2013-11-05 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob757eded9a1422eefa346e85b9dd556d1f5344056
1 2013-11-05  Richard Biener  <rguenther@suse.de>
3         PR middle-end/58941
4         * tree-dfa.c (get_ref_base_and_extent): Merge common code
5         in MEM_REF and TARGET_MEM_REF handling.  Make sure to
6         process trailing array detection before diving into the
7         view-converted object (and possibly apply some extra offset).
9 2013-11-05  Joseph Myers  <joseph@codesourcery.com>
11         * config/i386/i386.c (ix86_float_exceptions_rounding_supported_p):
12         New function.
13         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Define.
15 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
17         PR tree-optimization/58958
18         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Use
19         get_addr_base_and_unit_offset instead of get_ref_base_and_extent.
21 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
23         * tree-ssa-alias.h (ranges_overlap_p): Handle negative offsets.
24         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
26 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
28         PR tree-optimization/58984
29         * ipa-prop.c (ipa_load_from_parm_agg_1): Add SIZE_P argument,
30         set *SIZE_P if non-NULL on success.
31         (ipa_load_from_parm_agg, ipa_analyze_indirect_call_uses): Adjust
32         callers.
33         (ipcp_transform_function): Likewise.  Punt if size of access
34         is different from TYPE_SIZE on v->value's type.
36 2013-11-05  Tobias Burnus  <burnus@net-b.de>
38         * doc/invoke.texi (-fopenmp-simd): Document new option.
39         * gimplify.c (gimplify_body): Accept -fopenmp-simd.
40         * omp-low.c (execute_expand_omp, execute_lower_omp): Ditto.
41         * tree.c (attribute_value_equal): Ditto.
43 2013-11-04  Wei Mi  <wmi@google.com>
45         * sched-rgn.c (add_branch_dependences): Keep insns in
46         a SCHED_GROUP at the end of BB to remain their location.
48 2013-11-04  Wei Mi  <wmi@google.com>
50         * gcc/config/i386/i386.c (memory_address_length): Extract a part
51         of code to rip_relative_addr_p.
52         (rip_relative_addr_p): New Function.
53         (ix86_macro_fusion_p): Ditto.
54         (ix86_macro_fusion_pair_p): Ditto.
55         * gcc/config/i386/i386.h: Add new tune features about macro-fusion.
56         * gcc/config/i386/x86-tune.def (DEF_TUNE): Ditto.
57         * gcc/doc/tm.texi: Generated.
58         * gcc/doc/tm.texi.in: Ditto.
59         * gcc/haifa-sched.c (try_group_insn): New Function.
60         (group_insns_for_macro_fusion): Ditto.
61         (sched_init): Call group_insns_for_macro_fusion.
62         * gcc/target.def: Add two hooks: macro_fusion_p and
63         macro_fusion_pair_p.
65 2013-11-04  Kostya Serebryany  <kcc@google.com>
67         Update to match the changed asan API.
68         * asan.c (asan_function_start): New function.
69         (asan_emit_stack_protection): Update the string stored in the
70         stack red zone to match new API.  Store the PC of the current
71         function in the red zone.
72         (asan_global_struct): Update the __asan_global definition to match
73         the new API.
74         (asan_add_global): Ditto.
75         * asan.h (asan_function_start): New prototype.
76         * final.c (final_start_function): Call asan_function_start.
77         * sanitizer.def (BUILT_IN_ASAN_INIT): Rename __asan_init_v1 to __asan_init_v3.
79 2013-11-04  Wei Mi  <wmi@google.com>
81         * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Separate
82         PROCESSOR_COREI7_AVX out from PROCESSOR_COREI7.
83         * gcc/config/i386/i386.c (ix86_option_override_internal): Ditto.
84         (ix86_issue_rate): Ditto.
85         (ix86_adjust_cost): Ditto.
86         (ia32_multipass_dfa_lookahead): Ditto.
87         (ix86_sched_init_global): Ditto.
88         (get_builtin_code_for_version): Ditto.
89         * gcc/config/i386/i386.h (enum target_cpu_default): Ditto.
90         (enum processor_type): Ditto.
91         * gcc/config/i386/x86-tune.def (DEF_TUNE): Ditto.
93 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
95         PR rtl-optimization/58967
96         * config/rs6000/rs6000.c (legitimate_lo_sum_address_p): Remove
97         !lra_in_progress for mode sizes bigger word.
99 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
101         * config/rs6000/altivec.md (vec_widen_umult_hi_v16qi): Swap
102         arguments to merge instruction for little endian.
103         (vec_widen_umult_lo_v16qi): Likewise.
104         (vec_widen_smult_hi_v16qi): Likewise.
105         (vec_widen_smult_lo_v16qi): Likewise.
106         (vec_widen_umult_hi_v8hi): Likewise.
107         (vec_widen_umult_lo_v8hi): Likewise.
108         (vec_widen_smult_hi_v8hi): Likewise.
109         (vec_widen_smult_lo_v8hi): Likewise.
111 2013-11-04  Ian Lance Taylor  <iant@google.com>
113         * builtins.def (ATTR_NOTHROWCALL_LEAF_LIST): Define.
114         * sync-builtins.def: Use ATTR_NOTHROWCALL_LEAF_LIST for all sync
115         builtins that take pointers.
116         * lto-opts.c (lto_write_options): Write -fnon-call-exceptions
117         if set.
118         * lto-wrapper.c (merge_and_complain): Collect
119         OPT_fnon_call_exceptions.
120         (run_gcc): Pass -fnon-call-exceptions.
122 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
124         * optabs.c (expand_vec_perm): Revert one incorrect line from
125         2013-10-31 change.
127         PR tree-optimization/58978
128         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): Don't modify
129         use_stmt by single_imm_use directly.  Only call single_imm_use
130         on SSA_NAMEs.
132 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
134         PR rtl-optimization/58968
135         * lra-spills.c (return_regno_p): New function.
136         (lra_final_code_change): Use it.
138 2013-11-04  Joseph Myers  <joseph@codesourcery.com>
140         * doc/cpp.texi (__GCC_IEC_559, __GCC_IEC_559_COMPLEX): Document
141         macros.
142         * target.def (float_exceptions_rounding_supported_p): New hook.
143         * targhooks.c (default_float_exceptions_rounding_supported_p): New
144         function.
145         * targhooks.h (default_float_exceptions_rounding_supported_p):
146         Declare.
147         * doc/tm.texi.in (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P):
148         New @hook.
149         * doc/tm.texi: Regenerate.
150         * config.gcc (powerpc*-*-linux*): Set extra_objs.
151         * config/rs6000/rs6000-linux.c: New file.
152         * config/rs6000/rs6000-protos.h
153         (rs6000_linux_float_exceptions_rounding_supported_p): Declare.
154         * config/rs6000/linux.h
155         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): New macro.
156         * config/rs6000/linux64.h
157         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Likewise.
158         * config/rs6000/t-linux (rs6000-linux.o): New rule.
159         * config/rs6000/t-linux64 (rs6000-linux.o): Likewise.
161 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
163         * config/rs6000/vsx.md (*vsx_le_perm_store_<mode> for VSX_D):
164         Replace the define_insn_and_split with a define_insn and two
165         define_splits, with the split after reload re-permuting the source
166         register to its original value.
167         (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
168         (*vsx_le_perm_store_v8hi): Likewise.
169         (*vsx_le_perm_store_v16qi): Likewise.
171 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
173         * config/rs6000/vector.md (vec_pack_trunc_v2df):  Adjust for
174         little endian.
176 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
178         PR tree-optimization/58946
179         * tree-ssa-reassoc.c (maybe_optimize_range_tests): Update all
180         bbs with bbinfo[idx].op != NULL before all blocks with
181         bbinfo[idx].op == NULL.
183 2013-11-04  Richard Sandiford  <rdsandiford@googlemail.com>
185         * config/avr/avr-log.c (avr_double_int_pop_digit): Delete.
186         (avr_dump_double_int_hex): Likewise.
187         (avr_log_vadump): Remove %D and %X handling.
188         * config/avr/avr.c (avr_double_int_push_digit): Delete.
189         (avr_map_op_t): Change map from double_int to unsigned int.
190         (avr_map_op): Update accordingly.
191         (avr_map, avr_map_metric, avr_has_nibble_0xf, avr_map_decompose)
192         (avr_move_bits, avr_out_insert_bits, avr_fold_builtin): Operate on
193         unsigned ints rather than double_ints.
195 2013-11-03  Marek Polacek  <polacek@redhat.com>
197         Implement -fsanitize=vla-bound.
198         * opts.c (common_handle_option): Handle vla-bound.
199         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE):
200         Define.
201         * flag-types.h (enum sanitize_code): Add SANITIZE_VLA.
202         * asan.c (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR.
204 2013-11-02  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
206         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Adjust for
207         little endian.
209 2013-11-02  Uros Bizjak  <ubizjak@gmail.com>
211         * config/i386/constraints.md (Ts, Tv): New address constrains.
212         * config/i386/i386.md (*lea<mode>, *<mode>_<bndcheck>): Use Ts
213         constraint for address_no_seg_operand.
214         * config/i386/sse.md (*avx512pf_gatherpf<mode>_mask)
215         (*avx512pf_gatherpf<mode>, *avx512pf_scatterpf<mode>_mask)
216         (*avx512pf_scatterpf<mode>, *avx2_gathersi<mode>)
217         (*avx2_gathersi<mode>_2, *avx2_gatherdi<mode>, *avx2_gatherdi<mode>_2)
218         (*avx2_gatherdi<mode>_3, *avx2_gatherdi<mode>_4)
219         (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2)
220         (*avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2)
221         (*avx512f_scattersi<mode> *avx512f_scatterdi<mode>): Use Tv
222         constraint for vsib_address_operand.
224 2013-11-02  Steven Bosscher  <steven@gcc.gnu.org>
226         * gcse.c (pre_delete): Remove references to regmove from comments.
227         * recog.c: (validate_replace_rtx_1): Likewise.
228         * config/rl78/rl78.c: Likewise.
229         * config/v850/v850.h: Likewise, and remove unused ENABLE_REGMOVE_PASS.
230         * common/config/m32r/m32r-common.c: Don't manipulate OPT_fregmove.
231         * common/config/mmix/mmix-common.c: Likewise.
233 2013-11-01  Trevor Saunders  <tsaunders@mozilla.com>
235         * function.c (reorder_blocks): Convert block_stack to a stack_vec.
236         * gimplify.c (gimplify_compound_lval): Likewise.
237         * graphite-clast-to-gimple.c (gloog): Likewise.
238         * graphite-dependences.c (loop_is_parallel_p): Likewise.
239         * graphite-scop-detection.c (scopdet_basic_block_info): Likewise.
240         (limit_scop); Likewise.
241         (build_scops): Likewise.
242         (dot_scop): Likewise.
243         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
244         (build_scop_drs): Likewise.
245         (insert_stmts): Likewise.
246         (insert_out_of_ssa_copy): Likewise.
247         (remove_phi): Likewise.
248         (rewrite_commutative_reductions_out_of_ssa_close_phi): Likewise.
249         * hw-doloop.c (discover_loop): Likewise.
250         * tree-call-cdce.c (shrink_wrap_one_built_in_call): Likewise.
251         * tree-dfa.c (dump_enumerated_decls): Likewise.
252         * tree-if-conv.c (if_convertable_loop_p): Likewise.
253         * tree-inline.c (tree_function_versioning): Likewise.
254         * tree-loop-distribution.c (build_rdg): Likewise.
255         (rdg_flag_vertex_and_dependent): Likewise.
256         (distribute_loop): Likewise.
257         * tree-parloops.c (loop_parallel_p): Likewise.
258         (eliminate_local_variables): Likewise.
259         (separate_decls_in_region): Likewise.
260         * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
261         * tree-ssa-phiopt.c (cond_if_else_store_replacement): Likewise.
262         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
263         * tree-vect-loop.c (vect_analyze_scaler_cycles_1): Likewise.
264         * tree-vect-patterns.c (vect_pattern_recog): Likewise.
265         * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
266         (vectorizable_condition): Likewise.
268 2013-11-01  Uros Bizjak  <ubizjak@gmail.com>
270         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): Always define as 0/1.
271         * configure: Regenerate.
272         * config/i386/i386.md (*movdi_internal): Change
273         HAVE_AS_IX86_INTERUNIT_MOVQ to runtime check.
274         (*movdf_internal): Ditto.
275         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
276         * config/i386/sse.md (vec_concatv2di): Output interunit movq
277         for HAVE_AS_IX86_INTERUNIT_MOVQ targets.
279 2013-10-31  Robert Suchanek  <Robert.Suchanek@imgtec.com>
281         * lra-spills.c (assign_spill_hard_regs): Remove statement terminator
282         after comment.
284 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
286         Automated part of renaming of symtab_node_base to symtab_node.
288         Patch autogenerated by rename_symtab.py from
289         https://github.com/davidmalcolm/gcc-refactoring-scripts
290         revision 58bb219cc090b2f4516a9297d868c245495ee622
291         with ChangeLog entry fixed up by hand.
293         * cgraph.c (x_cgraph_nodes_queue): Rename symtab_node_base to
294         symtab_node.
295         (cgraph_node_for_asm): Likewise.
296         * cgraph.h (symtab_node_base): Likewise.
297         (cgraph_node): Likewise.
298         (varpool_node): Likewise.
299         (is_a_helper <cgraph_node>::test): Likewise.
300         (is_a_helper <varpool_node>::test): Likewise.
301         (symtab_nodes): Likewise.
302         (symtab_register_node): Likewise.
303         (symtab_unregister_node): Likewise.
304         (symtab_remove_node): Likewise.
305         (symtab_get_node): Likewise.
306         (symtab_node_for_asm): Likewise.
307         (symtab_node_asm_name): Likewise.
308         (symtab_node_name): Likewise.
309         (symtab_insert_node_to_hashtable): Likewise.
310         (symtab_add_to_same_comdat_group): Likewise.
311         (symtab_dissolve_same_comdat_group_list): Likewise.
312         (dump_symtab_node): Likewise.
313         (debug_symtab_node): Likewise.
314         (dump_symtab_base): Likewise.
315         (verify_symtab_node): Likewise.
316         (verify_symtab_base): Likewise.
317         (symtab_used_from_object_file_p): Likewise.
318         (symtab_alias_ultimate_target): Likewise.
319         (symtab_resolve_alias): Likewise.
320         (fixup_same_cpp_alias_visibility): Likewise.
321         (symtab_for_node_and_aliases): Likewise.
322         (symtab_nonoverwritable_alias): Likewise.
323         (availability symtab_node_availability): Likewise.
324         (symtab_semantically_equivalent_p): Likewise.
325         (fixup_same_cpp_alias_visibility): Likewise.
326         (symtab_prevail_in_asm_name_hash): Likewise.
327         (cgraph): Likewise.
328         (varpool): Likewise.
329         (varpool_first_variable): Likewise.
330         (varpool_next_variable): Likewise.
331         (varpool_first_static_initializer): Likewise.
332         (varpool_next_static_initializer): Likewise.
333         (varpool_first_defined_variable): Likewise.
334         (varpool_next_defined_variable): Likewise.
335         (cgraph_first_defined_function): Likewise.
336         (cgraph_next_defined_function): Likewise.
337         (cgraph_first_function): Likewise.
338         (cgraph_next_function): Likewise.
339         (cgraph_first_function_with_gimple_body): Likewise.
340         (cgraph_next_function_with_gimple_body): Likewise.
341         (symtab_alias_target): Likewise.
342         (symtab_real_symbol_p): Likewise.
343         (symtab_can_be_discarded): Likewise.
344         * cgraphbuild.c (mark_address): Likewise.
345         (mark_load): Likewise.
346         (mark_store): Likewise.
347         * cgraphunit.c (decide_is_symbol_needed): Likewise.
348         (first): Likewise.
349         (enqueue_node): Likewise.
350         (referred_to_p): Likewise.
351         (cgraph_process_same_body_aliases): Likewise.
352         (analyze_functions): Likewise.
353         (handle_alias_pairs): Likewise.
354         (output_weakrefs): Likewise.
355         (compile): Likewise.
356         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
357         * ipa-inline-analysis.c (inline_write_summary): Likewise.
358         * ipa-prop.c (remove_described_reference): Likewise.
359         (try_decrement_rdesc_refcount): Likewise.
360         (ipa_edge_duplication_hook): Likewise.
361         * ipa-ref.c (ipa_record_reference): Likewise.
362         (ipa_maybe_record_reference): Likewise.
363         (ipa_clone_ref): Likewise.
364         (ipa_clone_references): Likewise.
365         (ipa_clone_referring): Likewise.
366         (ipa_find_reference): Likewise.
367         (ipa_remove_stmt_references): Likewise.
368         (ipa_clear_stmts_in_references): Likewise.
369         * ipa-ref.h (symtab_node_base): Likewise.
370         (ipa_ref): Likewise.
371         (ipa_record_reference): Likewise.
372         (ipa_maybe_record_reference): Likewise.
373         (ipa_clone_references): Likewise.
374         (ipa_clone_referring): Likewise.
375         (ipa_clone_ref): Likewise.
376         (ipa_find_reference): Likewise.
377         (ipa_remove_stmt_references): Likewise.
378         (ipa_clear_stmts_in_references): Likewise.
379         * ipa-reference.c (ipa_reference_write_optimization_summary):
380         Likewise.
381         * ipa.c (enqueue_node): Likewise.
382         (process_references): Likewise.
383         (walk_polymorphic_call_targets): Likewise.
384         (symtab_remove_unreachable_nodes): Likewise.
385         (address_taken_from_non_vtable_p): Likewise.
386         (comdat_can_be_unshared_p_1): Likewise.
387         (comdat_can_be_unshared_p): Likewise.
388         (can_replace_by_local_alias): Likewise.
389         (function_and_variable_visibility): Likewise.
390         * is-a.h: Likewise (within example in comment).
391         * lto-cgraph.c (input_cgraph_opt_summary): Likewise.
392         (lto_symtab_encoder_encode): Likewise.
393         (lto_symtab_encoder_delete_node): Likewise.
394         (lto_symtab_encoder_in_partition_p): Likewise.
395         (lto_set_symtab_encoder_in_partition): Likewise.
396         (output_refs): Likewise.
397         (compute_ltrans_boundary): Likewise.
398         (output_symtab): Likewise.
399         (input_node): Likewise.
400         (input_ref): Likewise.
401         (input_edge): Likewise.
402         (input_cgraph_1): Likewise.
403         (input_refs): Likewise.
404         (output_cgraph_opt_summary): Likewise.
405         (input_node_opt_summary): Likewise.
406         (input_cgraph_opt_section): Likewise.
407         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
408         Likewise.
409         * lto-streamer-out.c (lto_output): Likewise.
410         (output_symbol_p): Likewise.
411         (produce_symtab): Likewise.
412         * lto-streamer.h (lto_encoder_entry): Likewise.
413         (lto_free_function_in_decl_state_for_node): Likewise.
414         (lto_symtab_encoder_encode): Likewise.
415         (lto_symtab_encoder_delete_node): Likewise.
416         (lto_symtab_encoder_in_partition_p): Likewise.
417         (lto_set_symtab_encoder_in_partition): Likewise.
418         (lto_symtab_encoder_lookup): Likewise.
419         (lsei_node): Likewise.
420         (lto_symtab_encoder_deref): Likewise.
421         * symtab.c (symtab_hash): Likewise.
422         (assembler_name_hash): Likewise.
423         (symtab_nodes): Likewise.
424         (hash_node): Likewise.
425         (eq_node): Likewise.
426         (hash_node_by_assembler_name): Likewise.
427         (eq_assembler_name): Likewise.
428         (insert_to_assembler_name_hash): Likewise.
429         (unlink_from_assembler_name_hash): Likewise.
430         (symtab_prevail_in_asm_name_hash): Likewise.
431         (symtab_register_node): Likewise.
432         (symtab_insert_node_to_hashtable): Likewise.
433         (symtab_unregister_node): Likewise.
434         (symtab_get_node): Likewise.
435         (symtab_remove_node): Likewise.
436         (symtab_initialize_asm_name_hash): Likewise.
437         (symtab_node_for_asm): Likewise.
438         (symtab_add_to_same_comdat_group): Likewise.
439         (symtab_dissolve_same_comdat_group_list): Likewise.
440         (symtab_node_asm_name): Likewise.
441         (symtab_node_name): Likewise.
442         (dump_symtab_base): Likewise.
443         (dump_symtab_node): Likewise.
444         (dump_symtab): Likewise.
445         (debug_symtab_node): Likewise.
446         (verify_symtab_base): Likewise.
447         (verify_symtab_node): Likewise.
448         (verify_symtab): Likewise.
449         (symtab_used_from_object_file_p): Likewise.
450         (symtab_node_availability): Likewise.
451         (symtab_alias_ultimate_target): Likewise.
452         (fixup_same_cpp_alias_visibility): Likewise.
453         (symtab_resolve_alias): Likewise.
454         (symtab_for_node_and_aliases): Likewise.
455         (symtab_for_node_and_aliases): Likewise.
456         (symtab_nonoverwritable_alias_1): Likewise.
457         (symtab_nonoverwritable_alias): Likewise.
458         (symtab_semantically_equivalent_p): Likewise.
459         * value-prof.c (init_node_map): Likewise.
460         * varasm.c (find_decl): Likewise.
461         * varpool.c (varpool_node_for_asm): Likewise.
462         (varpool_remove_unreferenced_decls): Likewise.
464 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
466         Manual part of renaming of symtab_node_base to symtab_node.
468         * ipa-ref.h (symtab_node): Remove typedef to pointer type, as it
469         clashes with the preferred name for the base class.
470         (const_symtab_node): Remove redundant typedef.
472 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
474         * optabs.c (expand_vec_perm): Avoid vector mode punning
475         SUBREGs in SET_DEST.
476         * expmed.c (store_bit_field_1): Likewise.
477         * config/i386/sse.md (movdi_to_sse, vec_pack_sfix_trunc_v2df,
478         vec_pack_sfix_v2df, vec_shl_<mode>, vec_shr_<mode>,
479         vec_interleave_high<mode>, vec_interleave_low<mode>): Likewise.
480         * config/i386/i386.c (ix86_expand_vector_move_misalign,
481         ix86_expand_sse_movcc, ix86_expand_int_vcond, ix86_expand_vec_perm,
482         ix86_expand_sse_unpack, ix86_expand_args_builtin,
483         ix86_expand_vector_init_duplicate, ix86_expand_vector_set,
484         emit_reduc_half, expand_vec_perm_blend, expand_vec_perm_pshufb,
485         expand_vec_perm_interleave2, expand_vec_perm_pshufb2,
486         expand_vec_perm_vpshufb2_vpermq,
487         expand_vec_perm_vpshufb2_vpermq_even_odd, expand_vec_perm_even_odd_1,
488         expand_vec_perm_broadcast_1, expand_vec_perm_vpshufb4_vpermq2,
489         ix86_expand_sse2_mulv4si3, ix86_expand_pinsr): Likewise.
490         (expand_vec_perm_palignr): Likewise.  Modify a copy of *d rather
491         than *d itself.
493 2013-10-31  Uros Bizjak  <ubizjak@gmail.com>
495         * config/i386/i386.c (ix86_expand_sse2_abs): Rename function arguments.
496         Use gcc_unreachable for unhandled modes.  Do not check results of
497         expand_simple_binop.  If not expanded to target, move the result.
499 2013-10-31  Chung-Ju Wu  <jasonwucj@gmail.com>
500             Shiva Chen  <shiva0217@gmail.com>
502         * config.gcc (nds32*-*-*): Add nds32 target.
503         * config/nds32/nds32.c: New file.
504         * config/nds32/nds32.h: New file.
505         * config/nds32/nds32.md: New file.
506         * config/nds32/constants.md: New file.
507         * config/nds32/constraints.md: New file.
508         * config/nds32/iterators.md: New file.
509         * config/nds32/nds32-doubleword.md: New file.
510         * config/nds32/nds32-intrinsic.md: New file.
511         * config/nds32/nds32_intrinsic.h: New file.
512         * config/nds32/nds32-modes.def: New file.
513         * config/nds32/nds32-multiple.md: New file.
514         * config/nds32/nds32.opt: New file.
515         * config/nds32/nds32-opts.h: New file.
516         * config/nds32/nds32-protos.h: New file.
517         * config/nds32/nds32-peephole2.md: New file.
518         * config/nds32/pipelines.md: New file.
519         * config/nds32/predicates.md: New file.
520         * config/nds32/t-mlibs: New file.
521         * common/config/nds32: New directory and files.
523         * doc/invoke.texi (NDS32 options): Document nds32 specific options.
524         * doc/md.texi (NDS32 family): Document nds32 specific constraints.
525         * doc/install.texi (Cross-Compiler-Specific Options): Document
526         --with-nds32-lib for nds32 target.
527         * doc/extend.texi (Function Attributes, Target Builtins): Document
528         nds32 specific attributes.
530 2013-10-31  Vladimir Makarov  <vmakarov@redhat.com>
532         * lra-constraints (process_alt_operands): Use the result
533         elimination register for operand when matching constraints.
535 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
537         * tree-vrp.c (maybe_set_nonzero_bits): New function.
538         (remove_range_assertions): Call it.
540         * tree.c (tree_ctz): New function.
541         * tree.h (tree_ctz): New prototype.
542         * tree-ssanames.h (get_range_info, get_nonzero_bits): Change
543         first argument from tree to const_tree.
544         * tree-ssanames.c (get_range_info, get_nonzero_bits): Likewise.
545         * tree-vectorizer.h (vect_generate_tmps_on_preheader): New prototype.
546         * tree-vect-loop-manip.c (vect_generate_tmps_on_preheader): No longer
547         static.
548         * expr.c (highest_pow2_factor): Reimplemented using tree_ctz.
549         * tree-vect-loop.c (vect_analyze_loop_operations,
550         vect_transform_loop): Don't force scalar loop for bound just because
551         number of iterations is unknown, only do it if it is not known to be
552         a multiple of vectorization_factor.
553         * builtins.c (get_object_alignment_2): Use tree_ctz on offset.
555         * gimple-pretty-print.c (dump_ssaname_info): Print newline also
556         in case of VR_VARYING.  Print get_nonzero_bits if not all ones.
557         * tree-ssanames.h (struct range_info_def): Add nonzero_bits field.
558         (set_nonzero_bits, get_nonzero_bits): New prototypes.
559         * tree-ssa-ccp.c (get_default_value): Use get_range_info to see if
560         a default def isn't partially constant.
561         (ccp_finalize): If after IPA, set_range_info if integral SSA_NAME
562         is known to be partially zero.
563         (evaluate_stmt): If we'd return otherwise VARYING, use get_range_info
564         to see if a default def isn't partially constant.
565         * tree-ssanames.c (set_range_info): Initialize nonzero_bits upon
566         creation of a range, if VR_RANGE, try to improve nonzero_bits from
567         the range.
568         (set_nonzero_bits, get_nonzero_bits): New functions.
570         * tree-cfg.c (assert_unreachable_fallthru_edge_p): New function.
571         * tree-cfg.h (assert_unreachable_fallthru_edge_p): New prototype.
572         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): New function.
573         (remove_range_assertions): If ASSERT_EXPR_VAR has no other immediate
574         uses but in the condition and ASSERT_EXPR and the other successor of
575         the predecessor bb is __builtin_unreachable (), set_range_info of the
576         ASSERT_EXPR_VAR to the range info of the ASSERT_EXPR's lhs.
578 2013-10-31  Martin Jambor  <mjambor@suse.cz>
580         PR rtl-optimization/58934
581         Revert:
582         2013-10-30  Martin Jambor  <mjambor@suse.cz>
583         PR rtl-optimization/10474
584         * ira.c (find_moveable_pseudos): Do not calculate dominance info
585         nor df analysis.
586         (interesting_dest_for_shprep): New function.
587         (split_live_ranges_for_shrink_wrap): Likewise.
588         (ira): Calculate dominance info and df analysis. Call
589         split_live_ranges_for_shrink_wrap.
591 2013-10-31  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
592             Yury Gribov  <y.gribov@samsung.com>
594         PR sanitizer/58543
595         * asan.c (asan_clear_shadow): Allocate a new vreg for temporary
596         shadow pointer to avoid clobbering the main one.
598 2013-10-31  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
600         * lower-subreg.c (resolve_simple_move): Copy REG_INC note.
602 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
604         PR bootstrap/58933
605         * ira-color.c (update_costs_from_copies): Add new parameter.  Use
606         it for calling update_costs_from_allocno.
607         (assign_hard_reg): Call restore_costs_from_copies only for
608         !retry_p.  Pass new argument to update_costs_from_copies.
609         (color_pass): Pass new argument to update_costs_from_copies.
610         (ira_mark_allocation_change): Ditto.
612 2013-10-30  Sharad Singhai  <singhai@google.com>
614         PR middle-end/58134
615         * opts.c (common_handle_option): Remove deprecated option
616         -ftree-vectorizer-verbose.
617         * doc/invoke.texi (Debugging Options): Ditto.
618         * opts-global.c (handle_common_deferred_options): Ditto.
619         (dump_remap_tree_vectorizer_verbose): Delete.
620         * common.opt: Set -ftree-vectorizer-verbose as an ignored option.
622 2013-10-30  DJ Delorie  <dj@redhat.com>
624         * config/rx/rx.c (ADD_RX_BUILTIN0): New macro, used for builtins
625         that take no arguments.
627 2013-10-30  Joern Rennecke  <joern.rennecke@embecosm.com>
629         PR other/58545
630         * reload1.c (update_eliminables_and_spill): New function, broken
631         out of reload.
632         (reload): Use it.  Check for frame size change after frame size
633         alignment, and call update_eliminables_and_spill first if continue-ing.
635 2013-10-30  Cong Hou  <congh@google.com>
637         PR target/58762
638         * config/i386/i386-protos.h (ix86_expand_sse2_abs): New function.
639         * config/i386/i386.c (ix86_expand_sse2_abs): New function.
640         * config/i386/sse.md: Add SSE2 support to abs (8/16/32-bit-int).
642 2013-10-18  Mikael Pettersson  <mikpelinux@gmail.com>
644         PR rtl-optimization/58369
645         * reload1.c (compute_reload_subreg_offset): New function.
646         (choose_reload_regs): Use it to pass endian-correct
647         offset to subreg_regno_offset.
649 2013-10-30  Tobias Burnus  <burnus@net-b.de>
651         PR other/33426
652         * gcc/tree-cfg.c (replace_loop_annotate): Replace warning by
653         warning_at.
655 2013-10-30  Jason Merrill  <jason@redhat.com>
657         * configure.ac (loose_warn): Add -Wno-format if
658         --disable-build-format-warnings.
660 2013-10-30  David Malcolm  <dmalcolm@redhat.com>
662         * cgraphunit.c (analyze_functions): Split symtab_node declarations
663         onto multiple lines to make things easier for rename_symtab.py.
665         * symtab.c (symtab_dissolve_same_comdat_group_list): Likewise.
666         (symtab_semantically_equivalent_p): Likewise.
668 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
670         PR target/58784
671         * lra.c (check_rtl): Remove address check before LRA work.
673 2013-10-30  Marc Glisse  <marc.glisse@inria.fr>
675         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for a
676         POINTER_PLUS_EXPR in the defining statement.
678 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
680         * regmove.c: Remove.
681         * tree-pass.h (make_pass_regmove): Remove.
682         * timevar.def (TV_REGMOVE): Remove.
683         * passes.def (pass_regmove): Remove.
684         * opts.c (default_options_table): Remove entry for regmove.
685         * doc/passes.texi: Remove regmove pass description.
686         * doc/invoke.texi (-foptimize-register-move, -fregmove): Remove
687         options.
688         (-fdump-rtl-regmove): Ditto.
689         * common.opt (foptimize-register-move, fregmove): Ignore.
690         * Makefile.in (OBJS): Remove regmove.o.
691         * regmove.c: Remove.
692         * ira-int.h (struct ira_allocno_pref, ira_pref_t): New structure
693         and type.
694         (struct ira_allocno) New member allocno_prefs.
695         (ALLOCNO_PREFS): New macro.
696         (ira_prefs, ira_prefs_num): New external vars.
697         (ira_setup_alts, ira_get_dup_out_num, ira_debug_pref): New prototypes.
698         (ira_debug_prefs, ira_debug_allocno_prefs, ira_create_pref): Ditto.
699         (ira_add_allocno_pref, ira_remove_pref, ira_remove_allocno_prefs):
700         Ditto.
701         (ira_add_allocno_copy_to_list): Remove prototype.
702         (ira_swap_allocno_copy_ends_if_necessary): Ditto.
703         (ira_pref_iterator): New type.
704         (ira_pref_iter_init, ira_pref_iter_cond): New functions.
705         (FOR_EACH_PREF): New macro.
706         * ira.c (commutative_constraint_p): Move from ira-conflicts.c.
707         (ira_get_dup_out_num): Ditto. Rename from get_dup_num.  Modify the
708         code.
709         (ira_setup_alts): New function.
710         (decrease_live_ranges_number): New function.
711         (ira): Call the above function.
712         * ira-build.c (ira_prefs, ira_prefs_num): New global vars.
713         (ira_create_allocno): Initialize allocno prefs.
714         (pref_pool, pref_vec): New static vars.
715         (initiate_prefs, find_allocno_pref, ira_create_pref): New
716         functions.
717         (add_allocno_pref_to_list, ira_add_allocno_pref, print_pref): Ditto.
718         (ira_debug_pref, print_prefs, ira_debug_prefs): Ditto.
719         (print_allocno_prefs, ira_debug_allocno_prefs, finish_pref): Ditto.
720         (ira_remove_pref, ira_remove_allocno_prefs, finish_prefs): Ditto.
721         (ira_add_allocno_copy_to_list): Make static.  Rename to
722         add_allocno_copy_to_list.
723         (ira_swap_allocno_copy_ends_if_necessary): Make static.  Rename to
724         swap_allocno_copy_ends_if_necessary.
725         (remove_unnecessary_allocnos, remove_low_level_allocnos): Call
726         ira_remove_allocno_prefs.
727         (ira_flattening): Ditto.
728         (ira_build): Call initiate_prefs, print_prefs.
729         (ira_destroy): Call finish_prefs.
730         * ira-color.c (struct update_cost_record): New.
731         (struct allocno_color_data): Add new member update_cost_records.
732         (update_cost_record_pool): New static var.
733         (init_update_cost_records, get_update_cost_record): New functions.
734         (free_update_cost_record_list, finish_update_cost_records): Ditto.
735         (struct update_cost_queue_elem): Add member from.
736         (initiate_cost_update): Call init_update_cost_records.
737         (finish_cost_update): Call finish_update_cost_records.
738         (queue_update_cost, get_next_update_cost): Add new param from.
739         (Update_allocno_cost, update_costs_from_allocno): New functions.
740         (update_costs_from_prefs): Ditto.
741         (update_copy_costs): Rename to update_costs_from_copies.
742         (restore_costs_from_copies): New function.
743         (update_conflict_hard_regno_costs): Don't go back.
744         (assign_hard_reg): Call restore_costs_from_copies.  Add printing
745         more debug info.
746         (pop_allocnos): Add priniting more debug info.
747         (color_allocnos): Remove prefs for conflicting hard regs.
748         Call update_costs_from_prefs.
749         * ira-conflicts.c (commutative_constraint_p): Move to ira.c
750         (get_dup_num): Rename, modify, and move to ira.c
751         (process_regs_for_copy): Add prefs.
752         (add_insn_allocno_copies): Put src as first arg of
753         process_regs_for_copy.  Remove dead code.  Call ira_setup_alts.
754         * ira-costs.c (record_reg_classes): Modify and move code into
755         record_operands_costs.
756         (find_costs_and_classes): Create prefs for the hard reg of small
757         reg class.
758         (process_bb_node_for_hard_reg_moves): Add prefs.
760 2013-10-30  Richard Biener  <rguenther@suse.de>
762         PR middle-end/57100
763         * basic-block.h (pre_and_rev_post_order_compute_fn): New function.
764         * cfganal.c (pre_and_rev_post_order_compute_fn): New function
765         as worker for ...
766         (pre_and_rev_post_order_compute): ... which now wraps it.
767         * graph.c (draw_cfg_nodes_no_loops): Use
768         pre_and_rev_post_order_compute_fn to avoid ICEing and dependence
769         on cfun.
771 2013-10-30  Christian Bruel  <christian.bruel@st.com>
773         * gcc/config/sh/sh-mem.cc (sh_expand_cmpnstr): New function.
774         (sh_expand_cmpstr): Handle known align and schedule improvements.
775         * gcc/config/sh/sh-protos.h (sh_expand_cmpstrn): Declare.
776         * gcc/config/sh/sh.md (cmpstrnsi): New pattern.
778 2013-10-30  Martin Jambor  <mjambor@suse.cz>
780         PR rtl-optimization/10474
781         * ira.c (find_moveable_pseudos): Do not calculate dominance info
782         nor df analysis.
783         (interesting_dest_for_shprep): New function.
784         (split_live_ranges_for_shrink_wrap): Likewise.
785         (ira): Calculate dominance info and df analysis. Call
786         split_live_ranges_for_shrink_wrap.
788 2013-10-30  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
790         PR target/58854
791         * config/arm/arm.c (arm_expand_epilogue_apcs_frame): Emit blockage.
793 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
795         * tree-core.h (tree_index): Add TI_POINTER_BOUNDS_TYPE.
796         * tree.h (POINTER_BOUNDS_P): New.
797         (BOUNDED_TYPE_P): New.
798         (BOUNDED_P): New.
799         (pointer_bounds_type_node): New.
800         * tree.c (build_common_tree_nodes): Initialize
801         pointer_bounds_type_node.
802         * gimple.h (gimple_call_get_nobnd_arg_index): New.
803         (gimple_call_num_nobnd_args): New.
804         (gimple_call_nobnd_arg): New.
805         (gimple_return_retbnd): New.
806         (gimple_return_set_retbnd): New
807         * gimple.c (gimple_build_return): Increase number of ops
808         for return statement.
809         (gimple_call_get_nobnd_arg_index): New.
810         * gimple-pretty-print.c (dump_gimple_return): Print second op.
812 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
814         * ipa.c (cgraph_build_static_cdtor_1): Support contructors
815         with "chkp ctor" and "bnd_legacy" attributes.
816         * gimplify.c (gimplify_init_constructor): Avoid infinite
817         loop during gimplification of bounds initializer.
819 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
821         * c-family/c-common.c (handle_bnd_variable_size_attribute): New.
822         (handle_bnd_legacy): New.
823         (c_common_attribute_table): Add bnd_variable_size and bnd_legacy.
824         * doc/extend.texi: Document bnd_variable_size and bnd_legacy
825         attributes.
827 2013-10-29  Ilya Enkovich  <ilya.enkovich@intel.com>
829         * builtin-types.def (BT_FN_VOID_CONST_PTR): New.
830         (BT_FN_PTR_CONST_PTR): New.
831         (BT_FN_CONST_PTR_CONST_PTR): New.
832         (BT_FN_PTR_CONST_PTR_SIZE): New.
833         (BT_FN_PTR_CONST_PTR_CONST_PTR): New.
834         (BT_FN_VOID_PTRPTR_CONST_PTR): New.
835         (BT_FN_VOID_CONST_PTR_SIZE): New.
836         (BT_FN_PTR_CONST_PTR_CONST_PTR_SIZE): New.
837         * chkp-builtins.def: New.
838         * builtins.def: include chkp-builtins.def.
839         (DEF_CHKP_BUILTIN): New.
840         * builtins.c (expand_builtin): Support BUILT_IN_CHKP_INIT_PTR_BOUNDS,
841         BUILT_IN_CHKP_NULL_PTR_BOUNDS, BUILT_IN_CHKP_COPY_PTR_BOUNDS,
842         BUILT_IN_CHKP_CHECK_PTR_LBOUNDS, BUILT_IN_CHKP_CHECK_PTR_UBOUNDS,
843         BUILT_IN_CHKP_CHECK_PTR_BOUNDS, BUILT_IN_CHKP_SET_PTR_BOUNDS,
844         BUILT_IN_CHKP_NARROW_PTR_BOUNDS, BUILT_IN_CHKP_STORE_PTR_BOUNDS,
845         BUILT_IN_CHKP_GET_PTR_LBOUND, BUILT_IN_CHKP_GET_PTR_UBOUND,
846         BUILT_IN_CHKP_BNDMK, BUILT_IN_CHKP_BNDSTX, BUILT_IN_CHKP_BNDCL,
847         BUILT_IN_CHKP_BNDCU, BUILT_IN_CHKP_BNDLDX, BUILT_IN_CHKP_BNDRET,
848         BUILT_IN_CHKP_INTERSECT, BUILT_IN_CHKP_ARG_BND, BUILT_IN_CHKP_NARROW,
849         BUILT_IN_CHKP_EXTRACT_LOWER, BUILT_IN_CHKP_EXTRACT_UPPER.
850         * common.opt (fcheck-pointer-bounds): New.
851         * toplev.c (process_options): Check Pointer Bounds Checker is
852         supported.
853         * doc/extend.texi: Document Pointer Bounds Checker built-in functions.
855 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
857         * target.def (builtin_chkp_function): New.
858         (chkp_bound_type): New.
859         (chkp_bound_mode): New.
860         (fn_abi_va_list_bounds_size): New.
861         (load_bounds_for_arg): New.
862         (store_bounds_for_arg): New.
863         * targhooks.h (default_load_bounds_for_arg): New.
864         (default_store_bounds_for_arg): New.
865         (default_fn_abi_va_list_bounds_size): New.
866         (default_chkp_bound_type): New.
867         (default_chkp_bound_mode): New.
868         (default_builtin_chkp_function): New.
869         * targhooks.c (default_load_bounds_for_arg): New.
870         (default_store_bounds_for_arg): New.
871         (default_fn_abi_va_list_bounds_size): New.
872         (default_chkp_bound_type): New.
873         (default_chkp_bound_mode); New.
874         (default_builtin_chkp_function): New.
875         * doc/tm.texi.in (TARGET_FN_ABI_VA_LIST_BOUNDS_SIZE): New.
876         (TARGET_LOAD_BOUNDS_FOR_ARG): New.
877         (TARGET_STORE_BOUNDS_FOR_ARG): New.
878         (TARGET_BUILTIN_CHKP_FUNCTION): New.
879         (TARGET_CHKP_BOUND_TYPE): New.
880         (TARGET_CHKP_BOUND_MODE): New.
881         * doc/tm.texi: Regenerated.
882         * langhooks.h (lang_hooks): Add chkp_supported field.
883         * langhooks-def.h (LANG_HOOKS_CHKP_SUPPORTED): New.
884         (LANG_HOOKS_INITIALIZER); Add LANG_HOOKS_CHKP_SUPPORTED.
886 2013-10-29  Andrew Pinski <apinski@cavium.com>
888         * tree-ssa-ifcombine.c: Include rtl.h and tm_p.h.
889         (ifcombine_ifandif): Handle cases where maybe_fold_and_comparisons
890         fails, combining the branches anyways.
891         (tree_ssa_ifcombine): Inverse the order of the basic block walk,
892         increases the number of combinings.
893         * gimple.h (gsi_start_nondebug_after_labels_bb): New function.
895 2013-10-29  Mike Stump  <mikestump@comcast.net>
897         * machmode.def (PARTIAL_INT_MODE): Add precision and name.
898         * genmodes.c (PARTIAL_INT_MODE): Add precision and name.
899         (make_vector_mode): Increase namebuf to 16.
900         (emit_insn_modes_h): When processing BImode, don't
901         also match partial int modes.
902         (emit_class_narrowest_mode): Likewise.
904         * config/bfin/bfin-modes.def: Add precision to PDI.
905         * config/m32c/m32c-modes.def: Add precision to PSI.
906         * config/msp430/msp430-modes.def: Add precision to PSI.
907         * config/rs6000/rs6000-modes.def: Add precision to PTI.
908         * config/sh/sh-modes.def: Add precision to PSI and PDI.
910 2013-10-29  Oleg Endo  <olegendo@gcc.gnu.org>
912         PR target/54236
913         * config/sh/sh.md (*addc): Rename existing variations to ...
914         (*addc_r_r_1, *addc_2r_1, *addc_r_1): ... these.
915         (*addc_r_lsb, *addc_r_r_lsb, *addc_r_lsb_r, *addc_2r_lsb, *addc_r_msb,
916         *addc_r_r_msb, *addc_2r_msb): New insn_and_split patterns.
917         * config/sh/sh.c (addsubcosts): Handle some addc special cases.
919 2013-10-29  Teresa Johnson  <tejohnson@google.com>
921         PR ipa/58862
922         * tree-ssa-tail-merge.c (replace_block_by): Tolerate profile
923         insanities when updating probabilities.
925 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
927         * gdbhooks.py (CGraphNodePrinter.to_string): Update gdb
928         prettyprinter for cgraph_node to reflect the conversion of the
929         symtable types to a C++ class hierarchy: it now *is* a
930         symtab_node_base, rather than having one (named "symbol").
932 2013-10-29  Balaji V. Iyer  <balaji.v.iyer@intel.com>
934         * builtins.c (is_builtin_name): Added a check for __cilkrts_detach and
935         __cilkrts_pop_frame.  If matched, then return true for built-in
936         function name.
937         (expand_builtin): Added BUILT_IN_CILK_DETACH and
938         BUILT_IN_CILK_POP_FRAME case.
939         * langhooks-def.h (lhd_install_body_with_frame_cleanup): New prototype.
940         (lhs_cilk_detect_spawn): Likewise.
941         (LANG_HOOKS_DECLS): Added LANG_HOOKS_CILKPLUS.
942         (LANG_HOOKS_CILKPLUS_DETECT_SPAWN_AND_UNWRAP): New #define.
943         (LANG_HOOKS_CILKPLUS_FRAME_CLEANUP): Likewise.
944         (LANG_HOOKS_CILKPLUS_GIMPLIFY_SPAWN): Likewise.
945         (LANG_HOOKS_CILKPLUS): Likewise.
946         * tree.h (CILK_SPAWN_FN): Likewise.
947         * builtin.def (DEF_CILK_BUILTIN_STUB): Likewise.
948         * Makefile.in (C_COMMON_OBJS): Added c-family/cilk.o.
949         (OBJS): Added cilk-common.o.
950         (BUILTINS_DEF): Added cilk-builtins.def.
951         * langhooks.c (lhd_install_body_with_frame_cleanup): New function.
952         (lhd_cilk_detect_spawn): Likewise.
953         * langhooks.h (lang_hooks_for_cilkplus): New struct.
954         (struct lang_hooks): Added new field called "cilkplus."
955         * cilk-common.c: New file.
956         * cilk.h: Likewise.
957         * cilk-builtins.def: Likewise.
958         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Added
959         "__cilk" macro and set it to 200.
960         * function.h (struct function::cilk_frame_decl): New field.
961         (struct function::is_cilk_function): Likewise.
962         (struct function::calls_cilk_spawn): Likewise.
963         * gimplify.c (gimplify_call_expr): Added a check if the function call
964         being gimplified is a spawn detach point.  If so, then add pop_frame
965         and detach function calls.
966         (gimplify_expr): Added a CILK_SPAWN_STMT and CILK_SYNC_STMT case
967         for gimplifying _Cilk_spawn and _Cilk_sync statements.
968         (gimplify_return_expr): Added a check for _Cilk_spawn usage in
969         function.  If so, added a _Cilk_sync and gimplified it.
970         (gimplify_modify_expr): Added a check for _Cilk_spawn in MODIFY and
971         INIT_EXPRs.  If so, then call gimplify_cilk_spawn.
972         * ipa-inline-analysis (initialize_inline_failed): Prevent inlining of
973         spawner function.
974         (can_inline_edge_p): Prevent inling of spawnee function.
975         * ira.c (ira_setup_eliminable_regset): Force usage of frame pointer
976         for functions that use Cilk keywords.
977         * tree-inline.h (struct copy_body_data::remap_var_for_cilk): New field.
978         * tree-pretty-print.c (dump_generic_node): Added CILK_SPAWN_STMT and
979         CILK_SYNC_STMT cases.
980         * tree.def (DEFTREECODE): Added CILK_SPAWN_STMT and CILK_SYNC_STMT
981         trees.
982         * generic.texi (CILK_SPAWN_STMT): Added documentation for _Cilk_spawn.
983         (CILK_SYNC_STMT): Added documentation for _Cilk_sync.
984         * passes.texi (Cilk Keywords): New section that describes the compiler
985         code changes for handling Cilk Keywords.
987 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
989         Patch autogenerated by refactor_symtab.py from
990         https://github.com/davidmalcolm/gcc-refactoring-scripts
991         revision 58bb219cc090b2f4516a9297d868c245495ee622
993         * asan.c (asan_finish_file): Update for conversion of symtab types to
994         a true class hierarchy.
995         * cfgexpand.c (estimated_stack_frame_size): Likewise.
996         * cgraph.c (cgraph_get_body): Likewise.
997         (cgraph_get_create_real_symbol_node): Likewise.
998         (verify_cgraph_node): Likewise.
999         (verify_edge_corresponds_to_fndecl): Likewise.
1000         (verify_edge_count_and_frequency): Likewise.
1001         (cgraph_will_be_removed_from_program_if_no_direct_calls): Likewise.
1002         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
1003         (cgraph_can_remove_if_no_direct_calls_and_refs_p): Likewise.
1004         (cgraph_node_cannot_return): Likewise.
1005         (cgraph_set_pure_flag_1): Likewise.
1006         (cgraph_set_const_flag_1): Likewise.
1007         (cgraph_set_nothrow_flag_1): Likewise.
1008         (cgraph_make_node_local_1): Likewise.
1009         (cgraph_for_node_and_aliases): Likewise.
1010         (cgraph_for_node_thunks_and_aliases): Likewise.
1011         (cgraph_node_can_be_local_p): Likewise.
1012         (cgraph_node_cannot_be_local_p_1): Likewise.
1013         (cgraph_function_body_availability): Likewise.
1014         (dump_cgraph_node): Likewise.
1015         (cgraph_rtl_info): Likewise.
1016         (cgraph_mark_address_taken_node): Likewise.
1017         (cgraph_remove_node): Likewise.
1018         (cgraph_release_function_body): Likewise.
1019         (cgraph_update_edges_for_call_stmt_node): Likewise.
1020         (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
1021         (cgraph_make_edge_direct): Likewise.
1022         (cgraph_resolve_speculation): Likewise.
1023         (cgraph_speculative_call_info): Likewise.
1024         (cgraph_turn_edge_to_speculative): Likewise.
1025         (cgraph_create_edge_1): Likewise.
1026         (cgraph_set_call_stmt): Likewise.
1027         (cgraph_node_for_asm): Likewise.
1028         (cgraph_add_thunk): Likewise.
1029         (cgraph_same_body_alias): Likewise.
1030         (cgraph_create_function_alias): Likewise.
1031         (cgraph_create_node): Likewise.
1032         (cgraph_create_empty_node): Likewise.
1033         (record_function_versions): Likewise.
1034         (used_from_object_file_p): Likewise.
1035         * cgraph.h (symtab_can_be_discarded): Likewise.
1036         (symtab_real_symbol_p): Likewise.
1037         (cgraph_mark_force_output_node): Likewise.
1038         (cgraph_edge_recursive_p): Likewise.
1039         (symtab_alias_target): Likewise.
1040         (varpool_all_refs_explicit_p): Likewise.
1041         (varpool_can_remove_if_no_refs): Likewise.
1042         (cgraph_only_called_directly_or_aliased_p): Likewise.
1043         (cgraph_next_function_with_gimple_body): Likewise.
1044         (cgraph_first_function_with_gimple_body): Likewise.
1045         (cgraph_function_with_gimple_body_p): Likewise.
1046         (cgraph_next_function): Likewise.
1047         (cgraph_first_function): Likewise.
1048         (cgraph_next_defined_function): Likewise.
1049         (cgraph_first_defined_function): Likewise.
1050         (varpool_next_defined_variable): Likewise.
1051         (varpool_first_defined_variable): Likewise.
1052         (varpool_next_static_initializer): Likewise.
1053         (varpool_first_static_initializer): Likewise.
1054         (varpool_next_variable): Likewise.
1055         (varpool_first_variable): Likewise.
1056         (varpool_node_name): Likewise.
1057         (varpool): Likewise.
1058         (cgraph): Likewise.
1059         (is_a_helper <varpool_node>::test): Likewise.
1060         (is_a_helper <cgraph_node>::test): Likewise.
1061         (varpool_variable_node): Likewise.
1062         (cgraph_function_or_thunk_node): Likewise.
1063         (varpool_alias_target): Likewise.
1064         (cgraph_alias_target): Likewise.
1065         (cgraph_node_name): Likewise.
1066         (varpool_node_asm_name): Likewise.
1067         (cgraph_node_asm_name): Likewise.
1068         * cgraphbuild.c (remove_cgraph_callee_edges): Likewise.
1069         (cgraph_rebuild_references): Likewise.
1070         (rebuild_cgraph_edges): Likewise.
1071         (record_eh_tables): Likewise.
1072         (build_cgraph_edges): Likewise.
1073         (mark_store): Likewise.
1074         (mark_load): Likewise.
1075         (mark_address): Likewise.
1076         (record_type_list): Likewise.
1077         (record_reference): Likewise.
1078         * cgraphclones.c (cgraph_materialize_all_clones): Likewise.
1079         (cgraph_materialize_clone): Likewise.
1080         (cgraph_function_versioning): Likewise.
1081         (cgraph_copy_node_for_versioning): Likewise.
1082         (update_call_expr): Likewise.
1083         (cgraph_find_replacement_node): Likewise.
1084         (cgraph_create_virtual_clone): Likewise.
1085         (cgraph_clone_node): Likewise.
1086         * cgraphunit.c (compile): Likewise.
1087         (output_weakrefs): Likewise.
1088         (output_in_order): Likewise.
1089         (expand_function): Likewise.
1090         (assemble_thunks_and_aliases): Likewise.
1091         (expand_thunk): Likewise.
1092         (mark_functions_to_output): Likewise.
1093         (handle_alias_pairs): Likewise.
1094         (analyze_functions): Likewise.
1095         (walk_polymorphic_call_targets): Likewise.
1096         (varpool_finalize_decl): Likewise.
1097         (process_function_and_variable_attributes): Likewise.
1098         (cgraph_process_same_body_aliases): Likewise.
1099         (analyze_function): Likewise.
1100         (cgraph_add_new_function): Likewise.
1101         (cgraph_finalize_function): Likewise.
1102         (referred_to_p): Likewise.
1103         (cgraph_reset_node): Likewise.
1104         (cgraph_process_new_functions): Likewise.
1105         (enqueue_node): Likewise.
1106         (decide_is_symbol_needed): Likewise.
1107         * coverage.c (coverage_compute_profile_id): Likewise.
1108         * dbxout.c (dbxout_expand_expr): Likewise.
1109         * dwarf2out.c (premark_types_used_by_global_vars_helper): Likewise.
1110         (reference_to_unused): Likewise.
1111         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
1112         * gimplify.c (unvisit_body): Likewise.
1113         (unshare_body): Likewise.
1114         * ipa-cp.c (ipcp_generate_summary): Likewise.
1115         (ipcp_decision_stage): Likewise.
1116         (identify_dead_nodes): Likewise.
1117         (decide_whether_version_node): Likewise.
1118         (decide_about_value): Likewise.
1119         (perhaps_add_new_callers): Likewise.
1120         (create_specialized_node): Likewise.
1121         (update_profiling_info): Likewise.
1122         (ipcp_propagate_stage): Likewise.
1123         (estimate_local_effects): Likewise.
1124         (good_cloning_opportunity_p): Likewise.
1125         (devirtualization_time_bonus): Likewise.
1126         (propagate_constants_accross_call): Likewise.
1127         (initialize_node_lattices): Likewise.
1128         (ipcp_cloning_candidate_p): Likewise.
1129         (determine_versionability): Likewise.
1130         (print_all_lattices): Likewise.
1131         (print_lattice): Likewise.
1132         (ipcp_discover_new_direct_edges): Likewise.
1133         * ipa-devirt.c (ipa_devirt): Likewise.
1134         (likely_target_p): Likewise.
1135         (update_type_inheritance_graph): Likewise.
1136         (possible_polymorphic_call_target_p): Likewise.
1137         (dump_possible_polymorphic_call_targets): Likewise.
1138         (devirt_variable_node_removal_hook): Likewise.
1139         (record_binfo): Likewise.
1140         (maybe_record_node): Likewise.
1141         (build_type_inheritance_graph): Likewise.
1142         * ipa-inline-analysis.c (inline_write_summary): Likewise.
1143         (inline_generate_summary): Likewise.
1144         (inline_analyze_function): Likewise.
1145         (do_estimate_growth): Likewise.
1146         (simple_edge_hints): Likewise.
1147         (estimate_node_size_and_time): Likewise.
1148         (estimate_edge_devirt_benefit): Likewise.
1149         (compute_inline_parameters): Likewise.
1150         (estimate_function_body_sizes): Likewise.
1151         (compute_bb_predicates): Likewise.
1152         (initialize_inline_failed): Likewise.
1153         (dump_inline_summary): Likewise.
1154         (dump_inline_edge_summary): Likewise.
1155         * ipa-inline-transform.c (inline_transform): Likewise.
1156         (preserve_function_body_p): Likewise.
1157         (save_inline_function_body): Likewise.
1158         (inline_call): Likewise.
1159         (clone_inlined_nodes): Likewise.
1160         (can_remove_node_now_p): Likewise.
1161         (can_remove_node_now_p_1): Likewise.
1162         * ipa-inline.c (early_inliner): Likewise.
1163         (early_inline_small_functions): Likewise.
1164         (inline_always_inline_functions): Likewise.
1165         (ipa_inline): Likewise.
1166         (flatten_function): Likewise.
1167         (inline_small_functions): Likewise.
1168         (speculation_useful_p): Likewise.
1169         (recursive_inlining): Likewise.
1170         (update_caller_keys): Likewise.
1171         (reset_edge_caches): Likewise.
1172         (update_edge_key): Likewise.
1173         (edge_badness): Likewise.
1174         (relative_time_benefit): Likewise.
1175         (want_inline_self_recursive_call_p): Likewise.
1176         (want_inline_small_function_p): Likewise.
1177         (want_early_inline_function_p): Likewise.
1178         (num_calls): Likewise.
1179         (can_early_inline_edge_p): Likewise.
1180         (can_inline_edge_p): Likewise.
1181         (report_inline_failed_reason): Likewise.
1182         * ipa-profile.c (ipa_profile): Likewise.
1183         (ipa_propagate_frequency): Likewise.
1184         (ipa_propagate_frequency_1): Likewise.
1185         (ipa_profile_generate_summary): Likewise.
1186         * ipa-prop.c (ipcp_transform_function): Likewise.
1187         (read_replacements_section): Likewise.
1188         (ipa_prop_read_section): Likewise.
1189         (ipa_modify_call_arguments): Likewise.
1190         (ipa_print_node_params): Likewise.
1191         (propagate_controlled_uses): Likewise.
1192         (update_indirect_edges_after_inlining): Likewise.
1193         (remove_described_reference): Likewise.
1194         (ipa_make_edge_direct_to_target): Likewise.
1195         (ipa_analyze_node): Likewise.
1196         (ipa_analyze_params_uses): Likewise.
1197         (ipa_compute_jump_functions): Likewise.
1198         (ipa_get_callee_param_type): Likewise.
1199         (ipa_print_node_jump_functions): Likewise.
1200         (ipa_initialize_node_params): Likewise.
1201         (ipa_populate_param_decls): Likewise.
1202         (ipa_func_spec_opts_forbid_analysis_p): Likewise.
1203         (write_agg_replacement_chain): Likewise.
1204         (ipa_write_node_info): Likewise.
1205         (ipa_edge_duplication_hook): Likewise.
1206         (try_decrement_rdesc_refcount): Likewise.
1207         * ipa-pure-const.c (propagate_nothrow): Likewise.
1208         (propagate_pure_const): Likewise.
1209         (pure_const_read_summary): Likewise.
1210         (pure_const_write_summary): Likewise.
1211         (analyze_function): Likewise.
1212         * ipa-ref-inline.h (ipa_ref_referred_ref_list): Likewise.
1213         (ipa_ref_referring_ref_list): Likewise.
1214         * ipa-ref.c (ipa_clear_stmts_in_references): Likewise.
1215         (ipa_remove_stmt_references): Likewise.
1216         (ipa_find_reference): Likewise.
1217         (ipa_dump_referring): Likewise.
1218         (ipa_dump_references): Likewise.
1219         (ipa_record_reference): Likewise.
1220         * ipa-reference.c (ipa_reference_read_optimization_summary): Likewise.
1221         (ipa_reference_write_optimization_summary): Likewise.
1222         (write_node_summary_p): Likewise.
1223         (propagate): Likewise.
1224         (read_write_all_from_decl): Likewise.
1225         (generate_summary): Likewise.
1226         (analyze_function): Likewise.
1227         (propagate_bits): Likewise.
1228         (ipa_reference_get_not_written_global): Likewise.
1229         (ipa_reference_get_not_read_global): Likewise.
1230         * ipa-split.c (execute_split_functions): Likewise.
1231         (split_function): Likewise.
1232         * ipa-utils.c (ipa_merge_profiles): Likewise.
1233         (dump_cgraph_node_set): Likewise.
1234         (ipa_reverse_postorder): Likewise.
1235         (ipa_edge_within_scc): Likewise.
1236         (ipa_get_nodes_in_cycle): Likewise.
1237         (ipa_free_postorder_info): Likewise.
1238         (ipa_reduced_postorder): Likewise.
1239         (searchc): Likewise.
1240         (recursive_call_p): Likewise.
1241         * ipa.c (ipa_cdtor_merge): Likewise.
1242         (record_cdtor_fn): Likewise.
1243         (function_and_variable_visibility): Likewise.
1244         (varpool_externally_visible_p): Likewise.
1245         (cgraph_externally_visible_p): Likewise.
1246         (comdat_can_be_unshared_p): Likewise.
1247         (comdat_can_be_unshared_p_1): Likewise.
1248         (address_taken_from_non_vtable_p): Likewise.
1249         (ipa_discover_readonly_nonaddressable_vars): Likewise.
1250         (symtab_remove_unreachable_nodes): Likewise.
1251         (walk_polymorphic_call_targets): Likewise.
1252         (process_references): Likewise.
1253         (enqueue_node): Likewise.
1254         (has_addr_references_p): Likewise.
1255         (cgraph_non_local_node_p_1): Likewise.
1256         * is-a.h (varpool_analyze_node): Likewise.
1257         * lto-cgraph.c (input_symtab): Likewise.
1258         (merge_profile_summaries): Likewise.
1259         (input_cgraph_1): Likewise.
1260         (input_edge): Likewise.
1261         (input_varpool_node): Likewise.
1262         (input_node): Likewise.
1263         (input_overwrite_node): Likewise.
1264         (compute_ltrans_boundary): Likewise.
1265         (output_refs): Likewise.
1266         (lto_output_varpool_node): Likewise.
1267         (lto_output_node): Likewise.
1268         (reachable_from_other_partition_p): Likewise.
1269         (referenced_from_other_partition_p): Likewise.
1270         (lto_output_edge): Likewise.
1271         (output_node_opt_summary): Likewise.
1272         (add_node_to): Likewise.
1273         (reachable_from_this_partition_p): Likewise.
1274         (lto_set_symtab_encoder_in_partition): Likewise.
1275         (lto_symtab_encoder_in_partition_p): Likewise.
1276         (lto_set_symtab_encoder_encode_initializer): Likewise.
1277         (lto_symtab_encoder_encode_initializer_p): Likewise.
1278         (lto_set_symtab_encoder_encode_body): Likewise.
1279         (lto_symtab_encoder_encode_body_p): Likewise.
1280         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
1281         Likewise.
1282         * lto-streamer-in.c (lto_read_body): Likewise.
1283         (fixup_call_stmt_edges): Likewise.
1284         (fixup_call_stmt_edges_1): Likewise.
1285         * lto-streamer-out.c (produce_symtab): Likewise.
1286         (output_symbol_p): Likewise.
1287         (write_symbol): Likewise.
1288         (lto_output): Likewise.
1289         (copy_function): Likewise.
1290         (output_function): Likewise.
1291         * passes.c (function_called_by_processed_nodes_p): Likewise.
1292         (ipa_write_optimization_summaries): Likewise.
1293         (ipa_write_summaries): Likewise.
1294         (do_per_function_toporder): Likewise.
1295         (do_per_function): Likewise.
1296         (dump_passes): Likewise.
1297         * symtab.c (symtab_semantically_equivalent_p): Likewise.
1298         (symtab_nonoverwritable_alias): Likewise.
1299         (symtab_nonoverwritable_alias_1): Likewise.
1300         (symtab_for_node_and_aliases): Likewise.
1301         (symtab_resolve_alias): Likewise.
1302         (fixup_same_cpp_alias_visibility): Likewise.
1303         (symtab_alias_ultimate_target): Likewise.
1304         (symtab_used_from_object_file_p): Likewise.
1305         (verify_symtab_base): Likewise.
1306         (dump_symtab_base): Likewise.
1307         (symtab_node_name): Likewise.
1308         (symtab_node_asm_name): Likewise.
1309         (symtab_dissolve_same_comdat_group_list): Likewise.
1310         (symtab_add_to_same_comdat_group): Likewise.
1311         (symtab_unregister_node): Likewise.
1312         (symtab_insert_node_to_hashtable): Likewise.
1313         (symtab_register_node): Likewise.
1314         (unlink_from_assembler_name_hash): Likewise.
1315         (insert_to_assembler_name_hash): Likewise.
1316         (eq_assembler_name): Likewise.
1317         (hash_node_by_assembler_name): Likewise.
1318         (eq_node): Likewise.
1319         (hash_node): Likewise.
1320         * toplev.c (wrapup_global_declaration_2): Likewise.
1321         * trans-mem.c (ipa_tm_execute): Likewise.
1322         (ipa_tm_transform_clone): Likewise.
1323         (ipa_tm_transform_transaction): Likewise.
1324         (ipa_tm_transform_calls_redirect): Likewise.
1325         (ipa_tm_insert_gettmclone_call): Likewise.
1326         (ipa_tm_insert_irr_call): Likewise.
1327         (ipa_tm_create_version): Likewise.
1328         (ipa_tm_create_version_alias): Likewise.
1329         (ipa_tm_mark_forced_by_abi_node): Likewise.
1330         (ipa_tm_mark_force_output_node): Likewise.
1331         (ipa_tm_diagnose_tm_safe): Likewise.
1332         (ipa_tm_mayenterirr_function): Likewise.
1333         (ipa_tm_scan_irr_function): Likewise.
1334         (ipa_tm_note_irrevocable): Likewise.
1335         (ipa_tm_scan_calls_clone): Likewise.
1336         (get_cg_data): Likewise.
1337         * tree-eh.c (tree_could_trap_p): Likewise.
1338         * tree-emutls.c (ipa_lower_emutls): Likewise.
1339         (create_emultls_var): Likewise.
1340         (lower_emutls_function_body): Likewise.
1341         (gen_emutls_addr): Likewise.
1342         (emutls_decl): Likewise.
1343         (new_emutls_decl): Likewise.
1344         * tree-inline.c (tree_function_versioning): Likewise.
1345         (optimize_inline_calls): Likewise.
1346         (expand_call_inline): Likewise.
1347         (estimate_num_insns): Likewise.
1348         (copy_bb): Likewise.
1349         (delete_unreachable_blocks_update_callgraph): Likewise.
1350         * tree-nested.c (gimplify_all_functions): Likewise.
1351         (create_nesting_tree): Likewise.
1352         (check_for_nested_with_variably_modified): Likewise.
1353         * tree-pretty-print.c (dump_function_header): Likewise.
1354         * tree-profile.c (tree_profiling): Likewise.
1355         * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
1356         (modify_function): Likewise.
1357         (convert_callers): Likewise.
1358         (convert_callers_for_node): Likewise.
1359         * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
1360         (associate_varinfo_to_alias): Likewise.
1361         (create_variable_info_for): Likewise.
1362         (get_constraint_for_ssa_var): Likewise.
