Martin Liska <mliska@suse.cz>
19152 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19153 and symtab_node::debug_symtab to symbol_table::debug.
19154 * cgraphunit.c (analyze_functions): Use the renamed function.
19155 (symbol_table::compile): Likewise.
19156 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19157 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19158 * passes.c (execute_todo): Likewise.
19159 * symtab.c (symbol_table::dump): New function.
19160 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19162 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19164 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19165 that nonconst implies exec.
19167 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19169 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19170 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19171 (inline_edge_summary_vec): Turn into ...
19172 (ipa_call_summaries): ... this one.
19173 (redirect_to_unreachable, edge_set_predicate,
19174 evaluate_properties_for_edge, inline_summary_alloc,
19175 reset_ipa_call_summary, reset_inline_summary,
19176 inline_summary_t::duplicate): Update.
19177 (inline_edge_duplication_hook): Turn to ...
19178 (ipa_call_summary_t::duplicate): ... this one.
19179 (inline_edge_removal_hook): Turn to ...
19180 (ipa_call_summary_t::remove): ... this one.
19181 (dump_inline_edge_summary): Turn to ...
19182 (dump_ipa_call_summary): ... this one.
19183 (estimate_function_body_sizes): Update.
19184 (inline_update_callee_summaries): Update.
19185 (remap_edge_change_prob): Update.
19186 (remap_edge_summaries): Update.
19187 (inline_merge_summary): Update.
19188 (do_estimate_edge_time): Update.
19189 (inline_generate_summary): Update.
19190 (inline_read_section): Update.
19191 (inline_read_summary): Update.
19192 (inline_free_summary): Update.
19193 * ipa-inline.c (can_inline_edge_p): Update.
19194 (compute_inlined_call_time): Update.
19195 (want_inline_small_function_p): Update.
19196 (edge_badness): Update.
19197 (early_inliner): Update.
19198 * ipa-inline.h (inline_edge_summary): Turn to ...
19199 (ipa_call_summary): ... this one.
19200 (ipa_call_summary_t): New class.
19201 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19202 (ipa_call_summaries): New.
19203 (inline_edge_summary): Remove.
19204 (estimate_edge_growth): Update.
19205 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19206 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19207 * ipa-split.c (execute_split_functions): Update.
19208 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19210 2017-05-23 Tom de Vries <tom@codesourcery.com>
19212 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19213 attributes): Document rdrand effective target.
19215 2017-05-23 Tom de Vries <tom@codesourcery.com>
19217 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19218 attributes): Sort alphabetically.
19220 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19222 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19224 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19227 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19228 V2DF/V2DI splat into two separate patterns, one that handles
19229 registers, and the other that only handles memory. Drop support
19230 for splatting from a GPR on ISA 2.07 and then splitting the
19231 splat into direct move and splat.
19232 (vsx_splat_<mode>_reg): Likewise.
19233 (vsx_splat_<mode>_mem): Likewise.
19235 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19237 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19239 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19241 PR middle-end/80809
19242 * omp-low.c (finish_taskreg_remap): New function.
19243 (finish_taskreg_scan): If unit size of ctx->record_type
19244 is non-constant, unshare the size expression and replace
19245 decls in it with possible outer var refs.
19247 PR middle-end/80809
19248 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19249 GOVD_SHARED rather than GOVD_PRIVATE with it.
19250 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19251 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19253 PR middle-end/80853
19254 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19255 as last argument to build_outer_var_ref for pointer bases of array
19256 section reductions.
19258 2017-05-19 Martin Sebor <msebor@redhat.com>
19260 * print-tree.c (print_node): Print DECL_READ_P flag.
19262 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19264 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19265 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19266 * cgraph.c: Likewise.
19267 * cgraphunit.c: Likewise.
19268 * gengtype.c: Likewise.
19269 * ipa-cp.c: Likewise.
19270 * ipa-devirt.c: Likewise.
19271 * ipa-icf.c: Likewise.
19272 * ipa-predicate.c: Likewise.
19273 * ipa-profile.c: Likewise.
19274 * ipa-prop.c: Likewise.
19275 * ipa-split.c: Likewise.
19277 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19278 edge_predicate_pool, dump_inline_hints,
19279 inline_summary::account_size_time, redirect_to_unreachable,
19280 edge_set_predicate, set_hint_predicate,
19281 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19282 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19283 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19284 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19285 ipa_call_summary_t::remove, initialize_growth_caches,
19286 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19287 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19288 mark_modified, unmodified_parm_1, unmodified_parm,
19289 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19290 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19291 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19292 will_be_nonconstant_predicate, record_modified_bb_info,
19293 get_minimal_bb, record_modified, param_change_prob,
19294 phi_result_unknown_predicate, predicate_for_phi_result,
19295 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19296 estimate_function_body_sizes, compute_inline_parameters,
19297 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19298 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19299 inline_update_callee_summaries, remap_edge_change_prob,
19300 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19301 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19302 inline_analyze_function, inline_summary_t::insert,
19303 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19304 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19305 inline_free_summary): Move to ipa-fnsummary.h
19306 (predicate_t): Remove.
19307 * ipa-fnsummary.c: New file.
19308 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19309 (enum inline_hints_vals, inline_hints, agg_position_info,
19310 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19311 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19312 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19313 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19314 inline_read_summary, inline_write_summary, inline_free_summary,
19315 inline_analyze_function, initialize_inline_failed,
19316 inline_merge_summary, inline_update_overall_summary,
19317 compute_inline_parameters): Move to ipa-fnsummary.h
19318 * ipa-fnsummary.h: New file.
19319 * ipa-inline-transform.h: Include ipa-inline.h.
19320 * ipa-inline.c: LIkewise.
19322 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19324 * ipa-inline.c (edge_badness): Use inlined_time instead of
19325 inline_summaries->get.
19327 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19329 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19331 2017-05-22 Nathan Sidwell <nathan@acm.org>
19333 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19334 (fdump-lang): Document 'raw' option.
19335 * dumpfile.h (TDI_tu): Delete.
19336 * dumpfile.c (dump_files): Remove translation-unit.
19337 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19339 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19341 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19342 command option from $(AWK) call.
19343 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19344 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19345 [FORMAT]: Remove handling of variable.
19346 * config/avr/t-multilib: Regenerate.
19348 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19350 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19352 (dump_inline_summary): Do not print self_time.
19353 (estimate_function_body_sizes): Do not set self_time.
19354 (compute_inline_parameters): Likewise.
19355 (inline_read_section, inline_write_summary): Do not stream self_time.
19356 * ipa-inline.h (inline_summary): Drop self_time.
19358 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19360 * ipa-inline-analysis.c (account_size_time): Rename to ...
19361 (inline_summary::account_size_time): ... this one.
19362 (reset_ipa_call_summary): Turn to ...
19363 (ipa_call_summary::reset): ... this one.
19364 (reset_inline_summary): Turn to ...
19365 (inline_summary::reset): ... this one.
19366 (inline_summary_t::remove): Update.
19367 (inline_summary_t::duplicate): Update.
19368 (ipa_call_summary_t::remove): Update.
19369 (dump_inline_summary): Update.
19370 (estimate_function_body_sizes): Update.
19371 (compute_inline_parameters): Update.
19372 (estimate_node_size_and_time): Update.
19373 (inline_merge_summary): Update.
19374 (inline_update_overall_summary): Update.
19375 (inline_read_section): Update.
19376 (inline_write_summary): Update.
19377 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19378 add account_size_time and reset member functions.
19379 (ipa_call_summary): Add reset function.
19380 * ipa-predicate.h (predicate::operator &): Constify.
19382 2017-05-22 Richard Biener <rguenther@suse.de>
19384 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19386 2017-05-19 Jason Merrill <jason@redhat.com>
19388 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19390 2017-05-19 Marek Polacek <polacek@redhat.com>
19393 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19394 TYPE_OVERFLOW_WRAPS checks.
19396 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19398 * tree-core.h (enum omp_clause_default_kind): Add
19399 "OMP_CLAUSE_DEFAULT_PRESENT".
19400 * tree-pretty-print.c (dump_omp_clause): Handle it.
19401 * gimplify.c (enum gimplify_omp_var_data): Add
19402 "GOVD_MAP_FORCE_PRESENT".
19403 (gimplify_adjust_omp_clauses_1): Map it to
19404 "GOMP_MAP_FORCE_PRESENT".
19405 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19407 * gimplify.c (oacc_default_clause): Clarify.
19409 2017-05-19 Nathan Sidwell <nathan@acm.org>
19411 LANG_HOOK_REGISTER_DUMPS
19412 * toplev.c (general_init): Call register dump lang hook.
19413 * doc/invoke.texi: Document -fdump-lang option family.
19414 * dumpfile.c (dump_files): Remove class dump here.
19415 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19416 * dumpfile.h (tree_dump_index): Remove TDI_class.
19417 * langhooks-def.h (lhd_register_dumps): Declare.
19418 (LANG_HOOKS_REGISTER_DUMPS): Define.
19419 (LANG_HOOKS_INITIALIZER): Add it.
19420 * langhooks.c (lhd_register_dumps): Define.
19421 * langhooks.h (struct lang_hooks): Add register_dumps.
19423 2017-05-19 Nathan Sidwell <nathan@acm.org>
19425 * context.h (context::set_passes): New.
19426 * context.c (context::context): Do not create pass manager.
19427 * toplev.c (general_init): Create pass manager here.
19429 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19431 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19432 use this splitter if two add or or instructions would also work for
19433 the constant we want to generate.
19435 2017-05-19 Richard Biener <rguenther@suse.de>
19438 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19439 predicate evaluation.
19441 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19443 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19445 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19448 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19450 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19451 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19452 (inline_edge_summary_vec): Turn into ...
19453 (ipa_call_summaries): ... this one.
19454 (redirect_to_unreachable, edge_set_predicate,
19455 evaluate_properties_for_edge, inline_summary_alloc,
19456 reset_ipa_call_summary, reset_inline_summary,
19457 inline_summary_t::duplicate): Update.
19458 (inline_edge_duplication_hook): Turn to ...
19459 (ipa_call_summary_t::duplicate): ... this one.
19460 (inline_edge_removal_hook): Turn to ...
19461 (ipa_call_summary_t::remove): ... this one.
19462 (dump_inline_edge_summary): Turn to ...
19463 (dump_ipa_call_summary): ... this one.
19464 (estimate_function_body_sizes): Update.
19465 (inline_update_callee_summaries): Update.
19466 (remap_edge_change_prob): Update.
19467 (remap_edge_summaries): Update.
19468 (inline_merge_summary): Update.
19469 (do_estimate_edge_time): Update.
19470 (inline_generate_summary): Update.
19471 (inline_read_section): Update.
19472 (inline_read_summary): Update.
19473 (inline_free_summary): Update.
19474 * ipa-inline.c (can_inline_edge_p): Update.
19475 (compute_inlined_call_time): Update.
19476 (want_inline_small_function_p): Update.
19477 (edge_badness): Update.
19478 (early_inliner): Update.
19479 * ipa-inline.h (inline_edge_summary): Turn to ...
19480 (ipa_call_summary): ... this one.
19481 (ipa_call_summary_t): New class.
19482 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19483 (ipa_call_summaries): New.
19484 (inline_edge_summary): Remove.
19485 (estimate_edge_growth): Update.
19486 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19487 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19488 * ipa-split.c (execute_split_functions): Update.
19489 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19491 2017-05-19 Richard Biener <rguenther@suse.de>
19493 PR middle-end/80764
19494 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19496 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19498 * config/rs6000/rs6000.c (struct machine_function): Add field
19499 fpr_is_wrapped_separately.
19500 (rs6000_get_separate_components): Use 64 components. Handle the
19501 new FPR components.
19502 (rs6000_components_for_bb): Handle the FPR components.
19503 (rs6000_emit_prologue_components): Handle the FPR components.
19504 (rs6000_emit_epilogue_components): Handle the FPR components.
19505 (rs6000_set_handled_components): Handle the FPR components.
19506 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19507 that are already separately shrink-wrapped.
19508 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19509 that are already separately shrink-wrapped.
19511 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19514 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19517 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19518 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19519 cases where the register allocator uses a move and an offsettable
19520 memory operation to/from a FPR register on ISA 2.06/2.07.
19521 (define_peephole2 for Altivec d-form store): Likewise.
19523 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19526 * config/i386/mmx.md (*mov<mode>_internal): Enable
19527 alternatives 11, 12, 13 and 14 also for 32bit targets.
19528 Remove alternatives 15, 16, 17 and 18.
19529 * config/i386/sse.md (vec_concatv2di): Change
19530 alternative (!x, *y) to (x, ?!*Yn).
19532 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19534 * dumpfile.h (enum dump_kind): Remove stray comma.
19536 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19538 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19539 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19540 predicate::num_conditions
19541 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19542 (CHANGED): turn into predicate::changed.
19543 (agg_position_info): Move to ipa-predicate.h
19544 (add_condition, predicate::add_clause, predicate::operator &=,
19545 predicate::or_with, predicate::evaluate, predicate::probability,
19546 dump_condition, dump_clause, predicate::dump,
19547 predicate::remap_after_duplication, predicate::remap_after_inlining,
19548 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19549 (evaluate_conditions_for_known_args): Update.
19550 (set_cond_stmt_execution_predicate): Update.
19551 * ipa-inline.h: Include ipa-predicate.h
19552 (condition, inline_param_summary, conditions, agg_position_info,
19553 predicate): Move to ipa-predicate.h
19554 * ipa-predicate.c: New file.
19555 * ipa-predicate.h: New file.
19557 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19559 * final.c (leaf_function_p): Check we are not in a sequence.
19561 2017-05-18 Martin Liska <mliska@suse.cz>
19563 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19564 * dumpfile.c (dump_register): Use new enum dump_kind.
19565 (get_dump_file_name): Likewise.
19566 (dump_enable_all): Likewise.
19567 (dump_switch_p_1): Likewise.
19568 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19569 * dumpfile.h (enum dump_kind): New enum type.
19570 (struct dump_file_info): Create constructor and
19571 format fields and comments.
19572 * passes.c (pass_manager::register_one_dump_file):
19574 * statistics.c (statistics_early_init): Likewise.
19575 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19576 TDF_TREE with TDF_SLIM.
19577 (gather_memory_references_ref): Likewise.
19579 2017-05-18 Martin Liska <mliska@suse.cz>
19581 * vec.h (struct vnull): Use it.
19583 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19585 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19586 (true_predicate, false_predicate, true_predicate_p,
19587 false_predicate_p): Remove.
19588 (single_cond_predicate, not_inlined_predicate): Turn to member function
19590 (add_condition): Update.
19591 (add_clause): Turn to...
19592 (predicate::add_clause): ... this one; update; allow passing NULL
19594 (and_predicates): Turn to ...
19595 (predicate::operator &=): ... this one.
19596 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19597 (or_predicates): Turn to ...
19598 (predicate::or_with): ... this one.
19599 (evaluate_predicate): Turn to ...
19600 (predicate::evaluate): ... this one.
19601 (predicate_probability): Turn to ...
19602 (predicate::probability): ... this one.
19603 (dump_condition): Update.
19604 (dump_predicate): Turn to ...
19605 (predicate::dump): ... this one.
19606 (account_size_time): Update.
19607 (edge_set_predicate): Update.
19608 (set_hint_predicate): UPdate.
19609 (evaluate_conditions_for_known_args): Update.
19610 (evaluate_properties_for_edge): Update.
19611 (remap_predicate_after_duplication): Turn to...
19612 (predicate::remap_after_duplication): ... this one.
19613 (remap_hint_predicate_after_duplication): Update.
19614 (inline_summary_t::duplicate): UPdate.
19615 (dump_inline_edge_summary): Update.
19616 (dump_inline_summary): Update.
19617 (set_cond_stmt_execution_predicate): Update.
19618 (set_switch_stmt_execution_predicate): Update.
19619 (compute_bb_predicates): Update.
19620 (will_be_nonconstant_expr_predicate): Update.
19621 (will_be_nonconstant_predicate): Update.
19622 (phi_result_unknown_predicate): Update.
19623 (predicate_for_phi_result): Update.
19624 (array_index_predicate): Update.
19625 (estimate_function_body_sizes): Update.
19626 (estimate_node_size_and_time): Update.
19627 (estimate_ipcp_clone_size_and_time): Update.
19628 (remap_predicate): Rename to ...
19629 (predicate::remap_after_inlining): ... this one.
19630 (remap_hint_predicate): Update.
19631 (inline_merge_summary): Update.
19632 (inline_update_overall_summary): Update.
19633 (estimate_size_after_inlining): Update.
19634 (read_predicate): Rename to ...
19635 (predicate::stream_in): ... this one.
19636 (read_inline_edge_summary): Update.
19637 (write_predicate): Rename to ...
19638 (predicate::stream_out): ... this one.
19639 (write_inline_edge_summary): Update.
19640 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19641 (clause_t): Turn to uint32_t
19642 (predicate): Turn to class; implement constructor and operators
19644 (size_time_entry): Update.
19645 (inline_summary): Update.
19646 (inline_edge_summary): Update.
19648 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19650 * fold-const.c (fold_binary_loc): Move transformation...
19651 * match.pd (C - X CMP X): ... here.
19653 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19655 * config/sparc/sparc.c (sparc_option_override): Set function
19656 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19657 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19659 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19661 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19663 2017-05-18 Marek Polacek <polacek@redhat.com>
19666 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19667 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19668 gimple_assign_load_p.
19670 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19672 PR middle-end/80692
19673 * real.c (do_compare): Give decimal_do_compare preference over
19674 comparing just the signs.
19676 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19678 * doc/md.texi (Canonicalization of Instructions): Describe the
19679 canonical form of instructions that inherently set a condition
19682 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19684 PR middle-end/80775
19685 * tree-cfg.c: Move deletion of unreachable case statements to after
19686 the merging of consecutive case labels.
19688 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19690 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19691 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19692 restoring of callee-saved registers.
19694 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19696 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19697 * config/visium/visium.c (single_set_and_flags): Likewise.
19698 * config/visium/visium.md (Substitutions): Likewise.
19700 2017-05-17 Martin Liska <mliska@suse.cz>
19702 * cfg.c: Introduce dump_flags_t type and
19703 use it instead of int type.
19705 * cfghooks.c: Likewise.
19706 * cfghooks.h (struct cfg_hooks): Likewise.
19707 * cfgrtl.c: Likewise.
19708 * cfgrtl.h: Likewise.
19709 * cgraph.c (cgraph_node::get_body): Likewise.
19710 * coretypes.h: Likewise.
19711 * domwalk.c: Likewise.
19712 * domwalk.h: Likewise.
19713 * dumpfile.c (struct dump_option_value_info): Likewise.
19714 (dump_enable_all): Likewise.
19715 (dump_switch_p_1): Likewise.
19716 (opt_info_switch_p): Likewise.
19717 * dumpfile.h (enum tree_dump_index): Likewise.
19718 (struct dump_file_info): Likewise.
19719 * genemit.c: Likewise.
19720 * generic-match-head.c: Likewise.
19721 * gengtype.c (open_base_files): Likewise.
19722 * gimple-pretty-print.c: Likewise.
19723 * gimple-pretty-print.h: Likewise.
19724 * graph.c (print_graph_cfg): Likewise.
19725 * graphite-scop-detection.c (dot_all_sese): Likewise.
19726 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19727 * loop-unroll.c (report_unroll): Likewise.
19728 * passes.c (pass_manager::register_one_dump_file): Likewise.
19729 * print-tree.c: Likewise.
19730 * statistics.c: Likewise.
19731 * tree-cfg.c: Likewise.
19732 * tree-cfg.h: Likewise.
19733 * tree-dfa.c: Likewise.
19734 * tree-dfa.h: Likewise.
19735 * tree-dump.c (dump_function): Likewise.
19736 * tree-dump.h (struct dump_info): Likewise.
19737 * tree-pretty-print.c: Likewise.
19738 * tree-pretty-print.h: Likewise.
19739 * tree-ssa-live.c: Likewise.
19740 * tree-ssa-live.h: Likewise.
19741 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19742 * tree-vect-loop.c: Likewise.
19743 * tree-vect-slp.c: Likewise.
19745 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19746 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19748 PR tree-optimization/80457
19749 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19750 of all arguments to a statement as scalar_to_vec operations.
19751 (vectorizable_call): Adjust call to vect_model_simple_cost for
19753 (vectorizable_conversion): Likewise.
19754 (vectorizable_assignment): Likewise.
19755 (vectorizable_shift): Likewise.
19756 (vectorizable_operation): Likewise.
19757 (vectorizable_comparison): Likewise.
19758 (vect_is_simple_cond): Record the def types for operands.
19759 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19760 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19761 for statement argument count.
19763 2017-05-16 Carl Love <cel@us.ibm.com>
19765 * config/rs6000/rs6000-c: Add support for built-in functions
19766 vector unsigned long long vec_bperm (vector unsigned long long,
19767 vector unsigned char)
19768 vector signed long long vec_mule (vector signed int,
19770 vector unsigned long long vec_mule (vector unsigned int,
19771 vector unsigned int)
19772 vector signed long long vec_mulo (vector signed int,
19774 vector unsigned long long vec_mulo (vector unsigned int,
19775 vector unsigned int)
19776 vector signed char vec_sldw (vector signed char,
19777 vector signed char,
19779 vector unsigned char vec_sldw (vector unsigned char,
19780 vector unsigned char,
19782 vector signed short vec_sldw (vector signed short,
19783 vector signed short,
19785 vector unsigned short vec_sldw (vector unsigned short,
19786 vector unsigned short,
19788 vector signed int vec_sldw (vector signed int,
19791 vector unsigned int vec_sldw (vector unsigned int,
19792 vector unsigned int,
19794 vector signed long long vec_sldw (vector signed long long,
19795 vector signed long long,
19797 vector unsigned long long vec_sldw (vector unsigned long long,
19798 vector unsigned long long,
19800 * config/rs6000/rs6000-c: Add support for built-in functions
19801 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19802 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19803 * doc/extend.texi: Update the built-in documentation for the
19804 new built-in functions.
19806 2017-05-16 Marek Polacek <polacek@redhat.com>
19810 * tree.c (save_expr): Don't fold the expression.
19812 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19814 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19815 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19816 and (?*y,m). Update insn attributes.
19818 2017-05-16 Martin Liska <mliska@suse.cz>
19820 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19821 flags argument of print_gimple_stmt, print_gimple_expr,
19822 print_generic_stmt and print_generic_expr.
19823 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19824 * coretypes.h: Likewise.
19825 * except.c (dump_eh_tree): Likewise.
19826 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19827 * gimple-pretty-print.h: Likewise.
19828 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19829 (backprop::push_to_worklist): Likewise.
19830 (backprop::pop_from_worklist): Likewise.
19831 (backprop::process_use): Likewise.
19832 (backprop::intersect_uses): Likewise.
19833 (note_replacement): Likewise.
19834 * gimple-ssa-store-merging.c
19835 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19836 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19837 (pass_store_merging::execute): Likewise.
19838 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19839 (ssa_base_cand_dump_callback): Likewise.
19840 (dump_incr_vec): Likewise.
19841 (replace_refs): Likewise.
19842 (replace_mult_candidate): Likewise.
19843 (create_add_on_incoming_edge): Likewise.
19844 (create_phi_basis): Likewise.
19845 (insert_initializers): Likewise.
19846 (all_phi_incrs_profitable): Likewise.
19847 (introduce_cast_before_cand): Likewise.
19848 (replace_one_candidate): Likewise.
19849 * gimplify.c (gimplify_expr): Likewise.
19850 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19851 (set_rename): Likewise.
19852 (rename_uses): Likewise.
19853 (copy_loop_phi_nodes): Likewise.
19854 (add_close_phis_to_merge_points): Likewise.
19855 (copy_loop_close_phi_args): Likewise.
19856 (copy_cond_phi_args): Likewise.
19857 (graphite_copy_stmts_from_block): Likewise.
19858 (translate_pending_phi_nodes): Likewise.
19859 * graphite-poly.c (print_pdr): Likewise.
19860 (dump_gbb_cases): Likewise.
19861 (dump_gbb_conditions): Likewise.
19862 (print_scop_params): Likewise.
19863 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19864 (build_cross_bb_scalars_use): Likewise.
19865 (gather_bbs::before_dom_children): Likewise.
19866 * hsa-dump.c (dump_hsa_immed): Likewise.
19867 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19868 (get_replacement_map): Likewise.
19869 * ipa-inline-analysis.c (dump_condition): Likewise.
19870 (estimate_function_body_sizes): Likewise.
19871 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19872 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19873 * ipa-prop.c (ipa_dump_param): Likewise.
19874 (ipa_print_node_jump_functions_for_edge): Likewise.
19875 (ipa_modify_call_arguments): Likewise.
19876 (ipa_modify_expr): Likewise.
19877 (ipa_dump_param_adjustments): Likewise.
19878 (ipa_dump_agg_replacement_values): Likewise.
19879 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19880 * ipa-pure-const.c (check_stmt): Likewise.
19881 (pass_nothrow::execute): Likewise.
19882 * ipa-split.c (execute_split_functions): Likewise.
19883 * omp-offload.c (dump_oacc_loop_part): Likewise.
19884 (dump_oacc_loop): Likewise.
19885 * trans-mem.c (tm_log_emit): Likewise.
19886 (tm_memopt_accumulate_memops): Likewise.
19887 (dump_tm_memopt_set): Likewise.
19888 (dump_tm_memopt_transform): Likewise.
19889 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19890 (print_loop): Likewise.
19891 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19892 (chkp_gather_checks_info): Likewise.
19893 (chkp_get_check_result): Likewise.
19894 (chkp_remove_check_if_pass): Likewise.
19895 (chkp_use_outer_bounds_if_possible): Likewise.
19896 (chkp_reduce_bounds_lifetime): Likewise.
19897 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19898 (chkp_mark_completed_bounds): Likewise.
19899 (chkp_register_incomplete_bounds): Likewise.
19900 (chkp_mark_invalid_bounds): Likewise.
19901 (chkp_maybe_copy_and_register_bounds): Likewise.
19902 (chkp_build_returned_bound): Likewise.
19903 (chkp_get_bound_for_parm): Likewise.
19904 (chkp_build_bndldx): Likewise.
19905 (chkp_get_bounds_by_definition): Likewise.
19906 (chkp_generate_extern_var_bounds): Likewise.
19907 (chkp_get_bounds_for_decl_addr): Likewise.
19908 * tree-chrec.c (chrec_apply): Likewise.
19909 * tree-data-ref.c (dump_data_reference): Likewise.
19910 (dump_subscript): Likewise.
19911 (dump_data_dependence_relation): Likewise.
19912 (analyze_overlapping_iterations): Likewise.
19913 * tree-inline.c (expand_call_inline): Likewise.
19914 (tree_function_versioning): Likewise.
19915 * tree-into-ssa.c (dump_defs_stack): Likewise.
19916 (dump_currdefs): Likewise.
19917 (dump_names_replaced_by): Likewise.
19918 (dump_update_ssa): Likewise.
19919 (update_ssa): Likewise.
19920 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19921 * tree-parloops.c (build_new_reduction): Likewise.
19922 (try_create_reduction_list): Likewise.
19923 (ref_conflicts_with_region): Likewise.
19924 (oacc_entry_exit_ok_1): Likewise.
19925 (oacc_entry_exit_single_gang): Likewise.
19926 * tree-pretty-print.h: Likewise.
19927 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19928 (get_scalar_evolution): Likewise.
19929 (add_to_evolution): Likewise.
19930 (get_loop_exit_condition): Likewise.
19931 (analyze_evolution_in_loop): Likewise.
19932 (analyze_initial_condition): Likewise.
19933 (analyze_scalar_evolution): Likewise.
19934 (instantiate_scev): Likewise.
19935 (number_of_latch_executions): Likewise.
19936 (gather_chrec_stats): Likewise.
19937 (final_value_replacement_loop): Likewise.
19938 (scev_const_prop): Likewise.
19939 * tree-sra.c (dump_access): Likewise.
19940 (disqualify_candidate): Likewise.
19941 (create_access): Likewise.
19942 (reject): Likewise.
19943 (maybe_add_sra_candidate): Likewise.
19944 (create_access_replacement): Likewise.
19945 (analyze_access_subtree): Likewise.
19946 (analyze_all_variable_accesses): Likewise.
19947 (sra_modify_assign): Likewise.
19948 (initialize_constant_pool_replacements): Likewise.
19949 (find_param_candidates): Likewise.
19950 (decide_one_param_reduction): Likewise.
19951 (replace_removed_params_ssa_names): Likewise.
19952 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19953 * tree-ssa-copy.c (dump_copy_of): Likewise.
19954 (copy_prop_visit_cond_stmt): Likewise.
19955 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19956 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19957 (record_equivalences_from_stmt): Likewise.
19958 * tree-ssa-dse.c (compute_trims): Likewise.
19959 (delete_dead_call): Likewise.
19960 (delete_dead_assignment): Likewise.
19961 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19962 (forward_propagate_into_cond): Likewise.
19963 (pass_forwprop::execute): Likewise.
19964 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19965 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19967 (move_computations_worker): Likewise.
19968 (execute_sm): Likewise.
19969 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19970 (remove_exits_and_undefined_stmts): Likewise.
19971 (remove_redundant_iv_tests): Likewise.
19972 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19973 (adjust_iv_update_pos): Likewise.
19974 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19975 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19976 (value_replacement): Likewise.
19977 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19978 * tree-ssa-pre.c (print_pre_expr): Likewise.
19979 (get_representative_for): Likewise.
19980 (create_expression_by_pieces): Likewise.
19981 (insert_into_preds_of_block): Likewise.
19982 (eliminate_insert): Likewise.
19983 (eliminate_dom_walker::before_dom_children): Likewise.
19984 (eliminate): Likewise.
19985 (remove_dead_inserted_code): Likewise.
19986 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19987 * tree-ssa-reassoc.c (get_rank): Likewise.
19988 (eliminate_duplicate_pair): Likewise.
19989 (eliminate_plus_minus_pair): Likewise.
19990 (eliminate_not_pairs): Likewise.
19991 (undistribute_ops_list): Likewise.
19992 (eliminate_redundant_comparison): Likewise.
19993 (update_range_test): Likewise.
19994 (optimize_range_tests_var_bound): Likewise.
19995 (optimize_vec_cond_expr): Likewise.
19996 (rewrite_expr_tree): Likewise.
19997 (rewrite_expr_tree_parallel): Likewise.
19998 (linearize_expr): Likewise.
19999 (break_up_subtract): Likewise.
20000 (linearize_expr_tree): Likewise.
20001 (attempt_builtin_powi): Likewise.
20002 (attempt_builtin_copysign): Likewise.
20003 (transform_stmt_to_copy): Likewise.
20004 (transform_stmt_to_multiply): Likewise.
20005 (dump_ops_vector): Likewise.
20006 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20007 (print_scc): Likewise.
20008 (set_ssa_val_to): Likewise.
20009 (visit_reference_op_store): Likewise.
20010 (visit_use): Likewise.
20011 (sccvn_dom_walker::before_dom_children): Likewise.
20012 (run_scc_vn): Likewise.
20013 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20015 (expr_hash_elt::print): Likewise.
20016 (const_and_copies::pop_to_marker): Likewise.
20017 (const_and_copies::record_const_or_copy_raw): Likewise.
20018 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20019 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20020 (dump_predicates): Likewise.
20021 (find_uninit_use): Likewise.
20022 (warn_uninitialized_phi): Likewise.
20023 (pass_late_warn_uninitialized::execute): Likewise.
20024 * tree-ssa.c (verify_vssa): Likewise.
20025 (verify_ssa): Likewise.
20026 (maybe_optimize_var): Likewise.
20027 * tree-vrp.c (dump_value_range): Likewise.
20028 (dump_all_value_ranges): Likewise.
20029 (dump_asserts_for): Likewise.
20030 (register_edge_assert_for_2): Likewise.
20031 (vrp_visit_cond_stmt): Likewise.
20032 (vrp_visit_switch_stmt): Likewise.
20033 (vrp_visit_stmt): Likewise.
20034 (vrp_visit_phi_node): Likewise.
20035 (simplify_cond_using_ranges_1): Likewise.
20036 (fold_predicate_in): Likewise.
20037 (evrp_dom_walker::before_dom_children): Likewise.
20038 (evrp_dom_walker::push_value_range): Likewise.
20039 (evrp_dom_walker::pop_value_range): Likewise.
20040 (execute_early_vrp): Likewise.
20042 2017-05-16 Richard Biener <rguenther@suse.de>
20044 * dwarf2out.c (loc_list_from_tree_1): Do not create
20045 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20047 2017-05-16 Richard Biener <rguenther@suse.de>
20049 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20051 (note_variable_value_in_expr): If we resolved the decl ref
20052 do not push to the stack.
20054 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20056 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20057 operations in fast-math mode.
20058 (vaddq_f16): Likewise.
20059 (vmul_f16): Likewise.
20060 (vmulq_f16): Likewise.
20061 (vsub_f16): Likewise.
20062 (vsubq_f16): Likewise.
20063 * config/arm/neon.md (add<mode>3): New.
20065 (fma:<VH:mode>3): New. Also remove outdated comment.
20068 2017-05-16 Martin Liska <mliska@suse.cz>
20072 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20073 (odr_types_equivalent_p): Likewise.
20075 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20077 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20079 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20082 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20083 non-interunit SSE move alternatives with '?'.
20084 (zero-extendsidi peephole2): New peephole to skip intermediate
20085 general register in SSE zero-extend sequence.
20087 2017-05-15 Jeff Law <law@redhat.com>
20089 * reorg.c (relax_delay_slots): Create a new variable to hold
20090 the temporary target rather than clobbering TARGET_LABEL.
20092 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20093 missing argument to extract_bit_field call.
20094 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20096 2017-05-15 Martin Liska <mliska@suse.cz>
20099 * gcc.c (process_command): Do not allow empty argument of -o option.
20101 2017-05-15 Renlin Li <renlin.li@arm.com>
20103 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20104 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20105 * config/aarch64/constraints.md (Usf): Add long call check.
20106 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20107 (call_value): Likewise.
20108 (sibcall): Likewise.
20109 (sibcall_value): Likewise.
20111 (call_value_insn): New.
20112 (sibcall_insn): Update rtx pattern.
20113 (sibcall_value_insn): Likewise.
20114 (call_internal): Remove.
20115 (call_value_internal): Likewise.
20116 (sibcall_internal): Likewise.
20117 (sibcall_value_internal): Likewise.
20118 (call_reg): Likewise.
20119 (call_symbol): Likewise.
20120 (call_value_reg): Likewise.
20121 (call_value_symbol): Likewise.
20123 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20126 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20128 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20130 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20131 compatible with CCGOCmode and with CCZmode.
20133 2017-05-14 Martin Sebor <msebor@redhat.com>
20135 PR middle-end/77671
20136 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20137 (gimple_fold_builtin_snprintf): Same.
20138 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20139 (gimple_fold_builtin_snprintf): Same.
20140 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20141 of character types.
20142 (is_call_safe): New function.
20143 (try_substitute_return_value): Call it.
20144 (try_simplify_call): New function.
20145 (pass_sprintf_length::handle_gimple_call): Call it.
20147 2017-05-14 Martin Sebor <msebor@redhat.com>
20149 PR middle-end/80669
20150 * builtins.c (expand_builtin_stpncpy): Simplify.
20152 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20154 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20155 * config/i386/i386.h
20156 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20157 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20158 (struct machine_function): Add new members call_ms2sysv,
20159 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20160 (struct machine_frame_state): New fields sp_realigned and
20161 sp_realigned_offset.
20162 * config/i386/i386.c
20163 (enum xlogue_stub): New enum.
20164 (enum xlogue_stub_sets): New enum.
20165 (class xlogue_layout): New class.
20166 (struct ix86_frame): New fields stack_realign_allocate_offset,
20167 stack_realign_offset and outlined_save_offset. Modify comments to
20168 detail stack layout when using out-of-line stubs.
20169 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20170 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20171 -mcall-ms2sysv-xlogues.
20172 (stub_managed_regs): New static variable.
20173 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20174 registers managed by out-of-line stub.
20175 (disable_call_ms2sysv_xlogues): New function.
20176 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20177 m->call_ms2sysv when appropriate and compute frame layout for
20179 (sp_valid_at, fp_valid_at): New inline functions.
20180 (choose_basereg): New function.
20181 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20183 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20184 Use align parameter of choose_baseaddr to generated aligned SSE movs
20186 (pro_epilogue_adjust_stack): Modify to track
20187 machine_frame_state::sp_realigned.
20188 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20189 (ix86_nsaved_sseregs): Likewise.
20190 (ix86_emit_save_regs): Likewise.
20191 (ix86_emit_save_regs_using_mov): Likewise.
20192 (ix86_emit_save_sse_regs_using_mov): Likewise.
20193 (get_scratch_register_on_entry): Likewise.
20194 (gen_frame_set): New function.
20195 (gen_frame_load): Likewise.
20196 (gen_frame_store): Likewise.
20197 (emit_outlined_ms2sysv_save): Likewise.
20198 (emit_outlined_ms2sysv_restore): Likewise.
20199 (ix86_expand_prologue): Modify stack re-alignment code and call
20200 emit_outlined_ms2sysv_save when appropriate.
20201 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20202 parameter rtx_insn *insn, which allows the function to be used to only
20203 generate the notes.
20204 (ix86_expand_epilogue): Modify validity checks of frame and stack
20205 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20206 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20207 * config/i386/predicates.md
20208 (save_multiple): New predicate.
20209 (restore_multiple): Likewise.
20210 * config/i386/sse.md
20211 (save_multiple<mode>): New pattern.
20212 (save_multiple_realign<mode>): Likewise.
20213 (restore_multiple<mode>): Likewise.
20214 (restore_multiple_and_return<mode>): Likewise.
20215 (restore_multiple_leave_return<mode>): Likewise.
20216 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20218 2017-05-14 Julia Koval <julia.koval@intel.com>
20220 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20221 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20222 (__builtin_ia32_xsetbv): New builtins.
20223 * config/i386/i386.c (ix86_expand_special_args_builtin):
20225 (ix86_expand_builtin): Special expand for new intrinsics.
20226 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20227 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20228 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20230 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20232 * cfganal.c (inverted_post_order_compute): Change argument type
20234 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20235 * df-core.c (rest_of_handle_df_initialize): Adjust.
20236 (rest_of_handle_df_finish): Likewise.
20237 (df_analyze_1): Likewise.
20238 (df_analyze): Likewise.
20239 (loop_inverted_post_order_compute): Change argument to be a vec *.
20240 (df_analyze_loop): Adjust.
20241 (df_get_n_blocks): Likewise.
20242 (df_get_postorder): Likewise.
20243 * df.h (struct df_d): Change field to be a vec.
20244 * lcm.c (compute_laterin): Adjust.
20245 (compute_available): Likewise.
20246 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20247 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20248 * tree-ssa-pre.c (compute_antic): Likewise.
20250 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20252 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20253 (depth_first_search::depth_first_search): Change structure init
20254 function to this constructor.
20255 (depth_first_search::add_bb): Rename function to this member.
20256 (depth_first_search::execute): Likewise.