1363         * tree-vectorizer.c (increase_alignment): Likewise.
1364         * tree.c (find_decls_types_in_var): Likewise.
1365         (find_decls_types_in_node): Likewise.
1366         (free_lang_data_in_decl): Likewise.
1367         * value-prof.c (gimple_ic_transform): Likewise.
1368         (gimple_ic): Likewise.
1369         (check_ic_target): Likewise.
1370         (init_node_map): Likewise.
1371         * varasm.c (decl_binds_to_current_def_p): Likewise.
1372         (default_binds_local_p_1): Likewise.
1373         (dump_tm_clone_pairs): Likewise.
1374         (assemble_alias): Likewise.
1375         (find_decl): Likewise.
1376         (mark_decl_referenced): Likewise.
1377         * varpool.c (varpool_for_node_and_aliases): Likewise.
1378         (varpool_extra_name_alias): Likewise.
1379         (varpool_create_variable_alias): Likewise.
1380         (add_new_static_var): Likewise.
1381         (varpool_finalize_named_section_flags): Likewise.
1382         (varpool_remove_unreferenced_decls): Likewise.
1383         (enqueue_node): Likewise.
1384         (varpool_assemble_decl): Likewise.
1385         (assemble_aliases): Likewise.
1386         (varpool_analyze_node): Likewise.
1387         (cgraph_variable_initializer_availability): Likewise.
1388         (varpool_add_new_variable): Likewise.
1389         (ctor_for_folding): Likewise.
1390         (dump_varpool_node): Likewise.
1391         (varpool_remove_initializer): Likewise.
1392         (varpool_remove_node): Likewise.
1393         (varpool_node_for_decl): Likewise.
1394         (varpool_create_empty_node): Likewise.
1395         * config/i386/i386.c (ix86_generate_version_dispatcher_body): Likewise.
1396         (ix86_get_function_versions_dispatcher): Likewise.
1398 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1400         * cgraph.h (symtab_node_base): Convert to a class;
1401         add GTY((desc ("%h.type"), tag ("SYMTAB_SYMBOL"))), and take
1402         chain_next/prev from symtab_node_def.
1403         (cgraph_node): Inherit from symtab_node; add GTY option
1404         tag ("SYMTAB_FUNCTION").
1405         (varpool_node): Inherit from symtab_node; add GTY option
1406         tag ("SYMTAB_VARIABLE").
1407         (symtab_node_def): Remove.
1408         (is_a_helper <cgraph_node>::test (symtab_node_def *)): Convert to...
1409         (is_a_helper <cgraph_node>::test (symtab_node_base *)): ...this.
1410         (is_a_helper <varpool_node>::test (symtab_node_def *)): Convert to...
1411         (is_a_helper <varpool_node>::test (symtab_node_base *)): ...this.
1413         * ipa-ref.h (symtab_node_def): Drop.
1414         (symtab_node): Change underlying type from symtab_node_def to
1415         symtab_node_base.
1416         (const_symtab_node): Likwise.
1418         * is-a.h: Update examples in comment.
1420         * symtab.c (symtab_hash): Change symtab_node_def to symtab_node_base.
1421         (assembler_name_hash): Likewise.
1423 2013-10-29  Martin Liska  <marxin.liska@gmail.com>
1425         * doc/tree-ssa.texi (gimple_phi_result): Document.
1426         (gimple_phi_num_args, gimple_phi_arg): Likewise.
1427         (gimple_phi_arg_edge, gimple_phi_arg_def): Likewise.
1428         (PHI_RESULT, PHI_NUM_ARGS): Remove.
1429         (PHI_ARG_ELT, PHI_ARG_EDGE, PHI_ARG_DEF): Likewise.
1431 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1433         * expr.h: Revert change and include tree-core.h.
1434         * rtl.h: Revert change and don't include tree-core.h.
1436 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1438         * config/darwin.c: Include gimple.h.
1439         * config/i386/winnt.c: Likewise.
1441 2013-10-29  Marc Glisse  <marc.glisse@inria.fr>
1443         PR tree-optimization/19831
1444         * tree-ssa-alias.c (stmt_kills_ref_p_1): Handle BUILT_IN_FREE.
1446 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1448         * tree-outof-ssa.h: Remove include files.
1449         * tree-outof-ssa.c: Add required include files from tree-outof-ssa.h.
1450         * expr.c: Likewise.
1451         * tree-ssa-coalesce.c: Likewise.
1452         * cfgexpand.c: Likewise.
1453         * tree-ssa-ter.c: Likewise.
1454         * ipa-prop.h: Remove gimple.h and tree-core.h from include list.
1455         * lto-streamer.h: Likewise.
1456         * cgraphbuild.c: Add gimple.h to include list.
1457         * data-streamer-in.c: Likewise.
1458         * ipa-cp.c: Likewise.
1459         * tree-streamer.c: Likewise.
1460         * lto-compress.c: Likewise.
1461         * ipa-reference.c: Likewise.
1462         * data-streamer-out.c: Likewise.
1463         * lto-cgraph.c: Likewise.
1464         * cgraphclones.c: Likewise.
1465         * ipa-utils.c: Likewise.
1466         * data-streamer.c: Likewise.
1467         * ipa-split.c: Likewise.
1468         * lto-section-in.c: Likewise.
1469         * tree-streamer-out.c: Likewise.
1470         * ipa-prop.c: Likewise.
1471         * tree-streamer-in.c: Likewise.
1472         * symtab.c: Likewise.
1473         * opts-global.c: Likewise.
1474         * lto-opts.c: Likewise.
1475         * lto-section-out.c: Likewise.
1476         * lto-streamer.c: Likewise.
1477         * rtl.h: Add tree-core.h to include list.
1478         * expr.h: Remove tree-core.h from include list.
1479         * gimple.h: Likewise.
1480         * ipa-utils.h: Likewise.
1481         * streamer-hooks.h: Likewise.
1482         * streamer-hooks.c: Include input.h.
1484 2013-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1486         * config/arm/arm.c (cortexa7_extra_costs): New table.
1487         (arm_cortex_a7_tune): New.
1488         * config/arm/arm-cores.def: Use cortex_a7 tuning for cortex-a7.
1490 2013-10-29  Eric Botcazou  <ebotcazou@adacore.com>
1492         * expr.c (expand_expr_real_1) <MEM_EXPR>: Eliminate small redundancy.
1494 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1496         * doc/gty.texi ("Inheritance and GTY"): Make it clear that
1497         to use autogenerated markers for a class-hierarchy, every class
1498         must have a GTY marker.
1499         * gengtype.h (struct type): Add linked list of subclasses to
1500         the "s" member of the union.
1501         (add_subclass): New decl.
1502         * gengtype-state.c (read_state_struct_type): Set up subclass
1503         linked list.
1504         * gengtype.c (get_ultimate_base_class): New.
1505         (add_subclass): New.
1506         (new_structure): Set up subclass linked list.
1507         (set_gc_used_type): Propagate usage information to subclasses.
1508         (output_mangled_typename): Use get_ultimate_base_class.
1509         (walk_subclasses): Use the subclass linked list, avoiding an
1510         O(N^2) when writing out all types.
1511         (walk_type): Issue an error if the base class is missing a tag,
1512         rather than generating bogus C code.  Add a gcc_unreachable
1513         default case, in case people omit tags from concrete subclasses,
1514         or get the values wrong.
1515         (write_func_for_structure): Issue an error for subclasses for
1516         which the base doesn't have a "desc", since otherwise the
1517         autogenerated routines for the base would silently fail to visit
1518         any subclass fields.
1519         (write_root): Use get_ultimate_base_class, tweaking constness of
1520         tp to match that function's signature.
1522 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1524         * doc/gty.texi (GTY Options): Add note about inheritance to
1525         description of desc and tag.
1526         (Inheritance and GTY): New.
1528 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1530         * gengtype-parse.c (opts_have): Drop "static" so that
1531         we can use this from gengtype.c.
1532         * gengtype.c (set_gc_used_type): Mark any base class as used;
1533         update field traversal to visit inherited fields.
1534         (output_mangled_typename):  Convert references to classes within
1535         an inheritance hierarchy to reference the ultimate base class,
1536         since only it will have gt_ functions.
1537         (get_string_option): New.
1538         (walk_subclasses): New.
1539         (walk_type): Treat GTY structs that have a "desc" as being the
1540         root of an inheritance hierarchy.  Generate a switch on it
1541         within the marking function which walks all subclasses, adding
1542         cases for them via walk_subclasses.  For subclasses, visit all
1543         fields of the type (including inherited ones).
1544         (write_func_for_structure): Don't write fns for subclasses, only
1545         for the ultimate base class within an inheritance hierarchy.
1546         Subclasses-marking will be handled by the base class marking functions.
1547         (write_types): Likewise.
1548         (write_local_func_for_structure): Likewise.
1549         (USED_BY_TYPED_GC_P): Emit allocators for subclasses that have
1550         a "tag" option (and are thus concrete subclasses).
1551         (write_root): Use the marker function for the ultimate base class.
1552         * gengtype.h (FOR_ALL_INHERITED_FIELDS): New.
1553         (opts_have): Add declaration.
1555 2013-10-28  Vladimir Makarov  <vmakarov@redhat.com>
1557         * lra-spills.c (lra_final_code_change): Remove useless move insns
1558         originated from moves of pseudos.
1560 2013-10-28  Jeff Law  <law@redhat.com>
1562         * lower-subreg.c (resolve_simple_move): Fix comment typo.
1564 2013-10-28  Trevor Saunders  <tsaunders@mozilla.com>
1566         * df-scan.c (df_collection_rec): Adjust.
1567         (copy_defs): New constant.
1568         (copy_uses): Likewise.
1569         (copy_eq_uses): Likewise.
1570         (copy_mw): Likewise.
1571         (copy_all): Likewise.
1572         (df_insn_rescan): Adjust.
1573         (df_notes_rescan): Likewise.
1574         (df_swap_refs): Likewise.
1575         (df_sort_and_compress_refs): Likewise.
1576         (df_sort_and_compress_mws): Likewise.
1577         (df_install_refs): Likewise.
1578         (df_install_mws): Likewise.
1579         (df_refs_add_to_chains): Add flags parameter controlling which vectors
1580         are coppied.
1581         (df_bb_refs_record): Adjust.
1582         (df_record_entry_block_defs): Likewise.
1583         (df_record_exit_block_defs): Likewise.
1584         (df_refs_verify): Likewise.
1585         (df_mws_verify): Likewise.
1586         (df_insn_refs_verify): Likewise.
1587         (df_bb_verify): Likewise.
1588         * ipa-pure-const.c (finish_state): Remove.
1589         (propagate): Adjust.
1590         * tree-data-ref.c tree-ssa-alias.c tree-ssa-loop-ivcanon.c
1591         tree-ssa-threadedge.c tree-vect-loop-manip.c tree-vect-slp.c
1592         var-tracking.c: Adjust.
1593         * vec.c (stack_vecs): Remove.
1594         (register_stack_vec): Likewise.
1595         (stack_vec_register_index): Likewise.
1596         (unregister_stack_vec): Likewise.
1597         * vec.h (struct va_stack): Remove.
1598         (struct vec<T, A, vl_ptr>): Specialize as
1599         struct vec<T, va_heap, vl_ptr> instead since va_heap is the only
1600         allocation strategy compatable with the vl_ptr layout.
1601         (struct vec<T, va_gc, vl_ptr>): Remove because it now gets an empty
1602         specialization anyway.
1603         (class stack_vec): New class.
1604         (vec_stack_alloc): Remove.
1605         (vec<T, va_heap, vl_ptr>::using_auto_storage): New method.
1607 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1608             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1609             Sergey Lega  <sergey.s.lega@intel.com>
1610             Anna Tikhonova  <anna.tikhonova@intel.com>
1611             Ilya Tocar  <ilya.tocar@intel.com>
1612             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1613             Ilya Verbin  <ilya.verbin@intel.com>
1614             Kirill Yukhin  <kirill.yukhin@intel.com>
1615             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1617         * config/i386/i386.md (prefetch): Allow TARGET_AVX512PF.
1618         (*prefetch_avx512pf_<mode>): New.
1619         * config/i386/sse.md (avx512f_vmcmp<mode>3): Ditto.
1620         (avx512f_maskcmp<mode>3): Ditto.
1621         (vashrv16si3): Ditto.
1623 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1624             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1625             Sergey Lega  <sergey.s.lega@intel.com>
1626             Anna Tikhonova  <anna.tikhonova@intel.com>
1627             Ilya Tocar  <ilya.tocar@intel.com>
1628             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1629             Ilya Verbin  <ilya.verbin@intel.com>
1630             Kirill Yukhin  <kirill.yukhin@intel.com>
1631             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1633         * config/i386/i386.md (any_truncate): New.
1634         (trunsuffix): Ditto.
1635         * config/i386/predicates.md (const_8_to_9_operand): New.
1636         (const_10_to_11_operand): Ditto.
1637         (const_12_to_13_operand): Ditto.
1638         (const_14_to_15_operand): Ditto.
1639         (const_16_to_19_operand): Ditto.
1640         (const_20_to_23_operand): Ditto.
1641         (const_24_to_27_operand): Ditto.
1642         (const_28_to_31_operand): Ditto.
1643         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_FIX_NOTRUNC.
1644         (cvtusi2<ssescalarmodesuffix>32): New.
1645         (cvtusi2<ssescalarmodesuffix>64): Ditto.
1646         (ufloatv16siv16sf2): Ditto.
1647         (avx512f_fix_notruncv16sfv16si): Ditto.
1648         (avx512f_ufix_notruncv16sfv16si): Ditto.
1649         (avx512f_vcvtss2usi): Ditto.
1650         (avx512f_vcvtss2usiq): Ditto.
1651         (avx512f_vcvttss2usi): Ditto.
1652         (avx512f_vcvttss2usiq): Ditto.
1653         (avx512f_vcvtsd2usi): Ditto.
1654         (avx512f_vcvtsd2usiq): Ditto.
1655         (avx512f_vcvttsd2usi): Ditto.
1656         (avx512f_vcvttsd2usiq): Ditto.
1657         (ufloatv8siv8df): Ditto.
1658         (avx512f_cvtdq2pd512_2): Ditto.
1659         (avx512f_cvtpd2dq512): Ditto.
1660         (avx512f_ufix_notruncv8dfv8si): Ditto.
1661         (avx512f_cvtpd2ps512): Ditto.
1662         (vec_unpacks_lo_v16sf): Ditto.
1663         (vec_unpacks_hi_v16sf): Ditto.
1664         (vec_unpacks_float_hi_v16si): Ditto.
1665         (vec_unpacks_float_lo_v16si): Ditto.
1666         (avx512f_unpckhps512): Ditto.
1667         (avx512f_unpcklps512): Ditto.
1668         (avx512f_movshdup512): Ditto.
1669         (avx512f_movsldup512): Ditto.
1670         (vec_extract_lo_v32hi): Ditto.
1671         (vec_extract_hi_v32hi): Ditto.
1672         (vec_extract_lo_v64qi): Ditto.
1673         (vec_extract_hi_v64qi): Ditto.
1674         (avx512f_unpckhpd512): Ditto.
1675         (avx512f_movddup512): Ditto.
1676         (avx512f_unpcklpd512): Ditto.
1677         (*avx512f_unpcklpd512): Ditto.
1678         (avx512f_shufps512_1): Ditto.
1679         (avx512f_shufpd512_1): Ditto.
1680         (avx512f_interleave_highv8di): Ditto.
1681         (avx512f_interleave_lowv8di): Ditto.
1682         (PMOV_DST_MODE): Ditto.
1683         (pmov_src_mode): Ditto.
1684         (pmov_src_lower): Ditto.
1685         (pmov_suff): Ditto.
1686         (*avx512f_<code><pmov_src_lower><mode>2): Ditto.
1687         (*avx512f_<code>v8div16qi2): Ditto.
1688         (*avx512f_<code>v8div16qi2_store): Ditto.
1689         (vec_widen_umult_even_v16si): Ditto.
1690         (*vec_widen_umult_even_v16si): Ditto.
1691         (vec_widen_smult_even_v16si): Ditto.
1692         (*vec_widen_smult_even_v16si): Ditto.
1693         (avx512f_interleave_highv16si): Ditto.
1694         (avx512f_interleave_lowv16si): Ditto.
1695         (avx512f_<code>v16qiv16si2): Ditto.
1696         (avx512f_<code>v16hiv16si2): Ditto.
1697         (avx512f_<code>v8qiv8di2): Ditto.
1698         (avx512f_<code>v8hiv8di2): Ditto.
1699         (avx512f_<code>v8siv8di2): Ditto.
1700         (avx512cd_maskb_vec_dupv8di): Ditto.
1701         (avx512cd_maskw_vec_dupv16si): Ditto.
1702         (avx512f_vcvtph2ps512): Ditto.
1703         (avx512f_vcvtps2ph512): Ditto.
1704         (VEC_EXTRACT_MODE): Extened with wider modes.
1705         (VEC_PERM_AVX2): Ditto.
1706         (VEC_PERM_CONST): Ditto.
1708 2013-10-28  Joern Rennecke  <joern.rennecke@embecosm.com>
1710         * config/arc/arc.c (arc_ccfsm_post_advance):
1711         Add comment about TYPE_RETURN.
1713 2013-10-28  Bin Cheng  <bin.cheng@arm.com>
1715         * tree-ssa-loop-ivopts.c (strip_offset_1): Change parameter type.
1716         Count DECL_FIELD_BIT_OFFSET for COMPONENT_REF.
1717         (strip_offset): Convert offset to unsigned number.
1719 2013-10-27  Tom de Vries  <tom@codesourcery.com>
1721         * cfgexpand.c (gimple_expand_cfg): Remove test for parm_birth_insn.
1722         Don't commit insertions after NOTE_INSN_FUNCTION_BEG.
1724 2013-10-27  Oleg Endo  <olegendo@gcc.gnu.org>
1726         * config/sh/sh.c (MSW, LSW): Move and rename macros to...
1727         * config/sh/sh.h (SH_REG_MSW_OFFSET, SH_REG_LSW_OFFSET): ... here.
1728         (TARGET_BIG_ENDIAN): New macro.
1729         * config/sh/sh.md: Use it instead of !TARGET_LITTLE_ENDIAN.
1730         Use SH_REG_MSW_OFFSET and SH_REG_LSW_OFFSET.
1731         * config/sh/sh.c: Likewise.
1732         * config/sh/sh.h: Likewise.
1734 2013-10-27  Hans-Peter Nilsson  <hp@axis.com>
1736         * config/cris/cris.c (cris_emit_trap_for_misalignment): Replace the
1737         removed PRED_MUDFLAP with PRED_NORETURN.  Correct file-path in comment.
1739 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1741         * config/sh/sh.md (movmemsi): Remove empty constraints and predicates.
1742         Fix formatting.
1743         (cmpstr_t, cmpstrsi): Fix formatting.
1745 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1747         PR target/52483
1748         * config/sh/predicates.md (general_movdst_operand): Allow reg+reg
1749         addressing, do not use general_operand for memory operands.
1751 2013-10-26  Vladimir Makarov  <vmakarov@redhat.com>
1753         Revert:
1754         2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1755         * lra-spills.c (lra_final_code_change): Remove useless move insns.
1757 2013-10-26  Jeff Law  <law@redhat.com>
1759         * predict.c (PRED_MUDFLAP): Remove.
1760         * targhooks.c (build_va_arg_indirect_ref): Remove mudflap support.
1762         * Makefile.in (C_COMMON_OBJS): Remove tree-mudflap.
1763         (OBJS): Remove tree-nomudflap.o
1764         (GTFILES): Remove tree-mudflap.c
1765         * builtins.c (expand_builtin_alloc): Remove mudflap support.
1766         * gcc.c (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
1767         (mfwrap_spec, mflib_spec): Likewise.
1768         (cpp_unique_options, cc1_options, static_specs): Likewise.
1769         * gimplify (gimplify_vla_decl, build_va_arg_indirect_ref): Likewise.
1770         * passes.def: Likewise.
1771         * toplev.c (compile_file, process_options): Likewise.
1772         * tree-inline.c (copy_tree_r): Likewise.
1773         * tree-pass.,h (make_pass_mudflap_1, make_pass_mudflap_2): Likewise.
1774         * varasm.c (make_decl_rtl, make_decl_rtl_for_debug): Likewise.
1775         (build_constant_desc, output_constant_def_contents): Likewise.
1776         (categorize_decl_for_section): Likewise.
1777         * tree-mudflap.c: Removed.
1778         * tree-mudflap.h: Removed.
1779         * tree-nomudflap.c: Removed.
1780         * bfin/uclinux.h (MFWRAP_SPEC): Remove.
1781         * moxie/uclinux.h (MFWRAP_SPEC): Likewise.
1782         * rs6000/aix.h (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
1783         * config/sol2.h (MFLIB_SPEC): Likewise.
1784         * doc/install.texi: Remove mudflap references.
1785         * doc/passes.texi: Similarly.
1786         * doc/sourcebuild.texi: Similarly.
1787         * doc/invoke.texi: Remove mudlfap related options.
1789 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1791         PR rtl-optimization/58759
1792         * lra-constraints.c (lra_constraints): Remove wrong condition to
1793         remove insn setting up an equivalent pseudo.
1795 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1797         * config/rs6000/rs6000-protos.h
1798         (rs6000_secondary_memory_needed_mode): New prototype.
1799         * config/rs6000/rs6000.c: Include ira.h.
1800         (TARGET_LRA_P): Redefine.
1801         (rs6000_legitimate_offset_address_p): Call
1802         legitimate_constant_pool_address_p in strict mode for LRA.
1803         (rs6000_legitimate_address_p): Ditto.
1804         (legitimate_lo_sum_address_p): Add code for LRA.  Use lra_in_progress.
1805         (rs6000_emit_move): Add LRA version of code to generate load/store
1806         of SDmode values.
1807         (rs6000_secondary_memory_needed_mode): New.
1808         (rs6000_alloc_sdmode_stack_slot): Do nothing for LRA.
1809         (rs6000_secondary_reload_class): Return NO_REGS for LRA for
1810         constants, memory, and FP registers.
1811         (rs6000_lra_p): New.
1812         * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): New macro.
1813         * config/rs6000/rs6000.opt (mlra): New option.
1814         * lra-spills.c (lra_final_code_change): Remove useless move insns.
1816 2013-10-25  Yufeng Zhang  <yufeng.zhang@arm.com>
1818         * tree-ssa-math-opts.c (convert_plusminus_to_widen): Call
1819         has_single_use () and not do the conversion if has_single_use ()
1820         returns false for the multiplication result.
1822 2013-10-25  David Malcolm  <dmalcolm@redhat.com>
1824         * tree.h (EXCEPTIONAL_CLASS_P): Rename parameter from "CODE"
1825         to "NODE", since this works on a "tree", not an
1826         "enum tree_code".
1827         (CONSTANT_CLASS_P): Likewise.
1828         (TYPE_P): Likewise.
1829         (DECL_P): Likewise.
1830         (INDIRECT_REF_P): Likewise.
1831         (REFERENCE_CLASS_P): Likewise.
1832         (COMPARISON_CLASS_P): Likewise.
1833         (UNARY_CLASS_P): Likewise.
1834         (BINARY_CLASS_P): Likewise.
1835         (STATEMENT_CLASS_P): Likewise.
1836         (VL_EXP_CLASS_P): Likewise.
1837         (EXPRESSION_CLASS_P): Likewise.
1838         (IS_TYPE_OR_DECL_P): Likewise.
1840 2013-10-25  Marc Glisse  <marc.glisse@inria.fr>
1842         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for an
1843         ADDR_EXPR in the defining statement.
1845 2013-10-25  Richard Biener  <rguenther@suse.de>
1847         PR tree-optimization/58626
1848         * tree-loop-distribution.c (enum rdg_dep_type): Remove
1849         anti_dd, output_dd and input_dd.
1850         (struct rdg_edge): Remove level and relation members.
1851         (RDGE_LEVEL, RDGE_RELATION): Remove.
1852         (dot_rdg_1): Adjust.
1853         (create_rdg_edge_for_ddr): Remove.
1854         (create_rdg_edges_for_scalar): Adjust.
1855         (create_edge_for_control_dependence): Likewise.
1856         (create_rdg_edges): Split into ...
1857         (create_rdg_flow_edges): ... this
1858         (create_rdg_cd_edges): ... and this.
1859         (free_rdg): Adjust.
1860         (build_rdg): Likewise, do not compute data dependences or
1861         add edges for them.
1862         (pg_add_dependence_edges): New function.
1863         (pgcmp): Likewise.
1864         (distribute_loop): First apply all non-dependence based
1865         partition mergings.  Then compute dependences between partitions
1866         and merge and order partitions according to them.
1868 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
1870         PR rtl-optimization/58831
1871         * alias.c (init_alias_analysis): At the beginning of each iteration,
1872         set the reg_seen[N] bit if static_reg_base_value[N] is non-null.
1874 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
1876         * recog.c (search_ofs): New static variable moved from...
1877         (peep2_find_free_register): ...here.
1878         (peephole2_optimize): Initialize it.
1880 2013-10-25  Tobias Burnus  <burnus@net-b.de>
1882         * doc/invoke.texi (fopenmp): Change supported OpenMP version to 4.0.
1884 2013-10-25  Uros Bizjak  <ubizjak@gmail.com>
1886         * config/i386/i386.h (TARGET_MPX): New define.
1887         (TARGET_MPX_P): Ditto.
1889 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
1891         * config/i386/constraints.md (B): New.
1892         (Ti): New.
1893         (Tb): New.
1894         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
1895         * config/i386/i386-modes.def (BND32): New.
1896         (BND64): New.
1897         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
1898         * config/i386/i386.c (isa_opts): Add mmpx.
1899         (regclass_map): Add bound registers.
1900         (dbx_register_map): Likewise.
1901         (dbx64_register_map): Likewise.
1902         (svr4_dbx_register_map): Likewise.
1903         (PTA_MPX): New.
1904         (ix86_option_override_internal): Support MPX ISA.
1905         (ix86_conditional_register_usage): Support bound registers.
1906         (print_reg): Likewise.
1907         (ix86_code_end): Add MPX bnd prefix.
1908         (output_set_got): Likewise.
1909         (ix86_output_call_insn): Likewise.
1910         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
1911         (ix86_print_operand_punct_valid_p): Likewise.
1912         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
1913         UNSPEC_BNDMK_ADDR.
1914         (ix86_class_likely_spilled_p): Add bound regs support.
1915         (ix86_hard_regno_mode_ok): Likewise.
1916         (x86_order_regs_for_local_alloc): Likewise.
1917         (ix86_bnd_prefixed_insn_p): New.
1918         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
1919         (FIXED_REGISTERS): Add bound registers.
1920         (CALL_USED_REGISTERS): Likewise.
1921         (REG_ALLOC_ORDER): Likewise.
1922         (HARD_REGNO_NREGS): Likewise.
1923         (TARGET_MPX): New.
1924         (VALID_BND_REG_MODE): New.
1925         (FIRST_BND_REG): New.
1926         (LAST_BND_REG): New.
1927         (reg_class): Add BND_REGS.
1928         (REG_CLASS_NAMES): Likewise.
1929         (REG_CLASS_CONTENTS): Likewise.
1930         (BND_REGNO_P): New.
1931         (ANY_BND_REG_P): New.
1932         (BNDmode): New.
1933         (HI_REGISTER_NAMES): Add bound registers.
1934         * config/i386/i386.md (UNSPEC_BNDMK): New.
1935         (UNSPEC_BNDMK_ADDR): New.
1936         (UNSPEC_BNDSTX): New.
1937         (UNSPEC_BNDLDX): New.
1938         (UNSPEC_BNDLDX_ADDR): New.
1939         (UNSPEC_BNDCL): New.
1940         (UNSPEC_BNDCU): New.
1941         (UNSPEC_BNDCN): New.
1942         (UNSPEC_MPX_FENCE): New.
1943         (BND0_REG): New.
1944         (BND1_REG): New.
1945         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
1946         (length_immediate): Likewise.
1947         (prefix_0f): Likewise.
1948         (memory): Likewise.
1949         (prefix_rep): Check for bnd prefix.
1950         (length_nobnd): New.
1951         (length): Use length_nobnd if specified.
1952         (BND): New.
1953         (bnd_ptr): New.
1954         (BNDCHECK): New.
1955         (bndcheck): New.
1956         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
1957         (*jcc_2): Likewise.
1958         (jump): Likewise.
1959         (simple_return_internal): Likewise.
1960         (simple_return_pop_internal): Likewise.
1961         (*indirect_jump): Add MPX bnd prefix.
1962         (*tablejump_1): Likewise.
1963         (simple_return_internal_long): Likewise.
1964         (simple_return_indirect_internal): Likewise.
1965         (<mode>_mk): New.
1966         (*<mode>_mk): New.
1967         (mov<mode>): New.
1968         (*mov<mode>_internal_mpx): New.
1969         (<mode>_<bndcheck>): New.
1970         (*<mode>_<bndcheck>): New.
1971         (<mode>_ldx): New.
1972         (*<mode>_ldx): New.
1973         (<mode>_stx): New.
1974         (*<mode>_stx): New.
1975         * config/i386/predicates.md (lea_address_operand): Rename to...
1976         (address_no_seg_operand): ... this.
1977         (address_mpx_no_base_operand): New.
1978         (address_mpx_no_index_operand): New.
1979         (bnd_mem_operator): New.
1980         * config/i386/i386.opt (mmpx): New.
1981         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
1982         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
1984 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
1986         * mode-classes.def (MODE_POINTER_BOUNDS): New.
1987         * tree.def (POINTER_BOUNDS_TYPE): New.
1988         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
1989         (POINTER_BOUNDS_MODE): New.
1990         (make_pointer_bounds_mode): New.
1991         * machmode.h (POINTER_BOUNDS_MODE_P): New.
1992         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
1993         (layout_type): Support POINTER_BOUNDS_TYPE.
1994         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
1995         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
1996         (type_contains_placeholder_1): Likewise.
1997         * tree.h (POINTER_BOUNDS_TYPE_P): New.
1998         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
1999         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
2001 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2003         * expr.c (expand_expr_real_1): Use mode of memory reference rather than
2004         mode of address computation when calling memory_address_addr_space.
2006 2013-08-24  Richard Henderson  <rth@twiddle.net>
2008         PR rtl/58542
2009         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
2010         instead of create_convert_operand_to.
2011         (maybe_emit_sync_lock_test_and_set): Likewise.
2012         (expand_atomic_compare_and_swap): Likewise.
2013         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
2015 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
2017         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
2018         Change param_is to use the struct and not the pointer to the struct.
2020 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
2022         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
2023         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2024         Move to targhooks.c.
2025         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
2026         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2027         Relocate from gimplify.c.
2028         * targhooks.h: Add 2 prototypes.
2029         * tree.h: Delete 2 prototypes.
2031 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2033         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
2034         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
2036 2013-10-24  Cong Hou  <congh@google.com>
2038         * convert.c (convert_to_real): Guard those unsafe math function
2039         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
2040         specially.
2042 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
2044         PR ipa/58712
2045         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
2046         as argument.
2047         (cgraph_create_edge): Use the new argument.
2048         (cgraph_create_indirect_edge): Likewise.
2050 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2052         * gcc/config/arc/arc.c (arc_ccfsm_post_advance): Also handle
2053         TYPE_UNCOND_BRANCH.
2054         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
2055         changes statep->state.
2057 2013-10-24  Tobias Burnus  <burnus@net-b.de>
2059         PR other/33426
2060         * tree-cfg.c (replace_loop_annotate): New function.
2061         (execute_build_cfg): Call it.
2062         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
2063         * internal-fn.c (expand_ANNOTATE): New function.
2064         * internal-fn.def (ANNOTATE): Define as new internal function.
2065         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
2066         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
2067         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
2068         * doc/extend.texi (Pragmas): Document #pragma ivdep.
2069         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
2071 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
2072             Marcus Shawcroft  <marcus.shawcroft@arm.com>
2074         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
2075         Special case reload SP+C into none GENERAL_REGS.
2077 2013-10-24  Michael Matz  <matz@suse.de>
2079         * gengtype.c (is_file_equal): Check that files will be same length.
2081 2013-10-25  Christian Bruel  <christian.bruel@st.com>
2083         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
2084         * gcc/config/sh/t-sh (sh-mem.o): New rule.
2085         * gcc/config/sh/sh-mem.cc (expand_block_move): Moved here.
2086         (sh_expand_cmpstr): New function.
2087         * gcc/config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c.
2088         * gcc/config/sh/sh-protos.h (sh_expand_cmpstr): Declare.
2089         * gcc/config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
2090         (rotlhi3_8): Rename.
2092 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2094         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
2095         * Makefile.in (CXXDEPMODE): Assign and change users.
2096         (CCDEPMODE): Delete.
2097         * configure: Regenerate.
2099 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
2101         * gengtype-parse.c (require_without_advance): New.
2102         (type): For GTY-marked types that are not GTY((user)), parse any
2103         base classes, requiring them to be single-inheritance, and not
2104         be templates.  For non-GTY-marked types and GTY((user)),
2105         continue to skip over any C++ inheritance specification.
2106         * gengtype-state.c (state_writer::write_state_struct_type):
2107         Write base class of type (if any).
2108         (read_state_struct_type): Read base class of type (if any).
2109         * gengtype.c (new_structure): Add a "base_class" parameter.
2110         (create_optional_field_): Update for new parameter to new_structure.
2111         (adjust_field_rtx_def): Likewise.
2112         (adjust_field_tree_exp): Likewise.
2113         * gengtype.h (struct type): Add "base_class" field to the s
2114         union field.
2115         (new_structure): Add "base" parameter.
2117 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
2119         PR target/57756
2120         * config/i386/i386.c (ix86_option_override_internal):
2121         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
2122         (ix86_valid_target_attribute_tree):
2123         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
2124         Change TARGET_SSE to TARGET_SSE_P (opts->...)
2126 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2128         * tree-ssa-loop.h: Remove include files.
2129         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2130         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
2131         * graphite-clast-to-gimple.c: Likewise.
2132         * graphite-scop-detection.c: Likewise.
2133         * graphite-sese-to-poly.c: Likewise.
2134         * ipa-inline-analysis.c: Likewise.
2135         * ipa-pure-const.c: Likewise.
2136         * loop-init.c: Likewise.
2137         * passes.c: Likewise.
2138         * predict.c: Likewise.
2139         * tree-cfg.c: Likewise.
2140         * tree-cfgcleanup.c: Likewise.
2141         * tree-chrec.c: Likewise.
2142         * tree-data-ref.c: Likewise.
2143         * tree-loop-distribution.c: Likewise.
2144         * tree-parloops.c: Likewise.
2145         * tree-predcom.c: Likewise.
2146         * tree-scalar-evolution.c: Likewise.
2147         * tree-ssa-address.c: Likewise.
2148         * tree-ssa.c: Likewise.
2149         * tree-ssa-dce.c: Likewise.
2150         * tree-ssa-loop.c: Likewise.
2151         * tree-ssa-loop-im.c: Likewise.
2152         * tree-ssa-loop-ivcanon.c: Likewise.
2153         * tree-ssa-loop-ivopts.c: Likewise.
2154         * tree-ssa-loop-manip.c: Likewise.
2155         * tree-ssa-loop-niter.c: Likewise.
2156         * tree-ssa-loop-prefetch.c: Likewise.
2157         * tree-ssa-loop-unswitch.c: Likewise.
2158         * tree-ssa-reassoc.c: Likewise.
2159         * tree-vect-data-refs.c: Likewise.
2160         * tree-vect-loop.c: Likewise.
2161         * tree-vect-loop-manip.c: Likewise.
2162         * tree-vectorizer.c: Likewise.
2163         * tree-vect-stmts.c: Likewise.
2164         * tree-vrp.c: Likewise.
2166 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2168         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
2170 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2172         PR tree-optimization/58775
2173         PR tree-optimization/58791
2174         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
2175         (insert_stmt_after): Rewritten, don't move the stmt, but really
2176         insert it.
2177         (get_stmt_uid_with_default): Remove.
2178         (build_and_add_sum): Use insert_stmt_after and
2179         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only labels.
2180         (update_range_test): Set uid on stmts added by
2181         force_gimple_operand_gsi.  Don't immediately modify statements
2182         in inter-bb optimization, just update oe->op values.
2183         (optimize_range_tests): Return bool whether any changed have been made.
2184         (update_ops): New function.
2185         (struct inter_bb_range_test_entry): New type.
2186         (maybe_optimize_range_tests): Perform statement changes here.
2187         (not_dominated_by, appears_later_in_bb, get_def_stmt,
2188         ensure_ops_are_available): Remove.
2189         (find_insert_point): Rewritten.
2190         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
2191         return LHS of the (new resp. old) stmt.  Don't call
2192         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
2193         instead of last, move new stmt at the right place.
2194         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
2195         (negate_value): Likewise.  Set uids.
2196         (break_up_subtract_bb): Initialize uids.
2197         (reassociate_bb): Adjust rewrite_expr_tree caller.
2198         (do_reassoc): Don't call renumber_gimple_stmt_uids.
2200 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
2202         PR target/58838
2203         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
2204         TARGET_32BIT final condition.
2205         (mulsi3_internal2 and splitter): Same.
2207 2013-10-23  Jeff Law  <law@redhat.com>
2209         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
2210         through joiner blocks with abnormal outgoing edges.
2212         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
2213         Add parameter JOINERS, to allow/disallow threading through joiner
2214         blocks.
2215         (thread_block): New.  Call thread_block_1.
2216         (mark_threaded_blocks): Remove code to filter out certain cases
2217         of threading through joiner blocks.
2218         (thread_through_all_blocks): Document how we can have a dangling
2219         edge AUX field and clear it.
2221 2013-10-23  Ian Lance Taylor  <iant@google.com>
2223         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
2224         (C++ Dialect Options): Likewise.
2225         (Optimize Options): Likewise.
2227 2013-10-23  Tom de Vries  <tom@codesourcery.com>
2229         PR tree-optimization/58805
2230         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
2232 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2234         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
2235         sequence based on get_range_info returned range.
2237 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2239         * tree-ssa.h: Remove all #include's
2240         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2241         * alias.c: Move required includes from tree-ssa.h.
2242         * asan.c: Likewise.
2243         * builtins.c: Likewise.
2244         * calls.c: Likewise.
2245         * cfgexpand.c: Likewise.
2246         * cfghooks.c: Likewise.
2247         * cfgloop.c: Likewise.
2248         * cfgloopmanip.c: Likewise.
2249         * cgraph.c: Likewise.
2250         * cgraphbuild.c: Likewise.
2251         * cgraphclones.c: Likewise.
2252         * cgraphunit.c: Likewise.
2253         * dse.c: Likewise.
2254         * except.c: Likewise.
2255         * expr.c: Likewise.
2256         * final.c: Likewise.
2257         * fold-const.c: Likewise.
2258         * ggc-page.c: Likewise.
2259         * gimple-builder.c: Likewise.
2260         * gimple-fold.c: Likewise.
2261         * gimple-iterator.c: Likewise.
2262         * gimple-low.c: Likewise.
2263         * gimple-pretty-print.c: Likewise.
2264         * gimple-ssa-strength-reduction.c: Likewise.
2265         * gimple-streamer-in.c: Likewise.
2266         * gimple-streamer-out.c: Likewise.
2267         * gimplify.c: Likewise.
2268         * graphite-blocking.c: Likewise.
2269         * graphite-clast-to-gimple.c: Likewise.
2270         * graphite-dependences.c: Likewise.
2271         * graphite-interchange.c: Likewise.
2272         * graphite-optimize-isl.c: Likewise.
2273         * graphite-poly.c: Likewise.
2274         * graphite-scop-detection.c: Likewise.
2275         * graphite-sese-to-poly.c: Likewise.
2276         * graphite.c: Likewise.
2277         * ipa-cp.c: Likewise.
2278         * ipa-inline-analysis.c: Likewise.
2279         * ipa-inline-transform.c: Likewise.
2280         * ipa-inline.c: Likewise.
2281         * ipa-prop.c: Likewise.
2282         * ipa-pure-const.c: Likewise.
2283         * ipa-reference.c: Likewise.
2284         * ipa-split.c: Likewise.
2285         * ipa-utils.c: Likewise.
2286         * loop-init.c: Likewise.
2287         * lto-cgraph.c: Likewise.
2288         * lto-section-in.c: Likewise.
2289         * lto-section-out.c: Likewise.
2290         * lto-streamer-in.c: Likewise.
2291         * lto-streamer-out.c: Likewise.
2292         * lto-streamer.c: Likewise.
2293         * omp-low.c: Likewise.
2294         * passes.c: Likewise.
2295         * predict.c: Likewise.
2296         * print-tree.c: Likewise.
2297         * profile.c: Likewise.
2298         * sese.c: Likewise.
2299         * targhooks.c: Likewise.
2300         * tracer.c: Likewise.
2301         * trans-mem.c: Likewise.
2302         * tree-call-cdce.c: Likewise.
2303         * tree-cfg.c: Likewise.
2304         * tree-cfgcleanup.c: Likewise.
2305         * tree-chrec.c: Likewise.
2306         * tree-complex.c: Likewise.
2307         * tree-data-ref.c: Likewise.
2308         * tree-dfa.c: Likewise.
2309         * tree-eh.c: Likewise.
2310         * tree-emutls.c: Likewise.
2311         * tree-if-conv.c: Likewise.
2312         * tree-inline.c: Likewise.
2313         * tree-into-ssa.c: Likewise.
2314         * tree-loop-distribution.c: Likewise.
2315         * tree-mudflap.c: Likewise.
2316         * tree-nested.c: Likewise.
2317         * tree-nrv.c: Likewise.
2318         * tree-object-size.c: Likewise.
2319         * tree-outof-ssa.c: Likewise.
2320         * tree-parloops.c: Likewise.
2321         * tree-phinodes.c: Likewise.
2322         * tree-predcom.c: Likewise.
2323         * tree-pretty-print.c: Likewise.
2324         * tree-profile.c: Likewise.
2325         * tree-scalar-evolution.c: Likewise.
2326         * tree-sra.c: Likewise.
2327         * tree-ssa-address.c: Likewise.
2328         * tree-ssa-alias.c: Likewise.
2329         * tree-ssa-ccp.c: Likewise.
2330         * tree-ssa-coalesce.c: Likewise.
2331         * tree-ssa-copy.c: Likewise.
2332         * tree-ssa-copyrename.c: Likewise.
2333         * tree-ssa-dce.c: Likewise.
2334         * tree-ssa-dom.c: Likewise.
2335         * tree-ssa-dse.c: Likewise.
2336         * tree-ssa-forwprop.c: Likewise.
2337         * tree-ssa-ifcombine.c: Likewise.
2338         * tree-ssa-live.c: Likewise.
2339         * tree-ssa-loop-ch.c: Likewise.
2340         * tree-ssa-loop-im.c: Likewise.
2341         * tree-ssa-loop-ivcanon.c: Likewise.
2342         * tree-ssa-loop-ivopts.c: Likewise.
2343         * tree-ssa-loop-manip.c: Likewise.
2344         * tree-ssa-loop-niter.c: Likewise.
2345         * tree-ssa-loop-prefetch.c: Likewise.
2346         * tree-ssa-loop-unswitch.c: Likewise.
2347         * tree-ssa-loop.c: Likewise.
2348         * tree-ssa-math-opts.c: Likewise.
2349         * tree-ssa-operands.c: Likewise.
2350         * tree-ssa-phiopt.c: Likewise.
2351         * tree-ssa-phiprop.c: Likewise.
2352         * tree-ssa-pre.c: Likewise.
2353         * tree-ssa-propagate.c: Likewise.
2354         * tree-ssa-reassoc.c: Likewise.
2355         * tree-ssa-sccvn.c: Likewise.
2356         * tree-ssa-sink.c: Likewise.
2357         * tree-ssa-strlen.c: Likewise.
2358         * tree-ssa-structalias.c: Likewise.
2359         * tree-ssa-tail-merge.c: Likewise.
2360         * tree-ssa-ter.c: Likewise.
2361         * tree-ssa-threadedge.c: Likewise.
2362         * tree-ssa-threadupdate.c: Likewise.
2363         * tree-ssa-uncprop.c: Likewise.
2364         * tree-ssa-uninit.c: Likewise.
2365         * tree-ssa.c: Likewise.
2366         * tree-ssanames.c: Likewise.
2367         * tree-stdarg.c: Likewise.
2368         * tree-streamer-in.c: Likewise.
2369         * tree-switch-conversion.c: Likewise.
2370         * tree-tailcall.c: Likewise.
2371         * tree-vect-data-refs.c: Likewise.
2372         * tree-vect-generic.c: Likewise.
2373         * tree-vect-loop-manip.c: Likewise.
2374         * tree-vect-loop.c: Likewise.
2375         * tree-vect-patterns.c: Likewise.
2376         * tree-vect-slp.c: Likewise.
2377         * tree-vect-stmts.c: Likewise.
2378         * tree-vectorizer.c: Likewise.
2379         * tree-vrp.c: Likewise.
2380         * tree.c: Likewise.
2381         * tsan.c: Likewise.
2382         * value-prof.c: Likewise.
2383         * var-tracking.c: Likewise.
2384         * varpool.c: Likewise.
2385         * vtable-verify.c: Likewise.
2387 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2389         * config/tilegx/tilegx.c: Include "tree.h".
2391 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2393         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
2394         the info, not after it.
2395         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
2396         dump_ssaname_info call.
2397         (pp_gimple_stmt_1): Adjust caller.
2398         (dump_phi_nodes): Likewise.  Don't print "# " here.
2400 2013-10-22  Jan Hubicka  <jh@suse.cz>
2402         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New
2403         tuning flag.
2404         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
2405         * i386.c (expand_small_movmem_or_setmem): New function.
2406         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New
2407         function.
2408         (alg_usable_p): Add support for value ranges; cleanup.
2409         (ix86_expand_set_or_movmem): Add support for misaligned moves.
2411 2013-10-22  Sterling Augustine  <saugustine@google.com>
2413         * doc/invoke.texi: Document -ggnu-pubnames.
2414         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
2415         logic.
2416         * dwarf2out.c: Include gdb/gdb-index.h.
2417         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
2418         debug_generate_pub_sections.
2419         (is_java, output_pubtables, output_pubname): New functions.
2420         (include_pubname_in_output): Handle debug_generate_pub_sections at
2421         level 2.
2422         (size_of_pubnames): Use new local space_for_flags based on
2423         debug_generate_pub_sections.
2424         (output_pubnames): Unify pubnames and pubtypes output logic.
2425         Genericize comments.  Call output_pubname.
2426         (dwarf2out_finish): Move logic to output_pubnames and call it.
2428 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
2430         PR target/58779
2431         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
2432         Remove CCCmode handling.
2433         <case LTU>: Return 'c' suffix for CCCmode.
2434         <case GEU>: Return 'nc' suffix for CCCmode.
2435         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
2436         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
2437         (*sub<mode>3_cc_overflow): Ditto.
2438         (*subsi3_zext_cc_overflow): Ditto.
2440 2013-10-22  Steve Ellcey  <sellcey@mips.com>
2442         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
2443         (AND (NOT OP1) (NOT OP2)).
2445 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
2447         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2448         meaning of merge-high and merge-low masks for little endian; avoid
2449         use of vector-pack masks for little endian for mismatched modes.
2451 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2453         * config/tilepro/tilepro.c: Include "tree.h".
2455 2013-10-22  Andreas Schwab  <schwab@suse.de>
2457         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
2458         with PRE_DEC.
2460 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
2462         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
2463         [Ed Smith-Rowland]): New entries.
2464         ([Stephen M. Webb]): Update.
2466 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
2468         * tree-ssa-ter.h: Remove duplicate copy of file contents.
2470 2013-10-21  Marek Polacek  <polacek@redhat.com>
2472         PR middle-end/58809
2473         * fold-const.c (fold_range_test): Return 0 if the type is not
2474         an integral type.
2476 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
2478         * system.h: Move hwint.h include further down.
2479         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
2480         gcc_checking_asserts.
2481         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
2483 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2485         Fix volatile issues in optimize_bit_field_compare.
2486         * fold-const.c (optimize_bit_field_compare): Bail out if
2487         lvolatilep or rvolatilep.
2489 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2491         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
2492         and get_inner_reference returning different pmode for non-volatile
2493         bit-field members dependent on flag_strict_volatile_bitfields.
2494         * stor-layout.c (layout_decl): Remove special handling of
2495         flag_strict_volatile_bitfields.
2496         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
2497         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
2499 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
2501         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
2502         calculations.
2504 2013-10-21  Jeff Law  <law@redhat.com>
2506         * tree-ssa-threadedge.c (thread_through_normal_block): New
2507         argument VISITED.  Remove VISISTED as a local variable.  When we
2508         have a threadable jump, verify the destination of the jump has not
2509         been visised.
2510         (thread_across_edge): Allocate VISITED bitmap once at function
2511         scope and use it throughout.  Make sure to set appropriate bits in
2512         VISITED for E (start of jump thread path).
2513         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
2514         through a joiner if any edge on the path has a recorded jump thread.
2516 2013-10-21  Ian Lance Taylor  <iant@google.com>
2518         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
2519         don't imply that attributes can solve all problems.
2520         (Directory Options): Fix typo.
2522 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2524         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
2525         extend and extend_add.
2527 2013-10-21  Richard Biener  <rguenther@suse.de>
2529         PR tree-optimization/58794
2530         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
2531         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
2533 2013-10-21  Richard Biener  <rguenther@suse.de>
2535         PR middle-end/58742
2536         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
2537         to (T) X for sign-changing conversions (or no conversion).
2539 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
2541         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
2543 2013-10-20  Jan Hubicka  <jh@suse.cz>
2545         * config/i386/i386-tune.def: Add comment; organize into categories
2547 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
2549         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
2550         argument.  Update function comment.
2551         (expand_set_or_movmem_via_rep): New function combining
2552         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
2553         (expand_movmem_via_rep_mov): Remove.
2554         expand_setmem_via_rep_stos): Remove.
2555         (expand_movmem_epilogue): Update calls correspondingly.
2556         (expand_setmem_epilogue_via_loop): Likewise.
2557         (emit_memset): New.
2558         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
2559         (expand_set_or_movmem_prologue): New function combining
2560         expand_movmem_prologue and expand_setmem_prologue.
2561         (expand_movmem_prologue): Remove.
2562         (expand_setmem_prologue): Remove.
2563         (expand_set_or_movmem_constant_prologue): New function combining
2564         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
2565         (expand_constant_movmem_prologue): Remove.
2566         (expand_constant_setmem_prologue): Remove.
2567         (promote_duplicated_reg): Allow vector-const0 value.
2568         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
2569         and ix86_expand_setmem.
2570         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
2571         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
2573 2013-10-21  Diego Novillo  <dnovillo@google.com>
2575         * asan.c: Include tree.h
2576         * bb-reorder.c: Likewise.
2577         * cfgcleanup.c: Likewise.
2578         * cfgloopmanip.c: Likewise.
2579         * data-streamer-in.c: Likewise.
2580         * data-streamer-out.c: Likewise.
2581         * data-streamer.c: Likewise.
2582         * dwarf2cfi.c: Likewise.
2583         * graphite-blocking.c: Likewise.
2584         * graphite-clast-to-gimple.c: Likewise.
2585         * graphite-dependences.c: Likewise.
2586         * graphite-interchange.c: Likewise.
2587         * graphite-optimize-isl.c: Likewise.
2588         * graphite-poly.c: Likewise.
2589         * graphite-scop-detection.c: Likewise.
2590         * graphite-sese-to-poly.c: Likewise.
2591         * graphite.c: Likewise.
2592         * ipa-devirt.c: Likewise.
2593         * ipa-profile.c: Likewise.
2594         * ipa.c: Likewise.
2595         * ira.c: Likewise.
2596         * loop-init.c: Likewise.
2597         * loop-unroll.c: Likewise.
2598         * lower-subreg.c: Likewise.
2599         * lto/lto-object.c: Likewise.
2600         * recog.c: Likewise.
2601         * reginfo.c: Likewise.
2602         * tree-loop-distribution.c: Likewise.
2603         * tree-parloops.c: Likewise.
2604         * tree-ssa-strlen.c: Likewise.
2605         * tree-streamer.c: Likewise.
2606         * value-prof.c: Likewise.
2607         * target-globals.c: Likewise.
2608         * expr.h: Include tree-core.h instead of tree.h.
2609         * gimple.h: Likewise.
2610         * ipa-prop.h: Likewise.
2611         * ipa-utils.h: Likewise.
2612         * lto-streamer.h: Likewise.
2613         * streamer-hooks.h: Likewise.
2614         * ipa-reference.h: Include cgraph.h instead of tree.h.
2615         * cgraph.h: Include basic-block.h instead of tree.h.
2616         * tree-streamer.h: Do not include tree.h.
2617         * genattrtab.c (write_header): Generate inclusion of
2618         tree.h.
2619         * genautomata.c (main): Likewise.
2620         * genemit.c: Likewise.
2621         * genopinit.c: Likewise.
2622         * genoutput.c (output_prologue): Likewise.
2623         * genpeep.c: Likewise.
2625 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2627         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
2628         little endian.
2629         (vec_unpacku_hi_v8hi): Likewise.
2630         (vec_unpacku_lo_v16qi): Likewise.
2631         (vec_unpacku_lo_v8hi): Likewise.
2633 2013-10-20  Jan Hubicka  <jh@suse.cz>
2635         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
2636         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
2637         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
2639 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
2641         * config/mips/mips.h (ISA_HAS_WSBH): Define.
2642         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
2643         constants.
2644         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
2646 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
2648         PR target/58603
2649         * system.h: Undef m_slot.
2651 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2653         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
2654         all elements for both endian flavors.
2656 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2658         PR target/58792
2659         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
2660         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
2661         and SI_REG for 64bit SYSV ABI targets.
2663 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2665         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
2666         to multi_reg_return.  Clarify that we are skipping USEs of multiple
2667         return registers.  Use bool type where appropriate.
2669 2013-10-18  Jan Hubicka  <jh@suse.cz>
2671         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
2672         for cold functions.
2673         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
2674         (X86_TUNE_PUSH_MEMORY): Likewise.
2675         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
2676         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
2677         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
2678         New.
2679         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
2680         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
2681         Remove.
2682         (ix86_option_override_internal): Update to use tune features instead
2683         of variables.
2685 2013-10-18  Cong Hou  <congh@google.com>
2687         PR tree-optimization/58508
2688         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
2689         statement that contains data refs with zero-step.
2691 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2693         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
2694         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
2695         * gimple-low.c (gimple_check_call_arg,
2696         gimple_check_call_matching_types): Move to cgraph.c.
2697         * gimple-low.h: Remove prototype.
2698         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
2699         Relocate from gimple-low.c.
2700         * cgraph.h: Add prototype.  Don't include basic-block.h.
2701         * gimplify.c: Add gimple-low to include list.
2702         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
2703         * tree-eh.c: Add gimple-low to include list.
2704         * tree-nested.c: Likewise.
2705         * cfgexpand.c: Add tree-ssa-address.h to include list.
2706         * expr.c: Likewise.
2707         * gimple-fold.c: Likewise.
2708         * gimple-ssa-strength-reduction.c: Likewise.
2709         * trans-mem.c: Likewise.
2710         * tree-mudflap.c: Likewise.
2711         * tree-ssa-loop-ivopts.c: Likewise.
2712         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
2713         (degenerate_phi_result): Move to tree-phinodes.c.
2714         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
2715         * tree-ssa-threadedge.c: Likewise.
2716         * tree-vrp.c: Likewise.
2717         * tree-phinodes.c (degenerate_phi_result): Relocate here.
2718         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
2719         * tree-phinodes.h (degenerate_phi_result): Add prototype.
2720         * tree-ssa-copy.c: Include tree-ssa-dom.h.
2721         * tree-ssa-forwprop.c: Likewise.
2722         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
2723         pass_data_cleanup_cfg_post_optimizing,
2724         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
2725         * tree-optimize.c: Delete File.
2726         * graphite.c: Include tree-cfgcleanup.h.
2727         * passes.c: Likewise.
2728         * tree-cfg.c: Likewise.
2729         * tree-profile.c: Likewise.
2730         * tree-ssa-dse.c: Likewise.
2731         * tree-ssa-loop-ivcanon.c: Likewise.
2732         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
2733         * tree-outof-ssa.c: Include sbitmap.h.
2734         * tree-ssa-live.c: Likewise.
2735         * tree-ssa-propagate.c: Likewise.
2736         * tree-ssa-structalias.c: Likewise.
2737         * tree-stdarg.c: Likewise.
2738         * Makefile.in (OBJS): Delete tree-optimize.o.
2739         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
2740         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
2741         * varasm.c: Include basic-block.h.
2742         * cfgloop.h: Include function.h instead of basic-block.h
2743         (bb_loop_depth): Move to cfgloop.c.
2744         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
2746 2013-10-18  Teresa Johnson  <tejohnson@google.com>
2748         * predict.c (probably_never_executed): Compare frequency-based
2749         count to number of training runs.
2750         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
2752 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2754         * config/arm/arm.c (cortexa9_extra_costs): New table.
2755         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
2757 2013-10-18  Jeff Law  <law@redhat.com>
2759         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
2761         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
2762         (register_jump_thread): Add "registered_jump_thread" debug
2763         counter support.
2764         * dbgcnt.def (registered_jump_thread): New debug counter.
2766 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2768         * config/rs6000/rs6000.c: Include cgraph.h.
2770 2013-10-18  Teresa Johnson  <tejohnson@google.com>
2772         * tree-ssa-tail-merge.c (replace_block_by): Update edge
2773         weights during merging.
2775 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2777         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
2778         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
2779         * tree-ssa-operands.h: Remove prototype.
2780         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
2781         * gimple.c (virtual_operand_p): Relocate from gimple.c.
2782         * gimple.h: Add prototype.
2783         * gimple-ssa.h: Include tree-ssa-operands.h.
2784         * tree-dump.c: Add tree-cfg.h to include list.
2785         * tree-ssa-alias.c: Add ipa-reference.h to include list.
2786         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
2787         * config/i386/i386.c: Don't include tree-flow.h.
2788         * config/rs6000/rs6000.c: Likewise.
2790 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2792         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
2794 2013-10-18  Richard Biener  <rguenther@suse.de>
2796         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
2797         or TYPE_UNSIGNED of integral types.
2798         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
2799         NULL_TREE for zero-precision integral types.
2801 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
2803         * config/aarch64/arm_neon.h
2804         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
2806 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
2808         PR target/57756
2809         * opth-gen.awk: Define target_flags_explicit.
2811 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
2813         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
2814         fields to the reg_addr array that describes the valid addressing
2815         mode for any register, general purpose registers, floating point
2816         registers, and Altivec registers.
2817         (FIRST_RELOAD_REG_CLASS): Likewise.
2818         (LAST_RELOAD_REG_CLASS): Likewise.
2819         (struct reload_reg_map_type): Likewise.
2820         (reload_reg_map_type): Likewise.
2821         (RELOAD_REG_VALID): Likewise.
2822         (RELOAD_REG_MULTIPLE): Likewise.
2823         (RELOAD_REG_INDEXED): Likewise.
2824         (RELOAD_REG_OFFSET): Likewise.
2825         (RELOAD_REG_PRE_INCDEC): Likewise.
2826         (RELOAD_REG_PRE_MODIFY): Likewise.
2827         (reg_addr): Likewise.
2828         (mode_supports_pre_incdec_p): New helper functions to say whether
2829         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
2830         (mode_supports_pre_modify_p): Likewise.
2831         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
2832         print the valid address mode bits for each mode.
2833         (rs6000_debug_print_mode): Likewise.
2834         (rs6000_debug_reg_global): Likewise.
2835         (rs6000_setup_reg_addr_masks): New function to set up the address
2836         mask bits for each type.
2837         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
2838         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
2839         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
2840         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
2841         PRE_MODIFY are supported.
2842         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
2843         registers, instead of {src,dest}_av_p.
2844         (rs6000_print_options_internal): Tweak the debug output slightly.
2846 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
2848         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
2849         isa attribute.
2851 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
2853         * tree-flow.h (struct omp_region): Move to omp-low.c.
2854         Remove omp_ prototypes and variables.
2855         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
2856         (copy_var_decl): Relocate prototype from tree-flow.h.
2857         * gimple.c (copy_var_decl): Relocate from omp-low.c.
2858         * tree.h: Move prototype to omp-low.h.
2859         * omp-low.h: New File.  Relocate prototypes here.
2860         * omp-low.c (struct omp_region): Make local here.
2861         (root_omp_region): Make static.
2862         (copy_var_decl) Move to gimple.c.
2863         (new_omp_region): Make static.
2864         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
2865         * tree-cfg.c: Include omp-low.h.
2866         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
2867         * gimplify.c: Include omp-low.h.
2868         * tree-parloops.c: Likewise.
2870 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
2872         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
2873         operands are in memory, prefer to force non-matched operand 1 to
2874         the register.
2876 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
2878         PR target/58673
2879         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
2880         restrict TImode addresses to single indirect registers if both
2881         -mquad-memory and -mvsx-timode are used.
2882         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
2883         we should emit load/store quad.  Remove using %y for quad memory
2884         addresses.
2886         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
2887         constraints to allow load/store quad on machines where TImode is
2888         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
2889         for TImode to load integer constants.
2891 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2893         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
2895 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2897         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
2898         handling of STACK_REG.
2900 2013-10-17  Richard Biener  <rguenther@suse.de>
2902         PR tree-optimization/58143
2903         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
2904         New function.
2905         (rewrite_to_defined_overflow): Likewise.
2906         (move_computations_dom_walker::before_dom): Rewrite stmts
2907         with undefined signed overflow that are not always executed
2908         into unsigned arithmetic.
2910 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
2912         PR target/57756
2913         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
2914         explicit isa flag to be an options variable, instead of using
2915         global_options_set.  Remove define from rs6000.h.
2916         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
2918         * config/rs6000/rs6000.c (rs6000_option_override_internal):
2919         Initialize rs6000_isa_flags_explicit.
2920         (rs6000_function_specific_save): Add gcc_options* parameter, so
2921         that the powerpc builds after the 2013-10-15 changes.
2922         (rs6000_function_specific_restore): Likewise.
2924 2013-10-16  DJ Delorie  <dj@redhat.com>
2926         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
2927         op is a REG before checking REGNO.
2928         (rl78_alloc_physical_registers): Verify pattern is a SET before
2929         checking SET_SRC.
2931 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2933         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
2934         endianness.
2935         (vec_unpacks_lo_v4sf): Likewise.
2936         (vec_unpacks_float_hi_v4si): Likewise.
2937         (vec_unpacks_float_lo_v4si): Likewise.
2938         (vec_unpacku_float_hi_v4si): Likewise.
2939         (vec_unpacku_float_lo_v4si): Likewise.
2941 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2943         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
2944         (vsx_concat_v2sf): Likewise.
2946 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
2948         * config/aarch64/aarch64.md
2949         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
2951 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
2953         PR tree-optimization/58697
2954         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
2955         estimated_loop_iterations_int.
2956         (max_stmt_executions_int): Call get_max_loop_iterations_int.
2957         (get_max_loop_iterations_int): New.  HWINT version of
2958         get_max_loop_iterations.
2959         * cfgloop.h: Add prototypes.
2960         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
2961         * loop-unroll.c (decide_peel_once_rolling): Call
2962         get_estimated_loop_iterations_int.
2963         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
2964         * tree-ssa-loop-niter.h: Tweak prototypes.
2966 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
2968         * gengtype-parse.c (struct_field_seq): Ignore access-control
2969         keywords ("public:" etc).
2971 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2973         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
2974         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
2976 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
2978         * config/arm/arm.opt (mlra): New option.
2979         * config/arm/arm.c (arm_lra_p): New function.
2980         (TARGET_LRA_P): Define.
2982 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
2984         * tree-core.h (tree_code_name): Remove.
2985         * tree.h (get_tree_code_name): New prototype.
2986         * tree.c (tree_code_name): Make static.
2987         (get_tree_code_name): New function.
2988         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
2989         tree_class_check_failed, tree_range_check_failed,
2990         tree_not_class_check_failed, omp_clause_check_failed,
2991         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
2992         wrapper get_tree_code_name instead of calling tree_code_name directly.
2993         * tree-vrp.c (dump_asserts_for): Likewise.
2994         * tree-dump.c (dequeue_and_dump): Likewise.
2995         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
2996         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
2997         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
2998         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
2999         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
3000         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
3001         dump_gimple_omp_for): Likewise.
3002         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3003         * tree-ssa-pre.c (print_pre_expr): Likewise.
3004         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3005         * print-tree.c (print_node_brief, print_node): Likewise.
3006         * gimple.c (gimple_check_failed): Likewise.
3007         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
3008         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
3009         * config/mep/mep.c (mep_validate_vliw): Likewise.
3010         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
3011         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
3013 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
3015         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
3016         for AMD bdver3.
3018 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
3020         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
3021         (MULTILIB_MATCHES): Add multilib for -march=v8.
3023 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
3025         PR target/57756
3026         * optc-save-gen.awk: Add extra parameter to the save and restore
3027         target calls.
3028         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
3029         * tree.c (build_optimization_node): New parameter.  Add extra parameter
3030         to call to cl_optimization_save.
3031         (build_target_option_node): New parameter. Add extra parameter
3032         to call to cl_target_option_save.
3033         * tree.h (build_optimization_node): New parameter.
3034         (build_target_option_node): New parameter.
3035         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
3036         build_optimization_node and build_target_option_node.
3037         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
3038         (handle_pragma_push_options): Ditto.
3039         * toplev.c (process_options): Ditto.
3040         * opts.c (init_options_struct): Check for opts_set non-null.
3041         * target.def (target_option.save): New parameter.
3042         (target_option.restore): New parameter.
3043         * tm.texi: Generate.
3044         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3045         (ix86_pragma_target_parse): Ditto.
3046         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
3047         parameters.
3048         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
3049         to  build_optimization_node and build_target_option_node.
3050         (rs6000_valid_attribute_p): Ditto.
3051         (rs6000_pragma_target_parse): Ditto.
3052         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
3053         data.
3054         * config/i386/i386.h:
3055         TARGET_64BIT_P: New Macro
3056         TARGET_MMX_P: New Macro.
3057         TARGET_3DNOW_P: New Macro.
3058         TARGET_3DNOW_A_P: New Macro.
3059         TARGET_SSE_P: New Macro.
3060         TARGET_SSE2_P: New Macro.
3061         TARGET_SSE3_P: New Macro.
3062         TARGET_SSSE3_P: New Macro.
3063         TARGET_SSE4_1_P: New Macro.
3064         TARGET_SSE4_2_P: New Macro.
3065         TARGET_AVX_P: New Macro.
3066         TARGET_AVX2_P: New Macro.
3067         TARGET_AVX512F_P: New Macro.
3068         TARGET_AVX512PF_P: New Macro.
3069         TARGET_AVX512ER_P: New Macro.
3070         TARGET_AVX512CD_P: New Macro.
3071         TARGET_FMA_P: New Macro.
3072         TARGET_SSE4A_P: New Macro.
3073         TARGET_FMA4_P: New Macro.
3074         TARGET_XOP_P: New Macro.
3075         TARGET_LWP_P: New Macro.
3076         TARGET_ABM_P: New Macro.
3077         TARGET_BMI_P: New Macro.
3078         TARGET_BMI2_P: New Macro.
3079         TARGET_LZCNT_P: New Macro.
3080         TARGET_TBM_P: New Macro.
3081         TARGET_POPCNT_P: New Macro.
3082         TARGET_SAHF_P: New Macro.
3083         TARGET_MOVBE_P: New Macro.
3084         TARGET_CRC32_P: New Macro.
3085         TARGET_AES_P: New Macro.
3086         TARGET_PCLMUL_P: New Macro.
3087         TARGET_CMPXCHG16B_P: New Macro.
3088         TARGET_FSGSBASE_P: New Macro.
3089         TARGET_RDRND_P: New Macro.
3090         TARGET_F16C_P: New Macro.
3091         TARGET_RTM_P: New Macro.
3092         TARGET_HLE_P: New Macro.
3093         TARGET_RDSEED_P: New Macro.
3094         TARGET_PRFCHW_P: New Macro.
3095         TARGET_ADX_P: New Macro.
3096         TARGET_FXSR_P: New Macro.
3097         TARGET_XSAVE_P: New Macro.
3098         TARGET_XSAVEOPT_P: New Macro.
3099         TARGET_LP64_P: New Macro.
3100         TARGET_X32_P: New Macro.
3101         TARGET_FPMATH_DEFAULT_P: New Macro.
3102         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
3103         * config/i386/i386.c (ix86_option_override_internal): New parameters.
3104         opts and opts_set.
3105         Change ix86_tune_string to access opts->x_ix86_tune_string.
3106         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
3107         Change ix86_arch_string to access opts->x_ix86_arch_string.
3108         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
3109         Change ix86_pmode to access opts->x_ix86_pmode.
3110         Change ix86_abi to access opts->x_ix86_abi.
3111         Change ix86_cmodel to access opts->x_ix86_cmodel.
3112         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
3113         Change ix86_isa_flags_explicit to access
3114         opts->x_ix86_isa_flags_explicit.
3115         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
3116         Change ix86_regparm to access opts->x_ix86_regparm.
3117         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
3118         Change ix86_preferred_stack_boundary_arg to access
3119         opts->x_ix86_preferred_stack_boundary_arg.
3120         Change ix86_force_align_arg_pointer to access
3121         opts->x_ix86_force_align_arg_pointer.
3122         Change ix86_incoming_stack_boundar_arg to access
3123         opts->x_ix86_incoming_stack_boundar_arg.
3124         Change ix86_fpmath to access opts->x_ix86_fpmath.
3125         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
3126         Change ix86_recip_name to access opts->x_ix86_recip_name.
3127         Change ix86_stack_protector_guard to access
3128         opts->x_ix86_stack_protector_guard.
3129         Change ix86_tune_memcpy_strategy to access
3130         opts->x_ix86_tune_memcpy_strategy.
3131         Change ix86_tune_memset_strategy to access
3132         opts->x_ix86_tune_memset_strategy.
3133         Change global_options to access opts.
3134         Change global_options_set to access opts_set.
3135         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
3136         Change TARGET_MMX to TARGET_MMX_P (opts->...).
3137         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
3138         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
3139         Change TARGET_SSE to TARGET_SSE_P (opts->...).
3140         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
3141         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
3142         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
3143         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
3144         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
3145         Change TARGET_AVX to TARGET_AVX_P (opts->...).
3146         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
3147         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
3148         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
3149         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
3150         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
3151         Change TARGET_FMA to TARGET_FMA_P (opts->...).
3152         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
3153         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
3154         Change TARGET_XOP to TARGET_XOP_P (opts->...).
3155         Change TARGET_LWP to TARGET_LWP_P (opts->...).
3156         Change TARGET_ABM to TARGET_ABM_P (opts->...).
3157         Change TARGET_BMI to TARGET_BMI_P (opts->...).
3158         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
3159         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
3160         Change TARGET_TBM to TARGET_TBM_P (opts->...).
3161         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
3162         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
3163         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
3164         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
3165         Change TARGET_AES to TARGET_AES_P (opts->...).
3166         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
3167         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
3168         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
3169         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
3170         Change TARGET_F16C to TARGET_F16C_P (opts->...).
3171         Change TARGET_RTM to TARGET_RTM_P (opts->...).
3172         Change TARGET_HLE to TARGET_HLE_P (opts->...).
3173         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
3174         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
3175         Change TARGET_ADX to TARGET_ADX_P (opts->...).
3176         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
3177         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
3178         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
3179         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
3180         Change TARGET_X32 to TARGET_X32_P (opts->...).
3181         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
3182         Change TARGET_FLOAT_RETURNS_IN_80387 to
3183         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
3184         (ix86_function_specific_save): New parameter. Use opts-> fields
3185         to replace global fields.
3186         (ix86_function_specific_restore): Ditto.
3187         (ix86_valid_target_attribute_inner_p): New parameters.
3188         Fix recursive call.
3189         Fix call to ix86_handle_option and set_option.
3190         (ix86_valid_target_attribute_tree): New parameters.
3191         Change global_options to access opts.
3192         Change global_options_set to access opts_set.
3193         Fix call to ix86_valid_target_attribute_inner_p.
3194         Change ix86_tune_string to access opts->x_ix86_tune_string.
3195         Change ix86_arch_string to access opts->x_ix86_arch_string.
3196         Change ix86_fpmath to access opts->x_ix86_fpmath
3197         Fix call to ix86_option_override_internal.
3198         Fix call to ix86_add_new_builtins.
3199         Fix calls to build_optimization_node and build_target_option_node.
3200         (ix86_valid_target_attribute_p): Remove access to global_options.
3201         Use new gcc_options structure func_options.
3202         Fix call to ix86_valid_target_attribute_tree.
3203         Fix call to  build_optimization_node.
3204         (get_builtin_code_for_version): Fix call to
3205         ix86_valid_target_attribute_tree.
3207 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
3209         * Makefile.in (PICFLAG): New.
3210         (enable_host_shared): New.
3211         (INTERNAL_CFLAGS): Use PICFLAG.
3212         (LIBIBERTY): Use pic build of libiberty.a if configured with
3213         --enable-host-shared.
3214         * configure.ac: Add --enable-host-shared, setting up new
3215         PICFLAG variable.
3216         * configure: Regenerate.
3217         * doc/install.texi (--enable-shared): Add note contrasting it with ...
3218         (--enable-host-shared): New option.
3220 2013-10-15  Richard Biener  <rguenther@suse.de>
3222         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
3223         for built-in functions.
3225 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3227         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
3228         (optimize_range_tests_1): New function,
3229         extracted from optimize_range_tests.
3230         (optimize_range_tests_xor): Similarly.
3231         (optimize_range_tests_diff): New function.
3232         (optimize_range_tests): Use optimize_range_tests_1.
3234 2013-10-15  Cong Hou  <congh@google.com>
3236         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
3237         requirement of the reduction pattern so that one operand of the
3238         reduction operation can come from outside of the loop.
3240 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3242         * config/arm/neon-schedgen.ml: Remove.
3243         * config/arm/cortex-a9-neon.md: Remove comment regarding
3244         neon-schedgen.ml.
3246 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3248         * config/arm/types: Remove old neon types.
3250 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3252         * config/arm/cortex-a7.md
3253         (cortex_a7_neon_type): New.
3254         (cortex_a7_neon_mul): Update for new types.
3255         (cortex_a7_neon_mla): Likewise.
3256         (cortex_a7_neon): Likewise.
3258 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3260         * config/arm/cortex-a15-neon.md
3261         (cortex_a15_neon_type): New,
3263         (cortex_a15_neon_int_1): Remove.
3264         (cortex_a15_neon_int_2): Likewise.
3265         (cortex_a15_neon_int_3): Likewise.
3266         (cortex_a15_neon_int_4): Likewise.
3267         (cortex_a15_neon_int_5): Likewise.
3268         (cortex_a15_neon_vqneg_vqabs): Likewise.
3269         (cortex_a15_neon_vmov): Likewise.
3270         (cortex_a15_neon_vaba): Likewise.
3271         (cortex_a15_neon_vaba_qqq): Likewise.
3272         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3273         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3274         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3275         Likewise.
3276         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3277         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3278         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
3279         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3280         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3281         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3282         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3283         (cortex_a15_neon_shift_1): Likewise.
3284         (cortex_a15_neon_shift_2): Likewise.
3285         (cortex_a15_neon_shift_3): Likewise.
3286         (cortex_a15_neon_vshl_ddd): Likewise.
3287         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3288         (cortex_a15_neon_vsra_vrsra): Likewise.
3289         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3290         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3291         (cortex_a15_neon_bp_3cycle): Likewise.
3292         (cortex_a15_neon_ldm_2): Likewise.
3293         (cortex_a15_neon_stm_2): Likewise.
3294         (cortex_a15_neon_mcr): Likewise.
3295         (cortex_a15_neon_mrc): Likewise.
3296         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3297         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3298         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3299         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3300         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3301         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3302         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3303         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3304         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3305         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3306         (cortex_a15_neon_bp_simple): Likewise.
3307         (cortex_a15_neon_bp_2cycle): Likewise.
3308         (cortex_a15_neon_bp_3cycle): Likewise.
3309         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3310         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3311         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3312         (cortex_a15_neon_vld2_4_regs): Likewise.
3313         (cortex_a15_neon_vld3_vld4): Likewise.
3314         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3315         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3316         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
3317         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
3318         (cortex_a15_neon_vst3_vst4): Rename to...
3319         (cortex_a15_neon_vst4): This, update for new attributes.
3320         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
3321         (cortex_a15_neon_vld3_vld4_lane): Likewise.
3322         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3323         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3324         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3325         (cortex_a15_neon_ldm_2): Likewise.
3326         (cortex_a15_neon_stm_2): Likewise.
3327         (cortex_a15_neon_mcr): Likewise.
3328         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3329         (cortex_a15_neon_mrc): Likewise.
3330         (cortex_a15_neon_mrrc): Likewise.
3332         (cortex_a15_neon_abd): New.
3333         (cortex_a15_neon_abd_q): Likewise.
3334         (cortex_a15_neon_aba): Likewise.
3335         (cortex_a15_neon_aba_q): Likewise.
3336         (cortex_a15_neon_acc): Likewise.
3337         (cortex_a15_neon_acc_q): Likewise.
3338         (cortex_a15_neon_arith_basic): Likewise.
3339         (cortex_a15_neon_arith_complex): Likewise.
3340         (cortex_a15_neon_multiply): Likewise.
3341         (cortex_a15_neon_multiply_q): Likewise.
3342         (cortex_a15_neon_mla): Likewise.
3343         (cortex_a15_neon_mla_q): Likewise.
3344         (cortex_a15_neon_sat_mla_long): Likewise.
3345         (cortex_a15_neon_shift_acc): Likewise.
3346         (cortex_a15_neon_shift_imm_basic): Likewise.
3347         (cortex_a15_neon_shift_imm_complex): Likewise.
3348         (cortex_a15_neon_shift_reg_basic): Likewise.
3349         (cortex_a15_neon_shift_reg_basic_q): Likewise.
3350         (cortex_a15_neon_shift_reg_complex): Likewise.
3351         (cortex_a15_neon_shift_reg_complex_q): Likewise.
3352         (cortex_a15_neon_fp_negabs): Likewise
3353         (cortex_a15_neon_fp_arith): Likewise
3354         (cortex_a15_neon_fp_arith_q): Likewise
3355         (cortex_a15_neon_fp_cvt_int): Likewise
3356         (cortex_a15_neon_fp_cvt_int_q): Likewise
3357         (cortex_a15_neon_fp_cvt_16): Likewise
3358         (cortex_a15_neon_fp_mul): Likewise
3359         (cortex_a15_neon_fp_mul_q): Likewise
3360         (cortex_a15_neon_fp_mla): Likewise
3361         (cortex_a15_neon_fp_mla_q): Likewise
3362         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
3363         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
3364         (cortex_a15_neon_bitops): Likewise.
3365         (cortex_a15_neon_bitops_q): Likewise.
3366         (cortex_a15_neon_from_gp): Likewise.
3367         (cortex_a15_neon_from_gp_q): Likewise.
3368         (cortex_a15_neon_tbl3_tbl4): Likewise.
3369         (cortex_a15_neon_zip_q): Likewise.
3370         (cortex_a15_neon_to_gp): Likewise.
3371         (cortex_a15_neon_load_a): Likewise.
3372         (cortex_a15_neon_load_b): Likewise.
3373         (cortex_a15_neon_load_c): Likewise.
3374         (cortex_a15_neon_load_d): Likewise.
3375         (cortex_a15_neon_load_e): Likewise.
3376         (cortex_a15_neon_load_f): Likewise.
3377         (cortex_a15_neon_store_a): Likewise.
3378         (cortex_a15_neon_store_b): Likewise.
3379         (cortex_a15_neon_store_c): Likewise.
3380         (cortex_a15_neon_store_d): Likewise.
3381         (cortex_a15_neon_store_e): Likewise.
3382         (cortex_a15_neon_store_f): Likewise.
3383         (cortex_a15_neon_store_g): Likewise.
3384         (cortex_a15_neon_store_h): Likewise.
3385         (cortex_a15_vfp_to_from_gp): Likewise.
3387 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3389         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
3391         (cortex_a9_neon_vshl_ddd): Remove.
3392         (cortex_a9_neon_vst3_vst4): Likewise.
3393         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3395         (cortex_a9_neon_bit_ops_q): New.
3397         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
3398         (cortex_a9_neon_int_2): Likewise.
3399         (cortex_a9_neon_int_3): Likewise.
3400         (cortex_a9_neon_int_4): Likewise.
3401         (cortex_a9_neon_int_5): Likewise.
3402         (cortex_a9_neon_vqneg_vqabs): Likewise.
3403         (cortex_a9_neon_vmov): Likewise.
3404         (cortex_a9_neon_vaba): Likewise.
3405         (cortex_a9_neon_vaba_qqq): Likewise.
3406         (cortex_a9_neon_shift_1): Likewise.
3407         (cortex_a9_neon_shift_2): Likewise.
3408         (cortex_a9_neon_shift_3): Likewise.
3409         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3410         (cortex_a9_neon_vsra_vrsra): Likewise.
3411         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3412         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3413         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3414         Likewise.
3415         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3416         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3417         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3418         Likewise.
3419         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3420         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3421         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3422         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3423         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3424         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3425         (cortex_a9_neon_fp_vsum): Likewise.
3426         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3427         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3428         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3429         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3430         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3431         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3432         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3433         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3434         (cortex_a9_neon_bp_simple): Likewise.
3435         (cortex_a9_neon_bp_2cycle): Likewise.
3436         (cortex_a9_neon_bp_3cycle): Likewise.
3437         (cortex_a9_neon_ldr): Likewise.
3438         (cortex_a9_neon_str): Likewise.
3439         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3440         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3441         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3442         (cortex_a9_neon_vld2_4_regs): Likewise.
3443         (cortex_a9_neon_vld3_vld4): Likewise.
3444         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3445         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3446         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3447         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3448         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3449         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3450         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3451         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3452         (cortex_a9_neon_mcr): Likewise.
3453         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3454         (cortex_a9_neon_mrc): Likewise.
3455         (cortex_a9_neon_mrrc): Likewise.
3457 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3459         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
3461         (cortex_a8_neon_vshl_ddd): Remove.
3462         (cortex_a8_neon_vst3_vst4): Likewise.
3463         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3465         (cortex_a8_neon_bit_ops_q): New.
3467         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
3468         (cortex_a8_neon_int_2): Likewise..
3469         (cortex_a8_neon_int_3): Likewise.
3470         (cortex_a8_neon_int_5): Likewise.
3471         (cortex_a8_neon_vqneg_vqabs): Likewise.
3472         (cortex_a8_neon_int_4): Likewise.
3473         (cortex_a8_neon_vaba): Likewise.
3474         (cortex_a8_neon_vaba_qqq): Likewise.
3475         (cortex_a8_neon_shift_1): Likewise.
3476         (cortex_a8_neon_shift_2): Likewise.
3477         (cortex_a8_neon_shift_3): Likewise.
3478         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3479         (cortex_a8_neon_vsra_vrsra): Likewise.
3480         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3481         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3482         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3483         Likewise.
3484         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3485         (cortex_a8_neon_mla_qqq_8_16): Likewise.
3486         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3487         Likewise.
3488         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3489         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3490         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3491         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3492         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3493         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3494         (cortex_a8_neon_fp_vsum): Likewise.
3495         (cortex_a8_neon_fp_vmul_ddd): Likewise.
3496         (cortex_a8_neon_fp_vmul_qqd): Likewise.
3497         (cortex_a8_neon_fp_vmla_ddd): Likewise.
3498         (cortex_a8_neon_fp_vmla_qqq): Likewise.
3499         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3500         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3501         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3502         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3503         (cortex_a8_neon_bp_simple): Likewise.
3504         (cortex_a8_neon_bp_2cycle): Likewise.
3505         (cortex_a8_neon_bp_3cycle): Likewise.
3506         (cortex_a8_neon_ldr): Likewise.
3507         (cortex_a8_neon_str): Likewise.
3508         (cortex_a8_neon_vld1_1_2_regs): Likewise.
3509         (cortex_a8_neon_vld1_3_4_regs): Likewise.
3510         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3511         (cortex_a8_neon_vld2_4_regs): Likewise.
3512         (cortex_a8_neon_vld3_vld4): Likewise.
3513         (cortex_a8_neon_vld1_vld2_lane): Likewise.
3514         (cortex_a8_neon_vld3_vld4_lane): Likewise.
3515         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3516         (cortex_a8_neon_vst1_3_4_regs): Likewise.
3517         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3518         (cortex_a8_neon_vst1_vst2_lane): Likewise.
3519         (cortex_a8_neon_vst3_vst4_lane): Likewise.
3520         (cortex_a8_neon_mcr): Likewise.
3521         (cortex_a8_neon_mcr_2_mcrr): Likewise.
3522         (cortex_a8_neon_mrc): Likewise.
3523         (cortex_a8_neon_mrrc): Likewise.
3525 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3527         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
3528         (fp): New.
3529         * config/aarch64/aarch64-simd.md (neon_type): Remove.
3530         (aarch64_simd_dup<mode>): Add "type" attribute.
3531         (aarch64_dup_lane<mode>): Likewise.
3532         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
3533         (*aarch64_simd_mov<mode>): Likewise.
3534         (aarch64_simd_mov_from_<mode>low): Likewise.
3535         (aarch64_simd_mov_from_<mode>high): Likewise.
3536         (orn<mode>3): Likewise.
3537         (bic<mode>3): Likewise.
3538         (add<mode>3): Likewise.
3539         (sub<mode>3): Likewise.
3540         (mul<mode>3): Likewise.
3541         (*aarch64_mul3_elt<mode>): Likewise.
3542         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3543         (*aarch64_mul3_elt_to_128df): Likewise.
3544         (*aarch64_mul3_elt_to_64v2df): Likewise.
3545         (neg<mode>2): Likewise.
3546         (abs<mode>2): Likewise.
3547         (abd<mode>_3): Likewise.
3548         (aba<mode>_3): Likewise.
3549         (fabd<mode>_3): Likewise.
3550         (*fabd_scalar<mode>3): Likewise.
3551         (and<mode>3): Likewise.
3552         (ior<mode>3): Likewise.
3553         (xor<mode>3): Likewise.
3554         (one_cmpl<mode>2): Likewise.
3555         (aarch64_simd_vec_set<mode>): Likewise.
3556         (aarch64_simd_lshr<mode>): Likewise.
3557         (aarch64_simd_ashr<mode>): Likewise.
3558         (aarch64_simd_imm_shl<mode>): Likewise.
3559         (aarch64_simd_reg_sshl<mode): Likewise.
3560         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
3561         (aarch64_simd_reg_shl<mode>_signed): Likewise.
3562         (aarch64_simd_vec_setv2di): Likewise.
3563         (aarch64_simd_vec_set<mode>): Likewise.
3564         (aarch64_mla<mode>): Likewise.
3565         (*aarch64_mla_elt<mode>): Likewise.
3566         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3567         (aarch64_mls<mode>): Likewise.
3568         (*aarch64_mls_elt<mode>): Likewise.
3569         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3570         (<su><maxmin><mode>3): Likewise.
3571         (move_lo_quad_<mode>): Likewise.
3572         (aarch64_simd_move_hi_quad_<mode>): Likewise.
3573         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
3574         (vec_pack_trunc_<mode>): Likewise.
3575         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
3576         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
3577         (*aarch64_<su>mlal_lo<mode>): Likewise.
3578         (*aarch64_<su>mlal_hi<mode>): Likewise.
3579         (*aarch64_<su>mlsl_lo<mode>): Likewise.
3580         (*aarch64_<su>mlsl_hi<mode>): Likewise.
3581         (*aarch64_<su>mlal<mode>): Likewise.
3582         (*aarch64_<su>mlsl<mode>): Likewise.
3583         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
3584         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
3585         (add<mode>3): Likewise.
3586         (sub<mode>3): Likewise.
3587         (mul<mode>3): Likewise.
3588         (div<mode>3): Likewise.
3589         (neg<mode>2): Likewise.
3590         (abs<mode>2): Likewise.
3591         (fma<mode>4): Likewise.
3592         (*aarch64_fma4_elt<mode>): Likewise.
3593         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3594         (*aarch64_fma4_elt_to_128df): Likewise.
3595         (*aarch64_fma4_elt_to_64v2df): Likewise.
3596         (fnma<mode>4): Likewise.
3597         (*aarch64_fnma4_elt<mode>): Likewise.
3598         (*aarch64_fnma4_elt_<vswap_width_name><mode>
3599         (*aarch64_fnma4_elt_to_128df): Likewise.
3600         (*aarch64_fnma4_elt_to_64v2df): Likewise.
3601         (<frint_pattern><mode>2): Likewise.
3602         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
3603         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
3604         (vec_unpacks_lo_v4sf): Likewise.
3605         (aarch64_float_extend_lo_v2df): Likewise.
3606         (vec_unpacks_hi_v4sf): Likewise.
3607         (aarch64_float_truncate_lo_v2sf): Likewise.
3608         (aarch64_float_truncate_hi_v4sf): Likewise.
3609         (aarch64_vmls<mode>): Likewise.
3610         (<su><maxmin><mode>3): Likewise.
3611         (<maxmin_uns><mode>3): Likewise.
3612         (reduc_<sur>plus_<mode>): Likewise.
3613         (reduc_<sur>plus_v2di): Likewise.
3614         (reduc_<sur>plus_v2si): Likewise.
3615         (reduc_<sur>plus_<mode>): Likewise.
3616         (aarch64_addpv4sf): Likewise.
3617         (clz<mode>2): Likewise.
3618         (reduc_<maxmin_uns>_<mode>): Likewise.
3619         (reduc_<maxmin_uns>_v2di): Likewise.
3620         (reduc_<maxmin_uns>_v2si): Likewise.
3621         (reduc_<maxmin_uns>_<mode>): Likewise.
3622         (reduc_<maxmin_uns>_v4sf): Likewise.
3623         (aarch64_simd_bsl<mode>_internal): Likewise.
3624         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3625         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3626         (aarch64_get_lane<mode>): Likewise.
3627         (*aarch64_combinez<mode>): Likewise.
3628         (aarch64_combine<mode>): Likewise.
3629         (aarch64_simd_combine<mode>): Likewise.
3630         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
3631         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
3632         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
3633         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
3634         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
3635         (aarch64_<sur>h<addsub><mode>): Likewise.
3636         (aarch64_<sur><addsub>hn<mode>): Likewise.
3637         (aarch64_<sur><addsub>hn2<mode>): Likewise.
3638         (aarch64_pmul<mode>): Likewise.
3639         (aarch64_<su_optab><optab><mode>): Likewise.
3640         (aarch64_<sur>qadd<mode>): Likewise.
3641         (aarch64_sqmovun<mode>): Likewise.
3642         (aarch64_<sur>qmovn<mode>): Likewise.
3643         (aarch64_s<optab><mode>): Likewise.
3644         (aarch64_sq<r>dmulh<mode>): Likewise.
3645         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3646         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3647         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3648         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
3649         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3650         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3651         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
3652         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
3653         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3654         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
3655         (aarch64_sqdmull<mode>): Likewise.
3656         (aarch64_sqdmull_lane<mode>_internal): Likewise.
3657         (aarch64_sqdmull_n<mode>): Likewise.
3658         (aarch64_sqdmull2<mode>_internal): Likewise.
3659         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3660         (aarch64_sqdmull2_n<mode>_internal): Likewise.
3661         (aarch64_<sur>shl<mode>): Likewise.
3662         (aarch64_<sur>q<r>shl<mode>
3663         (aarch64_<sur>shll_n<mode>): Likewise.
3664         (aarch64_<sur>shll2_n<mode>): Likewise.
3665         (aarch64_<sur>shr_n<mode>): Likewise.
3666         (aarch64_<sur>sra_n<mode>): Likewise.
3667         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
3668         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
3669         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
3670         (aarch64_cm<optab><mode>): Likewise.
3671         (aarch64_cm<optab>di): Likewise.
3672         (aarch64_cm<optab><mode>): Likewise.
3673         (aarch64_cm<optab>di): Likewise.
3674         (aarch64_cmtst<mode>): Likewise.
3675         (aarch64_cmtstdi): Likewise.
3676         (aarch64_cm<optab><mode>): Likewise.
3677         (*aarch64_fac<optab><mode>): Likewise.
3678         (aarch64_addp<mode>): Likewise.
3679         (aarch64_addpdi): Likewise.
3680         (sqrt<mode>2): Likewise.
3681         (vec_load_lanesoi<mode>): Likewise.
3682         (vec_store_lanesoi<mode>): Likewise.
3683         (vec_load_lanesci<mode>): Likewise.
3684         (vec_store_lanesci<mode>): Likewise.
3685         (vec_load_lanesxi<mode>): Likewise.
3686         (vec_store_lanesxi<mode>): Likewise.
3687         (*aarch64_mov<mode>): Likewise.
3688         (aarch64_ld2<mode>_dreg): Likewise.
3689         (aarch64_ld2<mode>_dreg): Likewise.
3690         (aarch64_ld3<mode>_dreg): Likewise.
3691         (aarch64_ld3<mode>_dreg): Likewise.
3692         (aarch64_ld4<mode>_dreg): Likewise.
3693         (aarch64_ld4<mode>_dreg): Likewise.
3694         (aarch64_tbl1<mode>): Likewise.
3695         (aarch64_tbl2v16qi): Likewise.
3696         (aarch64_combinev16qi): Likewise.
3697         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
3698         (aarch64_st2<mode>_dreg): Likewise.
3699         (aarch64_st2<mode>_dreg): Likewise.
3700         (aarch64_st3<mode>_dreg): Likewise.
3701         (aarch64_st3<mode>_dreg): Likewise.
3702         (aarch64_st4<mode>_dreg): Likewise.
3703         (aarch64_st4<mode>_dreg): Likewise.
3704         (*aarch64_simd_ld1r<mode>): Likewise.
3705         (aarch64_frecpe<mode>): Likewise.
3706         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
3707         (aarch64_frecps<mode>): Likewise.
3709 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3711         * config/arm/iterators.md (V_elem_ch): New.
3712         (q): Likewise.
3713         (VQH_type): Likewise.
3714         * config/arm/arm.md (is_neon_type): New.
3715         (conds): Use is_neon_type.
3716         (anddi3_insn): Update type attribute.
3717         (xordi3_insn): Likewise.
3718         (one_cmpldi2): Likewise.
3719         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
3720         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
3721         (*movmisalign<mode>_neon_store): Likewise.
3722         (*movmisalign<mode>_neon_load): Likewise.
3723         (vec_set<mode>_internal): Likewise.
3724         (vec_set<mode>_internal): Likewise.
3725         (vec_setv2di_internal): Likewise.
3726         (vec_extract<mode>): Likewise.
3727         (vec_extract<mode>): Likewise.
3728         (vec_extractv2di): Likewise.
3729         (*add<mode>3_neon): Likewise.
3730         (adddi3_neon): Likewise.
3731         (*sub<mode>3_neon): Likewise.
3732         (subdi3_neon): Likewise.
3733         (fma<VCVTF:mode>4): Likewise.
3734         (fma<VCVTF:mode>4_intrinsic): Likewise.
3735         (*fmsub<VCVTF:mode>4): Likewise.
3736         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
3737         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
3738         (ior<mode>3): Likewise.
3739         (and<mode>3): Likewise.
3740         (orn<mode>3_neon): Likewise.
3741         (orndi3_neon): Likewise.
3742         (bic<mode>3_neon): Likewise.
3743         (bicdi3_neon): Likewise.
3744         (xor<mode>3): Likewise.
3745         (one_cmpl<mode>2): Likewise.
3746         (abs<mode>2): Likewise.
3747         (neg<mode>2): Likewise.
3748         (negdi2_neon): Likewise.
3749         (*umin<mode>3_neon): Likewise.
3750         (*umax<mode>3_neon): Likewise.
3751         (*smin<mode>3_neon): Likewise.
3752         (*smax<mode>3_neon): Likewise.
3753         (vashl<mode>3): Likewise.
3754         (vashr<mode>3_imm): Likewise.
3755         (vlshr<mode>3_imm): Likewise.
3756         (ashl<mode>3_signed): Likewise.
3757         (ashl<mode>3_unsigned): Likewise.
3758         (neon_load_count): Likewise.
3759         (ashldi3_neon_noclobber): Likewise.
3760         (ashldi3_neon): Likewise.
3761         (signed_shift_di3_neon): Likewise.
3762         (unsigned_shift_di3_neon): Likewise.
3763         (ashrdi3_neon_imm_noclobber): Likewise.
3764         (lshrdi3_neon_imm_noclobber): Likewise.
3765         (<shift>di3_neon): Likewise.
3766         (widen_ssum<mode>3): Likewise.
3767         (widen_usum<mode>3): Likewise.
3768         (quad_halves_<code>v4si): Likewise.
3769         (quad_halves_<code>v4sf): Likewise.
3770         (quad_halves_<code>v8hi): Likewise.
3771         (quad_halves_<code>v16qi): Likewise.
3772         (reduc_splus_v2di): Likewise.
3773         (neon_vpadd_internal<mode>): Likewise.
3774         (neon_vpsmin<mode>): Likewise.
3775         (neon_vpsmax<mode>): Likewise.
3776         (neon_vpumin<mode>): Likewise.
3777         (neon_vpumax<mode>): Likewise.
3778         (*ss_add<mode>_neon): Likewise.
3779         (*us_add<mode>_neon): Likewise.
3780         (*ss_sub<mode>_neon): Likewise.
3781         (*us_sub<mode>_neon): Likewise.
3782         (neon_vadd<mode>_unspec): Likewise.
3783         (neon_vaddl<mode>): Likewise.
3784         (neon_vaddw<mode>): Likewise.
3785         (neon_vhadd<mode>): Likewise.
3786         (neon_vqadd<mode>): Likewise.
3787         (neon_vaddhn<mode>): Likewise.
3788         (neon_vmul<mode>): Likewise.
3789         (neon_vfms<VCVTF:mode>): Likewise.
3790         (neon_vmlal<mode>): Likewise.
3791         (neon_vmls<mode>): Likewise.
3792         (neon_vmlsl<mode>): Likewise.
3793         (neon_vqdmulh<mode>): Likewise.
3794         (neon_vqdmlal<mode>): Likewise.
3795         (neon_vqdmlsl<mode>): Likewise.
3796         (neon_vmull<mode>): Likewise.
3797         (neon_vqdmull<mode>): Likewise.
3798         (neon_vsub<mode>_unspec): Likewise.
3799         (neon_vsubl<mode>): Likewise.
3800         (neon_vsubw<mode>): Likewise.
3801         (neon_vqsub<mode>): Likewise.
3802         (neon_vhsub<mode>): Likewise.
3803         (neon_vsubhn<mode>): Likewise.
3804         (neon_vceq<mode>): Likewise.
3805         (neon_vcge<mode>): Likewise.
3806         (neon_vcgeu<mode>): Likewise.
3807         (neon_vcgt<mode>): Likewise.
3808         (neon_vcgtu<mode>): Likewise.
3809         (neon_vcle<mode>): Likewise.
3810         (neon_vclt<mode>): Likewise.
3811         (neon_vcage<mode>): Likewise.
3812         (neon_vcagt<mode>): Likewise.
3813         (neon_vtst<mode>): Likewise.
3814         (neon_vabd<mode>): Likewise.
3815         (neon_vabdl<mode>): Likewise.
3816         (neon_vaba<mode>): Likewise.
3817         (neon_vabal<mode>): Likewise.
3818         (neon_vmax<mode>): Likewise.
3819         (neon_vmin<mode>): Likewise.
3820         (neon_vpaddl<mode>): Likewise.
3821         (neon_vpadal<mode>): Likewise.
3822         (neon_vpmax<mode>): Likewise.
3823         (neon_vpmin<mode>): Likewise.
3824         (neon_vrecps<mode>): Likewise.
3825         (neon_vrsqrts<mode>): Likewise.
3826         (neon_vqabs<mode>): Likewise.
3827         (neon_vqneg<mode>): Likewise.
3828         (neon_vcls<mode>): Likewise.
3829         (clz<mode>2): Likewise.
3830         (popcount<mode>2): Likewise.
3831         (neon_vrecpe<mode>): Likewise.
3832         (neon_vrsqrte<mode>): Likewise.
3833         (neon_vget_lane<mode>_sext_internal): Likewise.
3834         (neon_vget_lane<mode>_zext_internal): Likewise.
3835         (neon_vdup_n<mode>): Likewise.
3836         (neon_vdup_n<mode>): Likewise.
3837         (neon_vdup_nv2di): Likewise.
3838         (neon_vdup_lane<mode>_interal): Likewise.
3839         (*neon_vswp<mode>): Likewise.
3840         (neon_vcombine<mode>): Likewise.
3841         (float<mode><V_cvtto>2): Likewise.
3842         (floatuns<mode><V_cvtto>2): Likewise.
3843         (fix_trunc<mode><V_cvtto>2): Likewise.
3844         (fixuns_trunc<mode><V_cvtto>2
3845         (neon_vcvt<mode>): Likewise.
3846         (neon_vcvt<mode>): Likewise.
3847         (neon_vcvtv4sfv4hf): Likewise.
3848         (neon_vcvtv4hfv4sf): Likewise.
3849         (neon_vcvt_n<mode>): Likewise.
3850         (neon_vcvt_n<mode>): Likewise.
3851         (neon_vmovn<mode>): Likewise.
3852         (neon_vqmovn<mode>): Likewise.
3853         (neon_vqmovun<mode>): Likewise.
3854         (neon_vmovl<mode>): Likewise.
3855         (neon_vmul_lane<mode>): Likewise.
3856         (neon_vmul_lane<mode>): Likewise.
3857         (neon_vmull_lane<mode>): Likewise.
3858         (neon_vqdmull_lane<mode>): Likewise.
3859         (neon_vqdmulh_lane<mode>): Likewise.
3860         (neon_vqdmulh_lane<mode>): Likewise.
3861         (neon_vmla_lane<mode>): Likewise.
3862         (neon_vmla_lane<mode>): Likewise.
3863         (neon_vmlal_lane<mode>): Likewise.
3864         (neon_vqdmlal_lane<mode>): Likewise.
3865         (neon_vmls_lane<mode>): Likewise.
3866         (neon_vmls_lane<mode>): Likewise.
3867         (neon_vmlsl_lane<mode>): Likewise.
3868         (neon_vqdmlsl_lane<mode>): Likewise.
3869         (neon_vext<mode>): Likewise.
3870         (neon_vrev64<mode>): Likewise.
3871         (neon_vrev32<mode>): Likewise.
3872         (neon_vrev16<mode>): Likewise.
3873         (neon_vbsl<mode>_internal): Likewise.
3874         (neon_vshl<mode>): Likewise.
3875         (neon_vqshl<mode>): Likewise.
3876         (neon_vshr_n<mode>): Likewise.
3877         (neon_vshrn_n<mode>): Likewise.
3878         (neon_vqshrn_n<mode>): Likewise.
3879         (neon_vqshrun_n<mode>): Likewise.
3880         (neon_vshl_n<mode>): Likewise.
3881         (neon_vqshl_n<mode>): Likewise.
3882         (neon_vqshlu_n<mode>): Likewise.
3883         (neon_vshll_n<mode>): Likewise.
3884         (neon_vsra_n<mode>): Likewise.
3885         (neon_vsri_n<mode>): Likewise.
3886         (neon_vsli_n<mode>): Likewise.
3887         (neon_vtbl1v8qi): Likewise.
3888         (neon_vtbl2v8qi): Likewise.
3889         (neon_vtbl3v8qi): Likewise.
3890         (neon_vtbl4v8qi): Likewise.
3891         (neon_vtbl1v16qi): Likewise.
3892         (neon_vtbl2v16qi): Likewise.
3893         (neon_vcombinev16qi): Likewise.
3894         (neon_vtbx1v8qi): Likewise.
3895         (neon_vtbx2v8qi): Likewise.
3896         (neon_vtbx3v8qi): Likewise.
3897         (neon_vtbx4v8qi): Likewise.
3898         (*neon_vtrn<mode>_insn): Likewise.
3899         (*neon_vzip<mode>_insn): Likewise.
3900         (*neon_vuzp<mode>_insn): Likewise.
3901         (neon_vld1<mode>): Likewise.
3902         (neon_vld1_lane<mode>): Likewise.
3903         (neon_vld1_lane<mode>): Likewise.
3904         (neon_vld1_dup<mode>): Likewise.
3905         (neon_vld1_dup<mode>): Likewise.
3906         (neon_vld1_dupv2di): Likewise.
3907         (neon_vst1<mode>): Likewise.
3908         (neon_vst1_lane<mode>): Likewise.
3909         (neon_vst1_lane<mode>): Likewise.
3910         (neon_vld2<mode>): Likewise.
3911         (neon_vld2<mode>): Likewise.
3912         (neon_vld2_lane<mode>): Likewise.
3913         (neon_vld2_lane<mode>): Likewise.
3914         (neon_vld2_dup<mode>): Likewise.
3915         (neon_vst2<mode>): Likewise.
3916         (neon_vst2<mode>): Likewise.
3917         (neon_vst2_lane<mode>): Likewise.
3918         (neon_vst2_lane<mode>): Likewise.
3919         (neon_vld3<mode>): Likewise.
3920         (neon_vld3qa<mode>): Likewise.
3921         (neon_vld3qb<mode>): Likewise.
3922         (neon_vld3_lane<mode>): Likewise.
3923         (neon_vld3_lane<mode>): Likewise.
3924         (neon_vld3_dup<mode>): Likewise.
3925         (neon_vst3<mode>): Likewise.
3926         (neon_vst3qa<mode>): Likewise.
3927         (neon_vst3qb<mode>): Likewise.
3928         (neon_vst3_lane<mode>): Likewise.
3929         (neon_vst3_lane<mode>): Likewise.
3930         (neon_vld4<mode>): Likewise.
3931         (neon_vld4qa<mode>): Likewise.
3932         (neon_vld4qb<mode>): Likewise.
3933         (neon_vld4_lane<mode>): Likewise.
3934         (neon_vld4_lane<mode>): Likewise.
3935         (neon_vld4_dup<mode>): Likewise.
3936         (neon_vst4<mode>): Likewise.
3937         (neon_vst4qa<mode>): Likewise.
3938         (neon_vst4qb<mode>): Likewise.
3939         (neon_vst4_lane<mode>): Likewise.
3940         (neon_vst4_lane<mode>): Likewise.
3941         (neon_vec_unpack<US>_lo_<mode>): Likewise.
3942         (neon_vec_unpack<US>_hi_<mode>): Likewise.
3943         (neon_vec_<US>mult_lo_<mode>): Likewise.
3944         (neon_vec_<US>mult_hi_<mode>): Likewise.
3945         (neon_vec_<US>shiftl_<mode>): Likewise.
3946         (neon_unpack<US>_<mode>): Likewise.
3947         (neon_vec_<US>mult_<mode>): Likewise.
3948         (vec_pack_trunc_<mode>): Likewise.
3949         (neon_vec_pack_trunc_<mode>): Likewise.
3950         (neon_vabd<mode>_2): Likewise.
3951         (neon_vabd<mode>_3): Likewise.
3953 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3955         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
3956         (load_pair): Update type attribute.
3957         (store_pair): Update type attribute.
3958         * config/aarch64/iterators.md (q): New.
3960 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3962         * config/arm/types.md: Add new types for Neon insns.
3964 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3965             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3966             Sergey Lega  <sergey.s.lega@intel.com>
3967             Anna Tikhonova  <anna.tikhonova@intel.com>
3968             Ilya Tocar  <ilya.tocar@intel.com>
3969             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3970             Ilya Verbin  <ilya.verbin@intel.com>
3971             Kirill Yukhin  <kirill.yukhin@intel.com>
3972             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3974         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
3975         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
3976         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
3977         (rcp14<mode>): New.
3978         (srcp14<mode>): Ditto.
3979         (rsqrt14<mode>): Ditto.
3980         (rsqrt14<mode>): Ditto.
3981         (avx512f_vmscalef<mode>): Ditto.
3982         (avx512f_scalef<mode>): Ditto.
3983         (avx512f_getexp<mode>): Ditto.
3984         (avx512f_sgetexp<mode>): Ditto.
3985         (avx512f_fixupimm<mode>): Ditto.
3986         (avx512f_sfixupimm<mode>): Ditto.
3987         (avx512f_rndscale<mode>): Ditto.
3988         (*avx512er_exp2<mode>): Ditto.
3989         (*avx512er_rcp28<mode>): Ditto.
3990         (avx512er_rsqrt28<mode>): Ditto.
3991         (avx512f_getmant<mode>): Ditto.
3992         (avx512f_getmant<mode>): Ditto.
3993         (avx512f_rndscale<mode>): Fix formatting.
3995 2013-10-15  Martin Jambor  <mjambor@suse.cz>
3997         * ipa-utils.h (ipa_edge_within_scc): Declare.
3998         * ipa-cp.c (edge_within_scc): Moved...
3999         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
4001 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4002             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4003             Sergey Lega  <sergey.s.lega@intel.com>
4004             Anna Tikhonova  <anna.tikhonova@intel.com>
4005             Ilya Tocar  <ilya.tocar@intel.com>
4006             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4007             Ilya Verbin  <ilya.verbin@intel.com>
4008             Kirill Yukhin  <kirill.yukhin@intel.com>
4009             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4011         * config/i386/predicates.md (const_8_to_15_operand): New.
4012         (const_16_to_31_operand): Ditto.
4013         * config/i386/sse.md (V8FI): New.
4014         (V16FI): Ditto.
4015         (reduc_splus_v8df): Ditto.
4016         (reduc_splus_v16sf): Ditto.
4017         (avx512f_vextract<shuffletype>32x4_1): Ditto.
4018         (vec_extract_hi_<mode>): Ditto.
4019         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
4020         (vec_set_lo_<mode>): Ditto.
4021         (vec_set_hi_<mode>): Ditto.
4022         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
4023         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
4024         (avx512f_pshufd_1): Ditto.
4025         (avx512f_broadcast<mode>): Ditto.
4026         (avx512f_broadcast<mode>): Ditto.
4027         (define_split): Split vec_extract_lo into move.
4028         (ssequartermode): Ditto.
4029         (ssedoublemode): Extened with wider modes.
4030         (vec_extract_lo_<mode>): Ditto.
4032 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4033             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4034             Sergey Lega  <sergey.s.lega@intel.com>
4035             Anna Tikhonova  <anna.tikhonova@intel.com>
4036             Ilya Tocar  <ilya.tocar@intel.com>
4037             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4038             Ilya Verbin  <ilya.verbin@intel.com>
4039             Kirill Yukhin  <kirill.yukhin@intel.com>
4040             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4042         * config/i386/predicates.md (register_or_constm1_operand): New.
4043         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
4044         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
4045         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
4046         UNSPEC_SCATTER_PREFETCH
4047         (VI48_512): New.
4048         (avx512f_ucmp<mode>3): Ditto.
4049         (avx512f_vternlog<mode>): Ditto.
4050         (avx512f_align<mode>): Ditto.
4051         (<shift_insn><mode>3): Ditto.
4052         (avx512f_<rotate>v<mode>): Ditto.
4053         (avx512f_<rotate><mode>): Ditto.
4054         (avx512f_eq<mode>3): Ditto.
4055         (avx512f_eq<mode>3_1): Ditto.
4056         (avx512f_gt<mode>3): Ditto.
4057         (avx512f_testm<mode>3): Ditto.
4058         (avx512f_testnm<mode>3): Ditto.
4059         (avx512pf_gatherpf<mode>): Ditto.
4060         (*avx512pf_gatherpf<mode>_mask): Ditto.
4061         (*avx512pf_gatherpf<mode>): Ditto.
4062         (avx512pf_scatterpf<mode>): Ditto.
4063         (*avx512pf_scatterpf<mode>_mask): Ditto.
4064         (*avx512pf_scatterpf<mode>): Ditto.
4065         (avx512f_vec_dup_gpr<mode>): Ditto.
4066         (clz<mode>2): Ditto.
4067         (conflict<mode>): Ditto.
4068         (REDUC_SMINMAX_MODE): Extened with wider modes.
4069         (reduc_<code>_<mode>): Ditto.
4070         (vlshr<mode>3): Ditto.
4071         (vashl<mode>3): Ditto.
4073 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4074             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4075             Sergey Lega  <sergey.s.lega@intel.com>
4076             Anna Tikhonova  <anna.tikhonova@intel.com>
4077             Ilya Tocar  <ilya.tocar@intel.com>
4078             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4079             Ilya Verbin  <ilya.verbin@intel.com>
4080             Kirill Yukhin  <kirill.yukhin@intel.com>
4081             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4083         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
4084         UNSPEC_SCATTER.
4085         (VI48F_512): New.
4086         (avx512fmaskmode): Ditto.
4087         (bcstscalarsuff): Ditto.
4088         (avx512f_blendm<mode>): Ditto.
4089         (cmp_imm_predicate): Ditto.
4090         (avx512f_cmp<mode>3): Ditto.
4091         (avx512f_vec_dup<mode>): Ditto.
4092         (avx512f_vec_dup_mem<mode>): Ditto.
4093         (avx512f_vpermi2var<mode>3): Ditto.
4094         (avx512f_vpermt2var<mode>3): Ditto.
4095         (vec_init<mode>): Ditto.
4096         (avx512f_gathersi<mode>): Ditto.
4097         (*avx512f_gathersi<mode>): Ditto.
4098         (*avx512f_gathersi<mode>_2): Ditto.
4099         (avx512f_gatherdi<mode>): Ditto.
4100         (*avx512f_gatherdi<mode>): Ditto.
4101         (*avx512f_gatherdi<mode>_2): Ditto.
4102         (avx512f_scattersi<mode>): Ditto.
4103         (*avx512f_scattersi<mode>): Ditto.
4104         (avx512f_scatterdi<mode>): Ditto.
4105         (*avx512f_scatterdi<mode>): Ditto.
4106         (sseintprefix): Extened with wider modes.
4107         (VEC_GATHER_IDXSI): Ditto.
4108         (VEC_GATHER_IDXDI): Ditto.
4109         (VEC_GATHER_SRCDI): Ditto.
4111 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
4112             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4114         * config/arm/t-aprofile: New file.
4115         * config.gcc: Handle --with-multilib-list option.
4117 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
4119         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
4120         the input for r2.
4122 2013-10-15  Richard Biener  <rguenther@suse.de>
4124         * tree-loop-distribution.c (build_empty_rdg): Inline into
4125         single user.
4126         (rdg_flag_vertex): Inline into single user.
4127         (rdg_flag_vertex_and_dependent): Likewise.
4128         (build_rdg_partition_for_vertex): Remove processed bitmap.
4129         (rdg_build_partitions): Simplify.
4131 2013-10-15  Richard Biener  <rguenther@suse.de>
4133         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
4134         Restructure forwarding through conversions and copies to
4135         avoid performing copy-propagation the wrong way.  Adjust
4136         recursion invocations.
4137         (forward_propagate_addr_expr): Add argument stating if we
4138         are recursing from a single-use.
4139         (ssa_forward_propagate_and_combine): Adjust.
4141 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
4143         * dumpfile.h (gcc::dump_manager): New class, to hold state
4144         relating to dumpfile management.
4145         (get_dump_file_name): Remove in favor of method of dump_manager.
4146         (dump_initialized_p): Likewise.
4147         (dump_start): Likewise.
4148         (dump_finish): Likewise.
4149         (dump_switch_p): Likewise.
4150         (dump_register): Likewise.
4151         (get_dump_file_info): Likewise.
4152         * context.c (gcc::context::context): Construct the dump_manager
4153         instance.
4154         * context.h (gcc::context::get_dumps): New.
4155         (gcc::context::m_dumps): New.
4156         * coverage.c (coverage_init): Port to dump_manager API.
4157         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
4158         (extra_dump_files_in_use): Likewise.
4159         (extra_dump_files_alloced): Likewise.
4160         (gcc::dump_manager::dump_manager): New.
4161         (dump_register): Convert to...
4162         (gcc::dump_manager::dump_register): ...method, replacing
4163         function-static next_dump with m_next_dump field.
4164         (get_dump_file_info): Convert to...
4165         (gcc::dump_manager::get_dump_file_info): ...method.
4166         (get_dump_file_name): Convert to...
4167         (gcc::dump_manager::get_dump_file_name): ...method.
4168         (dump_start): Convert to...
4169         (gcc::dump_manager::dump_start): ...method.
4170         (dump_finish): Convert to...
4171         (gcc::dump_manager::dump_finish): ...method.
4172         (dump_begin): Replace body with...
4173         (gcc::dump_manager::dump_begin): ...new method.
4174         (dump_phase_enabled_p): Convert to...
4175         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4176         (dump_phase_enabled_p): Convert to...
4177         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4178         (dump_initialized_p):  Convert to...
4179         (gcc::dump_manager::dump_initialized_p): ...method.
4180         (dump_flag_name): Replace body with...
4181         (gcc::dump_manager::dump_flag_name): ...new method.
4182         (dump_enable_all): Convert to...
4183         (gcc::dump_manager::dump_enable_all): ...new method.
4184         (opt_info_enable_passes): Convert to...
4185         (gcc::dump_manager::opt_info_enable_passes): ...new method.
4186         (dump_switch_p_1): Convert to...
4187         (gcc::dump_manager::dump_switch_p_1): ...new method.
4188         (dump_switch_p):  Convert to...
4189         (gcc::dump_manager::dump_switch_p): ...new method.
4190         (opt_info_switch_p): Port to dump_manager API.
4191         (enable_rtl_dump_file): Likewise.
4192         * opts-global.c (handle_common_deferred_options): Port to new
4193         dump_manager API.
4194         * passes.c (pass_manager::finish_optimization_passes): Likewise.
4195         (pass_manager::register_one_dump_file): Likewise.
4196         (pass_manager::register_pass): Likewise.
4197         (pass_init_dump_file): Likewise.
4198         (pass_fini_dump_file): Likewise.
4199         * statistics.c (statistics_early_init): Likewise.
4201 2013-10-14  Richard Biener  <rguenther@suse.de>
4203         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
4204         iterative_hash_canonical_type, gimple_canonical_type_hash,
4205         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
4206         gimple_register_canonical_type, print_gimple_types_stats,
4207         free_gimple_type_tables): Move to lto/lto.c
4208         (gt-gimple.h): Do not include.
4209         * gimple.h (gimple_register_canonical_type,
4210         print_gimple_types_stats, free_gimple_type_tables): Remove.
4211         * Makefile.in (GTFILES): Remove gimple.c.
4213 2013-10-14  Travis Snoozy  <quandary@remstate.com>
4215         PR target/58716
4216         * config/msp430/msp430.c (msp430_option_override): Correct thinko
4217         scanning for msp430x targets.
4219 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
4221         PR bootstrap/58509
4222         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
4223         (registers_ok_for_ldd_peep): Move around.
4224         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
4225         * config/sparc/sparc.md (widening peepholes): Use it.
4227 2013-10-14  Richard Biener  <rguenther@suse.de>
4229         PR middle-end/58712
4230         PR middle-end/55358
4231         * gimple.c (iterative_hash_canonical_type): Make sure to
4232         record the hash into the correct hashtable slot.
4234 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
4236         PR rtl-optimization/58662
4237         * combine.c (try_combine): Take into account death nodes on I2 when
4238         splitting a PARALLEL of two independent SETs.  Fix dump message.
4240 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
4242         PR target/51244
4243         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
4244         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
4245         * config/sh/t-sh (sh_treg_combine.o): New entry.
4246         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
4247         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
4248         (register_sh_passes): New function.  Register sh_treg_combine pass.
4249         (sh_option_override): Invoke it.
4250         (sh_canonicalize_comparison): Handle op0_preserve_value.
4251         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
4252         opportunities.  Canonicalize branch condition.
4253         (nott): Allow only if pseudos can be created for non-SH2A.
4255 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
4257         PR target/58690
4258         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
4259         (ix86_expand_movmem): Replace copy_addr_to_reg with
4260         ix86_copy_addr_to_reg.
4261         (ix86_expand_setmem): Likewise.
4263 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
4265         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
4266         provided by ix86_fp_compare_mode instead of CCFPUmode.
4268 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
4270         * config/aarch64/arm_neon.h
4271         (vtbx<1,3>_<psu>8): Fix register constriants.
4273 2013-10-11  Jeff Law  <law@redhat.com>
4275         PR tree-optimization/58640
4276         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
4277         threading paths that cross over two loop entry points.
4279 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4281         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
4282         handle vector float as well.
4283         (*vsx_le_perm_load_v4si): Likewise.
4284         (*vsx_le_perm_store_v2di): Likewise.
4285         (*vsx_le_perm_store_v4si): Likewise.
4287 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4289         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
4290         directly to circumvent subtract from splat{31} workaround.
4291         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
4292         prototype.
4293         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
4294         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
4295         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
4296         (altivec_vperm_<mode>): Convert to define_insn_and_split to
4297         separate big and little endian logic.
4298         (*altivec_vperm_<mode>_internal): New define_insn.
4299         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
4300         separate big and little endian logic.
4301         (*altivec_vperm_<mode>_uns_internal): New define_insn.
4302         (vec_permv16qi): Add little endian logic.
4304 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
4306         * doc/extend.texi (returns_nonnull): Remove arguments.
4308 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4309             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4310             Sergey Lega  <sergey.s.lega@intel.com>
4311             Anna Tikhonova  <anna.tikhonova@intel.com>
4312             Ilya Tocar  <ilya.tocar@intel.com>
4313             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4314             Ilya Verbin  <ilya.verbin@intel.com>
4315             Kirill Yukhin  <kirill.yukhin@intel.com>
4316             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4318         * config/i386/sse.md (VI48F_256_512): New.
4319         (avx2_permvar<mode>): Change to ...
4320         (<avx2_avx512f>_permvar<mode>): This.
4322 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4323             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4324             Sergey Lega  <sergey.s.lega@intel.com>
4325             Anna Tikhonova  <anna.tikhonova@intel.com>
4326             Ilya Tocar  <ilya.tocar@intel.com>
4327             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4328             Ilya Verbin  <ilya.verbin@intel.com>
4329             Kirill Yukhin  <kirill.yukhin@intel.com>
4330             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4332         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
4333         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
4334         * config/i386/sse.md (VI4_AVX): New.
4335         (sf2simodelower): Ditto.
4336         (sse2_cvtps2dq): Change to ...
4337         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
4339 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4340             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4341             Sergey Lega  <sergey.s.lega@intel.com>
4342             Anna Tikhonova  <anna.tikhonova@intel.com>
4343             Ilya Tocar  <ilya.tocar@intel.com>
4344             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4345             Ilya Verbin  <ilya.verbin@intel.com>
4346             Kirill Yukhin  <kirill.yukhin@intel.com>
4347             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4349         * config/i386/sse.md (V_512): New.
4350         (VI_512): Ditto.
4351         (vcond<V_512:mode><VF_512:mode>): Ditto.
4352         (vcond<V_512:mode><VI_512:mode>): Ditto.
4353         (vcondu<V_512:mode><VI_512:mode>): Ditto.
4355 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4356             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4357             Sergey Lega  <sergey.s.lega@intel.com>
4358             Anna Tikhonova  <anna.tikhonova@intel.com>
4359             Ilya Tocar  <ilya.tocar@intel.com>
4360             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4361             Ilya Verbin  <ilya.verbin@intel.com>
4362             Kirill Yukhin  <kirill.yukhin@intel.com>
4363             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4365         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
4366         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
4367         (FMAMODE): Ditto.
4368         (fma<mode>4): Removed condition.
4369         (fms<mode>4): Ditto.
4370         (fnma<mode>4): Ditto.
4371         (fnms<mode>4): Ditto.
4372         (fma4i_fmadd_<mode>): Ditto.
4373         (*fma_fmadd_<mode>): Ditto.
4374         (*fma_fmsub_<mode>): Ditto.
4375         (*fma_fnmadd_<mode>): Ditto.
4376         (*fma_fnmsub_<mode>): Ditto.
4377         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
4378         (*fma_fmaddsub_<mode>): Ditto.
4379         (*fma_fmsubadd_<mode>): Ditto.
4380         (*fmai_fmadd_<mode>): Ditto.
4381         (*fmai_fmsub_<mode>): Ditto.
4382         (*fmai_fnmadd_<mode>): Ditto.
4383         (*fmai_fnmsub_<mode>): Ditto.
4385 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4386             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4387             Sergey Lega  <sergey.s.lega@intel.com>
4388             Anna Tikhonova  <anna.tikhonova@intel.com>
4389             Ilya Tocar  <ilya.tocar@intel.com>
4390             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4391             Ilya Verbin  <ilya.verbin@intel.com>
4392             Kirill Yukhin  <kirill.yukhin@intel.com>
4393             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4395         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
4396         (VI124_256): Changed to ...
4397         (VI124_256_48_512): This.
4398         (ssepackmode): Extended with wider modes.
4399         (<code><mode>3): Changed iterator.
4400         (*avx2_<code><mode>3): Ditto.
4401         (vec_pack_trunc_<mode>): Ditto.
4403 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4404             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4405             Sergey Lega  <sergey.s.lega@intel.com>
4406             Anna Tikhonova  <anna.tikhonova@intel.com>
4407             Ilya Tocar  <ilya.tocar@intel.com>
4408             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4409             Ilya Verbin  <ilya.verbin@intel.com>
4410             Kirill Yukhin  <kirill.yukhin@intel.com>
4411             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4413         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
4414         (VI8F_256_512): Ditto.
4415         (abs<mode>2): Changed iterator.
4416         (avx2_perm<mode>): Changed to ...
4417         (<avx2_avx512f>_perm<mode>): This.
4418         (avx2_perm<mode>_1): Changed to ...
4419         (<avx2_avx512f>_perm<mode>_1): This.
4421 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4422             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4423             Sergey Lega  <sergey.s.lega@intel.com>
4424             Anna Tikhonova  <anna.tikhonova@intel.com>
4425             Ilya Tocar  <ilya.tocar@intel.com>
4426             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4427             Ilya Verbin  <ilya.verbin@intel.com>
4428             Kirill Yukhin  <kirill.yukhin@intel.com>
4429             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4431         * config/i386/sse.md (VI48_AVX512F): New.
4432         (VI48_AVX2): Changed to ...
4433         (VI48_AVX2_48_AVX512F): This.
4434         (avx2_ashrv<mode>): Changed to ...
4435         (<avx2_avx512f>_ashrv<mode>): This.
4436         (avx2_<shift_insn>v<mode>): Changed to ...
4437         (<avx2_avx512f>_<shift_insn>v<mode>): This.
4439 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4440             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4441             Sergey Lega  <sergey.s.lega@intel.com>
4442             Anna Tikhonova  <anna.tikhonova@intel.com>
4443             Ilya Tocar  <ilya.tocar@intel.com>
4444             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4445             Ilya Verbin  <ilya.verbin@intel.com>
4446             Kirill Yukhin  <kirill.yukhin@intel.com>
4447             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4449         * config/i386/sse.md (VI4_AVX512F): New.
4450         (VI8_AVX2_AVX512F): Ditto.
4451         (mul<mode>3): Extended with wider modes.
4452         (*<sse4_1_avx2>_mul<mode>3): Ditto.
4453         (mul<mode>3): Ditto.
4454         (vec_widen_<s>mult_odd_<mode>): Ditto.
4456 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4457             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4458             Sergey Lega  <sergey.s.lega@intel.com>
4459             Anna Tikhonova  <anna.tikhonova@intel.com>
4460             Ilya Tocar  <ilya.tocar@intel.com>
4461             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4462             Ilya Verbin  <ilya.verbin@intel.com>
4463             Kirill Yukhin  <kirill.yukhin@intel.com>
4464             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4466         * config/i386/sse.md (VI2_AVX512F): New.
4467         (VI124_AVX512F): Ditto.
4468         (sseunpackmode): Extended with wider modes.
4469         (sseunpackfltmode): Ditto.
4470         (vec_unpacks_float_hi_<mode>): Ditto.
4471         (vec_unpacks_float_lo_<mode>): Ditto.
4472         (vec_unpacku_float_hi_<mode>): Ditto.
4473         (vec_unpacku_float_lo_<mode>): Ditto.
4474         (vec_unpacks_lo_<mode>): Ditto.
4475         (vec_unpacks_hi_<mode>): Ditto.
4476         (vec_unpacku_lo_<mode>): Ditto.
4477         (vec_unpacku_hi_<mode>): Ditto.
4479 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4480             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4481             Sergey Lega  <sergey.s.lega@intel.com>
4482             Anna Tikhonova  <anna.tikhonova@intel.com>
4483             Ilya Tocar  <ilya.tocar@intel.com>
4484             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4485             Ilya Verbin  <ilya.verbin@intel.com>
4486             Kirill Yukhin  <kirill.yukhin@intel.com>
4487             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4489         * config/i386/i386.md (multdiv): New.
4490         (multdiv_mnemonic): Ditto.
4491         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
4492         (<sse>_vm<multdiv_mnemonic><mode>3): This.
4493         (<sse>_vmdiv<mode>3): Removed.
4495 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4496             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4497             Sergey Lega  <sergey.s.lega@intel.com>
4498             Anna Tikhonova  <anna.tikhonova@intel.com>
4499             Ilya Tocar  <ilya.tocar@intel.com>
4500             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4501             Ilya Verbin  <ilya.verbin@intel.com>
4502             Kirill Yukhin  <kirill.yukhin@intel.com>
4503             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4505         * config/i386/sse.md (V): Extended with wider modes.
4506         (VF2): Ditto.
4507         (ssehalfvecmode): Ditto.
4508         (i128): Ditto.
4509         (ssepackfltmode): Ditto.
4510         (avx_vec_concat<mode>): Ditto.
4511         (V_256_512): New iterator.
4512         (VF2_512_256): Ditto.
4513         (si2dfmode): New attribute.
4514         (si2dfmodelower): Ditto.
4515         (sf2dfmode): Ditto.
4516         (concat_tg_mode): Ditto.
4517         (floatv4siv4df2): Changed to ...
4518         (float<si2dfmodelower><mode>2): This.
4519         (avx_cvtps2pd256): Changed to ...
4520         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
4521         (vec_pack_trunc_v4df): Changed to ...
4522         (vec_pack_trunc_<mode>): This.
4523         (avx_vpermil<mode>): Changed to ...
4524         (<sse2_avx_avx512f>_vpermil<mode>): This.
4525         (<fixsuffix>fix_truncv8dfv8si2): New.
4526         (vec_pack_sfix_trunc_v8df): Ditto.
4527         (avx512f_rndscale<mode>): Ditto.
4528         (avx512f_roundpd512): Ditto.
4529         (vec_pack_ufix_trunc_<mode>): Updated iterator.
4531 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4532             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4533             Sergey Lega  <sergey.s.lega@intel.com>
4534             Anna Tikhonova  <anna.tikhonova@intel.com>
4535             Ilya Tocar  <ilya.tocar@intel.com>
4536             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4537             Ilya Verbin  <ilya.verbin@intel.com>
4538             Kirill Yukhin  <kirill.yukhin@intel.com>
4539             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4541         * config/i386/i386.md (any_fix): New iterator.
4542         (fixsuffix): New attribute.
4543         * config/i386/sse.md (VF1): Extened with wider modes.
4544         (VI): Ditto.
4545         (VI_AVX2): Ditto.
4546         (VI8): Ditto.
4547         (sseintvecmodelower): Ditto.
4548         (ssescalarmode): Ditto.
4549         (ssescalarnum): Ditto.
4550         (VF1_128_256): New.
4551         (ssexmmmode): Ditto.
4552         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
4553         (<sse>_rcp<mode>2): Change iterator.
4554         (rsqrt<mode>2): Ditto.
4555         (<sse>_rsqrt<mode>2): Ditto.
4556         (avx2_vec_dup<mode>): Ditto.
4557         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
4558         (round<mode>2_sfix): Ditto.
4559         (avx2_pbroadcast<mode>): Ditto.
4560         (*andnot<mode>3): Handle XI mode.
4561         (*<code><mode>3): Ditto.
4562         (AVXTOSSEMODE): Removed.
4563         (avx_vpermil<mode>): Changed to ...
4564         (<sse2_avx_avx512f>_vpermil<mode>): This.
4566 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4567             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4568             Sergey Lega  <sergey.s.lega@intel.com>
4569             Anna Tikhonova  <anna.tikhonova@intel.com>
4570             Ilya Tocar  <ilya.tocar@intel.com>
4571             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4572             Ilya Verbin  <ilya.verbin@intel.com>
4573             Kirill Yukhin  <kirill.yukhin@intel.com>
4574             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4576         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
4577         (<sse>_comi): Ditto.
4578         (<sse>_ucomi): Ditto.
4579         (sse_cvtss2siq_2): Ditto.
4580         (sse2_cvtsd2si): Ditto.
4581         (sse2_cvtsd2siq): Ditto.
4582         (sse2_cvttsd2si): Ditto.
4583         (sse2_cvttsd2siq): Ditto.
4584         (<shift_insn><mode>3): Ditto.
4585         (sse2_cvtsi2sdq): Update constraint and prefix.
4586         (sse_cvtsi2ss): Update prefix.
4587         (sse_cvtsi2ssq): Ditto.
4589 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4591         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
4592         ignore internal calls.
4594 2013-10-11  Richard Biener  <rguenther@suse.de>
4596         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
4597         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
4598         TREE_OVERFLOW is set.
4600 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
4602         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
4604         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
4605         * gimple.h: Likewise.
4607         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
4609         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
4610         macros.
4612         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
4614         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
4615         * configure: Regenerate.
4617 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4619         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
4620         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
4621         formatting fixes, use pp_colon instead of pp_character (..., ':'),
4622         similarly pp_right_paren.
4623         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
4624         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
4625         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
4626         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
4627         clauses.
4628         (omp_declare_simd_clauses_equal,
4629         omp_remove_redundant_declare_simd_attrs): New functions.
4630         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
4631         (walk_tree_1): Handle new OpenMP 4.0 clauses.
4632         * tree.h (OMP_LOOP_CHECK): Define.
4633         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
4634         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
4635         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
4636         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
4637         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
4638         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
4639         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
4640         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
4641         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
4642         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
4643         OMP_CLAUSE_SIMDLEN_EXPR): Define.
4644         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
4645         (omp_remove_redundant_declare_simd_attrs): New prototype.
4646         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
4647         GIMPLE_OMP_TEAMS): New codes.
4648         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
4649         * omp-low.c (struct omp_context): Add cancel_label and cancellable
4650         fields.
4651         (target_nesting_level): New variable.
4652         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
4653         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
4654         for collapse > 1 static schedule unless ordered.
4655         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
4656         (determine_parallel_type): Adjust get_ws_args_for caller.
4657         (install_var_field): Handle mask & 4 for double indirection.
4658         (scan_sharing_clauses): Ignore shared clause on teams construct.
4659         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
4660         (create_omp_child_function): If inside target or declare target
4661         constructs, set "omp declare target" attribute on the child function.
4662         (find_combined_for): New function.
4663         (scan_omp_parallel): Handle combined loops.
4664         (scan_omp_target, scan_omp_teams): New functions.
4665         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
4666         restrictions and set ctx->cancellable for cancellable constructs.
4667         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
4668         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
4669         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
4670         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
4671         (omp_clause_aligned_alignment): New function.
4672         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
4673         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
4674         on teams constructs.  Handle user defined reductions and new
4675         OpenMP 4.0 clauses.
4676         (lower_reduction_clauses): Don't set placeholder to address of ref
4677         if it has already the right type.
4678         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
4679         (expand_parallel_call): Use the new non-_start suffixed builtins,
4680         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
4681         and GOMP_parallel_end after the call.
4682         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
4683         (expand_omp_for_init_counts): Handle combined loops.
4684         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
4685         loops.
4686         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
4687         end of cancellable loops.
4688         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
4689         Likewise.  Handle collapse > 1 loops.
4690         (expand_omp_simd): Handle combined loops.