20257 (flow_dfs_compute_reverse_finish): Adjust.
20259 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20261 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20262 (longest_simple_path): Likewise.
20263 * shrink-wrap.c (spread_components): Likewise.
20264 (disqualify_problematic_components): Likewise.
20265 (emit_common_heads_for_components): Likewise.
20266 (emit_common_tails_for_components): Likewise.
20267 (insert_prologue_epilogue_for_components): Likewise.
20269 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20271 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20274 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20276 * df-core.c (df_set_blocks): Start using auto_bitmap.
20277 (df_compact_blocks): Likewise.
20278 * df-problems.c (df_rd_confluence_n): Likewise.
20279 * df-scan.c (df_insn_rescan_all): Likewise.
20280 (df_process_deferred_rescans): Likewise.
20281 (df_update_entry_block_defs): Likewise.
20282 (df_update_exit_block_uses): Likewise.
20283 (df_entry_block_bitmap_verify): Likewise.
20284 (df_exit_block_bitmap_verify): Likewise.
20285 (df_scan_verify): Likewise.
20286 * lra-constraints.c (lra_constraints): Likewise.
20287 (undo_optional_reloads): Likewise.
20288 (lra_undo_inheritance): Likewise.
20289 * lra-remat.c (calculate_gen_cands): Likewise.
20290 (do_remat): Likewise.
20291 * lra-spills.c (assign_spill_hard_regs): Likewise.
20292 (spill_pseudos): Likewise.
20293 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20294 (bitmap_set_subtract_values): Likewise.
20296 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20298 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20299 management with auto_bitmap.
20300 (fix_inter_tick): Likewise.
20301 (fix_recovery_deps): Likewise.
20302 * ira.c (add_store_equivs): Likewise.
20303 (find_moveable_pseudos): Likewise.
20304 (split_live_ranges_for_shrink_wrap): Likewise.
20305 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20306 (rtx_reuse_manager::seen_def_p): Likewise.
20307 (rtx_reuse_manager::set_seen_def): Likewise.
20308 * print-rtl.h (class rtx_reuse_manager): Likewise.
20310 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20312 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20314 (migrate_btr_def): Likewise.
20315 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20316 * df-core.c (loop_post_order_compute): Likewise.
20317 (loop_inverted_post_order_compute): Likewise.
20318 * hsa-common.h: Likewise.
20319 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20320 * init-regs.c (initialize_uninitialized_regs): Likewise.
20321 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20322 (inline_small_functions): Likewise.
20323 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20324 * ira.c (combine_and_move_insns): Likewise.
20325 (build_insn_chain): Likewise.
20326 * loop-invariant.c (find_invariants): Likewise.
20327 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20328 * predict.c (tree_predict_by_opcode): Likewise.
20329 (predict_paths_leading_to): Likewise.
20330 (predict_paths_leading_to_edge): Likewise.
20331 (estimate_loops_at_level): Likewise.
20332 (estimate_loops): Likewise.
20333 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20334 (spread_components): Likewise.
20335 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20336 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20337 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20338 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20339 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20340 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20341 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20342 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20343 (mark_threaded_blocks): Likewise.
20344 (thread_through_all_blocks): Likewise.
20345 * tree-ssa.c (verify_ssa): Likewise.
20346 (execute_update_addresses_taken): Likewise.
20347 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20351 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20353 (post_order_compute): Likewise.
20354 (inverted_post_order_compute): Likewise.
20355 (pre_and_rev_post_order_compute_fn): Likewise.
20357 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20359 * genrecog.c (int_set::int_set): Explicitly construct our
20360 auto_vec base class.
20361 * vec.h (auto_vec::auto_vec): New constructor.
20363 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20365 * bitmap.h (class auto_bitmap): New constructor taking
20366 bitmap_obstack * argument.
20368 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20370 * bitmap.h (class auto_bitmap): Change type of m_bits to
20371 bitmap_head, and adjust ctor / dtor and member operators.
20373 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20375 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20376 when returned register mode doesn't match original mode.
20378 2017-05-12 Jeff Law <law@redhat.com>
20379 Jakub Jelinek <jakub@redhat.com>
20381 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20382 we look for cc setter after the compare-elim changes.
20383 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20384 within the vector to match what compare-elim now expects.
20385 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20386 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20388 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20389 after the compare-elim changes.
20390 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20391 the vector to match what compare-elim now expects.
20392 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20393 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20394 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20395 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20396 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20398 * config/visium/visium.c (single_set_and_flags): Fix where
20399 we look for cc setter after the compare-elim changes.
20400 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20401 with the vector to match what compare-elim now expects.
20402 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20403 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20404 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20405 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20406 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20407 (neg<mode>2_insn_set_overflow): Likewise.
20409 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20411 PR middle-end/79794
20412 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20413 maybe_expand_insn call, set ops[0].target. If still set after call,
20414 set alt_rtl. Add extra arg to recursive calls.
20415 (extract_bit_field): Add alt_rtl argument. Pass to
20417 * expmed.h (extract_bit_field): Fix prototype.
20418 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20419 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20420 to extract_bit_field_calls.
20421 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20422 Pass alt_rtl to extract_bit_field calls.
20423 * calls.c (store_unaligned_arguments_into_psuedos)
20424 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20425 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20427 * optabs.h (struct expand_operand): Add target bitfield.
20429 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20431 * compare-elim.c (try_eliminate_compare): Canonicalize
20432 operation with embedded compare to
20433 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20434 (set (reg) (operation)].
20436 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20438 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20441 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20442 cost of adding a carry flag for ADC instruction.
20443 [case MINUS]: Ignore the cost of subtracting a carry flag
20444 for SBB instruction.
20446 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20448 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20450 * config/rs6000/bmiintrin.h: New file.
20451 * config/rs6000/bmi2intrin.h: New file.
20452 * config/rs6000/x86intrin.h: New file.
20454 2017-05-12 Jeff Law <law@redhat.com>
20456 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20459 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20461 PR middle-end/80707
20462 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20464 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20466 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20467 early expansion of vector divide builtins.
20468 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20469 builtins identified as having unsigned arguments.
20471 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20473 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20474 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20475 expansion of vector logical operations (and, andc, or, xor,
20478 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20480 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20481 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20483 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20485 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20486 early GIMPLE expansion of vector multiplies.
20488 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20490 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20491 TARGET_HAVE_MOVT conditional.
20492 (movt splitter): Likewise.
20494 2017-05-12 Richard Biener <rguenther@suse.de>
20496 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20497 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20498 Fold all stmts not inplace.
20500 2017-05-12 Richard Biener <rguenther@suse.de>
20502 PR tree-optimization/80713
20503 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20504 inserted_exprs bit for not removed stmts.
20506 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20508 PR middle-end/69921
20509 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20510 parallelized" attribute for parallelized OpenACC kernels.
20511 * omp-offload.c (execute_oacc_device_lower): Use it.
20513 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20514 Set "oacc kernels" attribute.
20515 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20516 parameter. Adjust all users.
20517 (oacc_fn_attrib_kernels_p): Remove function.
20518 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20519 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20520 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20521 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20522 assert "oacc kernels" attribute is set.
20524 2017-05-11 Carl Love <cel@us.ibm.com>
20526 * config/rs6000/rs6000-c: Add support for built-in functions
20527 vector unsigned char vec_popcnt (vector signed char)
20528 vector unsigned char vec_popcnt (vector unsigned char)
20529 vector unsigned short vec_popcnt (vector signed short)
20530 vector unsigned short vec_popcnt (vector unsigned short)
20531 vector unsigned int vec_popcnt (vector signed int)
20532 vector unsigned int vec_popcnt (vector unsigned int)
20533 vector unsigned long long vec_popcnt (vector signed long long)
20534 vector unsigned long long vec_popcnt (vector unsigned long long)
20535 vector signed long long vec_slo (vector signed long long,
20536 vector signed char)
20537 vector signed long long vec_slo (vector signed long long,
20538 vector unsigned char)
20539 vector unsigned long long vec_slo (vector unsigned long long,
20540 vector signed char)
20541 vector unsigned long long vec_slo (vector unsigned long long,
20542 vector unsigned char)
20543 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20544 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20545 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20546 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20547 * doc/extend.texi: Update the built-in documentation file for the
20548 new built-in functions.
20550 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20552 * attribs.h (sorted_attr_string): Move machine independent
20553 functions for target clone support from the i386 port to common
20554 code. Rename ix86_function_versions to common_function_versions.
20555 Rename make_name to make_unique_name.
20556 (common_function_versions): Likewise.
20557 (make_unique_name): Likewise.
20558 (make_dispatcher_decl): Likewise.
20559 (is_function_default_version): Likewise.
20560 * attribs.c (attr_strcmp): Likewise.
20561 (sorted_attr_string): Likewise.
20562 (common_function_versions): Likewise.
20563 (make_unique_name): Likewise.
20564 (make_dispatcher_decl): Likewise.
20565 (is_function_default_version): Likewise.
20566 * config/i386/i386.c (attr_strcmp): Likewise.
20567 (sorted_attr_string): Likewise.
20568 (ix86_function_versions): Likewise.
20569 (make_name): Likewise.
20570 (make_dispatcher_decl): Likewise.
20571 (is_function_default_version): Likewise.
20572 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20574 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20577 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20578 Account for direct move costs for vec_construct of integer
20581 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20584 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20585 (UNSPEC_STX_ATOMIC): Ditto.
20586 (loaddi_via_sse): New insn.
20587 (storedi_via_sse): Ditto.
20588 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20589 Update corresponding peephole2 patterns.
20590 (atomic_storedi_fpu): Ditto.
20592 2017-05-11 Julia Koval <julia.koval@intel.com>
20594 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20595 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20597 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20598 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20599 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20601 2017-05-11 Nathan Sidwell <nathan@acm.org>
20603 * graphite-poly.c: Include dumpfile.h.
20605 * dumpfle.h (dump_function): Declare here ...
20606 * tree-dump.h (dump_function): ... not here.
20607 * dumpfile.c: #include tree-cfg.h.
20608 (dump_function): Move here from ...
20609 * tree-dump.c (dump_function): ... here.
20610 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20611 * graphite-poly.c: Don't include tree-dump.h.
20612 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20613 * print-tree.c: Likewise.
20614 * stor-layout.c: Likewise.
20615 * tree-nested.c: Likewise.
20617 * dumpfile.c (dump_start): Use TDF_FLAGS.
20618 (dump_enable_all): Fix TDF_KIND check thinko.
20620 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20622 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20623 array entries to represent two legal parameterizations of the
20624 overloaded __builtin_cmpb function, as represented by the
20625 P6_OV_BUILTIN_CMPB constant.
20626 (altivec_resolve_overloaded_builtin): Add special case handling
20627 for the __builtin_cmpb function, as represented by the
20628 P6_OV_BUILTIN_CMPB constant.
20629 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20630 (BU_P6_64BIT_2): New macro.
20631 (BU_P6_OVERLOAD_2): New macro
20632 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20633 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20634 (CMPB): Add overload support to represent both 32-bit and 64-bit
20635 compare-bytes function.
20636 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20637 support for TARGET_CMPB.
20638 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20639 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20640 documentation of the __builtin_cmpb overloaded built-in function.
20642 2017-05-11 Richard Biener <rguenther@suse.de>
20644 PR tree-optimization/80705
20645 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20646 bases are not vectorizable.
20648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20650 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20651 when counting register pressure.
20653 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20655 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20656 (iv_ca_more_deps): Renamed to ...
20657 (iv_ca_compare_deps): ... this.
20658 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20660 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20662 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20664 (determine_group_iv_costs): ... here.
20665 (find_inv_vars_cb): Record inv var if it's not recorded before.
20667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20669 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20670 (get_shiftadd_cost): Ditto.
20672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20674 * tree-ssa-address.c: Include header file.
20675 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20677 (add_to_parts): Refactor.
20678 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20679 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20684 PR tree-optimization/53090
20685 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20687 (extract_cond_operands): Detect condition with IV on both sides
20688 and return COMP_IV_EXPR_2.
20689 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20690 (rewrite_use_compare): Simplify by removing call to function
20691 extract_cond_operands.
20693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20695 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20696 (extract_cond_operands): Detect condition comparing against non-
20697 invariant bound and return appropriate enum value.
20698 (find_interesting_uses_cond): Update use of extract_cond_operands.
20699 Handle its return value accordingly.
20700 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20704 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20705 nonlinear iv_use computation in loop invariant sensitive way.
20707 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20709 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20710 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20714 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20715 (dump_cand): Support iv_cand.inv_exprs.
20716 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20718 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20721 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20723 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20725 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20726 as local function. Include necessary header files.
20727 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20729 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20731 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20735 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20736 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20737 RSHIFT_EXPR and BIT_NOT_EXPR.
20739 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20741 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20742 (adjust_setup_cost): New parameter supporting round up adjustment.
20743 (struct address_cost_data): Delete.
20744 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20745 (split_address_cost, ptr_difference_cost): Delete.
20746 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20747 (struct ainc_cost_data): New struct.
20748 (get_address_cost_ainc): New function.
20749 (get_address_cost, get_computation_cost): Reimplement.
20750 (determine_group_iv_cost_address): Record inv_expr for all uses of
20752 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20753 (iv_ca_has_deps): Reimplemented to ...
20754 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20756 (iv_ca_extend): Call iv_ca_more_deps.
20758 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20760 * tree-ssa-address.c (struct mem_address): Move to header file.
20761 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20762 * tree-ssa-address.h (struct mem_address): Move from C file.
20763 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20767 * tree-affine.h (aff_combination_type): New interface.
20768 (aff_combination_zero_p): Remove static.
20769 (aff_combination_const_p): New interface.
20770 (aff_combination_singleton_var_p): New interfaces.
20772 2017-05-11 Richard Biener <rguenther@suse.de>
20774 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20775 Skip unreachable blocks and destinations.
20776 (eliminate): Move stmt removal and fixup ...
20777 (fini_eliminate): ... here. Skip inserted exprs.
20778 (pass_pre::execute): Move fini_pre after fini_eliminate.
20779 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20780 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20781 PRE to get rid of dead code that has invalid SSA form and
20782 split critical edges again.
20784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20786 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20788 2017-05-11 Richard Biener <rguenther@suse.de>
20790 * passes.c (execute_function_todo): Verify loops if they are
20791 said to be up-to-date.
20792 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20793 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20795 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20798 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20799 handle calling assemble_external ourself.
20802 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20803 modes with zero size. Enhance comment.
20805 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20807 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20808 built-ins for vec_xl and vec_xst with short and char pointer
20811 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20813 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20814 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20815 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20816 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20817 (_mm_maskz_min_round_ss): New intrinsics.
20818 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20819 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20820 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20821 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20822 (__builtin_ia32_minss_mask_round): New builtins.
20823 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20824 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20825 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20827 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20828 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20830 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20833 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20835 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20836 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20837 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20838 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20839 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20840 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20841 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20842 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20843 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20844 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20845 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20846 (__builtin_ia32_mulss_mask_round): New builtins.
20847 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20848 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20849 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20851 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20852 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20854 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20857 2017-05-10 Julia Koval <julia.koval@intel.com>
20859 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20860 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20861 (_mm256_setr_m128i): New intrinsics.
20863 2017-05-10 Julia Koval <julia.koval@intel.com>
20865 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20866 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20867 (_mm_maskz_rcp14_ss): New intrinsics.
20868 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20869 (__builtin_ia32_rcp14ss_mask): New builtins.
20870 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20872 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20874 PR tree-optimization/51513
20875 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20876 (assert_unreachable_fallthru_edge_p): Use it.
20877 (group_case_labels_stmt): Likewise.
20878 * tree-cfg.h: Prototype it.
20879 * stmt.c: Include cfghooks.h and tree-cfg.h.
20880 (emit_case_dispatch_table) <gap_label>: New local variable.
20881 Use it to fill dispatch table gaps.
20882 Test for default_label before updating probabilities.
20883 (expand_case) <default_label>: Remove unneeded initialization.
20884 Test for unreachable default case statement and remove its edge.
20885 Set default_label accordingly.
20886 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20888 2017-05-10 Carl Love <cel@us.ibm.com>
20890 * config/rs6000/rs6000-c: Add support for built-in functions
20891 vector signed char vec_neg (vector signed char)
20892 vector signed short int vec_neg (vector short int)
20893 vector signed int vec_neg (vector signed int)
20894 vector signed long long vec_neg (vector signed long long)
20895 vector float vec_neg (vector float)
20896 vector double vec_neg (vector double)
20897 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20899 * config/rs6000/altivec.h: Add define for vec_neg
20900 * doc/extend.texi: Update the built-in documentation for the
20901 new built-in functions.
20903 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20905 PR tree-optimization/77644
20906 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20908 2017-05-10 Nathan Sidwell <nathan@acm.org>
20910 * dumpfile.h (TDI_lang_all): New.
20911 (TDF_KIND): New. Renumber others
20912 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20914 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20916 (get_dump_file_name): Adjust suffix generation.
20917 (dump_enable_all): Use TDF_KIND.
20918 * doc/invoke.texi (-fdump-lang-all): Document.
20920 * dumpfile.h: Tabify.
20922 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20925 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20926 Move member access before delete.
20928 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20930 * tree-inline.c (expand_call_inline): Split block at stmt
20933 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20936 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20937 are now unused after splitting mov{sf,sd}_hardfloat.
20938 (f32_lr2): Likewise.
20939 (f32_lm): Likewise.
20940 (f32_lm2): Likewise.
20941 (f32_li): Likewise.
20942 (f32_li2): Likewise.
20943 (f32_lv): Likewise.
20944 (f32_sr): Likewise.
20945 (f32_sr2): Likewise.
20946 (f32_sm): Likewise.
20947 (f32_sm2): Likewise.
20948 (f32_si): Likewise.
20949 (f32_si2): Likewise.
20950 (f32_sv): Likewise.
20951 (f32_dm): Likewise.
20952 (f32_vsx): Likewise.
20953 (f32_av): Likewise.
20954 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20955 For movsf, order stores so the VSX stores occur before the GPR
20956 store which encourages the register allocator to use a traditional
20957 FPR instead of a GPR. For movsd, order the stores so that the GPR
20958 store comes before the VSX stores to allow the power6 to work.
20959 This is due to the power6 not having a 32-bit integer store
20960 instruction from a FPR.
20961 (movsf_hardfloat): Likewise.
20962 (movsd_hardfloat): Likewise.
20964 2017-05-09 Martin Sebor <msebor@redhat.com>
20966 PR translation/80280
20967 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20970 PR translation/80280
20971 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20972 data member added in r247778.
20973 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20975 2017-05-09 Nathan Sidwell <nathan@acm.org>
20977 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20979 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20982 2017-05-09 Marek Polacek <polacek@redhat.com>
20984 * doc/invoke.texi: Fix typo.
20986 2017-05-09 Richard Biener <rguenther@suse.de>
20988 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20989 (vrp_val_is_min): Likewise.
20990 (set_value_range_to_value): Likewise.
20991 (set_value_range_to_nonnegative): Likewise.
20992 (gimple_assign_nonzero_p): Likewise.
20993 (gimple_stmt_nonzero_p): Likewise.
20994 (vrp_int_const_binop): Likewise. Remove unreachable case.
20995 (adjust_range_with_scev): Adjust comments.
20996 (compare_range_with_value): Likewise.
20997 (extract_range_from_phi_node): Likewise.
20998 (test_for_singularity): Likewise.
21000 2017-05-09 Richard Biener <rguenther@suse.de>
21002 * tree-vrp.c (get_single_symbol): Add assert that we don't
21003 get overflowed constants as invariant part.
21004 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21005 checks. Use wi::cmp instead of recursing for integer constants.
21006 (compare_values): Just ignore whether we assumed undefined
21007 overflow instead of failing the compare.
21008 (extract_range_for_var_from_comparison_expr): Add comment before the
21009 TREE_NO_WARNING sets.
21010 (test_for_singularity): Likewise.
21011 (extract_range_from_comparison): Do not disable optimization
21012 when we assumed undefined overflow.
21013 (extract_range_basic): Remove init of unused var.
21015 2017-05-09 Richard Biener <rguenther@suse.de>
21017 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21018 (extract_range_from_multiplicative_op_1): Adjust.
21019 (extract_range_from_binary_expr_1): Use int_const_binop.
21021 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21024 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21025 rs6000_store_data_bypass_p in seven define_bypass directives and
21026 in several comments.
21027 * config/rs6000/rs6000-protos.h: Add prototype for
21028 rs6000_store_data_bypass_p function.
21029 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21030 function implements slightly different (rs6000-specific) semantics
21031 than store_data_bypass_p, returning false rather than aborting
21032 with assertion error when arguments do not satisfy the
21033 requirements of store data bypass.
21034 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21035 rs6000_store_data_bypass_p.
21037 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21039 * config/xtensa/xtensa-protos.h
21040 (xtensa_initial_elimination_offset): New declaration.
21041 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21042 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21043 macro definition, add case for FRAME_POINTER_REGNUM when
21044 FRAME_GROWS_DOWNWARD.
21045 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21046 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21047 xtensa_initial_elimination_offset.
21049 2017-05-08 Nathan Sidwell <nathan@acm.org>
21051 * doc/invoke.texi: Alphabetize -fdump options.
21053 2017-05-08 Martin Sebor <msebor@redhat.com>
21055 PR translation/80280
21056 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21058 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21060 * target.def (compute_frame_layout): New optional target hook.
21061 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21062 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21063 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21065 * reload1.c (verify_initial_elim_offsets): Likewise.
21066 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21067 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21068 (arm_get_frame_offsets): Split up into this ...
21069 (arm_compute_frame_layout): ... and this function.
21071 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21073 * config/aarch64/constraints.md (Usa): New constraint.
21074 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21076 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21078 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21079 with_multilib_list after it has been checked.
21081 2017-05-08 Richard Biener <rguenther@suse.de>
21083 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21084 (bitmap_set_subtract_values): Likewise.
21086 2017-05-08 Richard Biener <rguenther@suse.de>
21088 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21089 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21091 (gimple_stmt_nonzero_warnv_p): Rename to ...
21092 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21094 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21095 (extract_range_basic): Adjust, do not disable propagation on
21096 strict overflow sensitive simplification.
21097 (vrp_visit_cond_stmt): Likewise.
21099 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21101 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21102 body size unconditionally.
21104 2017-05-07 Jeff Law <law@redhat.com>
21107 2017-05-06 Jeff Law <law@redhat.com>
21108 PR tree-optimization/78496
21109 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21112 PR tree-optimization/78496
21113 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21114 (simplify_stmt_using_ranges): Call it.
21115 (vrp_dom_walker::before_dom_children): Extract equivalences
21116 from an ASSERT_EXPR with an equality comparison against a
21119 2017-05-06 Jeff Law <law@redhat.com>
21121 PR tree-optimization/78496
21122 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21125 PR tree-optimization/78496
21126 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21127 (simplify_stmt_using_ranges): Call it.
21128 (vrp_dom_walker::before_dom_children): Extract equivalences
21129 from an ASSERT_EXPR with an equality comparison against a
21132 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21134 * lra-constraints.c (lra_copy_reg_equiv): New function.
21135 (split_reg): Use it to copy equivalence information from the
21136 original register to the spill register.
21138 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21140 PR rtl-optimization/75964
21141 * simplify-rtx.c (simplify_const_relational_operation): Remove
21142 invalid handling of comparisons of integer ABS.
21144 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21146 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21147 initialize to zero.
21148 (init_regs): Remove declaration.
21149 (function_arg_advance_32): Initialize error_p as boolean variable.
21151 2017-05-05 Nathan Sidwell <nathan@acm.org>
21153 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21154 lines. Use for (;;).
21156 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21158 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21159 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21160 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21161 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21162 VF=2 that require versioning.
21164 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21166 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21169 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21171 * diagnostic.h (diagnostic_override_option_index): Convert from
21172 macro to inline function.
21174 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21176 * diagnostic.c (last_module_changed_p): New function.
21177 (set_last_module): New function.
21178 (diagnostic_report_current_module): Convert macro usage to
21179 the above functions.
21180 * diagnostic.h (diagnostic_context::last_module): Strengthen
21181 from const line_map * to const line_map_ordinary *.
21182 (diagnostic_last_module_changed): Delete macro.
21183 (diagnostic_set_last_module): Delete macro.
21185 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21187 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21188 with diagnostic_report_diagnostic.
21189 (diagnostic_n_impl_richloc): Likewise.
21190 * diagnostic.h (report_diagnostic): Delete macro.
21191 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21192 with diagnostic_report_diagnostic.
21193 * substring-locations.c (format_warning_va): Likewise.
21195 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21197 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21198 save/restor of format_spec. Move option-printing code to...
21199 (print_option_information): ...this new function, and
21200 reimplement by simply printing to the pretty_printer,
21201 rather than appending to the format string.
21203 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21205 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21206 handling logic into...
21207 (update_effective_level_from_pragmas): ...this new function.
21209 2017-05-04 Andrew Waterman <andrew@sifive.com>
21211 * config/riscv/riscv.opt (mstrict-align): New option.
21212 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21213 (SLOW_UNALIGNED_ACCESS): Define.
21214 (riscv_slow_unaligned_access): Declare.
21215 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21217 (riscv_slow_unaligned_access): New variable.
21218 (rocket_tune_info): Set slow_unaligned_access to true.
21219 (optimize_size_tune_info): Set slow_unaligned_access to false.
21220 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21221 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21222 (riscv_option_override): Set riscv_slow_unaligned_access.
21223 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21225 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21227 * config/riscv/riscv.md: Unify indentation.
21229 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21234 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21236 (extendsi<mode>2): Add support for doing sign extension via
21237 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21238 don't have ISA 3.0 instructions.
21239 (extendsi<mode>2 splitter): Likewise.
21240 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21241 generate the normal insns since SImode can now go in vector
21242 registers. Disallow the special UNSPECs needed for previous
21243 machines to hide SImode being used. Add new insns
21244 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21245 (fix_trunc<mode>si2_stfiwx): Likewise.
21246 (fix_trunc<mode>si2_internal): Likewise.
21247 (fixuns_trunc<mode>si2): Likewise.
21248 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21249 (fctiw<u>z_<mode>_smallint): Likewise.
21250 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21251 of floating point to 32-bit integer from doing a direct move to
21252 the GPR registers to do a store.
21253 (fctiwz_<mode>): Break long line.
21255 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21257 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21258 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21259 (addr_list, addr_offset_valid_p): New.
21260 (split_address_groups): Check offset validity with above function.
21261 (gt-tree-ssa-loop-ivopts.h): Include header file.
21263 2017-05-05 Nathan Sidwell <nathan@acm.org>
21265 * config.gcc (arm*-*-*): Add missing 'fi'.
21267 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21269 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21270 included in -fopt-info does not matter.
21271 * doc/optinfo.texi (-fopt-info): Fix description of default
21272 behavour. Explicitly say order of options included in -fopt-info
21275 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21277 * config.gcc: Allow combinations of aprofile and rmprofile values for
21278 --with-multilib-list.
21279 * config/arm/t-multilib: New file.
21280 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21281 variables. Remove setting of ISA and floating-point ABI in
21282 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21283 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21284 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21285 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21287 * config/arm/t-rmprofile: Likewise except for the matches changes.
21288 * doc/install.texi (--with-multilib-list): Document the combination of
21289 aprofile and rmprofile values and warn about pitfalls in doing that.
21291 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21293 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21294 (movdi_aarch64): Likewise.
21296 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21298 PR tree-optimization/80632
21299 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21301 (build_arrays): Initialize it for virtual phis.
21302 (fix_phi_nodes): Use it for virtual phis.
21304 PR tree-optimization/80558
21305 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21306 [x, y] op z into [x op, y op z] for op & or | if conditions
21309 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21310 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21313 * config/arm/arm.md (use_literal_pool): Remove.
21314 (64-bit immediate split): No longer takes cost into consideration
21315 if arm_disable_literal_pool is enabled.
21316 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21317 used when arm_disable_literal_pool is enabled.
21318 (arm_max_const_double_inline_cost): Remove use of
21319 arm_disable_literal_pool.
21320 (push_minipool_fix): Add assert.
21321 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21322 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21323 (no_literal_pool_sf_immediate): New.
21325 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21327 PR tree-optimization/80613
21328 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21329 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21331 2017-05-05 Richard Biener <rguenther@suse.de>
21333 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21335 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21337 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21338 of this flag from insn conditions due to removal from r247495.
21340 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21342 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21344 (arm_early_store_addr_dep_ptr): Likewise.
21345 * config/arm/aarch-common-protos.h
21346 (arm_early_load_addr_dep_ptr): Add prototype.
21347 (arm_early_store_addr_dep_ptr): Likewise.
21348 * config/arm/cortex-a53.md: Add new bypasses.
21350 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21352 * tree.c (next_type_uid): Change type to unsigned.
21353 (type_hash_canon): Decrement back next_type_uid if
21354 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21355 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21358 2017-05-04 Martin Sebor <msebor@redhat.com>
21360 * builtins.c: Fix a trivial typo in a comment.
21362 PR middle-end/79234
21363 * builtins.c (check_sizes): Adjust to handle reading past the end.
21364 Avoid printing excessive upper bound of ranges. Use %E to print
21365 tree nodes instead of converting them to %wu.
21366 (expand_builtin_memchr): New function.
21367 (compute_dest_size): Rename...
21368 (compute_objsize): ...to this.
21369 (expand_builtin_memcpy): Adjust.
21370 (expand_builtin_mempcpy): Adjust.
21371 (expand_builtin_strcat): Adjust.
21372 (expand_builtin_strcpy): Adjust.
21373 (check_strncat_sizes): Adjust.
21374 (expand_builtin_strncat): Adjust.
21375 (expand_builtin_strncpy): Adjust and simplify.
21376 (expand_builtin_memset): Adjust.
21377 (expand_builtin_bzero): Adjust.
21378 (expand_builtin_memcmp): Adjust.
21379 (expand_builtin): Handle memcmp.
21380 (maybe_emit_chk_warning): Check strncat just once.
21382 2017-05-04 Martin Sebor <msebor@redhat.com>
21384 PR preprocessor/79214
21385 PR middle-end/79222
21386 PR middle-end/79223
21387 * builtins.c (check_sizes): Add inlining context and issue
21388 warnings even when -Wno-system-headers is set.
21389 (check_strncat_sizes): Same.
21390 (expand_builtin_strncat): Same.
21391 (expand_builtin_memmove): New function.
21392 (expand_builtin_stpncpy): Same.
21393 (expand_builtin): Handle memmove and stpncpy.
21395 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21397 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21398 which is not used any more.
21400 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21402 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21404 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21406 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21407 (cortexa53_tunings): Likewise.
21408 (cortexa57_tunings): Likewise.
21409 (cortexa72_tunings): Likewise.
21410 (cortexa73_tunings): Likewise.
21412 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21414 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21415 Set loop alignment to 8.
21417 2017-05-04 Martin Sebor <msebor@redhat.com>
21419 PR translation/80280
21420 * builtins.c (expand_builtin_object_size): Add missing quoting to
21421 %D and like directives.
21422 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21423 (hsa_type_for_tree_type): Same.
21424 (verify_function_arguments): Same.
21425 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21426 * varasm.c (get_section): Same.
21429 2017-05-04 Martin Sebor <msebor@redhat.com>
21431 PR translation/80280
21432 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21434 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21436 * config/aarch64/aarch64.c (generic_addrcost_table):
21437 Change HI/TI mode setting.
21439 2017-05-04 Martin Jambor <mjambor@suse.cz>
21441 PR tree-optimization/80622
21442 * tree-sra.c (comes_initialized_p): New function.
21443 (build_accesses_from_assign): Only set write lazily when
21444 comes_initialized_p is false.
21445 (analyze_access_subtree): Use comes_initialized_p.
21446 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21447 instead of testing for PARM_DECL.
21449 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21451 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21452 constraint on operand 0 to allow more general addressing modes.
21453 Adjust output template.
21454 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21456 * config/aarch64/aarch64-protos.h
21457 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21458 * config/aarch64/constraints.md (Dp): New address constraint.
21459 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21462 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21464 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21465 update use of estimate_ipcp_clone_size_and_time.
21466 (estimate_local_effects): Update use of
21467 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21468 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21469 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21470 Return nonspecialized time.
21472 2017-05-04 Richard Biener <rguenther@suse.de>
21474 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21475 for the last VUSE which def dominates the PHI. Directly call
21477 (get_continuation_for_phi_1): Remove.
21479 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21481 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21482 to explain the use of truncating division. Cap the number of
21483 iterations to the maximum given by nb_iterations_upper_bound,
21486 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21488 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21489 * configure: Regenerate.
21490 * config.in: Regenerate.
21491 * config/i386/driver-mingw32.c: new file.
21492 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21493 * config.host: Link driver-mingw32.o on MinGW host.
21494 * doc/install.texi: Document new --enable-mingw-wildcard configure
21497 2017-05-04 Marek Polacek <polacek@redhat.com>
21499 PR tree-optimization/80612
21500 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21502 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21503 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21505 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21506 (movt splitter): Likewise.
21507 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21508 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21509 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21510 block for Thumb-1 with MOVT.
21511 (thumb2_legitimate_address_p): Move code block ...
21512 (can_avoid_literal_pool_for_label_p): ... into this new function.
21513 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21515 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21516 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21517 "M-profile targets with the MOVT instruction".
21519 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21521 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21522 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21523 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21525 2017-05-04 Martin Liska <mliska@suse.cz>
21527 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21528 variable cond_code.
21530 2017-05-04 Richard Biener <rguenther@suse.de>
21532 * tree.c (array_at_struct_end_p): Handle arrays at struct
21533 end with flexarrays more conservatively. Refactor and treat
21534 arrays of arrays or aggregates more strict. Fix
21535 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21536 * tree.c (array_at_struct_end_p): Adjust prototype.
21537 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21538 * gimple-fold.c (get_range_strlen): Likewise.
21539 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21541 2017-05-04 Richard Biener <rguenther@suse.de>
21543 PR tree-optimization/31130
21544 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21546 (supports_overflow_infinity): Likewise.
21547 (is_negative_overflow_infinity): Likewise.
21548 (is_positive_overflow_infinity): Likewise.
21549 (is_overflow_infinity): Likewise.
21550 (stmt_overflow_infinity): Likewise.
21551 (overflow_infinity_range_p): Likewise.
21552 (usable_range_p): Remove as always returning true.
21553 (make_overflow_infinity): Remove.
21554 (negative_overflow_infinity): Likewise.
21555 (positive_overflow_infinity): Likewise.
21556 (avoid_overflow_infinity): Likewise.
21557 (set_value_range): Adjust accordingly.
21558 (set_value_range_to_nonnegative): Likewise, remove now unused
21559 overflow_infinity arg.
21560 (vrp_operand_equal_p): Adjust.
21561 (update_value_range): Likewise.
21562 (range_int_cst_singleton_p): Likewise.
21563 (operand_less_p): Likewise.
21564 (compare_values_warnv): Likewise.
21565 (extract_range_for_var_from_comparison_expr): Likewise.
21566 (vrp_int_const_binop): Likewise.
21567 (zero_nonzero_bits_from_vr): Likewise.
21568 (extract_range_from_multiplicative_op_1): Likewise.
21569 (extract_range_from_binary_expr_1): Likewise.
21570 (extract_range_from_unary_expr): Likewise.
21571 (extract_range_from_comparison): Likewise.
21572 (extract_range_basic): Likewise.
21573 (adjust_range_with_scev): Likewise.
21574 (compare_ranges): Likewise.
21575 (compare_range_with_value): Likewise.
21576 (dump_value_range): Likewise.
21577 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21579 (simplify_cond_using_ranges): Adjust.
21581 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21583 * brig-builtins.def: Added a builtin for class_f64.
21584 * builtin-types.def: Added a builtin type needed by class_f64.
21586 2017-05-03 Jason Merrill <jason@redhat.com>
21588 * timevar.def: Add TV_CONSTEXPR.
21590 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21592 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21594 2017-05-03 Martin Jambor <mjambor@suse.cz>
21596 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21597 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21598 * ipa-cp.c (ipcp_propagate_stage): Do not call
21599 ipa_update_after_lto_read.
21600 * ipa-inline.c (ipa_inline): Likewise.
21602 2017-05-03 Martin Jambor <mjambor@suse.cz>
21604 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21605 tag. Added a default constructor and a destructor.
21606 (ipa_edge_args_sum_t): New class;
21607 (ipa_edge_args_sum): Declare.
21608 (ipa_edge_args_vector): Remove declaration.
21609 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21610 (ipa_free_edge_args_substructures): Remove declaration.
21611 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21612 (ipa_edge_args_info_available_for_edge_p): Likewise.
21613 * ipa-prop.c (ipa_edge_args_vector): Removed.
21614 (edge_removal_hook_holder): Likewise.
21615 (edge_duplication_hook_holder): Likewise.
21616 (ipa_edge_args_sum): New variable.
21617 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21618 ipa_edge_args_vector.
21619 (ipa_free_edge_args_substructures): Likewise.
21620 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21621 ipa_edge_args_vector.
21622 (ipa_edge_removal_hook): Turned into method
21623 ipa_edge_args_sum_t::remove.
21624 (ipa_edge_duplication_hook): Turned into method
21625 ipa_edge_args_sum_t::duplicate.
21626 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21627 registering edge hooks.
21628 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21629 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21630 ipa_edge_args_sum instead of ipa_edge_args_vector.
21631 * ipa-profile.c (ipa_profile): Likewise.
21633 2017-05-03 Martin Jambor <mjambor@suse.cz>
21635 * symbol-summary.h (function_summary): New method exists.
21636 (function_summary::symtab_removal): Deallocate through release.
21637 (call_summary): New class.
21638 (gt_ggc_mx): New overload.
21639 (gt_pch_nx): Likewise.
21640 (gt_pch_nx): Likewise.
21642 2017-05-03 Jeff Law <law@redhat.com>
21644 PR tree-optimization/78496
21645 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21646 from simplify_cond_using_ranges. Split off code to walk
21647 backwards through casts into ...
21648 (simplify_cond_using_ranges_2): New function.
21649 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21650 (execute_vrp): After identifying jump threads, call
21651 simplify_cond_using_ranges_2.
21653 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21656 * ipa-inline.h (inline_summary): Add ctor.
21657 (create_ggc): Do not use ggc_cleared_alloc.
21659 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21660 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21662 * gcc.c (handle_braces): Support escaping in switch matching
21664 * doc/invoke.texi (Spec Files): Document it.
21665 Remove superfluous @code markup in items.
21667 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21669 * diagnostic-show-locus.c (struct column_range): New struct.
21670 (get_affected_columns): New function.
21671 (get_printed_columns): New function.
21672 (struct correction): New struct.
21673 (correction::ensure_capacity): New function.
21674 (correction::ensure_terminated): New function.
21675 (struct line_corrections): New struct.
21676 (line_corrections::~line_corrections): New dtor.
21677 (line_corrections::add_hint): New function.
21678 (layout::print_trailing_fixits): Reimplement in terms of the new
21680 (selftest::test_overlapped_fixit_printing): New function.
21681 (selftest::diagnostic_show_locus_c_tests): Call it.