4691         (expand_omp_for): Add inner_stmt argument, adjust callers of
4692         expand_omp_for* functions, use expand_omp_for_static*chunk even
4693         for collapse > 1 unless ordered.
4694         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
4695         of cancellable sections.
4696         (expand_omp_single): Remove need_barrier variable, just rely on
4697         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
4698         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
4699         (expand_omp_atomic_load, expand_omp_atomic_store,
4700         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
4701         (expand_omp_target): New function.
4702         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
4703         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
4704         (build_omp_regions_1): Immediately close region for
4705         GF_OMP_TARGET_KIND_UPDATE.
4706         (maybe_add_implicit_barrier_cancel): New function.
4707         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
4708         cancellation.
4709         (lower_omp_single): Likewise.  Add clobber after the barrier.
4710         (lower_omp_taskgroup): New function.
4711         (lower_omp_for): Handle combined loops.  Adjust
4712         lower_rec_input_clauses caller.  Handle cancellation.
4713         (lower_depend_clauses): New function.
4714         (lower_omp_taskreg): Lower depend clauses.  Adjust
4715         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
4716         cancellation.
4717         (lower_omp_target, lower_omp_teams): New functions.
4718         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
4719         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
4720         and GOMP_cancellation_point calls.
4721         (lower_omp): Fold stmts inside of target region.
4722         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
4723         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4724         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
4725         (BT_FN_VOID_OMPFN_PTR_UINT,
4726         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
4727         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
4728         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
4729         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
4730         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
4731         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
4732         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
4733         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
4734         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
4735         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
4736         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
4737         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
4738         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
4739         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
4740         BUILT_IN_GOMP_PARALLEL_END.
4741         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
4742         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4743         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
4744         GF_OMP_FOR_KIND_DISTRIBUTE.
4745         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
4746         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
4747         (dump_gimple_omp_return): Print lhs if it has any.
4748         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
4749         gimple_omp_atomic_seq_cst_p.
4750         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
4751         and GIMPLE_OMP_TEAMS.
4752         * langhooks.c (lhd_omp_mappable_type): New function.
4753         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
4754         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
4755         hook.
4756         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
4757         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
4758         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
4759         (struct gimplify_omp_ctx): Add combined_loop field.
4760         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
4761         on stmts inside of target region.
4762         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
4763         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
4764         ORT_TARGET and ORT_TARGET_DATA.
4765         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
4766         Handle GOVD_MAP.
4767         (omp_notice_threadprivate_variable): Complain about threadprivate
4768         variables in target region.
4769         (omp_notice_variable): Complain about vars with non-mappable type
4770         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
4771         (omp_check_private): Ignore ORT_TARGET* regions.
4772         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
4773         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
4774         (find_combined_omp_for): New function.
4775         (gimplify_omp_for): Handle gimplification of combined loops.
4776         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
4777         OMP_TEAMS.
4778         (gimplify_omp_target_update): New function.
4779         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
4780         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
4781         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
4782         (gimplify_body): If fndecl has "omp declare target" attribute, add
4783         implicit ORT_TARGET context around it.
4784         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
4785         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
4786         * tree-nested.c (convert_nonlocal_reference_stmt,
4787         convert_local_reference_stmt, convert_gimple_call): Handle
4788         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4789         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
4790         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
4791         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
4792         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
4793         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
4794         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
4795         BUILT_IN_GOMP_LOOP_END_CANCEL,
4796         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
4797         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
4798         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
4799         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
4800         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
4801         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
4802         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
4803         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
4804         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
4805         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
4806         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
4807         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
4808         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
4809         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
4810         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
4811         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4812         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
4813         gimple_build_omp_teams): New functions.
4814         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
4815         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
4816         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
4817         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4818         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
4819         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
4820         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
4821         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
4822         GF_OMP_ATOMIC_SEQ_CST): New.
4823         (gimple_build_omp_taskgroup, gimple_build_omp_target,
4824         gimple_build_omp_teams): New prototypes.
4825         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
4826         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4827         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
4828         GIMPLE_OMP_SINGLE as end of range.
4829         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
4830         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
4831         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
4832         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
4833         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
4834         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
4835         gimple_omp_target_kind, gimple_omp_target_set_kind,
4836         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
4837         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
4838         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
4839         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
4840         gimple_omp_teams_set_clauses): New inlines.
4841         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
4842         and GIMPLE_OMP_TASKGROUP.
4843         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
4844         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
4845         enum omp_clause_proc_bind_kind): New.
4846         (union omp_clause_subcode): Add depend_kind, map_kind and
4847         proc_bind_kind fields.
4848         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
4849         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
4850         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
4851         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
4852         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
4854 2013-10-10  Teresa Johnson  <tejohnson@google.com>
4856         * predict.c (tree_estimate_probability): Add new parameter
4857         for estimate_bb_frequencies.
4858         (estimate_bb_frequencies): Add new parameter to force estimation.
4859         (rebuild_frequencies): When max frequency in function is small,
4860         recompute counts from frequencies.
4861         * predict.h (estimate_bb_frequencies): New parameter.
4863 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
4865         * ipa-inline.c (ipa_inline): Fix leak of "order" when
4866         optimizations are disabled.
4868 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
4870         * coverage.c (coverage_finish): Fix leak of da_file_name.
4872 2013-10-10  Jan Hubicka  <jh@suse.cz>
4874         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
4875         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
4876         Bobcat and generic.
4878 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
4880         PR middle-end/58670
4881         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
4882         if any labels are in FALLTHRU_BB, use a special label emitted
4883         immediately after the asm goto insn rather than label_rtx
4884         of the LABEL_DECL.
4885         (expand_asm_stmt): Adjust caller.
4886         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
4887         edge if the last insn in predecessor is a jump with single successor,
4888         but it isn't simplejump_p.
4890 2013-10-10  Richard Biener  <rguenther@suse.de>
4892         PR tree-optimization/58656
4893         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
4895 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4897         * gimplify.c: Include expr.h and tm_p.h for targets with special
4898         va-arg padding requirements.
4900 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4902         * tree-flow.h: Move some prototypes to gimple.h.
4903         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
4904         * gimple.h: Relocate some prototypes from tree-flow.h
4905         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
4906         Move to gimplify.c.
4907         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
4908         (build_va_arg_indirect_ref): Relocate and make static.
4909         (std_gimplify_va_arg_expr): Relocate here.
4910         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
4911         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
4913 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4915         * doc/md.texi: Document the mnemonic attribute.
4917 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4919         PR target/57377
4920         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
4921         (set_attr_alternative x ...) when searching for user defined
4922         mnemonic attribute.
4924 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
4926         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
4928 2013-10-09  Easwaran Raman <eraman@google.com>
4930         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
4931         * cfgexpand.c (defer_stack_allocation): ...use here
4932         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
4934 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
4936         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
4937         (operand_equal_for_value_replacement): New function, extracted from
4938         value_replacement and enhanced to catch more cases.
4939         (value_replacement): Use operand_equal_for_value_replacement.
4941 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4943         * loop-doloop.c (doloop_modify, doloop_optimize): Use
4944         get_max_loop_iterations.
4946 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4948         * config/arm/aarch-common.c (arm_early_load_addr_dep):
4949         Place comment above function.
4951 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
4953         * tree-flow.h: Remove all remaining prototypes, enums and structs that
4954         are not related to tree-cfg.c.
4955         * tree-ssa-address.h: New file.  Relocate prototypes.
4956         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
4957         (addr_for_mem_ref): New.  Combine call to get_address_description and
4958         return addr_for_mem_ref.
4959         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
4960         * tree-ssa-live.h: Adjust prototypes.
4961         * passes.c: Include tree-ssa-live.h.
4962         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
4963         * graphite.c (graphite_transform_loops): Make static.
4964         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
4965         make_pass_graphite, pass_data_graphite_transforms,
4966         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
4967         * ipa-pure-const.c (warn_function_noreturn): Make static.
4968         (execute_warn_function_noreturn, gate_warn_function_noreturn,
4969         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
4970         Relocate from tree-cfg.c
4971         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
4972         static.
4973         (execute_warn_function_noreturn, gate_warn_function_noreturn,
4974         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
4975         Move to ipa-pure-const.c.
4976         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
4977         Relocate from tree-optimize.c.
4978         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
4979         make_pass_fixup_cfg): Move to tree-cfg.c.
4980         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
4981         Relocate some prototypes.
4982         * tree-data-ref.h (tree_check_data_deps) Add prototype.
4983         * tree-dump.c (dump_function_to_file): Remove prototype.
4984         Add tree-flow.h to the include file.
4985         * tree-dump.h: Remove prototype.
4986         * tree-parloops.h: New File.  Add prototypes.
4987         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
4988         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
4989         from tree-ssa-loop.c.
4990         * tree-predcom.c (run_tree_predictive_commoning,
4991         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
4992         Relocate here from tree-ssa-loop.c.
4993         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
4994         ssa_name_values.release ().
4995         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
4996         (ssa_name_values): Relocate from tree-flow.h.
4997         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
4998         * tree-ssa-loop.c (run_tree_predictive_commoning,
4999         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
5000         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5001         make_pass_graphite, pass_data_graphite_transforms,
5002         make_pass_graphite_transforms, gate_tree_parallelize_loops,
5003         tree_parallelize_loops, pass_data_parallelize_loops,
5004         make_pass_parallelize_loops): Move to other files.
5005         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
5006         moved here.
5007         * tree.h: Remove prototypes from tree-address.c.
5009 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5011         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
5012         (struct int_tree_map): Move to tree-hasher.h
5013         (SCALE, LABEL, PERCENT): Move to gimple.h
5014         * tree-flow-inline.h: Delete.  Move functions to other files.
5015         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
5016         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
5017         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
5018         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
5019         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
5020         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
5021         * tree-hasher.h: Don't include tree-flow.h.
5022         (struct int_tree_map): Relocate from tree-flow.h.
5023         * tree-sra.c (contains_view_convert_expr_p): Relocate from
5024         tree-flow-inline.h and make static.
5025         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
5026         tree-flow-inline.h.
5027         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
5028         tree-flow-inline.h and make static.
5029         * tree.h (is_global_var, may_be_aliased): Relocate from
5030         tree-flow-inline.h.
5031         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
5032         * value-prof.c: No longer include tree-flow-inline.h.
5033         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
5035 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5037         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
5038         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
5039         (enum move_pos): Move to tree-ssa-loop-im.h
5040         * cfgloop.h: Move some prototypes.
5041         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
5042         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
5043         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
5044         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
5045         (loop_containing_stmt): Relocate from tree-flow-inline.h.
5046         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
5047         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
5048         (enum move_pos): Relocate here.
5049         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
5050         tree-ssa-loop.c.
5051         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
5052         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
5053         make_pass_lim): Relocate here from tree-ssa-loop.c.
5054         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
5055         tree-ssa-loop.c.
5056         (loop_edge_to_cancel, unloop_loops): Make static.
5057         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5058         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
5059         (tree_complete_unroll, gate_tree_complete_unroll,
5060         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
5061         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5062         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
5063         * tree-ssa-loop-ivopts.c: Remove local prototypes.
5064         (stmt_invariant_in_loop_p): Remove unused function.
5065         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
5066         * tree-ssa-loop-manip.h: New file.  Add prototypes.
5067         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
5068         (gcov_type_to_double_int): Move to cfgloop.h.
5069         (double_int_cmp, bound_index,
5070         estimate_numbers_of_iterations_loop): Make static.
5071         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
5072         (max_loop_iterations): Factor out get_max_loop_iterations.
5073         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
5074         cfgloop.c.
5075         * tree-ssa-loop-niter.h: New file.  Add prototypes.
5076         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
5077         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
5078         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
5079         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
5080         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
5081         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
5082         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
5083         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
5084         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
5085         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
5086         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5087         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
5088         pass_data_complete_unroll, make_pass_complete_unroll,
5089         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5090         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
5091         tree-ssa-loop-ivcanon.c.
5092         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
5093         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
5094         tree-ssa-loop-prefetch.c.
5095         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
5096         tree-ssa-loop-im.c.
5097         (get_lsm_tmp_name): Relocate and add suffix parameter.
5098         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
5099         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
5100         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
5101         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
5102         (get_estimated_loop_iterations): Factor out accessor from
5103         estimated_loop_iterations in tree-ssa-loop-niter.c.
5104         (get_max_loop_iterations): Factor out accessor from
5105         _max_loop_iterations in tree-ssa-niter.c.
5106         * loop-unroll.c (decide_unroll_constant_iterations,
5107         decide_unroll_runtime_iterations, decide_peel_simple,
5108         decide_unroll_stupid): Use new get_* accessors.
5110 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
5112         PR tree-optimization/20318
5113         * doc/extend.texi (returns_nonnull): New function attribute.
5114         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
5115         attribute.
5116         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
5117         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
5119 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
5121         PR middle-end/58570
5122         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
5123         false if both components are bitfields.
5125 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5127         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
5128         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
5129         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
5130         * config/aarch64/aarch64.h
5131         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
5132         * config/aarch64/aarch64-simd-builtins.def
5133         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
5135 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5137         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
5139 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5141         * config/aarch64/arm_neon.h (vdiv_f64): Added.
5143 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5145         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
5146         (vneg_f64): New intrinsic.
5147         (vneg_s8): Asm replaced with C.
5148         (vneg_s16): Likewise.
5149         (vneg_s32): Likewise.
5150         (vneg_s64): New intrinsic.
5151         (vnegq_f32): Asm replaced with C.
5152         (vnegq_f64): Likewise.
5153         (vnegq_s8): Likewise.
5154         (vnegq_s16): Likewise.
5155         (vnegq_s32): Likewise.
5156         (vnegq_s64): Likewise.
5158 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
5160         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
5162 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5164         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
5165         packed stack special handling.
5166         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
5167         back to fixed stack slots for FPRs saved due to stdarg.
5169 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5171         * config/s390/s390.c (s390_frame_info): Restructure function.
5173 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5175         * config/s390/s390.c (struct s390_frame_layout): New field
5176         gpr_save_slots.
5177         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
5178         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
5179         regs_ever_clobbered to char*.
5180         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
5181         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
5182         (s390_register_info): Enable FPR save slots.  Move/Copy some
5183         functionality into ...
5184         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
5185         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
5186         function.
5187         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
5188         not imply a stack frame.
5189         (s390_init_frame_layout): Remove variable clobbered_regs.
5190         (s390_update_register_info): Remove function.
5191         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
5192         cannot be used for register renaming.
5193         (s390_hard_regno_scratch_ok): New function.
5194         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
5195         (s390_initial_elimination_offset): Change offset calculation of
5196         the return address pointer.
5197         (save_gprs): Deal with only r6 being saved from the call-saved regs.
5198         (restore_gprs): Set frame related flag.
5199         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
5200         (s390_emit_prologue): Call s390_register_info instead of
5201         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
5202         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
5203         (s390_optimize_prologue): Call s390_optimize_register_info.
5204         Try to remove also FPR slot save/restore INSNs.  Remove frame
5205         related flags from restore INSNs.
5207 2013-10-08  DJ Delorie  <dj@redhat.com>
5209         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
5210         (movhi): Likewise.
5212         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
5213         avoid conflict with the MI use of %c.
5214         * config/rl78/rl78-real.md: change %c to %C throughout.
5215         * config/rl78/rl78-virt.md: Likewise.
5217 2013-10-08  Jan Hubicka  <jh@suse.cz>
5219         * config/i386/i386.c (ix86_option_override_internal): Switch
5220         to SSE math for -ffast-math when target ISA supports SSE2.
5222 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
5224         * tree-flow.h: Remove some prototypes.
5225         * tree.h: Remove some protypes, add a couple.
5226         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
5227         using_eh_for_cleanups_p): Add interface routines for front ends.
5228         * tree-eh.h: New file.  Add protoptyes.
5229         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
5230         (add_stmt_to_eh_lp_fn): Make static.
5231         (lower_try_finally): Use new using_eh_for_cleanups_p.
5232         * emit-rtl.c: Include tree-eh.h.
5233         * gimple.h: Include tree-eh.h.
5235 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
5237         PR tree-optimization/58480
5238         * tree-vrp.c (infer_nonnull_range): New function.
5239         (infer_value_range): Call infer_nonnull_range.
5241 2013-10-08  Dehao Chen  <dehao@google.com>
5243         PR tree-optimization/58619
5244         * tree-inline.c (copy_phis_for_bb): Combine location data
5245         only if non-null.
5247 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5249         PR target/58423
5250         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
5251         RTX_FRAME_RELATED_P on INSN.
5253 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5255         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
5256         (altivec_expand_vec_perm_const): Call it.
5258 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5260         * config/rs6000/vector.md (mov<mode>): Emit permuted move
5261         sequences for LE VSX loads and stores at expand time.
5262         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
5263         prototype.
5264         * config/rs6000/rs6000.c (rs6000_const_vec): New.
5265         (rs6000_gen_le_vsx_permute): New.
5266         (rs6000_gen_le_vsx_load): New.
5267         (rs6000_gen_le_vsx_store): New.
5268         (rs6000_gen_le_vsx_move): New.
5269         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
5270         (*vsx_le_perm_load_v4si): New.
5271         (*vsx_le_perm_load_v8hi): New.
5272         (*vsx_le_perm_load_v16qi): New.
5273         (*vsx_le_perm_store_v2di): New.
5274         (*vsx_le_perm_store_v4si): New.
5275         (*vsx_le_perm_store_v8hi): New.
5276         (*vsx_le_perm_store_v16qi): New.
5277         (*vsx_xxpermdi2_le_<mode>): New.
5278         (*vsx_xxpermdi4_le_<mode>): New.
5279         (*vsx_xxpermdi8_le_V8HI): New.
5280         (*vsx_xxpermdi16_le_V16QI): New.
5281         (*vsx_lxvd2x2_le_<mode>): New.
5282         (*vsx_lxvd2x4_le_<mode>): New.
5283         (*vsx_lxvd2x8_le_V8HI): New.
5284         (*vsx_lxvd2x16_le_V16QI): New.
5285         (*vsx_stxvd2x2_le_<mode>): New.
5286         (*vsx_stxvd2x4_le_<mode>): New.
5287         (*vsx_stxvd2x8_le_V8HI): New.
5288         (*vsx_stxvd2x16_le_V16QI): New.
5290 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
5292         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
5294 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5296         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
5297         loop to work also for 31bit ABI.
5298         Save the stack pointer for frame_size > 0.
5300 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5302         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
5303         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
5304         constraint letters from expanders.
5305         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
5306         retry count to general_operand.
5307         ("tabort"): Give operand 0 a mode.
5308         ("tabort_1"): Add mode and constraint letter for operand 0.
5309         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
5311 2013-10-04  Jeff Law  <law@redhat.com>
5313         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
5315         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
5316         out of ...
5317         (thread_across_edge): Here.  Call it.
5319 2013-10-04  Cary Coutant  <ccoutant@google.com>
5321         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
5322         discarding a location list expression (or a piece of one).
5324 2013-10-03  Jan Hubicka  <jh@suse.cz>
5326         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
5327         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
5328         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
5330 2013-10-03  Jan Hubicka  <jh@suse.cz>
5332         * config/i386/i386.c (ix86_option_override_internal): Do not enable
5333         accumulate-outgoing-args when producing unwind info.
5335 2013-10-03  Wei Mi  <wmi@google.com>
5337         * lra-constraints.c (insert_move_for_subreg): New function
5338         extracted from simplify_operand_subreg.
5339         (simplify_operand_subreg): Add reload for paradoxical subreg.
5341 2013-10-03  Rong Xu  <xur@google.com>
5343         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
5344         the candidate of builtin_expect such that we should fix the
5345         size/time estimation.
5346         (estimate_function_body_sizes): Do the acutally size/time fix-up
5347         for builtin_expect.
5349 2013-10-03  Rong Xu  <xur@google.com>
5351         * predict.c (tree_predict_by_opcode): Get the probability
5352         for builtin_expect from param builtin_expect_probability.
5353         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
5354         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
5355         * doc/invoke.texi: Add documentation for builtin-expect-probability.
5357 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
5359         PR c++/19476
5360         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
5361         * calls.c (alloca_call_p): Use get_callee_fndecl.
5362         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
5363         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
5364         Likewise.
5365         (vrp_visit_stmt): Remove duplicated code.
5367 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
5369         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
5370         ceildf2, btruncdf2, instead of vsx_* name.
5372         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
5373         iterators to only do V2DF and V4SF here.  Move the DF code to
5374         rs6000.md where it is combined with SF mode.  Replace <VSv> with
5375         just 'v' since only vector operations are handled with these insns
5376         after moving the DF support to rs6000.md.
5377         (vsx_sub<mode>3): Likewise.
5378         (vsx_mul<mode>3): Likewise.
5379         (vsx_div<mode>3): Likewise.
5380         (vsx_fre<mode>2): Likewise.
5381         (vsx_neg<mode>2): Likewise.
5382         (vsx_abs<mode>2): Likewise.
5383         (vsx_nabs<mode>2): Likewise.
5384         (vsx_smax<mode>3): Likewise.
5385         (vsx_smin<mode>3): Likewise.
5386         (vsx_sqrt<mode>2): Likewise.
5387         (vsx_rsqrte<mode>2): Likewise.
5388         (vsx_fms<mode>4): Likewise.
5389         (vsx_nfma<mode>4): Likewise.
5390         (vsx_copysign<mode>3): Likewise.
5391         (vsx_btrunc<mode>2): Likewise.
5392         (vsx_floor<mode>2): Likewise.
5393         (vsx_ceil<mode>2): Likewise.
5394         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
5395         (vsx_sminsf3): Likewise.
5396         (vsx_fmadf4): Likewise.
5397         (vsx_fmsdf4): Likewise.
5398         (vsx_nfmadf4): Likewise.
5399         (vsx_nfmsdf4): Likewise.
5400         (vsx_cmpdf_internal1): Likewise.
5402         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
5403         simpler to select whether a target has SPE or traditional floating
5404         point support in iterators.
5405         (TARGET_DF_SPE): Likewise.
5406         (TARGET_SF_FPR): Likewise.
5407         (TARGET_DF_FPR): Likewise.
5408         (TARGET_SF_INSN): Macros to say whether floating point support
5409         exists for a given operation for expanders.
5410         (TARGET_DF_INSN): Likewise.
5412         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
5413         combining of SF/DF mode operations, using both traditional and VSX
5414         registers.
5415         (Fvsx): Likewise.
5416         (Ff): Likewise.
5417         (Fv): Likewise.
5418         (Fs): Likewise.
5419         (Ffre): Likewise.
5420         (FFRE): Likewise.
5421         (abs<mode>2): Combine SF/DF modes using traditional floating point
5422         instructions.  Add support for using the upper DF registers with
5423         VSX support, and SF registers with power8-vector support.  Update
5424         expanders for operations supported by both the SPE and traditional
5425         floating point units.
5426         (abs<mode>2_fpr): Likewise.
5427         (nabs<mode>2): Likewise.
5428         (nabs<mode>2_fpr): Likewise.
5429         (neg<mode>2): Likewise.
5430         (neg<mode>2_fpr): Likewise.
5431         (add<mode>3): Likewise.
5432         (add<mode>3_fpr): Likewise.
5433         (sub<mode>3): Likewise.
5434         (sub<mode>3_fpr): Likewise.
5435         (mul<mode>3): Likewise.
5436         (mul<mode>3_fpr): Likewise.
5437         (div<mode>3): Likewise.
5438         (div<mode>3_fpr): Likewise.
5439         (sqrt<mode>3): Likewise.
5440         (sqrt<mode>3_fpr): Likewise.
5441         (fre<Fs>): Likewise.
5442         (rsqrt<mode>2): Likewise.
5443         (cmp<mode>_fpr): Likewise.
5444         (smax<mode>3): Likewise.
5445         (smin<mode>3): Likewise.
5446         (smax<mode>3_vsx): Likewise.
5447         (smin<mode>3_vsx): Likewise.
5448         (negsf2): Delete SF operations that are merged with DF.
5449         (abssf2): Likewise.
5450         (addsf3): Likewise.
5451         (subsf3): Likewise.
5452         (mulsf3): Likewise.
5453         (divsf3): Likewise.
5454         (fres): Likewise.
5455         (fmasf4_fpr): Likewise.
5456         (fmssf4_fpr): Likewise.
5457         (nfmasf4_fpr): Likewise.
5458         (nfmssf4_fpr): Likewise.
5459         (sqrtsf2): Likewise.
5460         (rsqrtsf_internal1): Likewise.
5461         (smaxsf3): Likewise.
5462         (sminsf3): Likewise.
5463         (cmpsf_internal1): Likewise.
5464         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
5465         (negdf2): Delete DF operations that are merged with SF.
5466         (absdf2): Likewise.
5467         (nabsdf2): Likewise.
5468         (adddf3): Likewise.
5469         (subdf3): Likewise.
5470         (muldf3): Likewise.
5471         (divdf3): Likewise.
5472         (fred): Likewise.
5473         (rsqrtdf_internal1): Likewise.
5474         (fmadf4_fpr): Likewise.
5475         (fmsdf4_fpr): Likewise.
5476         (nfmadf4_fpr): Likewise.
5477         (nfmsdf4_fpr): Likewise.
5478         (sqrtdf2): Likewise.
5479         (smaxdf3): Likewise.
5480         (smindf3): Likewise.
5481         (cmpdf_internal1): Likewise.
5482         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
5483         (btrunc<mode>2): Delete separate expander, and combine with the
5484         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
5485         (btrunc<mode>2_fpr): Likewise.
5486         (ceil<mode>2): Likewise.
5487         (ceil<mode>2_fpr): Likewise.
5488         (floor<mode>2): Likewise.
5489         (floor<mode>2_fpr): Likewise.
5490         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
5491         Add support for using the upper registers with VSX and
5492         power8-vector.  Move insns to be closer to the define_expands. On
5493         VSX systems, prefer the traditional form of FMA over the VSX
5494         version, since the traditional form allows the target not to
5495         overlap with the inputs.
5496         (fms<mode>4_fpr): Likewise.
5497         (nfma<mode>4_fpr): Likewise.
5498         (nfms<mode>4_fpr): Likewise.
5500 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5501             Richard Earnshaw  <richard.earnshaw@arm.com>
5503         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
5504         (struct mult_cost_table): Likewise.
5505         (struct mem_cost_table): Likewise.
5506         (struct fp_cost_table): Likewise.
5507         (struct vector_cost_table): Likewise.
5508         (cpu_cost_table): Likewise.
5509         * config/arm/arm.opt (mold-rts-costs): New option.
5510         (mnew-generic-costs): Likewise.
5511         * config/arm/arm.c (generic_extra_costs): New table.
5512         (cortexa15_extra_costs): Likewise.
5513         (arm_slowmul_tune): Use NULL as new costs.
5514         (arm_fastmul_tune): Likewise.
5515         (arm_strongarm_tune): Likewise.
5516         (arm_xscale_tune): Likewise.
5517         (arm_9e_tune): Likewise.
5518         (arm_v6t2_tune): Likewise.
5519         (arm_cortex_a5_tune): Likewise.
5520         (arm_cortex_a9_tune): Likewise.
5521         (arm_v6m_tune): Likewise.
5522         (arm_fa726te_tune): Likewise.
5523         (arm_cortex_a15_tune): Use cortex15_extra_costs.
5524         (arm_cortex_tune): Use generict_extra_costs.
5525         (shifter_op_p): New function.
5526         (arm_unspec_cost): Likewise.
5527         (LIBCALL_COST): Define.
5528         (arm_new_rtx_costs): New function.
5529         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
5530         table is available. Use old costs otherwise unless mnew-generic-costs
5531         is specified.
5532         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
5533         (cpu_cost_table): Declare.
5535 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5537         PR target/58460
5538         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
5539         (*subs_mul_imm_<mode>)
5540         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
5541         (*sub_<shift>_<mode>)
5542         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
5543         Remove k constraint.
5545 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
5547         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
5548         code.
5549         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
5551 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5553         * predict.c (probably_never_executed): New function.
5554         (probably_never_executed_bb_p): Invoke probably_never_executed.
5555         (probably_never_executed_edge_p): Ditto.
5556         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5557         Treat profile insanities conservatively.
5559 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
5561         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
5563 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
5565         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
5566         use it.  Use smaller increase for scratch.  Don't increase reject
5567         for early clobber scratch.
5568         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
5569         setting eliminated regs except setting fp from hfp.
5570         (lra_eliminate): Check lra_insn_recog_data on NULL.
5572 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5574         PR target/58587
5575         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
5576         setting -mvsx-timode by default until the underlying problem is fixed.
5577         (RS6000_CPU, power7 defaults): Likewise.
5579 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5581         * config/x-linux (host-linux.o): Remove header dependencies.
5582         Use $(COMPILE) and $(POSTCOMPILE).
5583         * config/t-linux-android (linux-android.o): Ditto.
5585 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5587         * Makefile.in (expmed.o-warn): Remove.
5589 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5591         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
5592         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
5594 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5596         * dojump.c (do_jump_1): Divide probability between
5597         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
5599 2013-10-02  Tom Tromey  <tromey@redhat.com>
5601         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
5603 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5605         * tree-flow.h: Remove some prototypes.
5606         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
5607         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
5608         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
5609         mark_virtual_phi_result_for_renaming): Relocate here.
5610         * tree-into-ssa.h: Add prototypes.
5611         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
5612         single_pred_before_succ_order.
5613         (blocks_in_phiopt_order): Rename and move to cfganal.c.
5614         (nonfreeing_call_p) Move to gimple.c.
5615         * cfganal.c (single_pred_before_succ_order): Move and renamed from
5616         tree-ssa-phiopt.c.
5617         * basic-block.h (single_pred_before_succ_order): Add prototype.
5618         * gimple.c (nonfreeing_call_p): Relocate here.
5619         * gimple.h: Add prototype.
5620         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
5621         * tree-ssa-dom.h: New file.  Relocate prototypes here.
5622         * tree-ssa.h: Include tree-ssa-dom.h.
5624 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5626         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
5627         Use $(COMPILE) and $(POSTCOMPILE).
5629         * config/alpha/x-alpha (driver-alpha.o): Ditto.
5631 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5633         * tree-flow.h: Remove some prototypes.
5634         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
5635         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5636         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
5637         propagate_tree_value*): Move from here to...
5638         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
5639         propagate_tree_value*): Relocate here.
5640         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5641         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
5642         * gimple-fold.c: Remove gimple-fold.h from include list.
5643         * tree-vrp.c: Remove gimple-fold.h from include list.
5644         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
5645         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
5646         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
5647         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
5648         * sese.c: Add tree-ssa-propagate.h to include list.
5650 2013-10-02  Richard Biener  <rguenther@suse.de>
5652         * tree-loop-distribution.c: Include tree-vectorizer.h for
5653         find_loop_location.
5654         (enum partition_kind): Remove PKIND_REDUCTION.
5655         (struct partition_s): Remove has_writes member, add reduction_p member.
5656         (partition_alloc): Adjust.
5657         (partition_builtin_p): Likewise.
5658         (partition_has_writes): Remove.
5659         (partition_reduction_p): New function.
5660         (partition_merge_into): Likewise.
5661         (generate_code_for_partition): Commonize builtin partition
5662         handling tail.
5663         (rdg_cannot_recompute_vertex_p): Remove.
5664         (already_processed_vertex_p): Likewise.
5665         (rdg_flag_vertex): Do not set has_writes.
5666         (classify_partition): Adjust.
5667         (rdg_build_partitions): Do not set has_writes, treat all
5668         partitions as useful.
5669         (distribute_loop): Record number of library calls generated.  Adjust.
5670         (tree_loop_distribution): Report number of loops and library
5671         calls generated as opt-info.
5673 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5675         * tree-flow.h: Include new .h files.  Move prototypes.
5676         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
5677         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
5678         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
5679         * tree-pretty-print.h: Add prototypes from tree-flow.h.
5680         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
5681         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
5682         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
5683         * tree.h (get_ref_base_and_extent): Move prototype out.
5684         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
5685         tree-dfa.h.
5686         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
5687         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
5688         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
5689         * tree-scalar-evolution.c: Include tree.h.
5690         * sese.c: Include tree.h.
5691         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
5692         * dwarf2out.c: Include tree-dfa.h.
5693         * tree-chrec.c: Include tree.h.
5694         * tree-data-ref.c: Include tree.h.
5696 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
5698         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
5699         Fix whitespace.
5701 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5703         * config/t-sol2 (sol2-c.o): Remove header dependencies.
5704         Use $(COMPILE) and $(POSTCOMPILE).
5705         (sol2-cxx.o): Likewise.
5706         (sol2-stubs.o): Likewise.
5707         (sol2.o): Likewise.
5708         * config/x-solaris (host-solaris.o): Likewise.
5710         * config/sparc/t-sparc (sparc.o): Remove.
5711         (sparc-c.o): Remove header dependencies.
5712         Use $(COMPILE) and $(POSTCOMPILE).
5713         * config/sparc/x-sparc: Likewise.
5715 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
5717         * config/arc/arc-opts.h: Add 2013 to Copyright years.
5718         * config/arc/arc700.md: Likewise.
5719         * config/arc/arc-modes.def: Likewise.
5720         * config/arc/arc-simd.h: Likewise.
5721         * config/arc/t-arc-uClibc: Likewise.
5722         * config/arc/t-arc-newlib: Likewise.
5724 2013-10-02  Renlin Li  <renlin.li@arm.com>
5726         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5727         plus_constant.
5728         (aarch64_expand_epilogue): Likewise.
5730 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5731             Yufeng Zhang  <yufeng.zhang@arm.com>
5733         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
5734         declaration.
5735         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
5736         'base_in' represent a conversion and legal_cast_p_1 holds; set
5737         'base_in' with the returned value from get_unwidened.
5739 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5741         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
5742         plus_constant is not used.
5744 2013-10-01  Wei Mi  <wmi@google.com>
5746         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
5747         * config/i386/i386.md: Add define_peephole2 to
5748         break partial reg stall for cvtss2sd/cvtsd2ss.
5750 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5752         * config/arc/arc.c (pass_arc_ifcvt::clone):
5753         Update for ctxt_ -> m_ctxt change.
5755 2013-10-01  Jeff Law  <law@redhat.com>
5757         * tree-ssa-threadupdate.c (struct redirection_data): Delete
5758         outgoing_edge and intermediate_edge fields.  Instead store the path.
5759         (redirection_data::hash): Hash on the last edge's destination index.
5760         (redirection_data::equal): Check the entire thread path.
5761         (lookup_redirectio_data): Corresponding changes.
5762         (create_edge_and_update_destination_phis): Likewise.
5763         (thread_single_edge): Likewise.
5765 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5766             Diego Novillo <dnovillo@google.com>
5768         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
5769         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
5770         (vld32wh_insn, vld32wl_insn): Delete commented-out old
5771         versions of these patterns.
5773         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
5774         (__builtin_arc_aligned): Likewise.
5776         * config/arc/arc.md: Expand adc_0 comment stating the intended
5777         purpose and why it isn't ready.
5778         Replace commented out call_value_via_label_mixed with a
5779         plain comment about bl_s.
5781         * config/arc/arc.c (stdio.h): Don't include directly.
5782         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
5783         Remove if (1) condition.
5784         (arc_encode_section_info): Fix comment.
5786 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5788         * config/arc/arc.c (arc_conditional_register_usage):
5789         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
5790         Also set reg_alloc_order for DMA config regs.
5792 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5793             Jeremy Bennett  <jeremy.bennett@embecosm.com>
5795         * doc/install.texi (--with-cpu): Mention ARC.
5796         (arc-*-elf32): New paragraph.
5797         (arc-linux-uclibc): Likewise.
5798         * doc/md.texi (Machine Constraints): Add ARC part.
5799         * doc/invoke.texi: (menu): Add ARC Options.
5800         (Machine Dependent Options) <ARC Options>: Add synopsis.
5801         (node ARC Options): Add.
5802         * doc/extend.texi (long_call / short_call attribute): Add ARC.
5803         (ARC Built-in Functions): New section defining
5804         generic ARC built-in functions.
5805         (ARC SIMD Built-in Functions): New section defining SIMD specific
5806         built-in functions.
5807         (Declaring Attributes of Functions): Extended
5808         description of short_call and long_call attributes for ARC and
5809         added index entries.
5811 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
5812             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
5813             Joern Rennecke  <joern.rennecke@embecosm.com>
5814             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
5815             Brendan Kehoe  <brendan@zen.org>
5816             Michael Eager  <eager@eagercon.com>
5817             Simon Cook  <simon.cook@embecosm.com>
5818             Jeremy Bennett  <jeremy.bennett@embecosm.com>
5820         * config/arc, common/config/arc: New directories.
5822 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5823             Brendan Kehoe  <brendan@zen.org>
5824             Simon Cook  <simon.cook@embecosm.com>
5826         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
5828 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
5830         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
5831         * tree-ssa-coalesce.h: New. Move prototype to here.
5832         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
5833         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
5834         (gimple_can_coalesce_p): Move to...
5835         * gimple.c (gimple_can_coalesce_p): Here.
5837 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
5839         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
5840         (dump_decl_set): Move to gimple.c.
5841         * gimple.h: Don't include tree-ssa-operands.h.
5842         (dump_decl_set): Add prototype.
5843         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
5844         Move to gimple-ssa.h.
5845         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
5846         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
5847         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
5848         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
5849         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
5850         than PHI_ARG_DEF.
5851         (dump_decl_set): Relocate here.
5852         * gimple-ssa.h: New file.
5853         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
5854         Relocate from gimple.h.
5855         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
5856         * tree-ssa-operands.c (swap_ssa_operands): Rename from
5857         swap_tree_operands and remove non-ssa path.
5858         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
5859         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
5860         swap_ssa_operands.
5861         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
5862         vect_is_simple_reduction_1): Use swap_ssa_operands.
5863         * tree-flow.h: Move various prototypes to tree-phinodes.h.
5864         (enum need_phi_state): Move to tree-into-ssa.c.
5865         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
5866         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
5867         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
5868         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
5869         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
5870         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
5871         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
5872         num_imm_uses): Move to ssa-iterators.h.
5873         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
5874         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
5875         tree-phinodes.h.
5876         (op_iter_done, op_iter_next_def, op_iter_next_tree,
5877         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
5878         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
5879         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
5880         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
5881         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
5882         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
5883         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
5884         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
5885         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
5886         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
5887         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
5888         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
5889         (set_phi_nodes): Move to tree-phinodes.h.
5890         * tree-ssa-operands.h (enum ssa_op_iter_type,
5891         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
5892         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
5893         (dump_decl_set): Remove prototype.
5894         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
5895         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
5896         (set_phi_nodes): Relocate from tree-flow-inline.h.
5897         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
5898         tree-flow-inline.h
5899         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
5900         include list.  Temporarily add gimple.h to include list.
5901         * ssa-iterators.h: New file.
5902         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
5903         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
5904         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
5905         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
5906         Relocate from tree-ssa-operands.h.
5907         (delink_imm_use, link_imm_use_to_list, link_imm_use,
5908         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
5909         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
5910         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
5911         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
5912         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
5913         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
5914         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
5915         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
5916         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
5917         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
5918         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
5919         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
5920         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
5921         Relocate from tree-flow-inline.h.
5922         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
5924 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
5926         * aarch64-simd.md
5927         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
5928         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
5929         Insert '\t' to output template.
5930         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
5931         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
5932         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
5933         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
5935 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
5937         * doc/install.texi (Host/target specific installation notes for GCC):
5938         Put @anchor before @heading.
5939         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
5940         Use @email for email addresses.
5942 2013-10-01  Jeff Law  <law@redhat.com>
5944         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
5945         a vec.  Only delete the path if we create one without successfully
5946         registering a jump thread.
5947         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
5948         as a pointer.
5949         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
5950         (paths): New vector of jump threading paths.
5951         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
5952         (THREAD_PATH): New accessor macro for the entire thread path.
5953         (lookup_redirection_data): Get intermediate and final outgoing edge
5954         from the thread path.
5955         (create_edge_and_update_destination_phis): Copy the threading path.
5956         (ssa_fix_duplicate_block_edges): Get edges and block types from the
5957         jump threading path.
5958         (ssa_redirect_edges): Get edges and block types from the jump threading
5959         path.  Free the path vector.
5960         (thread_block): Get edges from the jump threading path.  Look at the
5961         entire path to see if we thread to a loop exit.  If we cancel a jump
5962         thread request, then free the path vector.
5963         (thread_single_edge): Get edges and block types from the jump threading
5964         path.  Free the path vector.
5965         (thread_through_loop_header): Get edges and block types from the jump
5966         threading path.  Free the path vector.
5967         (mark_threaded_blocks): Iterate over the vector of paths and store
5968         the path on the appropriate edge.  Get edges and block types from the
5969         jump threading path.
5970         (mark_threaded_blocks): Get edges and block types from the jump
5971         threading path.  Free the path vector.
5972         (thread_through_all_blocks): Use the vector of paths rather than
5973         a vector of 3-edge sets.
5974         (register_jump_thread): Accept pointer to a path vector rather
5975         than the path vector itself.  Store the path vector for later use.
5976         Simplify.
5978 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
5979             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5981         PR target/58574
5982         * config/s390/s390.c (s390_split_branches): Modify check for table
5983         jump insns.
5984         (s390_chunkify_start): Rearrange table jump insn check in order to
5985         deal with compare and branch insns correctly.
5987 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
5989         PR target/58578
5990         Revert
5991         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
5992         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
5993         define_insn_and_split.
5994         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
5995         (shiftsi3_compare): New pattern.
5996         (rrx): New pattern.
5997         * config/arm/unspecs.md (UNSPEC_RRX): New.
5999 2013-10-01  Alan Modra  <amodra@gmail.com>
6001         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
6002         casing inout operands.
6004 2013-10-01  Richard Biener  <rguenther@suse.de>
6006         PR tree-optimization/58553
6007         * tree-loop-distribution.c (struct partition_s): Add niter member.
6008         (classify_partition): Populate niter member for the partition
6009         and properly identify whether the relevant store happens before
6010         or after the loop exit.
6011         (generate_memset_builtin): Use niter member from the partition.
6012         (generate_memcpy_builtin): Likewise.
6014 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6016         * vec.h (vec_prefix, vec): Prefix member names with "m_".
6017         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
6019 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6021         * basic-block.h (edge_list): Prefix member names with "m_".
6022         * context.h (context): Likewise.
6023         * domwalk.h (dom_walker): Likewise.
6024         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
6025         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
6026         * hash-table.h (hash_table): Likewise.
6027         * machmode.h (bit_field_mode_iterator): Likewise.
6028         * pass_manager.h (pass_list): Likewise.
6029         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
6030         * tree-pass.h (pass_data): Likewise.
6031         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
6032         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
6033         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
6034         * asan.c (pass_data_asan): Update accordingly.
6035         * cfganal.c (control_dependences::find_control_dependence): Likewise.
6036         (control_dependences::control_dependences): Likewise.
6037         (control_dependences::~control_dependences): Likewise.
6038         (control_dependences::~control_dependences): Likewise.
6039         (control_dependences::get_edges_dependent_on): Likewise.
6040         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
6041         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
6042         * context.c (gcc::context::context): Likewise.
6043         * cprop.c (pass_rtl_cprop::clone): Likewise.
6044         * domwalk.c (dom_walker::walk): Likewise.
6045         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
6046         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
6047         * mode-switching.c (pass_mode_switching::clone): Likewise.
6048         * passes.c (opt_pass::opt_pass): Likewise.
6049         (pass_manager::pass_manager): Likewise.
6050         * predict.c (pass_strip_predict_hints::clone): Likewise.
6051         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
6052         (pass_split_all_insns::clone): Likewise.
6053         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6054         Likewise.
6055         (bit_field_mode_iterator::next_mode): Likewise.
6056         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
6057         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
6058         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
6059         * tree-complex.c (pass_lower_complex::clone): Likewise.
6060         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
6061         * tree-object-size.c (pass_object_sizes::clone): Likewise.
6062         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
6063         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
6064         (pass_fold_builtins::clone): Likewise.
6065         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
6066         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
6067         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
6068         (pass_cd_dce::clone): Likewise.
6069         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
6070         (pass_phi_only_cprop::clone): Likewise.
6071         * tree-ssa-dse.c (pass_dse::clone): Likewise.
6072         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
6073         * tree-ssa-loop.c (pass_lim::clone): Likewise.
6074         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
6075         * tree-ssa-pre.c (pass_fre::clone): Likewise.
6076         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
6077         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
6078         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
6079         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
6080         * tree-vrp.c (pass_vrp::clone): Likewise.
6081         * tsan.c (pass_tsan::clone): Likewise.
6083 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
6085         PR middle-end/58564
6086         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
6087         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
6089         PR middle-end/58564
6090         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
6091         optimization, punt if sign_bit_p looked through any zero extension.
6093 2013-09-30  Teresa Johnson  <tejohnson@google.com>
6095         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
6096         Update redirected out edge count in joiner case.
6097         (ssa_redirect_edges): Common the joiner and non-joiner cases
6098         so that joiner case gets profile updates.
6100 2013-09-30  Richard Biener  <rguenther@suse.de>
6102         PR tree-optimization/58554
6103         * tree-loop-distribution.c (classify_partition): Require
6104         unconditionally executed stores for memcpy and memset recognition.
6105         (tree_loop_distribution): Calculate dominance info.
6107 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
6109         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
6110         (NO_PROFILE_COUNTERS): Likewise.
6111         (PROFILE_HOOK): Likewise.
6112         (FUNCTION_PROFILER): Likewise.
6113         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
6115 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
6117         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
6118         calls and defines in TARGET_MACHO conditional.
6119         (load_macho_picbase_di): Likewise.
6120         (reload_macho_picbase): Likewise.
6121         (reload_macho_picbase_si): Likewise.
6122         (reload_macho_picbase_di): Likewise.
6123         (nonlocal_goto_receiver): Likewise.
6125 2013-09-30  Nick Clifton  <nickc@redhat.com>
6127         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
6128         that use the MSP430X ISA.
6129         (msp430_option_override): Scan -mmcu command line option for any
6130         MCU name that supports the MSP430X ISA.
6131         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
6132         -mmcu options which enable the MSP430X ISA.
6134 2013-09-30  Richard Biener  <rguenther@suse.de>
6136         PR middle-end/58532
6137         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
6138         before looking for setjmp-like calls.
6140 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6142         PR target/10901
6143         * config/darwin-protos.h (machopic_get_function_picbase): New.
6144         * config/darwin.c (machopic_get_function_picbase): New.
6145         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
6146         label for a new func.  (load_macho_picbase_di): Likewise.
6147         (reload_macho_picbase): New expand.
6148         (reload_macho_picbase_si): New insn.
6149         (reload_macho_picbase_di): New insn.
6150         (nonlocal_goto_receiver): New define and split.
6151         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
6152         (unspecv enum): Add UNSPECV_NLGR.
6154 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6156         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
6157         that altivec registers are correctly sized on Darwin.
6159 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6161         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
6162         darwin-driver.o): Use COMPILE and POSTCOMPILE.
6163         * config/x-darwin (host-darwin.o): Likewise.
6164         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
6165         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
6166         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
6168 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
6170         * doc/invoke.texi: Fix usage of @tie{} command.
6172 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
6174         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
6176 2013-09-28  Jan Hubicka  <jh@suse.cz>
6178         * config/i386/x86-tune.def: Add documentation for each of the options;
6179         add whitespace.
6181 2013-09-28  Jan Hubicka  <jh@suse.cz>
6183         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
6184         generic.
6185         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
6186         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
6187         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
6188         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
6189         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
6191 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
6193         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
6194         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
6195         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
6196         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
6197         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
6198         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
6199         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
6200         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
6201         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
6202         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
6203         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
6204         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
6205         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
6206         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
6207         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
6208         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
6209         gimple.h, godump.c, graphite-clast-to-gimple.c,
6210         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
6211         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
6212         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
6213         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
6214         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
6215         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
6216         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
6217         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
6218         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
6219         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
6220         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
6221         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
6222         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
6223         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
6224         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
6225         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
6226         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
6227         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
6228         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
6229         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
6230         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
6231         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
6232         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
6233         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
6234         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
6235         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
6236         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
6237         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
6238         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
6239         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
6240         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
6241         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
6242         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
6243         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
6244         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6245         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
6246         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
6247         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
6248         whitespace before "(".
6250 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
6252         * expr.h (extract_bit_field): Remove packedp parameter.
6253         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
6254         from forward declaration.
6255         (store_split_bit_field): Remove packedp arg from calls to
6256         extract_fixed_bit_field.
6257         (extract_bit_field_1): Remove packedp parameter and packedp
6258         argument from recursive calls and calls to extract_fixed_bit_field.
6259         (extract_bit_field): Remove packedp parameter and corresponding
6260         arg to extract_bit_field_1.
6261         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
6262         to issue warnings.
6263         (extract_split_bit_field): Remove packedp arg from call to
6264         extract_fixed_bit_field.
6265         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
6266         (copy_blkmode_from_reg): Likewise.
6267         (copy_blkmode_to_reg): Likewise.
6268         (read_complex_part): Likewise.
6269         (store_field): Likewise.
6270         (expand_expr_real_1): Likewise.
6271         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
6272         to extract_bit_field.
6273         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
6274         call to extract_bit_field.
6275         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
6276         call to extract_bit_field.
6277         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
6278         and special packedp behavior from -fstrict-volatile-bitfields
6279         documentation.
6281 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6283         * lra-eliminations.c (init_elim_table): Guard value_p.
6285 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
6287         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
6288         DFmode, DImode, and SFmode in the upper VSX registers based on the
6289         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
6290         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
6291         of the VSX register handling.
6293         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
6294         (f32_sv): Likewise.
6295         (zero_extendsidi2_lfiwzx): Add support for loading into the
6296         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
6297         only do VSX memory options on Altivec registers.
6298         (extendsidi2_lfiwax): Likewise.
6299         (extendsfdf2_fpr): Likewise.
6300         (mov<mode>_hardfloat, SF/SD modes): Likewise.
6301         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
6302         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
6303         (movdi_internal64): Likewise.
6305 2013-09-27  Xinliang David Li  <davidxl@google.com>
6307         * opts.c (finish_options): Adjust parameters
6308         according to vect cost model.
6309         (common_handle_option): Set dynamic vect cost
6310         model for FDO.
6311         targhooks.c (default_add_stmt_cost): Compute stmt cost
6312         unconditionally.
6313         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
6314         Use helper function.
6315         * tree-vectorizer.h (unlimited_cost_model): New function.
6316         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
6317         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
6318         function.
6319         (vect_enhance_data_refs_alignment): Ditto.
6320         * flag-types.h: New enum.
6321         * common/config/i386/i386-common.c (ix86_option_init_struct):
6322         No need to initialize vect_cost_model flag.
6323         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
6324         unconditionally.
6326 2013-09-27  Diego Novillo  <dnovillo@google.com>
6328         * gimple.h (enum ssa_mode): Remove.
6330 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
6332         * cfgloop.h (number_of_loops): Fix typo in check for null.
6334 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
6336         PR middle-end/58551
6337         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
6338         are children of outermost saved_cfun's loop, and set it up to
6339         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
6340         if loop != loop0 and SESE region contains bbs that belong to loop0.
6342 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6344         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
6345         (binary_scale_code_p, get_base_term, get_index_term): New functions.
6346         (set_address_segment, set_address_base, set_address_index)
6347         (set_address_disp): Accept the argument unconditionally.
6348         (baseness): Remove must_be_base_p and must_be_index_p checks.
6349         (decompose_normal_address): Classify as much as possible in the
6350         main loop.
6352 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6354         * cse.c (count_reg_usage): Handle INT_LIST.
6355         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
6356         * reginfo.c (reg_scan_mark_refs): Likewise.
6357         * reload1.c (eliminate_regs_1): Likewise.
6359 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
6361         PR middle-end/58547
6362         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
6363         signed.
6365 2013-09-27  Richard Biener  <rguenther@suse.de>
6367         PR tree-optimization/58459
6368         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
6369         restriction not propagating into loops.
6371 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
6373         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
6374         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
6375         * doc/tree-ssa.texi (Walking use-def chains): Delete.
6377 2013-09-26  Richard Biener  <rguenther@suse.de>
6379         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
6381 2013-09-26  Richard Biener  <rguenther@suse.de>
6383         * alias.h (component_uses_parent_alias_set): Rename to ...
6384         (component_uses_parent_alias_set_from): ... this.
6385         * alias.c (component_uses_parent_alias_set): Rename to ...
6386         (component_uses_parent_alias_set_from): ... this and return
6387         the desired parent.
6388         (reference_alias_ptr_type_1): Use the result from
6389         component_uses_parent_alias_set_from instead of stripping
6390         components one at a time.
6391         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
6393 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
6395         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
6396         Move prototypes to...
6397         * tree-ssa-ter.h: New File.  Move prototypes here.
6398         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
6399         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
6400         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
6401         from is_replaceable_p.
6402         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
6403         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
6404         (process_replaceable): Use ter_is_replaceable_p.
6405         (find_replaceable_in_bb): Use ter_is_replaceable_p.
6406         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
6407         newly refactored ssa_is_replaceable_p.
6408         * cfgexpand.c: Include tree-outof-ssa.h.
6409         * ssaexpand.h: Delete.
6411 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
6413         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
6414         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
6415         tree-ssa.c
6416         (create_gimple_tmp): Delete.
6417         (get_expr_type, build_assign, build_type_cast): Move to...
6418         * gimple-builder.c: New File.
6419         (get_expr_type): Relocate from gimple.c.
6420         (build_assign, build_type_cast): Change to only create ssanames.
6421         * gimple.h: Move prototypes to...
6422         * gimple-builder.h: New File. Here.
6423         * tree-ssa.h: And here.
6424         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
6425         count_uses_and_derefs): Relocate from gimple.c.
6426         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
6427         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
6428         * tree-ssa-math-opts (execute_cse_reciprocals): Use
6429         gimple_replace_ssa_lhs.
6430         * asan.c: Include gimple-builder.h.
6431         * Makefile.in: Add gimple-builder.o.
6433 2013-09-26  Richard Biener  <rguenther@suse.de>
6435         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
6436         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
6437         (loe_visit_block): Use gcc_checking_assert.
6438         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
6439         gimple_assign_ssa_name_copy_p.
6440         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
6441         change.
6443 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
6445         * config/rs6000/t-rs6000 (rs6000.o): Remove.
6446         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
6448 2013-09-26  Richard Biener  <rguenther@suse.de>
6450         PR tree-optimization/58539
6451         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
6452         the fact that debug statements are not taking part in loop-closed
6453         SSA construction.
6455 2013-09-26  Nick Clifton  <nickc@redhat.com>
6457         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
6458         time warning message.
6459         (msp430_print_operand_raw): Delete unused letter parameter.
6460         (TARGET_PRINT_OPERAND_ADDRESS): Define.
6461         (msp430_print_operand_address): New function.
6462         (msp430_print_operand): Move address printing code from here to
6463         new function.
6464         * config/msp430/msp430.md (movsipsi2): Add comment in generated
6465         assembler.
6466         (zero_extendpsisi2): Likewise.
6467         (extendpsisi2): New pattern.
6468         (andneghi3): New pattern.
6470 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
6472         * config/aarch64/aarch64.opt (mlra): New option.
6473         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
6474         (TARGET_LRA_P): Define.
6476 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
6478         * expr.c (expand_assignment): Remove obsolete comment.
6480 2013-09-25  Jeff Law  <law@redhat.com>
6482         * tree-flow.h (thread_through_all_blocks): Prototype moved into
6483         tree-ssa-threadupdate.h.
6484         (register_jump_thread): Similarly.
6485         * tree-ssa-threadupdate.h: New header file.
6486         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
6487         * tree-vrp.c: Likewise.
6488         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
6489         (thread_around_empty_blocks): Change type of path vector argument to
6490         an edge,type pair from just an edge.  Initialize both elements when
6491         appending to a jump threading path.  Tweak references to elements
6492         appropriately.
6493         (thread_across_edge): Similarly.  Release memory for the elements
6494         as needed.
6495         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
6496         (dump_jump_thread_path): New function broken out from
6497         register_jump_thread.
6498         (register_jump_thread): Use dump_jump_thread_path.  Change type of
6499         path vector entries.  Search the path for NULL edges and dump
6500         the path if one is found.  Tweak the conversion of path to 3-edge
6501         form to use the block copy type information embedded in the path.
6503 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6505         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
6507 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6508             Vladimir Makarov  <vmakarov@redhat.com>
6510         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
6511         from the least significant bit.
6512         (strip_address_mutations): Add bitfield operations handling.
6513         (must_be_index_p): Add shifting and rotate operations handling.
6514         (set_address_base): Use must_be_base_p predicate.
6515         (set_address_index): Use must_be_index_p predicate.
6517 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6518             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
6519             Sergey Lega  <sergey.s.lega@intel.com>
6520             Anna Tikhonova  <anna.tikhonova@intel.com>
6521             Ilya Tocar  <ilya.tocar@intel.com>
6522             Andrey Turetskiy  <andrey.turetskiy@intel.com>
6523             Ilya Verbin  <ilya.verbin@intel.com>
6524             Kirill Yukhin  <kirill.yukhin@intel.com>
6525             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
6527         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
6528         Use new names.
6529         (ix86_expand_vector_move_misalign): Support new unaligned load and
6530         stores and use new names.
6531         (CODE_FOR_sse2_storedqu): Rename to ...
6532         (CODE_FOR_sse2_storedquv16qi): ... this.
6533         (CODE_FOR_sse2_loaddqu): Rename to ...
6534         (CODE_FOR_sse2_loaddquv16qi): ... this.
6535         (CODE_FOR_avx_loaddqu256): Rename to ...
6536         (CODE_FOR_avx_loaddquv32qi): ... this.
6537         (CODE_FOR_avx_storedqu256): Rename to ...
6538         (CODE_FOR_avx_storedquv32qi): ... this.
6539         * config/i386/i386.md (fpint_logic): New.
6540         * config/i386/sse.md (VMOVE): Extend for AVX512.
6541         (VF): Ditto.
6542         (VF_128_256): New.
6543         (VF_512): Ditto.
6544         (VI_UNALIGNED_LOADSTORE): Ditto.
6545         (sse2_avx_avx512f): Ditto.
6546         (sse2_avx2): Extend for AVX512.
6547         (sse4_1_avx2): Ditto.
6548         (avx2_avx512f): New.
6549         (sse): Extend for AVX512.
6550         (sse2): Ditto.
6551         (sse4_1): Ditto.
6552         (avxsizesuffix): Ditto.
6553         (sseintvecmode): Ditto.
6554         (ssePSmode): Ditto.
6555         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
6556         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
6557         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6558         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
6559         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6560         (<sse2_avx_avx512f>_storedqu<mode): ... this.
6561         (<sse>_movnt<mode>): Replace constraint "x" with "v".
6562         (STORENT_MODE): Extend for AVX512.
6563         (*absneg<mode>2): Replace constraint "x" with "v".
6564         (*mul<mode>3): Ditto.
6565         (*ieee_smin<mode>3): Ditto.
6566         (*ieee_smax<mode>3): Ditto.
6567         (avx_cmp<mode>3): Replace VF with VF_128_256.
6568         (*<sse>_maskcmp<mode>3_comm): Ditto.
6569         (<sse>_maskcmp<mode>3): Ditto.
6570         (<sse>_andnot<mode>3): Extend for AVX512.
6571         (<code><mode>3, anylogic): Replace VF with VF_128_256.
6572         (<code><mode>3, fpint_logic): New.
6573         (*<code><mode>3): Extend for AVX512.
6574         (avx512flogicsuff): New.
6575         (avx512f_<logic><mode>): Ditto.
6576         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
6577         VF_128_256.
6578         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
6579         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
6580         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
6581         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
6582         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
6583         (xop_vpermil2<mode>3): Ditto.
6584         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
6585         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
6586         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
6587         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
6589 2013-09-25  Tom Tromey  <tromey@redhat.com>
6591         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
6592         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
6593         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
6594         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
6595         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
6596         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
6597         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
6598         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
6599         (GRAPHITE_HTAB_H): Remove.
6601 2013-09-25  Tom Tromey  <tromey@redhat.com>
6603         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
6605 2013-09-25  Tom Tromey  <tromey@redhat.com>
6607         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
6609 2013-09-25  Tom Tromey  <tromey@redhat.com>
6611         * config/i386/t-i386 (i386.o): Remove.
6612         (i386-c.o): Use COMPILE and POSTCOMPILE.
6614 2013-09-25  Tom Tromey  <tromey@redhat.com>
6616         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
6618 2013-09-25  Tom Tromey  <tromey@redhat.com>
6620         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
6621         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
6622         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
6623         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
6624         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
6625         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
6626         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
6627         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
6628         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
6629         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
6630         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
6631         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
6632         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
6633         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
6634         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
6635         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
6636         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
6637         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
6638         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
6639         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
6640         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
6641         (tree-ssa-pre.o, tree-ssa-sccvn.o)
6642         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
6643         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
6644         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
6645         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
6646         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
6647         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
6648         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
6649         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
6650         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
6651         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
6652         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
6653         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
6654         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
6655         (graphite-clast-to-gimple.o, graphite-dependences.o)
6656         (graphite-interchange.o, graphite-poly.o)
6657         (graphite-scop-detection.o, graphite-sese-to-poly.o)
6658         (graphite-optimize-isl.o, tree-vect-loop.o)
6659         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
6660         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
6661         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
6662         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
6663         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
6664         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
6665         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
6666         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
6667         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
6668         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
6669         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
6670         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
6671         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
6672         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
6673         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
6674         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
6675         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
6676         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
6677         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
6678         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
6679         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
6680         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
6681         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
6682         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
6683         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
6684         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
6685         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
6686         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
6687         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
6688         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
6689         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
6690         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
6691         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
6692         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
6693         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
6694         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
6695         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
6696         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
6697         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
6698         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
6699         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
6700         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
6701         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
6702         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
6703         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
6704         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
6705         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
6706         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
6707         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
6708         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
6709         (gcov-dump.o): Remove.
6710         (default-c.o): Use COMPILE and POSTCOMPILE.
6711         (CFLAGS-gcc.o): New variable.
6712         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
6714 2013-09-25  Tom Tromey  <tromey@redhat.com>
6716         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
6717         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
6718         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
6719         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
6720         (c-family/c-pragma.o, c-family/c-pretty-print.o)
6721         (c-family/c-semantics.o, c-family/c-ada-spec.o)
6722         (c-family/array-notation-common.o, c-family/stub-objc.o)
6723         (c-family/c-ubsan.o): Remove.
6725 2013-09-25  Tom Tromey  <tromey@redhat.com>
6727         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
6729 2013-09-25  Tom Tromey  <tromey@redhat.com>
6731         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
6732         to add -DENABLE_SHARED_LIBGCC.
6733         (gcc.o): Don't use subshell.
6735 2013-09-25  Tom Tromey  <tromey@redhat.com>
6737         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
6738         * configure.ac: Don't invoke AM_PROG_CC_C_O.
6739         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
6740         * configure, config.in: Rebuild.
6742 2013-09-25  Tom Tromey  <tromey@redhat.com>
6744         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
6745         (COMPILE, POSTCOMPILE): New variables.
6746         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
6747         (DEPFILES): New variable.
6748         Include ".Po" files.
6749         * configure.ac: Add checks for dependency checking.
6750         * configure, aclocal.m4: Regenerate.
6752 2013-09-25  Tom Tromey  <tromey@redhat.com>
6754         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
6755         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
6757 2013-09-25  Tom Tromey  <tromey@redhat.com>
6759         * Makefile.in (generated_files): Add options.h,
6760         target-hooks-def.h, insn-opinit.h,
6761         common/common-target-hooks-def.h, pass-instances.def,
6762         c-family/c-target-hooks-def.h.
6764 2013-09-25  Jeff Law  <law@redhat.com>
6766         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
6767         than foo[foo.length () - 1] to access last member in a vec.
6768         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
6770 2013-09-25  Richard Biener  <rguenther@suse.de>
6772         PR middle-end/58521
6773         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
6775 2013-09-25  Jan Hubicka  <jh@suse.cz>
6777         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
6778         test.
6780 2013-09-25  Marek Polacek  <polacek@redhat.com>
6782         PR sanitizer/58420
6783         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
6784         when determining the type name.
6786 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
6788         * config/sh/sh.md: Fix formatting.
6790 2013-09-24  Xinliang David Li  <davidxl@google.com>
6792         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
6793         max peel iterations parameter.
6794         * param.def: New parameter.
6795         * doc/invoke.texi: Document New parameter.
6797 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
6799         * gimple-pretty-print.c: Various whitespace tweaks.
6800         * tree-core.h: Likewise.
6801         * tree-pretty-print.c: Likewise.
6802         * tree-ssa-alias.c: Likewise.
6803         * tree-ssa-copy.c: Likewise.
6804         * tree-ssanames.c: Likewise.
6805         * tree-ssanames.h: Likewise.
6806         * tree-vrp.c: Likewise.
6808 2013-09-24  Alan Modra  <amodra@gmail.com>
6810         PR middle-end/57134
6811         PR middle-end/57586
6812         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
6813         for output operands that disallow regs.  Don't use EXPAND_WRITE on
6814         inout operands.
6816 2013-09-24  Richard Biener  <rguenther@suse.de>
6818         PR middle-end/58513
6819         * tree.c (reference_alias_ptr_type): Move ...
6820         * alias.c (reference_alias_ptr_type): ... here and implement
6821         in terms of the new reference_alias_ptr_type_1.
6822         (ref_all_alias_ptr_type_p): New helper.
6823         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
6824         use ref_all_alias_ptr_type_p.
6825         (get_deref_alias_set): Add flag_strict_aliasing check here.
6826         (reference_alias_ptr_type_1): New function, split out from ...
6827         (get_alias_set): ... here.
6828         (alias_ptr_types_compatible_p): New function.
6829         * alias.h (reference_alias_ptr_type): Declare.
6830         (alias_ptr_types_compatible_p): Likewise.
6831         * tree.h (reference_alias_ptr_type): Remove.
6832         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
6833         to compare MEM_REF alias types.
6835 2013-09-24  Richard Biener  <rguenther@suse.de>
6837         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
6839 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6841         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
6842         reload helper function arrays into a single array reg_addr.
6843         (reload_fpr_gpr): Likewise.
6844         (reload_gpr_vsx): Likewise.
6845         (reload_vsx_gpr): Likewise.
6846         (struct rs6000_reg_addr): Likewise.
6847         (reg_addr): Likewise.
6848         (rs6000_debug_reg_global): Change rs6000_vector_reload,
6849         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
6850         (rs6000_init_hard_regno_mode_ok): Likewise.
6851         (rs6000_secondary_reload_direct_move): Likewise.
6852         (rs6000_secondary_reload): Likewise.
6854         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
6855         constraints: wu, ww, and wy.  Repurpose wv constraint added during
6856         power8 changes.  Put wg constraint in alphabetical order.
6858         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
6859         for future work to add ISA 2.07 VSX single precision support.
6860         (-mvsx-scalar-double): Change default from -1 to 1, update
6861         documentation comment.
6862         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
6863         (-mupper-regs-df): New debug switch to control whether DF values
6864         can go in the traditional Altivec registers.
6865         (-mupper-regs-sf): New debug switch to control whether SF values
6866         can go in the traditional Altivec registers.
6868         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
6869         and wy constraints.
6870         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
6871         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
6872         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
6873         (rs6000_debug_legitimate_address_p): Print if we are running
6874         before, during, or after reload.
6875         (rs6000_secondary_reload): Add a comment.
6876         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
6878         * config/rs6000/constraints.md (wa constraint): Sort w<x>
6879         constraints.  Update documentation string.
6880         (wd constraint): Likewise.
6881         (wf constraint): Likewise.
6882         (wg constraint): Likewise.
6883         (wn constraint): Likewise.
6884         (ws constraint): Likewise.
6885         (wt constraint): Likewise.
6886         (wx constraint): Likewise.
6887         (wz constraint): Likewise.
6888         (wu constraint): New constraint for ISA 2.07 SFmode scalar
6889         instructions.
6890         (ww constraint): Likewise.
6891         (wy constraint): Likewise.
6892         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
6893         in the previous submissions.
6894         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
6896 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
6898         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
6899         (REG_BR_PROB): Say that the probability is stored in an int_list.