21683 2017-05-03 Nathan Sidwell <nathan@acm.org>
21685 Canonicalize canonical type hashing
21686 * tree.h (type_hash_canon_hash): Declare.
21687 * tree.c (type_hash_list, attribute_hash_list): Move into
21688 type_hash_canon_hash.
21689 (build_type_attribute_qual_variant): Break out hash code calc into
21690 type_hash_canon_hash.
21691 (type_hash_canon_hash): New. Generic type hash computation.
21692 (build_range_type_1, build_array_type_1, build_function_type,
21693 build_method_type_directly, build_offset_type, build_complex_type,
21694 make_vector_type): Call it.
21696 2017-05-03 Richard Biener <rguenther@suse.de>
21698 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21699 When all DRs have unknown misaligned do not always peel
21700 when there is a store but apply the same costing model as if
21701 there were only loads.
21703 2017-05-03 Richard Biener <rguenther@suse.de>
21706 PR tree-optimization/80492
21707 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21708 compare_base_decls returning dont-know properly.
21710 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21712 * config/arm/iterators.md (CCSI): New mode iterator.
21713 (arch): New mode attribute.
21714 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21715 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21716 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21717 code iterator for success result mode.
21718 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21719 the corresponding new insn generators.
21721 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21724 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21725 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21727 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21729 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21730 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21731 (DDR_A): Wrap DDR argument in brackets.
21732 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21733 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21734 (DDR_REVERSED_P): Likewise.
21736 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21738 PR tree-optimization/79472
21739 * tree-switch-conversion.c (struct switch_conv_info): Add
21740 contiguous_range and default_case_nonstandard fields.
21741 (collect_switch_conv_info): Compute contiguous_range and
21742 default_case_nonstandard fields, don't clear final_bb if
21743 contiguous_range and only the default case doesn't have the required
21745 (check_all_empty_except_final): Set default_case_nonstandard instead
21746 of failing if contiguous_range and the default case doesn't have empty
21748 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21749 and only the default case doesn't have the required constants. Skip
21751 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21752 if default_case_nonstandard.
21753 (build_constructors): Build constant 1 just once. Assert that default
21754 values aren't inserted in between cases if contiguous_range. Skip
21756 (build_arrays): Skip virtual phis.
21757 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21758 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21759 Handle virtual phis.
21760 (gen_inbound_check): Handle default_case_nonstandard case.
21761 (process_switch): Adjust check_final_bb caller. Call
21762 gather_default_values with the first non-default case instead of
21763 default case if default_case_nonstandard.
21765 2017-05-02 Nathan Sidwell <nathan@acm.org>
21767 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21768 check. Fix formatting.
21770 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21772 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21773 errors when comparing specialized and unspecialized times.
21775 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21777 * diagnostic-show-locus.c
21778 (layout::should_print_annotation_line_p): Make private.
21779 (layout::print_annotation_line): Make private.
21780 (layout::annotation_line_showed_range_p): Make private.
21781 (layout::show_ruler): Make private.
21782 (layout::print_source_line): Make private. Pass in line and
21783 line_width, rather than calling location_get_source_line. Drop
21785 (layout::print_leading_fixits): New method.
21786 (layout::print_any_fixits): Rename to...
21787 (layout::print_trailing_fixits): ...this, and make private.
21788 Don't print newline fixits.
21789 (diagnostic_show_locus): Move logic for printing one row into...
21790 (layout::print_line): ...this new function. Move the
21791 location_get_source_line call and error-handling from
21792 print_source_line to here. Call print_leading_fixits, and rename
21793 print_any_fixits to print_trailing_fixits.
21794 (selftest::test_fixit_insert_containing_newline): Update now that
21795 newlines are partially supported.
21796 (selftest::test_fixit_insert_containing_newline_2): New test.
21797 (selftest::test_fixit_replace_containing_newline): Update comments.
21798 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21799 * edit-context.c (class added_line): New class.
21800 (class edited_line): Describe newline handling in comment.
21801 (edited_line::actually_edited_p): New method.
21802 (edited_line::print_content): Delete redundant decl.
21803 (edited_line::m_predecessors): New field.
21804 (edited_file::print_content): Call edited_line::print_content.
21805 (edited_file::print_diff): Update to support newlines.
21806 (edited_file::print_diff_hunk): Likewise.
21807 (edited_file::print_run_of_changed_lines): New function.
21808 (edited_file::print_diff_line): Convert to...
21809 (print_diff_line): ...this.
21810 (edited_file::get_effective_line_count): New function.
21811 (edited_line::edited_line): Initialize new field m_predecessors.
21812 (edited_line::~edited_line): Clean up m_predecessors.
21813 (edited_line::apply_fixit): Handle newlines.
21814 (edited_line::get_effective_line_count): New function.
21815 (edited_line::print_content): New function.
21816 (edited_line::print_diff_lines): New function.
21817 (selftest::test_applying_fixits_insert_containing_newline): New
21819 (selftest::test_applying_fixits_replace_containing_newline): New
21821 (selftest::insert_line): New function.
21822 (selftest::test_applying_fixits_multiple_lines): Add example of
21824 (selftest::edit_context_c_tests): Call the new tests.
21826 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21828 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21829 parameter cand. Update dump information.
21830 (get_computation_cost): Update uses.
21832 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21834 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21835 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21836 (get_computation_at, rewrite_use_address): Update use of
21837 get_computation_aff.
21839 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21841 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21842 (get_computation): Delete.
21843 (get_computation_cost): Implement like get_computation_cost_at.
21844 Use get_computation_at.
21845 (get_computation_cost_at): Delete.
21846 (rewrite_use_nonlinear_expr): Use get_computation_at.
21847 (rewrite_use_compare, remove_unused_ivs): Ditto.
21849 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21851 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21853 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21855 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21856 (ivopts_global_cost_for_size): Rename parameter and update uses.
21857 (iv_ca_recount_cost): Update uses.
21858 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21859 candidates seperately in n_invs and n_cands.
21860 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21862 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21864 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21865 (find_inv_vars_cb): New.
21866 (find_depends): Renamed to ...
21867 (find_inv_vars): ... this.
21868 (add_candidate_1, force_var_cost): Call find_inv_vars.
21869 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21871 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21873 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21874 inv_vars. Add inv_exprs.
21875 (struct iv_cand): Rename depends_on to inv_vars.
21876 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21877 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21878 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21879 (dump_cand): Dump inv_vars.
21880 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21881 (record_invariant, find_depends, add_candidate_1): Ditto.
21882 (set_group_iv_cost, force_var_cost): Ditto.
21883 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21884 (get_computation_cost_at, get_computation_cost): Ditto.
21885 (determine_group_iv_cost_generic): Ditto.
21886 (determine_group_iv_cost_address): Ditto.
21887 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21888 (determine_group_iv_costs): Ditto.
21889 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21890 (iv_ca_set_remove_invariants): Renamed to ...
21891 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21892 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21893 (iv_ca_set_add_invariants): Renamed to ...
21894 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21895 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21896 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21897 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21898 (create_new_ivs): Remove useless dump.
21900 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21902 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21904 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21905 (iv_ca_set_no_cp, create_new_iv): Ditto.
21907 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21909 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21911 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21913 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21914 function tree_check2.
21916 2017-05-02 Martin Liska <mliska@suse.cz>
21918 * doc/gcov.texi: Add missing preposition.
21919 * gcov.c (function_info::function_info): Properly fill up
21920 all member variables.
21922 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21924 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21926 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21928 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21930 2017-05-02 Martin Liska <mliska@suse.cz>
21933 * lto-streamer-in.c (lto_read_tree_1): Remove
21934 LTO_STREAMER_DEBUG.
21935 * lto-streamer.c (struct tree_hash_entry): Likewise.
21936 (struct tree_entry_hasher): Likewise.
21937 (tree_entry_hasher::hash): Likewise.
21938 (tree_entry_hasher::equal): Likewise.
21939 (lto_streamer_init): Likewise.
21940 (lto_orig_address_map): Likewise.
21941 (lto_orig_address_get): Likewise.
21942 (lto_orig_address_remove): Likewise.
21943 * lto-streamer.h: Likewise.
21944 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21945 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21947 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21949 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21950 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21951 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21952 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21953 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21954 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21955 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21956 (mm_maskz_sub_ss): New intrinsics.
21957 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21958 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21959 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21960 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21961 (__builtin_ia32_subss_mask_round): New builtins.
21962 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21963 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21964 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21966 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21967 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21969 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21972 2017-05-02 Martin Jambor <mjambor@suse.cz>
21974 PR tree-optimization/78687
21975 * tree-sra.c (access): New field parent.
21976 (process_subtree_disqualification): New function.
21977 (disqualify_candidate): Call it.
21978 (build_accesses_from_assign): Reset write flag if creating an
21980 (build_access_subtree): Fill in parent field and also prpagate
21981 down grp_write flag.
21982 (create_artificial_child_access): New parameter set_grp_write, set
21983 grp_write to its value.
21984 (propagate_subaccesses_across_link): Also propagate grp_write flag
21986 (propagate_all_subaccesses): Push the closest parent back to work
21987 queue if add_access_to_work_queue returned true.
21989 2017-05-02 Richard Biener <rguenther@suse.de>
21991 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21992 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21993 -fstrict-overflow documentation.
21994 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21995 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21996 flag_strict_overflow.
21997 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21998 * lto-opts.c (lto_write_options): Do not stream it.
21999 * lto-wrapper.c (merge_and_complain): Do not handle it.
22000 * opts.c (default_options_table): Do not set -fstrict-overflow.
22001 (finish_options): Likewise do not clear it when sanitizing.
22002 * simplify-rtx.c (simplify_const_relational_operation): Do not
22003 test flag_strict_overflow.
22005 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22007 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22008 using enabled attribute.
22009 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22010 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22011 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22012 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22013 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22014 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22015 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22016 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22017 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22018 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22020 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22022 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22024 2017-05-02 Richard Biener <rguenther@suse.de>
22026 PR tree-optimization/80591
22028 2017-04-10 Richard Biener <rguenther@suse.de>
22030 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22033 2017-05-02 Richard Biener <rguenther@suse.de>
22035 PR tree-optimization/80549
22036 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22037 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22038 headers if they do not have a preheader.
22040 2017-05-02 Martin Liska <mliska@suse.cz>
22043 * common.opt: Fix typo.
22044 * doc/invoke.texi: Likewise.
22046 2017-05-01 Jan Beulich <jbeulich@suse.com>
22048 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22049 swapping, add (x,x,m,x,n) alternative.
22051 2017-05-01 Nathan Sidwell <nathan@acm.org>
22053 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22054 unnecessary unadjusted_alignment check.
22056 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22059 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22061 * gimplify.c (gimplify_cilk_detach): New function.
22062 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22063 * tree-core.h: Document EXPR_CILK_SPAWN.
22064 * tree.h (EXPR_CILK_SPAWN): Define.
22066 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22068 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22069 to use new fixit_hint representation, using the "replace" logic.
22070 (get_line_span_for_fixit_hint): Likewise.
22071 (layout::print_any_fixits): Likewise.
22072 (selftest::test_one_liner_many_fixits): Rename to...
22073 (selftest::test_one_liner_many_fixits_1): ...this, and update
22074 comment and expected output to reflect that the multiple fix-it
22075 hints are now consolidated into one insertion.
22076 (selftest::test_one_liner_many_fixits_2): New test.
22077 (selftest::test_diagnostic_show_locus_one_liner): Update for
22079 (selftest::test_fixit_consolidation): Update for fix-it API
22081 * diagnostic.c (print_parseable_fixits): Likewise.
22082 * edit-context.c (edited_line::m_line_events): Convert from
22083 auto_vec <line_event *> to auto_vec <line_event>.
22084 (class line_event): Convert from abstract base class to a concrete
22085 class, taking over the role of replace_event.
22086 (class insert_event): Delete.
22087 (class replace_event): Rename to class line_event. Convert to
22089 (edit_context::add_fixits): Reimplement.
22090 (edit_context::apply_insert): Delete.
22091 (edit_context::apply_replace): Rename to...
22092 (edit_context::apply_fixit): ...this. Convert to half-open range.
22093 (edited_file::apply_insert): Delete.
22094 (edited_file::apply_replace): Rename to...
22095 (edited_file::apply_fixit): ...this.
22096 (edited_line::~edited_line): Drop deletion of events.
22097 (edited_line::apply_insert): Delete.
22098 (edited_line::apply_replace): Rename to...
22099 (edited_line::apply_fixit): ...this. Convert to half-open range.
22100 Update for change to type of m_line_events.
22101 * edit-context.h (edit_context::apply_insert): Delete.
22102 (edit_context::apply_replace): Rename to...
22103 (edit_context::apply_fixit): ...this.
22105 2017-05-01 Martin Sebor <msebor@redhat.com>
22107 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22110 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22113 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22114 __get_cpuid_max returns 0.
22115 (__get_cpuid_count): Ditto.
22117 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22119 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22120 replacement expression is another instance of one of its arguments.
22122 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22125 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22126 check for stack push/pop autoinc.
22127 * config/i386/i386.c (ix86_agi_dependent): Return false
22128 if the only reason why modified_in_p returned true is that
22129 addr is SP based and set_insn is a push or pop.
22131 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22133 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22136 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22139 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22140 (account_size_time): Use two predicates - exec_pred and
22142 (evaluate_conditions_for_known_args): Compute both clause and
22144 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22145 (inline_summary_t::duplicate): Update.
22146 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22148 (compute_inline_parameters): Likewise.
22149 (estimate_edge_size_and_time): Update caluclation of time.
22150 (estimate_node_size_and_time): Compute both time and nonspecialized
22152 (estimate_ipcp_clone_size_and_time): Update.
22153 (inline_merge_summary): Update.
22154 (do_estimate_edge_time): Update.
22155 (do_estimate_edge_size): Update.
22156 (do_estimate_edge_hints): Update.
22157 (inline_read_section, inline_write_summary): Stream both new predicates.
22158 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22160 (compute_inlined_call_time): Cleanup.
22161 (big_speedup_p): Update.
22162 (edge_badness): Update.
22163 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22164 (size_time_entry): Replace predicate by exec_predicate and
22165 nonconst_predicate.
22166 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22167 (estimate_edge_time): Return also nonspec_time.
22168 (reset_edge_growth_cache): Update.
22170 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22172 PR rtl-optimization/80491
22173 * ifcvt.c (noce_process_if_block): When looking for x setter
22174 with missing else_bb, don't check only the insn right before
22175 cond_earliest, but look for the last insn that x is modified in
22176 within the same bb.
22178 PR rtl-optimization/80491
22179 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22181 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22183 PR tree-optimization/80487
22184 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22186 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22188 PR tree-optimization/79697
22189 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22190 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22191 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22193 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22194 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22196 2017-04-28 Martin Sebor <msebor@redhat.com>
22198 PR tree-optimization/80523
22199 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22200 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22202 (maybe_warn, format_directive, parse_directive): Use new functions.
22203 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22205 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22207 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22209 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22211 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22212 target_header_dir): Set correctly.
22213 * configure: Regenerated.
22214 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22215 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22216 instead of SYSTEM_HEADER_DIR.
22218 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22220 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22221 (estimate_local_effects): Likewise.
22222 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22223 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22224 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22225 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22226 do_estimate_edge_time, estimate_edge_time): Likewise.
22227 * ipa-inline-analysis.c (estimate_node_size_and_time,
22228 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22229 (estimate_time_after_inlining): Remove.
22231 2017-04-28 Martin Liska <mliska@suse.cz>
22233 * doc/gcov.texi: Enhance documentation of gcov.
22235 2017-04-28 Martin Liska <mliska@suse.cz>
22237 * doc/gcov.texi: Sort options in alphabetic order.
22238 * doc/gcov-dump.texi: Likewise.
22239 * doc/gcov-tool.texi: Likewise.
22240 * gcov.c (print_usage): Likewise.
22241 * gcov-dump.c (print_usage): Likewise.
22242 * gcov-tool.c (print_merge_usage_message): Likewise.
22243 (print_rewrite_usage_message): Likewise.
22244 (print_overlap_usage_message): Likewise.
22246 2017-04-28 Martin Liska <mliska@suse.cz>
22248 PR gcov-profile/53915
22249 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22251 2017-04-28 Martin Liska <mliska@suse.cz>
22253 PR gcov-profile/79891
22254 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22255 is marked by compiler as living on a line.
22256 (get_cycles_count): Remove usage of the union.
22257 (output_intermediate_file): Likewise.
22258 (find_source): Fix GNU coding style.
22259 (accumulate_line_counts): Remove old non-all block mode.
22260 (output_lines): Remove usage of the union.
22261 * profile.c (output_location): Include all BBs, even if
22262 belonging to a same line (and file) as a previous BB.
22264 2017-04-28 Martin Liska <mliska@suse.cz>
22266 * gcov.c (process_args): Handle new argument 'w'.
22267 (read_graph_file): Assign ID to BBs.
22268 (output_branch_count): Display BB # if verbose flag is set.
22269 (output_lines): Likewise for arcs.
22270 (print_usage): Add '--verbose' option help.
22271 * doc/gcov.texi: Document --verbose (-w) option.
22273 2017-04-28 Martin Liska <mliska@suse.cz>
22275 * gcov.c (struct block_location_info): New struct.
22276 (process_file): Fill up the new structure.
22277 (read_graph_file): Replace usage of encoding by the newly added
22279 (add_line_counts): Likewise.
22280 (accumulate_line_counts): Remove usage of the union.
22281 (function_info::function_info): New function.
22282 (function_info::~function_info): Likewise.
22283 (process_file): Call delete instead of release_function.
22284 (release_function): Release the function.
22285 (release_structures): Call delete instead of release_function.
22286 (solve_flow_graph): Replace usage of num_blocks.
22287 (find_exception_blocks): Likewise.
22288 (output_lines): Fix GNU coding style.
22290 2017-04-28 Martin Liska <mliska@suse.cz>
22293 * coverage.c (coverage_remove_note_file): New function.
22294 * coverage.h: Declare the function.
22295 * toplev.c (finalize): Clean if an error has been seen.
22297 2017-04-28 Martin Liska <mliska@suse.cz>
22299 PR gcov-profile/80031
22300 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22301 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22302 * gcov.c (read_graph_file): Read just number of blocks.
22303 * profile.c (branch_prob): Do not stream 0 flags per a basic
22306 2017-04-28 Martin Liska <mliska@suse.cz>
22308 * gcov-dump.c (tag_*): Add new argument to declarations.
22309 (dump_gcov_file): Likewise.
22310 (tag_blocks): Add and use new argument depth.
22311 (tag_arcs): Likewise.
22312 (tag_lines): Likewise.
22313 (tag_counters): Likewise.
22314 (tag_summary): Likewise.
22315 (dump_working_sets): Use depth to do a proper indentation.
22317 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22320 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22321 * symtab.c (symtab_node::debug_symtab): Move definition here.
22323 2017-04-28 Richard Biener <rguenther@suse.de>
22325 * lto-streamer.h (LTO_major_version): Bump to 7.
22327 2017-04-28 Richard Biener <rguenther@suse.de>
22329 * tree-vrp.c (assert_info): New struct.
22330 (add_assert_info): New helper.
22331 (register_edge_assert_for_2): Refactor to add asserts to a vector
22333 (register_edge_assert_for_1): Likewise.
22334 (register_edge_assert_for): Likewise.
22335 (finish_register_edge_assert_for): New helper actually registering
22336 asserts where live on edge.
22337 (find_conditional_asserts): Adjust.
22338 (find_switch_asserts): Likewise.
22339 (evrp_dom_walker::try_find_new_range): Generalize.
22340 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22342 2017-04-27 Marek Polacek <polacek@redhat.com>
22345 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22346 arg10 and arg11 to itype.
22348 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22350 * doc/extend.texi (Object Size Checking): Improve grammar.
22352 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22355 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22356 that the logic for permitting reciprocal estimates matches that
22359 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22362 * tree.c (type_cache_hasher::equal): Only compare
22363 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22364 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22365 non-aggregate element types.
22366 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22367 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22369 2017-04-27 Richard Biener <rguenther@suse.de>
22371 PR middle-end/80533
22372 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22373 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22374 keeping a reference to a trailing array.
22376 2017-04-27 Richard Biener <rguenther@suse.de>
22378 PR middle-end/80539
22379 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22380 being in loop-closed SSA form conservatively.
22381 (chrec_fold_multiply_poly_poly): Likewise.
22383 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22385 PR middle-end/79665
22386 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22387 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22389 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22392 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22393 (aarch64_function_arg_alignment): Return unsigned int again, but still
22394 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22395 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22396 Don't emit -Wpsabi note.
22397 (aarch64_function_arg_boundary): Likewise.
22398 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22401 2017-04-26 Nathan Sidwell <nathan@acm.org>
22403 * tree.h (crc32_unsigned_n): Declare.
22404 (crc32_unsigned, crc32_unsigned): Make inline.
22405 * tree.c (crc32_unsigned_bits): Replace with ...
22406 (crc32_unsigned_n): ... this.
22407 (crc32_unsigned, crc32_byte): Remove.
22408 (crc32_string): Remove unnecessary braces.
22410 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22412 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22413 * ipa-inline-analysis.c (MAX_TIME): Remove.
22414 (account_size_time): Use sreal for time.
22415 (dump_inline_summary): Update.
22416 (estimate_function_body_sizes): Update.
22417 (estimate_edge_size_and_time): Update.
22418 (estimate_calls_size_and_time): Update.
22419 (estimate_node_size_and_time): Update.
22420 (inline_merge_summary): Update.
22421 (inline_update_overall_summary): Update.
22422 (estimate_time_after_inlining): Update.
22423 (inline_read_section): Update.
22424 (inline_write_summary): Update.
22425 * ipa-inline.c (compute_uninlined_call_time): Update.
22426 (compute_inlined_call_time): Update.
22427 (recursive_inlining): Update.
22428 (inline_small_functions): Update.
22429 (dump_overall_stats): Update.
22430 * ipa-inline.h: Include sreal.h.
22431 (size_time_entry): Turn time to sreal.
22432 (inline_summary): Turn self_time nad time to sreal.
22434 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22436 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22438 (sreal::stream_out, sreal::stream_in): New.
22439 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22441 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22443 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22446 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22449 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22451 (dimode_scalar_chain::compute_convert_gain): Ditto.
22452 (dimode_scalar_chain::make_vector_copies): Ditto.
22453 (dimode_scalar_chain::convert_reg): Ditto.
22454 (dimode_scalar_chain::convert_insn): Ditto.
22455 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22456 (VI248_AVX512BW_1): New mode iterator.
22457 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22458 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22461 2017-04-25 Martin Sebor <msebor@redhat.com>
22463 PR tree-optimization/80497
22464 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22465 constants are representable in HOST_WIDE_INT.
22466 (parse_directive): Ditto.
22468 2017-04-25 Martin Sebor <msebor@redhat.com>
22471 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22472 (new_zero_array): Adjust signature.
22473 (dom_info::dom_init): Used unsigned rather that size_t.
22474 (dom_info::dom_info): Same.
22476 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22477 Jakub Jelinek <jakub@redhat.com>
22480 * config/arm/arm.c: Include gimple.h.
22481 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22482 returns negative, increment ncrn only if it returned positive.
22483 (arm_needs_doubleword_align): Return int instead of bool,
22484 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22485 members, but if there is any such non-FIELD_DECL
22486 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22487 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22488 returns negative, increment nregs only if it returned positive.
22489 (arm_setup_incoming_varargs): Likewise.
22490 (arm_function_arg_boundary): Emit -Wpsabi note if
22491 arm_needs_doubleword_align returns negative, return
22492 DOUBLEWORD_ALIGNMENT only if it returned positive.
22494 2017-04-25 Marek Polacek <polacek@redhat.com>
22497 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22498 first argument to type.
22500 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22504 type checks to test for compatibility instead of equality.
22506 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22507 Jakub Jelinek <jakub@redhat.com>
22510 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22512 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22513 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22514 the alignment computation, but return their maximum in warn_alignment.
22515 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22516 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22518 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22519 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22522 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22524 * config/arc/simdext.md (dmpyh): Fix typo.
22526 2017-04-25 Richard Biener <rguenther@suse.de>
22528 PR tree-optimization/80492
22529 * alias.c (compare_base_decls): Handle registers with asm
22530 specification conservatively.
22531 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22532 compare_base_decls returning dont-know properly.
22534 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22536 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22537 (legitimate_offset_address_p): New function.
22538 (arc_legitimate_address_p): Use above function.
22540 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22542 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22546 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22547 ACCH registers whenever they are available.
22549 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22551 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22552 double regs fix when not used.
22554 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22556 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22558 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22559 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22561 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22563 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22564 when emitting short ADD instructions.
22566 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22568 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22570 (cmpsi_cc_c_insn): Likewise.
22571 (cbranchsi4_scratch): Compute proper instruction length using
22572 compact_hreg_operand.
22573 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22575 2017-04-25 Richard Biener <rguenther@suse.de>
22577 PR middle-end/80509
22578 * passes.c (pass_manager::pass_manager): Initialize
22579 m_name_to_pass_map.
22581 2017-04-25 Richard Biener <rguenther@suse.de>
22583 PR tree-optimization/79201
22584 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22586 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22589 * config/s390/vector.md: Split MEM->GPR vector moves for
22590 non-s_operand addresses.
22592 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22595 * config/s390/predicates.md (reload_const_wide_int_operand): New
22597 * config/s390/s390.md ("movti"): Remove d/P alternative.
22598 ("movti_bigconst"): New pattern definition.
22600 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22603 * s390-protos.h (s390_expand_cs_hqi): Removed.
22604 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22605 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22606 modes as well as CCZ1mode and CCZmode.
22607 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22608 signature of s390_emit_compare_and_swap.
22609 (s390_expand_cs_hqi): Likewise, make static.
22610 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22611 compare-and-swap, in some cases.
22612 (s390_expand_cs): Wrapper function.
22613 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22615 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22616 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22617 patterns for small and large integers. Forbid symref memory operands.
22618 Move expander to s390.c. Require cc register.
22619 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22620 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22621 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22622 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22623 symref memory operands. Remove CC mode and call s390_match_ccmode
22625 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22627 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22629 * config/s390/s390.md (define_peephole2): New peephole to help
22630 combining the load-and-test pattern with volatile memory.
22632 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22634 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22635 with CCZmode for TARGET_Z196.
22637 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22639 PR rtl-optimization/80501
22640 * combine.c (make_compound_operation_int): Set subreg_code to SET
22641 even for AND with mask of the sign bit of mode.
22643 PR rtl-optimization/80500
22644 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22645 sum's initial value.
22647 2017-04-25 Julian Brown <julian@codesourcery.com>
22648 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22650 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22652 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22654 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22656 2017-04-25 Julian Brown <julian@codesourcery.com>
22657 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22659 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22660 (thunderx2t99_sha): New Reservation.
22662 2017-04-25 Julian Brown <julian@codesourcery.com>
22663 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22665 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22666 type for 1-element load.
22668 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22670 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22672 2017-04-24 Martin Jambor <mjambor@suse.cz>
22674 PR tree-optimization/80293
22675 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22676 char arrays not totally scalarizable if it is false.
22677 (analyze_all_variable_accesses): Pass correct value in the new
22678 parameter. Add a statistics counter.
22680 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22682 PR middle-end/79931
22683 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22685 2017-04-24 Richard Biener <rguenther@suse.de>
22687 PR tree-optimization/80494
22688 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22689 out for complex types.
22691 2017-04-24 Richard Biener <rguenther@suse.de>
22693 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22694 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22695 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22696 (DFS): Adjust and never fail.
22697 (sccvn_dom_walker::fail): Remove.
22698 (sccvn_dom_walker::before_dom_children): Adjust.
22699 (run_scc_vn): Likewise and never fail.
22700 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22701 (pass_fre::execute): Likewise.
22703 2017-04-24 Richard Biener <rguenther@suse.de>
22705 PR tree-optimization/79725
22706 * tree-ssa-sink.c (statement_sink_location): Return whether
22707 failure reason was zero uses. Move that check later.
22708 (sink_code_in_bb): Deal with zero uses by removing the stmt
22711 2017-04-24 Richard Biener <rguenther@suse.de>
22714 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22715 pointer-based references.
22717 2017-04-24 Richard Biener <rguenther@suse.de>
22720 * pass_manager.h (pass_manager::operator new): Remove.
22721 (pass_manager::operator delete): Likewise.
22722 * passes.c (pass_manager::operator new): Remove.
22723 (pass_manager::operator delete): Likewise.
22724 (pass_manager::pass_manager): Zero individual pass members.
22726 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22729 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22730 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22731 Check "XEXP (src, 1)" operand here.
22732 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22733 Check "XEXP (src, 1)" operand here.
22734 (dimode_scalar_chain::make_vector_copies): Detect count register
22735 of a shift instruction. Zero extend count register from QImode
22736 to DImode to satisfy vector shift pattern count operand predicate.
22737 Substitute vector shift count operand with a DImode copy.
22738 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22741 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22743 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22744 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22745 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22746 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22747 (UNSPEC_NOREX_MEM): Remove definition.
22749 2017-04-21 Richard Biener <rguenther@suse.de>
22751 PR tree-optimization/79547
22752 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22753 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22754 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22755 without any constraints.
22757 2017-04-21 Richard Biener <rguenther@suse.de>
22759 PR tree-optimization/78847
22760 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22762 2017-04-21 Richard Biener <rguenther@suse.de>
22764 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22765 (build_distinct_type_copy): Likewise.
22766 (build_variant_type_copy): Likewise.
22767 * tree.c (build_qualified_type): Pass down mem-stat info.
22768 (build_distinct_type_copy): Likewise.
22769 (build_variant_type_copy): Likewise.
22771 2017-04-21 Richard Biener <rguenther@suse.de>
22773 PR tree-optimization/80237
22774 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22776 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22777 for a simplified result.
22779 2016-04-21 Richard Biener <rguenther@suse.de>
22781 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22782 sth as strict as a simple_iv but a chrec without symbols and an
22783 operand defined in the loop we are peeling (and not some subloop).
22784 (propagate_constants_for_unrolling): Propagate all constants.
22786 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22789 * config/i386/i386.c (print_reg): Remove assert for disalowed
22790 regno values, call output_operand_lossage instead.
22792 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22795 * config/i386/constraints.md (Yc): New register constraint.
22796 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22797 Use Yc constraint for alternative 2 of operand 0. Remove
22798 preferred_for_speed attribute.
22800 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22802 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22803 lastprivate clauses in SIMT case.
22805 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22807 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22809 2017-04-20 Richard Biener <rguenther@suse.de>
22811 PR tree-optimization/57796
22812 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22813 as N scalar stores.
22814 (vect_model_load_cost): Cost gathers as N scalar loads.
22816 2017-04-20 Richard Biener <rguenther@suse.de>
22818 * ggc-page.c (ggc_allocated_p): Rename to ...
22819 (safe_lookup_page_table_entry): ... this and return the lookup
22821 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22823 2017-04-20 Richard Biener <rguenther@suse.de>
22825 PR tree-optimization/80453
22826 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22827 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22828 from the conditions.
22829 (vn_phi_eq): Pass them down.
22830 (vn_phi_lookup): Record them.
22831 (vn_phi_insert): Likewise.
22833 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22835 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22836 uninitialized variable warning to avoid buffer overrun.
22838 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22841 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22842 is suppressed for '{ 0 }' in C.
22844 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22846 * BASE-VER: Set to 8.0.0.
22848 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22850 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22851 priority .init_array and .fini_array section with SECTION_NOTYPE
22854 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22856 PR middle-end/80423
22857 * tree.h (build_array_type): Add typeless_storage default argument.
22858 * tree.c (type_cache_hasher::equal): Also compare
22859 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22860 (build_array_type): Add typeless_storage argument, set
22861 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22863 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22864 (build_array_type): Likewise. Add typeless_storage argument.
22866 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22867 Jakub Jelinek <jakub@redhat.com>
22869 PR tree-optimization/80426
22870 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22871 operation on symbolic operands, also compute the overflow for the
22872 invariant part when the operation degenerates into a negation.
22874 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22877 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22878 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22881 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22883 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22886 * config/avr/avr.c (tree.h): Include it.
22887 (cgraph.h): Include it.
22888 (avr_encode_section_info): Don't warn for uninitialized progmem
22889 variable if it's just an alias.
22891 2017-04-19 Richard Biener <rguenther@suse.de>
22894 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22895 when needed by AutoPGO.
22897 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22900 * doc/lto.texi: Remove an extra 'that'.
22902 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22904 PR rtl-optimization/80429
22905 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22906 are only used in debug insns.
22908 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22909 Vladimir Makarov <vmakarov@redhat.com>
22911 * config/sparc/predicates.md (input_operand): Add comment. Return
22912 true for any memory operand when LRA is in progress.
22913 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22915 2017-04-18 Jeff Law <law@redhat.com>
22918 * mips.md ({return,simple_return}_internal): Do not overwrite
22921 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22923 PR tree-optimization/80443
22924 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22925 instead of adding 1, subtract -1 and similarly instead of subtracting
22928 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22930 PR rtl-optimization/80357
22931 * haifa-sched.c (tmp_bitmap): New variable.
22932 (model_recompute): Handle duplicate use records.
22933 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22934 (free_global_sched_pressure_data): Free it.
22936 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22939 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22940 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22941 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22942 instead of SYSTEM_HEADER_DIR.
22944 2017-04-18 Jeff Law <law@redhat.com>
22946 PR middle-end/80422
22947 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22948 predecessors after walking up the insn chain.
22950 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22953 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22954 sizetype type into debug info.
22956 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22959 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22960 unneeded test for TARGET_UPPER_REGS_SF.
22961 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22963 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22966 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22967 instead of gsi_after_labels.
22969 2017-04-18 Jeff Law <law@redhat.com>
22971 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22975 2017-04-13 Jeff Law <law@redhat.com>
22976 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22977 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22979 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22982 * config/avr/avr.c (intl.h): Include it.
22983 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22985 2017-04-18 Martin Liska <mliska@suse.cz>
22987 PR gcov-profile/78783
22988 * gcov-tool.c (gcov_output_files): Validate that destination
22989 file is either removed by the tool or by a user.
22991 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22992 Guy Benyei <guybe@mellanox.com>
22994 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22995 block, and do not negate it, the stored id is already negative.
22997 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22999 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23001 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23004 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23005 masks of options that should be turned off if the VSX vector
23006 options are turned off.
23007 (OTHER_P8_VECTOR_MASKS): Likewise.
23008 (OTHER_VSX_VECTOR_MASKS): Likewise.
23009 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23010 rs6000_disable_incompatible_switches to validate no type switches
23012 (rs6000_incompatible_switch): New function to disallow turning on
23013 other vector options if -mno-vsx, -mno-power8-vector, or
23014 -mno-power9-vector are specified.
23016 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23018 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23020 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23022 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23023 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23024 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23025 (ARG_POINTER_CFA_OFFSET): Likewise.
23027 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23029 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23030 conditions to take advantage of various optimizations.
23032 2017-04-13 Jeff Law <law@redhat.com>
23034 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23035 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23036 (zero_extendsidi2_dext): Likewise.
23038 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23041 * fold-const.c (fold_ternary_loc): Revert
23042 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23045 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23047 PR rtl-optimization/80343
23048 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23049 new scratch pseudo.
23051 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23054 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23055 to ubsan_encode_value.
23057 2017-04-13 Jeff Law <law@redhat.com>
23059 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23060 appearing in DEBUG_INSNs.
23062 2017-04-13 Martin Liska <mliska@suse.cz>
23064 PR gcov-profile/80413
23065 * gcov-io.c (gcov_write_string): Copy to buffer just when
23066 allocated size is greater than zero.
23068 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23071 * dwarf2out.c (decls_for_scope): Ignore declarations of
23072 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23074 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23077 * ipa-visibility.c (non_local_p): Fix typos.
23078 (localize_node): When localizing symbol in same comdat group,
23079 dissolve the group only when we know external symbols are going
23081 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23083 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23085 PR tree-optimization/79390
23086 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23087 order does not result in usable sequence, retry with reversed operand
23093 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23094 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23095 op0 instead of fold_convert_loc (loc, type, arg0).
23097 2017-04-12 Jeff Law <law@redhat.com>
23099 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23100 has a delay slot in the generated code.
23102 * config/cris/cris.md (cris_preferred_reload_class): Return
23103 GENNONACR_REGS rather than GENERAL_REGS.
23105 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23108 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23109 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23110 signedness of the result type.
23112 2017-04-12 Richard Biener <rguenther@suse.de>
23113 Jeff Law <law@redhat.com>
23115 PR tree-optimization/80359
23116 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23117 trim stores to TARGET_MEM_REFs.
23119 2017-04-12 Richard Biener <rguenther@suse.de>
23121 PR tree-optimization/79390
23122 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23123 threading case even more.
23125 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23128 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23129 for quad_address_p for TImode, instead of just not indexed_address.
23131 2017-04-12 Richard Biener <rguenther@suse.de>
23132 Bernd Edlinger <bernd.edlinger@hotmail.de>
23134 PR middle-end/79671
23135 * alias.c (component_uses_parent_alias_set_from): Handle
23136 TYPE_TYPELESS_STORAGE.
23137 (get_alias_set): Likewise.
23138 * tree-core.h (tree_type_common): Add typeless_storage flag.
23139 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23140 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23141 for types containing members with TYPE_TYPELESS_STORAGE.
23142 (place_field): Likewise.
23143 (layout_type): Likewise for ARRAY_TYPE.
23144 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23145 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23146 TYPE_TYPELESS_STORAGE.
23147 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23149 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23152 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23153 first argument to type.
23155 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23159 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23160 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23161 (rs6000_expand_binop_builtin): Likewise.
23162 (rs6000_expand_ternop_builtin): Likewise; also add missing
23163 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23164 vshasigma built-ins.
23165 * doc/extend.texi: Document that vec_xxpermdi's third argument
23166 must be a constant.
23168 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23170 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23171 Use shift_const cost parameter when calculating gain of STV shifts.
23173 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23175 PR rtl-optimization/70478
23176 * lra-constraints.c (process_alt_operands): Check memory for
23177 disfavoring memory insn operand.
23179 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23181 PR middle-end/80100
23182 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23183 left shift in unsigned HOST_WIDE_INT type.
23185 PR rtl-optimization/80385
23186 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23187 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23190 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23191 if they have any depend clauses.
23193 2017-04-11 Martin Liska <mliska@suse.cz>
23196 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23197 * ipa-split.c (split_function): Create a local comdat symbol
23198 if caller is in a comdat group.
23200 2017-04-11 Martin Liska <mliska@suse.cz>
23203 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23206 2017-04-11 Martin Sebor <msebor@redhat.com>
23208 PR middle-end/80364
23209 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23210 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23212 (directive::set_width, directive::set_precision, format_character):
23214 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23217 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23220 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23221 conflict, set target->arch_name instead of target->cpu_name.
23223 2017-04-11 Richard Biener <rguenther@suse.de>
23225 PR tree-optimization/80374
23226 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23227 build_zero_cst, remove fold_convertible_p check again.
23229 2017-04-11 Martin Liska <mliska@suse.cz>
23232 * ubsan.c (instrument_object_size): Do not instrument register
23235 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23238 * config/i386/i386-builtin-types.def
23239 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23240 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23241 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23242 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23243 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23244 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23245 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23246 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23247 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23248 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23249 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23250 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23251 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23252 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23253 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23254 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23255 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23256 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23257 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23258 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23259 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23260 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23261 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23262 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23263 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23264 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23265 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23266 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23267 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23268 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23269 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23270 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23271 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23272 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23273 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23274 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23275 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23276 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23277 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23278 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23279 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23280 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23281 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23282 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23283 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23284 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23286 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23287 flag to second_arg_count, handle 4 argument function type _COUNT
23288 aliases, handle second_arg_count on second argument rather than last.
23290 2017-04-10 Jeff Law <law@redhat.com>
23292 PR tree-optimization/80374
23293 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23294 record anything if we can not convert integer_zero_node to the
23297 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23300 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23301 Enhance special handling given to the TARGET_P9_MINMAX option in
23302 relation to certain other options.
23304 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23306 PR tree-optimization/80153
23307 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23308 remove POINTER_PLUS_EXPR's base part directly, rather than through
23311 2017-04-10 Richard Biener <rguenther@suse.de>
23312 Bin Cheng <bin.cheng@arm.com>
23314 PR tree-optimization/80153
23315 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23316 the first element of pointer type aff_tree. Build result expr in
23318 (add_elt_to_tree): Convert to type unconditionally. Remove other
23319 fold_convert calls.
23320 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23321 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23323 2017-04-10 Richard Biener <rguenther@suse.de>
23325 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23328 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23330 PR rtl-optimization/70478
23331 * lra-constraints.c (curr_small_class_check): New.
23332 (update_and_check_small_class_inputs): New.
23333 (process_alt_operands): Update curr_small_class_check. Disfavor
23334 alternative insn memory operands. Check available regs for small
23337 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23340 * config/mips/mips.opt (-mvirt): Update description.
23341 * doc/invoke.texi (-mvirt): Likewise.
23343 2017-04-10 Richard Biener <rguenther@suse.de>
23345 PR middle-end/80362
23346 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23347 looking for NEGATE_EXPR in -A / -B to A / B folding.
23349 2017-04-10 Martin Liska <mliska@suse.cz>
23351 PR gcov-profile/80224
23352 * gcov.c (print_usage): Fix usage string.
23353 (get_gcov_intermediate_filename): Remove.
23354 (output_gcov_file): Use both for normal and intermediate format.
23355 (generate_results): Do not initialize special file for
23356 intermediate format.
23358 2017-04-10 Richard Biener <rguenther@suse.de>
23360 PR tree-optimization/80304
23361 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23364 2017-04-10 Nathan Sidwell <nathan@acm.org>
23367 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23368 (rs6000_init_builtins): Use it.
23370 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23372 * config/arm/arm.md (<mrc>): Add mode to SET source.
23373 (<mrrc>): Likewise.
23375 2017-04-10 Richard Biener <rguenther@suse.de>
23377 PR middle-end/80344
23378 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23380 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23383 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23384 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23385 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23386 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23387 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23388 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23389 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23390 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23391 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23392 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23393 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23394 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23395 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23396 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23397 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23398 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23399 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23400 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23401 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23402 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23403 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23404 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23405 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23407 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23409 PR rtl-optimization/70478
23410 * lra-constraints.c: Reverse the last patch.
23412 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23414 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23415 Add comment for WCHAR_T.
23417 2017-04-08 Martin Liska <mliska@suse.cz>
23420 2017-04-07 Martin Liska <mliska@suse.cz>
23423 * ipa-split.c (split_function): Add function part to a same comdat
23426 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23429 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23431 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23433 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23434 (vec_pairing): ...to this.
23435 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23436 (rs6000_sched_init): Adjust for name change.
23437 (struct rs6000_sched_context): Likewise.
23438 (rs6000_init_sched_context): Likewise.
23439 (rs6000_set_sched_context): Likewise.
23441 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23447 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23449 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23450 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23451 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23453 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23455 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23457 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23459 PR rtl-optimization/70703
23460 * ira-color.c (update_conflict_hard_regno_costs): Use
23461 int64_t instead of HOST_WIDE_INT.
23463 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23465 PR rtl-optimization/70478
23466 * lra-constraints.c (process_alt_operands): Disfavor alternative
23467 insn memory operands.
23469 2017-04-07 Jeff Law <law@redhat.com>
23471 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23472 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23474 2017-04-07 Martin Liska <mliska@suse.cz>
23477 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23478 Show error message instead of an ICE.
23480 2017-04-07 Martin Liska <mliska@suse.cz>
23483 * ipa-split.c (split_function): Add function part to a same comdat
23486 2017-04-07 Richard Biener <rguenther@suse.de>
23488 PR middle-end/80341
23489 * tree.c (get_unwidened): Also handle ! for_type case for
23491 * convert.c (do_narrow): Split out from ...
23492 (convert_to_integer_1): ... here. Do not pass final truncation
23493 type to get_unwidened for TRUNC_DIV_EXPR.
23495 2017-04-07 Richard Biener <rguenther@suse.de>
23497 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23499 (aff_combination_const): Adjust.
23500 (aff_combination_scale): Likewise.
23501 (aff_combination_add_elt): Likewise.
23502 (aff_combination_add_cst): Likewise.
23503 (aff_combination_convert): Likewise.
23504 (add_elt_to_tree): Likewise. Remove unused argument.
23505 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23507 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23509 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23511 * config/arm/arm.c (arm_default_short_enums): Use
23512 ARM_DEFAULT_SHORT_ENUMS.
23513 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23515 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23518 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23519 members with redundant out-of-class redeclaration.
23521 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23524 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23525 * config/i386/i386.md (*zero_extendsidi2):
23526 Add (?*x,*x) and (?*v,*v) alternatives.
23528 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23531 * config/i386/i386.c (ix86_expand_builtin)
23532 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23533 mode from insn data. Convert operands to insn operand mode.
23534 Copy operands that don't satisfy insn predicate to a register.
23536 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23538 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23541 2017-04-06 Richard Biener <rguenther@suse.de>
23543 PR tree-optimization/80334
23544 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23545 preserve alignment of accesses.
23547 2017-04-06 Richard Biener <rguenther@suse.de>
23549 PR tree-optimization/80262
23550 * tree-sra.c (build_ref_for_offset): Preserve address-space
23552 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23553 Drop useless address-space information on MEM_REF offsets.
23555 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23557 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23559 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23561 PR rtl-optimization/70703
23562 * ira-color.c (update_conflict_hard_regno_costs): Use
23563 HOST_WIDE_INT instead of long.
23565 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23568 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23569 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23571 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23572 for x86_64 target. Handle -m3dnowa option.
23574 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23576 PR rtl-optimization/70703
23577 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23578 (update_conflict_hard_regno_costs): Use long instead of unsigned
23579 arithmetic for cost calculation.
23581 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23582 Bernd Edlinger <bernd.edlinger@hotmail.de>
23585 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23588 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23591 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23592 ptr_mode with Pmode throughout.
23593 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23594 into probe_stack_range and use DImode.
23596 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23599 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23600 call_eh_return is true.
23602 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23604 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23605 Initialize last_match_fntype_index.
23607 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23610 * tree-nvr.c: Include internal-fn.h.
23611 (pass_return_slot::execute): Ignore internal calls without
23614 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23615 Richard Biener <rguenther@suse.de>
23618 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23619 captures used multiple times, except for the last use.
23620 * generic-match-head.c: Include gimplify.h.
23622 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23624 PR tree-optimization/79390
23625 * target.h (struct noce_if_info): Declare.
23626 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23627 * target.def (noce_conversion_profitable_p): New target hook.
23628 * ifcvt.h (struct noce_if_info): New type, moved from ...
23629 * ifcvt.c (struct noce_if_info): ... here.
23630 (noce_conversion_profitable_p): Renamed to ...
23631 (default_noce_conversion_profitable_p): ... this. No longer
23633 (noce_try_store_flag_constants, noce_try_addcc,
23634 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23635 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23636 instead of noce_conversion_profitable_p.
23637 * config/i386/i386.c: Include ifcvt.h.
23638 (ix86_option_override_internal): Don't override
23639 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23640 (ix86_noce_conversion_profitable_p): New function.
23641 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23642 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23643 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23644 * doc/tm.texi: Regenerated.
23646 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23648 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23651 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23654 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23655 instructions for small multiply cores.
23657 2017-04-04 Jeff Law <law@redhat.com>
23659 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23661 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23662 that are not set by the loop over the elements.
23664 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23667 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23668 int mode, convert_modes it to mode as unsigned, otherwise use
23669 lowpart_subreg to mode rather than SImode.
23670 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23671 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23672 Use DImode instead of SImode for the shift count operand.
23673 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23676 2017-04-04 Richard Biener <rguenther@suse.de>
23678 PR middle-end/80281
23679 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23680 arithmetic done for the negate or the plus. Simplify.
23681 (A - (-B) -> A + B): Likewise.
23682 * fold-const.c (split_tree): Make sure to not negate pointers.
23684 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23686 PR rtl-optimization/60818
23687 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23688 a compare of comparisons with the thing compared if this results
23689 in a different machine mode.
23691 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23693 * alias.c (base_alias_check): Fix typo in comment.
23694 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23695 * cgraphunit.c (symbol_table::compile): Likewise.
23696 * collect2.c (maybe_run_lto_and_relink): Likewise.
23697 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23698 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23699 * config/avr/avr.c (avr_map_op_t): Likewise.
23700 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23701 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23702 * config/epiphany/epiphany.md (movcc): Likewise.
23703 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23704 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23706 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23707 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23708 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23709 * config/sh/sh.c (sh_rtx_costs): Likewise.
23710 * fold-const.c (fold_truth_andor): Likewise.
23711 * genautomata.c (collapse_flag): Likewise.
23712 * gengtype.h (struct type::u::s): Likewise.
23713 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23714 * input.c (FORMAT_AMOUNT): Likewise.
23715 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23716 (known_aggs_to_agg_replacement_list): Likewise.
23717 * ipa-inline-analysis.c: Likewise.
23718 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23719 * ipa-polymorphic-call.c
23720 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23721 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23722 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23724 * modulo-sched.c (apply_reg_moves): Likewise.
23725 * omp-expand.c (build_omp_regions_1): Likewise.
23726 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23727 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23728 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23729 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23730 * value-prof.c: Likewise.
23731 * var-tracking.c (val_reset): Likewise.
23733 2017-04-03 Richard Biener <rguenther@suse.de>
23735 PR tree-optimization/80275
23736 * fold-const.c (split_address_to_core_and_offset): Handle
23739 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23741 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23742 descriptors is at least equal to that of functions.
23744 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23746 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23748 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23751 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23752 (mov<IMOD4:mode>): New expander.
23753 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23755 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23757 PR rtl-optimization/79405
23758 * fwprop.c (propagations_left): New variable.
23759 (forward_propagate_into): Decrement it.
23760 (fwprop_init): Initialize it.
23761 (fw_prop): If the variable has reached zero, stop propagating.
23762 (fwprop_addr): Ditto.
23764 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23767 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23768 a FUNCTION_DECL, pass it as decl instead of origin to
23771 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23773 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23776 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23779 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23780 TARGET_VSX_SMALL_INTEGER.
23782 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23784 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23785 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23787 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23789 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23790 extraction from odd-numbered MSA register.
23792 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23794 PR middle-end/80173
23795 * expmed.c (store_bit_field_1): Don't attempt to create
23796 a word subreg out of hard registers wider than word if they
23797 have HARD_REGNO_NREGS of 1 for their mode.
23799 PR middle-end/80163
23800 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23801 conversions to integer types wider than word and pointer.
23804 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23805 (rtx_equal_for_cselib_p): Pass 0 to it.
23806 * cselib.c (cselib_hasher::equal): Likewise.
23807 (rtx_equal_for_cselib_1): Add depth argument. If depth
23808 is 128, don't look up VALUE locs and punt. Increment
23809 depth in recursive calls when walking VALUE locs.
23811 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23813 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23814 (make_gcov_file_name): Use the canonical path name for generating
23816 (read_line): Fix handling of files with ascii null bytes.
23818 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23820 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23821 to initialise a vector register instead
23822 of using a const_int.
23824 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23826 PR translation/80189
23827 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23828 diagnostic messages.
23830 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23833 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23834 (dfp_diex_<mode>): Update mode of operand 1.
23835 * doc/extend.texi (dxex, dxexq): Document change to return type.
23836 (diex, diexq): Document change to argument type.
23838 2017-03-30 Martin Jambor <mjambor@suse.cz>
23841 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23842 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23843 it reflects the signature changes performed at the callee side.
23844 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23845 to cgraph_build_function_type_skip_args.
23846 (build_function_decl_skip_args): Adjust call to the above function.
23848 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23851 * config/i386/sse.md
23852 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23853 register as dest whenever it is a MEM not rtx_equal_p to the
23854 corresponding dup operand, and when forcing into reg move the
23855 reg into the memory afterwards.
23856 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23857 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23858 for the force_reg mode.
23859 (avx512vl_vextractf128<mode>): Use register as dest either
23860 always when a MEM, or when it is a MEM not rtx_equal_p to the
23861 corresponding dup operand, or even not when it is a CONST_VECTOR
23862 depending on the mode and lo vs. hi.
23863 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23865 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23866 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23867 Likewise. Require that operands[2] is even.
23868 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23869 Remove extraneous parens. Require that operands[2] is a multiple
23871 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23872 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23873 disallow memory then.
23875 2017-03-30 Richard Biener <rguenther@suse.de>
23877 PR tree-optimization/77498
23878 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23879 to non-constants over backedges.
23881 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23883 PR rtl-optimization/80233
23884 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23885 as last_combined_insn. Do not test for BARRIER_P separately.
23887 2017-03-29 Andreas Schwab <schwab@suse.de>
23890 * calls.c (prepare_call_address): Convert funexp to Pmode before
23891 copying to temp reg.
23893 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23895 PR tree-optimization/80158
23896 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23897 Handle possible future case of more than one alternate
23899 (replace_rhs_if_not_dup): Likewise.
23900 (replace_one_candidate): Likewise.
23902 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23904 PR rtl-optimization/80193
23905 * ira.c (ira): Do not check allocation for LRA.
23907 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23909 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23910 (nvptx_output_simt_exit): Declare.
23911 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23912 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23913 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23914 (nvptx_declare_function_name): ...here. Emit declaration of local
23915 memory space buffer for omp_simt_enter insn.
23916 (nvptx_output_unisimt_switch): New.
23917 (nvptx_output_softstack_switch): New.
23918 (nvptx_output_simt_enter): New.
23919 (nvptx_output_simt_exit): New.
23920 * config/nvptx/nvptx.h (struct machine_function): New fields
23921 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23922 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23923 (UNSPECV_SIMT_EXIT): Ditto.
23924 (omp_simt_enter_insn): New insn.
23925 (omp_simt_enter): New expansion.
23926 (omp_simt_exit): New insn.
23927 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23929 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23930 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23931 (expand_GOMP_SIMT_EXIT): New.
23932 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23933 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23934 (GOMP_SIMT_EXIT): Ditto.
23935 * target-insns.def (omp_simt_enter): New insn.
23936 (omp_simt_exit): Ditto.
23937 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23939 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23940 (lower_rec_input_clauses): Likewise.
23941 (lower_lastprivate_clauses): Handle SIMT privatization.
23943 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23944 (ompdevlow_adjust_simt_enter): New.
23945 (find_simtpriv_var_op): New.
23946 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23947 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23949 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23950 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23951 (copy_decl_for_dup_finish): Ditto.
23953 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23955 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23958 * config/i386/i386.c (ix86_option_override_internal): Always
23959 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23961 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23963 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23965 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23967 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23968 mark new edge's irreducible flag accordign to it.
23969 (vect_do_peeling): Check loop preheader edge's irreducible flag
23970 and pass it to function slpeel_add_loop_guard.
23972 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23974 PR tree-optimization/80218
23975 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23976 Update block frequencies and counts.
23978 2017-03-28 Richard Biener <rguenther@suse.de>
23980 PR tree-optimization/78644
23981 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23982 of a simplification result we may not use it at all.
23984 2017-03-28 Richard Biener <rguenther@suse.de>
23987 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23988 without arguments, generate default definition of a SSA name.
23990 2017-03-28 Richard Biener <rguenther@suse.de>
23992 PR middle-end/80222
23993 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23994 TYPE_REF_CAN_ALIAS_ALL references.
23995 * fold-const.c (fold_indirect_ref_1): Likewise.
23997 2017-03-28 Martin Liska <mliska@suse.cz>
24000 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24001 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24003 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24004 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24006 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24007 (EXTRA_SPECS): Define.
24008 (SUBTARGET_EXTRA_SPECS): Likewise.
24009 (SUBTARGET_CPP_SPEC): Likewise.
24010 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24011 SUBTARGET_EXTRA_SPECS.
24012 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24014 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24016 * config/arc/simdext.md (vst64_insn): Update pattern.
24017 (vld32wh_insn): Likewise.
24018 (vld32wl_insn): Likewise.
24019 (vld64_insn): Likewise.
24020 (vld32_insn): Likewise.
24022 2017-03-28 Marek Polacek <polacek@redhat.com>
24025 * fold-const.c (fold_comparison): Use protected_set_expr_location
24026 instead of SET_EXPR_LOCATION.
24028 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24030 * tree.c (add_expr): Avoid name lookup warning.
24032 2017-03-27 Jeff Law <law@redhat.com>
24034 PR tree-optimization/80216
24035 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24036 function name. Limit recursion depth.
24037 (record_temporary_equivalences): Corresponding changes.
24039 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24041 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24044 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24047 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24049 * cfgcleanup.c (reg_note_cfa_p): New array.
24050 (insns_have_identical_cfa_notes): New function.
24051 (old_insns_match_p): Don't cross-jump in between /f
24052 and non-/f instructions. If both i1 and i2 are frame related,
24053 verify all CFA notes, their order and content.
24055 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24058 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24059 HImode and SImode with zero extend to DImode to one insn.
24060 (bswap<mode>2_extenddi): Likewise.
24061 (bswapsi2_extenddi): Likewise.
24062 (bswaphi2_extendsi): Likewise.
24063 (bswaphi2): Combine bswap HImode and SImode into one insn.
24064 Separate memory insns from swapping register.
24065 (bswapsi2): Likewise.
24066 (bswap<mode>2): Likewise.
24067 (bswaphi2_internal): Delete, no longer used.
24068 (bswapsi2_internal): Likewise.
24069 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24070 store, and gpr<-gpr swap insns.
24071 (bswap<mode>2_store): Likewise.
24072 (bswaphi2_reg): Register only splitter, combine with the splitter.
24073 (bswaphi2 splitter): Likewise.
24074 (bswapsi2_reg): Likewise.
24075 (bswapsi2 splitter): Likewise.
24076 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24077 the insns into load, store, and register/register insns.
24078 (bswapdi2_ldbrx): Likewise.
24079 (bswapdi2_load): Likewise.
24080 (bswapdi2_store): Likewise.
24081 (bswapdi2_reg): Likewise.
24083 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24085 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24086 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24088 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24091 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24094 special handling for target option conflicts between dform
24095 options (-mpower9-dform, -mpower9-dform-vector,
24096 -mpower9-dform-scalar) and -mno-direct-move.
24098 2017-03-27 Richard Biener <rguenther@suse.de>
24100 PR tree-optimization/80181
24101 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24103 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24105 * config/arc/predicates.md (move_double_src_operand): Replace the
24106 call to move_double_src_operand with a call to address_operand.
24108 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24110 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24111 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24112 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24114 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24116 * config/arc/predicates.md (long_immediate_loadstore_operand):
24117 Consider scaled addresses cases.
24119 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24121 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24122 restored when in interrupt.
24123 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24124 doesn't have delay slot.
24126 2017-03-27 Richard Biener <rguenther@suse.de>
24129 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24130 inlined thunk clones.
24132 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24135 * asan.c (instrument_derefs): Copy over last operand from
24136 original COMPONENT_REF to the new COMPONENT_REF with
24137 DECL_BIT_FIELD_REPRESENTATIVE.
24138 * ubsan.c (instrument_object_size): Likewise.
24140 2017-03-27 Richard Biener <rguenther@suse.de>
24142 PR tree-optimization/80170
24143 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24144 sure DR/SCEV didnt fold in constants we do not see when looking
24145 at the reference base alignment.
24147 2017-03-27 Richard Biener <rguenther@suse.de>
24149 PR middle-end/80171
24150 * gimple-fold.c (fold_ctor_reference): Properly guard against
24151 NULL return value from canonicalize_constructor_val.
24153 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24156 * config/i386/i386.c (ix86_expand_builtin)
24157 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24158 flags reg setting and flags reg using instructions.
24159 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24160 clobbering instructions to zero extend op2.
24162 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24164 * doc/install.texi (Configuration) <--with-aix-soname>:
24165 Update link to AIX ld.
24167 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24169 PR rtl-optimization/80160
24170 PR rtl-optimization/80159
24171 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24172 reg_alternate_class into account.
24174 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24177 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24178 to consider in curr_insn_transform.
24180 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24182 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24183 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24184 and emit_mode_inner.
24186 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24188 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24189 argument to the overloaded builtin variants. Use the new flag to
24190 deprecate certain builtin variants.
24191 * config/s390/s390-builtin-types.def: Add new builtin types.
24192 * config/s390/s390-builtins.h: Support new flags field for
24193 overloaded builtins.
24194 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24195 (s390_macro_to_expand): Enable vector float data type.
24196 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24197 builtins by incrementing the __VEC__ version number.
24198 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24200 (s390_resolve_overloaded_builtin): Emit error messages depending
24201 on the builtin flags.
24202 * config/s390/s390.c (s390_expand_builtin): Support additional
24203 flags argument. Change error message to match the messages
24204 emitted in s390-c.c.
24205 * config/s390/s390.md: New UNSPEC_* constants.
24206 (op_type): Add new instruction types.
24207 * config/s390/vecintrin.h: Add new builtins and test data class
24209 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24210 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24211 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24212 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24213 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24214 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24216 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24217 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24218 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24219 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24221 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24222 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24223 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24224 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24226 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24227 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24228 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24229 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24230 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24231 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24232 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24234 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24235 ("vec_scatter_element<V_HW_4:mode>_DI")
24236 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24237 ("vec_fpint<mode>", "vflls")
24238 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24239 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24240 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24241 ("*vec_cmphe<mode>_cc"): ... these.
24243 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24244 mode constant instead of magic value.
24246 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24248 * config/s390/s390.c (s390_expand_vec_compare): Support other
24249 vector floating point modes than just V2DF.
24250 (s390_expand_vcond): Likewise.
24251 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24252 (s390_cannot_change_mode_class): Prevent mode changes between TF
24253 and V1TF in vector registers.
24254 * config/s390/s390.md (DF, SF): New mode attributes.
24255 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24256 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24257 SFmode support for VRs.
24258 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24260 (VFT, VF_HW): New mode iterators.
24261 (vw, sdx): New mode attributes.
24262 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24263 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24264 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24265 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24266 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24267 also the new vector floating point modes. Renaming to ...
24269 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24270 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24271 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24272 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24273 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24274 ("vec_unordered<mode>"): ... these.
24276 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24277 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24278 ("*vec_extendv2df"): New insn definitions.
24280 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24282 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24283 ("mulditi3_2", "*muldi3_sign"): New patterns.
24284 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24285 rename the pattern definition.
24287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24289 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24291 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24293 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24295 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24296 instruction if possible.
24297 * config/s390/vector.md (vec_halfnumelts): New mode
24299 ("*vec_vllezlf<mode>"): New pattern.
24301 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24303 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24304 ("popcountv4si2", "popcountv2di2"): Rename to ...
24305 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24306 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24308 ("popcount<mode>2_vxe"): New pattern.
24310 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24312 * common/config/s390/s390-common.c (processor_flags_table): Add
24314 * config.gcc: Add arch12.
24315 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24316 Default to arch12 for unknown CPU model numbers.
24317 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24318 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24319 PROCESSOR_max sanity check.
24320 * config/s390/s390-opts.h (enum processor_type): Add
24322 * config/s390/s390.c (processor_table): Add arch12.
24323 (s390_expand_builtin): Add check for B_VXE flag.
24324 (s390_issue_rate): Add PROCESSOR_ARCH12.
24325 (s390_get_sched_attrmask): Likewise.
24326 (s390_get_unit_mask): Likewise.
24327 (s390_sched_score): Enable z13 scheduling for arch12.
24328 (s390_sched_reorder): Likewise.
24329 (s390_sched_variable_issue): Likewise.
24330 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24332 (s390_tune_attr): Use z13 scheduling also for arch12.
24333 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24334 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24335 (TARGET_VXE_P): New macros.
24336 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24337 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24338 * config/s390/s390.opt: Add arch12 as processor_type.
24340 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24342 * config/s390/s390.md
24343 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24344 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24345 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24347 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24348 Rename expanders to ...
24350 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24351 ("fixuns_truncdddi2_emu"): ... these.
24353 ("fixuns_trunc<mode>si2_emu"): New expander.
24355 ("*fixuns_truncdfdi2_z13"): Rename to ...
24356 ("*fixuns_truncdfdi2_vx"): ... this.
24358 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24360 * config/s390/2964.md: Remove the single element vector compare
24361 instructions which are no longer used.
24362 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24364 (s390_canonicalize_comparison): Remove handling of DFmode
24366 (s390_expand_vec_compare_scalar): Remove function.
24367 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24368 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24370 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24372 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24374 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24376 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24377 will anyway by matched by mov<mode>_64dfp.
24379 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24381 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24382 vlef/vstef. Add missing operand to vleif.
24384 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24386 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24387 pair for all vector types with 64 bit elements.
24388 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24389 * config/s390/vector.md (V_HW_64): ... here.
24390 (V_128_NOSINGLE): New mode iterator.
24391 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24392 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24393 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24394 ("*vec_load_pairv2di"): Change to ...
24395 ("*vec_load_pair<mode>"): ... this one.
24397 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24399 * config/s390/constraints.md: Add comments.
24400 (jKK): Reject element sizes > 8 bytes.
24401 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24403 * config/s390/s390.md: Add the s_operand checks formerly in
24404 s390_split_ok_p to various splitters where they are still
24406 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24407 for 128 bit vectors. Plus two splitters.
24409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24411 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24414 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24417 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24418 error if the boundary argument is not constant.
24420 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24422 PR rtl-optimization/80112
24423 * loop-doloop.c (doloop_condition_get): Don't check condition
24424 if cmp isn't SET with IF_THEN_ELSE src.
24426 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24428 PR tree-optimization/80158
24429 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24430 replacing a candidate statement, also replace it for the
24431 candidate's alternate interpretation.
24432 (replace_rhs_if_not_dup): Likewise.
24433 (replace_one_candidate): Likewise.
24435 2017-03-24 Richard Biener <rguenther@suse.de>
24437 PR tree-optimization/80167
24438 * graphite-isl-ast-to-gimple.c
24439 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24441 (translate_isl_ast_to_gimple::get_rename): Likewise.
24443 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24445 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24446 handling of certain combinations of target options, including the
24447 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24448 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24450 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24453 * config/arm/arm.md (*load_multiple): Add reload_completed to
24454 matching condition.
24456 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24457 Richard Biener <rguenth@suse.de>
24459 PR tree-optimization/79908
24460 PR tree-optimization/80136
24461 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24462 been cast away, gimplify_and_add suffices.
24464 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24466 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24468 2017-03-23 Richard Biener <rguenther@suse.de>
24470 PR tree-optimization/80032
24471 * gimplify.c (gimple_push_cleanup): Forced unconditional
24472 cleanups still have to go to the conditional_cleanups
24475 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24477 PR tree-optimization/80072
24478 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24480 (next_operand_entry_id): Change type to unsigned int.
24481 (sort_by_operand_rank): Make sure to return the right return value
24482 even if unsigned fields are bigger than INT_MAX.
24483 (struct oecount): Change cnt and id type to unsigned int.
24484 (oecount_hasher::equal): Formatting fix.
24485 (oecount_cmp): Make sure to return the right return value
24486 even if unsigned fields are bigger than INT_MAX.
24487 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24490 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24491 TREE_READONLY on result if writing it more than once.
24494 * doc/invoke.texi (-fsanitize=thread): Document that with
24495 -fnon-call-exceptions atomics are not able to throw
24499 * tsan.c: Include tree-eh.h.
24500 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24501 maybe_clean_or_replace_eh_stmt where needed.
24502 (instrument_memory_accesses): Add cfg_changed argument.
24503 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24504 if it returned true.
24505 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24507 PR rtl-optimization/63191
24508 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24509 wrapper function, moved the whole old content into ...
24510 (ix86_delegitimize_address_1): ... this. New inline function.
24511 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24512 true as last argument instead of ix86_delegitimize_address.
24514 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24516 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24517 cortexa57_branch_cost.
24519 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24521 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24523 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24526 * doc/md.texi (Constraints): Document wA constraint.
24527 * config/rs6000/constraints.md (wA): New.
24528 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24529 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24530 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24531 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24533 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24536 * gimplify.c (is_oacc_declared): New function.
24537 (oacc_default_clause): Use it to set default flags for acc declared
24538 variables inside parallel regions.
24539 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24540 declared variables.
24541 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24542 declare attribute to any decl as necessary.
24544 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24547 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24548 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24549 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24550 (arm_arch_lpae): This.
24551 * config/arm/arm.c (arm_arch7ve): Rename into ...
24552 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24553 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24556 2017-03-22 Martin Liska <mliska@suse.cz>
24559 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24560 error message instead of an ICE.
24562 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24564 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24566 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24568 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24570 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24573 2017-03-21 Martin Sebor <msebor@redhat.com>
24575 * doc/extend.texi: Use "cannot" instead of "can't."
24576 * doc/hostconfig.texi: Same.
24577 * doc/install.texi: Same.
24578 * doc/invoke.texi: Same.
24579 * doc/loop.texi: Same.
24580 * doc/md.texi: Same.
24581 * doc/objc.texi: Same.
24582 * doc/rtl.texi: Same.
24583 * doc/tm.texi: Same.
24584 * doc/tm.texi.in: Same.
24585 * doc/trouble.texi: Same.
24587 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24590 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24591 (collect_checksum_attributes): Set it.
24592 (die_checksum_ordered): Use it.
24594 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24596 PR tree-optimization/79908
24597 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24598 change: For a VA_ARG whose LHS has been cast away, use
24599 force_gimple_operand to construct the side effects.
24601 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24603 PR translation/80001
24604 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24605 more amenable to translation.
24606 (oacc_loop_auto_partitions): Likewise.
24608 2017-03-21 Marek Polacek <polacek@redhat.com>
24609 Martin Sebor <msebor@redhat.com>
24611 PR tree-optimization/80109
24612 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24613 on INTEGRAL_TYPE_P.
24615 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24616 Segher Boessenkool <segher@kernel.crashing.org>
24619 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24620 check reg_used_between_p between insn and one of succ or succ2
24621 depending on if succ is artificial insn not inserted into insn
24624 2017-03-21 Martin Liska <mliska@suse.cz>
24626 PR gcov-profile/80081
24627 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24628 * doc/gcc.texi: Include gcov-dump stuff.
24629 * doc/gcov-dump.texi: New file.
24631 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24633 PR rtl-optimization/79150
24634 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24635 conditional jump, if the jump is the last insn of the loop.
24637 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24638 Richard Biener <rguenth@suse.de>
24640 PR tree-optimization/79908
24641 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24642 been cast away, use force_gimple_operand to construct the side
24645 2017-03-21 Martin Liska <mliska@suse.cz>
24647 PR libfortran/79956
24648 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24651 2017-03-21 Brad Spengler <spender@grsecurity.net>
24654 * plugin.c (htab_hash_plugin): New function.
24655 (add_new_plugin): Use it and adjust.
24656 (parse_plugin_arg_opt): Adjust.
24657 (init_one_plugin): Likewise.
24659 2017-03-21 Richard Biener <rguenther@suse.de>
24661 PR tree-optimization/80032
24662 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24663 if set force the cleanup to happen unconditionally.
24664 (gimplify_target_expr): Push inserted clobbers with force_uncond
24665 to avoid them being removed by control-dependent DCE.
24667 2017-03-21 Richard Biener <rguenther@suse.de>
24669 PR tree-optimization/80122
24670 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24671 va_arg_pack_len when the inlined call stmt requires pack
24673 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24675 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24678 * tsan.c (instrument_builtin_call): If the memory model argument
24679 is not a constant, assume it is valid.
24682 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24685 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24687 PR rtl-optimization/79910
24688 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24689 if its dest is used by an insn before I2 (other than the combined
24690 insns themselves, which are properly handled already).
24692 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24695 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24697 * combine.c (record_used_regs): New static function.
24698 (try_combine): Handle situations where there is an additional
24699 instruction between I2 and I3 which needs to have a LOG_LINK
24703 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24705 * combine.c (try_combine): Delete redundant i1 test. Call
24706 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24708 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24711 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24712 alternatives 13/14.
24714 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24716 PR tree-optimization/80054
24717 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24718 the optimization if a PHI or any of its arguments is not dominated
24719 by the candidate's basis. Use gphi* rather than gimple* as
24721 (replace_profitable_candidates): Clean up a gimple* variable that
24722 should be a gphi* variable.
24724 2017-03-20 Martin Sebor <msebor@redhat.com>
24727 * doc/extend.texi (attribute constructor): Document present limitation.
24729 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24732 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24733 __POWER9_VECTOR__ #ifdef control, change template definition to
24734 use Power9-specific built-in function.
24735 (vec_any_eq): Likewise.
24736 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24737 to control outcomes from this test.
24738 (vector_ae_<mode>p): For VEC_F modes, likewise.
24740 2017-03-20 Ian Lance Taylor <iant@google.com>
24742 * config/i386/i386.c (ix86_function_regparm): Save an extra
24743 register for -fsplit-stack with DECL_STATIC_CHAIN.
24745 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24748 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24749 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24751 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24753 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24755 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24758 2017-03-20 Marek Polacek <polacek@redhat.com>
24761 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24763 2017-03-20 Richard Biener <rguenther@suse.de>
24765 PR tree-optimization/80113
24766 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24767 allocate extra SSA name for PHI def.
24768 (add_close_phis_to_outer_loops): Likewise.
24769 (add_close_phis_to_merge_points): Likewise.
24770 (copy_loop_close_phi_args): Likewise.
24771 (copy_cond_phi_nodes): Likewise.
24773 2017-03-20 Martin Liska <mliska@suse.cz>
24775 PR middle-end/79753
24776 * tree-chkp.c (chkp_build_returned_bound): Do not build
24777 returned bounds for a LHS that's not a BOUNDED_P type.
24779 2017-03-20 Martin Liska <mliska@suse.cz>
24783 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24784 COMPLEX_CST and VECTOR_CST.
24786 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24789 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24790 target operand. A new splitter adds the clobber statement in case
24791 the target operand is dead anyway.
24793 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24795 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24796 to age-old versions of binutils and glibc.
24798 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24800 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24802 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24804 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24806 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24808 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24809 requirement for binutils 2.13.
24811 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24813 * combine.c (try_combine): Delete redundant i1 test. Call
24814 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24816 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24818 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24819 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24821 <riscv64-*-elf>: Re-arrange section
24822 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24823 <riscv32-*-linux>: Likewise.
24824 <riscv64-*-elf>: Likewise
24825 <riscv64-*-linux>: Likewise.
24827 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24830 * aarch64.opt(verbose-cost-dump): Fix typo.
24832 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24835 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24836 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24838 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24840 * reload.c (find_reloads): When reloading a nonoffsettable address,
24841 use RELOAD_OTHER for it and its address reloads.
24843 PR rtl-optimization/79910
24844 * combine.c (record_used_regs): New static function.
24845 (try_combine): Handle situations where there is an additional
24846 instruction between I2 and I3 which needs to have a LOG_LINK
24849 2017-03-17 Jeff Law <law@redhat.com>
24851 PR tree-optimization/71437
24852 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24853 conditional in the hash table first.
24854 (vrp_dom_walker::before_dom_children): Extract condition from
24855 ASSERT_EXPR. Record condition, its inverion and any implied
24856 conditions as well.
24858 2017-03-17 Marek Polacek <polacek@redhat.com>
24859 Markus Trippelsdorf <markus@trippelsdorf.de>
24861 PR tree-optimization/80079
24862 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24865 2017-03-17 Richard Biener <rguenther@suse.de>
24867 PR middle-end/80075
24868 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24869 Properly verify the LHS before the RHS possibly claims to be
24871 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24874 2017-03-17 Martin Jambor <mjambor@suse.cz>
24876 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24877 (List of -O2 options): Likewise.
24878 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24881 2017-03-17 Tom de Vries <tom@codesourcery.com>
24883 * gcov-dump.c (print_usage): Print bug_report_url.
24885 2017-03-17 Richard Biener <rguenther@suse.de>
24887 PR middle-end/80050
24888 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24889 (parser::peek): Likewise.
24891 2017-03-17 Richard Biener <rguenther@suse.de>
24893 PR tree-optimization/80048
24894 * sese.c (free_sese_info): Properly release rename_map and
24895 copied_bb_map elements.
24897 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24899 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24900 Add linked-list forward and backlinks. Insert on
24901 construction, remove on destruction.
24902 (class pass_store_merging): Add m_stores_head field.
24903 (pass_store_merging::terminate_and_process_all_chains):
24904 Iterate over m_stores_head list.
24905 (pass_store_merging::terminate_all_aliasing_chains):
24907 (pass_store_merging::execute): Check for debug stmts first.
24908 Push new chains onto the m_stores_head stack.
24910 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24913 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24914 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24915 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24917 2017-03-16 Jeff Law <law@redhat.com>
24919 PR tree-optimization/71437
24920 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24921 member function. Implementation moved into after_dom_children
24922 member function and into the threader's thread_outgoing_edges
24924 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24925 some code into new thread_outgoing_edges.
24926 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24927 definition. Simplify marker handling (do it here). Assume we always
24928 have the available expression and the const/copies tables.
24929 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24931 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24932 * tree-vrp.c (equiv_stack): No longer file scoped.
24933 (vrp_dom_walker): New class.
24934 (vrp_dom_walker::before_dom_children): New member function.
24935 (vrp_dom_walker::after_dom_children): Likewise.
24936 (identify_jump_threads): Setup domwalker. Use it rather than
24937 walking edges in a random order by hand. Simplify setup/finalization.
24938 (finalize_jump_threads): Remove.
24939 (vrp_finalize): Do not call identify_jump_threads here.
24940 (execute_vrp): Do it here instead and call thread_through_all_blocks
24943 PR tree-optimization/71437
24944 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24946 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24948 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24949 (dom_opt_dom_walker::thread_across_edge): Remove
24950 handle_dominating_asserts argument. All callers changed.
24951 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24952 changes. Remove calls to lhs_of_dominating_assert. Other
24953 uses of handle_dominating_asserts turn into unconditional code
24954 (simplify_control_stmt_condition_1): Likewise.
24955 (simplify_control_stmt_condition): Likewise.