6900         * reg-notes.def: Update commentary to mention INT_LIST.
6901         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
6902         (INT_LIST): New rtx.
6903         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
6904         * rtlanal.c (int_reg_note_p): New function.
6905         (alloc_reg_note): Assert that the note does not have an int argument.
6906         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
6907         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
6908         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
6909         rather than an INSN_LIST.  Handle INT_LIST.
6910         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
6911         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
6912         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
6913         Manipulate them as ints rather than rtxes.
6914         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
6915         * regmove.c (copy_src_to_dest): Likewise.
6916         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
6918         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
6919         into the cases that need it.
6920         * config/arm/arm.c (arm_unwind_emit): Likewise.
6922         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
6923         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
6924         * loop-doloop.c (add_test, doloop_modify): Likewise.
6925         * loop-unswitch.c (compare_and_jump_seq): Likewise.
6926         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
6927         * predict.c (combine_predictions_for_insn): Likewise.
6928         * print-rtl.c (print_rtx): Handle INT_LIST.
6929         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
6930         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
6931         * config/arm/arm.c (emit_unlikely_jump): Likewise.
6932         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
6933         (ix86_split_fp_branch, predict_jump): Likewise.
6934         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
6935         * config/sh/sh.c (expand_cbranchsi4): Likewise.
6936         * config/spu/spu.c (ea_load_store_inline): Likewise.
6938         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
6939         value of a REG_BR_PROB note.
6940         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
6941         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
6942         * emit-rtl.c (try_split): Likewise.
6943         * predict.c (br_prob_note_reliable_p): Likewise.
6944         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
6945         * reorg.c (mostly_true_jump): Likewise.
6946         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
6947         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
6948         * config/i386/i386.c (ix86_print_operand): Likewise.
6949         * config/ia64/ia64.c (ia64_print_operand): Likewise.
6950         * config/mmix/mmix.c (mmix_print_operand): Likewise.
6951         * config/rs6000/rs6000.c (output_cbranch): Likewise.
6952         * config/s390/s390.c (s390_expand_tbegin): Likewise.
6953         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
6954         * config/sparc/sparc.c (output_cbranch): Likewise.
6955         * config/spu/spu.c (get_branch_target): Likewise.
6956         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
6957         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
6959 2013-09-23  Jan Hubicka  <jh@suse.cz>
6961         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
6962         for ipa-devirt.
6963         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
6964         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
6965         of external calls
6966         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
6967         (gimple_fold_call): Dump inconsistent devirtualizations; add
6968         sanity check for type based devirtualizations.
6969         * ipa-prop.c: Include ipa-utils.h
6970         (ipa_intraprocedural_devirtualization): Add sanity check.
6971         (try_make_edge_direct_virtual_call): Likewise.
6973 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
6975         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
6976         assignment statements.
6978 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
6980         * gimple-pretty-print.c (dump_ssaname_info): New function.
6981         (dump_gimple_phi): Call it.
6982         (pp_gimple_stmt_1): Likewise.
6983         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
6984         (range_info_def): Declare.
6985         * tree-pretty-print.c (pp_double_int): New function.
6986         (dump_generic_node): Call it.
6987         * tree-pretty-print.h (pp_double_int): Declare.
6988         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
6989         * tree-ssanames.h (range_info_def): New structure.
6990         (value_range_type): Move definition here.
6991         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
6992         Declare.
6993         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
6994         initialization.
6995         (set_range_info): New function.
6996         (get_range_info): Likewise.
6997         (duplicate_ssa_name_range_info): Likewise.
6998         (duplicate_ssa_name_fn): Check pointer type and call
6999         duplicate_ssa_name_range_info.
7000         * tree-ssa-copy.c (fini_copy_prop): Likewise.
7001         * tree-vrp.c (value_range_type): Remove definition, now in
7002         tree-ssanames.h.
7003         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
7004         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
7005         (SSA_NAME_RANGE_INFO): New macro.
7007 2013-09-23  Richard Biener  <rguenther@suse.de>
7009         PR tree-optimization/58464
7010         * tree-ssa-pre.c (phi_trans_lookup): Remove.
7011         (phi_trans_add): Change to add conditionally on being not present.
7012         (phi_translate_1): Remove recursion detection here.
7013         (phi_translate): Pre-seed the cache with NULL to catch
7014         recursion here in a more generic way.
7015         (bitmap_find_leader): Adjust comment.
7016         (get_representative_for): Dump value-numbers.
7017         (create_expression_by_pieces): Likewise.
7018         (insert_into_preds_of_block): Likewise.
7020 2013-09-23  Christian Bruel  <christian.bruel@st.com>
7022         PR target/58475
7023         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
7024         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
7026 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
7028         Revert r202780:
7029         2013-09-20  Renlin Li  <renlin.li@arm.com>
7031         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7032         plus_constant.
7033         (aarch64_expand_epilogue): Likewise.
7034         (aarch64_legitimize_reload_address): Likewise.
7036 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
7038         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
7039         NULL_TREE before pushing them onto the vector.  Likewise for labels.
7041 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
7043         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
7044         comparison operators when -fno-trapping-math is in effect.
7045         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
7046         comparison operators in TFmode and assert that unsupported operators
7047         cannot reach here.
7048         (ia64_print_operand): Likewise.
7050 2013-09-21  Jan Hubicka  <jh@suse.cz>
7052         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
7053         (sse_typeless_stores): Enable for core
7054         (sse_load0_by_pxor): Likewise.
7055         (four_jump_limit): Disable for core.
7056         (pad_returns): Likewise.
7057         (avoid_vector_decode): Likewise.
7058         (fuse_cmp_and_branch): Enable for cores.
7059         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
7061 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7063         PR middle-end/56791
7064         * config/pa/pa.c (pa_option_override): Disable auto increment and
7065         decrement instructions until reload is completed.
7067         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
7068         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
7069         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
7071 2013-09-20  DJ Delorie  <dj@redhat.com>
7072             Nick Clifton  <nickc@redhat.com>
7074         * config/rl78/rl78.c: Various whitespace and comment tweaks.
7075         (need_to_save): Save bank 0 on interrupts.
7076         (characterize_address): Strip far address wrappers.
7077         (rl78_as_legitimate_address): Likewise.
7078         (transcode_memory_rtx): Likewise.
7079         (rl78_peep_movhi_p): Disable this peephole after devirt.
7080         (rl78_propogate_register_origins): Forget all origins when a
7081         CLOBBER is seen.
7082         * config/rl78/rl78-virt.md: Various whitespace tweaks.
7083         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
7084         peephole2's.
7085         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
7086         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
7087         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
7088         relocating.
7089         * config/rl78/constraints.md: Various whitespace and paren tweaks.
7091 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7093         * config/pa/pa.md: In "scc" insn patterns, change output template to
7094         handle const0_rtx in reg_or_0_operand operands.
7096 2013-09-20  Martin Husemann  <martin@NetBSD.org>
7098         PR target/56875
7099         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
7100         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
7102 2013-09-20  Richard Biener  <rguenther@suse.de>
7104         PR middle-end/58484
7105         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
7106         remembering SSA name version and block index.
7107         (new_scev_info_str): Adjust.
7108         (hash_scev_info): Likewise.  Also hash the block index.
7109         (eq_scev_info): Adjust.
7110         (find_var_scev_info): Likewise.
7111         (struct instantiate_cache_entry): Remove.
7112         (struct instantiate_cache_type): Use a htab to map name, block
7113         to chrec.
7114         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7115         (get_instantiated_value_entry): Likewise.
7116         (hash_idx_scev_info, eq_idx_scev_info): New functions.
7117         (instantiate_scev_name): Adjust.
7119 2013-09-20  Jeff Law  <law@redhat.com>
7121         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
7123 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
7125         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7126         Call aarch64_simd_expand_args to update op[argc].
7128 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7130         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
7131         plugin argument.
7133 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7135         * gengtype.c (file_rules): Added rule for *.cc files.
7136         (get_output_file_with_visibility): Give fatal message when no
7137         rules found.
7139 2013-09-20  Renlin Li  <renlin.li@arm.com>
7141         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
7142         (aarch64_expand_epilogue): Likewise.
7143         (aarch64_legitimize_reload_address): Likewise.
7145 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7147         PR middle-end/57748
7148         * expr.c (expand_assignment): Remove misalignp code path.
7150 2013-09-20  Marek Polacek  <polacek@redhat.com>
7152         PR sanitizer/58413
7153         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
7154         TYPE_PRECISION.  Add asserts.
7156 2013-09-20  Richard Biener  <rguenther@suse.de>
7158         PR tree-optimization/58453
7159         * tree-loop-distribution.c (distribute_loop): Apply the cost
7160         model for -ftree-loop-distribute-patterns, too.
7162 2013-09-20  Richard Biener  <rguenther@suse.de>
7164         PR middle-end/58473
7165         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
7166         make type comparison less strict.
7168 2013-09-20  Alan Modra  <amodra@gmail.com>
7170         * configure: Regenerate.
7171         * aclocal.m4: Regenerate.
7173 2013-09-20  Marek Polacek  <polacek@redhat.com>
7175         PR other/58467
7176         * doc/extend.texi: Document that attribute used is meant to be used
7177         on variables with static storage duration.
7179 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7181         PR tree-optimization/58472
7182         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
7183         simd_lane_access set inv_p = false.
7184         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
7185         the simduid magic VAR_DECL.
7187 2013-09-19  Jan Hubicka  <jh@suse.cz>
7189         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
7191 2013-09-17  Jeff Law  <law@redhat.com>
7193         * tree-ssa-dom.c (record_temporary_equivalences): New function
7194         split out of dom_opt_dom_walker::after_dom_children.
7195         (dom_opt_dom_walker::thread_across_edge): Move common code
7196         in here from dom_opt_dom_walker::after_dom_children.
7197         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
7199 2013-09-19  Jan Hubicka  <jh@suse.cz>
7201         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
7202         (TARGET_GENERIC): Use PROCESOR_GENERIC
7203         (enum processor_type): Unify generic32 and 64.
7204         * i386.md (cpu): Likewise.
7205         * x86-tune.def (use_leave): Enable for generic32.
7206         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
7207         * athlon.md: Change generic64 to generic in all occurences.
7208         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
7209         (ix86_target_macros_internal): Likewise.
7210         * driver-i386.c (host_detect_local_cpu): Likewise.
7211         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
7212         to ..
7213         (generic_memcpy, generic_memset, generic_cost): This one.
7214         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
7215         (m_GENERIC32, m_GENERIC64): Remove.
7216         (m_GENERIC): Turn into one flag.
7217         (processor_target): Unify generic tunnings.
7218         (ix86_option_override_internal): Replace generic32/64 by generic.
7219         (ix86_issue_rate): Likewise.
7220         (ix86_adjust_cost): Likewise.
7222 2013-09-19  Jan Hubicka  <jh@suse.cz>
7224         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
7225         of speculative flag.
7227 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7229         * omp-low.c (expand_omp_sections): Always pass len - 1 to
7230         GOMP_sections_start, even if !exit_reachable.
7232 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7234         * lra-constraints.c (need_for_all_save_p): Use macro
7235         HARD_REGNO_CALL_PART_CLOBBERED.
7236         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
7237         set up pseudo conflict hard regs.
7239 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7241         PR target/58452
7242         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
7243         operands.
7245 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7247         PR rtl-optimization/58438
7248         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
7249         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
7250         from previous subpasses.
7252 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
7254         * arm.c (arm_get_frame_offsets): Validate architecture supports
7255         LDRD/STRD before accepting the tuning preference.
7256         (arm_expand_prologue): Likewise.
7257         (arm_expand_epilogue): Likewise.
7259 2013-09-18  Richard Biener  <rguenther@suse.de>
7261         PR tree-optimization/58417
7262         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
7263         have chrecs with symbols defined in the loop as operands.
7264         (chrec_fold_multiply): Likewise.
7265         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
7266         parameters before folding binary operations.
7267         (struct instantiate_cache_entry_hasher): Remove.
7268         (struct instantiate_cache_type): Use a pointer-map.
7269         (instantiate_cache_type::instantiate_cache_type): New function.
7270         (instantiate_cache_type::get): Likewise.
7271         (instantiate_cache_type::set): Likewise.
7272         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7273         (get_instantiated_value_entry): Likewise.
7274         (global_cache): New global.
7275         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
7276         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
7277         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
7278         (instantiate_scev_name): Adjust.
7279         (instantiate_scev): Construct global instead of local cache.
7280         (resolve_mixers): Likewise.
7282 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
7283             Paolo Carlini  <paolo.carlini@oracle.com>
7285         PR c++/58458
7286         * doc/implement-cxx.texi: Fix references to the C++ standards.
7288 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
7290         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
7291         * tree-vect-data-refs.c (vect_analyze_data_refs): For
7292         simd_lane_access drs, update also DR_ALIGNED_TO.
7294 2013-09-18  Marek Polacek  <polacek@redhat.com>
7296         PR sanitizer/58411
7297         * doc/extend.texi: Document no_sanitize_undefined attribute.
7298         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
7299         no_sanitize_undefined attribute.
7301 2013-09-18  Nick Clifton  <nickc@redhat.com>
7303         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
7304         (ASM_DECLARE_FUNCTION_NAME): Define.
7306 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
7308         * compare-elim.c (find_comparison_dom_walker): New class
7309         (find_comparisons_in_bb): Rename to
7310         find_comparison_dom_walker::before_dom_children
7311         (find_comparisons): Adjust
7312         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
7313         adjust.
7314         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
7315         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
7316         (init_walk_dominator_tree): Remove declaration.
7317         (fini_walk_dominator_tree): Remove declaration.
7318         * fwprop.c (single_def_use_dom_walker): New class
7319         (single_def_use_enter_block): Convert to
7320         single_def_use_dom_walker::before_dom_children.
7321         (single_def_use_leave_block): Convert to
7322         single_def_use_dom_walker::after_dom_children.
7323         (build_single_def_use_links): Adjust.
7324         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
7325         class.
7326         (find_candidates_in_block): Convert to
7327         find_candidates_dom_walker::before_dom_children.
7328         (execute_strength_reduction): Adjust.
7329         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
7330         (sese_dom_walker): New class.
7331         (sese_dom_walker::sese_dom_walker): New constructor.
7332         (sese_dom_walker::~sese_dom_walker): New destructor.
7333         (build_sese_conditions_before): Convert to
7334         sese_dom_walker::before_dom_children.
7335         (build_sese_conditions_after): Convert to
7336         sese_dom_walker::after_dom_children.
7337         (build_poly_scop): Adjust
7338         * tree-into-ssa.c (rewrite_dom_walker): New class
7339         (rewrite_enter_block): Convert to
7340         rewrite_dom_walker::before_dom_children.
7341         (rewrite_leave_block): Convert to
7342         rewrite_dom_walker::after_dom_children.
7343         (rewrite_update_dom_walker): New class.
7344         (rewrite_update_enter_block): Convert to
7345         rewrite_update_dom_walker::before_dom_children.
7346         (rewrite_update_leave_block): Convert to
7347         rewrite_update_dom_walker::after_dom_children.
7348         (rewrite_blocks, rewrite_into_ssa): Adjust.
7349         (mark_def_dom_walker): New class.
7350         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
7351         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
7352         (mark_def_sites_blocks): Convert to
7353         mark_def_dom_walker::before_dom_children.
7354         (mark_def_site_blocks): Remove.
7355         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
7356         (tree_ssa_dominator_optimize): Adjust.
7357         (dom_thread_across_edge): Convert to method
7358         dom_opt_dom_walker::thread_across_edge.
7359         (dom_opt_enter_block): Convert to member function
7360         dom_opt_dom_walker::before_dom_children.
7361         (dom_opt_leave_block): Convert to member function
7362         dom_opt_dom_walker::after_dom_children.
7363         * tree-ssa-dse.c (dse_dom_walker): New class.
7364         (dse_enter_block): Convert to member function
7365         dse_dom_walker::before_dom_children.
7366         (tree_ssa_dse): Adjust.
7367         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
7368         (determine_invariantness_stmt): Convert to method
7369         invariantness_dom_walker::before_dom_children.
7370         (determine_invariantness): Remove
7371         (move_computations_dom_walker): New class.
7372         (move_computations_stmt): Convert to method
7373         move_computations_dom_walker::before_dom_children.
7374         (move_computations, tree_ssa_lim): Adjust.
7375         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
7376         (nt_init_block): Convert to method
7377         notrappping_dom_walker::before_dom_children.
7378         (nt_fini_block): Convert to method
7379         method nontrapping_dom_walker::after_dom_children.
7380         (get_non_trapping): Adjust.
7381         * tree-ssa-pre.c (eliminate_dom_walker): New class.
7382         (eliminate_bb): Convert to method
7383         eliminate_dom_walker::before_dom_children.
7384         (eliminate_leave_block): Convert to method
7385         eliminate_dom_walker::after_dom_children.
7386         (eliminate): Adjust.
7387         * tree-ssa-strlen.c (strlen_dom_walker): New class.
7388         (strlen_enter_block): Convert to method
7389         strlen_dom_walker::before_dom_children.
7390         (strlen_leave_block): Convert to method
7391         method strlen_dom_walker::after_dom_children.
7392         (tree_ssa_strlen): Adjust.
7393         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
7394         (tree_ssa_uncprop): Adjust.
7395         (uncprop_leave_block): Convert to method
7396         uncprop_dom_walker::after_dom_children.
7397         (uncprop_leave_block): Convert to method
7398         uncprop_dom_walker::before_dom_children.
7400 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
7402         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
7403         branch in same basic block.  Check both src and dest of the move insn.
7405 2013-09-17  Nick Clifton  <nickc@redhat.com>
7407         * config/rl78/rl78-real.md (bf): New pattern.
7408         (bt): New pattern.
7409         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
7410         (rl78_print_operand): Do not put a # before a %B.
7411         * config/rl78/rl78.opt: Tweak doc strings.
7413 2013-09-17  DJ Delorie  <dj@redhat.com>
7415         * config/rl78/constraints.md (Wcv): Allow up to $r31.
7416         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
7417         (rl78_option_override): Likewise, if -mallregs.
7418         (is_virtual_register): Likewise.
7419         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
7420         (REGNO_OK_FOR_BASE_P): Likewise.
7421         * config/rl78/rl78.opt (-mallregs): New.
7423 2013-09-17  Nick Clifton  <nickc@redhat.com>
7425         * config/rl78/rl78.c (need_to_save): Change return type to bool.
7426         For interrupt functions: save all call clobbered registers if the
7427         interrupt handler is not a leaf function.
7428         (rl78_expand_prologue): Always recompute the frame information.
7429         For interrupt functions: only select bank 0 if one of the bank 0
7430         registers is going to be psuhed.
7432 2013-09-17  DJ Delorie  <dj@redhat.com>
7434         * config/rl78/constraints.md: For each W* constraint, rename to C*
7435         and create a W* constraint that checks for an optional ES: prefix
7436         pattern also.
7437         * config/rl78/rl78.md (UNS_ES_ADDR): New.
7438         (es_addr): New.  Used to wrap far addresses.
7439         * config/rl78/rl78-protos.h (rl78_es_addr): New.
7440         (rl78_es_base): New.
7441         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
7442         wrapped far addresses.
7443         (rl78_print_operand_1): Unwrap far addresses before processing.
7444         (rl78_lo16): Wrap far addresses in unspecs.
7445         (rl78_es_addr): New.
7446         (rl78_es_base): New.
7447         (insn_ok_now): Check for not-yet-wrapped far addresses.
7448         (transcode_memory_rtx): Properly re-wrap far addresses.
7450 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7452         * config/sparc/t-rtems: Add leon3 multilibs.
7454 2013-09-17  Cong Hou  <congh@google.com>
7456         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
7457         when checking the dot production pattern. The type of rhs operand
7458         of multiply is now checked correctly.
7460 2013-09-17  Jeff Law  <law@redhat.com>
7462         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7463         edge implied equivalences into successor phis.
7464         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
7465         here from tree-ssa-threadedge.c.
7466         (mark_threaded_blocks): When threading through a joiner, if both
7467         successors of the joiner's clone reach the same block, verify the
7468         PHI arguments are equal.  If not, cancel the jump threading request.
7469         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
7470         tree-ssa-threadupdate.c
7471         (thread_across_edge): Don't check PHI argument equality when
7472         threading through joiner block here.
7474 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
7476         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
7477         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
7478         (warn_uninit, warn_uninitialized_vars,
7479         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7480         Move to tree-ssa-uninit.c.
7481         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
7482         (has_undefined_value_p): New.  Pass dependant parts of
7483         ssa_undefined_value_p.
7484         (uninit_undefined_value_p): Use has_undefined_value_p.
7485         (warn_uninit, warn_uninitialized_vars,
7486         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7487         Move from tree-ssa.c.
7488         * tree-ssa.h: Adjust prototypes.
7490 2013-09-17  Jan Hubicka  <jh@suse.cz>
7492         PR middle-end/58332
7493         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
7494         * ipa-inline.c (can_inline_edge_p): Do not downgrade
7495         FUNCTION_NOT_OPTIMIZED.
7496         * ipa-inline-analysis.c (compute_inline_parameters): Function
7497         not optimized is not inlinable unless it is alwaysinline.
7498         (inline_analyze_function): Force calls in not optimized
7499         function not inlinable.
7501 2013-09-17  Jan Hubicka  <jh@suse.cz>
7503         PR middle-end/58329
7504         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
7505         to return NULL.
7506         * ipa.c (function_and_variable_visibility): Likewise.
7507         * ipa-profile.c (ipa_profile): Likewise.
7509 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7511         PR ipa/58398
7512         * cgraph.c (cgraph_function_body_availability): Check for ifunc
7513         attribute, and don't inline the resolver in this case.
7515 2013-09-17  Teresa Johnson  <tejohnson@google.com>
7517         * coverage.c (get_coverage_counts): Add missing newline.
7519 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7521         PR tree-optimization/58088
7522         * fold-const.c (mask_with_trailing_zeros): New function.
7523         (fold_binary_loc): Make sure we don't recurse infinitely
7524         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
7525         Use mask_with_trailing_zeros where appropriate.
7527 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
7529         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
7530         of 'prev' var to get better distance estimation.
7532 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
7534         * tree-inline.h (struct copy_body_data): Add transform_parameter.
7535         * tree-inline.c (is_parameter_of): New predicate.
7536         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
7537         a parameter has been remapped.
7538         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
7539         (optimize_inline_calls): Initialize transform_parameter.
7540         (copy_gimple_seq_and_replace_locals): Likewise.
7541         (tree_function_versioning): Likewise.
7542         (maybe_inline_call_in_expr): Likewise.
7544 2013-09-17  Nick Clifton  <nickc@redhat.com>
7546         * config/msp430/msp430-protos.h: Add prototypes for new functions.
7547         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
7548         interrupt handlers.
7549         (is_attr_func): New function.
7550         (msp430_is_interrupt_func): New function.
7551         (is_naked_func): New function.
7552         (is_reentrant_func): New function.
7553         (is_critical_func): New function.
7554         (msp430_start_function): Add annotations for function attributes.
7555         (msp430_attr): New function.
7556         (msp430_attribute_table): New.
7557         (msp430_function_section): New function.
7558         (TARGET_ASM_FUNCTION_SECTION): Define.
7559         (msp430_builtin): New enum.
7560         (msp430_init_builtins): New function.
7561         (msp430_builtin_devl): New function.
7562         (msp430_expand_builtin): New function.
7563         (TARGET_INIT_BUILTINS): Define.
7564         (TARGET_EXPAND_BUILTINS): Define.
7565         (TARGET_BUILTIN_DECL): Define.
7566         (msp430_expand_prologue): Add support for naked, interrupt,
7567         critical and reentrant functions.
7568         (msp430_expand_epilogue): Likewise.
7569         (msp430_print_operand): Handle 'O' character.
7570         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
7571         NO_TRAMPOLINES.
7572         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
7573         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
7574         (pushm): Use a 'n' rather than an 'i' constraint.
7575         (msp_return): Add generation of the interrupt return instruction.
7576         (disable_interrupts): New pattern.
7577         (enable_interrupts): New pattern.
7578         (push_intr_state): New pattern.
7579         (pop_intr_state): New pattern.
7580         (bic_SR): New pattern.
7581         (bis_SR): New pattern.
7582         * doc/extend.texi: Document MSP430 function attributes and builtin
7583         functions.
7585 2013-09-17  Richard Biener  <rguenther@suse.de>
7587         PR tree-optimization/58432
7588         * tree-loop-distribution.c (tree_loop_distribution): Also
7589         scan PHIs for outside loop uses and seed a partition from them.
7591 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
7593         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
7594         (restructure_reference): Call backtrace_base_for_ref.
7596 2013-09-17  Alan Modra  <amodra@gmail.com>
7598         PR target/57589
7599         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
7600         patch.
7602 2013-09-16  DJ Delorie  <dj@redhat.com>
7604         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
7605         vregs location for RL78/G10.
7606         (rl78_expand_prologue): Avoid SEL on G10.
7607         (rl78_expand_epilogue): Likewise.
7608         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
7609         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
7610         __RL78_G10__ when appropriate.
7611         (ASM_SPEC): Pass -mg10 along to the assembler.
7612         * config/rl78/rl78.md (sel_rb): Disable for G10.
7613         * config/rl78/rl78.opt: Add -mg10 option.
7614         * config/rl78/t-rl78: Add -mg10 multilib.
7616 2013-09-16  Xinliang David Li  <davidxl@google.com>
7618         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
7619         * omp-low.c (omp_max_vf): Ditto.
7620         (expand_omp_simd): Ditto.
7621         * tree-vectorizer.c (vectorize_loops): Ditto.
7622         (gate_vect_slp): Ditto.
7623         (gate_increase_alignment): Ditto.
7624         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
7625         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
7626         (gate_tree_vectorize): Name change.
7627         (tree_vectorize): Ditto.
7628         (pass_vectorize::gate): Call new function.
7629         (pass_vectorize::execute): Ditto.
7630         * opts.c: O3 default setting change.
7631         (finish_options): Check new flag.
7632         * doc/invoke.texi: Document new flags.
7633         * common.opt: New flags.
7635 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
7637         * doc/tm.texi.in (Cond Exec Macros): Remove node.
7638         (Condition Code): Don't reference it.
7639         * doc/tm.texi: Regenerate.
7641 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
7643         PR middle-end/58418
7644         * lra-constraints.c (undo_optional_reloads): Consider all optional
7645         reload even if it did not get a hard reg.
7647 2013-09-16  Teresa Johnson  <tejohnson@google.com>
7649         * dumpfile.c (dump_loc): Remove newline emission.
7650         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
7651         emission to dump_printf_loc calls where missing.
7652         (vect_mark_for_runtime_alias_test): Ditto.
7653         (vect_analyze_data_ref_dependence): Ditto.
7654         (vect_analyze_data_ref_dependences): Ditto.
7655         (vect_slp_analyze_data_ref_dependence): Ditto.
7656         (vect_slp_analyze_data_ref_dependences): Ditto.
7657         (vect_compute_data_ref_alignment): Ditto.
7658         (vect_update_misalignment_for_peel): Ditto.
7659         (vect_verify_datarefs_alignment): Ditto.
7660         (vector_alignment_reachable_p): Ditto.
7661         (vect_get_data_access_cost): Ditto.
7662         (vect_enhance_data_refs_alignment): Ditto.
7663         (vect_find_same_alignment_drs): Ditto.
7664         (vect_analyze_data_refs_alignment): Ditto.
7665         (vect_analyze_group_access): Ditto.
7666         (vect_analyze_data_ref_access): Ditto.
7667         (vect_analyze_data_ref_accesses): Ditto.
7668         (vect_prune_runtime_alias_test_list): Ditto.
7669         (vect_analyze_data_refs): Ditto.
7670         (vect_create_addr_base_for_vector_ref): Ditto.
7671         (vect_create_data_ref_ptr): Ditto.
7672         (vect_grouped_store_supported): Ditto.
7673         (vect_grouped_load_supported): Ditto.
7674         * value-prof.c (check_counter): Ditto.
7675         (check_ic_target): Ditto.
7676         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
7677         (vect_recog_widen_mult_pattern): Ditto.
7678         (vect_recog_widen_sum_pattern): Ditto.
7679         (vect_recog_over_widening_pattern): Ditto.
7680         (vect_recog_widen_shift_pattern): Ditto.
7681         (vect_recog_rotate_pattern): Ditto.
7682         (vect_recog_vector_vector_shift_pattern): Ditto.
7683         (vect_recog_divmod_pattern): Ditto.
7684         (vect_recog_mixed_size_cond_pattern): Ditto.
7685         (vect_recog_bool_pattern): Ditto.
7686         (vect_pattern_recog_1): Ditto.
7687         (vect_pattern_recog): Ditto.
7688         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
7689         (vect_is_simple_iv_evolution): Ditto.
7690         (vect_analyze_scalar_cycles_1): Ditto.
7691         (vect_get_loop_niters): Ditto.
7692         (vect_analyze_loop_1): Ditto.
7693         (vect_analyze_loop_form): Ditto.
7694         (vect_analyze_loop_operations): Ditto.
7695         (vect_analyze_loop_2): Ditto.
7696         (vect_analyze_loop): Ditto.
7697         (report_vect_op): Ditto.
7698         (vect_is_slp_reduction): Ditto.
7699         (vect_is_simple_reduction_1): Ditto.
7700         (vect_get_known_peeling_cost): Ditto.
7701         (vect_estimate_min_profitable_iters): Ditto.
7702         (vect_model_reduction_cost): Ditto.
7703         (vect_model_induction_cost): Ditto.
7704         (get_initial_def_for_induction): Ditto.
7705         (vect_create_epilog_for_reduction): Ditto.
7706         (vectorizable_reduction): Ditto.
7707         (vectorizable_induction): Ditto.
7708         (vectorizable_live_operation): Ditto.
7709         (vect_loop_kill_debug_uses): Ditto.
7710         (vect_transform_loop): Ditto.
7711         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
7712         (vect_stmt_relevant_p): Ditto.
7713         (process_use): Ditto.
7714         (vect_mark_stmts_to_be_vectorized): Ditto.
7715         (vect_model_simple_cost): Ditto.
7716         (vect_model_promotion_demotion_cost): Ditto.
7717         (vect_model_store_cost): Ditto.
7718         (vect_get_store_cost): Ditto.
7719         (vect_model_load_cost): Ditto.
7720         (vect_get_load_cost): Ditto.
7721         (vect_init_vector_1): Ditto.
7722         (vect_get_vec_def_for_operand): Ditto.
7723         (vect_finish_stmt_generation): Ditto.
7724         (vectorizable_call): Ditto.
7725         (vectorizable_conversion): Ditto.
7726         (vectorizable_assignment): Ditto.
7727         (vectorizable_shift): Ditto.
7728         (vectorizable_operation): Ditto.
7729         (vectorizable_store): Ditto.
7730         (vectorizable_load): Ditto.
7731         (vectorizable_condition): Ditto.
7732         (vect_analyze_stmt): Ditto.
7733         (vect_transform_stmt): Ditto.
7734         (vect_is_simple_use): Ditto.
7735         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
7736         (vect_can_advance_ivs_p): Ditto.
7737         (vect_update_ivs_after_vectorizer): Ditto.
7738         (vect_do_peeling_for_loop_bound): Ditto.
7739         (vect_gen_niters_for_prolog_loop): Ditto.
7740         (vect_update_inits_of_drs): Ditto.
7741         (vect_create_cond_for_alias_checks): Ditto.
7742         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
7743         (vect_build_slp_tree_1): Ditto.
7744         (vect_supported_load_permutation_p): Ditto.
7745         (vect_analyze_slp_instance): Ditto.
7746         (vect_analyze_slp): Ditto.
7747         (vect_make_slp_decision): Ditto.
7748         (vect_detect_hybrid_slp): Ditto.
7749         (vect_bb_vectorization_profitable_p): Ditto.
7750         (vect_slp_analyze_bb_1): Ditto.
7751         (vect_update_slp_costs_according_to_vf): Ditto.
7752         (vect_get_mask_element): Ditto.
7753         (vect_transform_slp_perm_load): Ditto.
7754         (vect_schedule_slp_instance): Ditto.
7755         (vect_schedule_slp): Ditto.
7756         (vect_slp_transform_bb): Ditto.
7757         * profile.c (read_profile_edge_counts): Ditto.
7758         (compute_branch_probabilities): Ditto.
7759         * coverage.c (get_coverage_counts): Ditto.
7761 2013-09-16  Diego Novillo  <dnovillo@google.com>
7763         * tree-core.h: Add missing comment lines from refactoring of tree.h.
7765 2013-09-16  Jan Hubicka  <jh@suse.cz>
7767         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
7768         abstract functions; for static functions check the presence of body.
7770 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7772         * config/aarch64/aarch64-simd-builtins.def (fma): New.
7773         * config/aarch64/aarch64-simd.md
7774         (aarch64_mla_elt<mode>): New.
7775         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
7776         (aarch64_mls_elt<mode>): Likewise.
7777         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
7778         (aarch64_fma4_elt<mode>): Likewise.
7779         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
7780         (aarch64_fma4_elt_to_128v2df): Likewise.
7781         (aarch64_fma4_elt_to_64df): Likewise.
7782         (fnma<mode>4): Likewise.
7783         (aarch64_fnma4_elt<mode>): Likewise.
7784         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7785         (aarch64_fnma4_elt_to_128v2df): Likewise.
7786         (aarch64_fnma4_elt_to_64df): Likewise.
7787         * config/aarch64/iterators.md (VDQSF): New.
7788         * config/aarch64/arm_neon.h
7789         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
7790         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
7792 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7794         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
7795         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7796         (aarch64_mul3_elt_to_128df): Likewise.
7797         (aarch64_mul3_elt_to_64v2df): Likewise.
7798         * config/aarch64/iterators.md (VEL): Also handle DFmode.
7799         (VMUL): New.
7800         (VMUL_CHANGE_NLANES) Likewise.
7801         (h_con): Likewise.
7802         (f): Likewise.
7803         * config/aarch64/arm_neon.h
7804         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
7806 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
7808         * config/aarch64/arm_neon.h
7809         (vcvtx_high_f32_f64): Fix parameters.
7811 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7812             Uros Bizjak  <ubizjak@gmail.com>
7814         * config/alpha.c: Include tree-ssanames.h.
7816 2013-09-16  Richard Biener  <rguenther@suse.de>
7818         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
7819         (dot_rdg_1): Handle control_dd.
7820         (create_edge_for_control_dependence): New function.
7821         (create_rdg_edges): Add control dependences if asked for.
7822         (build_rdg): Likewise.
7823         (generate_loops_for_partition): If there are not necessary
7824         control stmts remove all their dependencies.
7825         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
7826         (distribute_loop): Pass on control dependences.
7827         (tree_loop_distribution): Compute control dependences and remove
7828         restriction on number of loop nodes.
7830 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
7832         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
7833         for internal calls.
7835 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
7837         * cse.c (try_const_anchors): Punt on CC modes.
7839 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7841         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
7843 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
7845         PR target/58382
7846         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
7847         calls to word_mode.
7849 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
7851         PR target/48094
7852         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
7853         seen.
7854         (darwin_objc1_section): Likewise.
7855         (darwin_file_end): Emit Image Info section when required.
7857 2013-09-14  Jan Hubicka  <jh@suse.cz>
7859         * tree-into-ssa.c (gate_into_ssa): New.
7860         (pass_data_build_ssa): Use it.
7861         * cgraph.h (expand_thunk): Update prototype.
7862         * cgraphunit.c (analyze_function): Expand thunks early.
7863         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
7864         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
7865         set lowered flag; do not add new function.
7866         (assemble_thunks_and_aliases): Update.
7867         * tree-ssa.c (gate_init_datastructures): New gate.
7868         (pass_data_init_datastructures): Use it.
7870 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
7872         PR target/58269
7873         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
7874         xmm register set described in the psABI.
7876 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
7878         * dwarf2out.c (should_emit_struct_debug): Add check
7879         for type_decl variable is not NULL.
7881 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
7883         * config.gcc: Use new winnt-c.c target hooks
7884         * config/t-winnt: New file
7885         * config/winnt-c.c: New file
7886         * doc/tm.texi.in: Document new hook
7887         * doc/tm.texi: Regenerated
7889 2013-09-13  Jan Hubicka  <jh@suse.cz>
7891         PR middle-end/58094
7892         * ipa-inline.c (check_callers): New function.
7893         (check_caller_edge): Remove.
7894         (want_inline_function_to_all_callers_p): Also permit alises that are
7895         called dirrectly.
7896         (inline_to_all_callers): Terminate the walk when devirtualization
7897         introduce new calls.
7899 2013-09-13  Jan Hubicka  <jh@suse.cz>
7901         * ipa-inline-analysis.c (struct growth_data): Add node.
7902         (do_estimate_growth_1): Fix detection of recursion.
7904 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
7906         PR tree-optimization/58392
7907         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
7908         to avoid shadowing of outer loop variable.  If
7909         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
7910         replace_by_duplicate_decl simduid of loops that have it set and
7911         set dest_cfun->has_simduid_loops and/or
7912         dest_cfun->has_force_vect_loops.
7913         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
7914         instead of maybe_lookup_decl.
7915         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
7916         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
7917         call.  Copy over force_vect and copy and remap simduid.  Set
7918         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
7919         (copy_cfg_body): Remove blocks_to_copy argument.  Use
7920         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
7921         caller.  Don't set cfun->has_simduid_loops and/or
7922         cfun->has_force_vect_loops here.
7923         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
7924         caller.
7925         (expand_call_inline, tree_function_versioning): Adjust copy_body
7926         callers.
7928 2013-09-13  Martin Jambor  <mjambor@suse.cz>
7930         PR bootstrap/58388
7931         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
7932         the assert if the edge was a speculative one.
7934 2013-09-13  Richard Biener  <rguenther@suse.de>
7936         * tree-data-ref.h (known_dependences_p): Move here ...
7937         * tree-loop-distribution.c (known_dependences_p): ... from here.
7938         (dump_rdg_component, debug_rdg_component): Remove.
7939         (dump_rdg): Adjust.
7940         (generate_loops_for_partition): Use gimple_uid instead of
7941         relying on matching stmt visit order.
7942         (rdg_build_partitions): Take starting stmt vector.
7943         (ldist_gen): Merge into ...
7944         (distribute_loop): ... this function.  Do not compute starting
7945         vertices vector.
7946         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
7948 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7950         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
7951         Set type attribute correctly. Set predicable_short_it attribute.
7952         (cmpsi_shiftsi): Remove %? from output template.
7954 2013-09-13  Richard Biener  <rguenther@suse.de>
7956         * tree-loop-distribution.c (struct rdg_component,
7957         rdg_defs_used_in_other_loops_p, free_rdg_components,
7958         rdg_build_components): Remove.
7959         (stmts_from_loop): Do not record virtual PHIs.
7960         (generate_loops_for_partition): Skip virtual PHIs.
7961         (build_rdg_partition_for_component): Rename to ...
7962         (build_rdg_partition_for_vertex): ... this and adjust.
7963         (rdg_build_partitions): Take a vector of starting vertices
7964         instead of components.  Remove unnecessary leftover handling.
7965         (ldist_gen): Do not build components or record other stores.
7966         (distribute_loop): Do not distribute loops containing stmts
7967         with side-effects.
7969 2013-09-13  Christian Bruel  <christian.bruel@st.com>
7971         PR target/58314
7972         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
7974 2013-09-13  Kai Tietz  <ktietz@redhat.com>
7976         * config.gcc: Separate cases for mingw and cygwin targets,
7977         and add 64-bit cygwin target case.
7979         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
7980         dll-export inline-functions.
7981         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
7984 2013-09-13  Jeff Law  <law@redhat.com>
7986         PR middle-end/58387
7987         Revert:
7988         2013-09-06  Jeff Law  <law@redhat.com>
7990         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7991         edge implied equivalences into successor phis.
7993 2013-09-12  DJ Delorie  <dj@redhat.com>
7995         * config/rl78/rl78-virt.md: Change from | to \; for asm line
7996         separators.
7998 2013-09-12  Brooks Moses  <bmoses@google.com>
8000         PR driver/42955
8001         * Makefile.in: Do not install driver binaries in $(target)/bin.
8003 2013-09-12  DJ Delorie  <dj@redhat.com>
8005         * config/rl78/rl78.opt (mrelax): New.
8006         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
8007         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
8009         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
8010         between SP and FP.
8011         (rl78_expand_epilogue): Likewise.
8013 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
8015         PR middle-end/58335
8016         * lra-eliminations.c (remove_reg_equal_offset_note): New.
8017         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
8018         pointer elimination with using remove_reg_equal_offset_note.
8020 2013-09-12  DJ Delorie  <dj@redhat.com>
8022         * config/msp430/: New port.
8023         * config.gcc (msp430): Added.
8024         * doc/invoke.texi: Document MSP430 options.
8025         * doc/install.texi: Document msp430-elf
8026         * doc/md.texi: Document msp430-elf
8027         * doc/contrib.texi: Document msp430-elf
8029         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
8030         PSImode.
8032 2013-09-12  Martin Jambor  <mjambor@suse.cz>
8034         PR ipa/58389
8035         * ipa-prop.c (remove_described_reference): Give up if the edge in the
8036         reference descriptor is NULL.
8037         (ipa_edge_removal_hook): If owning a reference descriptor, set its
8038         edge to NULL.
8040 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8042         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
8043         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
8044         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
8045         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
8046         make_temp_ssa_name): move to tree-ssanames.h
8047         * tree-ssa-alias.h: Move prototype.
8048         * tree-ssa.h: Include tree-ssanames.h.
8049         * tree-ssanames.c (FREE_SSANAMES): Move to here.
8050         * tree-ssanames.h: New.  Move items from tree-flow*.h
8051         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
8053 2013-09-12  Richard Biener  <rguenther@suse.de>
8055         PR tree-optimization/58404
8056         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
8057         propagate non-invariant addresses into dereferences wrapped
8058         in component references.
8060 2013-09-12  Richard Biener  <rguenther@suse.de>
8062         PR tree-optimization/58402
8063         * passes.def: Move pass_late_warn_uninitialized later.
8065 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8067         * tree-ssa.h: New.  Move content from tree-flow.h and
8068         tree-flow-inline.h.
8069         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
8070         Move prototypes belonging to tree-ssa.c.
8071         * tree-flow-inline.h (redirect_edge_var_map_def,
8072         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
8073         tree-ssa.h.
8074         * gimple.h: Adjust prototypes.
8075         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
8076         to...
8077         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
8078         * tree.h: Move prototype to tree-ssa.h.
8079         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
8080         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
8081         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
8082         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
8083         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
8084         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
8085         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
8086         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
8087         graphite-blocking.c, graphite-clast-to-gimple.c,
8088         graphite-dependences.c, graphite-interchange.c,
8089         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
8090         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
8091         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
8092         ipa-reference.c, ipa-split.c, ipa-utils.c,
8093         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
8094         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
8095         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
8096         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
8097         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
8098         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
8099         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
8100         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
8101         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
8102         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
8103         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
8104         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
8105         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
8106         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
8107         value-prof.c, var-tracking.c,
8108         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
8110 2013-09-12  Richard Biener  <rguenther@suse.de>
8112         PR tree-optimization/58396
8113         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
8114         (build_rdg): Take a loop-nest parameter, fix memleaks.
8115         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
8117 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8119         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
8120         for SLM.
8122 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
8124         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
8125         The FMA instruction names should have a 'v' prefix.
8127 2013-09-12  Richard Biener  <rguenther@suse.de>
8129         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
8130         (dot_rdg): Use popen instead of system in optional code.
8131         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
8132         (already_processed_vertex_p): Adjust.
8133         (has_anti_or_output_dependence, predecessor_has_mem_write,
8134         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
8135         rdg_flag_uses): Remove.
8136         (rdg_flag_vertex): Simplify.
8137         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
8138         remove recursion.
8139         (build_rdg_partition_for_component): Process the first vertex
8140         of a component only.
8141         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
8143 2013-09-12  Alan Modra  <amodra@gmail.com>
8145         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
8147 2013-09-11  DJ Delorie  <dj@redhat.com>
8148             Nick Clifton  <nickc@redhat.com>
8150         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
8151         (rl78_stack_based_mem): New.
8152         * config/rl78/constraints.md (Iv08): New.
8153         (Iv16): New.
8154         (Iv24): New.
8155         (Is09): New.
8156         (Is17): New.
8157         (Is25): New.
8158         (ISsi): New.
8159         (IShi): New.
8160         (ISqi): New.
8161         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
8162         (movhi): Likewise.
8163         (movsi): Change from expand to insn-and-split.
8164         (ashrsi3): Clobber AX.
8165         (lshrsi3): New.
8166         (ashlsi3): New.
8167         (cbranchsi4): New.
8168         * config/rl78/rl78.md (CC_REG): Fix.
8169         (addsi3): Allow memory and immediate operands.
8170         (addsi3_internal): Split into...
8171         (addsi3_internal_virt): ...new, and ...
8172         (addsi3_internal_real): ...new.
8173         (subsi): New.
8174         (subsi3_internal_virt): New.
8175         (subsi3_internal_real): New.
8176         (mulsi3): Add memory operand.
8177         (mulsi3_rl78): Likewise.
8178         (mulsi3_g13): Likewise.
8179         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
8180         (cbranchqi4_real): Add more constraint options.
8181         (cbranchhi4_real): Expand pattern.
8182         (cbranchhi4_real_signed): New.
8183         (cbranchhi4_real_inverted): New.
8184         (cbranchsi4_real_lt): New.
8185         (cbranchsi4_real_ge): New.
8186         (cbranchsi4_real_signed): New.
8187         (cbranchsi4_real): New.
8188         (peephole2): New.
8189         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
8190         constant shifts.
8191         (lshrsi3_virt): Likewise.
8192         (ashlsi3_virt): Likewise.
8193         (cbranchqi4_virt_signed): New.
8194         (cbranchhi4_virt_signed): New.
8195         (cbranchsi4_virt): New.
8196         * config/rl78/rl78.c: Whitespace fixes throughout.
8197         (move_elim_pass): New.
8198         (pass_data_rl78_move_elim): New.
8199         (pass_rl78_move_elim): New.
8200         (make_pass_rl78_move_elim): New.
8201         (rl78_devirt_info): Run devirt earlier.
8202         (rl78_move_elim_info): New.
8203         (rl78_asm_file_start): Register it.
8204         (rl78_split_movsi): New.
8205         (rl78_as_legitimate_address): Allow virtual base registers when
8206         appropriate.
8207         (rl78_addr_space_convert): Remove spurious debug stuff.
8208         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
8209         (rl78_print_operand): More cases for not printing '#'.
8210         (rl78_expand_compare): Remove most of the logic.
8211         (content_memory): New.
8212         (clear_content_memory): New.
8213         (get_content_index): New.
8214         (get_content_name): New.
8215         (display_content_memory): New.
8216         (update_content): New.
8217         (record_content): New.
8218         (already_contains): New.
8219         (insn_ok_now): Re-recog insns with virtual registers.
8220         (add_postponed_content_update): New.
8221         (process_postponed_content_update): New.
8222         (gen_and_emit_move): New.
8223         (transcode_memory_rtx): Record new location content.
8224         Use gen_and_emit_move.
8225         (force_into_acc): New.
8226         (move_to_acc): Use gen_and_emit_move.
8227         (move_from_acc): Likewise.
8228         (move_acc_to_reg): Likewise.
8229         (move_to_x): Likewise.
8230         (move_to_hl): Likewise.
8231         (move_to_de): Likewise.
8232         (rl78_alloc_physical_registers_op1): Record location content.
8233         (has_constraint): New.
8234         (rl78_alloc_physical_registers_op2): Record location content.
8235         Optimize use of HL.
8236         (rl78_alloc_physical_registers_ro1): Likewise.
8237         (rl78_alloc_physical_registers_cmp): Likewise.
8238         (rl78_alloc_physical_registers_umul): Likewise.
8239         (rl78_alloc_address_registers_macax): New.
8240         (rl78_alloc_physical_registers): Initialize and set location
8241         content memory as needed.
8242         (rl78_reorg): Make sure split2 is called.
8243         (rl78_rtx_costs): New.
8245 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
8247         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
8248         for (not (neg ...)) and (neg (not ...)) cases.
8250 2013-09-11  Richard Biener  <rguenther@suse.de>
8252         PR middle-end/58377
8253         * passes.def: Split critical edges before late uninit warning passes.
8254         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
8256 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
8258         PR tree-optimization/58385
8259         * fold-const.c (build_range_check): If both low and high are NULL,
8260         use omit_one_operand_loc to preserve exp side-effects.
8262 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8264         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
8266 2013-09-11  Richard Biener  <rguenther@suse.de>
8268         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
8269         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
8270         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
8271         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
8272         stmts_from_loop, known_dependences_p, build_empty_rdg,
8273         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
8274         * tree-loop-distribution.c: ... here.
8275         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
8276         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
8277         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
8278         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
8279         * tree-loop-distribution.c: ... here.
8280         * tree-loop-distribution.c: Include gimple-pretty-print.h.
8281         (struct partition_s): Add loops member.
8282         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
8283         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
8284         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
8286 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8287             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8288             Sergey Lega  <sergey.s.lega@intel.com>
8289             Anna Tikhonova  <anna.tikhonova@intel.com>
8290             Ilya Tocar  <ilya.tocar@intel.com>
8291             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8292             Ilya Verbin  <ilya.verbin@intel.com>
8293             Kirill Yukhin  <kirill.yukhin@intel.com>
8294             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8296         * config/i386/constraints.md (k): New.
8297         (Yk): Ditto.
8298         * config/i386/i386.c (const regclass_map): Add new mask registers.
8299         (dbx_register_map): Ditto.
8300         (dbx64_register_map): Ditto.
8301         (svr4_dbx_register_map): Ditto.
8302         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
8303         disabled.
8304         (ix86_preferred_reload_class): Disable constants for mask registers.
8305         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
8306         (ix86_hard_regno_mode_ok): Support new mask registers.
8307         (x86_order_regs_for_local_alloc): Ditto.
8308         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
8309         (FIXED_REGISTERS): Add new mask registers.
8310         (CALL_USED_REGISTERS): Ditto.
8311         (REG_ALLOC_ORDER): Ditto.
8312         (VALID_MASK_REG_MODE): New.
8313         (FIRST_MASK_REG): Ditto.
8314         (LAST_MASK_REG): Ditto.
8315         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
8316         (MAYBE_MASK_CLASS_P): New.
8317         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
8318         (REG_CLASS_CONTENTS): Ditto.
8319         (MASK_REGNO_P): New.
8320         (ANY_MASK_REG_P): Ditto.
8321         (HI_REGISTER_NAMES): Add new mask registers.
8322         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
8323         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
8324         mask registers.
8325         (attribute "type"): Add mskmov, msklog.
8326         (attribute "length_immediate"): Support them.
8327         (attribute "memory"): Ditto.
8328         (attribute "prefix_0f"): Ditto.
8329         (*movhi_internal): Support new mask registers.
8330         (*movqi_internal): Ditto.
8331         (define_split): Split out clobber pattern is a logic
8332         insn on mask registers.
8333         (*k<logic><mode>): New.
8334         (*andhi_1): Extend to support mask regs.
8335         (*andqi_1): Extend to support mask regs.
8336         (kandn<mode>): New.
8337         (define_split): Split and-not to and and not if operands
8338         are not mask regs.
8339         (*<code><mode>_1): Separate HI mode to new pattern...
8340         (*<code>hi_1): This.
8341         (*<code>qi_1): Extend to support mask regs.
8342         (kxnor<mode>): New.
8343         (kortestzhi): Ditto.
8344         (kortestchi): Ditto.
8345         (kunpckhi): Ditto.
8346         (*one_cmpl<mode>2_1): Remove HImode and handle it...
8347         (*one_cmplhi2_1): ...Here, now with mask registers support.
8348         (*one_cmplqi2_1): Support new mask registers.
8349         (HI/QImode arithmetics splitter): Don't split if mask registers
8350         are used.
8351         (HI/QImode not splitter): Ditto.
8352         * config/i386/predicated.md (mask_reg_operand): New.
8353         (general_reg_operand): Ditto.
8355 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8357         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
8358         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
8360 2013-09-10  Jeff Law  <law@redhat.com>
8362         PR tree-optimization/58380
8363         * tree-ssa-threadupdate.c (thread_block): Recognize another case
8364         of threading through a buried loop header.
8366         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
8367         return value for single successor case.
8369 2013-09-10  Jan Hubicka  <jh@suse.cz>
8371         * ipa-devirt.c (ipa_devirt): Enable with LTO.
8373 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
8375         PR target/58361
8376         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
8377         support conditional execution.
8378         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
8380 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
8382         * lra.c (lra): Clear lra_optional_reload_pseudos before every
8383         constraint pass.
8384         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
8385         Check destination too to check move insn.
8386         (undo_optional_reloads): Add check that the original peudo did not
8387         changed its allocation and the optional reload was inherited on last
8388         inheritance pass.  Break loop after deciding to keep optional reload.
8389         (lra_undo_inheritance): Add check that inherited pseudo still in
8390         memory.
8392 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
8394         * config/aarch64/aarch64.md (generic_sched): New.
8395         * config/aarch64/aarch64-generic.md (load): Make conditional
8396         on generic_sched attribute.
8397         (nonload): Likewise.
8399 2013-09-10  Jan Hubicka  <jh@suse.cz>
8401         * lto-cgraph.c: Include ipa-utils.h.
8402         (compute_ltrans_boundary): Also add possible targets into the boundary.
8404 2013-09-10  Jan Hubicka  <jh@suse.cz>
8406         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
8407         VAR_DECL of vtable rather than full expression.
8409 2013-09-10  Jan Hubicka  <jh@suse.cz>
8410             Paolo Carlini  <paolo.carlini@oracle.com>
8412         * cgraphunit.c (analyze_functions): Save input_location, set it
8413         to UNKNOWN_LOCATION and restore it at the end.
8415 2013-09-10  Martin Jambor  <mjambor@suse.cz>
8417         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
8418         represented by a thunk.
8420 2013-09-10  Jeff Law  <law@redhat.com>
8422         PR tree-optimization/58343
8423         * tree-ssa-threadupdate.c (thread_block): Identify and disable
8424         jump threading requests through loop headers buried in the middle
8425         of a jump threading path.
8427         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
8428         in return value/type.
8430 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
8432         PR rtl-optimization/58365
8433         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
8434         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
8435         it differs.
8437 2013-09-10  Richard Biener  <rguenther@suse.de>
8439         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
8440         * tree-data-ref.c (create_rdg_vertices): Collect all data
8441         references, signal failure to the caller, use data-ref API.
8442         (build_rdg): Compute data references only once.  Maintain lifetime
8443         of data references and data dependences from within RDG.
8444         (free_rdg): Free dependence relations.
8445         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
8446         inventing extra dependences.
8447         (distribute_loop): Update for RDG API changes.
8449 2013-09-10  Kai Tietz  <ktietz@redhat.com>
8451         * doc/invoke.texi (fms-extensions): Document changed
8452         behavior for ms-abi targets.
8453         * config/i386/i386.c (ix86_option_override_internal):
8454         Set default value of option -fms-extension for ms-abi targets.
8456 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8458         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
8460 2013-09-10  Alan Modra  <amodra@gmail.com>
8462         PR target/58330
8463         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
8465 2013-09-10  Alan Modra  <amodra@gmail.com>
8467         * config/rs6000/predicates.md (add_cint_operand): New.
8468         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
8469         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
8470         using add_cint_operand.
8471         (largetoc_high_plus_aix): Likewise.
8473 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
8475         PR tree-optimization/58364
8476         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
8477         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
8478         the current range can't be an unconditional true or false.
8480 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
8482         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
8484 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
8486         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
8488 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
8490         PR c++/43452
8491         * doc/invoke.texi (-Wdelete-incomplete): Document it.
8493 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
8495         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
8496         NO_REGS for immediate that can't be moved directly into FP_REGS.
8498 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8500         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
8501         comparison with negated operand.
8502         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
8503         RTL form.
8505 2013-09-09  Richard Biener  <rguenther@suse.de>
8507         PR middle-end/58326
8508         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
8509         of a subloop record all its block as affecting loop-closed SSA form.
8511 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8513         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
8514         of an rtx/bitpos pair.
8515         (store_fixed_bit_field): Update accordingly.
8517 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8519         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
8520         GEN_INT.
8521         * builtins.c (expand_errno_check): Likewise.
8522         * dwarf2cfi.c (init_return_column_size): Likewise.
8523         * except.c (sjlj_mark_call_sites): Likewise.
8524         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
8525         * lra-constraints.c (emit_inc): Likewise.
8526         * ree.c (combine_set_extension): Likewise.
8527         * regmove.c (fixup_match_2): Likewise.
8528         * reload1.c (inc_for_reload): Likewise.
8530 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8532         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
8533         (force_to_mode, simplify_shift_const_1, simplify_comparison):
8534         Use gen_int_mode with the mode of the associated simplify_* call.
8535         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
8536         * expmed.c (expand_shift_1): Likewise.
8537         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8538         * loop-iv.c (iv_number_of_iterations): Likewise.
8539         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8540         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
8542 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8544         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
8545         of the associated expand_* call.
8546         (asan_emit_stack_protection): Likewise.
8547         * builtins.c (round_trampoline_addr): Likewise.
8548         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
8549         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
8550         (emit_store_flag): Likewise.
8551         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
8552         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
8553         Likewise.
8554         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8555         * ifcvt.c (noce_try_store_flag_constants): Likewise.
8556         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8557         * modulo-sched.c (generate_prolog_epilog): Likewise.
8558         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
8559         (expand_ctz, expand_ffs, expand_unop): Likewise.
8561 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8563         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
8564         of the associated gen_rtx_* call.
8565         * caller-save.c (init_caller_save): Likewise.
8566         * combine.c (find_split_point, make_extraction): Likewise.
8567         (make_compound_operation): Likewise.
8568         * dwarf2out.c (mem_loc_descriptor): Likewise.
8569         * explow.c (plus_constant, probe_stack_range): Likewise.
8570         * expmed.c (expand_mult_const): Likewise.
8571         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
8572         * reload1.c (init_reload): Likewise.
8573         * valtrack.c (cleanup_auto_inc_dec): Likewise.
8574         * var-tracking.c (adjust_mems): Likewise.
8575         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
8576         rather than gen_rtx_fmt_ee.
8578 2013-09-09  Jan Hubicka  <jh@suse.cz>
8580         PR middle-end/58294
8581         * value-prof.c (gimple_ic): Copy also abnormal edges.
8583 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8585         * asan.c (asan_shadow_cst): Use gen_int_mode.
8587 2013-09-08  Jan Hubicka  <jh@suse.cz>
8589         * ipa-profile.c: Add toplevel comment.
8590         (ipa_propagate_frequency_1): Be more conservative when profile is read.
8591         (contains_hot_call_p): New function.
8592         (ipa_propagate_frequency): Set frequencies based on counts when
8593         profile is read.
8594         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
8595         profile; do not tamper with profile after inlining if it is read.
8597 2013-09-08  Jan Hubicka  <jh@suse.cz>
8599         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
8600         speculative edges.
8602 2013-09-08  Jan Hubicka  <jh@suse.cz>
8604         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
8605         summary generation.
8607 2013-09-08  Jeff Law  <law@redhat.com>
8609         PR bootstrap/58340
8610         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
8611         of 'found'.
8613 2013-09-08  Andi Kleen  <ak@linux.intel.com>
8615         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
8617 2013-09-08  Jan Hubicka  <jh@suse.cz>
8619         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
8621 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
8623         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
8624         for non-debug insns.
8625         * lra.c (new_insn_reg): Take the containing insn as a parameter.
8626         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
8627         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
8628         accordingly.
8630 2013-09-08  Jan Hubicka  <jh@suse.cz>
8632         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
8633         targets and devirtualize to BUILT_IN_UNREACHABLE.
8634         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
8635         * ipa.c (walk_polymorphic_call_targets): New function.
8636         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
8637         functions; use the new timevar.
8638         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
8639         was removed from the program.
8640         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
8641         not consider it in the walk when its vtable is dead.
8642         (possible_polymorphic_call_targets_1): Pass anonymous flag to
8643         record_binfo.
8644         (devirt_variable_node_removal_hook): New function.
8645         (possible_polymorphic_call_targets): Also register
8646         devirt_variable_node_removal_hook.
8647         (ipa_devirt): Do not do non-speculative devirtualization.
8648         (gate_ipa_devirt): One execute if devirtualizing speculatively.
8650 2013-09-08  Jan Hubicka  <jh@suse.cz>
8652         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
8653         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
8654         varpool_remove_variable_insertion_hook): Declare.
8655         * varpool.c (varpool_node_hook_list): New structure.
8656         (first_varpool_node_removal_hook,
8657         first_varpool_variable_insertion_hook): New variables.
8658         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
8659         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
8660         varpool_remove_variable_insertion_hook,
8661         varpool_call_variable_insertion_hooks): New functions.
8662         (varpool_remove_node): Use it.
8664 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
8666         PR c++/54941
8667         * diagnostic.c (diagnostic_build_prefix): When s.file is
8668         "<built-in>" don't output line and column numbers.
8670 2013-09-06  Jan Hubicka  <jh@suse.cz>
8672         * cgraphunit.c (expand_thunk): Get body before touching arguments.
8673         * lto-streamer-out.c: Stream thunks, too.
8674         * lto-streamer-in.c (input_function): Pop cfun here
8675         (lto_read_body): Instead of here.
8677 2013-09-06  Caroline Tice  <cmtice@google.com>
8679         * doc/install.texi: Add documentation for the --enable-vtable-verify
8680         and the --disable-libvtv configure options.
8682 2013-09-06  Jeff Law  <law@redhat.com>
8684         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8685         edge implied equivalences into successor phis.
8687 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
8689         * resource.c (mark_referenced_resources): Handle COND_EXEC.
8691 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
8693         * resource.c (mark_target_live_regs): Compute resources taking
8694         into account if a call is predicated or not.
8696 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
8698         * toplev.c (output_stack_usage): Be prepared for suffixes created by
8699         the compiler in the function names.
8701 2013-09-06  Jan Hubicka  <jh@suse.cz>
8703         PR middle-end/58094
8704         * ipa-inline.c (has_caller_p): New function.
8705         (want_inline_function_to_all_callers_p): Use it.
8706         (sum_callers, inline_to_all_callers): Break out from ...
8707         (ipa_inline): ... here.
8709 2013-09-06  Jan Hubicka  <jh@suse.cz>
8711         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
8712         only when AVX is enabled.
8714 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8716         * config/aarch64/aarch64.md
8717         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
8718         is fpsimd_<load/store>2.
8719         (load_pair<mode>): Likewise.
8720         (store_pair<mode>): Likewise.
8722 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8724         * config/arm/types.md (type): Add "mrs" type.
8725         * config/aarch64/aarch64.md
8726         (aarch64_load_tp_hard): Make type "mrs".
8727         * config/arm/arm.md
8728         (load_tp_hard): Make type "mrs".
8729         * config/arm/cortex-a15.md: Update with new attributes.
8730         * config/arm/cortex-a5.md: Update with new attributes.
8731         * config/arm/cortex-a53.md: Update with new attributes.
8732         * config/arm/cortex-a7.md: Update with new attributes.
8733         * config/arm/cortex-a8.md: Update with new attributes.
8734         * config/arm/cortex-a9.md: Update with new attributes.
8735         * config/arm/cortex-m4.md: Update with new attributes.
8736         * config/arm/cortex-r4.md: Update with new attributes.
8737         * config/arm/fa526.md: Update with new attributes.
8738         * config/arm/fa606te.md: Update with new attributes.
8739         * config/arm/fa626te.md: Update with new attributes.
8740         * config/arm/fa726te.md: Update with new attributes.
8742 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8744         * config/aarch64/aarch64.md
8745         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
8746         (*movtf_aarch64): Likewise.
8747         * config/arm/arm.md
8748         (thumb1_movdi_insn): Use "multiple" for type where more than one
8749         instruction is used for a move.
8750         (*arm32_movhf): Likewise.
8751         (*thumb_movdf_insn): Likewise.
8753 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8755         * config/arm/types.md (type): Rename fcpys to fmov.
8756         * config/arm/vfp.md
8757         (*arm_movsi_vfp): Rename type fcpys as fmov.
8758         (*thumb2_movsi_vfp): Likewise
8759         (*movhf_vfp_neon): Likewise
8760         (*movhf_vfp): Likewise
8761         (*movsf_vfp): Likewise
8762         (*thumb2_movsf_vfp): Likewise
8763         (*movsfcc_vfp): Likewise
8764         (*thumb2_movsfcc_vfp): Likewise
8765         * config/aarch64/aarch64-simd.md
8766         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
8767         * config/aarch64/aarch64.md
8768         (*movsi_aarch64): Replace type mov_reg with fmovs.
8769         (*movdi_aarch64): Likewise
8770         (*movsf_aarch64): Likewise
8771         (*movdf_aarch64): Likewise
8772         * config/arm/arm.c
8773         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
8774         * config/arm/iwmmxt.md
8775         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
8776         * config/arm/arm1020e.md: Update with new attributes.
8777         * config/arm/cortex-a15-neon.md: Update with new attributes.
8778         * config/arm/cortex-a5.md: Update with new attributes.
8779         * config/arm/cortex-a53.md: Update with new attributes.
8780         * config/arm/cortex-a7.md: Update with new attributes.
8781         * config/arm/cortex-a8-neon.md: Update with new attributes.
8782         * config/arm/cortex-a9.md: Update with new attributes.
8783         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8784         * config/arm/cortex-r4f.md: Update with new attributes.
8785         * config/arm/marvell-pj4.md: Update with new attributes.
8786         * config/arm/vfp11.md: Update with new attributes.
8788 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8790         * config/aarch64/aarch64.md
8791         (*madd<mode>): Fix type attribute.
8792         (*maddsi_uxtw): Likewise.
8793         (*msub<mode>): Likewise.
8794         (*msubsi_uxtw): Likewise.
8795         (<su_optab>maddsidi4): Likewise.
8796         (<su_optab>msubsidi4): Likewise.
8798 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8800         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
8801         * config/arm/arm.md (core_cycles): Remove fdiv.
8802         * config/arm/vfp.md:
8803         (*sqrtsf2_vfp): Update for attribute changes.
8804         (*sqrtdf2_vfp): Likewise.
8805         * config/aarch64/aarch64.md:
8806         (sqrt<mode>2): Update for attribute changes.
8807         * config/arm/arm1020e.md: Update with new attributes.
8808         * config/arm/cortex-a15-neon.md: Update with new attributes.
8809         * config/arm/cortex-a5.md: Update with new attributes.
8810         * config/arm/cortex-a53.md: Update with new attributes.
8811         * config/arm/cortex-a7.md: Update with new attributes.
8812         * config/arm/cortex-a8-neon.md: Update with new attributes.
8813         * config/arm/cortex-a9.md: Update with new attributes.
8814         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8815         * config/arm/cortex-r4f.md: Update with new attributes.
8816         * config/arm/marvell-pj4.md: Update with new attributes.
8817         * config/arm/vfp11.md: Update with new attributes.
8819 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8821         * config/arm/types.md
8822         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
8823         * config/aarch64/aarch64.md
8824         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
8825         new attributes.
8826         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
8827         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
8828         (float<GPI:mode><GPF:mode>2): Likewise.
8829         * config/arm/vfp.md
8830         (*truncsisf2_vfp): Update with new attributes.
8831         (*truncsidf2_vfp): Likewise.
8832         (fixuns_truncsfsi2): Likewise.
8833         (fixuns_truncdfsi2): Likewise.
8834         (*floatsisf2_vfp): Likewise.
8835         (*floatsidf2_vfp): Likewise.
8836         (floatunssisf2): Likewise.
8837         (floatunssidf2): Likewise.
8838         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
8839         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
8840         * config/arm/arm1020e.md: Update with new attributes.
8841         * config/arm/cortex-a15-neon.md: Update with new attributes.
8842         * config/arm/cortex-a5.md: Update with new attributes.
8843         * config/arm/cortex-a53.md: Update with new attributes.
8844         * config/arm/cortex-a7.md: Update with new attributes.
8845         * config/arm/cortex-a8-neon.md: Update with new attributes.