24956 (thread_through_normal_block, thread_across_edge): Likewise.
24957 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24958 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24959 object if it is not an SSA_NAME.
24960 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24961 before calling into the VRP specific simplifiers.
24962 (identify_jump_threads): Remove handle_dominating_asserts
24965 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24968 * tree-diagnostic.c (default_tree_printer): No longer static.
24969 * tree-diagnostic.h (default_tree_printer): New prototype.
24971 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24973 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24974 Change ins into fmov.
24976 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24978 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24979 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24980 Use h_con constraint for operand 1.
24981 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24982 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24984 2017-03-15 Jeff Law <law@redhat.com>
24986 PR tree-optimization/71437
24987 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24988 (record_temporary_equivalences): Use it.
24990 PR tree-optimization/71437
24991 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24992 tree-ssa-scopedtables.
24993 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24994 (record_conditions, record_cond, vuse_eq): Likewise.
24995 (record_edge_info): Adjust to API tweak of record_conditions.
24996 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24997 (record_temporary_equivalences, optimize_stmt): Likewise.
24998 (eliminate_redundant_computations): Likewise.
24999 (record_equivalences_from_stmt): Likewise.
25000 * tree-ssa-scopedtables.c: Include options.h and params.h.
25001 (vuse_eq): New function, moved from tree-ssa-dom.c
25002 (build_and_record_new_cond): Likewise.
25003 (record_conditions): Likewise. Accept vector of conditions rather
25004 than edge_equivalence structure for first argument.
25005 for the first argument.
25006 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25007 from tree-ssa-dom.c.
25008 (avail_exprs_stack::record_cond): Likewise.
25009 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25010 from tree-ssa-dom.c.
25011 (avail_exprs_stack): Add new member functions lookup_avail_expr
25013 (record_conditions): Declare.
25015 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25018 * lra-constraints.c (process_alt_operands): Increase reject for
25019 reloading an input/output operand.
25021 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25024 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25025 insns to convert from signed/unsigned char/short to IEEE 128-bit
25027 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25029 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25032 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25033 subreg of inner mode for values already in registers.
25035 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25037 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25038 iteration reg is used after the loop.
25040 2017-03-14 Martin Sebor <msebor@redhat.com>
25042 PR tree-optimization/79800
25043 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25044 precision in negative-positive range.
25045 (format_floating): Call non-const overload with adjusted precision.
25047 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25050 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25053 2017-03-14 Martin Sebor <msebor@redhat.com>
25055 PR middle-end/80020
25056 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25057 * builtins.def (aligned_alloc): Use it.
25060 * Makefile.in (GTFILES): Add calls.c.
25061 * calls.c: Include "gt-calls.h".
25063 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25065 PR rtl-optimization/79728
25066 * regs.h (struct target_regs): New field
25067 x_contains_allocatable_regs_of_mode.
25068 (contains_allocatable_regs_of_mode): New macro.
25069 * reginfo.c (init_reg_sets_1): Initialize it, and change
25070 contains_reg_of_mode so it includes global regs as well.
25071 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25072 rather than contains_regs_of_mode.
25074 2017-03-14 Martin Liska <mliska@suse.cz>
25076 * doc/invoke.texi: Document options that can't be combined with
25077 -fcheck-pointer-bounds.
25079 2017-03-14 Martin Liska <mliska@suse.cz>
25081 PR middle-end/79831
25082 * doc/invoke.texi (-Wchkp): Document the option.
25084 2017-03-14 Martin Liska <mliska@suse.cz>
25086 * Makefile.in: Install gcov-dump.
25088 2017-03-14 Martin Liska <mliska@suse.cz>
25090 * multiple_target.c (expand_target_clones): Bail out for
25091 an invalid attribute.
25093 2017-03-14 Richard Biener <rguenther@suse.de>
25095 * alias.c (struct alias_set_entry): Pack properly.
25096 * cfgloop.h (struct loop): Likewise.
25097 * cse.c (struct set): Likewise.
25098 * ipa-utils.c (struct searchc_env): Likewise.
25099 * loop-invariant.c (struct invariant): Likewise.
25100 * lra-remat.c (struct cand): Likewise.
25101 * recog.c (struct change_t): Likewise.
25102 * rtl.h (struct address_info): Likewise.
25103 * symbol-summary.h (function_summary): Likewise.
25104 * tree-loop-distribution.c (struct partition): Likewise.
25105 * tree-object-size.c (struct object_size_info): Likewise.
25106 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25107 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25108 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25109 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25110 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25111 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25112 (struct _stmt_vec_info): Likewise.
25114 2017-03-14 Martin Liska <mliska@suse.cz>
25117 * multiple_target.c (create_dispatcher_calls): Check that
25118 a target can create a function dispatcher.
25120 2017-03-14 Martin Liska <mliska@suse.cz>
25123 * multiple_target.c (expand_target_clones): Drop local.local
25124 flag for default implementation.
25126 2017-03-14 Richard Biener <rguenther@suse.de>
25128 PR tree-optimization/80030
25129 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25131 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25133 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25134 gcc_fallthrough() instead of __attribute__((fallthrough));
25136 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25138 * doc/gcc.texi: Remove "up" link to (DIR).
25139 * doc/gccint.texi: Ditto.
25141 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25143 * doc/install.texi (Specific) <avr>: Remove reference to
25146 2017-03-13 Jeff Law <law@redhat.com>
25148 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25149 attribute rather than comments.
25151 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25152 match_scratch operand is highest.
25154 2017-03-13 Martin Liska <mliska@suse.cz>
25156 PR middle-end/78339
25157 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25158 is a CHKP clone, use original declaration.
25160 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25162 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25163 (arc_conditional_register_usage): Use a different allocation order
25164 when optimizing for size.
25165 * common/config/arc/arc-common.c (arc_option_optimization_table):
25166 Section anchors default on when optimizing for size.
25168 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25170 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25172 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25174 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25175 * config/arc/arc.md (cpu_facility): Add cd variant.
25176 (*movqi_insn): Add code density variant.
25177 (*movhi_insn): Likewise.
25178 (*movqi_insn): Likewise.
25179 (*addsi3_mixed): Likewise.
25180 (subsi3_insn): Likewise.
25182 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25184 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25186 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25188 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25189 expressions with MINUS and UNARY ops.
25191 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25194 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25196 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25197 between vec_select and vector argument.
25198 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25199 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25200 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25201 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25202 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25203 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25205 2017-03-13 Richard Biener <rguenther@suse.de>
25208 * params.def (vect-max-peeling-for-alignment): Fix typo.
25210 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25212 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25213 issue that only occurred with binutils below 2.18.
25215 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25217 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25218 refer to binutils 2.11/2.12 minimum.
25220 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25222 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25223 ftp.kernel.org and simplify binutils requirement.
25225 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25227 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25229 (Optimize Options): Ditto. Also remove redundancy.
25231 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25233 PR translation/79848
25234 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25236 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25237 to G_ to avoid double translation.
25239 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25241 PR translation/79923
25242 * auto-profile.c (get_combined_location): Convert leading
25243 character of diagnostics to lower case and remove trailing period.
25244 (read_profile): Likewise for various diagnostics.
25245 * config/arm/arm.c (arm_option_override): Remove trailing period
25246 from various diagnostics.
25247 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25248 (msp430_expand_delay_cycles): Likewise.
25250 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25253 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25254 full command-line argument, rather than just "str".
25255 (aarch64_validate_march): Likewise.
25256 (aarch64_validate_mtune): Likewise.
25258 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25260 PR rtl-optimization/78911
25261 * lra-assigns.c (must_not_spill_p): New function.
25262 (spill_for): Use it.
25264 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25266 PR tree-optimization/79981
25267 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25268 ATOMIC_COMPARE_EXCHANGE ifn result.
25269 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25270 IFN_ATOMIC_COMPARE_EXCHANGE.
25272 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25275 * opts.c (parse_sanitizer_options): Add missing question mark to
25276 "did you mean" message.
25278 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25280 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25282 (VMULEUH_UNS): Likewise.
25283 (VMULOUB_UNS): Likewise.
25284 (VMULOUH_UNS): Likewise.
25285 * config/rs6000/rs6000.c (builtin_function_type): Remove
25286 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25288 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25291 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25292 x with result of extra_parsing_for_operand_code_0.
25293 (function_reader::extra_parsing_for_operand_code_0): Convert
25294 return type from void to rtx, returning x. When reading
25295 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25296 larger size containing struct block_symbol.
25298 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25300 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25301 -mfloat128-hardware without -m64.
25303 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25306 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25307 entries to the case statement that marks unsigned arguments to
25308 overloaded functions.
25310 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25312 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25313 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25315 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25318 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25319 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25321 2017-03-10 Martin Liska <mliska@suse.cz>
25325 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25326 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25329 2017-03-10 Olivier Hainque <hainque@adacore.com>
25331 * tree-switch-conversion (array_value_type): Start by resetting
25332 candidate type to it's main variant.
25334 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25336 PR rtl-optimization/79909
25337 * combine.c (try_combine): Use simplify_replace_rtx on individual
25338 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25339 of the whole CALL_INSN_FUNCTION_USAGE.
25341 PR tree-optimization/79972
25342 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25343 get_range_info on SSA_NAMEs. Formatting fixes.
25345 2017-03-10 Richard Biener <rguenther@suse.de>
25346 Jakub Jelinek <jakub@redhat.com>
25348 PR tree-optimization/77975
25349 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25350 edge to be constant.
25351 (get_val_for): For constant x return it. Formatting fix.
25352 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25353 would use the same bases as the current one.
25355 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25357 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25358 instead of vec_select for V1TImode.
25359 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25361 (VSX_LE_128): Add V1TI to this mode iterator.
25362 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25363 (*vsx_le_perm_store_<mode>): Likewise.
25364 (pre-reload splitter for VSX stores): Likewise.
25365 (post-reload splitter for VSX stores): Likewise.
25366 (*vsx_xxpermdi2_le_<mode>): Likewise.
25367 (*vsx_lxvd2x2_le_<mode>): Likewise.
25368 (*vsx_stxvd2x2_le_<mode>): Likewise.
25370 2017-03-09 Michael Eager <eager@eagercon.com>
25372 Correct failures with --enable-checking=yes,rtl.
25374 * config/microblaze/microblaze.c (microblaze_expand_shift):
25375 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25376 test for const0_rtx.
25377 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25378 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25380 2017-03-09 Richard Biener <rguenther@suse.de>
25382 PR tree-optimization/79977
25383 * graphite-scop-detection.c (scop_detection::merge_sese):
25384 Handle the case of extra exits to blocks dominating the entry.
25386 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25388 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25391 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25393 PR rtl-optimization/79949
25394 * lra-constraints.c (process_alt_operands): Check memory when
25395 trying to predict a cycle. Print about the overall increase.
25397 2017-03-09 Richard Biener <rguenther@suse.de>
25399 PR middle-end/79971
25400 * gimple-expr.c (useless_type_conversion_p): Preserve
25401 TYPE_SATURATING for fixed-point types.
25403 2017-03-09 Richard Biener <rguenther@suse.de>
25406 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25407 alignment of BLKmode params.
25409 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25412 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25413 (VALL_NO_V2Q): Likewise.
25415 * config/aarch64/aarch64-simd.md
25416 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25418 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25419 VALL_NO_V2Q mode iterator.
25420 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25422 2017-03-09 Martin Liska <mliska@suse.cz>
25424 PR tree-optimization/79631
25425 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25426 tree_int_cst_sign_bit just for INTEGER constants.
25428 2017-03-09 Martin Liska <mliska@suse.cz>
25432 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25435 2017-03-09 Marek Polacek <polacek@redhat.com>
25438 * tree.c (inchash::add_expr): Handle TREE_VEC.
25440 2017-03-09 Martin Liska <mliska@suse.cz>
25443 (chkp_narrow_size_and_offset): New function.
25444 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25445 (void chkp_parse_bit_field_ref): New function.
25446 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25447 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25449 2017-03-09 Martin Liska <mliska@suse.cz>
25452 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25453 (chkp_find_bounds_1): Remove gcc_unreachable.
25455 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25458 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25459 BUILT_IN_SYNC*, determine the access type from the size suffix and
25460 always build a MEM_REF with that type. Handle forgotten
25461 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25464 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25465 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25466 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25467 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25468 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25469 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25470 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25471 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25472 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25473 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25474 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25475 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25476 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25477 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25478 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25479 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25480 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25481 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25482 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25483 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25484 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25485 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25486 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25487 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25488 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25489 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25490 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25491 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25492 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25493 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25494 definitions outside of __OPTIMIZE__ guarded section.
25497 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25498 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25499 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25500 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25503 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25505 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25506 ("vfenez<mode>"): Add missing constraints.
25508 2017-03-08 Martin Sebor <msebor@redhat.com>
25511 * config/nds32/nds32.c (nds32_option_override):
25512 Fix misspelled diagnostic.
25514 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25517 * gimplify.c (gimplify_omp_for): Replace index var in outer
25518 taskloop statement with an artificial variable and add
25519 OMP_CLAUSE_PRIVATE clause for it.
25521 2017-03-08 Richard Biener <rguenther@suse.de>
25523 PR tree-optimization/79955
25524 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25525 for accesses that are completely outside of the variable.
25527 2017-03-08 Andrew Haley <aph@redhat.com>
25529 PR tree-optimization/79943
25530 * tree-ssa-loop-split.c (compute_new_first_bound): When
25531 calculating the new upper bound, (END-BEG) should be added, not
25534 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25536 * config/avr/avr.md (setmemhi): Make sure match_dup
25537 operand number comes before match_scratch.
25539 2017-03-08 Richard Biener <rguenther@suse.de>
25541 PR tree-optimization/79920
25542 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25543 with ncopies == 1 to ...
25544 (vect_transform_slp_perm_load): ... here. Properly compute
25545 all element loads by iterating VF times over the group. Do
25546 not handle ncopies (computed in a broken way) in
25547 vect_create_mask_and_perm.
25549 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25552 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25553 is a uniform vector, use uniform_vector_p return value instead of
25554 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25556 2017-03-07 Marek Polacek <polacek@redhat.com>
25558 PR middle-end/79809
25559 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25560 (alloca_call_type): Likewise.
25562 2017-03-07 Martin Liska <mliska@suse.cz>
25564 * gcov.c (process_args): Put comment to correct location.
25566 2017-03-07 Martin Liska <mliska@suse.cz>
25568 PR middle-end/68270
25569 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25570 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25571 (chkp_narrow_bounds_for_field): Likewise.
25572 (chkp_parse_array_and_component_ref): Pass one more argument to
25575 2017-03-07 Richard Biener <rguenther@suse.de>
25577 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25580 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25582 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25583 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25585 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25588 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25589 to end of description.
25590 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25592 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25594 PR rtl-optimization/79901
25595 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25597 (*avx512f_<code><mode>3<mask_name>): ... this.
25598 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25599 iterator instead of VI8_AVX2_AVX512BW.
25601 PR rtl-optimization/79901
25602 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25603 min/max expander, expand it using expand_vec_cond_expr.
25606 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25609 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25612 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25613 to void * for PCH reasons.
25614 * dwarf2out.c (output_loc_operands, output_die): Cast
25615 v.val_vec.array to unsigned char *.
25617 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25620 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25623 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25625 PR rtl-optimization/79571
25626 * lra-constraints.c (process_alt_operands): Calculate static
25627 reject and subtract it from overall when only addresses will be
25630 2017-03-06 Julia Koval <julia.koval@intel.com>
25633 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25634 incoming stack boundary to 128 for 64-bit targets.
25636 2017-03-06 Richard Biener <rguenther@suse.de>
25638 PR tree-optimization/79894
25639 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25640 to NULL after folding it.
25642 2017-03-06 Richard Biener <rguenther@suse.de>
25644 PR tree-optimization/79824
25645 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25646 check disabling peeling for gaps.
25648 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25650 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25651 attributes): Document gettimeofday.
25653 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25655 * config/s390/s390.c (s390_option_override_internal): Set
25656 PARAM_MIN_VECT_LOOP_BOUND
25658 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25660 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25661 * config/s390/s390.md: Likewise.
25663 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25666 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25667 (<avx2_avx512>_perm<mode>): Rename to ...
25668 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25670 (<avx512>_perm<mode>_mask): Rename to ...
25671 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25673 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25674 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25675 instead of VI8F_256_512.
25676 (avx512f_perm<mode>): New define_expand.
25677 (avx512f_perm<mode>_mask): Likewise.
25678 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25679 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25681 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25683 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25684 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25686 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25688 2017-03-06 Martin Liska <mliska@suse.cz>
25691 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25692 when having a SSA NAME w/o VAR_DECL assigned to it.
25694 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25696 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25697 msa_dpsub_<su>_d): Fix MODE for vec_select.
25699 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25701 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25703 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25705 2017-03-06 Richard Biener <rguenther@suse.de>
25707 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25708 * plugin.c (register_plugin_info): Likewise.
25709 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25711 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25713 * config/i386/sse.md (sse_storehps, sse_storelps,
25714 avx_<castmode><avxsizesuffix>_<castmode>,
25715 avx512f_<castmode><avxsizesuffix>_<castmode>,
25716 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25717 in condition that at least one operand is not a MEM.
25719 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25721 PR middle-end/79805
25722 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25723 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25725 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25726 gimple_call_nothrow_p flag based on whether original builtin can throw.
25727 If it can, emit following stmts on the fallthrough edge.
25728 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25729 don't create new bb if inserting just debug stmts on the edge, try to
25730 insert them on the fallthru bb or just reset debug stmts.
25732 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25735 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25736 restore recog_data (including the operand rtxes inside it) around
25737 the call to get_insn_template.
25739 2017-03-03 Martin Sebor <msebor@redhat.com>
25741 PR tree-optimization/79699
25742 * context.c (context::~context): Free MPFR caches to avoid
25743 a memory leak on program exit.
25745 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25747 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25748 Use wide_int::ulow () instead of .elt (0).
25750 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25752 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25753 (*pushxf): Limit oF constraint to 32bit targets and add oC
25754 constraint for 64bit targets.
25755 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25756 (*pushdf): Change rmF constraint to rmC.
25758 2017-03-03 Martin Liska <mliska@suse.cz>
25760 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25761 Remove unused variable.
25763 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25766 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25767 is a memory operand, increase num_memory.
25768 (ix86_expand_args_builtin): Likewise.
25770 2017-03-03 Jan Hubicka <jh@suse.cz>
25773 * ipa-devirt.c (maybe_record_node): Properly handle
25774 __cxa_pure_virtual visibility.
25776 2017-03-03 Martin Liska <mliska@suse.cz>
25778 PR tree-optimization/79803
25779 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25781 (pass_loop_prefetch::execute): Disabled optimization if an
25782 assumption about L1 cache size is not met.
25784 2017-03-03 Martin Liska <mliska@suse.cz>
25786 PR rtl-optimization/79574
25787 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25788 (hash_scan_set): Likewise.
25789 (dump_hash_table): Likewise.
25790 (hoist_code): Likewise.
25792 2017-03-03 Richard Biener <rguenther@suse.de>
25794 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25795 in place of uhigh ().
25796 (fixed_convert_from_real): Likewise.
25798 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25801 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25803 2017-03-03 Richard Biener <rguenther@suse.de>
25805 PR middle-end/79818
25806 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25807 TYPE_OVERFLOW_UNDEFINED check.
25809 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25811 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25813 (vector_ae_<mode>_p): Likewise.
25814 (vector_nez_<mode>_p): Likewise.
25815 (vector_ne_v2di_p): Likewise.
25816 (vector_ae_v2di_p): Likewise.
25817 (vector_ne_<mode>_p): Likewise.
25818 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25820 (vsx_tsqrt<mode>2_fe): Likewise.
25822 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25825 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25827 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25829 PR rtl-optimization/79780
25830 * cprop.c (one_cprop_pass): When second and further conditional trap
25831 in a single basic block is turned into an unconditional trap, turn it
25832 into a deleted note to avoid RTL verification failures.
25834 2017-03-02 Richard Biener <rguenther@suse.de>
25836 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25838 2017-03-02 Richard Biener <rguenther@suse.de>
25840 PR tree-optimization/79345
25842 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25843 param and abort the walk, returning -1 if it is hit.
25844 (walk_aliased_vdefs): Take a limit param and pass it on.
25845 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25846 defaulting to 0 and return a signed int.
25847 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25848 (check_defs): New helper.
25849 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25850 about uninitialized memory.
25851 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25852 bogus uninitialized warning.
25853 (fixed_convert_from_real): Likewise.
25855 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25857 PR tree-optimization/66768
25858 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25859 iv_use if base object can't be determined.
25861 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25863 PR tree-optimization/79345
25864 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25865 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25866 (get_pattern_stats): Initialize it.
25867 * genemit.c (gen_expand): Verify match_scratch numbers come after
25868 match_operand/match_dup numbers.
25869 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25870 match_scratch numbers.
25871 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25873 * config/s390/s390.md (trunctdsd2): Likewise.
25875 2017-03-02 Richard Biener <rguenther@suse.de>
25877 * wide-int.h (wide_int_storage::operator=): Implement in terms
25880 2017-03-02 Richard Biener <rguenther@suse.de>
25882 PR tree-optimization/79777
25883 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25884 the to insert expression to sth existing.
25886 2017-03-01 Martin Sebor <msebor@redhat.com>
25888 PR middle-end/79692
25889 * gimple-ssa-sprintf.c
25890 (directive::known_width_and_precision): New function.
25891 (format_integer): Use it.
25892 (get_mpfr_format_length): Consider the full range of precision
25893 when computing %g output with the # flag. Set the likely byte
25894 count to 3 rather than 1 when precision is indeterminate.
25895 (format_floating): Correct the lower bound of precision.
25897 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25899 * doc/invoke.texi: Document default code model for 64-bit Linux.
25901 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25904 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25905 udiv rather than div since input pattern is unsigned.
25907 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25909 * config/i386/i386.c (print_reg): Warn for values of
25910 unsupported size in integer register.
25912 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25915 * config/rs6000/predicates.md (current_file_function_operand): Do
25916 not allow self calls to be local if the function is replaceable.
25918 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25921 * config/rs6000/altivec.h (vec_ctz and others): Change the
25922 preprocessor macro that controls conditional compilation from
25923 _ARCH_PWR9 to __POWER9_VECTOR__.
25924 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25925 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25926 control (instead of _ARCH_PWR9 control) so that template
25927 definition uses power9-specific function.
25928 (vec_any_eq): Likewise.
25929 (vec_all_ne): Change macro definition to use a power9-specific
25930 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25931 _ARCH_PWR9 control).
25932 (vec_any_eq) Likewise.
25933 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25934 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25935 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25936 support for xvcmpnedp instruction.
25937 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25938 macro expansion so that Power9 implementation of vec_all_ne does
25939 not use the AltiVec predicate framework.
25940 (VCMPNEH_P): Likewise.
25941 (VCMPNEW_P): Likewise.
25942 (VCMPNED_P): Likewise.
25943 (VCMPNEFP_P): Likewise.
25944 (VCMPNEDP_P): Likewise.
25945 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25946 implementation of vec_any_eq to not use AltiVec predicate
25948 (VCMPAEH_P): Likewise.
25949 (VCMPAEW_P): Likewise.
25950 (VCMPAED_P): Likewise.
25951 (VCMPAEFP_P): Likewise.
25952 (VCMPAEDP_P): Likewise.
25953 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25954 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25955 not use the AltiVec predicate framework.
25956 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25957 of vec_any_eq to not use AltiVec predicate framework.
25958 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25959 support for predefined __POWER9_VECTOR__ macro to indicate that
25960 Power9 instruction selection is enabled.
25961 (altivec_overloaded_builtins): Remove extraneous
25962 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25963 function argument types RS6000_BTI_bool_V16QI and
25964 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25965 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25966 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25967 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25968 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25969 Power9 for implementations of vec_cmpne. Change the signature for
25970 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25971 (representing vec_all_ne) to remove the previously described first
25972 argument of type RS6000_BTI_INTSI, as this was an artifact of
25973 reliance on the AltiVec predicate framework, which is no longer
25974 used in the implementation of these functions. Add
25975 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25976 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25977 since, unlike the AltiVec predicate framework implementation, we
25978 do not share function descriptors between vec_alle and vec_anyeq.
25979 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25980 set of modes that receive special treatment even when
25981 TARGET_P9_VECTOR is true. The special treatment emits code that
25982 does not depend on Power9 instructions.
25983 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25984 define_expand to not rely on AltiVec predicate framework.
25985 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25987 (vector_ne_v2di_p): Change this define_expand to not rely on
25988 AltiVec predicate framework.
25989 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25991 (vector_ne_<mode>_p): Change this define_expand to not rely on
25992 AltiVec predicate framework.
25993 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25995 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25996 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25997 define_insn pattern.
25998 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25999 define_insn pattern because the xvcmpne<VSs>. instruction is not
26001 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26002 instruction is not supported.
26004 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26006 * config/nvptx/nvptx.c: Include intl.h.
26008 2017-03-01 Martin Jambor <mjambor@suse.cz>
26011 * ipa-prop.h (ipa_bits): Removed field known.
26012 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26013 to pointers. Adjusted their comments to warn about their sharing.
26014 (ipcp_transformation_summary): Change bits to a vector of pointers.
26015 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26016 (ipa_get_ipa_bits_for_value): Declare.
26017 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26018 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26019 (ipa_bits_hash_table): Likewise.
26020 (ipa_vr_ggc_hash_traits): Likewise.
26021 (ipa_vr_hash_table): Likewise.
26022 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26023 being pointers and vr_known being removed.
26024 (ipa_set_jf_unknown): Likewise.
26025 (ipa_get_ipa_bits_for_value): New function.
26026 (ipa_set_jfunc_bits): Likewise.
26027 (ipa_get_value_range): New overloaded functions.
26028 (ipa_set_jfunc_vr): Likewise.
26029 (ipa_compute_jump_functions_for_edge): Use the above functions to
26030 construct bits and vr parts of jump functions.
26031 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26032 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26034 (ipcp_grow_transformations_if_necessary): Also allocate
26035 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26037 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26038 them. Fix too long lines.
26039 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26040 vr_known being removed.
26041 (ipa_read_jump_function): Use new setter functions to construct bits
26042 and vr parts of jump functions or set them to NULL.
26043 (write_ipcp_transformation_info): Adjust for bits being pointers.
26044 (read_ipcp_transformation_info): Likewise.
26045 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26047 Include gt-ipa-prop.h.
26048 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26050 (ipcp_store_bits_results): Likewise.
26051 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26052 Do not write to existing jump functions but use a temporary instead.
26054 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26057 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26058 attempt to use its first operand as BIT_FIELD_REF base.
26060 2017-03-01 Richard Biener <rguenther@suse.de>
26062 PR middle-end/79721
26063 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26064 interpolating formula in wrapping arithmetic.
26065 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26067 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26069 PR tree-optimization/79734
26070 * tree-vect-generic.c (expand_vector_condition): Optimize
26071 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26072 Handle VEC_COND_EXPR where comparison has different inner width from
26073 type's inner width.
26075 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26077 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26078 markup, and similar issues. Remove @opindex entries for things
26079 that aren't options. Add missing -mmpy-option entries.
26081 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26083 PR tree-optimization/79737
26084 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26085 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26086 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26087 instead of byte_size. Formatting fix.
26088 (shift_bytes_in_array_right): Formatting fix.
26090 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26093 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26094 condition on optimize for the leaf function test.
26096 2017-02-28 Martin Liska <mliska@suse.cz>
26099 * read-rtl-function.c (function_reader::handle_unknown_directive):
26100 Bail out when one uses -flto.
26102 2017-02-28 Martin Liska <mliska@suse.cz>
26104 * common.opt: Replace space with tabular for options of <number>
26106 * config/i386/i386.opt: Show <number> value for
26107 -mlarge-data-threshold.
26108 * opts.c (print_filtered_help): Do not display number in hexadecimal
26111 2017-02-28 Martin Liska <mliska@suse.cz>
26113 * common.opt: Fix --help=option -Q for options which are of
26116 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26118 * config/i386/i386.c (print_reg): Error out for values
26119 of 8-bit size in invalid integer register.
26121 2017-02-28 Martin Sebor <msebor@redhat.com>
26123 PR tree-optimization/79691
26124 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26126 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26129 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26130 gcc_unreachable with output_operand_lossage.
26132 2017-02-28 Richard Biener <rguenther@suse.de>
26134 PR tree-optimization/79740
26135 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26137 (visit_nary_op): Insert the nary into the hashtable if we
26138 pattern-matched sth.
26139 * tree-ssa-pre.c (eliminate_insert): Robustify.
26141 2017-02-28 Richard Biener <rguenther@suse.de>
26143 PR middle-end/79731
26144 * fold-const.c (decode_field_reference): Reject out-of-bound
26147 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26149 * config/i386/i386.c: Include intl.h.
26150 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26151 instead of just cond ? "..." : "...".
26152 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26153 * coverage.c (read_counts_file): Likewise.
26154 * omp-offload.c: Include intl.h.
26155 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26156 of just cond ? "..." : "...".
26157 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26158 of just cond ? "..." : "...".
26160 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26163 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26165 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26166 'tune for' CPU name.
26167 * config/arm/arm-cpu-data.h: Regenerated.
26169 2017-02-28 Richard Biener <rguenther@suse.de>
26171 PR tree-optimization/79732
26172 * tree-inline.c (expand_call_inline): Do not shadow var.
26174 2017-02-28 Richard Biener <rguenther@suse.de>
26176 PR tree-optimization/79723
26177 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26178 address-space properly.
26180 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26182 * doc/optinfo.texi (Optimization groups): Fix option used for
26184 * doc/invoke.texi (-fopt-info): Document "omp".
26185 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26186 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26187 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26188 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26189 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26191 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26193 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26194 OPTGROUP_OMP with "omp".
26196 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26199 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26200 for arithmetic shift of unsigned V2DI.
26202 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26204 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26205 arc/linux.h headers.
26206 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26207 (LINK_SPEC): Likewise.
26208 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26209 (EXTRA_SPECS): Likewise.
26210 (STARTFILE_SPEC): Likewise.
26211 (ENDFILE_SPEC): Likewise.
26212 (LIB_SPEC): Likewise.
26213 (TARGET_SDATA_DEFAULT): Likewise.
26214 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26215 (MULTILIB_DEFAULTS): Likewise.
26216 (DWARF2_UNWIND_INFO): Likewise.
26217 * config/arc/big.h: New file.
26218 * config/arc/elf.h: Likewise.
26219 * config/arc/linux.h: Likewise.
26220 * config/arc/t-uClibc: Remove.
26222 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26224 PR tree-optimization/77536
26225 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26226 (tree_transform_and_unroll_loop): Use above function to compute the
26227 estimated niter of unrolled loop and use it when scaling profile.
26228 Also use count info rather than frequency if it's non-zero.
26229 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26230 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26231 (vect_transform_loop): Call above function.
26233 2017-02-27 Richard Biener <rguenther@suse.de>
26235 PR tree-optimization/45397
26236 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26237 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26238 (visit_nary_op): Add pattern matching for CSEing sign-changed
26239 or truncated operations with wider ones.
26241 2017-02-27 Richard Biener <rguenther@suse.de>
26243 PR tree-optimization/79690
26244 * tree-vect-stmts.c (vectorizable_store): Use vector type
26245 built from the DR with address-space.
26247 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26249 * doc/invoke.texi (Optimize Options): Refine the description
26250 of asan-use-after-return.
26252 2017-02-25 Alan Modra <amodra@gmail.com>
26254 PR rtl-optimization/79584
26255 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26256 base, not ad->base_term, the reg within base. Remove assertion
26257 that ad->base == ad->base_term. Replace gen_int_mode using
26258 bogus mode with const0_rtx.
26260 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26262 PR middle-end/79396
26263 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26264 FMA_EXPR like tcc_binary or tcc_unary.
26266 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26269 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26271 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26272 (output_loc_operands): Handle DW_OP_call_ref and
26273 DW_OP_GNU_variable_value.
26274 (struct variable_value_struct): New type.
26275 (struct variable_value_hasher): Likewise.
26276 (variable_value_hash): New variable.
26277 (string_types): Remove.
26278 (copy_loc_descr): New function.
26279 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26280 (prepend_loc_descr_to_each): New function.
26281 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26282 instead of add_loc_descr_to_each if the first argument is single
26283 location list and the second has multiple.
26284 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26285 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26286 when looking for variable value which doesn't have other location info.
26287 (loc_list_from_tree): Formatting fix.
26288 (gen_array_type_die): Simplify DW_AT_string_length handling.
26289 (adjust_string_types): Remove.
26290 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26291 string_types. Call resolve_variable_values.
26292 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26293 (resolve_addr_in_expr): Likewise. Add A argument.
26294 (copy_deref_exprloc): Remove deref argument. Adjust for the
26295 original expression being DW_OP_GNU_variable_value with optionally
26296 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26297 optionally after it.
26298 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26299 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26300 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26301 (variable_value_hasher::hash, variable_value_hasher::equal): New
26303 (resolve_variable_value_in_expr, resolve_variable_value,
26304 resolve_variable_values, note_variable_value_in_expr,
26305 note_variable_value): New functions.
26306 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26309 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26312 * opts.h (handle_generated_option): Add GENERATED_P argument.
26313 * opts-common.c (handle_option): Adjust function comment.
26314 (handle_generated_option): Add GENERATED_P argument, pass it to
26316 (control_warning_option): Pass false to handle_generated_option
26318 * opts.c (maybe_default_option): Pass true to handle_generated_option
26320 * optc-gen.awk: Likewise.
26322 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26324 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26325 a REG, look at the REG it is a SUBREG of.
26326 (splitter for cmpeqsi_t): Ditto.
26328 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26330 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26331 the special USEs with the pattern of the insn, not the insn itself.
26333 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26336 * doc/invoke.texi: Document -mload-store-pairs.
26338 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26339 Sandra Loosemore <sandra@codesourcery.com>
26341 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26342 argument isn't a CONST_INT.
26343 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26344 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26345 (nios2_validate_compare): Bypass alternate compare logic if *op2
26346 is not a CONST_INT.
26347 (ldstwm_operation_p): Return false if first_base is not a REG or
26348 if first_offset is not a CONST_INT.
26350 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26352 * config/cris/cris.md: Use correct operand in a define_peephole2.
26354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26356 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26358 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26360 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26361 this_insn if it is an INSN or JUMP_INSN.
26362 (force_offsettable): Look at base, not at addr.
26363 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26364 on things that aren't necessarily CONST_INTs.
26366 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26368 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26369 -mfpmath=sse is the default also for x86-32 targets with SSE2
26370 instruction set when @option{-ffast-math} is enabled
26372 2017-02-24 Jeff Law <law@redhat.com>
26374 PR rtl-optimizatoin/79286
26375 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26378 2017-02-24 Richard Biener <rguenther@suse.de>
26380 PR tree-optimization/79389
26381 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26384 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26386 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26387 function comment to reflect reality.
26388 (loop_exits_before_overflow): Fix typo in function description.
26390 2017-02-24 Richard Biener <rguenther@suse.de>
26392 PR tree-optimization/79389
26393 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26394 properly that a threading opportunity exists. Detect conditional
26395 copy/constant propagation opportunities.
26397 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26399 * config/visium/visium.md (type): Add trap.
26400 (b): New mode attribute.
26401 (*btst): Rename into...
26402 (*btst<mode>): ...this and adjust.
26403 (*cbranchsi4_btst_insn): Rename into...
26404 (*cbranch<mode>4_btst_insn): ...this and adjust.
26405 (trap): New define_insn.
26407 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26409 PR tree-optimization/79389
26410 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26411 (noce_reversed_cond_code): New function.
26412 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26413 reversed_comparison_code. Formatting fix.
26414 (noce_try_store_flag): Test rev_cond != NULL in addition to
26415 reversed_comparison_code.
26416 (noce_try_store_flag_constants): Likewise.
26417 (noce_try_store_flag_mask): Likewise.
26418 (noce_try_addcc): Use rev_cond if non-NULL instead of
26419 reversed_comparison_code.
26420 (noce_try_cmove_arith): Likewise. Formatting fixes.
26421 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26422 (noce_find_if_block): Initialize rev_cond.
26423 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26424 instead of false as last argument never attempt to reverse it
26427 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26429 PR tree-optimization/79663
26430 * tree-predcom.c (combine_chains): Process refs in reverse order
26431 only for ZERO length chains, and add explaining comment.
26433 2017-02-23 Jeff Law <law@redhat.com>
26435 PR tree-optimization/79578
26436 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26437 in call to operand_equal_p.
26439 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26442 * config/i386/cpuid.h: Fix another undefined behavior.
26444 2017-02-23 Richard Biener <rguenther@suse.de>
26446 PR tree-optimization/79683
26447 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26448 vector types for data-refs.
26450 2017-02-23 Martin Liska <mliska@suse.cz>
26452 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26454 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26456 PR middle-end/79665
26457 * internal-fn.c (get_range_pos_neg): Moved to ...
26458 * tree.c (get_range_pos_neg): ... here. No longer static.
26459 * tree.h (get_range_pos_neg): New prototype.
26460 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26461 are known to be in between 0 and signed maximum inclusive, try to
26462 expand both unsigned and signed divmod and use the cheaper one from
26465 2017-02-22 Jeff Law <law@redhat.com>
26467 PR tree-optimization/79578
26468 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26469 to compare base operands.
26471 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26474 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26475 gpc_reg_operand instead of fpr_reg_operand.
26477 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26479 * config/mips/mips.c (mips_return_in_memory): Force FP
26480 vector types to be returned in memory for o32 ABI.
26482 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26484 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26485 instead of DW_TAG_member for static data member declarations and don't
26486 set no_linkage_name for static inline data members.
26487 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26490 2017-02-22 Martin Liska <mliska@suse.cz>
26492 * doc/invoke.texi: Replace inequality signs with square brackets
26495 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26497 PR tree-optimization/68644
26498 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26500 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26503 * lra-constraints.c (simplify_operand_subreg): Handle
26504 WORD_REGISTER_OPERATIONS targets.
26506 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26509 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26510 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26511 elimination by swapping fld*.
26513 2017-02-22 Richard Biener <rguenther@suse.de>
26515 PR tree-optimization/79673
26516 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26517 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26518 irrelevant address-space qualifiers and avoiding a
26519 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26521 2017-02-22 Richard Biener <rguenther@suse.de>
26523 PR tree-optimization/79666
26524 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26525 to not symbolically negate if that may introduce undefined
26528 2017-02-22 Martin Liska <mliska@suse.cz>
26531 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26532 * data-streamer-out.c (streamer_write_gcov_count_stream):
26534 * value-prof.c (stream_out_histogram_value): Make assert more
26535 precise based on type of counter.
26537 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26540 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26541 Use nonimmediate_operand instead of memory_operand for operand 1.
26542 (float-extend standard_x87sse_constant_load splitter): Ditto.
26544 2017-02-21 Jeff Law <law@redhat.com>
26546 PR tree-optimization/79621
26547 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26548 blocks with edges to themselves.
26550 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26553 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26554 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26555 Use gimple_call_builtin_p.
26558 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26559 on temporarily removed DEBUG_INSNs.
26561 PR tree-optimization/79649
26562 * tree-loop-distribution.c (classify_partition): Give up on
26563 non-generic address space loads/stores.
26565 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26567 * doc/loop.texi (Loop manipulation): Remove nonexistent
26568 tree_ssa_loop_version from the documentation.
26569 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26571 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26574 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26575 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26576 * config/rs6000/rs6000.c: Include except.h.
26577 (rs6000_expand_split_stack_prologue): Call
26578 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26580 2017-02-21 Martin Jambor <mjambor@suse.cz>
26583 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26584 have been analyzed.
26586 2017-02-21 Martin Jambor <mjambor@suse.cz>
26588 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26589 for backward compatibility only.
26590 * doc/invoke.texi (Option Summary): Remove all references to
26591 -fipa-cp-alignment.
26593 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26597 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26599 * lra-constraints.c (curr_insn_transform): Handle
26600 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26602 2017-02-21 Martin Liska <mliska@suse.cz>
26604 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26606 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26608 PR translation/79638
26609 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26611 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26614 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26615 (arm_function_ok_for_sibcall): Return false for an indirect call by
26616 descriptor if all the argument registers are used.
26617 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26618 alignment of the function.
26620 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26622 PR tree-optimization/61441
26623 * simplify-rtx.c (simplify_const_unary_operation): For
26624 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26625 the sNaN unmodified.
26627 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26629 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26630 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26631 instead of SYSTEM_HEADER_DIR.
26633 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26634 Martin Liška <mliska@suse.cz>
26636 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26637 Fix typos and grammar, use active voice, and clarify.
26639 2017-02-20 Marek Polacek <polacek@redhat.com>
26641 PR middle-end/79537
26642 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26645 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26647 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26650 * config/i386/i386.c (ix86_expand_builtin): Handle
26651 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26652 ix86_builtins_isa[fcode].isa as a requirement of those
26653 flags and any other flag in the bitmask.
26654 (ix86_init_mmx_sse_builtins): Use 0 instead of
26655 ~OPTION_MASK_ISA_64BIT as mask.
26656 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26657 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26658 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26659 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26661 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26664 * lra-constraints.c (split_reg): Check requested split mode
26665 is supported by the register.
26667 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26669 * lra-constraints.c (simplify_operand_subreg): Remove early
26672 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26675 * lra-constraints.c (curr_insn_transform): Tighten condition
26676 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26678 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26681 * lra-constraints.c (curr_insn_transform): Handle
26682 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26684 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26687 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26689 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26691 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26694 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26697 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26699 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26702 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26704 * final.c (last_columnnum, override_columnnum): New variables.
26705 (final_start_function): Set last_columnnum, pass it to begin_prologue
26706 hook and pass 0 to dwarf2out_begin_prologue.
26707 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26708 to source_line debug hook.
26709 (notice_source_line): Compute last_columnnum and for debug_column_info
26710 return true on column changes.
26711 * debug.h (struct gcc_debug_hooks): Add column argument to
26712 source_line and begin_prologue hooks.
26713 (debug_nothing_int_charstar_int_bool): Remove prototype.
26714 (debug_nothing_int_int_charstar,
26715 debug_nothing_int_int_charstar_int_bool): New prototypes.
26716 (dwarf2out_begin_prologue): Add column argument.
26717 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26718 begin_prologue hooks.
26719 (debug_nothing_int_charstar_int_bool): Remove.
26720 (debug_nothing_int_int_charstar,
26721 debug_nothing_int_int_charstar_int_bool): New functions.
26722 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26723 through to dwarf2out_source_line.
26724 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26725 (dwarf2out_source_line): Add column argument, emit it if requested.
26726 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26728 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26729 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26730 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26731 through to dwarf2out_begin_prologue.
26732 (vmsdbgout_source_line): Add column argument, pass it through to
26733 dwarf2out_source_line.
26734 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26735 dbxout_source_line caller.
26736 (dbxout_source_line): Add column argument.
26738 * common.opt (gno-column-info, gcolumn-info): New options.
26739 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26740 (check_die): Also test for multiple DW_AT_decl_column attributes.
26741 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26742 DW_AT_decl_column if requested.
26743 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26745 (gen_variable_die): Likewise.
26746 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26747 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26750 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26751 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26752 (ix86_handle_option): Handle OPT_m3dnowa.
26753 * doc/invoke.texi (-m3dnowa): Document.
26754 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26755 -m3dnowa instead of -m3dnow -march=athlon.
26758 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26759 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26761 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26764 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26765 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26766 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26767 generator for vsx_xxpermdi_<mode>_be.
26768 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26769 force big-endian semantics.
26770 (vsx_xxpermdi_<mode>_be): New define_expand with same
26771 implementation as previous version of vsx_xxpermdi_<mode>.
26773 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26775 PR tree-optimization/79327
26776 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26777 variable, its initialization and use.
26779 2017-02-17 Julia Koval <julia.koval@intel.com>
26781 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26782 (OPTION_MASK_ISA_PKU_UNSET): New.
26783 (ix86_handle_option): Handle -mrdpid.
26784 * config/i386/cpuid.h (bit_RDPID): New.
26785 * config/i386/driver-i386.c (host_detect_local_cpu):
26786 Detect RDPID feature.
26787 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26788 * config/i386/i386-c.c (ix86_target_macros_internal):
26790 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26791 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26792 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26793 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26794 * config/i386/i386.md (define_insn "rdpid"): New.
26795 * config/i386/i386.opt Add -mrdpid.
26796 * config/i386/immintrin.h (_rdpid_u32): New.
26798 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26800 PR rtl-optimization/79541
26801 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26802 instead of transforming it into USE.
26804 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26806 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26807 If HONOR_SNANS (SFmode) force the input to a register.
26808 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26809 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26810 an frsp or similar insn.
26812 2017-02-17 Martin Liska <mliska@suse.cz>
26814 PR rtl-optimization/79577
26815 * params.def (selsched-max-sched-times): Increase minimum to 1.
26817 2017-02-17 Martin Liska <mliska@suse.cz>
26819 PR rtl-optimization/79574
26820 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26822 2017-02-17 Martin Liska <mliska@suse.cz>
26824 PR tree-optimization/79529
26825 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26826 ssa_defined_default_def_p to handle cases which are implicitly
26828 * tree-ssa.c (ssa_defined_default_def_p): New function.
26829 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26830 which are implicitly defined.
26831 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26833 2017-02-17 Richard Biener <rguenther@suse.de>
26835 PR middle-end/79576
26836 * params.def (max-ssa-name-query-depth): Limit to 10.
26838 2017-02-17 Richard Biener <rguenther@suse.de>
26840 PR tree-optimization/79552
26841 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26844 2017-02-17 Richard Biener <rguenther@suse.de>
26847 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26849 2017-02-17 Marek Polacek <polacek@redhat.com>
26851 PR middle-end/79536
26852 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26853 (fold_negate_expr): New wrapper.
26855 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26857 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26858 Correct terminology and de-emphasize pre-standard behavior.
26860 2017-02-16 Alan Modra <amodra@gmail.com>
26862 PR rtl-optimization/79286
26863 * ira.c (def_dominates_uses): New function.
26864 (update_equiv_regs): Don't create an equivalence for insns that
26865 may trap where the register def does not dominate the use.
26867 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26869 PR rtl-optimization/78127
26870 * lra.c (lra): Call lra_eliminate before finish the loop after
26873 2017-02-16 Richard Biener <rguenther@suse.de>
26875 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26877 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26878 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26879 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26880 (isl_val_int_from_wi): New function.
26881 (extract_affine_gmp): Rename to ...
26882 (extract_affine_wi): ... this, take a widest_int.
26883 (extract_affine_int): Just wrap extract_affine_wi.
26884 (add_param_constraints): Use isl_val_int_from_wi.
26885 (add_loop_constraints): Likewise, and extract_affine_wi.
26887 2017-02-15 Jeff Law <law@redhat.com>
26889 PR middle-end/79521
26890 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26891 ira_init_register_move_cost_if_necessary.
26893 2017-02-15 Martin Sebor <msebor@redhat.com>
26895 PR middle-end/32003
26896 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26897 removed in a prior commit.
26899 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26901 PR tree-optimization/79347
26902 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26903 counters during peeling.
26905 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26907 * Makefile.in (site.exp): Remove "set ISLVER".
26909 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26912 * real.c (real_from_integer): Call real_convert even for decimal.
26914 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26917 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26919 2017-02-14 Andrew Pinski <apinski@cavium.com>
26921 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26922 cores and change the partno/implementer to be correct.
26923 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26924 the 'B" as the implementer.
26925 * config/aarch64/aarch64-tune.md: Regenerate.
26927 2017-02-14 Carl Love <cel@us.ibm.com>
26929 * config/rs6000/rs6000.c: Add case statement entry to make the
26930 xvcvuxdsp built-in argument unsigned.
26931 * config/rs6000/vsx.md: Fix the source and return operand types so they
26932 match the instruction definitions from the ISA document. Fix typo
26933 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26936 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26939 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26940 member early_clobber_alts.
26941 * lra-lives.c (reg_early_clobber_p): New.
26942 (process_bb_lives): Use it.
26943 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26944 (debug_operand_data): Initialize early_clobber_alts.
26945 (setup_operand_alternative): Set up early_clobber_alts.
26946 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26947 alternatives to new_insn_reg.
26948 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26950 (lra_update_insn_regno_info): Pass the new arg.
26952 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26954 PR middle-end/79505
26955 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26956 (new_oacc_loop_raw): Don't clear already cleared fields.
26959 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26960 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26961 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26963 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26966 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26968 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26971 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26972 the extra instruction to the right place to store 128-bit constant
26975 2017-02-14 Martin Sebor <msebor@redhat.com>
26977 PR middle-end/79448
26978 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26979 warning for strings of unknown length.
26981 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26983 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26985 2017-02-14 Jeff Law <law@redhat.com>
26988 * ira-costs.c (scan_one_insn): Initialize register move costs
26989 for pseudos seen in USE/CLOBBER insns.
26991 PR tree-optimization/79095
26992 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26993 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26994 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26995 if the operands are known to be not equal, then the resulting range
26997 (intersect_ranges): If the new range is ~[0,0] and the old range is
26998 wide, then prefer ~[0,0].
26999 * tree-vrp.c (overflow_comparison_p_1): New function.
27000 (overflow_comparison_p): New function.
27001 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27002 if NAME is used in an overflow test.
27003 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27004 overflow check that can be expressed as an equality test, then adjust
27005 ops to be that equality test.
27007 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27009 * config/s390/s390-builtin-types.def: Remove flags argument.
27010 * config/s390/s390.c (s390_init_builtins): Likewise.
27012 2017-02-14 Martin Liska <mliska@suse.cz>
27014 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27015 vector. Fix trailing white spaces.
27017 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27019 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27022 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27024 PR rtl-optimization/68664
27025 * config/arm/arm.c (arm_sched_can_speculate_insn):
27026 New function. Declare prototype.
27027 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27029 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27031 PR rtl-optimization/68664
27032 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27034 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27036 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27038 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27039 max skip bytes for function, loop and jump.
27041 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27043 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27044 ABS_EXPR for gimple dump.
27046 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27049 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27051 PR tree-optimization/79408
27052 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27053 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27054 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27055 also if rhs1 is INTEGER_CST.
27057 2017-02-14 Richard Biener <rguenther@suse.de>
27059 PR middle-end/79432
27060 * tree-into-ssa.c (insert_phi_nodes): When the function can
27061 have abnormal edges rewrite SSA names with broken use-def
27062 dominance out of SSA and register them for PHI insertion.
27064 2017-02-13 Martin Sebor <msebor@redhat.com>
27066 PR middle-end/79496
27067 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27068 clearing info.nowrite flag when snprintf size argument is a range.
27070 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27072 * cprop.c (cprop_jump): Add missing space in string literal.
27073 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27074 (get_constraint_for_component_ref): Likewise.
27075 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27076 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27077 * lra-constraints.c (process_alt_operands): Likewise.
27078 * ipa-inline.c (inline_small_functions): Likewise.
27079 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27080 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27081 * trans-mem.c (diagnose_tm_1_op): Likewise.
27082 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27083 (grid_parallel_clauses_gridifiable): Likewise.
27085 * config/nvptx/mkoffload.c (process): Add space in between
27088 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27089 "MOD4_SSE_REGS" and "ALL_REGS".
27091 * spellcheck.c (test_data): Add , in between "foo" and "food".
27093 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27096 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27097 boundary crossing check and subsequent code generation agree.
27099 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27101 * config/aarch64/aarch64.c (has_memory_op): Delete.
27102 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27105 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27107 PR rtl-optimization/79388
27108 PR rtl-optimization/79450
27109 * combine.c (distribute_notes): When removing TEM_INSN for which
27110 corresponding dest has last value recorded, invalidate that last
27113 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27115 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27116 of explicit '@'. Add missing assembly comment marker on branch costs
27119 2017-02-13 Nathan Sidwell <nathan@acm.org>
27121 * gengtype-lex.l (<in_struct>): Add '/'.
27123 2017-02-13 Martin Liska <mliska@suse.cz>
27126 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27128 2017-02-13 Richard Biener <rguenther@suse.de>
27130 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27132 * configure: Re-generate.
27133 * config.in: Likewise.
27134 * graphite-dependences.c: Simplify as if
27135 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27136 * graphite-isl-ast-to-gimple.c: Likewise.
27137 * graphite-optimize-isl.c: Likewise.
27138 * graphite-poly.c: Likewise.
27139 * graphite-sese-to-poly.c: Likewise.
27140 * graphite.h: Likewise.
27141 * toplev.c: Include isl/version.h and use isl_version () for
27142 printing the ISL version.
27143 * doc/install.texi: Update ISL requirement.
27145 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27147 * doc/standards.texi (Standards): Update reference to
27150 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27152 * doc/extend.texi (Named Address Spaces): sourceware.org now
27154 * doc/install.texi (Binaries): Ditto.
27157 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27159 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27160 standard terminology "stringize"/"stringizing" throughout.
27161 * doc/cppinternals.texi: Likewise.
27163 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27165 * doc/extend.texi: Fix some spelling mistakes and typos.
27166 * doc/invoke.texi: Likewise.
27168 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27171 * params.def (inline-min-speedup) Change from 10 to 8.
27173 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27175 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27178 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27181 * ipa-inline-analysis.c (get_minimal_bb): New function.
27182 (record_modified): Use it.
27183 (remap_edge_change_prob): Handle also ancestor functions.
27185 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27187 * doc/contrib.texi (Contributors): Remove broken link into
27188 the Mauve CVS repository.
27190 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27192 PR middle-end/79454
27193 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27194 result computation whenever lhs doesn't have vector mode, not
27195 just when it has BLKmode.
27197 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27199 * doc/makefile.texi (profiledbootstrap): Refer to the
27200 installation instructions only in textual form.
27202 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27205 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27207 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27209 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27210 (Specific): Update mingw-w64 reference.
27212 (Specific): Remove broken link to Renesas RX processor.
27214 2017-02-10 Richard Biener <rguenther@suse.de>
27216 * toplev.c (process_options): Do not mention obsolete graphite
27217 options when printing sorry message about missing graphite support.
27218 Mention -floop-nest-optimize.
27220 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27222 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27223 (vtst_p16): Likewise.
27224 (vtstq_p8): Likewise.
27225 (vtstq_p16): Likewise.
27227 (vtstq_p64): Likewise.
27228 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27229 (vset_lane_p64): New.
27230 (vsetq_lane_p64): New.
27232 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27234 PR tree-optimization/79411
27235 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27236 stmt operands are SSA_NAMEs used in abnormal phis.
27237 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27240 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27243 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27246 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27248 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27250 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27252 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27256 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27257 not arbitrary TREE_CONSTANT.
27260 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27261 "omp declare target link" attribute unless is_global_var.
27262 * omp-offload.c (find_link_var_op): Likewise.
27264 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27265 Chung-Lin Tang <cltang@codesourcery.com>
27267 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27269 (gimplify_adjust_omp_clauses): Don't delete TILE.
27270 (gimplify_omp_for): Deal with TILE.
27271 * internal-fn.c (expand_GOACC_TILE): New function.
27272 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27274 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27275 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27277 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27278 avoid DIV for outermost collapse var.
27279 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27280 Remove out of date comments, fix whitespace.
27281 * omp-general.c (omp_extract_for_data): Deal with tiling.
27282 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27283 adjust OLF_DIM_BASE value.
27284 (struct omp_for_data): Add tiling field.
27285 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27286 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27287 for auto loops. Remove default auto determining, moved to
27288 oacc_loop_fixed_partitions.
27289 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27290 stmts, add e_mask field.
27291 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27292 (oacc_thread_numbers): Use oacc_dim_call.
27293 (oacc_xform_tile): New.
27294 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27295 (finish_oacc_loop): Adjust for ifns vector.
27296 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27297 add case for GOACC_TILE fns.
27298 (oacc_loop_xform_loop): Delete.
27299 (oacc_loop_process): Iterate over call list directly, and add
27300 handling for GOACC_TILE fns.
27301 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27303 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27304 vector partitioning to outer loops. Assign 2 partitions to loops
27305 when available. Add TILE handling.
27306 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27307 (execite_oacc_device_lower): Process GOACC_TILE fns,
27308 ignore unknown specs.
27309 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27310 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27311 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27313 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27315 * configure.ac (ACX_BUGURL): Update.
27316 * configure: Regenerate.
27318 2017-02-09 Richard Biener <rguenther@suse.de>
27320 PR tree-optimization/69823
27321 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27322 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27324 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27326 * config/arc/arc-c.def: Add __NPS400__ definition.
27327 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27328 (TARGET_NPS400): Define.
27330 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27332 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27334 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27335 pointer, arch_info.
27336 (arc_cpu_types): Fill the arch_info field with a pointer into the
27337 arc_arch_types table.
27338 (arc_selected_cpu): Declare.
27339 * config/arc/arc.c (arc_selected_cpu): Make global.
27340 (arc_selected_arch): Delete.
27341 (arc_base_cpu): Delete.
27342 (arc_override_options): Remove references to deleted variables,
27343 update access to arch information.
27344 (ARC_OPT): Update access to arch information.
27345 (ARC_OPTX): Likewise.
27346 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27347 (TARGET_ARC600): Update access to arch information.
27348 (TARGET_ARC601): Likewise.
27349 (TARGET_ARC700): Likewise.
27350 (TARGET_EM): Likewise.
27351 (TARGET_HS): Likewise.
27352 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27355 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27358 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27359 condition/operands for integer GE/LE/GEU/LEU operations.
27361 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27363 PR translation/79397
27364 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27367 2017-02-08 Martin Jambor <mjambor@suse.cz>
27370 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27371 whether allocation happened.
27372 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27373 nothing was allocated.
27375 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27377 PR tree-optimization/79408
27378 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27379 constant, but SSA_NAME with a known integer range, use the minimum
27380 of that range instead of op1 to determine if modulo can be replaced
27381 with its first operand.
27383 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27385 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27387 2017-02-08 Richard Biener <rguenther@suse.de>
27389 PR tree-optimization/71824
27390 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27391 Check all loops contained in the merged region.
27393 2017-02-07 Andrew Pinski <apinski@cavium.com>
27395 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27397 2017-02-07 Andrew Pinski <apinski@cavium.com>
27399 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27400 (thunderxt88): Likewise.
27401 (thunderxt81): Disable LSE and change v8.1 to v8.
27402 (thunderxt83): Likewise.
27404 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27405 Richard Biener <rguenther@suse.de>
27407 PR middle-end/79399
27408 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27409 type from int to size_t.
27410 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27412 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27414 PR rtl-optimization/79386
27415 * cprop.c (bypass_conditional_jumps): Initialize
27416 bypass_last_basic_block already before splitting bbs after
27417 unconditional traps...
27418 (bypass_conditional_jumps): ... rather than here.
27421 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27422 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27423 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27424 fix -masm=intel patterns.
27426 2017-02-07 Richard Biener <rguenther@suse.de>
27428 PR tree-optimization/79256
27429 PR middle-end/79278
27430 * builtins.c (get_object_alignment_2): Use min_align_of_type
27431 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27432 and ADJUST_FIELD_ALIGN.
27434 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27436 * doc/tm.texi: Regenerate.
27437 * stor-layout.c (layout_decl): Adjust.
27438 (update_alignment_for_field): Likewise.
27439 (place_field): Likewise.
27440 (min_align_of_type): Likewise.
27441 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27442 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27443 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27444 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27445 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27446 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27447 * config/i386/i386.c (x86_field_alignment): Likewise.
27448 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27449 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27450 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27451 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27452 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27453 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27457 2017-01-30 Richard Biener <rguenther@suse.de>
27459 PR tree-optimization/79256
27460 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27461 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27464 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27466 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27467 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27468 builtins to SImode and emit a zero-extend, if necessary.
27470 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27472 * docs/invoke.texi (RISC-V Options): Alphabetize.
27474 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27476 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27479 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27481 * config/riscv/riscv.c: New file.
27482 * common/config/riscv/riscv-common.c: Likewise.
27483 * config.gcc: Likewise.
27484 * config/riscv/constraints.md: Likewise.
27485 * config/riscv/elf.h: Likewise.
27486 * config/riscv/generic.md: Likewise.
27487 * config/riscv/linux.h: Likewise.
27488 * config/riscv/multilib-generator: Likewise.
27489 * config/riscv/peephole.md: Likewise.
27490 * config/riscv/pic.md: Likewise.
27491 * config/riscv/predicates.md: Likewise.
27492 * config/riscv/riscv-builtins.c: Likewise.
27493 * config/riscv/riscv-c.c: Likewise.
27494 * config/riscv/riscv-ftypes.def: Likewise.
27495 * config/riscv/riscv-modes.def: Likewise.
27496 * config/riscv/riscv-opts.h: Likewise.
27497 * config/riscv/riscv-protos.h: Likewise.
27498 * config/riscv/riscv.h: Likewise.
27499 * config/riscv/riscv.md: Likewise.
27500 * config/riscv/riscv.opt: Likewise.
27501 * config/riscv/sync.md: Likewise.
27502 * config/riscv/t-elf-multilib: Likewise.
27503 * config/riscv/t-linux: Likewise.
27504 * config/riscv/t-linux-multilib: Likewise.
27505 * config/riscv/t-riscv: Likewise.
27506 * configure.ac: Likewise.
27507 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27508 Waterman as RISC-V maintainers.
27509 * doc/install.texi: Add RISC-V entries.
27510 * doc/invoke.texi: Add RISC-V options section.
27511 * doc/md.texi: Add RISC-V constraints section.
27512 * configure: Regenerated.
27514 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27517 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27518 false values to be constant vectors with all 0 or all 1 bits set.
27519 (vcondu<mode><mode>): Likewise.
27520 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27522 (fpmask_comparison_operator): Update comment.
27523 (vecint_comparison_operator): New predicate.
27524 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27525 vector conditionals when the true and false values are constant
27526 vectors with all 0 bits or all 1 bits set.
27528 2017-02-06 Martin Sebor <msebor@redhat.com>
27530 PR tree-optimization/79376
27531 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27533 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27535 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27536 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27537 to simplify split condition.
27539 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27541 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27542 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27545 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27547 PR rtl-optimization/68664
27548 * target.def (can_speculate_insn): New hook.
27549 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27550 * doc/tm.texi: Regenerate.
27551 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27552 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27553 (rs6000_sched_can_speculate_insn): New function.
27555 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27557 PR tree-optimization/79284
27558 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27559 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27560 vectorizable_mask_load_store, vectorizable_operation,
27561 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27562 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27563 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27564 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27565 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27566 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27567 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27568 is_gimple_assign (stmt). Replace another such test with
27569 is_gimple_assign (stmt).
27571 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27574 * config/avr/avr.c (rtl-iter.h): Include it.
27575 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27576 (avr_legitimate_combined_insn): ...and implementation.
27578 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27580 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27581 * config/s390/s390.c (s390_const_operand_ok)
27582 (s390_canonicalize_comparison, s390_extract_part)
27583 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27584 (s390_contiguous_bitmask_p, s390_rtx_costs)
27585 (legitimize_pic_address): Likewise.
27586 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27587 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27588 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27589 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27590 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27592 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27594 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27595 REGNO($0) == REGNO($1).
27597 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27599 * config/s390/linux.h(SIZE_TYPE): Add comment.
27601 2017-02-06 Julian Brown <julian@codesourcery.com>
27602 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27603 Virendra Pathak <virendra.pathak@broadcom.com>
27605 * config/aarch64/aarch64-cores.def: Change the scheduler
27607 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27608 * config/aarch64/thunderx2t99.md: New file.
27610 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27612 * doc/standards.texi (Go Language): Update link to language
27615 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27617 * tree-eh.c (lower_resx): Sanitize profile.
27618 (cleanup_empty_eh_move_lp): Likewise.
27620 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27623 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27625 * cfgloopmanip.h (loop_version): Update prototype.
27626 * modulo-sched.c (sms_schedule): Update call of loop_version.
27627 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27628 * tree-parloops.c (gen_parallel_loop): Likewise.
27629 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27630 * tree-ssa-loop-split.c (split_loop): Likewise.
27631 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27632 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27634 2017-02-05 Martin Liska <mliska@suse.cz>
27637 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27639 (print_operand_address): Initialize a struct to zero.
27641 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27643 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27644 garbage collector only in textual form.
27646 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27648 * doc/extend.texi (x86 specific memory model extensions for
27649 transactional memory): Simplify a phrase.
27651 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27654 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27655 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27656 (atomic_storedi_1): Likewise.
27658 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27660 PR tree-optimization/79338
27661 * tree-parloops.c (gather_scalar_reductions): Don't call
27662 vect_analyze_loop_form for loop->inner before destroying loop's
27665 2017-02-03 Martin Sebor <msebor@redhat.com>
27667 PR tree-optimization/79327
27668 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27669 when precision has resulted in leading zeros.
27670 (format_integer): Adjust the likely counter to assume an unknown
27671 argument that may be zero is non-zero.
27673 2017-02-03 Jason Merrill <jason@redhat.com>
27676 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27677 avoid copying non-taken branch.
27679 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27681 PR tree-optimization/79340
27682 * tree-vect-loop.c (vectorizable_reduction): Release
27683 vec_defs elements after safe_splicing them into other vectors.
27686 PR tree-optimization/79327
27687 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27688 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27690 (format_integer): Use wide_int_to_tree instead of build_int_cst
27691 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27692 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27693 of shortest and longest sequence.
27695 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27697 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27698 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27700 2017-02-03 Walter Lee <walt@tilera.com>
27703 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27704 after initial stackframe link reg save.
27705 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27707 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27710 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27711 wu for stxssp alternative.
27713 2017-02-03 Martin Sebor <msebor@redhat.com>
27715 PR tree-optimization/79352
27716 * gimple-fold.c (get_range_strlen): Add argument.
27717 (get_range_strlen): Change return type to bool.
27718 (get_maxval_strlen): Pass in a dummy argument.
27719 * gimple-fold.h (get_range_strlen): Change return type to bool.
27720 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27721 * tree.h (array_at_struct_end_p): Add argument.
27722 * tree.c (array_at_struct_end_p): Handle it.
27724 2017-02-03 Martin Liska <mliska@suse.cz>
27727 * multiple_target.c (create_dispatcher_calls): Redirect edge
27728 from a caller of a dispatcher.
27729 (expand_target_clones): Make the clones local.
27730 (ipa_target_clone): Do both target clones and resolvers.
27731 (ipa_dispatcher_calls): Remove the pass.
27732 (pass_dispatcher_calls::gate): Likewise.
27733 (make_pass_dispatcher_calls): Likewise.
27734 * passes.def (pass_target_clone): Put as very first IPA early
27737 2017-02-03 Martin Liska <mliska@suse.cz>
27739 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27740 in case of a function with ifunc attribute.
27742 2017-02-03 Martin Liska <mliska@suse.cz>
27744 * cgraph.c (cgraph_node::dump): Dump function version info.
27745 * symtab.c (symtab_node::dump_base): Add missing new line.
27747 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27749 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27750 (ifcombine_ifandif): Use it.
27752 2017-02-03 Martin Liska <mliska@suse.cz>
27754 * doc/invoke.texi: Document default value for
27755 use-after-scope-direct-emission-threshold.
27757 2017-02-03 Martin Liska <mliska@suse.cz>
27759 PR tree-optimization/79339
27760 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27761 (format_floating): Likewise.
27763 2017-02-03 Martin Liska <mliska@suse.cz>
27766 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27768 (ipa_node_params_t::remove): Likewise.
27769 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27770 initialization from removed ipa_node_params_t::insert.
27771 (ipa_node_params::~ipa_node_params): Move from removed
27772 ipa_node_params_t::release.
27773 * symbol-summary.h (symbol_summary::m_released): New member.
27774 Do not release a summary twice. Do not allow to call finalizer
27775 for types of a summary that live in GGC memory.
27777 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27779 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27782 2017-02-02 Martin Sebor <msebor@redhat.com>
27784 PR middle-end/79275
27785 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27786 (format_string): Tighten up the range of output for non-constant
27787 strings and correct the expected range for wide non-constant strings.
27789 2017-02-02 Martin Sebor <msebor@redhat.com>
27791 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27793 PR middle-end/32003
27794 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27796 (-fdump-tree-@var): Add to index and document how to come up
27797 with pass-specific option and dump file names.
27798 (-fdump-passes): Clarify where to look for output.
27800 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27802 PR middle-end/77445
27803 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27804 statistics of the analyzed path; allow threading for speed when
27805 any of BBs along the path are optimized for speed.
27807 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27809 PR middle-end/78468
27810 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27811 settings of the virtual registers.
27814 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27816 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27817 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27820 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27822 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27823 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27825 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27827 * config/s390/s390.md: Add missing comments with the expanded
27829 * config/s390/vector.md: Likewise.
27830 * config/s390/vx-builtins.md: Likewise.
27832 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27835 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27836 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27837 conditions on a single line.
27839 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27841 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27842 __S390_VX__ to __VX__.
27844 2017-02-01 Andrew Pinski <apinski@cavium.com>
27846 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27847 stmt_info to record_stmt_cost.
27848 (vect_get_known_peeling_cost): Pass stmt_info if known to
27850 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27851 cpu_vector_cost field into
27852 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27853 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27854 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27855 splitting of scalar_stmt_cost and vec_stmt_cost.
27856 (thunderx_vector_cost): Likewise.
27857 (cortexa57_vector_cost): LIkewise.
27858 (exynosm1_vector_cost): Likewise.
27859 (xgene1_vector_cost): Likewise.
27860 (thunderx2t99_vector_cost): Improve after the splitting of the two
27862 (aarch64_builtin_vectorization_cost): Update for the splitting of
27863 scalar_stmt_cost and vec_stmt_cost.
27865 2017-02-01 Torvald Riegel <triegel@redhat.com>
27866 Richard Henderson <rth@redhat.com>
27868 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27869 conditional on existance of a fast atomic load.
27870 * optabs-query.c (can_atomic_load_p): New function.
27871 * optabs-query.h (can_atomic_load_p): Declare it.
27872 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27873 no fast atomic load is available for the particular size of access.
27874 (expand_atomic_compare_and_swap): Likewise.
27875 (expand_atomic_load): Likewise.
27876 (expand_atomic_store): Likewise.
27877 (expand_atomic_fetch_op): Likewise.
27878 * testsuite/lib/target-supports.exp
27879 (check_effective_target_sync_int_128): Remove x86 because it provides
27880 no fast atomic load.
27881 (check_effective_target_sync_int_128_runtime): Likewise.
27883 2017-02-01 Richard Biener <rguenther@suse.de>
27885 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27886 (graphite_transform_loops): Provide opt-info for optimized nests.
27887 * tree-parloop.c (parallelize_loops): Provide opt-info for
27888 parallelized loops.
27890 2017-02-01 Richard Biener <rguenther@suse.de>
27892 PR middle-end/79315
27893 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27894 was not set before.
27896 2017-02-01 Richard Biener <rguenther@suse.de>
27898 PR tree-optimization/71824
27899 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27900 Verify the loops are valid in the merged SESE region.
27901 (scop_detection::can_represent_loop_1): Check analyzing the
27902 evolution of the number of iterations in the region succeeds.
27904 2017-01-31 Ian Lance Taylor <iant@golang.org>
27906 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27907 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27909 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27911 PR preprocessor/79210
27912 * input.c (get_substring_ranges_for_loc): Replace line_width
27913 assertion with error-handling.
27915 2017-01-31 Richard Biener <rguenther@suse.de>
27917 PR tree-optimization/77318
27918 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27919 (create_pw_aff_from_tree): Take loop parameter.
27920 (add_condition_to_pbb): Pass loop of the condition to
27921 create_pw_aff_from_tree.
27923 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27925 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27926 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27928 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27932 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27934 (convert_int_to_float128): Likewise.
27935 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27936 (convert_int_to_float128): Likewise.
27937 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27938 (UNSPEC_IEEE128_CONVERT): Likewise.
27939 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27940 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27941 Use local variables for IBM extended format.
27942 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27943 (fix_trunc<mode>si2_fprs): Likewise.
27944 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27945 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27946 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27947 to know that we can now have integers of all sizes in vector
27949 (fix<uns>_<mode>di2_hw): Likewise.
27950 (float<uns>_<mode>si2_hw): Likewise.
27951 (fix_<mode>si2_hw): Likewise.
27952 (fixuns_<mode>si2_hw): Likewise.
27953 (float<uns>_<mode>di2_hw): Likewise.
27954 (float_<mode>di2_hw): Likewise.
27955 (float_<mode>si2_hw): Likewise.
27956 (floatuns_<mode>di2_hw): Likewise.
27957 (floatuns_<mode>si2_hw): Likewise.
27958 (xscvqp<su>wz_<mode>): Delete, no longer used.
27959 (xscvqp<su>dz_<mode>): Likewise.
27960 (xscv<su>dqp_<mode>): Likewise.
27961 (ieee128_mfvsrd_64bit): Likewise.
27962 (ieee128_mfvsrd_32bit): Likewise.
27963 (ieee128_mfvsrwz): Likewise.
27964 (ieee128_mtvsrw): Likewise.
27965 (ieee128_mtvsrd_64bit): Likewise.
27966 (ieee128_mtvsrd_32bit): Likewise.
27968 2017-01-31 Martin Liska <mliska@suse.cz>
27971 * ipa-prop.c (ipa_free_all_node_params): Call release method
27972 instead of ~sumbol_summary to not to trigger double times
27975 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27977 PR tree-optimization/71691
27978 * bitmap.h (class auto_bitmap): New.
27979 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27980 is_maybe_undefined instead of ssa_undefined_value_p.
27982 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27984 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27985 __S390_ARCH_LEVEL__ to __ARCH__.
27987 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27989 PR tree-optimization/79267
27990 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27991 if should_remove_lhs_p is true.
27993 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27996 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27997 (add_alignment_attribute): New.
27998 (base_type_die): Add alignment attribute.
27999 (subrange_type_die): Likewise.
28000 (modified_type_die): Likewise.
28001 (gen_array_type_die): Likewise.
28002 (gen_descr_array_type_die: Likewise.
28003 (gen_enumeration_type_die): Likewise.
28004 (gen_subprogram_die): Likewise.
28005 (gen_variable_die): Likewise.
28006 (gen_field_die): Likewise.
28007 (gen_ptr_to_mbr_type_die): Likewise.
28008 (gen_struct_or_union_type_die): Likewise.
28009 (gen_subroutine_type_die): Likewise.
28010 (gen_typedef_die): Likewise.
28011 (base_type_cmp): Compare alignment attribute.
28013 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28016 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28017 (setb_unsigned) New pattern for setb with CCUNS.
28018 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28019 subfc./subfe sequence to avoid overflow problems. Generate a
28020 shorter sequence with cmpld/setb for power9.
28021 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28022 for generating subfc. instruction.
28023 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28024 now uses this instruction.
28026 2017-01-30 Ian Lance Taylor <iant@google.com>
28029 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28030 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28032 2017-01-30 Martin Sebor <msebor@redhat.com>
28034 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28035 Move constant to the right of a relational operator.
28036 (get_mpfr_format_length, format_character, format_string): Ditto.
28037 (should_warn_p, maybe_warn): Same.
28039 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28041 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28044 * asan.c (get_translation_unit_decl): Remove function.
28045 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28047 2017-01-30 Martin Liska <mliska@suse.cz>
28049 PR gcov-profile/79259
28050 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28051 -fprofile-generate.
28053 2017-01-30 Martin Liska <mliska@suse.cz>
28056 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28057 Initialize variables with NULL value.
28059 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28062 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28064 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28066 2017-01-30 Richard Biener <rguenther@suse.de>
28068 PR tree-optimization/79276
28069 * tree-vrp.c (process_assert_insertions): Properly adjust common
28070 when removing a duplicate.
28072 * gcc.dg/torture/pr79276.c: New testcase.
28074 2017-01-30 Richard Biener <rguenther@suse.de>
28076 PR tree-optimization/79256
28077 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28078 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28080 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28082 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28085 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28086 ("*r<noxa>sbg_<mode>_sll_bitmask")
28087 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28088 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28089 Use contiguous_bitmask_nowrap_operand.
28091 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28094 * config/rs6000/altivec.h (vec_xl): Revise #define.
28095 (vec_xst): Likewise.
28097 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28099 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28101 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28103 PR rtl-optimization/79194
28104 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28105 traps before call to bypass_conditional_jumps.
28107 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28109 PR tree-optimization/71374
28110 * lra-constraints.c (check_conflict_input_operands): New.
28111 (match_reload): Use it.
28113 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28116 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28117 account to calculate conflict_set.
28119 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28121 PR rtl-optimization/78559
28122 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28123 other_insn in combine.
28125 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28127 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28128 uint16_type_node for BT_UINT16.
28130 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28132 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28133 "RTL Tests" to menu.
28134 (GIMPLE Tests): New node.
28135 (RTL Tests): New node.
28137 2017-01-27 Richard Biener <rguenther@suse.de>
28139 PR tree-optimization/79245
28140 * tree-loop-distribution.c (distribute_loop): Apply cost
28141 modeling also to detected patterns.
28143 2017-01-27 Richard Biener <rguenther@suse.de>
28145 PR tree-optimization/71433
28146 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28147 (compare_assert_loc): New function.
28148 (process_assert_insertions): Sort and optimize assert locations
28149 to remove duplicates and push down identical assertions on
28150 edges to their destination block.
28152 2017-01-27 Richard Biener <rguenther@suse.de>
28154 PR tree-optimization/79244
28155 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28156 out SSA names even if abnormal.
28158 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28160 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28161 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28162 instead of MPFR_RNDN.
28164 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28167 * arm.c (arm_option_override): Don't call build_target_option_node
28168 until after doing all option overrides.
28169 (arm_valid_target_attribute_tree): Likewise.
28171 2017-01-27 Martin Liska <mliska@suse.cz>
28173 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28174 for {cd}tors and C++ {cd}tors.
28176 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28178 * config/s390/s390.md ("*setmem_long_and")
28179 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28181 2017-01-26 Martin Sebor <msebor@redhat.com>
28183 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28186 2017-01-26 Martin Sebor <msebor@redhat.com>
28188 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28189 HAVE_DFmode before using XFmode or DFmode.