8846         * config/arm/cortex-a9.md: Update with new attributes.
8847         * config/arm/cortex-m4-fpu.md: Update with new attributes.
8848         * config/arm/cortex-r4f.md: Update with new attributes.
8849         * config/arm/marvell-pj4.md: Update with new attributes.
8850         * config/arm/vfp11.md: Update with new attributes.
8852 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8854         * config/aarch64/arm_neon.h
8855         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
8856         (vqtbx<1,2,3,4><q>_s8): Likewise.
8858 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8860         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
8861         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
8862         (add<mode>3): Add type attribute.
8863         (add<mode>3): Likewise.
8864         (usadd<mode>3): Likewise.
8865         (ssadd<mode>3): Likewise.
8866         (sub<mode>3): Likewise.
8867         (sub<mode>3): Likewise.
8868         (ussub<mode>3): Likewise.
8869         (sssub<mode>3): Likewise.
8870         (ssmulsa3): Likewise.
8871         (usmulusa3): Likewise.
8872         (arm_usatsihi): Likewise.
8873         * config/arm/vfp.md
8874         (*movdi_vfp): Add types for all instructions.
8875         (*movdi_vfp_cortexa8): Likewise.
8876         (*movhf_vfp_neon): Likewise.
8877         (*movhf_vfp): Likewise.
8878         (*movdf_vfp): Likewise.
8879         (*thumb2_movdf_vfp): Likewise.
8880         (*thumb2_movdfcc_vfp): Likewise.
8881         * config/arm/arm.md: Add type attribute to all insn patterns.
8882         (*thumb1_adddi3): Add type attribute.
8883         (*arm_adddi3): Likewise.
8884         (*adddi_sesidi_di): Likewise.
8885         (*adddi_zesidi_di): Likewise.
8886         (*thumb1_addsi3): Likewise.
8887         (addsi3_compare0): Likewise.
8888         (*addsi3_compare0_scratch): Likewise.
8889         (*compare_negsi_si): Likewise.
8890         (cmpsi2_addneg): Likewise.
8891         (*addsi3_carryin_<optab>): Likewise.
8892         (*addsi3_carryin_alt2_<optab>): Likewise.
8893         (*addsi3_carryin_clobercc_<optab>): Likewise.
8894         (*subsi3_carryin): Likewise.
8895         (*subsi3_carryin_const): Likewise.
8896         (*subsi3_carryin_compare): Likewise.
8897         (*subsi3_carryin_compare_const): Likewise.
8898         (*arm_subdi3): Likewise.
8899         (*thumb_subdi3): Likewise.
8900         (*subdi_di_zesidi): Likewise.
8901         (*subdi_di_sesidi): Likewise.
8902         (*subdi_zesidi_di): Likewise.
8903         (*subdi_sesidi_di): Likewise.
8904         (*subdi_zesidi_ze): Likewise.
8905         (thumb1_subsi3_insn): Likewise.
8906         (*arm_subsi3_insn): Likewise.
8907         (*anddi3_insn): Likewise.
8908         (*anddi_zesidi_di): Likewise.
8909         (*anddi_sesdi_di): Likewise.
8910         (*ne_zeroextracts): Likewise.
8911         (*ne_zeroextracts): Likewise.
8912         (*ite_ne_zeroextr): Likewise.
8913         (*ite_ne_zeroextr): Likewise.
8914         (*anddi_notdi_di): Likewise.
8915         (*anddi_notzesidi): Likewise.
8916         (*anddi_notsesidi): Likewise.
8917         (andsi_notsi_si): Likewise.
8918         (thumb1_bicsi3): Likewise.
8919         (*iordi3_insn): Likewise.
8920         (*iordi_zesidi_di): Likewise.
8921         (*iordi_sesidi_di): Likewise.
8922         (*thumb1_iorsi3_insn): Likewise.
8923         (*xordi3_insn): Likewise.
8924         (*xordi_zesidi_di): Likewise.
8925         (*xordi_sesidi_di): Likewise.
8926         (*arm_xorsi3): Likewise.
8927         (*andsi_iorsi3_no): Likewise.
8928         (*smax_0): Likewise.
8929         (*smax_m1): Likewise.
8930         (*arm_smax_insn): Likewise.
8931         (*smin_0): Likewise.
8932         (*arm_smin_insn): Likewise.
8933         (*arm_umaxsi3): Likewise.
8934         (*arm_uminsi3): Likewise.
8935         (*minmax_arithsi): Likewise.
8936         (*minmax_arithsi_): Likewise.
8937         (*satsi_<SAT:code>): Likewise.
8938         (arm_ashldi3_1bit): Likewise.
8939         (arm_ashrdi3_1bit): Likewise.
8940         (arm_lshrdi3_1bit): Likewise.
8941         (*arm_negdi2): Likewise.
8942         (*thumb1_negdi2): Likewise.
8943         (*arm_negsi2): Likewise.
8944         (*thumb1_negsi2): Likewise.
8945         (*negdi_extendsid): Likewise.
8946         (*negdi_zero_extend): Likewise.
8947         (*arm_abssi2): Likewise.
8948         (*thumb1_abssi2): Likewise.
8949         (*arm_neg_abssi2): Likewise.
8950         (*thumb1_neg_abss): Likewise.
8951         (one_cmpldi2): Likewise.
8952         (extend<mode>di2): Likewise.
8953         (*compareqi_eq0): Likewise.
8954         (*arm_extendhisi2addsi): Likewise.
8955         (*arm_movdi): Likewise.
8956         (*thumb1_movdi_insn): Likewise.
8957         (*arm_movt): Likewise.
8958         (*thumb1_movsi_insn): Likewise.
8959         (pic_add_dot_plus_four): Likewise.
8960         (pic_add_dot_plus_eight): Likewise.
8961         (tls_load_dot_plus_eight): Likewise.
8962         (*thumb1_movhi_insn): Likewise.
8963         (*thumb1_movsf_insn): Likewise.
8964         (*movdf_soft_insn): Likewise.
8965         (*thumb_movdf_insn): Likewise.
8966         (cbranchsi4_insn): Likewise.
8967         (cbranchsi4_scratch): Likewise.
8968         (*negated_cbranchsi4): Likewise.
8969         (*tbit_cbranch): Likewise.
8970         (*tlobits_cbranch): Likewise.
8971         (*tstsi3_cbranch): Likewise.
8972         (*cbranchne_decr1): Likewise.
8973         (*addsi3_cbranch): Likewise.
8974         (*addsi3_cbranch_scratch): Likewise.
8975         (*arm_cmpdi_insn): Likewise.
8976         (*arm_cmpdi_unsig): Likewise.
8977         (*arm_cmpdi_zero): Likewise.
8978         (*thumb_cmpdi_zero): Likewise.
8979         (*deleted_compare): Likewise.
8980         (*mov_scc): Likewise.
8981         (*mov_negscc): Likewise.
8982         (*mov_notscc): Likewise.
8983         (*cstoresi_eq0_thumb1_insn): Likewise.
8984         (cstoresi_nltu_thumb1): Likewise.
8985         (cstoresi_ltu_thu): Likewise.
8986         (thumb1_addsi3_addgeu): Likewise.
8987         (*arm_jump): Likewise.
8988         (*thumb_jump): Likewise.
8989         (*check_arch2): Likewise.
8990         (arm_casesi_internal): Likewise.
8991         (thumb1_casesi_dispatch): Likewise.
8992         (*arm_indirect_jump): Likewise.
8993         (*thumb1_indirect_jump): Likewise.
8994         (nop): Likewise.
8995         (*and_scc): Likewise.
8996         (*ior_scc): Likewise.
8997         (*compare_scc): Likewise.
8998         (*cond_move): Likewise.
8999         (*cond_arith): Likewise.
9000         (*cond_sub): Likewise.
9001         (*cmp_ite0): Likewise.
9002         (*cmp_ite1): Likewise.
9003         (*cmp_and): Likewise.
9004         (*cmp_ior): Likewise.
9005         (*ior_scc_scc): Likewise.
9006         (*ior_scc_scc_cmp): Likewise.
9007         (*and_scc_scc): Likewise.
9008         (*and_scc_scc_cmp): Likewise.
9009         (*and_scc_scc_nod): Likewise.
9010         (*negscc): Likewise.
9011         (movcond_addsi): Likewise.
9012         (movcond): Likewise.
9013         (*ifcompare_plus_move): Likewise.
9014         (*if_plus_move): Likewise.
9015         (*ifcompare_move_plus): Likewise.
9016         (*if_move_plus): Likewise.
9017         (*ifcompare_arith_arith): Likewise.
9018         (*if_arith_arith): Likewise.
9019         (*ifcompare_arith_move): Likewise.
9020         (*if_arith_move): Likewise.
9021         (*ifcompare_move_arith): Likewise.
9022         (*if_move_arith): Likewise.
9023         (*ifcompare_move_not): Likewise.
9024         (*if_move_not): Likewise.
9025         (*ifcompare_not_move): Likewise.
9026         (*if_not_move): Likewise.
9027         (*ifcompare_shift_move): Likewise.
9028         (*if_shift_move): Likewise.
9029         (*ifcompare_move_shift): Likewise.
9030         (*if_move_shift): Likewise.
9031         (*ifcompare_shift_shift): Likewise.
9032         (*ifcompare_not_arith): Likewise.
9033         (*ifcompare_arith_not): Likewise.
9034         (*if_arith_not): Likewise.
9035         (*ifcompare_neg_move): Likewise.
9036         (*if_neg_move): Likewise.
9037         (*ifcompare_move_neg): Likewise.
9038         (*if_move_neg): Likewise.
9039         (prologue_thumb1_interwork): Likewise.
9040         (*cond_move_not): Likewise.
9041         (*sign_extract_onebit): Likewise.
9042         (*not_signextract_onebit): Likewise.
9043         (stack_tie): Likewise.
9044         (align_4): Likewise.
9045         (align_8): Likewise.
9046         (consttable_end): Likewise.
9047         (consttable_1): Likewise.
9048         (consttable_2): Likewise.
9049         (consttable_4): Likewise.
9050         (consttable_8): Likewise.
9051         (consttable_16): Likewise.
9052         (*thumb1_tablejump): Likewise.
9053         (prefetch): Likewise.
9054         (force_register_use): Likewise.
9055         (thumb_eh_return): Likewise.
9056         (load_tp_hard): Likewise.
9057         (load_tp_soft): Likewise.
9058         (tlscall): Likewise.
9059         (*arm_movtas_ze): Likewise.
9060         (*arm_rev): Likewise.
9061         (*arm_revsh): Likewise.
9062         (*arm_rev16): Likewise.
9063         * config/arm/thumb2.md
9064         (*thumb2_smaxsi3): Likewise.
9065         (*thumb2_sminsi3): Likewise.
9066         (*thumb32_umaxsi3): Likewise.
9067         (*thumb2_uminsi3): Likewise.
9068         (*thumb2_negdi2): Likewise.
9069         (*thumb2_abssi2): Likewise.
9070         (*thumb2_neg_abss): Likewise.
9071         (*thumb2_movsi_insn): Likewise.
9072         (tls_load_dot_plus_four): Likewise.
9073         (*thumb2_movhi_insn): Likewise.
9074         (*thumb2_mov_scc): Likewise.
9075         (*thumb2_mov_negs): Likewise.
9076         (*thumb2_mov_negs): Likewise.
9077         (*thumb2_mov_nots): Likewise.
9078         (*thumb2_mov_nots): Likewise.
9079         (*thumb2_movsicc_): Likewise.
9080         (*thumb2_movsfcc_soft_insn): Likewise.
9081         (*thumb2_indirect_jump): Likewise.
9082         (*thumb2_and_scc): Likewise.
9083         (*thumb2_ior_scc): Likewise.
9084         (*thumb2_ior_scc_strict_it): Likewise.
9085         (*thumb2_cond_move): Likewise.
9086         (*thumb2_cond_arith): Likewise.
9087         (*thumb2_cond_ari): Likewise.
9088         (*thumb2_cond_sub): Likewise.
9089         (*thumb2_negscc): Likewise.
9090         (*thumb2_movcond): Likewise.
9091         (thumb2_casesi_internal): Likewise.
9092         (thumb2_casesi_internal_pic): Likewise.
9093         (*thumb2_alusi3_short): Likewise.
9094         (*thumb2_mov<mode>_shortim): Likewise.
9095         (*thumb2_addsi_short): Likewise.
9096         (*thumb2_subsi_short): Likewise.
9097         (thumb2_addsi3_compare0): Likewise.
9098         (*thumb2_cbz): Likewise.
9099         (*thumb2_cbnz): Likewise.
9100         (*thumb2_one_cmplsi2_short): Likewise.
9101         (*thumb2_negsi2_short): Likewise.
9102         (*orsi_notsi_si): Likewise.
9103         * config/arm/arm1020e.md: Update with new attributes.
9104         * config/arm/arm1026ejs.md: Update with new attributes.
9105         * config/arm/arm1136jfs.md: Update with new attributes.
9106         * config/arm/arm926ejs.md: Update with new attributes.
9107         * config/arm/cortex-a15.md: Update with new attributes.
9108         * config/arm/cortex-a5.md: Update with new attributes.
9109         * config/arm/cortex-a53.md: Update with new attributes.
9110         * config/arm/cortex-a7.md: Update with new attributes.
9111         * config/arm/cortex-a8.md: Update with new attributes.
9112         * config/arm/cortex-a9.md: Update with new attributes.
9113         * config/arm/cortex-m4.md: Update with new attributes.
9114         * config/arm/cortex-r4.md: Update with new attributes.
9115         * config/arm/fa526.md: Update with new attributes.
9116         * config/arm/fa606te.md: Update with new attributes.
9117         * config/arm/fa626te.md: Update with new attributes.
9118         * config/arm/fa726te.md: Update with new attributes.
9120 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9122         * config/aarch64/aarch64-simd.md
9123         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
9124         <vwx> iterator to ensure correct register choice.
9125         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
9126         (aarch64_sqdmull_n<mode>): Likewise.
9127         (aarch64_sqdmull2_n<mode>_internal): Likewise.
9128         * config/aarch64/arm_neon.h
9129         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
9130         (vml<as><q>_n_<su>16): Likewise.
9131         (vml<as>l_high_lane<q>_<su>16): Likewise.
9132         (vml<as>l_high_n_<su>16): Likewise.
9133         (vml<as>l_lane<q>_<su>16): Likewise.
9134         (vml<as>l_n_<su>16): Likewise.
9135         (vmul<q>_lane<q>_<su>16): Likewise.
9136         (vmul<q>_n_<su>16): Likewise.
9137         (vmull_lane<q>_<su>16): Likewise.
9138         (vmull_n_<su>16): Likewise.
9139         (vmull_high_lane<q>_<su>16): Likewise.
9140         (vmull_high_n_<su>16): Likewise.
9141         (vqrdmulh<q>_n_s16): Likewise.
9143 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
9145         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
9146         have the correct lane parameter.
9148 2013-09-06  Richard Biener <rguenther@suse.de>
9150         * cfganal.c (control_dependences::~control_dependences):
9151         Properly free all of the vector.
9153 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
9155         PR target/58269
9156         * config/i386/i386.c (ix86_conditional_register_usage):
9157         Proper initialize extended SSE registers.
9159 2013-09-06  Jan Hubicka  <jh@suse.cz>
9161         PR tree-optimization/58311
9162         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
9164 2013-09-06  Jan Hubicka  <jh@suse.cz>
9166         * Makefile.in (tree-sra.o): Update dependencies.
9167         * tree-sra.c: Include ipa-utils.h
9168         (scan_function): Use recursive_call_p.
9169         (has_caller_p): New function.
9170         (cgraph_for_node_and_aliases): Count also callers of aliases.
9172 2013-09-06  Jan Hubicka  <jh@suse.cz>
9174         PR middle-end/58094
9175         * cgraph.h (symtab_semantically_equivalent_p): Declare.
9176         * tree-tailcall.c: Include ipa-utils.h.
9177         (find_tail_calls): Use it.
9178         * ipa-pure-const.c (check_call): Likewise.
9179         * ipa-utils.c (recursive_call_p): New function.
9180         * ipa-utils.h (recursive_call_p): Dclare.
9181         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
9182         (symtab_semantically_equivalent_p): New function.
9183         * Makefile.in (tree-tailcall.o): Update dependencies.
9185 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9187         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
9188         non-inlinable part.
9190 2013-09-06  Richard Biener  <rguenther@suse.de>
9192         * lto-streamer.h (lto_global_var_decls): Remove.
9193         * Makefile.in (OBJS): Remove lto-symtab.o.
9194         (lto-symtab.o): Remove.
9195         (GTFILES): Remove lto-symtab.c
9196         * lto-symtab.c: Move to lto/
9198 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9200         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
9201         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
9202         (UNSPEC_FPINT_RINT): New constant definitions.
9203         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
9204         definition with 2 attributes.
9205         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
9206         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
9207         definitions.
9209 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9211         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
9212         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
9213         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
9214         attribute to "z196_alone".
9215         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
9216         "zEC12_simple".
9218 2013-09-06  Richard Biener  <rguenther@suse.de>
9220         * basic-block.h (class control_dependences): New.
9221         * tree-ssa-dce.c (control_dependence_map): Remove.
9222         (cd): New global.
9223         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
9224         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
9225         find_pdom, find_control_dependence, find_all_control_dependences):
9226         Move to cfganal.c.
9227         (mark_control_dependent_edges_necessary,
9228         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
9229         tree_dce_done, perform_tree_ssa_dce): Adjust.
9230         * cfganal.c (set_control_dependence_map_bit,
9231         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
9232         find_all_control_dependences): Move from tree-ssa-dce.c and
9233         implement as methods of control_dependences class.
9234         (control_dependences::control_dependences): New.
9235         (control_dependences::~control_dependences): Likewise.
9236         (control_dependences::get_edges_dependent_on): Likewise.
9237         (control_dependences::get_edge): Likewise.
9239 2013-09-04  Jan Hubicka  <jh@suse.cz>
9241         * tree.c (types_same_for_odr): Drop overactive check.
9242         * ipa-devirt.c (hash_type_name): Likewise.
9244 2013-09-04  Jan Hubicka  <jh@suse.cz>
9246         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
9247         (analyze_functions): ... here.
9249 2013-09-04  Jan Hubicka  <jh@suse.cz>
9251         PR middle-end/58201
9252         * cgraphunit.c (analyze_functions): Clear AUX fields
9253         after processing; initialize assembler name has.
9255 2013-09-05  Jeff Law  <law@redhat.com>
9257         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
9258         from thread_around_empty_block.  Record threading path into PATH.
9259         Recurse if threading through the initial block is successful.
9260         (thread_across_edge): Corresponding changes to slightly simplify.
9262 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9264         * config/aarch64/aarch64.md
9265         (type): Remove frecpe, frecps, frecpx.
9266         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
9267         fix to be a TARGET_SIMD instruction.
9268         (aarch64_frecps): Remove.
9269         * config/aarch64/aarch64-simd.md
9270         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
9271         (aarch64_frecps<mode>): Handle all float/vector of float modes.
9273 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9274             Sofiane Naci  <sofiane.naci@arm.com>
9276         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
9277         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
9278         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
9279         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
9280         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
9281         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
9282         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
9283         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
9284         into "clz, "rbit".  Rename "shift" to "shift_imm".
9285         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
9286         changes.  Update for attribute changes all occurrences of arlo_* and
9287         shift* types.
9288         * config/arm/arm-fixed.md: Update for attribute changes
9289         all occurrences of arlo_* types.
9290         * config/arm/thumb2.md: Update for attribute changes all occurrences
9291         of arlo_* types.
9292         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
9293         (cortexa7_older_only): Likewise.
9294         (cortexa7_younger):  Likewise.
9295         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
9296         (1020alu_shift_op): Likewise.
9297         (1020alu_shift_reg_op): Likewise.
9298         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
9299         (alu_shift_op): Likewise.
9300         (alu_shift_reg_op): Likewise.
9301         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
9302         (11_alu_shift_op): Likewise.
9303         (11_alu_shift_reg_op): Likewise.
9304         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
9305         (9_alu_shift_reg_op): Likewise.
9306         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
9307         attribute changes.
9308         (cortex_a15_alu_shift): Likewise.
9309         (cortex_a15_alu_shift_reg): Likewise.
9310         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
9311         attribute changes.
9312         (cortex_a5_alu_shift): Likewise.
9313         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
9314         attribute changes.
9315         (cortex_a53_alu_shift): Likewise.
9316         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
9317         attribute changes.
9318         (cortex_a7_alu_reg): Likewise.
9319         (cortex_a7_alu_shift): Likewise.
9320         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
9321         attribute changes.
9322         (cortex_a8_alu_shift): Likewise.
9323         (cortex_a8_alu_shift_reg): Likewise.
9324         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
9325         (cortex_a9_dp_shift): Likewise.
9326         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
9327         attribute changes.
9328         * config/arm/cortex-r4.md
9329         (cortex_r4_alu): Update for attribute changes.
9330         (cortex_r4_mov): Likewise.
9331         (cortex_r4_alu_shift_reg): Likewise.
9332         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
9333         (526_alu_shift_op): Likewise.
9334         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
9335         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
9336         (626te_alu_shift_op): Likewise.
9337         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
9338         (726te_alu_shift_op): Likewise.
9339         (726te_alu_shift_reg_op): Likewise.
9340         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
9341         (mp626_alu_shift_op): Likewise.
9342         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
9343         (pj4_alu_conds): Likewise.
9344         (pj4_shift): Likewise.
9345         (pj4_shift_conds): Likewise.
9346         (pj4_alu_shift): Likewise.
9347         (pj4_alu_shift_conds): Likewise.
9348         * config/aarch64/aarch64.md: Update for attribute change
9349         all occurrences of arlo_* and shift* types.
9351 2013-09-05  Mike Stump  <mikestump@comcast.net>
9353         * tree.h: Move documentation for tree_function_decl to tree-core.h
9354         with the declaration.
9356 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
9358         PR target/58139
9359         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
9360         looking for widest mode.
9362 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
9364         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
9366 2013-09-05  Richard Biener  <rguenther@suse.de>
9368         PR tree-optimization/58137
9369         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
9370         Do not create vectors of pointers.
9371         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
9372         types for the components of the vector initializer.
9373         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
9374         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
9376 2013-09-05  Martin Jambor  <mjambor@suse.cz>
9378         * ipa-prop.c (remove_described_reference): Accept missing references,
9379         return false if that hppens, otherwise return true.
9380         (cgraph_node_for_jfunc): New function.
9381         (try_decrement_rdesc_refcount): Likewise.
9382         (try_make_edge_direct_simple_call): Use them.
9383         (ipa_edge_removal_hook): Remove references from rdescs.
9384         (ipa_edge_duplication_hook): Clone rdescs and their references
9385         when the new edge has the same caller as the old one.
9386         * cgraph.c (cgraph_resolve_speculation): Remove speculative
9387         reference before removing any edges.
9389 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
9391         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
9392         initial store.
9393         * thumb2.md (thumb2_storewb_parisi): New pattern.
9395 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
9397         * config/aarch64/aarch64-option-extensions.def: Add
9398         AARCH64_OPT_EXTENSION of 'crc'.
9399         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
9400         (AARCH64_ISA_CRC): Ditto.
9401         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
9402         description of the CRC extension.
9404 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9406         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
9407         * config/rs6000/linux.h: Ditto.
9408         * alpha/linux.h: Ditto.
9409         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
9410         no_c99_libc_has_function.
9411         * config/c6x/uclinux-elf.h: Ditto.
9412         * config/lm32/uclinux-elf.h: Ditto.
9413         * config/m68k/uclinux.h: Ditto.
9414         * config/moxie/uclinux.h: Ditto.
9415         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
9416         (crisv32-*-linux*, cris-*-linux*): Ditto.
9417         * config/bfin/bfin.c: Include "tm_p.h".
9419 2013-09-05  Richard Biener  <rguenther@suse.de>
9421         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
9422         check for a definition without a basic-block.
9424 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9425             Sofiane Naci <sofiane.naci@arm.com>
9427         * config/aarch64/aarch64.md
9428         (*movti_aarch64): Rename r_2_f and f_2_r.
9429         (*movsf_aarch64): Likewise.
9430         (*movdf_aarch64): Likewise.
9431         (*movtf_aarch64): Likewise.
9432         (aarch64_movdi_<mode>low): Likewise.
9433         (aarch64_movdi_<mode>high): Likewise.
9434         (aarch64_mov<mode>high_di): Likewise.
9435         (aarch64_mov<mode>low_di): Likewise.
9436         (aarch64_movtilow_tilow): Likewise.
9437         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
9438         values to config/arm/types.md
9439         (attribute "conds"): Update for attribute change.
9440         (anddi3_insn): Likewise.
9441         (iordi3_insn): Likewise.
9442         (xordi3_insn): Likewise.
9443         (one_cmpldi2): Likewise.
9444         * config/arm/types.md (type): Add Neon types.
9445         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
9446         use "type" attribute.
9447         (movmisalign<mode>_neon_store): Likewise.
9448         (movmisalign<mode>_neon_load): Likewise.
9449         (vec_set<mode>_internal): Likewise.
9450         (vec_setv2di_internal): Likewise.
9451         (vec_extract<mode>): Likewise.
9452         (vec_extractv2di): Likewise.
9453         (add<mode>3_neon): Likewise.
9454         (adddi3_neon): Likewise.
9455         (sub<mode>3_neon): Likewise.
9456         (subdi3_neon): Likewise.
9457         (mul<mode>3_neon): Likewise.
9458         (mul<mode>3add<mode>_neon): Likewise.
9459         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
9460         (fma<VCVTF:mode>4)): Likewise.
9461         (fma<VCVTF:mode>4_intrinsic): Likewise.
9462         (fmsub<VCVTF:mode>4)): Likewise.
9463         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
9464         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
9465         (ior<mode>3): Likewise.
9466         (and<mode>3): Likewise.
9467         (anddi3_neon): Likewise.
9468         (orn<mode>3_neon): Likewise.
9469         (orndi3_neon): Likewise.
9470         (bic<mode>3_neon): Likewise.
9471         (bicdi3_neon): Likewise.
9472         (xor<mode>3): Likewise.
9473         (one_cmpl<mode>2): Likewise.
9474         (abs<mode>2): Likewise.
9475         (neg<mode>2): Likewise.
9476         (umin<mode>3_neon): Likewise.
9477         (umax<mode>3_neon): Likewise.
9478         (smin<mode>3_neon): Likewise.
9479         (smax<mode>3_neon): Likewise.
9480         (vashl<mode>3): Likewise.
9481         (vashr<mode>3_imm): Likewise.
9482         (vlshr<mode>3_imm): Likewise.
9483         (ashl<mode>3_signed): Likewise.
9484         (ashl<mode>3_unsigned): Likewise.
9485         (neon_load_count): Likewise.
9486         (ashldi3_neon_noclobber): Likewise.
9487         (signed_shift_di3_neon): Likewise.
9488         (unsigned_shift_di3_neon): Likewise.
9489         (ashrdi3_neon_imm_noclobber): Likewise.
9490         (lshrdi3_neon_imm_noclobber): Likewise.
9491         (widen_ssum<mode>3): Likewise.
9492         (widen_usum<mode>3): Likewise.
9493         (quad_halves_<code>v4si): Likewise.
9494         (quad_halves_<code>v4sf): Likewise.
9495         (quad_halves_<code>v8hi): Likewise.
9496         (quad_halves_<code>v16qi): Likewise.
9497         (reduc_splus_v2di): Likewise.
9498         (neon_vpadd_internal<mode>): Likewise.
9499         (neon_vpsmin<mode>): Likewise.
9500         (neon_vpsmax<mode>): Likewise.
9501         (neon_vpumin<mode>): Likewise.
9502         (neon_vpumax<mode>): Likewise.
9503         (ss_add<mode>_neon): Likewise.
9504         (us_add<mode>_neon): Likewise.
9505         (ss_sub<mode>_neon): Likewise.
9506         (us_sub<mode>_neon): Likewise.
9507         (neon_vadd<mode>_unspec): Likewise.
9508         (neon_vaddl<mode>): Likewise.
9509         (neon_vaddw<mode>): Likewise.
9510         (neon_vhadd<mode>): Likewise.
9511         (neon_vqadd<mode>): Likewise.
9512         (neon_vaddhn<mode>): Likewise.
9513         (neon_vmul<mode>): Likewise.
9514         (neon_vmla<mode>): Likewise.
9515         (neon_vmlal<mode>): Likewise.
9516         (neon_vmls<mode>): Likewise.
9517         (neon_vmlsl<mode>): Likewise.
9518         (neon_vqdmulh<mode>): Likewise.
9519         (neon_vqdmlal<mode>): Likewise.
9520         (neon_vqdmlsl<mode>): Likewise.
9521         (neon_vmull<mode>): Likewise.
9522         (neon_vqdmull<mode>): Likewise.
9523         (neon_vsub<mode>_unspec): Likewise.
9524         (neon_vsubl<mode>): Likewise.
9525         (neon_vsubw<mode>): Likewise.
9526         (neon_vqsub<mode>): Likewise.
9527         (neon_vhsub<mode>): Likewise.
9528         (neon_vsubhn<mode>): Likewise.
9529         (neon_vceq<mode>): Likewise.
9530         (neon_vcge<mode>): Likewise.
9531         (neon_vcgeu<mode>): Likewise.
9532         (neon_vcgt<mode>): Likewise.
9533         (neon_vcgtu<mode>): Likewise.
9534         (neon_vcle<mode>): Likewise.
9535         (neon_vclt<mode>): Likewise.
9536         (neon_vcage<mode>): Likewise.
9537         (neon_vcagt<mode>): Likewise.
9538         (neon_vtst<mode>): Likewise.
9539         (neon_vabd<mode>): Likewise.
9540         (neon_vabdl<mode>): Likewise.
9541         (neon_vaba<mode>): Likewise.
9542         (neon_vabal<mode>): Likewise.
9543         (neon_vmax<mode>): Likewise.
9544         (neon_vmin<mode>): Likewise.
9545         (neon_vpaddl<mode>): Likewise.
9546         (neon_vpadal<mode>): Likewise.
9547         (neon_vpmax<mode>): Likewise.
9548         (neon_vpmin<mode>): Likewise.
9549         (neon_vrecps<mode>): Likewise.
9550         (neon_vrsqrts<mode>): Likewise.
9551         (neon_vqabs<mode>): Likewise.
9552         (neon_vqneg<mode>): Likewise.
9553         (neon_vcls<mode>): Likewise.
9554         (clz<mode>2): Likewise.
9555         (popcount<mode>2): Likewise.
9556         (neon_vrecpe): Likewise.
9557         (neon_vrsqrte): Likewise.
9558         (neon_vget_lane<mode>_sext_internal): Likewise.
9559         (neon_vget_lane<mode>_zext_internal): Likewise.
9560         (neon_vdup_n<mode>): Likewise.
9561         (neon_vdup_nv2di): Likewise.
9562         (neon_vdpu_lane<mode>_internal): Likewise.
9563         (neon_vswp<mode>): Likewise.
9564         (float<mode><V_cvtto>2): Likewise.
9565         (floatuns<mode><V_cvtto>2): Likewise.
9566         (fix_trunc<mode><V_cvtto>)2): Likewise
9567         (fixuns_trunc<mode><V_cvtto)2): Likewise.
9568         (neon_vcvt<mode>): Likewise.
9569         (neon_vcvtv4sfv4hf): Likewise.
9570         (neon_vcvtv4hfv4sf): Likewise.
9571         (neon_vcvt_n<mode>): Likewise.
9572         (neon_vmovn<mode>): Likewise.
9573         (neon_vqmovn<mode>): Likewise.
9574         (neon_vqmovun<mode>): Likewise.
9575         (neon_vmovl<mode>): Likewise.
9576         (neon_vmul_lane<mode>): Likewise.
9577         (neon_vmull_lane<mode>): Likewise.
9578         (neon_vqdmull_lane<mode>): Likewise.
9579         (neon_vqdmulh_lane<mode>): Likewise.
9580         (neon_vmla_lane<mode>): Likewise.
9581         (neon_vmlal_lane<mode>): Likewise.
9582         (neon_vqdmlal_lane<mode>): Likewise.
9583         (neon_vmls_lane<mode>): Likewise.
9584         (neon_vmlsl_lane<mode>): Likewise.
9585         (neon_vqdmlsl_lane<mode>): Likewise.
9586         (neon_vext<mode>): Likewise.
9587         (neon_vrev64<mode>): Likewise.
9588         (neon_vrev32<mode>): Likewise.
9589         (neon_vrev16<mode>): Likewise.
9590         (neon_vbsl<mode>_internal): Likewise.
9591         (neon_vshl<mode>): Likewise.
9592         (neon_vqshl<mode>): Likewise.
9593         (neon_vshr_n<mode>): Likewise.
9594         (neon_vshrn_n<mode>): Likewise.
9595         (neon_vqshrn_n<mode>): Likewise.
9596         (neon_vqshrun_n<mode>): Likewise.
9597         (neon_vshl_n<mode>): Likewise.
9598         (neon_vqshl_n<mode>): Likewise.
9599         (neon_vqshlu_n<mode>): Likewise.
9600         (neon_vshll_n<mode>): Likewise.
9601         (neon_vsra_n<mode>): Likewise.
9602         (neon_vsri_n<mode>): Likewise.
9603         (neon_vsli_n<mode>): Likewise.
9604         (neon_vtbl1v8qi): Likewise.
9605         (neon_vtbl2v8qi): Likewise.
9606         (neon_vtbl3v8qi): Likewise.
9607         (neon_vtbl4v8qi): Likewise.
9608         (neon_vtbx1v8qi): Likewise.
9609         (neon_vtbx2v8qi): Likewise.
9610         (neon_vtbx3v8qi): Likewise.
9611         (neon_vtbx4v8qi): Likewise.
9612         (neon_vtrn<mode>_internal): Likewise.
9613         (neon_vzip<mode>_internal): Likewise.
9614         (neon_vuzp<mode>_internal): Likewise.
9615         (neon_vld1<mode>): Likewise.
9616         (neon_vld1_lane<mode>): Likewise.
9617         (neon_vld1_dup<mode>): Likewise.
9618         (neon_vld1_dupv2di): Likewise.
9619         (neon_vst1<mode>): Likewise.
9620         (neon_vst1_lane<mode>): Likewise.
9621         (neon_vld2<mode>): Likewise.
9622         (neon_vld2_lane<mode>): Likewise.
9623         (neon_vld2_dup<mode>): Likewise.
9624         (neon_vst2<mode>): Likewise.
9625         (neon_vst2_lane<mode>): Likewise.
9626         (neon_vld3<mode>): Likewise.
9627         (neon_vld3qa<mode>): Likewise.
9628         (neon_vld3qb<mode>): Likewise.
9629         (neon_vld3_lane<mode>): Likewise.
9630         (neon_vld3_dup<mode>): Likewise.
9631         (neon_vst3<mode>): Likewise.
9632         (neon_vst3qa<mode>): Likewise.
9633         (neon_vst3qb<mode>): Likewise.
9634         (neon_vst3_lane<mode>): Likewise.
9635         (neon_vld4<mode>): Likewise.
9636         (neon_vld4qa<mode>): Likewise.
9637         (neon_vld4qb<mode>): Likewise.
9638         (neon_vld4_lane<mode>): Likewise.
9639         (neon_vld4_dup<mode>): Likewise.
9640         (neon_vst4<mode>): Likewise.
9641         (neon_vst4qa<mode>): Likewise.
9642         (neon_vst4qb<mode>): Likewise.
9643         (neon_vst4_lane<mode>): Likewise.
9644         (neon_vec_unpack<US>_lo_<mode>): Likewise.
9645         (neon_vec_unpack<US>_hi_<mode>): Likewise.
9646         (neon_vec_<US>mult_lo_<mode>): Likewise.
9647         (neon_vec_<US>mult_hi_<mode>): Likewise.
9648         (neon_vec_<US>shiftl_<mode>): Likewise.
9649         (neon_unpack<US>_<mode>): Likewise.
9650         (neon_vec_<US>mult_<mode>): Likewise.
9651         (vec_pack_trunc_<mode>): Likewise.
9652         (neon_vec_pack_trunk_<mode>): Likewise.
9653         (neon_vabd<mode>_2): Likewise.
9654         (neon_vabd<mode>_3): Likewise.
9655         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
9656         (thumb2_movsi_vfp): Likewise.
9657         (movdi_vfp): Likewise.
9658         (movdi_vfp_cortexa8): Likewise.
9659         (movhf_vfp_neon): Likewise.
9660         (movhf_vfp): Likewiwse.
9661         (movsf_vfp): Likewiwse.
9662         (thumb2_movsf_vfp): Likewiwse.
9663         (movdf_vfp): Likewise.
9664         (thumb2_movdf_vfp): Likewise.
9665         (movsfcc_vfp): Likewise.
9666         (thumb2_movsfcc_vfp): Likewise.
9667         (movdfcc_vfp): Likewise.
9668         (thumb2_movdfcc_vfp): Likewise.
9669         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
9670         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
9671         (v10_v2c): Likewise.
9672         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
9673         attribute change.
9674         (cortex_a15_neon_int_2): Likewise.
9675         (cortex_a15_neon_int_3): Likewise.
9676         (cortex_a15_neon_int_4): Likewise.
9677         (cortex_a15_neon_int_5): Likewise.
9678         (cortex_a15_neon_vqneg_vqabs): Likewise.
9679         (cortex_a15_neon_vmov): Likewise.
9680         (cortex_a15_neon_vaba): Likewise.
9681         (cortex_a15_neon_vaba_qqq): Likewise.
9682         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9683         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9684         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
9685         scalar_64_32_long_scalar): Likewise.
9686         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9687         (cortex_a15_neon_mla_qqq_8_16): Likewise.
9688         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9689         lotype_qdd_64_32_long): Likewise.
9690         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9691         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9692         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
9693         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9694         (cortex_a15_neon_shift_1): Likewise.
9695         (cortex_a15_neon_shift_2): Likewise.
9696         (cortex_a15_neon_shift_3): Likewise.
9697         (cortex_a15_neon_vshl_ddd): Likewise.
9698         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9699         (cortex_a15_neon_vsra_vrsra): Likewise.
9700         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
9701         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
9702         (cortex_a15_neon_fp_vmul_ddd): Likewise.
9703         (cortex_a15_neon_fp_vmul_qqd): Likewise.
9704         (cortex_a15_neon_fp_vmla_ddd): Likewise.
9705         (cortex_a15_neon_fp_vmla_qqq): Likewise.
9706         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
9707         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
9708         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9709         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9710         (cortex_a15_neon_bp_simple): Likewise.
9711         (cortex_a15_neon_bp_2cycle): Likewise.
9712         (cortex_a15_neon_bp_3cycle): Likewise.
9713         (cortex_a15_neon_vld1_1_2_regs): Likewise.
9714         (cortex_a15_neon_vld1_3_4_regs): Likewise.
9715         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9716         (cortex_a15_neon_vld2_4_regs): Likewise.
9717         (cortex_a15_neon_vld3_vld4): Likewise.
9718         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9719         (cortex_a15_neon_vst1_3_4_regs): Likewise.
9720         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
9721         (cortex_a15_neon_vst3_vst4): Likewise.
9722         (cortex_a15_neon_vld1_vld2_lane): Likewise.
9723         (cortex_a15_neon_vld3_vld4_lane" 10
9724         (cortex_a15_neon_vst1_vst2_lane): Likewise.
9725         (cortex_a15_neon_vst3_vst4_lane): Likewise.
9726         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
9727         (cortex_a15_neon_ldm_2): Likewise.0
9728         (cortex_a15_neon_stm_2): Likewise.
9729         (cortex_a15_neon_mcr): Likewise.
9730         (cortex_a15_neon_mcr_2_mcrr): Likewise.
9731         (cortex_a15_neon_mrc): Likewise.
9732         (cortex_a15_neon_mrrc): Likewise.
9733         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
9734         change.
9735         (cortex_a15_alu_shift): Likewise.
9736         (cortex_a15_alu_shift_reg): Likewise.
9737         (cortex_a15_mult32): Likewise.
9738         (cortex_a15_mult64): Likewise.
9739         (cortex_a15_block): Likewise.
9740         (cortex_a15_branch): Likewise.
9741         (cortex_a15_load1): Likewise.
9742         (cortex_a15_load3): Likewise.
9743         (cortex_a15_store1): Likewise.
9744         (cortex_a15_store3): Likewise.
9745         (cortex_a15_call): Likewise.
9746         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
9747         (cortex_a5_f2r): Likewise.
9748         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
9749         change.
9750         (cortex_a53_f2r): Likewise.
9751         * config/arm/cortex-a7.md
9752         (cortex_a7_branch): Update for attribute change.
9753         (cortex_a7_call): Likewise.
9754         (cortex_a7_alu_imm): Likewise.
9755         (cortex_a7_alu_reg): Likewise.
9756         (cortex_a7_alu_shift): Likewise.
9757         (cortex_a7_mul): Likewise.
9758         (cortex_a7_load1): Likewise.
9759         (cortex_a7_store1): Likewise.
9760         (cortex_a7_load2): Likewise.
9761         (cortex_a7_store2): Likewise.
9762         (cortex_a7_load3): Likewise.
9763         (cortex_a7_store3): Likewise.
9764         (cortex_a7_load4): Likewise.
9765         (cortex_a7_store4): Likewise.
9766         (cortex_a7_fpalu): Likewise.
9767         (cortex_a7_fconst): Likewise.
9768         (cortex_a7_fpmuls): Likewise.
9769         (cortex_a7_neon_mul): Likewise.
9770         (cortex_a7_fpmacs): Likewise.
9771         (cortex_a7_neon_mla: Likewise.
9772         (cortex_a7_fpmuld: Likewise.
9773         (cortex_a7_fpmacd: Likewise.
9774         (cortex_a7_fpfmad: Likewise.
9775         (cortex_a7_fdivs: Likewise.
9776         (cortex_a7_fdivd: Likewise.
9777         (cortex_a7_r2f: Likewise.
9778         (cortex_a7_f2r: Likewise.
9779         (cortex_a7_f_flags: Likewise.
9780         (cortex_a7_f_loads: Likewise.
9781         (cortex_a7_f_loadd: Likewise.
9782         (cortex_a7_f_stores: Likewise.
9783         (cortex_a7_f_stored: Likewise.
9784         (cortex_a7_neon): Likewise.
9785         * config/arm/cortex-a8-neon.md
9786         (cortex_a8_neon_mrc): Update for attribute change.
9787         (cortex_a8_neon_mrrc): Likewise.
9788         (cortex_a8_neon_int_1): Likewise.
9789         (cortex_a8_neon_int_2): Likewise.
9790         (cortex_a8_neon_int_3): Likewise.
9791         (cortex_a8_neon_int_4): Likewise.
9792         (cortex_a8_neon_int_5): Likewise.
9793         (cortex_a8_neon_vqneg_vqabs): Likewise.
9794         (cortex_a8_neon_vmov): Likewise.
9795         (cortex_a8_neon_vaba): Likewise.
9796         (cortex_a8_neon_vaba_qqq): Likewise.
9797         (cortex_a8_neon_vsma): Likewise.
9798         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9799         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9800         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
9801         long_scalar): Likewise.
9802         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9803         (cortex_a8_neon_mla_qqq_8_16): Likewise.
9804         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9805         long_scalar_qdd_64_32_long): Likewise.
9806         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9807         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9808         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
9809         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9810         (cortex_a8_neon_shift_1): Likewise.
9811         (cortex_a8_neon_shift_2): Likewise.
9812         (cortex_a8_neon_shift_3): Likewise.
9813         (cortex_a8_neon_vshl_ddd): Likewise.
9814         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9815         (cortex_a8_neon_vsra_vrsra): Likewise.
9816         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
9817         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
9818         (cortex_a8_neon_fp_vsum): Likewise.
9819         (cortex_a8_neon_fp_vmul_ddd): Likewise.
9820         (cortex_a8_neon_fp_vmul_qqd): Likewise.
9821         (cortex_a8_neon_fp_vmla_ddd): Likewise.
9822         (cortex_a8_neon_fp_vmla_qqq): Likewise.
9823         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
9824         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
9825         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9826         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9827         (cortex_a8_neon_bp_simple): Likewise.
9828         (cortex_a8_neon_bp_2cycle): Likewise.
9829         (cortex_a8_neon_bp_3cycle): Likewise.
9830         (cortex_a8_neon_ldr): Likewise.
9831         (cortex_a8_neon_str): Likewise.
9832         (cortex_a8_neon_vld1_1_2_regs): Likewise.
9833         (cortex_a8_neon_vld1_3_4_regs): Likewise.
9834         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9835         (cortex_a8_neon_vld2_4_regs): Likewise.
9836         (cortex_a8_neon_vld3_vld4): Likewise.
9837         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9838         (cortex_a8_neon_vst1_3_4_regs): Likewise.
9839         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
9840         (cortex_a8_neon_vst3_vst4): Likewise.
9841         (cortex_a8_neon_vld1_vld2_lane): Likewise.
9842         (cortex_a8_neon_vld3_vld4_lane): Likewise.
9843         (cortex_a8_neon_vst1_vst2_lane): Likewise.
9844         (cortex_a8_neon_vst3_vst4_lane): Likewise.
9845         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
9846         (cortex_a8_neon_mcr): Likewise.
9847         (cortex_a8_neon_mcr_2_mcrr): Likewise.
9848         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
9849         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
9850         change.
9851         (ca9_neon_mrrc): Likewise.
9852         (cortex_a9_neon_int_1): Likewise.
9853         (cortex_a9_neon_int_2): Likewise.
9854         (cortex_a9_neon_int_3): Likewise.
9855         (cortex_a9_neon_int_4): Likewise.
9856         (cortex_a9_neon_int_5): Likewise.
9857         (cortex_a9_neon_vqneg_vqabs): Likewise.
9858         (cortex_a9_neon_vmov): Likewise.
9859         (cortex_a9_neon_vaba): Likewise.
9860         (cortex_a9_neon_vaba_qqq): Likewise.
9861         (cortex_a9_neon_vsma): Likewise.
9862         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9863         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9864         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
9865         long_scalar): Likewise.
9866         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9867         (cortex_a9_neon_mla_qqq_8_16): Likewise.
9868         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9869         long_scalar_qdd_64_32_long): Likewise.
9870         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9871         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9872         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
9873         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9874         (cortex_a9_neon_shift_1): Likewise.
9875         (cortex_a9_neon_shift_2): Likewise.
9876         (cortex_a9_neon_shift_3): Likewise.
9877         (cortex_a9_neon_vshl_ddd): Likewise.
9878         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9879         (cortex_a9_neon_vsra_vrsra): Likewise.
9880         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
9881         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
9882         (cortex_a9_neon_fp_vsum): Likewise.
9883         (cortex_a9_neon_fp_vmul_ddd): Likewise.
9884         (cortex_a9_neon_fp_vmul_qqd): Likewise.
9885         (cortex_a9_neon_fp_vmla_ddd): Likewise.
9886         (cortex_a9_neon_fp_vmla_qqq): Likewise.
9887         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
9888         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
9889         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9890         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9891         (cortex_a9_neon_bp_simple): Likewise.
9892         (cortex_a9_neon_bp_2cycle): Likewise.
9893         (cortex_a9_neon_bp_3cycle): Likewise.
9894         (cortex_a9_neon_ldr): Likewise.
9895         (cortex_a9_neon_str): Likewise.
9896         (cortex_a9_neon_vld1_1_2_regs): Likewise.
9897         (cortex_a9_neon_vld1_3_4_regs): Likewise.
9898         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9899         (cortex_a9_neon_vld2_4_regs): Likewise.
9900         (cortex_a9_neon_vld3_vld4): Likewise.
9901         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9902         (cortex_a9_neon_vst1_3_4_regs): Likewise.
9903         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
9904         (cortex_a9_neon_vst3_vst4): Likewise.
9905         (cortex_a9_neon_vld1_vld2_lane): Likewise.
9906         (cortex_a9_neon_vld3_vld4_lane): Likewise.
9907         (cortex_a9_neon_vst1_vst2_lane): Likewise.
9908         (cortex_a9_neon_vst3_vst4_lane): Likewise.
9909         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
9910         (cortex_a9_neon_mcr): Likewise.
9911         (cortex_a9_neon_mcr_2_mcrr): Likewise.
9912         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
9913         (cortex_a9_fps): Likewise.
9914         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
9915         change.
9916         (cortex_m4_fmuls): Likewise.
9917         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
9918         change.
9919         (cortex_r4_mrc): Likewise.
9920         * config/arm/iterators.md: Update comment referring to neon_type.
9921         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
9922         (iwmmxt_movsi_insn): Likewise.
9923         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
9924         attribute change.
9925         (pj4_core_to_vfp): Likewise.
9926         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
9927         attribute change.
9928         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
9929         (vfp_fstore): Likewise.
9930         * doc/md.texi: Change references to neon_type to refer to type.
9932 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
9934         * tree.h (DECL_BUILT_IN): Fix typo in comment.
9936 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
9938         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
9939         lglobl if not weak.
9941 2013-09-04  Easwaran Raman  <eraman@google.com>
9943         PR middle-end/57370
9944         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
9945         (build_and_add_sum): Use it.
9946         (appears_later_in_bb): Simplify code.
9948 2013-09-04  Teresa Johnson  <tejohnson@google.com>
9950         * dumpfile.c (dump_finish): Don't close stderr/stdout.
9952 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
9954         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
9956 2013-09-04  Jan Hubicka  <jh@suse.cz>
9958         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
9959         * ipa-devirt.c: Include diganostic.h
9960         (odr_type_d): Add types and types_set.
9961         (hash_type_name): Work for types with vtables during LTO.
9962         (odr_hasher::remove): Fix comment; destroy types_set.
9963         (add_type_duplicate): New function,
9964         (get_odr_type): Use it.
9965         (dump_type_inheritance_graph): Dump type duplicates.
9966         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
9967         graph.
9968         * tree.c (types_same_for_odr): Give exact answers on types with
9969         virtual tables.
9971 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
9973         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
9974         explaining their differences.
9976 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
9978         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
9980 2013-09-03  Jeff Law  <law@redhat.com>
9982         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
9983         when not threading through a joiner block.  Pass joiner/no joiner
9984         state to register_jump_thread.
9985         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
9986         state from argument rather than implying on path length.
9987         Dump the entire jump thread path into debugging dump.
9988         * tree-flow.h (register_jump_thread): Update prototype.
9990 2013-08-29  Xinliang David Li  <davidxl@google.com>
9992         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
9993         Remove a trivial gcc_assert.
9995 2013-08-29  Xinliang David Li  <davidxl@google.com>
9997         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
9998         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
9999         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10000         Delay base decl alignment adjustment.
10001         * tree-vectorizer.c (vect_destroy_datarefs): New function.
10002         * tree-vectorizer.h: New data structure.
10003         (set_dr_misalignment): New function.
10004         (dr_misalignment): Ditto.
10005         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
10006         (vectorizable_load): Ditto.
10007         (ensure_base_align): New function.
10008         (vectorize_loops): Add dbg_cnt support.
10009         (execute_vect_slp): Ditto.
10010         * dbgcnt.def: New debug counter.
10011         * Makefile: New dependency.
10013 2013-09-03  Meador Inge  <meadori@codesourcery.com>
10015         Revert:
10017         2013-08-30  Meador Inge  <meadori@codesourcery.com>
10019         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10021 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
10023         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
10024         function descriptor.
10026 2013-09-03  Richard Biener  <rguenther@suse.de>
10028         * tree-affine.c (add_elt_to_tree): Fix association issue,
10029         avoid useless converts and make sure to always return a
10030         properly typed result.
10032 2013-09-03  Richard Biener  <rguenther@suse.de>
10034         PR middle-end/57656
10035         * fold-const.c (negate_expr_p): Fix division case.
10036         (negate_expr): Likewise.
10038 2013-09-03  Richard Biener  <rguenther@suse.de>
10040         PR lto/58285
10041         * tree-streamer-out.c: Include tm.h.
10042         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
10044 2013-09-03  Jan Hubicka  <jh@suse.cz>
10046         * tree-profile.c (tree_profiling): Cleanup CFG when done.
10048 2013-09-03  Alan Modra  <amodra@gmail.com>
10050         * config.gcc (powerpc*-*-linux*): Add support for little-endian
10051         multilibs to big-endian target and vice versa.
10052         * config/rs6000/t-linux64: Use := assignment on all vars.
10053         (MULTILIB_EXTRA_OPTS): Remove fPIC.
10054         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
10055         * config/rs6000/t-linux64le: New file.
10056         * config/rs6000/t-linux64bele: New file.
10057         * config/rs6000/t-linux64lebe: New file.
10059 2013-09-02  Jan Hubicka  <jh@suse.cz>
10061         * ipa-inline-transform.c (inline_transform): Do not
10062         optimize_inline_calls when not optimizing.
10064 2013-09-02  Jan Hubicka  <jh@suse.cz>
10066         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
10067         duplicated nodes for assembler names.
10068         * symtab.c (symtab_unregister_node): Do not attempt to unlink
10069         hard registers from assembler name hash.
10071 2013-09-02  Jan Hubicka  <jh@suse.cz>
10073         * ipa-split.c (execute_split_functions): Split externally visible
10074         functions called once.
10076 2013-09-02  Martin Jambor  <mjambor@suse.cz>
10078         PR ipa/58106
10079         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
10080         linked list.  When finding the correct duplicate, also consider also
10081         the caller in additon to its inlined_to node.
10083 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
10085         * config/aarch64/aarch64-simd-builtins.def
10086         (dup_lane_scalar): Remove.
10087         * config/aarch64/aarch64-simd.md
10088         (aarch64_simd_dup): Add 'w->w' alternative.
10089         (aarch64_dup_lane<mode>): Allow for VALL.
10090         (aarch64_dup_lane_scalar<mode>): Remove.
10091         (aarch64_dup_lane_<vswap_width_name><mode>): New.
10092         (aarch64_get_lane_signed<mode>): Add w->w altenative.
10093         (aarch64_get_lane_unsigned<mode>): Likewise.
10094         (aarch64_get_lane<mode>): Likewise.
10095         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
10096         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
10097         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
10098         (VCON): Change container of V2SF.
10099         (vswap_width_name): Likewise.
10100         * config/aarch64/arm_neon.h
10101         (__aarch64_vdup_lane_any): New.
10102         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10103         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
10104         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10106 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
10108         PR middle-end/56382
10109         * expr.c (emit_move_complex): Do not move complex FP values as parts if
10110         the source or the destination is a single hard register.
10112 2013-09-02  Richard Biener  <rguenther@suse.de>
10114         PR middle-end/57511
10115         * tree-scalar-evolution.c (instantiate_scev_name): Allow
10116         non-linear SCEVs.
10118 2013-09-02  Richard Biener  <rguenther@suse.de>
10120         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
10121         arithmetic to sizetype.
10123 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
10125         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
10126         Find auto-increment use both before and after candidate.
10128 2013-09-02  Marek Polacek  <polacek@redhat.com>
10130         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
10132 2013-09-01  Jan Hubicka  <jh@suse.cz>
10134         * Makefile.in: Add ipa-profile.o
10135         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
10136         * cgraph.c (struct cgraph_propagate_frequency_data,
10137         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
10138         ipa-profile.c; replace cgraph_ by ipa_ prefix.
10139         * cgraph.h (cgraph_propagate_frequency): Remove.
10140         * ipa-inline-analysis.c: Include ipa-utils.h;
10141         drop duplicated cfgloop.h.
10142         (inline_update_callee_summaries): Update.
10143         * ipa-profile.c: New file.
10144         * ipa-utils.h (ipa_propagate_frequency): Declare.
10145         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
10146         data-streamer.h, value-prof.h.
10147         (symtab_remove_unreachable_nodes): Update profile.
10148         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
10149         account_time_size, cmp_counts, dump_histogram,
10150         ipa_profile_generate_summary, ipa_profile_write_summary,
10151         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
10152         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
10153         Move to ipa-profile.c.
10155 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
10157         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
10159 2013-09-01  Jan Hubicka  <jh@suse.cz>
10161         * common.opt (fdevirtualize-speculatively): New function.
10162         * invoke.texi (fdevirtualize-speculatively): Document.
10163         * ipa-devirt.c: Include ipa-inline.h
10164         (likely_target_p): New function.
10165         (ipa_devirt): New function.
10166         (gate_ipa_devirt): New function.
10167         (pass_data_ipa_devirt): New static var.
10168         (pass_ipa_devirt): Likewise.
10169         (make_pass_ipa_devirt): New function.
10170         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
10171         (common_handle_option): Disable devirtualization when
10172         value range profiling is available.
10173         * passes.def (pass_ipa_devirt): Add.
10174         * timever.def (TV_IPA_DEVIRT): New timevar.
10175         * tree-pass.h (make_pass_ipa_devirt):
10177 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
10179         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
10180         include sanitize(undefined).
10182 2013-08-31  Diego Novillo  <dnovillo@google.com>
10184         * Makefile.in (TREE_CORE_H): Define.
10185         (TREE_H): Use.
10186         (GTFILES): Add tree-core.h.
10187         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
10188         size the array.
10189         * tree-core.h: New file.
10190         Move all data structures, enum, typedefs, global
10191         declarations and constants from ...
10192         * tree.h: ... here.
10194 2013-08-31  Jan Hubicka  <jh@suse.cz>
10196         * bulitins.c (expand_builtin): Do not early exit for gcov
10197         instrumented functions.
10199 2013-08-31  Marek Polacek  <polacek@redhat.com>
10201         * ubsan.c: Include tm_p.h.
10203 2013-08-31  Jan Hubicka  <jh@suse.cz>
10205         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
10206         warning.
10208         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
10209         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
10210         * tree-cfg.c (verify_gimple_label): ... here.
10211         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
10212         (ipa_merge_profiles): New function.
10213         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
10214         (lto_input_function_body): Likewise.
10215         * ipa-utils.h (ipa_merge_profiles): Declare.
10216         * lto-streamer.h (lto_input_function_body): Update prototype.
10217         (emit_label_in_global_context_p): Remove.
10218         * lto-symtab.c: Include ipa-utils.h
10219         (lto_cgraph_replace_node): Use ipa_merge_profiles.
10221 2013-08-31  Jan Hubicka  <jh@suse.cz>
10223         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
10225 2013-08-31  Jan Hubicka  <jh@suse.cz>
10227         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
10229 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
10231         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
10232         "cmp" RTX before signed_comparison_operator check to account
10233         for "code" changes.
10235 2013-08-30  Jan Hubicka  <jh@suse.cz>
10237         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
10238         (detect_type_change_1): Rename to ...
10239         (detect_type_change): ... this one; early return on non-polymorphic
10240         types.
10241         (detect_type_change_ssa): Add comp_type parameter; update
10242         use of detect_type_change.
10243         (compute_complex_assign_jump_func): Add param_type parameter;
10244         update use of detect_type_change_ssa.
10245         (compute_complex_ancestor_jump_func): Likewise.
10246         (ipa_get_callee_param_type): New function.
10247         (ipa_compute_jump_functions_for_edge): Compute parameter type;
10248         update calls to the jump function computation functions.
10250 2013-08-30  Teresa Johnson  <tejohnson@google.com>
10251             Steven Bosscher  <steven@gcc.gnu.org>
10253         * cfgrtl.c (fixup_new_cold_bb): New routine.
10254         (commit_edge_insertions): Invoke fixup_partitions.
10255         (find_partition_fixes): New routine.
10256         (fixup_partitions): Ditto.
10257         (verify_hot_cold_block_grouping): Update comments.
10258         (rtl_verify_edges): Invoke find_partition_fixes.
10259         (rtl_verify_bb_pointers): Update comments.
10260         (rtl_verify_bb_layout): Ditto.
10261         * basic-block.h (probably_never_executed_edge_p): Declare.
10262         (fixup_partitions): Ditto.
10263         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
10264         * bb-reorder.c (sanitize_hot_paths): New function.
10265         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
10266         sanitize_hot_paths.
10267         * predict.c (probably_never_executed_edge_p): New routine.
10268         * cfg.c (check_bb_profile): Add partition insanity warnings.
10270 2013-08-30  Meador Inge  <meadori@codesourcery.com>
10272         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10274 2013-08-30  Marek Polacek  <polacek@redhat.com>
10276         * Makefile.in (ubsan.o): Add.
10277         (c-family/c-ubsan.o): Add.
10278         (builtins.o): Add ubsan.h dependency.
10279         * ubsan.h: New file.
10280         * ubsan.c: New file.
10281         * common.opt: Add -fsanitize=undefined option.
10282         (flag_sanitize): Add variable.
10283         (fsanitize=): Add option.  Add Driver.
10284         (fsanitize=thread): Remove option.
10285         (fsanitize=address): Likewise.
10286         (static-libubsan): New option.
10287         * doc/invoke.texi: Document the new flag and -static-libubsan.
10288         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
10289         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
10290         * builtin-attrs.def (ATTR_COLD): Define.
10291         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10292         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
10293         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
10294         * flag-types.h (sanitize_code): New enum.
10295         * opts.c (common_handle_option): Parse command line arguments
10296         of -fsanitize=.  Add -fsanitize=unreachable option.
10297         * varasm.c (get_variable_section): Adjust.
10298         (assemble_noswitch_variable): Likewise.
10299         (assemble_variable): Likewise.
10300         (output_constant_def_contents): Likewise.
10301         (categorize_decl_for_section): Likewise.
10302         (place_block_symbol): Likewise.
10303         (output_object_block): Likewise.
10304         * builtins.def: Likewise.
10305         * toplev.c (compile_file): Likewise.
10306         (process_options): Likewise.
10307         * cppbuiltin.c: Likewise.
10308         * tsan.c (tsan_pass): Likewise.
10309         (tsan_gate): Likewise.
10310         (tsan_gate_O0): Likewise.
10311         * cfgexpand.c (partition_stack_vars): Likewise.
10312         (expand_stack_vars): Likewise.
10313         (defer_stack_allocation): Likewise.
10314         (expand_used_vars): Likewise.
10315         * cfgcleanup.c (old_insns_match_p): Likewise.
10316         * asan.c (asan_finish_file): Likewise.
10317         (asan_instrument): Likewise.
10318         (gate_asan): Likewise.
10319         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
10320         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10321         (asan_global_struct): Use pointer_sized_int_node instead
10322         calling build_nonstandard_integer_type.
10323         (initialize_sanitizer_builtins): Likewise.
10324         (asan_finish_file): Likewise.
10325         * gcc.c: Document %{%:function(args):X}.
10326         (static_spec_functions): Add sanitize.
10327         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
10328         store funcval != NULL there.
10329         (do_spec_1): Adjust handle_spec_function caller.
10330         (handle_braces): Allow %:function(args) as condition.
10331         (sanitize_spec_function): New function.
10332         (ADD_STATIC_LIBUBSAN_LIBS): Define.
10333         (LIBUBSAN_SPEC): Likewise.
10334         (LIBUBSAN_EARLY_SPEC): Likewise.
10335         (SANITIZER_SPEC): Handle libubsan.
10336         (SANITIZER_EARLY_SPEC): Likewise.
10337         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
10338         instead of fsanitize=address.
10339         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
10340         instead of fsanitize=address*.
10341         * builtins.c: Include ubsan.h.
10342         (fold_builtin_0): Instrument __builtin_unreachable.
10343         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
10344         instead of flag_asan.
10345         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
10346         (pointer_sized_int_node): Define.
10347         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
10349 2013-08-30  Mike Stump  <mikestump@comcast.net>
10351         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
10352         with RE patterns.
10354 2013-08-29  Jan Hubicka  <jh@suse.cz>
10356         * cgraph.c (cgraph_function_body_availability): Handle weakref
10357         correctly.
10358         * passes.def: Remove pass_fixup_cfg.
10359         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
10360         track when we need to remove functions.
10361         (gate_ipa_inline): Execute inlining always; add comment why.
10362         (pass_data_ipa_inline): Remove TODO_remove_functions.
10363         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
10364         do not produce summaries.
10365         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
10366         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
10367         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
10368         constant pool and vtable.
10370 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
10372         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
10373         New arm_neon.h's internal macros to specify 64-bit constants.
10374         Avoid using stdint.h's macros.
10376 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10378         * recog.c (verify_changes): Verify that changes[i].old is non-zero
10379         before applying REG_P.
10381 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
10383         PR tree-optimization/58277
10384         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
10385         after seeing too many stmts with vdef in between dombb and current
10386         bb, invalidate everything.
10388 2013-08-30  Richard Biener  <rguenther@suse.de>
10390         * fold-const.c (fold_single_bit_test): Fix overflow test.
10392 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
10394         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
10395         and which can live in a register, always retrieve the value on entry.
10396         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
10397         passed by invisible reference specially.
10398         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
10399         (vt_add_function_parameter): Correctly deal with a parameter passed by
10400         invisible reference.
10402 2013-08-30  Jan Hubicka  <jh@suse.cz>
10404         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
10406 2013-08-30  Richard Biener  <rguenther@suse.de>
10408         PR tree-optimization/58228
10409         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
10410         allow invariant loads in nested loop vectorization.
10412 2013-08-30  Richard Biener  <rguenther@suse.de>
10414         PR tree-optimization/58223
10415         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
10416         (has_anti_or_output_dependence): ... this and adjust to also
10417         look for output dependences.
10418         (mark_nodes_having_upstream_mem_writes): Adjust.
10419         (rdg_flag_uses): Likewise.
10421 2013-08-30  Richard Biener  <rguenther@suse.de>
10423         PR tree-optimization/58010
10424         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
10425         assert that we have a loop-closed PHI.
10427 2013-08-29  Jan Hubicka  <jh@suse.cz>
10429         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
10430         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
10431         * lto-section-in.c (lto_free_function_in_decl_state): New function.
10432         (lto_free_function_in_decl_state_for_node): New function.
10434 2013-08-29  Xinliang David Li  <davidxl@google.com>
10436         * loop-unroll.c (report_unroll_peel): Minor message change.
10437         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
10438         Emit alignment peeling message with default -fopt-info.
10439         (vect_loop_versioning): Emit loop version info message.
10440         * tree-vectorizer.c (vectorize_loops): Minor message change.
10441         (execute_vect_slp): Ditto.
10443 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
10445         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
10446         of the clone from the DECL_NAME of the original function.
10448 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
10450         * passes.c (register_pass): Add overload.
10451         * tree-pass.h (register_pass): Forward declare it.  Add comment.
10453 2013-08-29  Jan Hubicka  <jh@suse.cz>
10455         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
10456         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
10457         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
10458         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10459         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
10460         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
10461         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10462         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
10464 2013-08-29  Teresa Johnson  <tejohnson@google.com>
10466         * dumpfile.c (dump_loc): Output column number.
10467         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
10468         * doc/invoke.texi: Document optall -fopt-info flag.
10469         * profile.c (read_profile_edge_counts): Use new dump framework.
10470         (compute_branch_probabilities): Ditto.
10471         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
10472         when pass not in any opt group.
10473         * pass_manager.h (pass_manager::get_pass_profile): New method.
10474         * value-prof.c (check_counter): Use new dump framework.
10475         (check_ic_target): Ditto.
10476         * coverage.c (get_coverage_counts): Ditto.
10477         (coverage_init): Setup new dump framework.
10479 2013-08-29  Richard Biener  <rguenther@suse.de>
10481         PR tree-optimization/58246
10482         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
10483         handle the dominance check inside a basic-block.
10485 2013-08-29  Richard Biener  <rguenther@suse.de>
10487         PR middle-end/57287
10488         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
10489         of default defs that appear in abnormal PHI nodes.
10491 2013-08-29  Richard Biener  <rguenther@suse.de>
10493         PR tree-optimization/57685
10494         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
10495         single-use operands to avoid exponential complexity.
10497 2013-08-28  Dehao Chen  <dehao@google.com>
10499         * ipa-inline.c (edge_badness): Fix integer underflow.
10501 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
10503         * gtm-builtins.def (_ITM_free): Declare leaf.
10505 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10507         PR target/58067
10508         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
10509         (*tls_local_dynamic_base_64_largepic): Likewise.