28190 (parse_directive): Avoid using the z length modifier to avoid
28191 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28193 PR middle-end/78703
28194 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28195 to accept adjustment as an array.
28196 (get_int_range): New function.
28197 (struct directive): Make width and prec arrays.
28198 (directive::set_width, directive::set_precision): Call get_int_range.
28199 (format_integer, format_floating): Handle width and precision ranges.
28200 (format_string, parse_directive): Same.
28202 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28205 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28206 comdat_type_p set, just clone them, but keep the children in the
28210 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28211 which have direct callers with -fvar-tracking-assignments enabled
28213 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28214 inside of type units.
28216 2017-01-26 Martin Sebor <msebor@redhat.com>
28218 PR middle-end/78703
28219 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28221 (struct format_result): Replace number_chars, number_chars_min,
28222 and number_chars_max with a single member of struct result_range.
28224 (format_result::operator+=): Adjust.
28225 (struct fmtresult): Remove bounded. Handle likely and unlikely
28227 (fmtresult::adjust_for_width_or_precision): New function.
28228 (fmtresult:type_max_digits): New function.
28229 (bytes_remaining): Handle likely and unlikely counters.
28230 (min_bytes_remaining): Remove.
28231 (format_percent): Simplify.
28232 (format_integer, format_floating): Set likely and unlikely counters.
28233 (get_string_length, format_character, format_string): Same.
28234 (format_plain, should_warn_p): New function.
28235 (maybe_warn): Call should_warn_p. Update diagnostic messages
28236 and handle those for all directives, including plain strings.
28237 (format_directive): Handle likely and unlikely counters.
28238 Remove unnecessary quoting from diagnostics. Add an informational
28240 (add_bytes): Remove.
28241 (pass_sprintf_length::compute_format_length): Simplify.
28242 (try_substitute_return_value): Handle likely and unlikely counters.
28244 2017-01-26 Carl Love <cel@us.ibm.com>
28246 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28247 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28249 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28252 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28253 endianess for subregs into account.
28254 * lra-constraints.c (lra_constraints): Do risky transformations
28255 always on the first iteration.
28256 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28257 last_call_used_reg_set.
28258 (process_bb_lives): Define and use last_call_used_reg_set.
28259 * lra.c (lra): Always continue after lra_constraints on the first
28262 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28264 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28266 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28268 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28270 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28271 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28272 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28273 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28274 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28275 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28276 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28277 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28278 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28280 2017-01-26 Marek Polacek <polacek@redhat.com>
28283 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28284 for the third operand.
28286 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28288 PR middle-end/79236
28289 * omp-low.c (struct omp_context): Add simt_stmt field.
28290 (scan_omp_for): Return omp_context *.
28291 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28292 context to the _simt_ SIMD stmt.
28293 (lower_omp_for): For combined SIMD with sibling _simt_
28294 SIMD, make sure to use the same decls in _looptemp_
28295 clauses as in the sibling.
28297 2017-01-26 David Sherwood <david.sherwood@arm.com>
28299 PR middle-end/79212
28300 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28303 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28306 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28307 emit fld b; fld a; if possible.
28309 * brig-builtins.def: Update copyright years.
28310 * config/arm/arm_acle_builtins.def: Update copyright years.
28312 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28315 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28316 constraint instead of o for the stxsd instruction.
28318 2017-01-25 Carl Love <cel@us.ibm.com>
28320 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28321 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28323 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28325 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28327 2017-01-25 Richard Biener <rguenther@suse.de>
28329 PR tree-optimization/69264
28330 * target.def (vector_alignment_reachable): Improve documentation.
28331 * doc/tm.texi: Regenerate.
28332 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28334 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28335 earlier changes with respect to TYPE_USER_ALIGN.
28336 (vector_alignment_reachable_p): Likewise. Improve dumping.
28338 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28341 * config/arm/arm.md (xordi3): Force constant operand into a register
28344 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28346 * doc/invoke.texi (-fstore-merging): Correct default optimization
28347 levels at which it is enabled.
28348 (-O): Move -fstore-merging from list to...
28351 2017-01-25 Richard Biener <rguenther@suse.de>
28354 * omp-expand.c: Include debug.h.
28355 (expand_omp_taskreg): Make sure to generate early debug before
28356 outlining anything from a function.
28357 (expand_omp_target): Likewise.
28358 (grid_expand_target_grid_body): Likewise.
28360 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28363 * asan.c (get_translation_unit_decl): New function.
28364 (asan_add_global): Extract modules file name from globals
28365 TRANSLATION_UNIT_DECL name.
28367 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28370 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28371 for long calls with APCS frame and VFP.
28373 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28375 * cfg.c (original_copy_tables_initialized_p): New function.
28376 * cfg.h (original_copy_tables_initialized_p): New decl.
28377 * cfgrtl.c (relink_block_chain): Guard the call to
28378 free_original_copy_tables with a call to
28379 original_copy_tables_initialized_p.
28380 * cgraph.h (symtab_node::native_rtl_p): New decl.
28381 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28382 (symtab_node::needed_p): Don't assert for early assembly output
28383 for __RTL functions.
28384 (cgraph_node::finalize_function): Set "force_output" for __RTL
28386 (cgraph_node::analyze): Bail out early for __RTL functions.
28387 (analyze_functions): Update assertion to support __RTL functions.
28388 (cgraph_node::expand): Bail out early for __RTL functions.
28389 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28391 * function.h (struct function): Update comment for field
28393 * gimple-expr.c: Include "tree-pass.h".
28394 (gimple_has_body_p): Return false for __RTL functions.
28395 * Makefile.in (OBJS): Add run-rtl-passes.o.
28396 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28398 (gcc::pass_manager::get_clean_slate): New accessor.
28399 * passes.c: Include "insn-addr.h".
28400 (should_skip_pass_p): Add logging. Update logic for running
28401 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28402 property-provider override so it is only done for gimple passes.
28404 (skip_pass): New function.
28405 (execute_one_pass): Call skip_pass when skipping passes.
28406 * read-md.c (md_reader::read_char): Support filtering
28407 the input to a subset of line numbers.
28408 (md_reader::md_reader): Initialize fields
28409 m_first_line and m_last_line.
28410 (md_reader::read_file_fragment): New function.
28411 * read-md.h (md_reader::read_file_fragment): New decl.
28412 (md_reader::m_first_line): New field.
28413 (md_reader::m_last_line): New field.
28414 * read-rtl-function.c (function_reader::create_function): Only
28415 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28416 curr_properties. Set DECL_INITIAL to a dummy block.
28417 (read_rtl_function_body_from_file_range): New function.
28418 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28420 * run-rtl-passes.c: New file.
28421 * run-rtl-passes.h: New file.
28423 2017-01-24 Jeff Law <law@redhat.com>
28425 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28428 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28430 PR tree-optimization/79159
28431 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28432 (record_nonwrapping_iv): Improve boundary using above function if no
28433 value range information.
28435 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28436 Martin Jambor <mjambor@suse.cz>
28438 * brig-builtins.def: New file.
28439 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28440 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28441 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28442 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28443 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28444 * builtin-types.def (BT_INT8): New.
28445 (BT_INT16): Likewise.
28446 (BT_UINT8): Likewise.
28447 (BT_UINT16): Likewise.
28448 (BT_FN_ULONG): Likewise.
28449 (BT_FN_UINT_INT): Likewise.
28450 (BT_FN_UINT_ULONG): Likewise.
28451 (BT_FN_UINT_LONG): Likewise.
28452 (BT_FN_UINT_PTR): Likewise.
28453 (BT_FN_ULONG_PTR): Likewise.
28454 (BT_FN_INT8_FLOAT): Likewise.
28455 (BT_FN_INT16_FLOAT): Likewise.
28456 (BT_FN_UINT32_FLOAT): Likewise.
28457 (BT_FN_UINT16_FLOAT): Likewise.
28458 (BT_FN_UINT8_FLOAT): Likewise.
28459 (BT_FN_UINT64_FLOAT): Likewise.
28460 (BT_FN_UINT16_UINT32): Likewise.
28461 (BT_FN_UINT32_UINT16): Likewise.
28462 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28463 (BT_FN_INT_PTR_INT): Likewise.
28464 (BT_FN_UINT_PTR_UINT): Likewise.
28465 (BT_FN_LONG_PTR_LONG): Likewise.
28466 (BT_FN_ULONG_PTR_ULONG): Likewise.
28467 (BT_FN_VOID_UINT64_UINT64): Likewise.
28468 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28469 (BT_FN_INT8_INT8_INT8): Likewise.
28470 (BT_FN_INT16_INT16_INT16): Likewise.
28471 (BT_FN_INT_INT_INT): Likewise.
28472 (BT_FN_UINT_FLOAT_UINT): Likewise.
28473 (BT_FN_FLOAT_UINT_UINT): Likewise.
28474 (BT_FN_ULONG_UINT_UINT): Likewise.
28475 (BT_FN_ULONG_UINT_PTR): Likewise.
28476 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28477 (BT_FN_UINT_UINT_UINT): Likewise.
28478 (BT_FN_VOID_UINT_PTR): Likewise.
28479 (BT_FN_UINT_UINT_PTR: Likewise.
28480 (BT_FN_UINT32_UINT64_PTR): Likewise.
28481 (BT_FN_INT_INT_UINT_UINT): Likewise.
28482 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28483 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28484 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28485 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28486 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28487 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28488 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28489 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28490 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28491 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28492 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28493 * doc/frontends.texi: List BRIG FE.
28494 * doc/install.texi (Testing): Add BRIG tesring requirements.
28495 * doc/invoke.texi (Overall Options): Mention BRIG.
28496 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28498 2017-01-24 Richard Biener <rguenther@suse.de>
28500 PR translation/79208
28501 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28503 2017-01-24 Martin Jambor <mjambor@suse.cz>
28506 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28507 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28508 and known_contexts.
28510 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28512 PR middle-end/79123
28513 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28514 casts from signed to unsigned really don't have a range.
28516 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28518 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28519 GMP_RNDx for compatiblity.
28521 2017-01-24 Martin Liska <mliska@suse.cz>
28524 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28525 that would prevent us to call alloca with -1 as argument.
28527 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28529 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28530 Avoid -Wformat-security warning.
28532 2017-01-23 Andrew Pinski <apinski@cavium.com>
28534 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28537 2017-01-23 Martin Sebor <msebor@redhat.com>
28539 PR middle-end/78703
28540 * gimple-ssa-sprintf.c (warn_level): New global.
28541 (format_integer): Use it here and throughout the rest of the file.
28542 Use the same switch to compute sign as base.
28543 (maybe_warn): New function.
28544 (format_directive): Factor out warnings into maybe_warn.
28545 Add debugging output. Use warn_level.
28546 (add_bytes): Use warn_level.
28547 (pass_sprintf_length::compute_format_length): Add debugging output.
28548 (try_substitute_return_value): Same.
28549 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28551 PR middle-end/78703
28552 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28553 (struct fmtresult, format_integer, format_floating): Adjust.
28554 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28555 (get_string_length, format_string,format_directive): Same.
28556 (pass_sprintf_length::compute_format_length): Same.
28557 (try_substitute_return_value): Simplify slightly.
28559 PR middle-end/78703
28560 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28561 (fmtresult::operator+=): Outlined.
28562 (struct fmtresult): Add ctors.
28563 (struct conversion_spec): Rename...
28564 (struct directive): ...to this. Add and remove data members.
28565 (directive::set_width, directive::set_precision): New functions.
28566 (format_percent): Use fmtresult ctor.
28567 (get_width_and_precision): Remove.
28568 (format_integer): Make naming changes. Avoid computing width and
28570 (format_floating): Same. Adjust indentation.
28571 (format_character, format_none): New functions.
28572 (format_string): Moved character handling to format_character.
28573 (format_directive): Remove arguments, change return type.
28574 (parse_directive): New function.
28575 (pass_sprintf_length::compute_format_length): Move directive
28576 parsing to parse_directive.
28578 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28580 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28581 (assign_assembler_name_if_needed): ... this.
28582 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28583 (assign_assembler_name_if_needed): ... this.
28584 (free_lang_data_in_cgraph): Adjust callers.
28585 * cgraphunit.c (cgraph_node::analyze): Likewise.
28586 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28589 2017-01-23 Richard Biener <rguenther@suse.de>
28591 PR tree-optimization/79088
28592 PR tree-optimization/79188
28593 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28594 resetting loop bounds after last path deletion. Reset loop
28595 bounds of the target loop, make code match the comments.
28596 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28597 Make sure loops need no fixups.
28599 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28601 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28602 exponent support with double type for first argument.
28603 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28604 type returned by __builtin_vec_extract_sig,
28605 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28606 functions from "vector int" to "vector unsigned int" or from
28607 "vector long long int" to "vector unsigned long long int".
28608 Changed type returned by __builtin_vec_extract_exp,
28609 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_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 return type of __builtin_vec_test_data_class,
28613 __builtin_vec_test_data_class_sp, and
28614 __builtin_vec_test_data_class_dp from "vector int" to
28615 "vector bool int" or from "vector long long int" to "vector bool
28616 long long int" and changed second argument type from "unsigned
28617 int" to "int". Added new overloaded function forms "vector float
28618 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28619 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28620 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28621 double, vector unsigned long long int)" and "vector double
28622 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28623 long int)". Changed return type of
28624 __builtin_scalar_test_data_class and
28625 __builtin_scalar_test_data_class_sp and
28626 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28627 int" and changed second argument from "unsigned int" to "int".
28628 Changed type returned by __builtin_scalar_test_neg,
28629 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28630 from "int" to "bool int". Added new overloaded function form
28631 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28632 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28633 exponent double-precision with floating point first argument.
28634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28635 documentation of scalar_test_data_class, scalar_test_neg,
28636 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28637 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28638 vec_test_data_class built-in functions to reflect refinements in
28639 their type signatures.
28641 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28643 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28645 (aarch64_elf_asm_destructor): Likewise.
28647 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28649 PR rtl-optimization/78634
28650 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28651 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28652 * ifcvt.c (noce_try_cmove): Add missing cost check.
28654 PR rtl-optimization/71724
28655 * combine.c (if_then_else_cond): Look for situations where it is
28656 beneficial to undo the work of one of the recursive calls.
28658 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28660 PR tree-optimization/70754
28661 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28662 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28663 combined stmt before it if not NULL.
28664 (combine_chains): Process refs reversely and compute dominance point
28667 2017-01-23 Martin Liska <mliska@suse.cz>
28669 PR tree-optimization/79196
28670 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28671 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28673 (strlen_optimize_stmt): Call the renamed function.
28675 2017-01-23 Michael Matz <matz@suse.de>
28677 PR tree-optimization/78384
28678 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28680 2017-01-23 Richard Biener <rguenther@suse.de>
28682 PR tree-optimization/79186
28683 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28684 both incoming edges before moving an assert.
28686 2017-01-23 Martin Jambor <mjambor@suse.cz>
28688 * ipa-prop.c (load_from_param_1): Removed.
28689 (load_from_unmodified_param): Bits from load_from_param_1 put back
28691 (load_from_param): Removed.
28692 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28693 with stmt. Reverted back to use of load_from_unmodified_param.
28695 2017-01-23 Martin Jambor <mjambor@suse.cz>
28698 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28699 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28700 field a pointer to garbage collected vector, mark lattices and
28701 ipcp_orig_node with GTY((skip)).
28702 (ipa_get_param_count): Adjust to descriptors being a pointer.
28703 (ipa_get_param): Likewise.
28704 (ipa_get_type): Likewise.
28705 (ipa_get_param_move_cost): Likewise.
28706 (ipa_set_param_used): Likewise.
28707 (ipa_get_controlled_uses): Likewise.
28708 (ipa_set_controlled_uses): Likewise.
28709 (ipa_is_param_used): Likewise.
28710 (ipa_node_params_t): Move into garbage collector. New methods insert
28712 (ipa_node_params_sum): Annotate wth GTY(()).
28713 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28715 (ipa_load_from_parm_agg): Adjust declaration.
28716 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28717 * ipa-profile.c (ipa_profile): Likewise.
28718 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28719 (ipa_populate_param_decls): Make descriptors parameter garbage
28721 (ipa_dump_param): Adjust to descriptors being a pointer.
28722 (ipa_alloc_node_params): Likewise.
28723 (ipa_initialize_node_params): Likewise.
28724 (load_from_param_1): Make descriptors parameter garbage collected.
28725 (load_from_unmodified_param): Likewise.
28726 (load_from_param): Likewise.
28727 (ipa_load_from_parm_agg): Likewise.
28728 (ipa_node_params::~ipa_node_params): Removed.
28729 (ipa_free_all_node_params): Remove call to delete operator.
28730 (ipa_node_params_t::insert): New.
28731 (ipa_node_params_t::remove): Likewise.
28732 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28733 copy known_csts and known_contexts vectors.
28734 (ipa_read_node_info): Adjust to descriptors being a pointer.
28735 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28737 (ipcp_transform_function): Make descriptors variable garbage
28740 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28742 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28743 * config/i386/avx512dqintrin.h: Ditto.
28744 * config/i386/avx512fintrin.h: Ditto.
28745 * config/i386/i386.c: Handle new builtins.
28746 * config/i386/i386-builtin.def: Add new builtins.
28747 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28748 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28750 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28751 Martin Liska <mliska@suse.cz>
28753 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28754 * asan.c (asan_expand_poison_ifn): Support stores and use
28755 appropriate ASAN report function.
28756 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28757 * internal-fn.def (ASAN_POISON_USE): Declare.
28758 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28759 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28760 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28761 ASAN_POISON calls w/o LHS.
28762 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28763 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28764 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28765 * gimplify.c (asan_poison_variables): Add attribute
28766 use_after_scope_memory to variables that really needs to live
28768 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28769 having the attribute.
28771 2017-01-23 Martin Liska <mliska@suse.cz>
28773 * asan.c (create_asan_shadow_var): New function.
28774 (asan_expand_poison_ifn): Likewise.
28775 * asan.h (asan_expand_poison_ifn): New declaration.
28776 * internal-fn.c (expand_ASAN_POISON): Likewise.
28777 * internal-fn.def (ASAN_POISON): New builtin.
28778 * sanopt.c (pass_sanopt::execute): Expand
28779 asan_expand_poison_ifn.
28780 * tree-inline.c (copy_decl_for_dup_finish): Make function
28782 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28783 * tree-ssa.c (is_asan_mark_p): New function.
28784 (execute_update_addresses_taken): Rewrite local variables
28785 (identified just by use-after-scope as addressable) into SSA.
28787 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28789 * doc/install.texi (Specific): opensource.apple.com uses https
28790 now. Remove trailing slash.
28792 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28794 * README.Portability: Remove note on an Irix compatibility issue.
28796 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28798 * gcov.c (INCLUDE_ALGORITHM): Define.
28799 (INCLUDE_VECTOR): Define.
28800 No longer include <vector> and <algorithm> directly.
28802 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28804 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28806 * doc/invoke.texi (Code Gen Options): Ditto.
28808 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28811 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28813 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28815 rtl-optimization/79125
28816 * cprop.c (local_cprop_pass): Handle cases where we make an
28817 unconditional trap.
28819 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28823 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28824 read from, for big endian.
28826 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28828 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28829 register pauth builtins for LP64 only.
28831 2017-01-20 Marek Polacek <polacek@redhat.com>
28834 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28837 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28839 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28841 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28842 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28843 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28845 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28848 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28849 in big-endian mode when they are not a single duplicated value.
28851 2017-01-20 Richard Biener <rguenther@suse.de>
28853 * BASE-VER: Bump to 7.0.1.
28855 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28857 * omp-low.c (omplow_simd_context): New struct. Use it...
28858 (lower_rec_simd_input_clauses): ...here and...
28859 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28860 references to idx, lane, max_vf, is_simt.
28862 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28864 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28867 2017-01-20 Martin Jambor <mjambor@suse.cz>
28869 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28870 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28872 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28873 (GTFILES): Rename hsa.c to hsa-common.c.
28874 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28875 * hsa-dump.c: Likewise.
28876 * hsa-gen.c: Likewise.
28877 * hsa-regalloc.c: Likewise.
28878 * ipa-hsa.c: Likewise.
28879 * omp-expand.c: Likewise.
28880 * omp-low.c: Likewise.
28881 * toplev.c: Likewise.
28883 2017-01-20 Marek Polacek <polacek@redhat.com>
28886 * doc/invoke.texi: Document -Wduplicated-branches.
28887 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28888 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28889 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28890 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28891 return 0 only when not OEP_LEXICOGRAPHIC.
28892 (fold_build_cleanup_point_expr): Use the expression
28893 location when building CLEANUP_POINT_EXPR.
28894 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28895 * tree.c (add_expr): Handle error_mark_node.
28897 2017-01-20 Martin Liska <mliska@suse.cz>
28900 * tree-profile.c (init_ic_make_global_vars): Do not call
28902 (gimple_init_gcov_profiler): Likewise.
28904 2017-01-20 Martin Liska <mliska@suse.cz>
28907 * cgraph.h (maybe_create_reference): Remove argument and
28909 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28911 * ipa-cp.c (create_specialized_node): Likewise.
28912 * symtab.c (symtab_node::maybe_create_reference): Handle
28913 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28915 2017-01-20 Martin Liska <mliska@suse.cz>
28917 * read-rtl-function.c (function_reader::create_function): Use
28918 build_decl instread of build_decl_stat.
28920 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28922 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28923 * config/i386/avx512dqintrin.h: Ditto.
28924 * config/i386/avx512fintrin.h: Ditto.
28925 * config/i386/i386-builtin-types.def: Add new types.
28926 * config/i386/i386.c: Handle new types.
28927 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28928 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28929 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28930 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28931 (__builtin_ia32_kshiftridi): New.
28932 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28934 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28938 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28939 define to rs6000_init_stack_protect_guard.
28940 (rs6000_init_stack_protect_guard): New function.
28942 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28943 Yunqiang Su <yunqiang.su@imgtec.com>
28945 * config.gcc (supported_defaults): Add madd4.
28946 (with_madd4): Add validation.
28947 (all_defaults): Add madd4.
28948 * config/mips/mips.opt (mmadd4): New option.
28949 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28951 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28953 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28954 (ISA_HAS_FUSED_MADD4): Likewise.
28955 * doc/invoke.texi (-mmadd4): Document the new option.
28956 * doc/install.texi (--with-madd4): Document the new option.
28958 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28960 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28961 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28962 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28963 (aarch64_init_pauth_hint_builtins): New.
28964 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28965 (aarch64_expand_builtin): Expand new builtins.
28967 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28969 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28970 * combine-stack-adj.c (no_unhandled_cfa): Handle
28971 REG_CFA_TOGGLE_RA_MANGLE.
28972 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28973 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28974 info for return address signing.
28975 (aarch64_expand_epilogue): Likewise.
28977 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28979 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28980 * config/aarch64/aarch64-protos.h
28981 (aarch64_return_address_signing_enabled): New declaration.
28982 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28984 (aarch64_expand_prologue): Sign return address before it's pushed onto
28986 (aarch64_expand_epilogue): Authenticate return address fetched from
28988 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28989 (aarch64_attributes): New function attributes for "sign-return-address".
28990 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28991 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28992 ("*do_return"): Generate combined instructions according to key index.
28993 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28994 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28996 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28997 * config/aarch64/aarch64.opt (msign-return-address=): New.
28998 * doc/extend.texi (AArch64 Function Attributes): Documents
28999 "sign-return-address=".
29000 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29002 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29004 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29005 overall option summary.
29007 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29009 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29010 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29011 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29012 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29014 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29016 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29017 -mpower9-minmax by default for -mcpu=power9.
29018 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29019 128-bit floating point.
29021 2017-01-20 Alan Modra <amodra@gmail.com>
29023 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29024 optimizing for size.
29026 2017-01-20 Alan Modra <amodra@gmail.com>
29029 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29030 for strcmp and strncmp from corresponding builtin decl.
29032 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29034 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29035 instead of i386/rtems-64.h.
29036 * config/i386/rtems-64.h: Remove.
29038 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29042 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29044 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29046 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29048 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29049 Change int to HOST_WIDE_INT.
29050 * config/aarch64/aarch64-protos.h
29051 (aarch64_simd_gen_const_vector_dup): Likewise.
29052 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29054 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29056 * langhooks-def.h (lhd_type_for_size): New decl.
29057 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29058 * langhooks.c (lhd_type_for_size): New function, taken from
29061 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29063 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29064 define_bypass for CR latency.
29065 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29066 (power9-alu2): Add define_bypass for CR latency.
29068 (power9-mul): Update insn latency.
29069 (power9-mul-compare): Update insn latency, bypass latency and remove
29072 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29074 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29076 * config/aarch64/aarch64.md
29077 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29078 aarch64_nopcrelative_literal_loads.
29079 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29081 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29083 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29084 TARGET_LOONGSON_3A.
29085 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29087 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29090 * config.gcc (supported_defaults): Add lxc1-sxc1.
29091 (with_lxc1_sxc1): Add validation.
29092 (all_defaults): Add lxc1-sxc1.
29093 * config/mips/mips.opt (mlxc1-sxc1): New option.
29094 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29096 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29097 __mips_no_lxc1_sxc1.
29098 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29099 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29100 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29102 2017-01-19 Richard Biener <rguenther@suse.de>
29104 PR tree-optimization/72488
29105 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29106 sure to restore SSA info.
29107 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29109 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29111 PR rtl-optimization/79121
29112 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29113 of the inner type when shifting an extended value.
29115 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29118 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29119 interposable aliases.
29121 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29124 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29125 Use the evmergelohi instruction.
29126 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29127 (mov_sitf_e500_subreg8_2_be): Likewise.
29128 (mov_sitf_e500_subreg12_2_le): Likewise.
29129 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29130 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29131 (mov_sitf_e500_subreg8_2_le): Likewise.
29132 (mov_sitf_e500_subreg12_2_be): Likewise.
29134 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29136 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29137 attribute from vecsimple to vecperm.
29138 (altivec_vbpermq2): Likewise.
29140 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29143 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29145 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29146 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29147 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29148 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29149 case where N arg is SIZE_MAX.
29150 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29151 (cmpstrsi): Add pattern.
29153 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29156 __builtin_vec_revb builtins.
29157 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29158 built-in functions to support generation of the ISA 3.0 XXBR<x>
29159 vector byte reverse instructions.
29160 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29161 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29162 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29163 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29164 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29165 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29166 (P9V_BUILTIN_VEC_REVB): Likewise.
29167 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29168 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29169 (p9_xxbrq_v16qi): Likewise.
29170 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29171 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29172 (p9_xxbrh_v8hi): Likewise.
29173 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29174 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29175 vec_revb built-in functions.
29177 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29179 PR rtl-optimization/78952
29180 * config/i386/i386.md (any_extract): New code iterator.
29181 (*insvqi_2): Use any_extract for source operand.
29182 (*insvqi_3): Use any_shiftrt for source operand.
29184 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29186 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29188 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29190 2017-01-18 Matthias Klose <doko@ubuntu.com>
29192 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29194 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29196 * config/rs6000/altivec.h (vec_bperm): Change #define.
29197 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29198 (altivec_vbpermq2): New define_insn.
29199 (altivec_vbpermd): Likewise.
29200 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29201 function interface.
29202 (VBPERMD): Likewise.
29203 (VBPERM): New polymorphic function interface.
29204 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29205 Add entries for P9V_BUILTIN_VEC_VBPERM.
29206 * doc/extend.texi: Add interfaces for vec_bperm.
29208 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29210 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29211 first letter of error messages.
29212 (s390_resolve_overloaded_builtin): Likewise.
29213 * config/s390/s390.c (s390_expand_builtin): Likewise.
29214 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29215 (s390_valid_target_attribute_inner_p): Likewise.
29216 * config/s390/s390.md ("tabort"): Likewise.
29218 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29220 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29221 (ISA_AVOID_DIV_HILO): New macro.
29222 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29223 (ISA_HAS_DDIV): Likewise.
29225 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29227 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29229 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29231 * doc/invoke.texi (fabi-version): Spelling fix.
29233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29236 * doc/invoke.texi (fabi-version): Mention mangling fix for
29239 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29242 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29244 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29247 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29250 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29252 (stack_protect_test): Ditto.
29253 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29254 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29256 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29257 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29258 -mstack-protector-guard-offset=.
29259 (RS/6000 and PowerPC Options): Ditto.
29261 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29263 * config/i386/i386.h (MASK_CLASS_P): New define.
29264 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29265 there are no registers from different register sets also when
29266 mask registers are used. Update function comment.
29267 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29268 to (*k/*r) and (*k/*km) alternatives.
29270 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29272 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29273 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29274 (EH_RETURN_HANDLER_RTX): New define.
29275 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29276 Force frame pointer in EH return functions.
29277 (aarch64_expand_epilogue): Add barrier for eh_return.
29278 (aarch64_final_eh_return_addr): Remove.
29279 (aarch64_eh_return_handler_rtx): New function.
29280 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29282 (aarch64_eh_return_handler_rtx): New prototype.
29284 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29286 * config/rs6000/altivec.h (vec_rlmi): New #define.
29287 (vec_vrlnm): Likewise.
29288 (vec_rlnm): Likewise.
29289 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29290 (UNSPEC_VRLNM): Likewise.
29291 (VIlong): New mode iterator.
29292 (altivec_vrl<VI_char>mi): New define_insn.
29293 (altivec_vrl<VI_char>nm): Likewise.
29294 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29296 (VRLDNM): Likewise.
29297 (RLNM): New polymorphic function entry.
29298 (VRLWMI): New monomorphic function entry.
29299 (VRLDMI): Likewise.
29300 (RLMI): New polymorphic function entry.
29301 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29302 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29303 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29306 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29309 * dwarf2out.c (field_byte_offset): Restore the
29310 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29311 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29312 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29315 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29318 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29320 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29323 * dwarf2out.c (add_data_member_location_attribute): For constant
29324 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29325 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29326 DW_AT_byte_size attributes.
29328 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29330 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29331 after forcing to constant memory when the code model is medium.
29333 2017-01-17 Julia Koval <julia.koval@intel.com>
29336 * config/i386/avx512fintrin.h
29337 (_mm512_i32gather_ps): Change __addr type to void const*.
29338 (_mm512_mask_i32gather_ps): Ditto.
29339 (_mm512_i32gather_pd): Ditto.
29340 (_mm512_mask_i32gather_pd): Ditto.
29341 (_mm512_i64gather_ps): Ditto.
29342 (_mm512_mask_i64gather_ps): Ditto.
29343 (_mm512_i64gather_pd): Ditto.
29344 (_mm512_mask_i64gather_pd): Ditto.
29345 (_mm512_i32gather_epi32): Ditto.
29346 (_mm512_mask_i32gather_epi32): Ditto.
29347 (_mm512_i32gather_epi64): Ditto.
29348 (_mm512_mask_i32gather_epi64): Ditto.
29349 (_mm512_i64gather_epi32): Ditto.
29350 (_mm512_mask_i64gather_epi32): Ditto.
29351 (_mm512_i64gather_epi64): Ditto.
29352 (_mm512_mask_i64gather_epi64): Ditto.
29353 (_mm512_i32scatter_ps): Change __addr type to void*.
29354 (_mm512_mask_i32scatter_ps): Ditto.
29355 (_mm512_i32scatter_pd): Ditto.
29356 (_mm512_mask_i32scatter_pd): Ditto.
29357 (_mm512_i64scatter_ps): Ditto.
29358 (_mm512_mask_i64scatter_ps): Ditto.
29359 (_mm512_i64scatter_pd): Ditto.
29360 (_mm512_mask_i64scatter_pd): Ditto.
29361 (_mm512_i32scatter_epi32): Ditto.
29362 (_mm512_mask_i32scatter_epi32): Ditto.
29363 (_mm512_i32scatter_epi64): Ditto.
29364 (_mm512_mask_i32scatter_epi64): Ditto.
29365 (_mm512_i64scatter_epi32): Ditto.
29366 (_mm512_mask_i64scatter_epi32): Ditto.
29367 (_mm512_i64scatter_epi64): Ditto.
29368 (_mm512_mask_i64scatter_epi64): Ditto.
29369 * config/i386/avx512pfintrin.h
29370 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29371 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29372 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29373 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29374 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29375 (_mm512_prefetch_i32scatter_ps): Ditto.
29376 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29377 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29378 (_mm512_prefetch_i64scatter_pd): Ditto.
29379 (_mm512_prefetch_i64scatter_ps): Ditto.
29380 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29381 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29382 * config/i386/avx512vlintrin.h
29383 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29384 (_mm_mmask_i32gather_ps): Ditto.
29385 (_mm256_mmask_i32gather_pd): Ditto.
29386 (_mm_mmask_i32gather_pd): Ditto.
29387 (_mm256_mmask_i64gather_ps): Ditto.
29388 (_mm_mmask_i64gather_ps): Ditto.
29389 (_mm256_mmask_i64gather_pd): Ditto.
29390 (_mm_mmask_i64gather_pd): Ditto.
29391 (_mm256_mmask_i32gather_epi32): Ditto.
29392 (_mm_mmask_i32gather_epi32): Ditto.
29393 (_mm256_mmask_i32gather_epi64): Ditto.
29394 (_mm_mmask_i32gather_epi64): Ditto.
29395 (_mm256_mmask_i64gather_epi32): Ditto.
29396 (_mm_mmask_i64gather_epi32): Ditto.
29397 (_mm256_mmask_i64gather_epi64): Ditto.
29398 (_mm_mmask_i64gather_epi64): Ditto.
29399 (_mm256_i32scatter_ps): Change __addr type to void*.
29400 (_mm256_mask_i32scatter_ps): Ditto.
29401 (_mm_i32scatter_ps): Ditto.
29402 (_mm_mask_i32scatter_ps): Ditto.
29403 (_mm256_i32scatter_pd): Ditto.
29404 (_mm256_mask_i32scatter_pd): Ditto.
29405 (_mm_i32scatter_pd): Ditto.
29406 (_mm_mask_i32scatter_pd): Ditto.
29407 (_mm256_i64scatter_ps): Ditto.
29408 (_mm256_mask_i64scatter_ps): Ditto.
29409 (_mm_i64scatter_ps): Ditto.
29410 (_mm_mask_i64scatter_ps): Ditto.
29411 (_mm256_i64scatter_pd): Ditto.
29412 (_mm256_mask_i64scatter_pd): Ditto.
29413 (_mm_i64scatter_pd): Ditto.
29414 (_mm_mask_i64scatter_pd): Ditto.
29415 (_mm256_i32scatter_epi32): Ditto.
29416 (_mm256_mask_i32scatter_epi32): Ditto.
29417 (_mm_i32scatter_epi32): Ditto.
29418 (_mm_mask_i32scatter_epi32): Ditto.
29419 (_mm256_i32scatter_epi64): Ditto.
29420 (_mm256_mask_i32scatter_epi64): Ditto.
29421 (_mm_i32scatter_epi64): Ditto.
29422 (_mm_mask_i32scatter_epi64): Ditto.
29423 (_mm256_i64scatter_epi32): Ditto.
29424 (_mm256_mask_i64scatter_epi32): Ditto.
29425 (_mm_i64scatter_epi32): Ditto.
29426 (_mm_mask_i64scatter_epi32): Ditto.
29427 (_mm256_i64scatter_epi64): Ditto.
29428 (_mm256_mask_i64scatter_epi64): Ditto.
29429 (_mm_i64scatter_epi64): Ditto.
29430 (_mm_mask_i64scatter_epi64): Ditto.
29431 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29432 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29433 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29434 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29435 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29436 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29437 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29438 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29439 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29440 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29441 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29442 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29443 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29444 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29445 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29446 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29447 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29448 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29449 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29450 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29451 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29452 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29453 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29454 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29455 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29456 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29457 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29458 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29459 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29460 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29461 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29462 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29463 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29464 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29465 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29466 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29467 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29468 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29469 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29470 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29471 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29472 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29473 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29474 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29475 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29476 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29477 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29478 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29479 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29480 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29481 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29482 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29483 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29484 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29485 definitions accordingly.
29487 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29488 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29491 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29494 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29497 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29498 hard regs for inner regno.
29500 2017-01-17 Martin Liska <mliska@suse.cz>
29503 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29504 assumption and add comment.
29506 2017-01-17 Nathan Sidwell <nathan@acm.org>
29508 * ipa-visibility.c (localize_node): New function, broken out of ...
29509 (function_and_variable_visibility): ... here. Call it.
29511 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29513 PR middle-end/77445
29514 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29515 correctly set frequency of oudgoing edge.
29516 (duplicate_thread_path): Fix profile updating.
29518 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29521 * configure.ac: Add GCC_BASE_VER.
29522 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29523 version from BASE-VER file.
29524 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29525 (gcc.o): Depend on $(BASEVER).
29526 * common.opt (dumpfullversion): New option.
29527 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29528 * doc/invoke.texi: Document -dumpfullversion.
29529 * doc/install.texi: Document --with-gcc-major-version-only.
29530 * configure: Regenerated.
29532 2017-01-17 Richard Biener <rguenther@suse.de>
29534 PR tree-optimization/71433
29535 * tree-vrp.c (register_new_assert_for): Merge same asserts
29536 on all incoming edges.
29537 (process_assert_insertions_for): Handle insertions at the
29540 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29542 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29543 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29545 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29548 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29551 2017-01-17 Alan Modra <amodra@gmail.com>
29554 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29555 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29556 symbolic stack limit when pic.
29558 2017-01-16 Martin Sebor <msebor@redhat.com>
29560 PR tree-optimization/78608
29561 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29563 2017-01-16 Jeff Law <law@redhat.com>
29566 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29567 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29568 for several include directories that may be relative to sysroot.
29569 * config/i386/x-mingw32 (gplus_includedir): Define.
29570 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29571 (native_system_includedir): Likewise.
29572 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29573 override if TARGET_SYSTEM_ROOT is defined.
29574 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29576 PR tree-optimization/79090
29577 PR tree-optimization/33562
29578 PR tree-optimization/61912
29579 PR tree-optimization/77485
29580 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29581 and computed trims into the dump file.
29583 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29585 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29587 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29590 * gimplify.c (gimplify_init_constructor): If want_value and
29591 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29595 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29596 sequence. Formatting fixes.
29597 (doloop_optimize): Formatting fixes.
29600 * gcc.c (debug_level_greater_than_spec_func): New function.
29601 (static_spec_functions): Add debug-level-gt spec function.
29602 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29604 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29605 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29606 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29607 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29608 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29609 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29611 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29613 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29614 QImode fixups to general and mask registers only.
29616 2017-01-16 Carl Love <cel@us.ibm.com>
29618 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29619 for built-in functions
29620 vector signed char vec_nabs (vector signed char)
29621 vector signed short vec_nabs (vector signed short)
29622 vector signed int vec_nabs (vector signed int)
29623 vector signed long long vec_nabs (vector signed long long)
29624 vector float vec_nabs (vector float)
29625 vector double vec_nabs (vector double)
29626 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29628 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29629 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29630 * doc/extend.texi: Update the documentation file for the new built-in
29633 2017-01-16 Martin Sebor <msebor@redhat.com>
29635 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29638 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29640 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29641 UNSPEC_VSX__XXSPLTD to require special splat handling.