10510         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
10511         Remove predicate from call operand.
10512         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
10513         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
10515 2013-08-28  Jeff Law  <law@redhat.com>
10517         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
10518         checks for the number of predecessors and successors allowed.
10519         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
10520         which require copying a joiner block if there is a request which
10521         is a subpath that requires no joiner block copying.
10523 2013-08-28  Jan Hubicka  <jh@suse.cz>
10525         * lto-streamer-out.c (DFS_write_tree_body): Drop
10526         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
10527         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
10528         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
10529         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
10530         Do not read DECL_ERROR_ISSUED.
10531         (unpack_ts_decl_with_vis_value_fields): Do not read
10532         DECL_DEFER_OUTPUT.
10533         (lto_input_ts_binfo_tree_pointers): Do not read
10534         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
10535         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
10536         write DECL_ERROR_ISSUED..
10537         (pack_ts_decl_with_vis_value_fields): Do not write
10538         DECL_DEFER_OUTPUT.
10539         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
10540         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
10541         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
10542         * tree.h (tree_decl_common): Update comment.
10543         (DECL_ERROR_ISSUED): Remove.
10545 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10547         PR middle-end/58257
10548         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
10550 2013-08-28  Jan Hubicka  <jh@suse.cz>
10552         * builtins.def (free): Declare leaf.
10554 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
10556         * gdbhooks.py: New.
10557         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
10558         * configure: Regenerate.
10560 2013-08-27  Martin Jambor  <mjambor@suse.cz>
10562         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
10563         (ipa_ancestor_jf_data): Likewise.
10564         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
10565         (ipa_get_jf_pass_through_type_preserved): New function.
10566         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
10567         (ipa_get_jf_ancestor_type_preserved): New function.
10568         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
10569         (ipa_get_jf_ancestor_result): Likewise.
10570         (propagate_vals_accross_pass_through): Use
10571         ipa_get_jf_pass_through_result to do all the value mappings.
10572         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
10573         type_preserved flag.
10574         (ipa_set_jf_cst_copy): New function.
10575         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
10576         (ipa_set_jf_arith_pass_through): Likewise.
10577         (ipa_set_ancestor_jf): Likewise.
10578         (compute_complex_assign_jump_func): Set type_preserved instead of
10579         punting.
10580         (ipa_compute_jump_functions_for_edge): Likewise.
10581         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
10582         (update_jump_functions_after_inlining): Update type_preserved.
10583         Explicitely create jump functions when combining one with pass_through.
10584         (ipa_write_jump_function): Stream the type_preserved flags.
10585         (ipa_read_jump_function): Likewise.
10587 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
10588             Aldy Hernandez  <aldyh@redhat.com>
10590         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
10591         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
10592         * function.h (struct function): Add has_force_vect_loops and
10593         has_simduid_loops.
10594         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
10595         * gimple.c (gimple_build_omp_critical): Add KIND argument and
10596         handle it.
10597         * gimple.def: Update CLAUSES comments.
10598         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
10599         (gimple_build_omp_for): Add argument to prototype.
10600         (gimple_omp_for_kind): New.
10601         (gimple_omp_for_set_kind): New.
10602         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
10603         GOVD_DATA_SHARE_CLASS.
10604         (enum omp_region_type): Add ORT_SIMD.
10605         (gimple_add_tmp_var): Handle ORT_SIMD.
10606         (gimplify_var_or_parm_decl): Same.
10607         (is_gimple_stmt): Same.
10608         (omp_firstprivatize_variable): Same.
10609         (omp_add_variable): Only use splay_tree_insert if lookup failed.
10610         (omp_notice_variable): Handle ORT_SIMD.
10611         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
10612         (omp_check_private): Handle ORT_SIMD.
10613         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10614         OMP_CLAUSE_SAFELEN.
10615         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
10616         Handle OMP_CLAUSE_LASTPRIVATE.
10617         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10618         OMP_CLAUSE_SAFELEN.
10619         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
10620         (gimplify_expr): Handle OMP_SIMD.
10621         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
10622         (expand_GOMP_SIMD_VF): New.
10623         (expand_GOMP_SIMD_LAST_LANE): New.
10624         * internal-fn.def (GOMP_SIMD_LANE): New.
10625         (GOMP_SIMD_VF): New.
10626         (GOMP_SIMD_LAST_LANE): New.
10627         * omp-low.c: Include target.h.
10628         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
10629         OMP_CLAUSE_SAFELEN.
10630         (check_omp_nesting_restrictions): Same.
10631         (omp_max_vf): New.
10632         (lower_rec_simd_input_clauses): New.
10633         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
10634         OMP_CLAUSE_LINEAR.
10635         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
10636         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
10637         (expand_omp_build_assign): New.
10638         (expand_omp_for_init_counts): New.
10639         (expand_omp_for_init_vars): New.
10640         (extract_omp_for_update_vars): New.
10641         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
10642         and rewrite accordingly.
10643         (expand_omp_simd): New.
10644         (expand_omp_for): Use expand_omp_simd.
10645         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
10646         (lower_omp_for): Do not lower the body.
10647         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
10648         in their own loops.
10649         * tree-flow.h (find_omp_clause): Remove prototype.
10650         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
10651         forcing vectorization of the loop, or if flag_tree_vectorize.
10652         (gate_tree_if_conversion): Similarly.
10653         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
10654         gimple_build_omp_for.
10655         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
10656         * tree-parloops (create_parallel_loop): Pass kind argument to
10657         gimple_build_omp_for.
10658         * tree-pretty-print.c (dump_omp_clause): Add cases for
10659         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
10660         OMP_CLAUSE__SIMDUID_.
10661         (dump_generic_node): Handle OMP_SIMD.
10662         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
10663         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
10664         unroll OMP_SIMD loops here.
10665         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
10666         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
10667         loop->safelen.
10668         (vect_analyze_data_refs): Handle simd loops.
10669         * tree-vect-loop.c (vectorizable_live_operation): Handle
10670         IFN_GOMP_SIMD*.
10671         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
10672         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
10673         (vectorizable_load): Same.
10674         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
10675         (struct simduid_to_vf): New.
10676         (simduid_to_vf::hash): New.
10677         (simduid_to-vf::equal): New.
10678         (struct simd_array_to_simduid): New.
10679         (simd_array_to_simduid::hash): New.
10680         (simd_array_to_simduid::equal): New.
10681         (adjust_simduid_builtins): New.
10682         (struct note_simd_array_uses_struct): New.
10683         (note_simd_array_uses_cb): New.
10684         (note_simd_array_uses): New.
10685         (vectorize_loops): Handle simd hints and adjust simd builtins
10686         accordingly.
10687         * tree-vectorizer.h (struct _stmt_vec_info): Add
10688         simd_lane_access_p field.
10689         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
10690         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
10691         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
10692         (omp_clause_code_name): Same.
10693         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
10694         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
10695         * tree.def (OMP_SIMD): New entry.
10696         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
10697         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
10698         (OMP_CLAUSE_DECL): Adjust range for new clauses.
10699         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
10700         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
10701         (OMP_CLAUSE_LINEAR_STEP): New.
10702         (OMP_CLAUSE_SAFELEN_EXPR): New.
10703         (OMP_CLAUSE__SIMDUID__DECL): New.
10704         (find_omp_clause): New prototype.
10706 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
10708         * config/i386/driver-i386.c (host_detect_local_cpu): Update
10709         Haswell processor detection.
10711 2013-08-27  Christian Widmer  <shadow@umbrox.de>
10713         PR target/57927
10714         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
10715         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
10716         AVX2 capable processors.
10718 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
10720         * config/aarch64/arm_neon.h: Replace all inline asm implementations
10721         of vget_low_* with implementations in terms of other intrinsics.
10723 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
10725         PR middle-end/57219
10726         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
10727         values to -1, 0 and 1.
10729 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
10731         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
10732         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
10733         (<optab><mode>3_insn): Remove.
10734         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
10735         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
10736         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
10737         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
10738         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
10739         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
10740         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
10741         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
10742         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
10743         (ror<mode>3_insn): Likewise.
10744         * config/aarch64/predicates.md (aarch64_simd_register): New.
10746 2013-08-27  Richard Biener  <rguenther@suse.de>
10748         PR tree-optimization/57521
10749         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
10750         one edge is non-critical.
10751         (find_phi_replacement_condition): Make sure to use a non-critical
10752         edge.  Cleanup and remove old bug workarounds.
10753         (bb_postdominates_preds): Remove.
10754         (if_convertible_loop_p_1): Do not compute post-dominators.
10755         (combine_blocks): Do not free post-dominators.
10756         (main_tree_if_conversion): Likewise.
10757         (pass_data_if_conversion): Add TODO_verify_ssa.
10759 2013-08-27  DJ Delorie  <dj@redhat.com>
10761         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
10763 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
10765         * function.c (assign_parm_find_data_types): Set passed_mode and
10766         nominal_mode to the TYPE_MODE of nominal_type for the built
10767         pointer type in case of the struct-pass-by-reference.
10769 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10771         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
10772         (UINT16_TYPE): Change default to "unsigned int".
10774         * config/avr/avr.opt (mfract-convert-truncate): New option.
10775         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
10776         is set, round negative fractional integers according to n1169
10777         when converting to integer types.
10779 2013-08-26  Jan Hubicka  <jh@suse.cz>
10781         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
10782         methods can not be called indirectly when their address is not taken.
10784 2013-08-26  Jan Hubicka  <jh@suse.cz>
10786         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
10787         ctor_for_folding.
10789 2013-08-26  Jan Hubicka  <jh@suse.cz>
10791         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
10792         can be unshared.
10794 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10796         * reload.c (find_valid_class): Allow classes that do not include
10797         FIRST_PSEUDO_REGISTER - 1.
10799 2013-08-26  Jan Hubicka  <jh@suse.cz>
10801         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
10802         fix edge count/frequency when speculation failed; fix type check
10803         for the direct call.
10805 2013-08-26  Jan Hubicka  <jh@suse.cz>
10807         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
10809 2013-08-26  Jan Hubicka  <jh@suse.cz>
10811         * ipa-inline-transform.c (inline_transform): Be ready for basic block
10812         to be changed by edge redirection.
10814 2013-08-26  Jan Hubicka  <jh@suse.cz>
10816         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
10817         formating; add sanity check.
10818         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
10819         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
10820         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
10821         (resolve_noninline_speculation): Update callee keys, too.
10823 2013-08-26  Jan Hubicka  <jh@suse.cz>
10825         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
10826         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
10828 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
10830         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
10831         into proper place.
10833 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
10835         * config/i386/i386.c (ix86_debug_options): Remove prototype.
10836         (x86_64_elf_select_section): Ditto.
10837         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
10838         arguments.
10839         (ix86_pass_by_reference): Ditto.
10840         (output_set_got): Ditto.
10841         (ix86_unary_operator_ok): Ditto.
10842         (ix86_expand_builtin): Ditto.
10844 2013-08-23  Jan Hubicka  <jh@suse.cz>
10846         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
10848 2013-08-23  Jan Hubicka  <jh@suse.cz>
10850         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
10851         (tree_decl_with_vis): Add FINAL field.
10853 2013-08-23  Jeff Law  <law@redhat.com>
10855         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
10856         the debugging dump when the expression is fully redundant.
10858 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10860         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
10861         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
10862         * pretty-print.c (pp_formatted_text_data): Likewise.
10863         (pp_write_text_to_stream): Likewise.
10864         (pp_write_text_as_dot_label_to_stream): Likewise.
10865         (pp_append_r): Likewise.
10866         (pp_format): Likewise.
10867         (pp_flush): Likewise.
10868         (pp_clear_output_area): Likewise.
10869         (pp_append_text): Likewise.
10870         (pp_formatted_text): Likewise.
10871         (pp_remaining_character_count_for_line): Likewise.
10872         (pp_newline): Likewise.
10873         (pp_character): Likewise.
10874         (output_buffer::~output_buffer): Define.
10875         (pretty_printer::~pretty_printer): Destruct output buffer.
10876         * pretty-print.h (output_buffer::~output_buffer): Declare.
10877         (pretty_printer::~pretty_printer): Declare virtual.
10879 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
10881         PR other/57324
10882         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
10883         HOST_WIDE_INT_M1U): New macros.
10884         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
10885         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
10886         unsigned -1 for lshift.
10887         * cse.c (cse_insn): Likewise.
10888         * double-int.c (rshift_double, lshift_double): Likewise.
10889         * builtins.c (fold_builtin_bitop): Likewise.
10890         * combine.c (force_to_mode): Likewise.
10891         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
10892         * simplify-rtx.c (simplify_const_unary_operation,
10893         simplify_const_binary_operation): Likewise.
10894         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
10895         check_va_list_escapes): Likewise.
10896         * rtlanal.c (nonzero_bits1): Likewise.
10897         * expmed.c (expand_smod_pow2): Likewise.
10898         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
10900 2013-08-23  Jan Hubicka  <jh@suse.cz>
10902         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
10903         as having address taken.
10905 2013-08-23  Jan Hubicka  <jh@suse.cz>
10907         * ipa-utils.h (method_class_type): Declare.
10908         * ipa-devirt.c (method_class_type): Export.
10910         * cgraphunit.c (analyze_functions): Do basic devirtualization;
10911         do not walk base classes of anonymous types.
10913 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
10915         PR rtl-optimization/58220
10916         PR regression/58221
10917         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
10918         handle SEQUENCE insns properly.
10920 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10922         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
10923         definition.
10924         (pp_newline_and_indent): Likewise.
10925         (pp_separate_with): Likewise.
10926         * pretty-print.c (pp_newline_and_flush): Define.
10927         (pp_newline_and_indent): Likewise.
10928         (pp_separate_with): Likewise.
10930 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
10932         PR target/58218
10933         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
10934         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
10936 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
10938         * config/i386/predicates.md (ext_sse_reg_operand): New.
10939         * config/i386/i386.md (*movti_internal): Use
10940         predicate to determine if EVEX is needed.
10941         (*movsi_internal): Ditto.
10942         (*movdf_internal): Ditto.
10943         (*movsf_internal): Ditto.
10944         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
10946 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
10948         PR tree-optimization/58209
10949         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
10950         (find_tail_calls): Give up for pointer result types if m is non-NULL.
10951         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
10952         emit POINTER_PLUS_EXPR.
10953         (create_tailcall_accumulator): For pointer result type accumulate in
10954         sizetype type.
10956 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
10958         * configure.ac: Add backslashes missing from the last change.
10959         * configure: Regenerate.
10961 2013-08-22  Jan Hubicka  <jh@susue.cz>
10963         * ipa.c (function_and_variable_visibility): First remember function
10964         was global and then make it local.
10966 2013-08-22  Julian Brown  <julian@codesourcery.com>
10968         * configure.ac: Add aarch64 to list of arches which use "nop" in
10969         debug_line test.
10970         * configure: Regenerate.
10972 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10974         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
10975         gnu_libc_has_function.
10976         * config/s390/tpf.h: Likewise.
10978 2013-08-22  Jan Hubicka  <jh@susue.cz>
10980         * timevar.c (validate_phases): Add cast.
10982 2013-08-22  Jan Hubicka  <jh@susue.cz>
10984         * timevar.c (validate_phases): Use size_t for memory.
10985         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
10987 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10989         * pretty-print.h (output_buffer::output_buffer): Declare.
10990         (pretty_printer::pretty_printer): Likewise.
10991         (pp_construct): Remove.
10992         * pretty-print.c (output_buffer::output_buffer): Define.
10993         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
10994         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
10995         (print_gimple_expr): Likewise.
10996         (print_gimple_seq): Likewise.
10997         (gimple_dump_bb): Likewise.
10998         * sched-vis.c (dump_value_slim): Likewise.
10999         (dump_insn_slim): Likewise.
11000         (dump_rtl_slim): Likewise.
11001         (str_pattern_slim): Likewise.
11002         * tree-mudflap.c (mf_varname_tree): Likewise.
11003         * graph.c (print_graph_cfg): Likewise.
11004         (start_graph_dump): Likewise.
11005         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
11006         placement-new.
11007         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
11008         pretty printer initialization.
11009         * coretypes.h (diagnostic_context): Remove superflous type alias
11010         declaration.
11011         (pretty_printer): Likewise.  Declare directly as a class.
11012         (pretty_print_info): Remove declaration as class.
11013         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
11014         and pp_clear_output_area.
11015         (asan_add_global): Likewise.
11017 2013-08-22  Jan Hubicka  <jh@suse.cz>
11019         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
11020         * ipa-utils.h (update_type_inheritance_graph): Declare.
11021         (possible_polymorphic_call_target_p): Declare.
11022         (possible_polymorphic_call_target_p): New.
11023         * ipa-devirt.c: Update toplevel comments.
11024         (cached_polymorphic_call_targets): Move up.
11025         (odr_type_d): Move ID down.
11026         (polymorphic_type_binfo_p): Update comment.
11027         (odr_hasher::remove): Likewise;
11028         (get_odr_type): Set anonymous_namespace.
11029         (dump_odr_type): Dump it.
11030         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
11031         (maybe_record_node): Record node in cached_polymorphic_call_targets.
11032         (record_binfo): Add comment.
11033         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
11034         built.
11035         (devirt_node_removal_hook): Do not iCE when cache is freed.
11036         (possible_polymorphic_call_target_p): New predicate.
11037         (update_type_inheritance_graph): New function.
11039 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11040             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
11041             Sergey Lega  <sergey.s.lega@intel.com>
11042             Anna Tikhonova  <anna.tikhonova@intel.com>
11043             Ilya Tocar  <ilya.tocar@intel.com>
11044             Andrey Turetskiy  <andrey.turetskiy@intel.com>
11045             Ilya Verbin  <ilya.verbin@intel.com>
11046             Kirill Yukhin  <kirill.yukhin@intel.com>
11047             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11049         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
11050         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
11051         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
11052         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
11053         (OPTION_MASK_ISA_AVX2_UNSET): Update.
11054         (OPTION_MASK_ISA_AVX512F_UNSET): New.
11055         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
11056         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
11057         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
11058         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
11059         OPT_mavx512pf, OPT_mavx512er cases.
11060         * config/i386/constraints.md (v): New constraint.
11061         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
11062         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
11063         (bit_AVX512CD): New.
11064         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11065         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
11066         * config/i386/i386-c.c (ix86_target_macros_internal):
11067         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
11068         __AVX512PF__.
11069         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
11070         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
11071         * config/i386/i386.c (regclass_map, dbx_register_map)
11072         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
11073         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
11074         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
11075         -mavx512pf options.
11076         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
11077         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
11078         -mavx512pf options.  Fix formatting.
11079         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
11080         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
11081         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
11082         -mavx512cd, -mavx512pf options.
11083         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
11084         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
11085         (ix86_preferred_output_reload_class): Replace SSE_REGS with
11086         ALL_SSE_REGS.
11087         (ix86_hard_regno_mode_ok): Support 512-bit registers.
11088         (ix86_set_reg_reg_cost): Ditto.
11089         (x86_order_regs_for_local_alloc): Ditto.
11090         (MAX_VECT_LEN): Extend to 64-byte.
11091         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
11092         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
11093         (TARGET_AVX512ER, TARGET_AVX512CD): New.
11094         (BIGGEST_ALIGNMENT): Extend to 512-bits.
11095         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
11096         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
11097         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
11098         (SSE_REG_MODE_P): Support new modes.
11099         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
11100         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
11101         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
11102         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
11103         (REG_CLASS_CONTENTS): Add new registers.
11104         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
11105         (EXT_REX_SSE_REGNO_P): New.
11106         (HI_REGISTER_NAMES): Add new registers.
11107         * config/i386/i386.md: Define constants for new registers.
11108         (mode): Add new 512-bit modes.
11109         (prefix): Support evex prefix.
11110         (isa): Support avx512f, noavx512f, fma_avx512f.
11111         (ssemodesuffix): Add new 512-bit modes.
11112         (movxi): New.
11113         (*movxi_internal_avx512f): Ditto.
11114         (*movdi_internal): Replace constraint "x" with the new constraint "v".
11115         Support MODE_XI.
11116         (*movsi_internal): Likewise.
11117         (*movdf_internal): Likewise.
11118         (*movsf_internal): Likewise.
11119         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
11120         (<code><mode>3): Likewise.
11121         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
11122         New.
11123         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
11124         with the new constraint "v".
11125         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
11126         modes.
11127         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
11128         with the new constraint "v".
11129         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
11130         (<sse2>_storedqu<avxsizesuffix>): Likewise.
11131         (*<plusminus_insn><mode>3): Likewise.
11132         (<sse>_vm<plusminus_insn><mode>3): Likewise.
11133         (*mul<mode>3): Likewise.
11134         (<sse>_vmmul<mode>3): Likewise.
11135         (<sse>_div<mode>3): Likewise.
11136         (<sse>_vmdiv<mode>3): Likewise.
11137         (<sse>_sqrt<mode>2): Likewise.
11138         (<sse>_vmsqrt<mode>2): Likewise.
11139         (*<code><mode>3_finite): Likewise.
11140         (*<code><mode>3) <smaxmin>: Likewise.
11141         (<sse>_vm<code><mode>3): Likewise.
11142         (*<code><mode>3) <any_logic>: Likewise.
11143         (*fma_fmadd_<mode>): Likewise.
11144         (*fma_fmsub_<mode>): Likewise.
11145         (*fma_fnmadd_<mode>): Likewise.
11146         (*fma_fnmsub_<mode>): Likewise.
11147         (*fma_fmaddsub_<mode>): Likewise.
11148         (*fma_fmsubadd_<mode>): Likewise.
11149         (*fmai_fmadd_<mode>): Likewise.
11150         (*fmai_fmsub_<mode>): Likewise.
11151         (*fmai_fnmadd_<mode>): Likewise.
11152         (*fmai_fnmsub_<mode>): Likewise.
11153         (sse_cvtsi2ss): Likewise.
11154         (sse_cvtsi2ssq): Likewise.
11155         (sse_cvtss2si): Likewise.
11156         (sse_cvtss2si_2): Likewise.
11157         (sse_cvtss2siq): Likewise.
11158         (sse_cvtss2siq_2): Likewise.
11159         (sse_cvttss2si): Likewise.
11160         (sse_cvtss2siq_2): Likewise.
11161         (float<sseintvecmodelower><mode>2): Likewise.
11162         (sse2_cvtsd2si_2): Likewise.
11163         (sse2_cvtsd2siq_2): Likewise.
11164         (*<plusminus_insn><mode>3): Likewise.
11165         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
11166         (*<sse4_1_avx2>_mul<mode>3): Likewise.
11167         (ashr<mode>3): Likewise.
11168         (<shift_insn><mode>3): Likewise.
11169         (avx2_<code><mode>3): Likewise.
11170         (*avx2_<code><mode>3): Likewise.
11171         (*andnot<mode>3): Likewise.
11172         (*<code><mode>3) <any_logic>: Likewise.
11173         (abs<mode>2): Likewise.
11174         (avx2_permvar<mode>): Likewise.
11175         (avx2_perm<mode>_1): Likewise.
11176         (*avx_vpermilp<mode>): Likewise.
11177         (avx_vpermilvar<mode>3): Likewise.
11178         (avx2_ashrv<mode>): Likewise.
11179         (avx2_<shift_insn>v<mode>): Likewise.
11180         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
11181         -mavx512cd.
11182         * doc/rtl.texi: Document XImode.
11184 2013-08-21  Jeff Law  <law@redhat.com>
11186         * tree-flow.h (register_jump_thread): Pass vector of edges
11187         instead of each important edge.
11188         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
11189         thread path into a vector and pass that to register_jump_thread.
11190         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
11191         passed in edge vector to the current 3-edge form.
11193         Revert:
11194         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11196         * dce.c (fini_dce): Call df_analyze again just in case
11197         delete_unmarked_insns removed anything.
11199 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
11201         * reload.h (struct reg_equivs): Rename to ..
11202         (struct reg_equivs_s): .. this.
11204 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
11206         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
11208 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11210         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
11212 2013-08-21  Jeff Law  <law@redhat.com>
11214         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
11215         simplify assignments too.  If the RHS collapses to a singleton
11216         range, then return the value for the range.
11218 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
11220         * config/i386/sse.md (V16): Rename to...
11221         (VMOVE): this.
11222         (mov<mode>): Update iterator name.
11223         (*mov<mode>_internal): Ditto.
11224         (push<mode>1): Ditto.
11225         (movmisalign<mode>): Ditto.
11227 2013-08-20  Jan Hubicka  <jh@suse.cz>
11229         PR bootstrap/58186
11230         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
11231         entry for direct edges.
11232         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
11234 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11236         Revert my last two changes, r201865 and r201864:
11238         Revert r201865:
11239         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11241         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11242         instances can own GC refs.
11244         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11245         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11246         (gcc::context::gt_pch_nx): Likewise.
11247         (gcc::context::gt_pch_nx):  Likewise.
11248         * ggc.h (gt_ggc_mx <T>): New.
11249         (gt_pch_nx_with_op <T>): New.
11250         (gt_pch_nx <T>): New.
11251         * passes.c (opt_pass::gt_ggc_mx): New.
11252         (opt_pass::gt_pch_nx): New.
11253         (opt_pass::gt_pch_nx_with_op): New.
11254         (pass_manager::gt_ggc_mx): New.
11255         (pass_manager::gt_pch_nx): New.
11256         (pass_manager::gt_pch_nx_with_op): New.
11257         (pass_manager::operator new): Use
11258         ggc_internal_cleared_alloc_stat rather than xcalloc.
11259         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11260         (pass_manager::gt_ggc_mx): New.
11261         (pass_manager::gt_pch_nx): New.
11262         (pass_manager::gt_pch_nx_with_op): New.
11263         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11264         (opt_pass::operator new): New.
11265         (opt_pass::gt_ggc_mx): New.
11266         (opt_pass::gt_pch_nx): New.
11267         (opt_pass::gt_pch_nx_with_op): New.
11269         Revert r201864:
11270         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11272         * Makefile.in (GTFILES): Add context.h.
11273         * context.c (gcc::context::operator new): New.
11274         (gcc::context::gt_ggc_mx): New.
11275         (gcc::context::gt_pch_nx): New.
11276         (gcc::context::gt_pch_nx): New.
11277         * context.h (gcc::context): Add GTY((user)) marking.
11278         (gcc::context::operator new): New.
11279         (gcc::context::gt_ggc_mx): New.
11280         (gcc::context::gt_pch_nx): New.
11281         (gcc::context::gt_pch_nx): New.
11282         (g): Add GTY marking.
11283         (gt_ggc_mx (gcc::context *)): New.
11284         (gt_pch_nx (gcc::context *)): New.
11285         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11286         void *cookie)): New.
11287         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11289 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11291         * dce.c (fini_dce): Call df_analyze again just in case
11292         delete_unmarked_insns removed anything.
11294 2013-08-20  Teresa Johnson  <tejohnson@google.com>
11296         PR rtl-optimizations/57451
11297         * final.c (reemit_insn_block_notes): Prevent lexical blocks
11298         from crossing split section boundaries.
11300 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
11302         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
11303         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
11304         with MULTLIB_DEFAULTS.
11306 2013-08-20  Nick Clifton  <nickc@redhat.com>
11308         * target.def (narrow_volatile_bitfield): Note that the default
11309         value is false, not !TARGET_STRICT_ALIGN.
11310         * doc/tm.texi: Regenerate.
11312 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
11314         Fix LIB_SPEC for systems without libpthread.
11316         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
11317         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
11318         for Android.
11319         * config/i386/linux-common.h: Likewise.
11320         * config/mips/linux-common.h: Likewise.
11322 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
11324         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
11325         checks.
11327 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11329         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11330         instances can own GC refs.
11332         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11333         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11334         (gcc::context::gt_pch_nx): Likewise.
11335         (gcc::context::gt_pch_nx):  Likewise.
11336         * ggc.h (gt_ggc_mx <T>): New.
11337         (gt_pch_nx_with_op <T>): New.
11338         (gt_pch_nx <T>): New.
11339         * passes.c (opt_pass::gt_ggc_mx): New.
11340         (opt_pass::gt_pch_nx): New.
11341         (opt_pass::gt_pch_nx_with_op): New.
11342         (pass_manager::gt_ggc_mx): New.
11343         (pass_manager::gt_pch_nx): New.
11344         (pass_manager::gt_pch_nx_with_op): New.
11345         (pass_manager::operator new): Use
11346         ggc_internal_cleared_alloc_stat rather than xcalloc.
11347         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11348         (pass_manager::gt_ggc_mx): New.
11349         (pass_manager::gt_pch_nx): New.
11350         (pass_manager::gt_pch_nx_with_op): New.
11351         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11352         (opt_pass::operator new): New.
11353         (opt_pass::gt_ggc_mx): New.
11354         (opt_pass::gt_pch_nx): New.
11355         (opt_pass::gt_pch_nx_with_op): New.
11357 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11359         * Makefile.in (GTFILES): Add context.h.
11360         * context.c (gcc::context::operator new): New.
11361         (gcc::context::gt_ggc_mx): New.
11362         (gcc::context::gt_pch_nx): New.
11363         (gcc::context::gt_pch_nx): New.
11364         * context.h (gcc::context): Add GTY((user)) marking.
11365         (gcc::context::operator new): New.
11366         (gcc::context::gt_ggc_mx): New.
11367         (gcc::context::gt_pch_nx): New.
11368         (gcc::context::gt_pch_nx): New.
11369         (g): Add GTY marking.
11370         (gt_ggc_mx (gcc::context *)): New.
11371         (gt_pch_nx (gcc::context *)): New.
11372         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11373         void *cookie)): New.
11374         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11376 2013-08-20  Alan Modra  <amodra@gmail.com>
11378         PR target/57865
11379         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
11380         (rs6000_emit_epilogue): Likewise.
11382 2013-08-19  Dehao Chen  <dehao@google.com>
11384         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
11386 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
11387             Jakub Jelinek  <jakub@redhat.com>
11389         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
11390         (BUILT_IN_FABSD64): Likewise.
11391         (BUILT_IN_FABSD128): Likewise.
11392         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
11393         (fold_builtin_1): Likewise.
11394         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
11395         destination and source operands.
11396         (*abstd2_fpr): Likewise.
11397         (*nabstd2_fpr): Likewise.
11399 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
11401         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
11402         JUMP_P and INSN_P.
11404 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
11406         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
11407         incomplete.
11409 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11411         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
11412         * builtins.c (default_libc_has_function): New.
11413         (gnu_libc_has_function): Ditto.
11414         (no_c99_libc_has_function): Ditto.
11415         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
11416         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
11417         (fold_builtin_sincos): Likewise.
11418         (fold_builtin_cexp): Likewise.
11419         * builtins.def (DEF_C94_BUILTIN): Likewise.
11420         (DEF_C99_BUILTIN): Likewise.
11421         (DEF_C99_C90RES_BUILTIN): Likewise.
11422         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
11423         definitions to using this define.
11424         * config/darwin-protos.h (darwin_libc_has_function): New.
11425         * config/darwin.c (darwin_libc_has_function): Ditto.
11426         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
11427         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
11428         * config/darwin.h: Ditto.
11429         * config/elfos.h: Ditto.
11430         * config/freebsd.h: Ditto.
11431         * config/i386/cygming.h: Ditto.
11432         * config/i386/djgpp.h: Ditto.
11433         * config/i386/i386-interix.h: Ditto.
11434         * config/microblaze/microblaze.h: Ditto.
11435         * config/mmix/mmix.h: Ditto.
11436         * config/gnu-user.h: Ditto.
11437         * config/ia64/hpux.h: Ditto.
11438         * config/pa/pa-hpux.h: Ditto.
11439         * config/pdp11/pdp11.h: Ditto.
11440         * config/picochip/picochip.h: Ditto.
11441         * config/linux.h: Ditto.
11442         * config/netbsd.h: Ditto.
11443         * config/openbsd.h: Ditto.
11444         * config/rs6000/aix43.h: Ditto.
11445         * config/rs6000/aix51.h: Ditto.
11446         * config/rs6000/aix52.h: Ditto.
11447         * config/rs6000/aix53.h: Ditto.
11448         * config/rs6000/aix61.h: Ditto.
11449         * config/rs6000/darwin.h: Ditto.
11450         * config/rs6000/linux.h: Ditto.
11451         * config/rs6000/linux64.h: Ditto.
11452         * config/s390/tpf.h: Ditto.
11453         * config/sol2-10.h: Ditto.
11454         * config/sol2.h: Ditto.
11455         * config/vms/vms.h: Ditto.
11456         * config/vxworks.h: Ditto.
11457         * config/linux-android.c (linux_android_libc_has_function):
11458         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
11459         * config/linux-protos.h (linux_android_libc_has_function):
11460         New declaration.
11461         * config/i386/i386.c (ix86_libc_has_function): New.
11462         * config/i386/i386-protos.h
11463         (ix86_libc_has_function): New declaration.
11464         * config/i386/i386.md
11465         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
11466         ("isinf<mode>2): Likewise.
11467         * convert.c (convert_to_integer): Using new target hook
11468         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
11469         TARGET_C99_FUNCTIONS.
11470         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
11471         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
11472         * coretypes.h (function_class): New enum for different
11473         classes of functions.
11474         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
11475         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
11476         (TARGET_HAS_SINCOS): Likewise.
11477         (TARGET_LIBC_HAS_FUNCTION): New.
11478         * doc/tm.texi: Regenerated.
11479         * targhooks.h (default_libc_has_function): New declaration.
11480         (no_c99_libc_has_function): Ditto.
11481         (gnu_libc_has_function): Ditto.
11482         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
11483         and TARGET_HAS_SINCOS.
11485 2013-08-18  Jan Hubicka  <jh@suse.cz>
11487         * Makeifle-in (ipa-devirt.o): New.
11488         (GTFILES): Add ipa-utils.h and ipa-devirt.c
11489         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
11490         (analyze_functions): Look into possible targets of polymorphic call.
11491         * dumpfile.c (dump_files): Add type-inheritance dump.
11492         * dumpfile.h (TDI_inheritance): New.
11493         * ipa-devirt.c: New file.
11494         * ipa-utils.h (odr_type_d): Forward declare.
11495         (odr_type): New type.
11496         (build_type_inheritance_graph): Declare.
11497         (possible_polymorphic_call_targets): Declare and introduce inline
11498         variant when only edge is pased.
11499         (dump_possible_polymorphic_call_targets): Likewise.
11500         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
11501         * tree.c (type_in_anonymous_namespace_p): Break out from ...
11502         (types_same_for_odr): ... here.
11503         * tree.h (type_in_anonymous_namespace_p): Declare.
11505 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
11507         PR tree-optimization/58006
11508         * tree-parloops.c (take_address_of): Don't ICE if get_name
11509         returns NULL.
11510         (eliminate_local_variables_stmt): Remove clobber stmts.
11512 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
11514         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
11515         error message is issued for an alias to undefined symbol.
11517 2013-08-18  Jan Hubicka  <jh@suse.cz>
11519         * cgraph.c (cgraph_create_indirect_edge): Discover
11520         polymorphic calls and record basic info into indirect_info.
11521         * gimple-fold.c (gimple_fold_call): When doing BINFO based
11522         devirtualization, ignore objc function calls.
11523         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
11524         call with no parm index info.
11525         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
11526         * tree.c (virtual_method_call_p): New function.
11527         * tree.h (virtual_method_call_p): Declare.
11529 2013-08-16  Jan Hubicka  <jh@suse.cz>
11531         PR middle-end/58179
11532         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
11534 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
11536         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
11537         attribute.
11539 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11541         * gengtype.c (type_for_name): Add special-case support for
11542         locating types within the "gcc::" namespace.
11543         (open_base_files): Emit a "using namespace gcc" directive.
11545 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
11547         PR target/58160
11548         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
11549         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
11551         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
11552         array instead of each individual operand as a separate argument.
11553         (emit_fusion_gpr_load): Likewise.
11554         (expand_fusion_gpr_load): Add new function declaration.
11556         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
11557         signature to have the operands passed as an array, instead of as
11558         separate arguments.  Allow ZERO_EXTEND to be in the memory
11559         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
11560         depend on the register live/dead flags when peepholes are run.
11561         (expand_fusion_gpr_load): New function to be called from the
11562         peephole2 pass, to change the register that addis sets to be the
11563         target register.
11564         (emit_fusion_gpr_load): Change the calling signature to have the
11565         operands passed as an array, instead of as separate arguments.
11566         Allow ZERO_EXTEND to be in the memory address, and also
11567         SIGN_EXTEND if -mpower8-fusion-sign.
11569         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
11570         unspec enumeration.
11571         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
11572         adjust the register addis loads up in the peephole2 pass.  Do not
11573         depend on the register live/dead state when the peephole pass is done.
11575 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11577         * gengtype.c (create_user_defined_type): Ensure that the kind
11578         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
11579         declaration is seen before the GTY((user)) marking.
11581 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
11583         PR target/58105
11584         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
11586 2013-08-16  Jan Hubicka  <jh@suse.cz>
11588         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
11589         arugment expected_type.
11590         (gimple_fold_call): Use it.
11591         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
11592         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
11593         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
11594         (try_make_edge_direct_virtual_call): Likewise.
11595         * tree.c (obj_type_ref_class): New.
11596         * tree.h (obj_type_ref_class): Use it.
11598 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11600         * sched-vis.c (rtl_slim_pp_initialized): Remove.
11601         (rtl_slim_pp): Likewise.
11602         (init_rtl_slim_pretty_print): Likewise.
11603         (dump_value_slim):  Don't call it.  Use local pretty printer.
11604         (dump_insn_slim): Likewise.
11605         (dump_rtl_slim): Likewise.
11606         (str_pattern_slim): Likewise.
11607         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
11608         Simplify.
11610 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
11612         PR tree-optimization/58164
11613         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
11614         walk gimple_goto_dest of GIMPLE_GOTO.
11616         PR tree-optimization/58165
11617         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
11618         bi_call must be the last stmt in a bb, don't split_block, instead
11619         use fallthru edge from it and give up if there is none.
11620         Release conds vector when returning early.
11622 2013-08-14  Xinliang David Li  <davidxl@google.com>
11624         * config/i386/i386.c (ix86_option_override_internal):
11625         Remove unused variable and field.
11627 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11629         PR target/57949
11630         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
11631         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
11632         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
11633         and Linux, correct BLKmode alignment when 128-bit alignment is
11634         required and compatibility flag is not set.
11635         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
11636         for zero-size arguments when compatibility flag is not set.
11638 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11640         PR tree-optimization/58145
11641         * tree-sra.c (build_ref_for_offset): If prev_base has
11642         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
11644 2013-08-14  Xinliang David Li  <davidxl@google.com>
11646         * config/i386/i386.c (ix86_option_override_internal):
11647         Fix uninitialized variable error.
11649 2013-08-14  Xinliang David Li  <davidxl@google.com>
11651         * config/i386/i386.opt: Define two new options.
11652         * config/i386/x86-tune.def: Add arch selector field in macros.
11653         * config/i386/i386.h: Adjust macro definition.
11654         * config/i386/i386.c (ix86_option_override_internal):
11655         Refactor the code.
11656         (parse_mtune_ctrl_str): New function.
11657         (set_ix86_tune_features): New function.
11658         (ix86_function_specific_restore): Call the new helper function.
11660 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
11662         PR rtl-optimization/57662
11663         * sel-sched.c (code_motion_process_successors): When the current insn
11664         is removed after the recursive traversal, break from the loop.
11665         Add comments and debug printouts.
11667 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11668             Alexandre Oliva  <aoliva@redhat.com>
11670         PR target/58067
11671         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
11672         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
11673         there also UNSPEC_PLTOFF.
11675 2013-08-14  Marek Polacek  <polacek@redhat.com>
11677         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
11678         NUM_CONDITIONS bit positions.
11680 2013-08-13  Cary Coutant  <ccoutant@google.com>
11682         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
11683         (attr_checksum): Hash vector contents instead of pointer.
11684         (attr_checksum_ordered): Likewise.
11686 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
11688         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
11689         when Pmode != word_mode.  Add length_address attribute.
11690         (sse3_monitor_<mode>): Merge from sse3_monitor and
11691         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
11692         Pmode != word_mode.  Update insn length attribute.
11693         * config/i386/i386.c (ix86_option_override_internal): Update
11694         ix86_gen_monitor selection for merged sse3_monitor insn.
11696 2013-08-13  Julian Brown  <julian@codesourcery.com>
11698         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
11699         perform invalid legitimization on greater-than-word-size modes for
11700         TARGET_E500_DOUBLE.
11702 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
11704         * ira.c (setup_class_translate_array): Use aclass instead of cl
11705         for classes not fully covered by allocno classes.
11707 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
11709         PR tree-optimization/57661
11710         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
11711         * tree-inline.c (tree_function_versioning): Initialize it.
11712         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
11713         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
11714         that is not being copied.
11716         PR sanitizer/56417
11717         * asan.c (instrument_strlen_call): Fix typo in comment.
11718         Use char * type even for the lhs of POINTER_PLUS_EXPR.
11720 2013-08-13  Steve Ellcey  <sellcey@mips.com>
11722         * config/mips/mips.md (prefetch): Use lw instead of ld on
11723         loongson in 32bit mode.
11725 2013-08-13  Nick Clifton  <nickc@redhat.com>
11727         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
11729 2013-08-13  Jan Hubicka  <jh@suse.cz>
11731         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
11732         introduced edge; fix typo in sanity check.
11733         (cgraph_resolve_speculation): Export; improve diagnostic.
11734         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
11735         speculation at type mismatch.
11736         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
11737         (cgraph_resolve_speculation): Declare.
11738         (symtab_can_be_discarded): New function.
11739         * value-prof.c (gimple_ic_transform): Remove actual transform code.
11740         * ipa-inline-transform.c (speculation_removed): New global var.
11741         (clone_inlined_nodes): See if speculation can be removed.
11742         (inline_call): If speculations was removed, we growths may not match.
11743         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
11744         (speculation_useful_p): New function.
11745         (resolve_noninline_speculation): New function.
11746         (inline_small_functions): Resolve useless speculations.
11747         * ipa-inline.h (speculation_useful_p): Declare
11748         * ipa.c (can_replace_by_local_alias): Simplify.
11749         (ipa_profile): Produce speculative calls in non-lto, too;
11750         add simple cost model; produce local aliases.
11752 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
11754         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
11755         PASS_MANAGER_H.
11757 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
11759         * config/i386/i386.c (ix86_function_versions): Use error + inform.
11761 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
11763         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
11764         iterator instead of X87MODEF.
11766 2013-08-12  Perez Read  <netfirewall@gmail.com>
11768         PR target/58132
11769         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
11770         operand 0 for intel asm alternative.
11771         (*movabs<mode>_2): Ditto for operand 1.
11773 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
11775         * config/aarch64/arm_none.h
11776         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
11778 2013-08-12  Nick Clifton  <nickc@redhat.com>
11780         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
11782 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
11784         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
11785         expand for QI/HImode operand to produce more effictive code for
11786         unsigned char(short) --> float(double) conversion.
11788 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
11790         * doc/invoke.texi: Mention that -ftls-model does not force the final
11791         model.
11793 2013-08-12  Marek Polacek  <polacek@redhat.com>
11794             Marc Glisse  <marc.glisse@inria.fr>
11796         PR tree-optimization/57980
11797         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
11798         when creating -1 constant.
11800 2013-08-10  Jan Hubicka  <jh@suse.cz>
11802         Workaround binutils PR14342.
11803         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
11804         (gimple_init_edge_profiler): Likewise.
11805         (gimple_gen_ic_func_profiler): Likewise.
11807 2013-08-09  Jan Hubicka  <jh@suse.cz>
11809         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
11811 2013-08-09  Xinliang David Li  <davidxl@google.com>
11813         * config/i386/stringop.def: New file.
11814         * config/i386/stringop.opt: New file.
11815         * config/i386/i386-opts.h: Include stringopt.def.
11816         * config/i386/i386.opt: Include stringopt.opt.
11817         * config/i386/i386.c (ix86_option_override_internal):
11818         Override default size based stringop inline strategies with options.
11819         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
11820         New function.
11822 2013-08-09  Jan Hubicka  <jh@suse.cz>
11824         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
11826 2013-08-09  Jan Hubicka  <jh@suse.cz>
11828         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
11829         CGRAPH_FREQ_MAX.
11830         (dump_cgraph_node): Dump profile-id.
11831         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
11832         and common_target_probability.
11833         * lto-cgraph.c (lto_output_edge): Stream common targets.
11834         (lto_output_node): Stream profile ids.
11835         (input_node): Stream profile ids.
11836         (input_edge): Stream common targets.
11837         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
11838         * ipa.c: Include value-prof.h
11839         (ipa_profile_generate_summary): Turn indirect call statement histograms
11840         into common targets.
11841         (ipa_profile): Turn common targets into speculative edges.
11843 2013-08-09  Jan Hubicka  <jh@suse.cz>
11845         * cgraph.h (cgraph_node): Add profile_id.
11846         * value-prof.c (cgraph_node_map): Turn into pointer_map.
11847         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
11848         (del_node_map): Update.
11849         (find_func_by_funcdef_no): Replace by ...
11850         (find_func_by_profile_id): ... this one.
11851         (gimple_ic_transform): Do not remove useful histograms when
11852         speculation is not done; dump info when indirect call removal
11853         can happen at LTO.
11854         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
11855         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
11856         (__gcov_indirect_call_profiler_v2): .. this one.
11857         * profile.h (init_node_map): Update.
11858         * coverage.c (coverage_compute_profile_id): New function.
11859         * coverage.h (coverage_compute_profile_id): Declare.
11860         * tree-profile.c (init_ic_make_global_vars): Make
11861         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
11862         (gimple_init_edge_profiler): Update prototype of
11863         __gcov_indirect_call_profiler.
11864         (gimple_gen_ic_func_profiler): Simplify.
11865         (tree_profiling): Use init_node_map
11867 2013-08-09  Jan Hubicka  <jh@suse.cz>
11869         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
11870         non-speculative refs.
11871         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
11872         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
11873         (cgraph_set_call_stmt): Likewise.
11874         (cgraph_create_edge_1): Fix release checking compilatoin;
11875         clear lto_stmt_uid.
11876         (cgraph_free_edge): Free indirect info.
11877         (cgraph_turn_edge_to_speculative): New function.
11878         (cgraph_speculative_call_info): New function.
11879         (cgraph_make_edge_direct): Return direct edge; handle speculation.
11880         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
11881         (dump_cgraph_node): Dump speculation.
11882         (verify_edge_count_and_frequency): Accept speculative edges.
11883         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
11884         (verify_cgraph_node): Handle speculation.
11885         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
11886         (cgraph_set_call_stmt): Update prototype.
11887         (cgraph_make_edge_direct): Update prototype.
11888         (cgraph_speculative_call_info): Declare.
11889         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
11890         to change; update call of ipa_find_references.
11891         * ipa-ref.c (ipa_record_reference): Fix return value; clear
11892         lto_stmt_uid and speculative flags.
11893         (ipa_dump_references): Dump speculation.
11894         (ipa_clone_references): Clone speculative flag.
11895         (ipa_clone_referring): Likewise.
11896         (ipa_clone_ref): New function.
11897         (ipa_find_reference): Look into lto_stmt_uids
11898         (ipa_clear_stmts_in_references): Do not clear speculative calls.
11899         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
11900         (ipa_find_reference): Update declaration.
11901         (ipa_clone_ref): Declare.
11902         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
11903         stream speculative flag.
11904         (lto_output_ref): Stream statements uids and speculation.
11905         (input_ref): Likewise.
11906         (input_edge): Stream speuclation.
11907         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
11908         (cgraph_set_call_stmt_including_clones): Handle speculation.
11909         * ipa-inline.c (heap_edge_removal_hook): New function.
11910         (inline_small_functions): Register it.
11911         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
11912         also initialize refs.
11913         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
11914         edge to change.
11915         (try_make_edge_direct_simple_call): Likewise.
11916         (try_make_edge_direct_simple_call): Likewise.
11917         (update_indirect_edges_after_inlining): Likewise.
11918         (remove_described_reference): Look proper lto_stmt_uid.
11919         (propagate_controlled_uses): Likewise.
11920         (propagate_controlled_uses): Liekwise.
11921         * tree-inline.c (copy_bb): Copy speculative edges.
11922         (redirect_all_calls): New function.
11923         (copy_cfg_body): Do redirection after loop info is updated.
11924         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
11926 2013-08-09  Jan Hubicka  <jh@suse.cz>
11928         * lto-streamer-out.c (output_function): Renumber PHIs.
11929         * lto-streamer-in.c (input_function): Likewise.
11931 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
11933         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
11934         (get_lane_unsigned): Likewise.
11935         (dup_lane_scalar): Likewise.
11936         (get_lane): enable for VALL.
11937         * config/aarch64/aarch64-simd.md
11938         (aarch64_dup_lane_scalar<mode>): Remove.
11939         (aarch64_get_lane_signed<mode>): Likewise.
11940         (aarch64_get_lane_unsigned<mode>): Likewise.
11941         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
11942         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
11943         (aarch64_get_lane<mode>): Enable for all vector modes.
11944         (aarch64_get_lanedi): Remove misleading constraints.
11945         * config/aarch64/arm_neon.h
11946         (__aarch64_vget_lane_any): Define.
11947         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
11948         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
11949         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
11950         * config/aarch64/iterators.md (VDQQH): New.
11951         (VDQQHS): Likewise.
11952         (vwcore): Likewise.
11954 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
11956         * configure.ac: Add GAS check for LEON instructions on SPARC.
11957         * configure: Regenerate.
11958         * config.in: Likewise.
11959         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
11960         sparc*-*-* block.
11961         * config/sparc/sparc.opt (LEON, LEON3): New masks.
11962         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
11963         for LEON or LEON3.
11964         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
11965         (AS_LEON_FLAG): New macro.
11966         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
11967         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
11968         Deal with LEON and LEON3 for the memory model.
11969         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
11970         (atomic_compare_and_swap<mode>_1): Likewise.
11971         (*atomic_compare_and_swap<mode>_1): Likewise.
11973 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11975         * config/arm/neon.md (vcond): Fix floating-point vector
11976         comparisons against 0.
11978 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
11980         * lra-constraints.c (emit_spill_move): Remove assert.
11981         (process_alt_operands): Add more debugging
11982         output.  Increase reject for spilling into memory.  Decrease
11983         reject for reloading scratch.
11984         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
11986 2013-08-08  Steve Ellcey  <sellcey@mips.com>
11988         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
11989         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
11990         micromips incompatible.  Add nan2008.
11991         (MULTILIB_DIRNAMES): Add nan2008.
11992         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
11993         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
11994         and micromips incompatible.  Add nan2008.
11995         (MULTILIB_DIRNAMES): Add nan2008.
11996         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
11998 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
12000         PR rtl-optimization/58079
12001         * combine.c (combine_simplify_rtx): Avoid using SUBST if
12002         simplify_comparison has widened a comparison with an integer.
12004 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12006         * config/arm/neon.md (movmisalign<mode>): Disable when we
12007         don't allow unaligned accesses.
12008         (*movmisalign<mode>_neon_store): Likewise.
12009         (*movmisalign<mode>_neon_load): Likewise.
12010         (*movmisalign<mode>_neon_store): Likewise.
12011         (*movmisalign<mode>_neon_load): Likewise.
12013 2013-08-08  Jan Hubicka  <jh@suse.cz>
12015         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
12016         (make_pass_rebuild_cgraph_edges): Also clear references.
12017         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
12018         * ipa-inline-transform.c (inline_transform): Remove all references
12019         after inlining.
12020         * cgraphunit.c (expand_function): Remove all references after
12021         expansion.
12022         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
12023         (ipa_find_reference): Rewrite to iterator.
12024         (remove_stmt_references): Likewise.
12025         (ipa_clear_stmts_in_references): New function.
12026         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
12027         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
12028         clear references.
12029         * ipa-split.c (split_function): Remove references in split function.
12031 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12033         PR target/57431
12034         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
12035         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
12037 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12039         PR target/56979
12040         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
12041         suggested mode for the assignment isn't compatible with the
12042         registers required.
12044 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12046         PR target/58065
12047         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
12049 2013-08-07  Xinliang David Li  <davidxl@google.com>
12051         * config/i386/i386.opt: New option -mtune-ctrl=.
12052         * config/i386/x86-tune.def: New file.
12053         * config/i386/i386.h: include x86-tune.def.
12054         * config/i386/i386.c (ix86_option_override_internal):
12055         Parsing -mtune-ctrl= option and set tune features.
12057 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
12059         PR other/12081
12060         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
12061         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
12062         to gen_2arg_fn_t.
12064 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12066         * rtl.h (update_alignments): Declare.
12067         * final.c (grow_label_align): New function extracted from...
12068         (shorten_branches): ...here.  Call it.
12069         (update_alignments): New function.
12070         * reorg.c (sibling_labels): New variable.
12071         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
12072         the new label along with it onto the sibling_labels vector.
12073         (fill_simple_delay_slots): Adjust call to get_label_before.
12074         (fill_slots_from_thread): Likewise.
12075         (relax_delay_slots): Likewise.
12076         (make_return_insns): Likewise.
12077         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
12079 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12081         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
12082         document its semantics.
12083         (diagnostic_report_diagnostic): Adjust accordingly.
12085 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12087         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
12088         (sparc_option_override): ...and port to new C++ pass API.
12089         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
12091 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
12093         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
12095 2013-08-06  Caroline Tice  <cmtice@google.com>
12097         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
12098         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
12099         * tree-pass.h: Add pass_vtable_verify.
12100         * varasm.c (assemble_variable): Add code to properly set the comdat
12101         section and name for the .vtable_map_vars section.
12102         (assemble_vtyv_preinit_initializer): New function.
12103         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
12104         LINK_ONCE flag.
12105         * output.h: Add function decl for assemble_vtv_preinit_initializer.
12106         * vtable-verify.c: New file.
12107         * vtable-verify.h: New file.
12108         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
12109         initialiation levels.
12110         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
12111         * passes.def: Insert pass_vtable_verify.
12112         * aclocal.m4: Reorder includes.
12113         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
12114         -fvtv-counts options.
12115         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
12116         as appropriate, if -fvtable-verify=... is used.
12117         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
12118         -fvtable-verify=... is used.
12119         * Makefile.in (OBJS):  Add vtable-verify.o to list.
12120         (vtable-verify.o): Add new build rule.
12121         (GTFILES): Add vtable-verify.c to list.
12122         * common.opt (fvtable-verify=): New flag.
12123         (vtv_priority): Values for fvtable-verify= flag.
12124         (fvtv-counts): New flag.
12125         (fvtv-debug): New flag.
12126         * tree.h (save_vtable_map_decl): New extern function decl.
12128 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12130         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
12131         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
12132         (pass_data_rl78_devirt): ...new pass_data instance and...
12133         (make_pass_rl78_devirt): ...new function.
12134         (rl78_asm_file_start): Port pass registration to new C++ API.
12136 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12138         * coretypes.h (rtl_opt_pass): Add.
12139         (gcc::context): Add.
12140         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
12141         (epiphany_init): Port to new C++ pass API.
12142         (epiphany_optimize_mode_switching): Likewise.
12143         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
12144         (pass_manager::get_pass_mode_switching): New.
12145         (pass_manager::get_pass_peephole2): New.
12146         * mode-switching.c (pass_mode_switching): Add clone method.
12147         * recog.c (pass_peephole2): Add clone method.
12148         (pass_split_all_insns): Add clone method.
12150 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
12152         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
12153         (mips_option_override): ...here, porting to new C++ API for passes.
12155 2013-08-06  Jan Hubicka  <jh@suse.cz>
12157         * cgraph.c (cgraph_get_body): New function based on lto.c
12158         implementation.
12159         * cgraph.h (cgraph_get_body): Declare.
12160         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
12161         LTO paths.
12162         * cgraphunit.c (expand_function): Get body prior expanding.
12163         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
12164         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
12165         really need.
12166         * passes.c (do_per_function_toporder): Get body.
12167         * tree-inline.c (expand_call_inline): Get body prior inlining it.
12168         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
12170 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12172         PR fortran/57987
12173         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
12174         is not re-finalized.  Rename second parameter to no_collect.
12176 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12178         PR middle-end/58041
12179         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
12180         MEM_REF has proper alignment information.
12182 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
12184         PR other/12081
12185         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
12186         class insn_gen_fn.
12187         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
12188         rtx (*) (rtx, ...) with insn_gen_fn.
12189         * genoutput.c (output_insn_data): Cast gen_? function pointers to
12190         insn_gen_fn::stored_funcptr.  Add initializer braces.
12192 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12194         Rewrite how instances of passes are cloned to remove assumptions
12195         about their sizes (thus allowing pass subclasses to have
12196         additional data fields, albeit non-GC-managed ones at this point).
12198         * passes.c (make_pass_instance): Now that passes have clone
12199         methods, rewrite this function to eliminate XNEW and memcpy
12200         calls that used hardcoded sizes.  Since this function no longer
12201         creates pass instances, rename it to...
12202         (add_pass_instance): ...this.  Document the old way that passes were
12203         numbered and flagged, and rework this function to continue using it.
12204         (next_pass_1): Add an initial_pass argument for use by
12205         add_pass_instance.
12206         (position_pass): When adding multiple instances of a pass, use
12207         the pass's clone method, rather than relying on the XNEW/memcpy
12208         within the former make_pass_instance (now add_pass_instance).
12209         (pass_manager::pass_manager): When invoking next_pass_1, also supply
12210         the initial instance of the current pass within the pass manager.
12212 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12214         This is the automated part of the conversion of passes from C
12215         structs to C++ classes.
12217         Patch autogenerated by refactor_passes.py from
12218         https://github.com/davidmalcolm/gcc-refactoring-scripts
12219         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
12221         * asan.c (pass_asan): Convert from a global struct to a subclass of
12222         gimple_opt_pass along with...
12223         (pass_data_asan): ...new pass_data instance and...
12224         (make_pass_asan): ...new function.
12225         (pass_asan_O0): Convert from a global struct to a subclass of
12226         gimple_opt_pass along with...
12227         (pass_data_asan_O0): ...new pass_data instance and...
12228         (make_pass_asan_O0): ...new function.
12229         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
12230         subclass of rtl_opt_pass along with...
12231         (pass_data_inc_dec): ...new pass_data instance and...
12232         (make_pass_inc_dec): ...new function.
12233         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
12234         a subclass of rtl_opt_pass along with...
12235         (pass_data_reorder_blocks): ...new pass_data instance and...
12236         (make_pass_reorder_blocks): ...new function.
12237         (pass_duplicate_computed_gotos): Convert from a global struct to a
12238         subclass of rtl_opt_pass along with...
12239         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
12240         (make_pass_duplicate_computed_gotos): ...new function.
12241         (pass_partition_blocks): Convert from a global struct to a subclass of
12242         rtl_opt_pass along with...
12243         (pass_data_partition_blocks): ...new pass_data instance and...
12244         (make_pass_partition_blocks): ...new function.
12245         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
12246         struct to a subclass of rtl_opt_pass along with...
12247         (pass_data_branch_target_load_optimize1): ...new pass_data instance
12248         and...
12249         (make_pass_branch_target_load_optimize1): ...new function.
12250         (pass_branch_target_load_optimize2): Convert from a global struct to a
12251         subclass of rtl_opt_pass along with...
12252         (pass_data_branch_target_load_optimize2): ...new pass_data instance
12253         and...
12254         (make_pass_branch_target_load_optimize2): ...new function.
12255         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
12256         of rtl_opt_pass along with...
12257         (pass_data_jump): ...new pass_data instance and...
12258         (make_pass_jump): ...new function.
12259         (pass_jump2): Convert from a global struct to a subclass of
12260         rtl_opt_pass along with...
12261         (pass_data_jump2): ...new pass_data instance and...
12262         (make_pass_jump2): ...new function.
12263         * cfgexpand.c (pass_expand): Convert from a global struct to a
12264         subclass of rtl_opt_pass along with...
12265         (pass_data_expand): ...new pass_data instance and...
12266         (make_pass_expand): ...new function.
12267         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
12268         of rtl_opt_pass along with...
12269         (pass_data_free_cfg): ...new pass_data instance and...
12270         (make_pass_free_cfg): ...new function.
12271         (pass_into_cfg_layout_mode): Convert from a global struct to a
12272         subclass of rtl_opt_pass along with...
12273         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
12274         (make_pass_into_cfg_layout_mode): ...new function.
12275         (pass_outof_cfg_layout_mode): Convert from a global struct to a
12276         subclass of rtl_opt_pass along with...
12277         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
12278         (make_pass_outof_cfg_layout_mode): ...new function.
12279         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
12280         struct to a subclass of gimple_opt_pass along with...
12281         (pass_data_build_cgraph_edges): ...new pass_data instance and...
12282         (make_pass_build_cgraph_edges): ...new function.
12283         (pass_rebuild_cgraph_edges): Convert from a global struct to a
12284         subclass of gimple_opt_pass along with...
12285         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
12286         (make_pass_rebuild_cgraph_edges): ...new function.
12287         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
12288         subclass of gimple_opt_pass along with...
12289         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
12290         and...
12291         (make_pass_remove_cgraph_callee_edges): ...new function.
12292         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
12293         struct to a subclass of rtl_opt_pass along with...
12294         (pass_data_stack_adjustments): ...new pass_data instance and...
12295         (make_pass_stack_adjustments): ...new function.
12296         * combine.c (pass_combine): Convert from a global struct to a subclass
12297         of rtl_opt_pass along with...
12298         (pass_data_combine): ...new pass_data instance and...
12299         (make_pass_combine): ...new function.
12300         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
12301         global struct to a subclass of rtl_opt_pass along with...
12302         (pass_data_compare_elim_after_reload): ...new pass_data instance
12303         and...
12304         (make_pass_compare_elim_after_reload): ...new function.
12305         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
12306         of rtl_opt_pass along with...
12307         (pass_data_rtl_cprop): ...new pass_data instance and...
12308         (make_pass_rtl_cprop): ...new function.
12309         * cse.c (pass_cse): Convert from a global struct to a subclass of
12310         rtl_opt_pass along with...
12311         (pass_data_cse): ...new pass_data instance and...
12312         (make_pass_cse): ...new function.
12313         (pass_cse2): Convert from a global struct to a subclass of
12314         rtl_opt_pass along with...
12315         (pass_data_cse2): ...new pass_data instance and...
12316         (make_pass_cse2): ...new function.
12317         (pass_cse_after_global_opts): Convert from a global struct to a
12318         subclass of rtl_opt_pass along with...
12319         (pass_data_cse_after_global_opts): ...new pass_data instance and...
12320         (make_pass_cse_after_global_opts): ...new function.
12321         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
12322         of rtl_opt_pass along with...
12323         (pass_data_ud_rtl_dce): ...new pass_data instance and...
12324         (make_pass_ud_rtl_dce): ...new function.
12325         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
12326         rtl_opt_pass along with...
12327         (pass_data_fast_rtl_dce): ...new pass_data instance and...
12328         (make_pass_fast_rtl_dce): ...new function.
12329         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
12330         a subclass of rtl_opt_pass along with...
12331         (pass_data_df_initialize_opt): ...new pass_data instance and...
12332         (make_pass_df_initialize_opt): ...new function.
12333         (pass_df_initialize_no_opt): Convert from a global struct to a
12334         subclass of rtl_opt_pass along with...
12335         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
12336         (make_pass_df_initialize_no_opt): ...new function.
12337         (pass_df_finish): Convert from a global struct to a subclass of
12338         rtl_opt_pass along with...
12339         (pass_data_df_finish): ...new pass_data instance and...
12340         (make_pass_df_finish): ...new function.
12341         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
12342         rtl_opt_pass along with...
12343         (pass_data_rtl_dse1): ...new pass_data instance and...
12344         (make_pass_rtl_dse1): ...new function.
12345         (pass_rtl_dse2): Convert from a global struct to a subclass of
12346         rtl_opt_pass along with...
12347         (pass_data_rtl_dse2): ...new pass_data instance and...
12348         (make_pass_rtl_dse2): ...new function.
12349         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
12350         subclass of rtl_opt_pass along with...
12351         (pass_data_dwarf2_frame): ...new pass_data instance and...
12352         (make_pass_dwarf2_frame): ...new function.
12353         * except.c (pass_set_nothrow_function_flags): Convert from a global
12354         struct to a subclass of rtl_opt_pass along with...
12355         (pass_data_set_nothrow_function_flags): ...new pass_data instance
12356         and...
12357         (make_pass_set_nothrow_function_flags): ...new function.
12358         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
12359         subclass of rtl_opt_pass along with...
12360         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
12361         and...
12362         (make_pass_convert_to_eh_region_ranges): ...new function.
12363         * final.c (pass_compute_alignments): Convert from a global struct to a
12364         subclass of rtl_opt_pass along with...
12365         (pass_data_compute_alignments): ...new pass_data instance and...
12366         (make_pass_compute_alignments): ...new function.
12367         (pass_final): Convert from a global struct to a subclass of
12368         rtl_opt_pass along with...
12369         (pass_data_final): ...new pass_data instance and...
12370         (make_pass_final): ...new function.
12371         (pass_shorten_branches): Convert from a global struct to a subclass of
12372         rtl_opt_pass along with...
12373         (pass_data_shorten_branches): ...new pass_data instance and...
12374         (make_pass_shorten_branches): ...new function.
12375         (pass_clean_state): Convert from a global struct to a subclass of
12376         rtl_opt_pass along with...
12377         (pass_data_clean_state): ...new pass_data instance and...
12378         (make_pass_clean_state): ...new function.
12379         * function.c (pass_instantiate_virtual_regs): Convert from a global
12380         struct to a subclass of rtl_opt_pass along with...
12381         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
12382         (make_pass_instantiate_virtual_regs): ...new function.
12383         (pass_leaf_regs): Convert from a global struct to a subclass of
12384         rtl_opt_pass along with...
12385         (pass_data_leaf_regs): ...new pass_data instance and...
12386         (make_pass_leaf_regs): ...new function.
12387         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
12388         subclass of rtl_opt_pass along with...
12389         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
12390         and...
12391         (make_pass_thread_prologue_and_epilogue): ...new function.
12392         (pass_match_asm_constraints): Convert from a global struct to a
12393         subclass of rtl_opt_pass along with...
12394         (pass_data_match_asm_constraints): ...new pass_data instance and...
12395         (make_pass_match_asm_constraints): ...new function.
12396         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
12397         subclass of rtl_opt_pass along with...
12398         (pass_data_rtl_fwprop): ...new pass_data instance and...
12399         (make_pass_rtl_fwprop): ...new function.
12400         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
12401         rtl_opt_pass along with...
12402         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
12403         (make_pass_rtl_fwprop_addr): ...new function.
12404         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
12405         rtl_opt_pass along with...
12406         (pass_data_rtl_pre): ...new pass_data instance and...
12407         (make_pass_rtl_pre): ...new function.
12408         (pass_rtl_hoist): Convert from a global struct to a subclass of
12409         rtl_opt_pass along with...
12410         (pass_data_rtl_hoist): ...new pass_data instance and...
12411         (make_pass_rtl_hoist): ...new function.
12412         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
12413         subclass of gimple_opt_pass along with...
12414         (pass_data_lower_cf): ...new pass_data instance and...
12415         (make_pass_lower_cf): ...new function.
12416         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
12417         from a global struct to a subclass of gimple_opt_pass along with...
12418         (pass_data_strength_reduction): ...new pass_data instance and...
12419         (make_pass_strength_reduction): ...new function.
12420         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
12421         of rtl_opt_pass along with...
12422         (pass_data_rtl_ifcvt): ...new pass_data instance and...
12423         (make_pass_rtl_ifcvt): ...new function.
12424         (pass_if_after_combine): Convert from a global struct to a subclass of
12425         rtl_opt_pass along with...
12426         (pass_data_if_after_combine): ...new pass_data instance and...
12427         (make_pass_if_after_combine): ...new function.
12428         (pass_if_after_reload): Convert from a global struct to a subclass of
12429         rtl_opt_pass along with...
12430         (pass_data_if_after_reload): ...new pass_data instance and...
12431         (make_pass_if_after_reload): ...new function.
12432         * init-regs.c (pass_initialize_regs): Convert from a global struct to
12433         a subclass of rtl_opt_pass along with...
12434         (pass_data_initialize_regs): ...new pass_data instance and...
12435         (make_pass_initialize_regs): ...new function.