29643 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29646 * system.h: Poison strndup.
29648 2017-01-16 Alan Modra <amodra@gmail.com>
29651 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29654 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29656 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29658 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29660 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29661 call recog here. Assert that INSN_CODE (insn) is non-negative.
29663 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29666 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29668 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29669 in the currently scheduled RTL fragment.
29671 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29673 PR rtl-optimization/78751
29674 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29677 2017-01-14 Jeff Law <law@redhat.com>
29679 PR tree-optimization/79090
29680 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29681 variable length stores.
29682 (compute_trims): Delete dead assignment to *trim_tail.
29683 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29686 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29688 PR rtl-optimization/78626
29689 PR rtl-optimization/78727
29690 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29691 of a block, and split such blocks after everything else is finished.
29693 2017-01-14 Alan Modra <amodra@gmail.com>
29696 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29697 target legitimate_combined_insn.
29698 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29699 (rs6000_legitimate_combined_insn): New function.
29700 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29702 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29703 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29704 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29706 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29708 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29710 2017-01-13 Jeff Law <law@redhat.com>
29712 PR tree-optimization/33562
29713 PR tree-optimization/61912
29714 PR tree-optimization/77485
29715 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29717 (delete_dead_assignment): Likewise.
29718 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29719 statement to delete_dead_call and delete_dead_assignment.
29721 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29724 * substring-locations.c (format_warning_va): Strengthen case 1 so
29725 that both endpoints of the substring must be within the format
29726 range for just the substring to be printed.
29728 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29730 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29731 * config/i386/i386.c (ix86_target_string): Add missing options
29732 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29733 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29734 flags_other and ix86_target_other to flags2_other. Display unknown
29736 (ix86_valid_target_attribute_inner_p): Add missing options and
29737 reorder options by implied ISAs, as in ix86_target_string.
29739 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29741 * hash-table.h (hash_table::too_empty_p): New function.
29742 (hash_table::expand): Use it.
29743 (hash_table::traverse): Likewise.
29744 (hash_table::empty_slot): Use sizeof (value_type) instead of
29745 sizeof (PTR) to convert bytes to elements. Shrink the table
29746 if the current size is excessive for the current number of
29749 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29751 * ira-costs.c (record_reg_classes): Break from the inner loop
29752 early once alt_fail is known to be true. Update outer loop
29753 handling accordingly.
29755 2017-01-13 Jeff Law <law@redhat.com>
29757 * tree-ssa-dse.c (decrement_count): New function.
29758 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29759 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29760 when we know the partially dead statement is a mem* function.
29762 PR tree-optimization/61912
29763 PR tree-optimization/77485
29764 * tree-ssa-dse.c: Include expr.h.
29765 (maybe_trim_constructor_store): New function.
29766 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29768 PR tree-optimization/33562
29769 PR tree-optimization/61912
29770 PR tree-optimization/77485
29771 * doc/invoke.texi: Document new dse-max-object-size param.
29772 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29773 * tree-ssa-dse.c: Include params.h.
29774 (dse_store_status): New enum.
29775 (initialize_ao_ref_for_dse): New, partially extracted from
29777 (valid_ao_ref_for_dse, normalize_ref): New.
29778 (setup_live_bytes_from_ref, compute_trims): Likewise.
29779 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29780 (maybe_trim_partially_dead_store): Likewise.
29781 (maybe_trim_complex_store): Likewise.
29782 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29783 Track what bytes live from the original store. Return tri-state
29784 for dead, partially dead or live.
29785 (dse_dom_walker): Add constructor, destructor and new private members.
29786 (delete_dead_call, delete_dead_assignment): New extracted from
29788 (dse_optimize_stmt): Make a member of dse_dom_walker.
29789 Use initialize_ao_ref_for_dse.
29791 PR tree-optimization/33562
29792 PR tree-optimization/61912
29793 PR tree-optimization/77485
29794 * sbitmap.h (bitmap_count_bits): Prototype.
29795 (bitmap_clear_range, bitmap_set_range): Likewise.
29796 * sbitmap.c (bitmap_clear_range): New function.
29797 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29799 2017-01-13 Martin Liska <mliska@suse.cz>
29802 * function.c (set_cfun): Add new argument force.
29803 * function.h (set_cfun): Likewise.
29804 * ipa-inline-transform.c (inline_call): Use the function when
29805 strict alising from is dropped for function we inline to.
29807 2017-01-13 Richard Biener <rguenther@suse.de>
29809 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29810 for dumping GIMPLE INTEGER_CSTs.
29812 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29814 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29815 to 201112L since C++17.
29817 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29820 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29821 if -fsanitize=kernel-address is present.
29823 2017-01-13 Richard Biener <rguenther@suse.de>
29825 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29826 as _Literal ( type ) number in case usual suffixes do not
29827 preserve all information.
29829 2017-01-13 Richard Biener <rguenther@suse.de>
29831 PR tree-optimization/77283
29832 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29833 and ssa-iterators.h.
29834 (is_feasible_trace): Implement a cost model based on joiner
29837 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29840 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29841 char or short to __float128/_Float128 directly.
29843 2017-01-12 Martin Sebor <msebor@redhat.com>
29845 to -Wformat-overflow.
29846 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29847 (min_bytes_remaining): Same.
29848 (get_string_length): Same.
29849 (format_string): Same.
29850 (format_directive): Same.
29852 (pass_sprintf_length::handle_gimple_call): Same.
29854 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29856 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29857 info.nowrite calls with no lhs that can't throw. Return bool
29858 whether gsi_remove has been called or not.
29859 (pass_sprintf_length::handle_gimple_call): Return bool whether
29860 try_substitute_return_value called gsi_remove. Formatting fix.
29861 (pass_sprintf_length::execute): Don't use gsi_remove if
29862 handle_gimple_call returned true.
29865 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29866 be removed due to side-effects, don't remove following barrier nor
29867 turn the successor edge into fallthru edge.
29869 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29872 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29873 element-reversing loads and stores as not swappable.
29875 2017-01-12 Nathan Sidwell <nathan@acm.org>
29876 Nicolai Stange <nicstange@gmail.com>
29878 * combine.c (try_combine): Don't ignore result of overlap checking
29879 loop. Combine overlap & asm check into single loop.
29881 2017-01-12 Richard Biener <rguenther@suse.de>
29883 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29884 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29886 2017-01-12 Richard Biener <rguenther@suse.de>
29888 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29889 and TS_TARGET_OPTION directly derive from TS_BASE.
29890 * tree-core.h (tree_optimization_option): Derive from tree_base.
29891 (tree_target_option): Likewise.
29893 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29895 * config/i386/i386.c (memory_address_length): Increase len
29896 only when rip_relative_addr_p returns false.
29898 2017-01-11 Julia Koval <julia.koval@intel.com>
29900 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29901 (OPTION_MASK_ISA_SGX_SET): New.
29902 (ix86_handle_option): Handle OPT_msgx.
29903 * config.gcc: Added sgxintrin.h.
29904 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29905 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29906 * config/i386/i386.c (ix86_target_string): Add -msgx.
29908 (ix86_option_override_internal): Handle new options.
29909 (ix86_valid_target_attribute_inner_p): Add sgx.
29910 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29911 * config/i386/i386.opt: Add msgx.
29912 * config/i386/sgxintrin.h: New file.
29913 * config/i386/x86intrin.h: Add sgxintrin.h.
29915 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29918 * fold-const.c (maybe_nonzero_address): Return 1 for function
29920 (tree_single_nonzero_warnv_p): Don't handle function local objects
29924 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29927 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29930 * opts.c: Include "spellcheck.h"
29931 (struct string_fragment): New struct.
29932 (struct edit_distance_traits<const string_fragment &>): New
29934 (get_closest_sanitizer_option): New function.
29935 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29937 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29939 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29941 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29942 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29943 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29944 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29945 for initial die_offset if dwarf_split_debug_info.
29946 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29947 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29949 (output_skeleton_debug_sections): Formatting fix. Use
29950 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29951 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29953 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29955 * config/arm/cortex-a53.md: Add bypasses for
29956 cortex_a53_r2f_cvt.
29957 (cortex_a53_r2f): Only use for transfers.
29958 (cortex_a53_f2r): Likewise.
29959 (cortex_a53_r2f_cvt): Add reservation for conversions.
29960 (cortex_a53_f2r_cvt): Likewise.
29962 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29964 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29965 to all inlined functions, change static to extern.
29967 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29970 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29972 (arm_assemble_integer): Likewise.
29974 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29976 * config.gcc: Use new awk script to check CPU, FPU and architecture
29977 parameters for --with-... options.
29978 * config/arm/parsecpu.awk: New file
29979 * config/arm/arm-cpus.in: New file.
29980 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29982 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29984 * config/arm/t-arm: Update dependency rules.
29985 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29986 of processing .def files.
29987 * config/arm/genopt.sh: Deleted.
29988 * config/arm/gentune.sh: Deleted.
29989 * config/arm/arm-cores.def: Deleted.
29990 * config/arm/arm-arches.def: Deleted.
29991 * config/arm/arm-fpus.def: Deleted.
29992 * config/arm/arm-tune.md: Regenerated.
29993 * config/arm/arm-tables.opt: Regenerated.
29994 * config/arm/arm-cpu.h: New generated file.
29995 * config/arm/arm-cpu-data.h: New generated file.
29996 * config/arm/arm-cpu-cdata.h: New generated file.
29998 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30001 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30003 (input_varpool_node): Unpack dynamically_initialized bit.
30005 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30007 PR rtl-optimization/79032
30008 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30009 the alignment of the adjusted memory reference against that of MODE,
30010 instead of the alignment of the original memory reference.
30012 2017-01-11 Martin Jambor <mjambor@suse.cz>
30014 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30016 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30017 decorated functions.
30019 2017-01-11 Richard Biener <rguenther@suse.de>
30021 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30022 set range/nonnull info for PHI results. Do not set it on
30023 stmts marked for removal.
30025 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30027 * expr.c (store_field): In the bitfield case, fetch the return value
30028 from the registers before applying a single big-endian adjustment.
30029 Always do a final load for a BLKmode value not larger than a word.
30031 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30034 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30035 that we correctly handle column numbers greater than
30036 LINE_MAP_MAX_COLUMN_NUMBER.
30038 2017-01-10 Martin Sebor <msebor@redhat.com>
30040 PR middle-end/78245
30041 * gimple-ssa-sprintf.c (get_destination_size): Call
30042 {init,fini}object_sizes.
30043 * tree-object-size.c (addr_object_size): Adjust.
30044 (pass_through_call): Adjust.
30045 (pass_object_sizes::execute): Adjust.
30046 * tree-object-size.h (fini_object_sizes): Declare.
30048 2017-01-10 Martin Sebor <msebor@redhat.com>
30050 PR tree-optimization/78775
30051 * builtins.c (get_size_range): Move...
30052 * calls.c: ...to here.
30053 (alloc_max_size): Accept zero argument.
30054 (operand_signed_p): Remove.
30055 (maybe_warn_alloc_args_overflow): Call get_size_range.
30056 * calls.h (get_size_range): Declare.
30058 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30060 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30061 from TI's devices.csv file as of September 2016.
30062 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30064 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30066 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30067 * doc/invoke.texi: Likewise.
30068 * doc/md.texi: Likewise.
30069 * doc/objc.texi: Likewise.
30071 2017-01-10 Joshua Conner <joshconner@google.com>
30073 * config/arm/fuchsia-elf.h: New file.
30074 * config/fuchsia.h: New file.
30075 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30076 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30078 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30080 2016-01-10 Richard Biener <rguenther@suse.de>
30082 PR tree-optimization/79034
30083 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30084 Propagate out degenerate PHIs in the joiner.
30086 2017-01-10 Martin Liska <mliska@suse.cz>
30088 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30089 (sort_congruence_classes_by_decl_uid): Likewise.
30090 (sort_congruence_class_groups_by_decl_uid): Likewise.
30091 (sem_item_optimizer::merge_classes): Sort class, groups in these
30092 classes and members in the groups by DECL_UID of declarations.
30093 This would make merge operations stable.
30095 2017-01-10 Martin Liska <mliska@suse.cz>
30097 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30098 usage of m_classes_vec.
30099 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30100 (sem_item_optimizer::get_group_by_hash): Likewise.
30101 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30102 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30103 (sem_item_optimizer::verify_classes): Likewise.
30104 (sem_item_optimizer::process_cong_reduction): Likewise.
30105 (sem_item_optimizer::dump_cong_classes): Likewise.
30106 (sem_item_optimizer::merge_classes): Likewise.
30107 * ipa-icf.h (congruence_class_hash): Rename from
30108 congruence_class_group_hash. Remove declaration of m_classes_vec.
30110 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30112 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30113 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30114 * config.gcc: Add avx512vpopcntdqintrin.h.
30115 * config/i386/avx512vpopcntdqintrin.h: New.
30116 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30117 * config/i386/i386-builtin-types.def: Add new types.
30118 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30119 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30120 __builtin_ia32_vpopcountq_v8di_mask): New.
30121 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30122 __AVX512VPOPCNTDQ__.
30123 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30124 (PTA_AVX512VPOPCNTDQ): Define.
30125 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30126 TARGET_AVX512VPOPCNTDQ_P): Define.
30127 * config/i386/i386.opt: Add mavx512vpopcntdq.
30128 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30129 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30131 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30133 PR middle-end/77484
30134 * predict.def (PRED_CALL): Set to 67.
30136 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30138 * expr.c (store_field): In the bitfield case, if the value comes from
30139 a function call and is of an aggregate type returned in registers, do
30140 not modify the field mode; extract the value in all cases if the mode
30141 is BLKmode and the size is not larger than a word.
30143 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30146 * config/i386/cpuid.h: Fix undefined behavior.
30148 2017-01-04 Jeff Law <law@redhat.com>
30150 PR tree-optimization/79007
30151 PR tree-optimization/67955
30152 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30153 conservative for pt.null when flag_non_call_exceptions is on.
30155 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30157 PR translation/79019
30158 PR translation/79020
30159 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30160 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30161 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30163 * config/avr/avr.opt (maccumulate-args): Likewise.
30164 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30165 * common.opt (freport-bug): Likewise.
30166 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30167 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30168 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30169 translatable string.
30170 * config/i386/i386.c (function_value_32): Likewise.
30171 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30172 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30174 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30175 * common/config/msp430/msp430-common.c (msp430_handle_option):
30177 * symtab.c (symtab_node::verify_base): Likewise.
30178 * opts.c (set_debug_level): Likewise.
30179 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30180 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30181 missing whitespace to translatable strings.
30182 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30183 * config/sh/superh.h: Likewise.
30184 * config/i386/xopintrin.h: Likewise.
30185 * config/i386/znver1.md: Likewise.
30186 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30187 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30188 * double-int.h (struct double_int): Likewise.
30189 * double-int.c (div_and_round_double): Likewise.
30190 * wide-int.cc: Likewise.
30191 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30192 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30193 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30194 (crossjumps_occurred): ... this.
30195 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30198 PR tree-optimization/78899
30199 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30200 returning bool return struct loop *, NULL for failure and the new
30202 (versionable_outer_loop_p): Don't version outer loop if it has
30203 dont_vectorized bit set.
30204 (tree_if_conversion): When versioning outer loop, ensure
30205 tree_if_conversion is performed also on the inner loop of the
30206 non-vectorizable outer loop copy.
30207 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30208 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30209 prevent vectorization of it.
30210 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30211 the outer loop vectorization of the non-scalar version is attempted
30212 before vectorization of the inner loop in scalar version. If
30213 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30214 vectorization of its inner loop.
30215 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30216 has 2 inner loops, rename also on edges from bb whose single pred
30217 is outer_loop->header. Fix typo in function comment.
30219 2017-01-09 Martin Sebor <msebor@redhat.com>
30222 * asan.c (asan_emit_stack_protection): Increase local buffer size
30223 to avoid snprintf truncation warning.
30225 2017-01-09 Andrew Pinski <apinski@cavium.com>
30227 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30228 to reference thunderx2t99 for the tuning structure
30229 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30231 (thunderx2t99_extra_costs): This.
30232 * config/aarch64/aarch64-tune.md: Regenerate.
30233 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30234 (vulcan_addrcost_table): This.
30235 (vulcan_regmove_cost): Rename to ...
30236 (thunderx2t99_regmove_cost): This.
30237 (vulcan_vector_cost): Rename to ...
30238 (thunderx2t99_vector_cost): this.
30239 (vulcan_branch_cost): Rename to ...
30240 (thunderx2t99_branch_cost): This.
30241 (vulcan_tunings): Rename to ...
30242 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30243 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30245 2017-01-09 Martin Jambor <mjambor@suse.cz>
30249 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30250 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30251 (propagate_vr_accross_jump_function): Use the above function for all
30252 value range computations for pass-through jump functions and type
30253 converasion from explicit value range values.
30254 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30255 parameters from TYPE_ARG_TYPES.
30256 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30257 (ipa_write_node_info): Stream type of the actual argument.
30258 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30260 2017-01-09 Martin Liska <mliska@suse.cz>
30263 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30264 (lookup_compiler): Do not show error message with have_E.
30266 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30268 PR tree-optimization/78938
30269 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30270 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30271 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30272 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30275 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30277 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30280 2017-01-09 Richard Biener <rguenther@suse.de>
30282 PR tree-optimization/78997
30283 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30284 name condition properly.
30286 2017-01-09 Richard Biener <rguenther@suse.de>
30289 * dwarf2out.c (is_cxx): New overload with context.
30290 (is_naming_typedef_decl): Use it.
30292 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30294 * invoke.texi (Option Summary): Correct spacing in option lists
30295 and add line breaks to fix over-long lines.
30297 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30299 PR middle-end/17660
30301 * extend.texi (Common Variable Attributes): Add xref to GCC
30302 Internals manual to explain mode attribute keywords.
30304 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30307 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30308 and Preprocessor Options.
30309 (Options for Linking): Document -pthread here....
30310 (RS/6000 and PowerPC Options): ...not here.
30311 (Solaris 2 Options): ...or here.
30312 * doc/cppopts.texi: Document -pthread.
30314 2017-01-08 Martin Sebor <msebor@redhat.com>
30316 PR middle-end/77708
30317 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30318 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30319 New member functions.
30320 (format_directive): Used them.
30322 (pass_sprintf_length::handle_gimple_call): Same.
30323 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30324 to avoid truncation for any argument.
30325 (extract_affine_mul): Same.
30326 * tree.c (get_file_function_name): Same.
30328 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30330 PR middle-end/77484
30331 * predict.def (PRED_INDIR_CALL): Set to 86.
30333 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30335 PR preprocessor/54124
30336 * doc/cppopts.texi: Reformat -d subtable to list the full name
30337 of the options. Add cross-reference to the docs for the general
30338 compiler -d options.
30339 * doc/invoke.texi (Developer Options): Add cross-reference to the
30340 preprocessor-specific -d option documentation.
30342 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30344 PR preprocessor/13498
30345 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30346 redudant material, and reflect new command-line options.
30347 (System Headers): Likewise.
30349 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30351 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30352 -isystem, and -idirafter. Copy-edit.
30353 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30354 default for -ftrack-macro-expansion. Delete obsolete and
30355 badly-formatted implementation details about -fdebug-cpp output.
30356 * doc/cppwarnopts.texi: Copy-edit.
30358 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30361 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30362 that the transition from a max line width >= 1<<10 to narrower
30363 lines works correctly.
30365 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30367 * doc/options.texi (PerFunction): New.
30368 * opt-functions.awk (switch_flags): Map both Optimization and
30369 PerFunction to CL_OPTIMIZATION.
30370 * opth-gen.awk: Test for PerFunction flag along with
30372 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30373 it only when the latter is present. Skip those that don't in
30374 the hash function generator.
30375 * common.opt (fvar-tracking): Mark as PerFunction instead of
30377 (fvar-tracking-assignments): Likewise.
30378 (fvar-tracking-assignments-toggle): Likewise.
30379 (fvar-tracking-uninit): Likewise.
30381 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30383 PR translation/79018
30384 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30387 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30390 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30391 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30392 TARGET_LONG_JUMP_TABLE_OFFSETS.
30393 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30394 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30395 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30396 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30397 * config/m68k/m68k.md (tablejump expander): Likewise.
30398 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30399 TARGET_LONG_JUMP_TABLE_OFFSETS.
30400 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30401 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30403 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30404 David Holsgrove <david.holsgrove@xilinx.com>
30406 * common/config/microblaze/microblaze-common.c
30407 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30408 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30410 * config/microblaze/microblaze.c (microblaze_must_save_register)
30411 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30413 (microblaze_eh_return): New function.
30414 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30415 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30416 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30417 * config/microblaze/microblaze.md (eh_return): New pattern.
30419 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30421 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30422 GCC_DIAGNOSTIC_STRINGIFY): Define.
30424 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30426 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30428 * config/arm/arm.md (<mcrr>): New.
30430 * config/arm/arm.c (arm_arch5te): New.
30431 (arm_option_override): Set arm_arch5te.
30432 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30434 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30435 (arm_mcrr_qualifiers): ... this. New.
30436 (MRRC_QUALIFIERS): Define to...
30437 (arm_mrrc_qualifiers): ... this. New.
30438 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30440 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30441 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30442 (MRRCI, mrrc, MRRC): New.
30443 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30444 VUNSPEC_MRRC2): New.
30446 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30448 * config/arm/arm.md (<mcr>): New.
30450 * config/arm/arm.c (arm_coproc_builtin_available): Add
30451 support for mcr, mrc, mcr2 and mrc2.
30452 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30453 (arm_mcr_qualifiers): ... this. New.
30454 (MRC_QUALIFIERS): Define to ...
30455 (arm_mrc_qualifiers): ... this. New.
30456 (MCR_QUALIFIERS): Define to ...
30457 (arm_mcr_qualifiers): ... this. New.
30458 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30460 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30461 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30462 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30463 VUNSPEC_MRC2): New.
30465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30467 * config/arm/arm.md (*ldc): New.
30471 * config/arm/arm.c (arm_coproc_builtin_available): Add
30472 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30473 (arm_coproc_ldc_stc_legitimate_address): New.
30474 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30475 'qualifier_const_pointer'.
30476 (LDC_QUALIFIERS): Define to...
30477 (arm_ldc_qualifiers): ... this. New.
30478 (STC_QUALIFIERS): Define to...
30479 (arm_stc_qualifiers): ... this. New.
30480 * config/arm/arm-protos.h
30481 (arm_coproc_ldc_stc_legitimate_address): New.
30482 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30483 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30484 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30485 stc2, stcl, stc2l): New.
30486 * config/arm/constraints.md (Uz): New.
30487 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30488 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30489 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30490 VUNSPEC_STC2L): New.
30492 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30494 * config/arm/arm.md (<cdp>): New.
30495 * config/arm/arm.c (neon_const_bounds): Rename this ...
30496 (arm_const_bounds): ... this.
30497 (arm_coproc_builtin_available): New.
30498 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30499 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30500 (CDP_QUALIFIERS): Define to...
30501 (arm_cdp_qualifiers): ... this. New.
30503 (arm_expand_builtin_args): Add case for 6 arguments.
30504 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30505 (arm_const_bounds): ... this.
30506 (arm_coproc_builtin_available): New.
30507 * config/arm/arm_acle.h (__arm_cdp): New.
30509 * config/arm/arm_acle_builtins.def (cdp): New.
30511 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30512 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30513 'arm_const_bounds'.
30514 * config/arm/types.md (coproc): New.
30515 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30516 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30517 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30518 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30520 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30522 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30523 (UBINOP_QUALIFIERS): New.
30525 (acle_builtin_data): New. Change comment.
30526 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30527 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30528 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30529 arm_acle_builtins.def.
30530 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30531 (arm_init_acle_builtins): New.
30532 (CRC32_BUILTIN): Remove.
30533 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30534 crc32cb, crc32ch and crc32cw.
30535 (arm_init_crc32_builtins): Remove.
30536 (arm_init_builtins): Use arm_init_acle_builtins rather
30537 than arm_init_crc32_builtins.
30538 (arm_expand_acle_builtin): New.
30539 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30540 * config/arm/arm_acle_builtins.def: New.
30542 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30544 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30545 (arm_builtin_datum): ... this.
30546 (arm_init_neon_builtin): Rename to ...
30547 (arm_init_builtin): ... this. Add a new parameters PREFIX
30548 and USE_SIG_IN_NAME.
30549 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30550 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30551 'arm_builtin_datum'.
30552 (arm_init_vfp_builtins): Likewise.
30553 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30554 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30555 (arm_expand_neon_args): Rename to ...
30556 (arm_expand_builtin_args): ... this. Rename builtin_arg
30557 enum values and differentiate between ARG_BUILTIN_MEMORY
30558 and ARG_BUILTIN_NEON_MEMORY.
30559 (arm_expand_neon_builtin_1): Rename to ...
30560 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30561 values, arm_expand_builtin_args and add bool parameter NEON.
30562 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30563 (arm_expand_vfp_builtin): Likewise.
30564 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30566 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30568 PR middle-end/77484
30569 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30570 * predict.c (tree_estimate_probability_bb): Reverse direction of
30571 polymorphic call predictor.
30573 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30575 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30576 (determine_pass_name_match): ...this new function and...
30577 (should_skip_pass_p): ...this new function.
30579 2017-01-06 Nathan Sidwell <nathan@acm.org>
30581 * ipa-visibility.c (function_and_variable_visibility): Reformat
30582 comments and long lines. Remove extrneous if.
30583 * symtab.c (symtab_node::make_decl_local): Fix code format.
30584 (symtab_node::set_section_for_node): Fix comment typo.
30586 2017-01-06 Martin Liska <mliska@suse.cz>
30589 * lra-constraints.c: Rename invariant to lra_invariant.
30590 * predict.c (set_even_probabilities): Initialize e to NULL.
30592 2017-01-05 Martin Sebor <msebor@redhat.com>
30594 PR tree-optimization/78910
30595 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30596 (format_integer): Correct off-by-one error in the handling
30597 of precision with negative numbers in signed conversions..
30599 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30601 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30603 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30605 PR tree-optimization/71016
30606 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30607 factor_out_conditional_conversion. Formatting fix.
30608 (factor_out_conditional_conversion): Add cond_stmt argument.
30609 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30610 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30613 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30615 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30616 read-rtl-function.o, and selftest-rtl.o.
30617 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30618 (selftest::aarch64_test_loading_full_dump): New function.
30619 (selftest::aarch64_run_selftests): New function.
30620 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30621 selftest::aarch64_run_selftests.
30622 * config/i386/i386.c
30623 (selftest::ix86_test_loading_dump_fragment_1): New function.
30624 (selftest::ix86_test_loading_call_insn): New function.
30625 (selftest::ix86_test_loading_full_dump): New function.
30626 (selftest::ix86_test_loading_unspec): New function.
30627 (selftest::ix86_run_selftests): Call the new functions.
30628 * emit-rtl.c (maybe_set_max_label_num): New function.
30629 * emit-rtl.h (maybe_set_max_label_num): New decl.
30630 * function.c (instantiate_decls): Guard call to
30631 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30632 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30634 * gensupport.c (gen_reader::gen_reader): Pass "false"
30635 for new "compact" param of rtx_reader.
30636 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30637 rather than an empty string for NULL strings.
30638 * read-md.c: Potentially include config.h rather than bconfig.h.
30639 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30640 (have_error): New global, copied from errors.c.
30641 (md_reader::read_name): Rename to...
30642 (md_reader::read_name_1): ...this, adding "out_loc" param,
30643 and converting "missing name or number" to returning false, rather
30645 (md_reader::read_name): Reimplement in terms of read_name_1.
30646 (md_reader::read_name_or_nil): New function.
30647 (md_reader::read_string): Handle "(nil)" by returning NULL.
30648 (md_reader::md_reader): Add new param "compact".
30649 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30650 (md_reader::read_file): New method.
30651 * read-md.h (md_reader::md_reader): Add new param "compact".
30652 (md_reader::read_file): New method.
30653 (md_reader::is_compact): New accessor.
30654 (md_reader::read_name): Convert return type from void to file_location.
30655 (md_reader::read_name_or_nil): New decl.
30656 (md_reader::read_name_1): New decl.
30657 (md_reader::m_compact): New field.
30658 (noop_reader::noop_reader): Pass "false" for new "compact" param
30660 (rtx_reader::rtx_reader): Add new "compact" param.
30661 (rtx_reader::read_rtx_operand): Make virtual and convert return
30662 type from void to rtx.
30663 (rtx_reader::read_until): New decl.
30664 (rtx_reader::handle_any_trailing_information): New virtual function.
30665 (rtx_reader::postprocess): New virtual function.
30666 (rtx_reader::finalize_string): New virtual function.
30667 (rtx_reader::m_in_call_function_usage): New field.
30668 (rtx_reader::m_reuse_rtx_by_id): New field.
30669 * read-rtl-function.c: New file.
30670 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30671 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30672 (selftest::verify_three_block_rtl_cfg): New decl.
30673 * read-rtl-function.h: New file.
30674 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30675 For host, include function.h, memmodel.h, and emit-rtl.h.
30676 (one_time_initialization): New function.
30677 (struct compact_insn_name): New struct.
30678 (compact_insn_names): New array.
30679 (find_code): Handle insn codes in compact dumps.
30680 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30681 (bind_subst_iter_and_attr): Likewise.
30682 (add_condition_to_string): Likewise.
30683 (add_condition_to_rtx): Likewise.
30684 (apply_attribute_uses): Likewise.
30685 (add_current_iterators): Likewise.
30686 (apply_iterators): Likewise.
30687 (initialize_iterators): Guard usage of apply_subst_iterator with
30688 #ifdef GENERATOR_FILE.
30689 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30690 (md_reader::read_mapping): Likewise.
30691 (add_define_attr_for_define_subst): Likewise.
30692 (add_define_subst_attr): Likewise.
30693 (read_subst_mapping): Likewise.
30694 (check_code_iterator): Likewise.
30695 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30697 (one_time_initialization): New function.
30698 (rtx_reader::read_until): New method.
30699 (read_flags): New function.
30700 (parse_reg_note_name): New function.
30701 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30702 Handle reuse_rtx ids.
30703 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30704 Add parsing support for RTL dumps, mirroring the special-cases in
30705 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30706 values, and calling handle_any_trailing_information.
30707 (rtx_reader::read_rtx_operand): Convert return type from void
30708 to rtx, returning return_rtx. Handle case 'e'. Call
30709 finalize_string on XSTR and XTMPL fields.
30710 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30711 "(nil)" values were omitted. Call the postprocess vfunc on the
30713 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30714 class ctor. Initialize m_in_call_function_usage. Call
30715 one_time_initialization.
30716 * rtl-tests.c (selftest::test_uncond_jump): Call
30717 set_new_first_and_last_insn.
30718 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30719 * selftest-rtl.c: New file.
30720 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30721 (selftest::get_insn_by_uid): New decl.
30722 * selftest-run-tests.c (selftest::run_tests): Call
30723 read_rtl_function_c_tests.
30724 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30725 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30728 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30730 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30731 operands in a special way. Assert that pos+len <= mode precision.
30733 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30735 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30736 3 argument Alias with unlimited for the negative form.
30737 (fno-vect-cost-model): Removed.
30739 2017-01-05 Martin Liska <mliska@suse.cz>
30741 * hsa-gen.c (gen_hsa_divmod): New function.
30742 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30744 2017-01-05 Martin Liska <mliska@suse.cz>
30747 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30750 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30752 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30753 small constant length operands.
30755 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30757 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30758 between loop iterations.
30760 2017-01-05 Martin Liska <mliska@suse.cz>
30763 * gimplify.c (gimplify_decl_expr): Compare to
30764 asan_poisoned_variables instread of checking flags.
30765 (gimplify_target_expr): Likewise.
30766 (gimplify_expr): Likewise.
30767 (gimplify_function_tree): Conditionally initialize
30768 asan_poisoned_variables.
30770 2017-01-04 Jeff Law <law@redhat.com>
30772 PR tree-optimizatin/78812
30773 * rtl.h (contains_mem_rtx_p): Prototype.
30774 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30775 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30776 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30777 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30778 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30780 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30782 * input.c (assert_char_at_range): Default-initialize actual_range.
30784 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30786 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30787 to match the caller.
30789 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30791 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30792 insns after final jump in test to emit dummy move.
30794 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30796 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30797 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30799 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30801 * multiple_target.c (create_dispatcher_calls): Init e_next.
30802 * tree-ssa-loop-split.c (split_loop): Init border.
30803 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30806 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30811 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30812 (altivec_register_operand): Do not return true if the operand
30813 contains a SUBREG mixing SImode and SFmode.
30814 (vsx_register_operand): Likewise.
30815 (vsx_reg_sfsubreg_ok): New predicate.
30816 (vfloat_operand): Do not return true if the operand contains a
30817 SUBREG mixing SImode and SFmode.
30818 (vint_operand): Likewise.
30819 (vlogical_operand): Likewise.
30820 (gpc_reg_operand): Likewise.
30821 (int_reg_operand): Likewise.
30822 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30823 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30824 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30826 (rs6000_emit_move_si_sf_subreg): New helper function.
30827 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30828 fixup SUBREGs involving SImode and SFmode.
30829 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30830 numbers for the new peephole2 optimization.
30831 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30832 the GLIBC math library that do AND/IOR/XOR operations on single
30833 precision floating point.
30834 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30835 target macros to say whether we need to avoid SUBREGs mixing
30837 (TARGET_ALLOW_SF_SUBREG): Likewise.
30838 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30839 (UNSPEC_SI_FROM_SF): Likewise.
30840 (iorxor): Change spacing.
30841 (and_ior_xor): New iterator for AND, IOR, and XOR.
30842 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30843 (movdi_from_sf_zero_ext): Likewise.
30844 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30845 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30846 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30847 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30848 (fms<mode>4): Likewise.
30849 (fnma<mode>4): Likewise.
30850 (fnms<mode>4): Likewise.
30851 (nfma<mode>4): Likewise.
30852 (nfms<mode>4): Likewise.
30854 2017-01-04 Marek Polacek <polacek@redhat.com>
30857 * doc/invoke.texi: Document -Wpointer-compare.
30859 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30861 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30864 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30865 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30868 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30871 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30872 documentation of the powerpc_popcntb_ok attribute.
30873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30874 code to issue warning messages if a requested CPU configuration is
30875 not supported by the binary (assembler and loader) toolchain.
30876 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30877 made to define a built-in function that has been disabled.
30878 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30879 made to define a built-in function that has been disabled.
30880 (altivec_init_builtins): Add comment explaining why definition
30881 of the DST built-in functions is not preceded by an assertion
30882 check. Add assertions to prevent ICE if attempts are made to
30883 define an altivec predicate or an abs* built-in function that has
30885 (htm_init_builtins): Add comment explaining why definition of the
30886 htm built-in functions is not preceded by an assertion check.
30888 2017-01-04 Jeff Law <law@redhat.com>
30890 PR tree-optimizatin/67955
30891 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30892 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30893 the points-to solution does not include pt_null. Use DECL_PT_UID
30896 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30898 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30899 Use gen_int_mode instead of gen_lopwart for const_int operands.
30901 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30903 PR tree-optimization/71563
30904 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30905 out of range value - has low bits known to be zero.
30907 2017-01-04 Alan Modra <amodra@gmail.com>
30909 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30910 * configure: Regenerate.
30911 * config.in: Regenerate.
30913 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30916 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30917 a substring of the message, but strcmp with the whole message. Ifdef
30918 ENABLE_NLS, translate the message first using dgettext.
30920 2017-01-03 Jeff Law <law@redhat.com>
30922 PR tree-optimizatin/78856
30923 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30924 (mark_threaded_blocks): Remove code to truncate thread paths that
30925 cross multiple loop headers. Instead invalidate the cached loop
30926 iteration information and handle case of a thread path walking
30927 into an irreducible region.
30929 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30932 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30933 assertions. Add support for doing the signbit if the IEEE 128-bit
30934 floating point value is in a GPR.
30935 * config/rs6000/rs6000.md (Fsignbit): Delete.
30936 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30937 Update the length attribute if the value is in a GPR.
30938 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30939 the sign or zero extension instruction, since the value is always 0/1.
30940 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30943 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30944 extracting SImode to a GPR register so that we can generate a
30945 store, limit the vector to be in a traditional Altivec register
30946 for the vextuwrx instruction.
30948 2017-01-03 Ian Lance Taylor <iant@google.com>
30950 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30952 2017-01-03 Martin Sebor <msebor@redhat.com>
30954 PR tree-optimization/78696
30955 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30956 precision. Use MPFR for %f for greater fidelity. Correct handling
30958 (pass_sprintf_length::compute_format_length): Set width and precision
30959 specified by asrerisk to void_node for vararg functions.
30960 (try_substitute_return_value): Adjust dump output.
30962 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30964 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30966 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30968 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30969 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30970 -mlra/-mno-lra was passed to the compiler.
30972 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30974 PR rtl-optimization/65618
30975 * emit-rtl.c (try_split): Move initialization of "before" and
30976 "after" to just before the call to emit_insn_after_setloc.
30978 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30980 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30982 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30984 * dwarf2out.c (gen_enumeration_type_die): When
30985 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30987 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30989 PR tree-optimization/78965
30990 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30991 Change first argument from const call_info & to call_info &. For %n
30992 set info.nowrite to false.
30994 PR middle-end/78901
30995 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30996 possibly throwing calls.
30998 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30999 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31000 and fns handling, rather than in a separate case SSA_NAME.
31002 2017-01-02 Jeff Law <law@redhat.com>
31004 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31005 fixes for first_period and second_period variables.
31007 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31010 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31011 (*insvqi_1): New insn pattern.
31012 (*insvqi_1_mem_rex64): Ditto.
31013 (*insvqi_2): Ditto.
31014 (*insvqi_3): Rename from *insvqi.
31016 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31018 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31020 * doc/cfg.texi (Edges): Remove reference to Java.
31021 (Maintaining the CFG): Ditto.
31023 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31025 PR middle-end/77674
31026 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31027 transparent aliases.
31029 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31031 PR middle-end/77484
31032 * predict.def (PRED_CALL): Update hitrate.
31033 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31034 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31035 into direct/indirect/polymorphic variants.
31037 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31039 Update copyright years.
31041 * gcc.c (process_command): Update copyright notice dates.
31042 * gcov-dump.c (print_version): Ditto.
31043 * gcov.c (print_version): Ditto.
31044 * gcov-tool.c (print_version): Ditto.
31045 * gengtype.c (create_file): Ditto.
31046 * doc/cpp.texi: Bump @copying's copyright year.
31047 * doc/cppinternals.texi: Ditto.
31048 * doc/gcc.texi: Ditto.
31049 * doc/gccint.texi: Ditto.
31050 * doc/gcov.texi: Ditto.
31051 * doc/install.texi: Ditto.
31052 * doc/invoke.texi: Ditto.
31054 Copyright (C) 2017 Free Software Foundation, Inc.
31056 Copying and distribution of this file, with or without modification,
31057 are permitted in any medium without royalty provided the copyright
31058 notice and this notice are preserved.