12436         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
12437         of ipa_opt_pass_d along with...
12438         (pass_data_ipa_cp): ...new pass_data instance and...
12439         (make_pass_ipa_cp): ...new function.
12440         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
12441         global struct to a subclass of gimple_opt_pass along with...
12442         (pass_data_inline_parameters): ...new pass_data instance and...
12443         (make_pass_inline_parameters): ...new function.
12444         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
12445         subclass of gimple_opt_pass along with...
12446         (pass_data_early_inline): ...new pass_data instance and...
12447         (make_pass_early_inline): ...new function.
12448         (pass_ipa_inline): Convert from a global struct to a subclass of
12449         ipa_opt_pass_d along with...
12450         (pass_data_ipa_inline): ...new pass_data instance and...
12451         (make_pass_ipa_inline): ...new function.
12452         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
12453         struct to a subclass of gimple_opt_pass along with...
12454         (pass_data_local_pure_const): ...new pass_data instance and...
12455         (make_pass_local_pure_const): ...new function.
12456         (pass_ipa_pure_const): Convert from a global struct to a subclass of
12457         ipa_opt_pass_d along with...
12458         (pass_data_ipa_pure_const): ...new pass_data instance and...
12459         (make_pass_ipa_pure_const): ...new function.
12460         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
12461         to a subclass of ipa_opt_pass_d along with...
12462         (pass_data_ipa_reference): ...new pass_data instance and...
12463         (make_pass_ipa_reference): ...new function.
12464         * ipa-split.c (pass_split_functions): Convert from a global struct to
12465         a subclass of gimple_opt_pass along with...
12466         (pass_data_split_functions): ...new pass_data instance and...
12467         (make_pass_split_functions): ...new function.
12468         (pass_feedback_split_functions): Convert from a global struct to a
12469         subclass of gimple_opt_pass along with...
12470         (pass_data_feedback_split_functions): ...new pass_data instance and...
12471         (make_pass_feedback_split_functions): ...new function.
12472         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
12473         global struct to a subclass of simple_ipa_opt_pass along with...
12474         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
12475         instance and...
12476         (make_pass_ipa_function_and_variable_visibility): ...new function.
12477         (pass_ipa_free_inline_summary): Convert from a global struct to a
12478         subclass of simple_ipa_opt_pass along with...
12479         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
12480         (make_pass_ipa_free_inline_summary): ...new function.
12481         (pass_ipa_whole_program_visibility): Convert from a global struct to a
12482         subclass of ipa_opt_pass_d along with...
12483         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
12484         and...
12485         (make_pass_ipa_whole_program_visibility): ...new function.
12486         (pass_ipa_profile): Convert from a global struct to a subclass of
12487         ipa_opt_pass_d along with...
12488         (pass_data_ipa_profile): ...new pass_data instance and...
12489         (make_pass_ipa_profile): ...new function.
12490         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
12491         ipa_opt_pass_d along with...
12492         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
12493         (make_pass_ipa_cdtor_merge): ...new function.
12494         * ira.c (pass_ira): Convert from a global struct to a subclass of
12495         rtl_opt_pass along with...
12496         (pass_data_ira): ...new pass_data instance and...
12497         (make_pass_ira): ...new function.
12498         (pass_reload): Convert from a global struct to a subclass of
12499         rtl_opt_pass along with...
12500         (pass_data_reload): ...new pass_data instance and...
12501         (make_pass_reload): ...new function.
12502         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
12503         subclass of rtl_opt_pass along with...
12504         (pass_data_cleanup_barriers): ...new pass_data instance and...
12505         (make_pass_cleanup_barriers): ...new function.
12506         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
12507         of rtl_opt_pass along with...
12508         (pass_data_loop2): ...new pass_data instance and...
12509         (make_pass_loop2): ...new function.
12510         (pass_rtl_loop_init): Convert from a global struct to a subclass of
12511         rtl_opt_pass along with...
12512         (pass_data_rtl_loop_init): ...new pass_data instance and...
12513         (make_pass_rtl_loop_init): ...new function.
12514         (pass_rtl_loop_done): Convert from a global struct to a subclass of
12515         rtl_opt_pass along with...
12516         (pass_data_rtl_loop_done): ...new pass_data instance and...
12517         (make_pass_rtl_loop_done): ...new function.
12518         (pass_rtl_move_loop_invariants): Convert from a global struct to a
12519         subclass of rtl_opt_pass along with...
12520         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
12521         (make_pass_rtl_move_loop_invariants): ...new function.
12522         (pass_rtl_unswitch): Convert from a global struct to a subclass of
12523         rtl_opt_pass along with...
12524         (pass_data_rtl_unswitch): ...new pass_data instance and...
12525         (make_pass_rtl_unswitch): ...new function.
12526         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
12527         subclass of rtl_opt_pass along with...
12528         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
12529         and...
12530         (make_pass_rtl_unroll_and_peel_loops): ...new function.
12531         (pass_rtl_doloop): Convert from a global struct to a subclass of
12532         rtl_opt_pass along with...
12533         (pass_data_rtl_doloop): ...new pass_data instance and...
12534         (make_pass_rtl_doloop): ...new function.
12535         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
12536         a subclass of rtl_opt_pass along with...
12537         (pass_data_lower_subreg): ...new pass_data instance and...
12538         (make_pass_lower_subreg): ...new function.
12539         (pass_lower_subreg2): Convert from a global struct to a subclass of
12540         rtl_opt_pass along with...
12541         (pass_data_lower_subreg2): ...new pass_data instance and...
12542         (make_pass_lower_subreg2): ...new function.
12543         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
12544         struct to a subclass of ipa_opt_pass_d along with...
12545         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
12546         (make_pass_ipa_lto_gimple_out): ...new function.
12547         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
12548         of ipa_opt_pass_d along with...
12549         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
12550         (make_pass_ipa_lto_finish_out): ...new function.
12551         * mode-switching.c (pass_mode_switching): Convert from a global struct
12552         to a subclass of rtl_opt_pass along with...
12553         (pass_data_mode_switching): ...new pass_data instance and...
12554         (make_pass_mode_switching): ...new function.
12555         * modulo-sched.c (pass_sms): Convert from a global struct to a
12556         subclass of rtl_opt_pass along with...
12557         (pass_data_sms): ...new pass_data instance and...
12558         (make_pass_sms): ...new function.
12559         * omp-low.c (pass_expand_omp): Convert from a global struct to a
12560         subclass of gimple_opt_pass along with...
12561         (pass_data_expand_omp): ...new pass_data instance and...
12562         (make_pass_expand_omp): ...new function.
12563         (pass_lower_omp): Convert from a global struct to a subclass of
12564         gimple_opt_pass along with...
12565         (pass_data_lower_omp): ...new pass_data instance and...
12566         (make_pass_lower_omp): ...new function.
12567         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
12568         of gimple_opt_pass along with...
12569         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
12570         (make_pass_diagnose_omp_blocks): ...new function.
12571         * passes.c (pass_early_local_passes): Convert from a global struct to
12572         a subclass of simple_ipa_opt_pass along with...
12573         (pass_data_early_local_passes): ...new pass_data instance and...
12574         (make_pass_early_local_passes): ...new function.
12575         (pass_all_early_optimizations): Convert from a global struct to a
12576         subclass of gimple_opt_pass along with...
12577         (pass_data_all_early_optimizations): ...new pass_data instance and...
12578         (make_pass_all_early_optimizations): ...new function.
12579         (pass_all_optimizations): Convert from a global struct to a subclass
12580         of gimple_opt_pass along with...
12581         (pass_data_all_optimizations): ...new pass_data instance and...
12582         (make_pass_all_optimizations): ...new function.
12583         (pass_all_optimizations_g): Convert from a global struct to a subclass
12584         of gimple_opt_pass along with...
12585         (pass_data_all_optimizations_g): ...new pass_data instance and...
12586         (make_pass_all_optimizations_g): ...new function.
12587         (pass_rest_of_compilation): Convert from a global struct to a subclass
12588         of rtl_opt_pass along with...
12589         (pass_data_rest_of_compilation): ...new pass_data instance and...
12590         (make_pass_rest_of_compilation): ...new function.
12591         (pass_postreload): Convert from a global struct to a subclass of
12592         rtl_opt_pass along with...
12593         (pass_data_postreload): ...new pass_data instance and...
12594         (make_pass_postreload): ...new function.
12595         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
12596         subclass of rtl_opt_pass along with...
12597         (pass_data_gcse2): ...new pass_data instance and...
12598         (make_pass_gcse2): ...new function.
12599         * postreload.c (pass_postreload_cse): Convert from a global struct to
12600         a subclass of rtl_opt_pass along with...
12601         (pass_data_postreload_cse): ...new pass_data instance and...
12602         (make_pass_postreload_cse): ...new function.
12603         * predict.c (pass_profile): Convert from a global struct to a subclass
12604         of gimple_opt_pass along with...
12605         (pass_data_profile): ...new pass_data instance and...
12606         (make_pass_profile): ...new function.
12607         (pass_strip_predict_hints): Convert from a global struct to a subclass
12608         of gimple_opt_pass along with...
12609         (pass_data_strip_predict_hints): ...new pass_data instance and...
12610         (make_pass_strip_predict_hints): ...new function.
12611         * recog.c (pass_peephole2): Convert from a global struct to a subclass
12612         of rtl_opt_pass along with...
12613         (pass_data_peephole2): ...new pass_data instance and...
12614         (make_pass_peephole2): ...new function.
12615         (pass_split_all_insns): Convert from a global struct to a subclass of
12616         rtl_opt_pass along with...
12617         (pass_data_split_all_insns): ...new pass_data instance and...
12618         (make_pass_split_all_insns): ...new function.
12619         (pass_split_after_reload): Convert from a global struct to a subclass
12620         of rtl_opt_pass along with...
12621         (pass_data_split_after_reload): ...new pass_data instance and...
12622         (make_pass_split_after_reload): ...new function.
12623         (pass_split_before_regstack): Convert from a global struct to a
12624         subclass of rtl_opt_pass along with...
12625         (pass_data_split_before_regstack): ...new pass_data instance and...
12626         (make_pass_split_before_regstack): ...new function.
12627         (pass_split_before_sched2): Convert from a global struct to a subclass
12628         of rtl_opt_pass along with...
12629         (pass_data_split_before_sched2): ...new pass_data instance and...
12630         (make_pass_split_before_sched2): ...new function.
12631         (pass_split_for_shorten_branches): Convert from a global struct to a
12632         subclass of rtl_opt_pass along with...
12633         (pass_data_split_for_shorten_branches): ...new pass_data instance
12634         and...
12635         (make_pass_split_for_shorten_branches): ...new function.
12636         * ree.c (pass_ree): Convert from a global struct to a subclass of
12637         rtl_opt_pass along with...
12638         (pass_data_ree): ...new pass_data instance and...
12639         (make_pass_ree): ...new function.
12640         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
12641         subclass of rtl_opt_pass along with...
12642         (pass_data_stack_regs): ...new pass_data instance and...
12643         (make_pass_stack_regs): ...new function.
12644         (pass_stack_regs_run): Convert from a global struct to a subclass of
12645         rtl_opt_pass along with...
12646         (pass_data_stack_regs_run): ...new pass_data instance and...
12647         (make_pass_stack_regs_run): ...new function.
12648         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
12649         subclass of rtl_opt_pass along with...
12650         (pass_data_cprop_hardreg): ...new pass_data instance and...
12651         (make_pass_cprop_hardreg): ...new function.
12652         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
12653         subclass of rtl_opt_pass along with...
12654         (pass_data_reginfo_init): ...new pass_data instance and...
12655         (make_pass_reginfo_init): ...new function.
12656         * regmove.c (pass_regmove): Convert from a global struct to a subclass
12657         of rtl_opt_pass along with...
12658         (pass_data_regmove): ...new pass_data instance and...
12659         (make_pass_regmove): ...new function.
12660         * regrename.c (pass_regrename): Convert from a global struct to a
12661         subclass of rtl_opt_pass along with...
12662         (pass_data_regrename): ...new pass_data instance and...
12663         (make_pass_regrename): ...new function.
12664         * reorg.c (pass_delay_slots): Convert from a global struct to a
12665         subclass of rtl_opt_pass along with...
12666         (pass_data_delay_slots): ...new pass_data instance and...
12667         (make_pass_delay_slots): ...new function.
12668         (pass_machine_reorg): Convert from a global struct to a subclass of
12669         rtl_opt_pass along with...
12670         (pass_data_machine_reorg): ...new pass_data instance and...
12671         (make_pass_machine_reorg): ...new function.
12672         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
12673         of rtl_opt_pass along with...
12674         (pass_data_sched): ...new pass_data instance and...
12675         (make_pass_sched): ...new function.
12676         (pass_sched2): Convert from a global struct to a subclass of
12677         rtl_opt_pass along with...
12678         (pass_data_sched2): ...new pass_data instance and...
12679         (make_pass_sched2): ...new function.
12680         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
12681         to a subclass of rtl_opt_pass along with...
12682         (pass_data_stack_ptr_mod): ...new pass_data instance and...
12683         (make_pass_stack_ptr_mod): ...new function.
12684         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
12685         to a subclass of rtl_opt_pass along with...
12686         (pass_data_rtl_store_motion): ...new pass_data instance and...
12687         (make_pass_rtl_store_motion): ...new function.
12688         * tracer.c (pass_tracer): Convert from a global struct to a subclass
12689         of gimple_opt_pass along with...
12690         (pass_data_tracer): ...new pass_data instance and...
12691         (make_pass_tracer): ...new function.
12692         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
12693         to a subclass of gimple_opt_pass along with...
12694         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
12695         (make_pass_diagnose_tm_blocks): ...new function.
12696         (pass_lower_tm): Convert from a global struct to a subclass of
12697         gimple_opt_pass along with...
12698         (pass_data_lower_tm): ...new pass_data instance and...
12699         (make_pass_lower_tm): ...new function.
12700         (pass_tm_init): Convert from a global struct to a subclass of
12701         gimple_opt_pass along with...
12702         (pass_data_tm_init): ...new pass_data instance and...
12703         (make_pass_tm_init): ...new function.
12704         (pass_tm_mark): Convert from a global struct to a subclass of
12705         gimple_opt_pass along with...
12706         (pass_data_tm_mark): ...new pass_data instance and...
12707         (make_pass_tm_mark): ...new function.
12708         (pass_tm_edges): Convert from a global struct to a subclass of
12709         gimple_opt_pass along with...
12710         (pass_data_tm_edges): ...new pass_data instance and...
12711         (make_pass_tm_edges): ...new function.
12712         (pass_tm_memopt): Convert from a global struct to a subclass of
12713         gimple_opt_pass along with...
12714         (pass_data_tm_memopt): ...new pass_data instance and...
12715         (make_pass_tm_memopt): ...new function.
12716         (pass_ipa_tm): Convert from a global struct to a subclass of
12717         simple_ipa_opt_pass along with...
12718         (pass_data_ipa_tm): ...new pass_data instance and...
12719         (make_pass_ipa_tm): ...new function.
12720         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
12721         subclass of gimple_opt_pass along with...
12722         (pass_data_call_cdce): ...new pass_data instance and...
12723         (make_pass_call_cdce): ...new function.
12724         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
12725         subclass of gimple_opt_pass along with...
12726         (pass_data_build_cfg): ...new pass_data instance and...
12727         (make_pass_build_cfg): ...new function.
12728         (pass_split_crit_edges): Convert from a global struct to a subclass of
12729         gimple_opt_pass along with...
12730         (pass_data_split_crit_edges): ...new pass_data instance and...
12731         (make_pass_split_crit_edges): ...new function.
12732         (pass_warn_function_return): Convert from a global struct to a
12733         subclass of gimple_opt_pass along with...
12734         (pass_data_warn_function_return): ...new pass_data instance and...
12735         (make_pass_warn_function_return): ...new function.
12736         (pass_warn_function_noreturn): Convert from a global struct to a
12737         subclass of gimple_opt_pass along with...
12738         (pass_data_warn_function_noreturn): ...new pass_data instance and...
12739         (make_pass_warn_function_noreturn): ...new function.
12740         (pass_warn_unused_result): Convert from a global struct to a subclass
12741         of gimple_opt_pass along with...
12742         (pass_data_warn_unused_result): ...new pass_data instance and...
12743         (make_pass_warn_unused_result): ...new function.
12744         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
12745         a subclass of gimple_opt_pass along with...
12746         (pass_data_merge_phi): ...new pass_data instance and...
12747         (make_pass_merge_phi): ...new function.
12748         * tree-complex.c (pass_lower_complex): Convert from a global struct to
12749         a subclass of gimple_opt_pass along with...
12750         (pass_data_lower_complex): ...new pass_data instance and...
12751         (make_pass_lower_complex): ...new function.
12752         (pass_lower_complex_O0): Convert from a global struct to a subclass of
12753         gimple_opt_pass along with...
12754         (pass_data_lower_complex_O0): ...new pass_data instance and...
12755         (make_pass_lower_complex_O0): ...new function.
12756         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
12757         subclass of gimple_opt_pass along with...
12758         (pass_data_lower_eh): ...new pass_data instance and...
12759         (make_pass_lower_eh): ...new function.
12760         (pass_refactor_eh): Convert from a global struct to a subclass of
12761         gimple_opt_pass along with...
12762         (pass_data_refactor_eh): ...new pass_data instance and...
12763         (make_pass_refactor_eh): ...new function.
12764         (pass_lower_resx): Convert from a global struct to a subclass of
12765         gimple_opt_pass along with...
12766         (pass_data_lower_resx): ...new pass_data instance and...
12767         (make_pass_lower_resx): ...new function.
12768         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
12769         of gimple_opt_pass along with...
12770         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
12771         (make_pass_lower_eh_dispatch): ...new function.
12772         (pass_cleanup_eh): Convert from a global struct to a subclass of
12773         gimple_opt_pass along with...
12774         (pass_data_cleanup_eh): ...new pass_data instance and...
12775         (make_pass_cleanup_eh): ...new function.
12776         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
12777         to a subclass of simple_ipa_opt_pass along with...
12778         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
12779         (make_pass_ipa_lower_emutls): ...new function.
12780         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
12781         a subclass of gimple_opt_pass along with...
12782         (pass_data_if_conversion): ...new pass_data instance and...
12783         (make_pass_if_conversion): ...new function.
12784         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
12785         subclass of gimple_opt_pass along with...
12786         (pass_data_build_ssa): ...new pass_data instance and...
12787         (make_pass_build_ssa): ...new function.
12788         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
12789         global struct to a subclass of gimple_opt_pass along with...
12790         (pass_data_loop_distribution): ...new pass_data instance and...
12791         (make_pass_loop_distribution): ...new function.
12792         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
12793         subclass of gimple_opt_pass along with...
12794         (pass_data_mudflap_1): ...new pass_data instance and...
12795         (make_pass_mudflap_1): ...new function.
12796         (pass_mudflap_2): Convert from a global struct to a subclass of
12797         gimple_opt_pass along with...
12798         (pass_data_mudflap_2): ...new pass_data instance and...
12799         (make_pass_mudflap_2): ...new function.
12800         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
12801         subclass of gimple_opt_pass along with...
12802         (pass_data_mudflap_1): ...new pass_data instance and...
12803         (make_pass_mudflap_1): ...new function.
12804         (pass_mudflap_2): Convert from a global struct to a subclass of
12805         gimple_opt_pass along with...
12806         (pass_data_mudflap_2): ...new pass_data instance and...
12807         (make_pass_mudflap_2): ...new function.
12808         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
12809         gimple_opt_pass along with...
12810         (pass_data_nrv): ...new pass_data instance and...
12811         (make_pass_nrv): ...new function.
12812         (pass_return_slot): Convert from a global struct to a subclass of
12813         gimple_opt_pass along with...
12814         (pass_data_return_slot): ...new pass_data instance and...
12815         (make_pass_return_slot): ...new function.
12816         * tree-object-size.c (pass_object_sizes): Convert from a global struct
12817         to a subclass of gimple_opt_pass along with...
12818         (pass_data_object_sizes): ...new pass_data instance and...
12819         (make_pass_object_sizes): ...new function.
12820         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
12821         global struct to a subclass of gimple_opt_pass along with...
12822         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
12823         and...
12824         (make_pass_cleanup_cfg_post_optimizing): ...new function.
12825         (pass_fixup_cfg): Convert from a global struct to a subclass of
12826         gimple_opt_pass along with...
12827         (pass_data_fixup_cfg): ...new pass_data instance and...
12828         (make_pass_fixup_cfg): ...new function.
12829         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
12830         (make_pass_mudflap_1): ...new function.
12831         (pass_mudflap_2): Replace declaration with that of...
12832         (make_pass_mudflap_2): ...new function.
12833         (pass_asan): Replace declaration with that of...
12834         (make_pass_asan): ...new function.
12835         (pass_asan_O0): Replace declaration with that of...
12836         (make_pass_asan_O0): ...new function.
12837         (pass_tsan): Replace declaration with that of...
12838         (make_pass_tsan): ...new function.
12839         (pass_tsan_O0): Replace declaration with that of...
12840         (make_pass_tsan_O0): ...new function.
12841         (pass_lower_cf): Replace declaration with that of...
12842         (make_pass_lower_cf): ...new function.
12843         (pass_refactor_eh): Replace declaration with that of...
12844         (make_pass_refactor_eh): ...new function.
12845         (pass_lower_eh): Replace declaration with that of...
12846         (make_pass_lower_eh): ...new function.
12847         (pass_lower_eh_dispatch): Replace declaration with that of...
12848         (make_pass_lower_eh_dispatch): ...new function.
12849         (pass_lower_resx): Replace declaration with that of...
12850         (make_pass_lower_resx): ...new function.
12851         (pass_build_cfg): Replace declaration with that of...
12852         (make_pass_build_cfg): ...new function.
12853         (pass_early_tree_profile): Replace declaration with that of...
12854         (make_pass_early_tree_profile): ...new function.
12855         (pass_cleanup_eh): Replace declaration with that of...
12856         (make_pass_cleanup_eh): ...new function.
12857         (pass_sra): Replace declaration with that of...
12858         (make_pass_sra): ...new function.
12859         (pass_sra_early): Replace declaration with that of...
12860         (make_pass_sra_early): ...new function.
12861         (pass_early_ipa_sra): Replace declaration with that of...
12862         (make_pass_early_ipa_sra): ...new function.
12863         (pass_tail_recursion): Replace declaration with that of...
12864         (make_pass_tail_recursion): ...new function.
12865         (pass_tail_calls): Replace declaration with that of...
12866         (make_pass_tail_calls): ...new function.
12867         (pass_tree_loop): Replace declaration with that of...
12868         (make_pass_tree_loop): ...new function.
12869         (pass_tree_loop_init): Replace declaration with that of...
12870         (make_pass_tree_loop_init): ...new function.
12871         (pass_lim): Replace declaration with that of...
12872         (make_pass_lim): ...new function.
12873         (pass_tree_unswitch): Replace declaration with that of...
12874         (make_pass_tree_unswitch): ...new function.
12875         (pass_predcom): Replace declaration with that of...
12876         (make_pass_predcom): ...new function.
12877         (pass_iv_canon): Replace declaration with that of...
12878         (make_pass_iv_canon): ...new function.
12879         (pass_scev_cprop): Replace declaration with that of...
12880         (make_pass_scev_cprop): ...new function.
12881         (pass_empty_loop): Replace declaration with that of...
12882         (make_pass_empty_loop): ...new function.
12883         (pass_record_bounds): Replace declaration with that of...
12884         (make_pass_record_bounds): ...new function.
12885         (pass_graphite): Replace declaration with that of...
12886         (make_pass_graphite): ...new function.
12887         (pass_graphite_transforms): Replace declaration with that of...
12888         (make_pass_graphite_transforms): ...new function.
12889         (pass_if_conversion): Replace declaration with that of...
12890         (make_pass_if_conversion): ...new function.
12891         (pass_loop_distribution): Replace declaration with that of...
12892         (make_pass_loop_distribution): ...new function.
12893         (pass_vectorize): Replace declaration with that of...
12894         (make_pass_vectorize): ...new function.
12895         (pass_slp_vectorize): Replace declaration with that of...
12896         (make_pass_slp_vectorize): ...new function.
12897         (pass_complete_unroll): Replace declaration with that of...
12898         (make_pass_complete_unroll): ...new function.
12899         (pass_complete_unrolli): Replace declaration with that of...
12900         (make_pass_complete_unrolli): ...new function.
12901         (pass_parallelize_loops): Replace declaration with that of...
12902         (make_pass_parallelize_loops): ...new function.
12903         (pass_loop_prefetch): Replace declaration with that of...
12904         (make_pass_loop_prefetch): ...new function.
12905         (pass_iv_optimize): Replace declaration with that of...
12906         (make_pass_iv_optimize): ...new function.
12907         (pass_tree_loop_done): Replace declaration with that of...
12908         (make_pass_tree_loop_done): ...new function.
12909         (pass_ch): Replace declaration with that of...
12910         (make_pass_ch): ...new function.
12911         (pass_ccp): Replace declaration with that of...
12912         (make_pass_ccp): ...new function.
12913         (pass_phi_only_cprop): Replace declaration with that of...
12914         (make_pass_phi_only_cprop): ...new function.
12915         (pass_build_ssa): Replace declaration with that of...
12916         (make_pass_build_ssa): ...new function.
12917         (pass_build_alias): Replace declaration with that of...
12918         (make_pass_build_alias): ...new function.
12919         (pass_build_ealias): Replace declaration with that of...
12920         (make_pass_build_ealias): ...new function.
12921         (pass_dominator): Replace declaration with that of...
12922         (make_pass_dominator): ...new function.
12923         (pass_dce): Replace declaration with that of...
12924         (make_pass_dce): ...new function.
12925         (pass_dce_loop): Replace declaration with that of...
12926         (make_pass_dce_loop): ...new function.
12927         (pass_cd_dce): Replace declaration with that of...
12928         (make_pass_cd_dce): ...new function.
12929         (pass_call_cdce): Replace declaration with that of...
12930         (make_pass_call_cdce): ...new function.
12931         (pass_merge_phi): Replace declaration with that of...
12932         (make_pass_merge_phi): ...new function.
12933         (pass_split_crit_edges): Replace declaration with that of...
12934         (make_pass_split_crit_edges): ...new function.
12935         (pass_pre): Replace declaration with that of...
12936         (make_pass_pre): ...new function.
12937         (pass_profile): Replace declaration with that of...
12938         (make_pass_profile): ...new function.
12939         (pass_strip_predict_hints): Replace declaration with that of...
12940         (make_pass_strip_predict_hints): ...new function.
12941         (pass_lower_complex_O0): Replace declaration with that of...
12942         (make_pass_lower_complex_O0): ...new function.
12943         (pass_lower_complex): Replace declaration with that of...
12944         (make_pass_lower_complex): ...new function.
12945         (pass_lower_vector): Replace declaration with that of...
12946         (make_pass_lower_vector): ...new function.
12947         (pass_lower_vector_ssa): Replace declaration with that of...
12948         (make_pass_lower_vector_ssa): ...new function.
12949         (pass_lower_omp): Replace declaration with that of...
12950         (make_pass_lower_omp): ...new function.
12951         (pass_diagnose_omp_blocks): Replace declaration with that of...
12952         (make_pass_diagnose_omp_blocks): ...new function.
12953         (pass_expand_omp): Replace declaration with that of...
12954         (make_pass_expand_omp): ...new function.
12955         (pass_expand_omp_ssa): Replace declaration with that of...
12956         (make_pass_expand_omp_ssa): ...new function.
12957         (pass_object_sizes): Replace declaration with that of...
12958         (make_pass_object_sizes): ...new function.
12959         (pass_strlen): Replace declaration with that of...
12960         (make_pass_strlen): ...new function.
12961         (pass_fold_builtins): Replace declaration with that of...
12962         (make_pass_fold_builtins): ...new function.
12963         (pass_stdarg): Replace declaration with that of...
12964         (make_pass_stdarg): ...new function.
12965         (pass_early_warn_uninitialized): Replace declaration with that of...
12966         (make_pass_early_warn_uninitialized): ...new function.
12967         (pass_late_warn_uninitialized): Replace declaration with that of...
12968         (make_pass_late_warn_uninitialized): ...new function.
12969         (pass_cse_reciprocals): Replace declaration with that of...
12970         (make_pass_cse_reciprocals): ...new function.
12971         (pass_cse_sincos): Replace declaration with that of...
12972         (make_pass_cse_sincos): ...new function.
12973         (pass_optimize_bswap): Replace declaration with that of...
12974         (make_pass_optimize_bswap): ...new function.
12975         (pass_optimize_widening_mul): Replace declaration with that of...
12976         (make_pass_optimize_widening_mul): ...new function.
12977         (pass_warn_function_return): Replace declaration with that of...
12978         (make_pass_warn_function_return): ...new function.
12979         (pass_warn_function_noreturn): Replace declaration with that of...
12980         (make_pass_warn_function_noreturn): ...new function.
12981         (pass_cselim): Replace declaration with that of...
12982         (make_pass_cselim): ...new function.
12983         (pass_phiopt): Replace declaration with that of...
12984         (make_pass_phiopt): ...new function.
12985         (pass_forwprop): Replace declaration with that of...
12986         (make_pass_forwprop): ...new function.
12987         (pass_phiprop): Replace declaration with that of...
12988         (make_pass_phiprop): ...new function.
12989         (pass_tree_ifcombine): Replace declaration with that of...
12990         (make_pass_tree_ifcombine): ...new function.
12991         (pass_dse): Replace declaration with that of...
12992         (make_pass_dse): ...new function.
12993         (pass_nrv): Replace declaration with that of...
12994         (make_pass_nrv): ...new function.
12995         (pass_rename_ssa_copies): Replace declaration with that of...
12996         (make_pass_rename_ssa_copies): ...new function.
12997         (pass_sink_code): Replace declaration with that of...
12998         (make_pass_sink_code): ...new function.
12999         (pass_fre): Replace declaration with that of...
13000         (make_pass_fre): ...new function.
13001         (pass_check_data_deps): Replace declaration with that of...
13002         (make_pass_check_data_deps): ...new function.
13003         (pass_copy_prop): Replace declaration with that of...
13004         (make_pass_copy_prop): ...new function.
13005         (pass_vrp): Replace declaration with that of...
13006         (make_pass_vrp): ...new function.
13007         (pass_uncprop): Replace declaration with that of...
13008         (make_pass_uncprop): ...new function.
13009         (pass_return_slot): Replace declaration with that of...
13010         (make_pass_return_slot): ...new function.
13011         (pass_reassoc): Replace declaration with that of...
13012         (make_pass_reassoc): ...new function.
13013         (pass_rebuild_cgraph_edges): Replace declaration with that of...
13014         (make_pass_rebuild_cgraph_edges): ...new function.
13015         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
13016         (make_pass_remove_cgraph_callee_edges): ...new function.
13017         (pass_build_cgraph_edges): Replace declaration with that of...
13018         (make_pass_build_cgraph_edges): ...new function.
13019         (pass_local_pure_const): Replace declaration with that of...
13020         (make_pass_local_pure_const): ...new function.
13021         (pass_tracer): Replace declaration with that of...
13022         (make_pass_tracer): ...new function.
13023         (pass_warn_unused_result): Replace declaration with that of...
13024         (make_pass_warn_unused_result): ...new function.
13025         (pass_diagnose_tm_blocks): Replace declaration with that of...
13026         (make_pass_diagnose_tm_blocks): ...new function.
13027         (pass_lower_tm): Replace declaration with that of...
13028         (make_pass_lower_tm): ...new function.
13029         (pass_tm_init): Replace declaration with that of...
13030         (make_pass_tm_init): ...new function.
13031         (pass_tm_mark): Replace declaration with that of...
13032         (make_pass_tm_mark): ...new function.
13033         (pass_tm_memopt): Replace declaration with that of...
13034         (make_pass_tm_memopt): ...new function.
13035         (pass_tm_edges): Replace declaration with that of...
13036         (make_pass_tm_edges): ...new function.
13037         (pass_split_functions): Replace declaration with that of...
13038         (make_pass_split_functions): ...new function.
13039         (pass_feedback_split_functions): Replace declaration with that of...
13040         (make_pass_feedback_split_functions): ...new function.
13041         (pass_strength_reduction): Replace declaration with that of...
13042         (make_pass_strength_reduction): ...new function.
13043         (pass_ipa_lower_emutls): Replace declaration with that of...
13044         (make_pass_ipa_lower_emutls): ...new function.
13045         (pass_ipa_function_and_variable_visibility): Replace declaration with
13046         that of...
13047         (make_pass_ipa_function_and_variable_visibility): ...new function.
13048         (pass_ipa_tree_profile): Replace declaration with that of...
13049         (make_pass_ipa_tree_profile): ...new function.
13050         (pass_early_local_passes): Replace declaration with that of...
13051         (make_pass_early_local_passes): ...new function.
13052         (pass_ipa_whole_program_visibility): Replace declaration with that
13053         of...
13054         (make_pass_ipa_whole_program_visibility): ...new function.
13055         (pass_ipa_lto_gimple_out): Replace declaration with that of...
13056         (make_pass_ipa_lto_gimple_out): ...new function.
13057         (pass_ipa_increase_alignment): Replace declaration with that of...
13058         (make_pass_ipa_increase_alignment): ...new function.
13059         (pass_ipa_inline): Replace declaration with that of...
13060         (make_pass_ipa_inline): ...new function.
13061         (pass_ipa_free_lang_data): Replace declaration with that of...
13062         (make_pass_ipa_free_lang_data): ...new function.
13063         (pass_ipa_free_inline_summary): Replace declaration with that of...
13064         (make_pass_ipa_free_inline_summary): ...new function.
13065         (pass_ipa_cp): Replace declaration with that of...
13066         (make_pass_ipa_cp): ...new function.
13067         (pass_ipa_reference): Replace declaration with that of...
13068         (make_pass_ipa_reference): ...new function.
13069         (pass_ipa_pure_const): Replace declaration with that of...
13070         (make_pass_ipa_pure_const): ...new function.
13071         (pass_ipa_pta): Replace declaration with that of...
13072         (make_pass_ipa_pta): ...new function.
13073         (pass_ipa_lto_finish_out): Replace declaration with that of...
13074         (make_pass_ipa_lto_finish_out): ...new function.
13075         (pass_ipa_tm): Replace declaration with that of...
13076         (make_pass_ipa_tm): ...new function.
13077         (pass_ipa_profile): Replace declaration with that of...
13078         (make_pass_ipa_profile): ...new function.
13079         (pass_ipa_cdtor_merge): Replace declaration with that of...
13080         (make_pass_ipa_cdtor_merge): ...new function.
13081         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
13082         of...
13083         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13084         (pass_init_datastructures): Replace declaration with that of...
13085         (make_pass_init_datastructures): ...new function.
13086         (pass_fixup_cfg): Replace declaration with that of...
13087         (make_pass_fixup_cfg): ...new function.
13088         (pass_expand): Replace declaration with that of...
13089         (make_pass_expand): ...new function.
13090         (pass_instantiate_virtual_regs): Replace declaration with that of...
13091         (make_pass_instantiate_virtual_regs): ...new function.
13092         (pass_rtl_fwprop): Replace declaration with that of...
13093         (make_pass_rtl_fwprop): ...new function.
13094         (pass_rtl_fwprop_addr): Replace declaration with that of...
13095         (make_pass_rtl_fwprop_addr): ...new function.
13096         (pass_jump): Replace declaration with that of...
13097         (make_pass_jump): ...new function.
13098         (pass_jump2): Replace declaration with that of...
13099         (make_pass_jump2): ...new function.
13100         (pass_lower_subreg): Replace declaration with that of...
13101         (make_pass_lower_subreg): ...new function.
13102         (pass_cse): Replace declaration with that of...
13103         (make_pass_cse): ...new function.
13104         (pass_fast_rtl_dce): Replace declaration with that of...
13105         (make_pass_fast_rtl_dce): ...new function.
13106         (pass_ud_rtl_dce): Replace declaration with that of...
13107         (make_pass_ud_rtl_dce): ...new function.
13108         (pass_rtl_dce): Replace declaration with that of...
13109         (make_pass_rtl_dce): ...new function.
13110         (pass_rtl_dse1): Replace declaration with that of...
13111         (make_pass_rtl_dse1): ...new function.
13112         (pass_rtl_dse2): Replace declaration with that of...
13113         (make_pass_rtl_dse2): ...new function.
13114         (pass_rtl_dse3): Replace declaration with that of...
13115         (make_pass_rtl_dse3): ...new function.
13116         (pass_rtl_cprop): Replace declaration with that of...
13117         (make_pass_rtl_cprop): ...new function.
13118         (pass_rtl_pre): Replace declaration with that of...
13119         (make_pass_rtl_pre): ...new function.
13120         (pass_rtl_hoist): Replace declaration with that of...
13121         (make_pass_rtl_hoist): ...new function.
13122         (pass_rtl_store_motion): Replace declaration with that of...
13123         (make_pass_rtl_store_motion): ...new function.
13124         (pass_cse_after_global_opts): Replace declaration with that of...
13125         (make_pass_cse_after_global_opts): ...new function.
13126         (pass_rtl_ifcvt): Replace declaration with that of...
13127         (make_pass_rtl_ifcvt): ...new function.
13128         (pass_into_cfg_layout_mode): Replace declaration with that of...
13129         (make_pass_into_cfg_layout_mode): ...new function.
13130         (pass_outof_cfg_layout_mode): Replace declaration with that of...
13131         (make_pass_outof_cfg_layout_mode): ...new function.
13132         (pass_loop2): Replace declaration with that of...
13133         (make_pass_loop2): ...new function.
13134         (pass_rtl_loop_init): Replace declaration with that of...
13135         (make_pass_rtl_loop_init): ...new function.
13136         (pass_rtl_move_loop_invariants): Replace declaration with that of...
13137         (make_pass_rtl_move_loop_invariants): ...new function.
13138         (pass_rtl_unswitch): Replace declaration with that of...
13139         (make_pass_rtl_unswitch): ...new function.
13140         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
13141         (make_pass_rtl_unroll_and_peel_loops): ...new function.
13142         (pass_rtl_doloop): Replace declaration with that of...
13143         (make_pass_rtl_doloop): ...new function.
13144         (pass_rtl_loop_done): Replace declaration with that of...
13145         (make_pass_rtl_loop_done): ...new function.
13146         (pass_web): Replace declaration with that of...
13147         (make_pass_web): ...new function.
13148         (pass_cse2): Replace declaration with that of...
13149         (make_pass_cse2): ...new function.
13150         (pass_df_initialize_opt): Replace declaration with that of...
13151         (make_pass_df_initialize_opt): ...new function.
13152         (pass_df_initialize_no_opt): Replace declaration with that of...
13153         (make_pass_df_initialize_no_opt): ...new function.
13154         (pass_reginfo_init): Replace declaration with that of...
13155         (make_pass_reginfo_init): ...new function.
13156         (pass_inc_dec): Replace declaration with that of...
13157         (make_pass_inc_dec): ...new function.
13158         (pass_stack_ptr_mod): Replace declaration with that of...
13159         (make_pass_stack_ptr_mod): ...new function.
13160         (pass_initialize_regs): Replace declaration with that of...
13161         (make_pass_initialize_regs): ...new function.
13162         (pass_combine): Replace declaration with that of...
13163         (make_pass_combine): ...new function.
13164         (pass_if_after_combine): Replace declaration with that of...
13165         (make_pass_if_after_combine): ...new function.
13166         (pass_ree): Replace declaration with that of...
13167         (make_pass_ree): ...new function.
13168         (pass_partition_blocks): Replace declaration with that of...
13169         (make_pass_partition_blocks): ...new function.
13170         (pass_match_asm_constraints): Replace declaration with that of...
13171         (make_pass_match_asm_constraints): ...new function.
13172         (pass_regmove): Replace declaration with that of...
13173         (make_pass_regmove): ...new function.
13174         (pass_split_all_insns): Replace declaration with that of...
13175         (make_pass_split_all_insns): ...new function.
13176         (pass_fast_rtl_byte_dce): Replace declaration with that of...
13177         (make_pass_fast_rtl_byte_dce): ...new function.
13178         (pass_lower_subreg2): Replace declaration with that of...
13179         (make_pass_lower_subreg2): ...new function.
13180         (pass_mode_switching): Replace declaration with that of...
13181         (make_pass_mode_switching): ...new function.
13182         (pass_sms): Replace declaration with that of...
13183         (make_pass_sms): ...new function.
13184         (pass_sched): Replace declaration with that of...
13185         (make_pass_sched): ...new function.
13186         (pass_ira): Replace declaration with that of...
13187         (make_pass_ira): ...new function.
13188         (pass_reload): Replace declaration with that of...
13189         (make_pass_reload): ...new function.
13190         (pass_clean_state): Replace declaration with that of...
13191         (make_pass_clean_state): ...new function.
13192         (pass_branch_prob): Replace declaration with that of...
13193         (make_pass_branch_prob): ...new function.
13194         (pass_value_profile_transformations): Replace declaration with that
13195         of...
13196         (make_pass_value_profile_transformations): ...new function.
13197         (pass_postreload_cse): Replace declaration with that of...
13198         (make_pass_postreload_cse): ...new function.
13199         (pass_gcse2): Replace declaration with that of...
13200         (make_pass_gcse2): ...new function.
13201         (pass_split_after_reload): Replace declaration with that of...
13202         (make_pass_split_after_reload): ...new function.
13203         (pass_branch_target_load_optimize1): Replace declaration with that
13204         of...
13205         (make_pass_branch_target_load_optimize1): ...new function.
13206         (pass_thread_prologue_and_epilogue): Replace declaration with that
13207         of...
13208         (make_pass_thread_prologue_and_epilogue): ...new function.
13209         (pass_stack_adjustments): Replace declaration with that of...
13210         (make_pass_stack_adjustments): ...new function.
13211         (pass_peephole2): Replace declaration with that of...
13212         (make_pass_peephole2): ...new function.
13213         (pass_if_after_reload): Replace declaration with that of...
13214         (make_pass_if_after_reload): ...new function.
13215         (pass_regrename): Replace declaration with that of...
13216         (make_pass_regrename): ...new function.
13217         (pass_cprop_hardreg): Replace declaration with that of...
13218         (make_pass_cprop_hardreg): ...new function.
13219         (pass_reorder_blocks): Replace declaration with that of...
13220         (make_pass_reorder_blocks): ...new function.
13221         (pass_branch_target_load_optimize2): Replace declaration with that
13222         of...
13223         (make_pass_branch_target_load_optimize2): ...new function.
13224         (pass_leaf_regs): Replace declaration with that of...
13225         (make_pass_leaf_regs): ...new function.
13226         (pass_split_before_sched2): Replace declaration with that of...
13227         (make_pass_split_before_sched2): ...new function.
13228         (pass_compare_elim_after_reload): Replace declaration with that of...
13229         (make_pass_compare_elim_after_reload): ...new function.
13230         (pass_sched2): Replace declaration with that of...
13231         (make_pass_sched2): ...new function.
13232         (pass_stack_regs): Replace declaration with that of...
13233         (make_pass_stack_regs): ...new function.
13234         (pass_stack_regs_run): Replace declaration with that of...
13235         (make_pass_stack_regs_run): ...new function.
13236         (pass_df_finish): Replace declaration with that of...
13237         (make_pass_df_finish): ...new function.
13238         (pass_compute_alignments): Replace declaration with that of...
13239         (make_pass_compute_alignments): ...new function.
13240         (pass_duplicate_computed_gotos): Replace declaration with that of...
13241         (make_pass_duplicate_computed_gotos): ...new function.
13242         (pass_variable_tracking): Replace declaration with that of...
13243         (make_pass_variable_tracking): ...new function.
13244         (pass_free_cfg): Replace declaration with that of...
13245         (make_pass_free_cfg): ...new function.
13246         (pass_machine_reorg): Replace declaration with that of...
13247         (make_pass_machine_reorg): ...new function.
13248         (pass_cleanup_barriers): Replace declaration with that of...
13249         (make_pass_cleanup_barriers): ...new function.
13250         (pass_delay_slots): Replace declaration with that of...
13251         (make_pass_delay_slots): ...new function.
13252         (pass_split_for_shorten_branches): Replace declaration with that of...
13253         (make_pass_split_for_shorten_branches): ...new function.
13254         (pass_split_before_regstack): Replace declaration with that of...
13255         (make_pass_split_before_regstack): ...new function.
13256         (pass_convert_to_eh_region_ranges): Replace declaration with that
13257         of...
13258         (make_pass_convert_to_eh_region_ranges): ...new function.
13259         (pass_shorten_branches): Replace declaration with that of...
13260         (make_pass_shorten_branches): ...new function.
13261         (pass_set_nothrow_function_flags): Replace declaration with that of...
13262         (make_pass_set_nothrow_function_flags): ...new function.
13263         (pass_dwarf2_frame): Replace declaration with that of...
13264         (make_pass_dwarf2_frame): ...new function.
13265         (pass_final): Replace declaration with that of...
13266         (make_pass_final): ...new function.
13267         (pass_rtl_seqabstr): Replace declaration with that of...
13268         (make_pass_rtl_seqabstr): ...new function.
13269         (pass_release_ssa_names): Replace declaration with that of...
13270         (make_pass_release_ssa_names): ...new function.
13271         (pass_early_inline): Replace declaration with that of...
13272         (make_pass_early_inline): ...new function.
13273         (pass_inline_parameters): Replace declaration with that of...
13274         (make_pass_inline_parameters): ...new function.
13275         (pass_update_address_taken): Replace declaration with that of...
13276         (make_pass_update_address_taken): ...new function.
13277         (pass_convert_switch): Replace declaration with that of...
13278         (make_pass_convert_switch): ...new function.
13279         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
13280         to a subclass of simple_ipa_opt_pass along with...
13281         (pass_data_ipa_tree_profile): ...new pass_data instance and...
13282         (make_pass_ipa_tree_profile): ...new function.
13283         * tree-sra.c (pass_sra_early): Convert from a global struct to a
13284         subclass of gimple_opt_pass along with...
13285         (pass_data_sra_early): ...new pass_data instance and...
13286         (make_pass_sra_early): ...new function.
13287         (pass_sra): Convert from a global struct to a subclass of
13288         gimple_opt_pass along with...
13289         (pass_data_sra): ...new pass_data instance and...
13290         (make_pass_sra): ...new function.
13291         (pass_early_ipa_sra): Convert from a global struct to a subclass of
13292         gimple_opt_pass along with...
13293         (pass_data_early_ipa_sra): ...new pass_data instance and...
13294         (make_pass_early_ipa_sra): ...new function.
13295         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
13296         subclass of gimple_opt_pass along with...
13297         (pass_data_ccp): ...new pass_data instance and...
13298         (make_pass_ccp): ...new function.
13299         (pass_fold_builtins): Convert from a global struct to a subclass of
13300         gimple_opt_pass along with...
13301         (pass_data_fold_builtins): ...new pass_data instance and...
13302         (make_pass_fold_builtins): ...new function.
13303         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
13304         subclass of gimple_opt_pass along with...
13305         (pass_data_copy_prop): ...new pass_data instance and...
13306         (make_pass_copy_prop): ...new function.
13307         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
13308         global struct to a subclass of gimple_opt_pass along with...
13309         (pass_data_rename_ssa_copies): ...new pass_data instance and...
13310         (make_pass_rename_ssa_copies): ...new function.
13311         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
13312         subclass of gimple_opt_pass along with...
13313         (pass_data_dce): ...new pass_data instance and...
13314         (make_pass_dce): ...new function.
13315         (pass_dce_loop): Convert from a global struct to a subclass of
13316         gimple_opt_pass along with...
13317         (pass_data_dce_loop): ...new pass_data instance and...
13318         (make_pass_dce_loop): ...new function.
13319         (pass_cd_dce): Convert from a global struct to a subclass of
13320         gimple_opt_pass along with...
13321         (pass_data_cd_dce): ...new pass_data instance and...
13322         (make_pass_cd_dce): ...new function.
13323         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
13324         subclass of gimple_opt_pass along with...
13325         (pass_data_dominator): ...new pass_data instance and...
13326         (make_pass_dominator): ...new function.
13327         (pass_phi_only_cprop): Convert from a global struct to a subclass of
13328         gimple_opt_pass along with...
13329         (pass_data_phi_only_cprop): ...new pass_data instance and...
13330         (make_pass_phi_only_cprop): ...new function.
13331         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
13332         subclass of gimple_opt_pass along with...
13333         (pass_data_dse): ...new pass_data instance and...
13334         (make_pass_dse): ...new function.
13335         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
13336         a subclass of gimple_opt_pass along with...
13337         (pass_data_forwprop): ...new pass_data instance and...
13338         (make_pass_forwprop): ...new function.
13339         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
13340         struct to a subclass of gimple_opt_pass along with...
13341         (pass_data_tree_ifcombine): ...new pass_data instance and...
13342         (make_pass_tree_ifcombine): ...new function.
13343         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
13344         subclass of gimple_opt_pass along with...
13345         (pass_data_ch): ...new pass_data instance and...
13346         (make_pass_ch): ...new function.
13347         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
13348         subclass of gimple_opt_pass along with...
13349         (pass_data_tree_loop): ...new pass_data instance and...
13350         (make_pass_tree_loop): ...new function.
13351         (pass_tree_loop_init): Convert from a global struct to a subclass of
13352         gimple_opt_pass along with...
13353         (pass_data_tree_loop_init): ...new pass_data instance and...
13354         (make_pass_tree_loop_init): ...new function.
13355         (pass_lim): Convert from a global struct to a subclass of
13356         gimple_opt_pass along with...
13357         (pass_data_lim): ...new pass_data instance and...
13358         (make_pass_lim): ...new function.
13359         (pass_tree_unswitch): Convert from a global struct to a subclass of
13360         gimple_opt_pass along with...
13361         (pass_data_tree_unswitch): ...new pass_data instance and...
13362         (make_pass_tree_unswitch): ...new function.
13363         (pass_predcom): Convert from a global struct to a subclass of
13364         gimple_opt_pass along with...
13365         (pass_data_predcom): ...new pass_data instance and...
13366         (make_pass_predcom): ...new function.
13367         (pass_vectorize): Convert from a global struct to a subclass of
13368         gimple_opt_pass along with...
13369         (pass_data_vectorize): ...new pass_data instance and...
13370         (make_pass_vectorize): ...new function.
13371         (pass_graphite): Convert from a global struct to a subclass of
13372         gimple_opt_pass along with...
13373         (pass_data_graphite): ...new pass_data instance and...
13374         (make_pass_graphite): ...new function.
13375         (pass_graphite_transforms): Convert from a global struct to a subclass
13376         of gimple_opt_pass along with...
13377         (pass_data_graphite_transforms): ...new pass_data instance and...
13378         (make_pass_graphite_transforms): ...new function.
13379         (pass_check_data_deps): Convert from a global struct to a subclass of
13380         gimple_opt_pass along with...
13381         (pass_data_check_data_deps): ...new pass_data instance and...
13382         (make_pass_check_data_deps): ...new function.
13383         (pass_iv_canon): Convert from a global struct to a subclass of
13384         gimple_opt_pass along with...
13385         (pass_data_iv_canon): ...new pass_data instance and...
13386         (make_pass_iv_canon): ...new function.
13387         (pass_scev_cprop): Convert from a global struct to a subclass of
13388         gimple_opt_pass along with...
13389         (pass_data_scev_cprop): ...new pass_data instance and...
13390         (make_pass_scev_cprop): ...new function.
13391         (pass_record_bounds): Convert from a global struct to a subclass of
13392         gimple_opt_pass along with...
13393         (pass_data_record_bounds): ...new pass_data instance and...
13394         (make_pass_record_bounds): ...new function.
13395         (pass_complete_unroll): Convert from a global struct to a subclass of
13396         gimple_opt_pass along with...
13397         (pass_data_complete_unroll): ...new pass_data instance and...
13398         (make_pass_complete_unroll): ...new function.
13399         (pass_complete_unrolli): Convert from a global struct to a subclass of
13400         gimple_opt_pass along with...
13401         (pass_data_complete_unrolli): ...new pass_data instance and...
13402         (make_pass_complete_unrolli): ...new function.
13403         (pass_parallelize_loops): Convert from a global struct to a subclass
13404         of gimple_opt_pass along with...
13405         (pass_data_parallelize_loops): ...new pass_data instance and...
13406         (make_pass_parallelize_loops): ...new function.
13407         (pass_loop_prefetch): Convert from a global struct to a subclass of
13408         gimple_opt_pass along with...
13409         (pass_data_loop_prefetch): ...new pass_data instance and...
13410         (make_pass_loop_prefetch): ...new function.
13411         (pass_iv_optimize): Convert from a global struct to a subclass of
13412         gimple_opt_pass along with...
13413         (pass_data_iv_optimize): ...new pass_data instance and...
13414         (make_pass_iv_optimize): ...new function.
13415         (pass_tree_loop_done): Convert from a global struct to a subclass of
13416         gimple_opt_pass along with...
13417         (pass_data_tree_loop_done): ...new pass_data instance and...
13418         (make_pass_tree_loop_done): ...new function.
13419         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
13420         struct to a subclass of gimple_opt_pass along with...
13421         (pass_data_cse_reciprocals): ...new pass_data instance and...
13422         (make_pass_cse_reciprocals): ...new function.
13423         (pass_cse_sincos): Convert from a global struct to a subclass of
13424         gimple_opt_pass along with...
13425         (pass_data_cse_sincos): ...new pass_data instance and...
13426         (make_pass_cse_sincos): ...new function.
13427         (pass_optimize_bswap): Convert from a global struct to a subclass of
13428         gimple_opt_pass along with...
13429         (pass_data_optimize_bswap): ...new pass_data instance and...
13430         (make_pass_optimize_bswap): ...new function.
13431         (pass_optimize_widening_mul): Convert from a global struct to a
13432         subclass of gimple_opt_pass along with...
13433         (pass_data_optimize_widening_mul): ...new pass_data instance and...
13434         (make_pass_optimize_widening_mul): ...new function.
13435         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
13436         subclass of gimple_opt_pass along with...
13437         (pass_data_phiopt): ...new pass_data instance and...
13438         (make_pass_phiopt): ...new function.
13439         (pass_cselim): Convert from a global struct to a subclass of
13440         gimple_opt_pass along with...
13441         (pass_data_cselim): ...new pass_data instance and...
13442         (make_pass_cselim): ...new function.
13443         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
13444         subclass of gimple_opt_pass along with...
13445         (pass_data_phiprop): ...new pass_data instance and...
13446         (make_pass_phiprop): ...new function.
13447         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
13448         subclass of gimple_opt_pass along with...
13449         (pass_data_pre): ...new pass_data instance and...
13450         (make_pass_pre): ...new function.
13451         (pass_fre): Convert from a global struct to a subclass of
13452         gimple_opt_pass along with...
13453         (pass_data_fre): ...new pass_data instance and...
13454         (make_pass_fre): ...new function.
13455         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
13456         subclass of gimple_opt_pass along with...
13457         (pass_data_reassoc): ...new pass_data instance and...
13458         (make_pass_reassoc): ...new function.
13459         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
13460         subclass of gimple_opt_pass along with...
13461         (pass_data_sink_code): ...new pass_data instance and...
13462         (make_pass_sink_code): ...new function.
13463         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
13464         subclass of gimple_opt_pass along with...
13465         (pass_data_strlen): ...new pass_data instance and...
13466         (make_pass_strlen): ...new function.
13467         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
13468         struct to a subclass of gimple_opt_pass along with...
13469         (pass_data_build_alias): ...new pass_data instance and...
13470         (make_pass_build_alias): ...new function.
13471         (pass_build_ealias): Convert from a global struct to a subclass of
13472         gimple_opt_pass along with...
13473         (pass_data_build_ealias): ...new pass_data instance and...
13474         (make_pass_build_ealias): ...new function.
13475         (pass_ipa_pta): Convert from a global struct to a subclass of
13476         simple_ipa_opt_pass along with...
13477         (pass_data_ipa_pta): ...new pass_data instance and...
13478         (make_pass_ipa_pta): ...new function.
13479         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
13480         subclass of gimple_opt_pass along with...
13481         (pass_data_uncprop): ...new pass_data instance and...
13482         (make_pass_uncprop): ...new function.
13483         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
13484         global struct to a subclass of gimple_opt_pass along with...
13485         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
13486         (make_pass_late_warn_uninitialized): ...new function.
13487         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
13488         to a subclass of gimple_opt_pass along with...
13489         (pass_data_init_datastructures): ...new pass_data instance and...
13490         (make_pass_init_datastructures): ...new function.
13491         (pass_early_warn_uninitialized): Convert from a global struct to a
13492         subclass of gimple_opt_pass along with...
13493         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
13494         (make_pass_early_warn_uninitialized): ...new function.
13495         (pass_update_address_taken): Convert from a global struct to a
13496         subclass of gimple_opt_pass along with...
13497         (pass_data_update_address_taken): ...new pass_data instance and...
13498         (make_pass_update_address_taken): ...new function.
13499         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
13500         struct to a subclass of gimple_opt_pass along with...
13501         (pass_data_release_ssa_names): ...new pass_data instance and...
13502         (make_pass_release_ssa_names): ...new function.
13503         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
13504         subclass of gimple_opt_pass along with...
13505         (pass_data_stdarg): ...new pass_data instance and...
13506         (make_pass_stdarg): ...new function.
13507         * tree-switch-conversion.c (pass_convert_switch): Convert from a
13508         global struct to a subclass of gimple_opt_pass along with...
13509         (pass_data_convert_switch): ...new pass_data instance and...
13510         (make_pass_convert_switch): ...new function.
13511         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
13512         to a subclass of gimple_opt_pass along with...
13513         (pass_data_tail_recursion): ...new pass_data instance and...
13514         (make_pass_tail_recursion): ...new function.
13515         (pass_tail_calls): Convert from a global struct to a subclass of
13516         gimple_opt_pass along with...
13517         (pass_data_tail_calls): ...new pass_data instance and...
13518         (make_pass_tail_calls): ...new function.
13519         * tree-vect-generic.c (pass_lower_vector): Convert from a global
13520         struct to a subclass of gimple_opt_pass along with...
13521         (pass_data_lower_vector): ...new pass_data instance and...
13522         (make_pass_lower_vector): ...new function.
13523         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
13524         gimple_opt_pass along with...
13525         (pass_data_lower_vector_ssa): ...new pass_data instance and...
13526         (make_pass_lower_vector_ssa): ...new function.
13527         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
13528         to a subclass of gimple_opt_pass along with...
13529         (pass_data_slp_vectorize): ...new pass_data instance and...
13530         (make_pass_slp_vectorize): ...new function.
13531         (pass_ipa_increase_alignment): Convert from a global struct to a
13532         subclass of simple_ipa_opt_pass along with...
13533         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
13534         (make_pass_ipa_increase_alignment): ...new function.
13535         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
13536         gimple_opt_pass along with...
13537         (pass_data_vrp): ...new pass_data instance and...
13538         (make_pass_vrp): ...new function.
13539         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
13540         subclass of simple_ipa_opt_pass along with...
13541         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
13542         (make_pass_ipa_free_lang_data): ...new function.
13543         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
13544         gimple_opt_pass along with...
13545         (pass_data_tsan): ...new pass_data instance and...
13546         (make_pass_tsan): ...new function.
13547         (pass_tsan_O0): Convert from a global struct to a subclass of
13548         gimple_opt_pass along with...
13549         (pass_data_tsan_O0): ...new pass_data instance and...
13550         (make_pass_tsan_O0): ...new function.
13551         * var-tracking.c (pass_variable_tracking): Convert from a global
13552         struct to a subclass of rtl_opt_pass along with...
13553         (pass_data_variable_tracking): ...new pass_data instance and...
13554         (make_pass_variable_tracking): ...new function.
13555         * web.c (pass_web): Convert from a global struct to a subclass of
13556         rtl_opt_pass along with...
13557         (pass_data_web): ...new pass_data instance and...
13558         (make_pass_web): ...new function.
13559         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
13560         declaration with that of...
13561         (make_pass_mode_switch_use): ...new function.
13562         (pass_resolve_sw_modes): Replace declaration with that of...
13563         (make_pass_resolve_sw_modes): ...new function.
13564         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
13565         from a global struct to a subclass of rtl_opt_pass along with...
13566         (pass_data_mode_switch_use): ...new pass_data instance and...
13567         (make_pass_mode_switch_use): ...new function.
13568         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
13569         from a global struct to a subclass of rtl_opt_pass along with...
13570         (pass_data_resolve_sw_modes): ...new pass_data instance and...
13571         (make_pass_resolve_sw_modes): ...new function.
13572         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
13573         struct to a subclass of rtl_opt_pass along with...
13574         (pass_data_insert_vzeroupper): ...new pass_data instance and...
13575         (make_pass_insert_vzeroupper): ...new function.
13576         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
13577         global struct to a subclass of rtl_opt_pass along with...
13578         (pass_data_work_around_errata): ...new pass_data instance and...
13579         (make_pass_work_around_errata): ...new function.
13580         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
13581         struct to a subclass of rtl_opt_pass along with...
13582         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
13583         (make_pass_mips_machine_reorg2): ...new function.
13585 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13587         * passes.c (pass_manager::operator new): New.
13589 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13591         Handwritten part of conversion of passes to C++ classes.
13593         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
13594         (toplev.o): Add dep on PASS_MANAGER_H.
13595         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
13596         of early local pases to reflect this moving from a global to a
13597         member of gcc::pass_manager.
13598         (cgraph_add_new_function): Likewise.
13599         * lto-cgraph.c (lto_output_node): Update for conversion of
13600         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
13601         * passes.c (opt_pass::clone): New.
13602         (opt_pass::gate): New.
13603         (opt_pass::execute): New.
13604         (opt_pass::opt_pass): New.
13605         (pass_manager::execute_early_local_passes): New.
13606         (pass_manager::execute_pass_mode_switching): new.
13607         (finish_optimization_passes): Convert to...
13608         (pass_manager::finish_optimization_passes): ...this.
13609         (finish_optimization_passes): Update for conversion of passes to
13610         C++ classes.
13611         (register_dump_files_1): Use has_gate since we cannot portably
13612         check a vtable entry against NULL.
13613         (dump_one_pass): Likewise.
13614         (ipa_write_summaries_2): Likewise.
13615         (ipa_write_optimization_summaries_1): Likewise.
13616         (ipa_read_summaries_1): Likewise.
13617         (ipa_read_optimization_summaries_1): Likewise.
13618         (execute_ipa_stmt_fixups): Likewise.
13619         (pass_manager::pass_manager): Rewrite pass-creation, invoking
13620         pass-creation functions rather than wiring up globals, and
13621         storing the results in fields of pass_manager generated using
13622         pass-instances.def.
13623         (pass_manager::dump_profile_report): Update for conversion of
13624         passes to C++ classes.
13625         (pass_manager::execute_ipa_summary_passes): Likewise.
13626         (execute_one_ipa_transform_pass): Likewise.
13627         (execute_one_pass): Use has_gate and has_execute since we cannot
13628         portably check a vtable entry against NULL.
13629         * pass_manager.h (pass_manager::finish_optimization_passes): New.
13630         (pass_manager): Use pass-instances.def to add fields for the
13631         various pass instances.
13632         * toplev.c (finalize): Update for move of
13633         finish_optimization_passes to a method of gcc::pass_manager.
13634         * toplev.h (finish_optimization_passes): Move to method of class
13635         pass_manager.
13636         * tree-pass.h (struct pass_data): New.
13637         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
13638         (opt_pass::gate): Convert to virtual function.
13639         (opt_pass::~opt_pass): New.
13640         (opt_pass::clone): New.
13641         (opt_pass::execute): Convert to virtual function.
13642         (opt_pass::opt_pass): New.
13643         (opt_pass::ctxt_): new.
13644         (gimple_opt_pass): Convert to subclass of opt_pass.
13645         (gimple_opt_pass::gimple_opt_pass): New.
13646         (rtl_opt_pass): Convert to subclass of opt_pass.
13647         (rtl_opt_pass::rtl_opt_pass): New.
13648         (ipa_opt_pass_d): Convert to subclass of opt_pass.
13649         (ipa_opt_pass_d::ipa_opt_pass_d): New.
13650         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
13651         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
13652         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
13653         invocation of pass_mode_switching to reflect this moving from a
13654         global to a member of gcc::pass_manager.
13655         (ix86_option_override): Rework how pass_insert_vzeroupper is
13656         added to the pass_manager to reflect autogenerated changes.
13657         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
13659 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
13661         PR rtl-optimization/57708
13662         * recog.c (peep2_find_free_register): Validate all regs in a
13663         multi-reg mode.
13665 2013-08-05  Jan Hubicka  <jh@suse.cz>
13667         PR lto/57602
13668         * cgraph.c (verify_cgraph_node): Accept local flags from other
13669         partitions.
13670         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
13671         (function_and_variable_visibility): Likewise.
13672         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
13674 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13676         * graph.c (init_graph_slim_pretty_print): Remove.
13677         (print_graph_cfg): Do not call it.  Use local pretty printer.
13678         (start_graph_dump): Likewise.
13680 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13682         * gimple-pretty-print.c (buffer): Remove.
13683         (initialized): Likewise.
13684         (maybe_init_pretty_print): Likewise.
13685         (print_gimple_stmt): Do not call it.  Use non-static local
13686         pretty_printer variable.
13687         (print_gimple_expr): Likewise.
13688         (print_gimple_seq): Likewise.
13689         (gimple_dump_bb): Likewise.
13691 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13693         * asan.c (asan_pp): Remove.
13694         (asan_pp_initialized): Likewise.
13695         (asan_pp_initialize): Likewise.
13696         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
13697         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
13698         (asan_add_global): Likewise.
13700 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13702         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
13703         * diagnostic.h (diagnostic_flush_buffer): Adjust.
13704         * pretty-print.c (pp_formatted_text_data): Likewise.
13705         (pp_indent): Rename from pp_base_indent.
13706         (pp_format): Rename from pp_base_format.
13707         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
13708         (pp_format_verbatim): Rename from pp_base_format_verbatim.
13709         (pp_flush): Rename from pp_base_flush.
13710         (pp_set_line_maximum_length): Rename from
13711         pp_base_set_line_maximum_length.
13712         (pp_clear_output_area): Rename from pp_base_clear_output_area.
13713         (pp_set_prefix): Rename from pp_base_set_prefix.
13714         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
13715         (pp_emit_prefix): Rename from pp_base_emit_prefix.
13716         (pp_append_text): Rename from pp_base_append_text.
13717         (pp_formatted_text): Rename from pp_base_formatted_text.
13718         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
13719         (pp_remaining_character_count_for_line): Rename from
13720         pp_base_remaining_character_count_for_line.
13721         (pp_newline): Rename from pp_base_newline.
13722         (pp_character): Rename from pp_base_character.
13723         (pp_string): Rename from pp_base_string.
13724         (pp_maybe_space): Rename from pp_base_maybe_space.
13725         * asan.c (asan_pp_string): Adjust.
13726         (asan_emit_stack_protection): Likewise.
13727         (asan_add_global): Likewise.
13728         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
13729         * tree-mudflap.c (mf_varname_tree): Likewise.
13730         * tree-pretty-print.c (pp_tree_identifier): Rename from
13731         pp_base_tree_identifier.
13732         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
13733         Declare as function.
13735 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13737         * pretty-print.h (pp_bar_bar): New.
13738         (pp_ampersand_ampersand): Likewise.
13739         (pp_less_equal): Likewise.
13740         (pp_greater_equal): Likewise.
13741         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
13742         printer functions instead of pp_string or operators and punctuators.
13743         (dump_gimple_call): Likewise.
13744         (dump_gimple_omp_for): Likewise.
13745         (dump_gimple_transaction): Likewise.
13746         (dump_gimple_phi): Likewise.
13747         (pp_gimple_stmt_1): Likewise.
13748         * sched-vis.c (print_insn): Likewise.
13749         * tree-mudflap.c (mf_varname_tree): Likewise.
13750         * tree-pretty-print.c (dump_block_node): Likewise.
13751         (dump_generic_node): Likewise.
13753 2013-08-02  Jan Hubicka  <jh@suse.cz>
13755         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
13756         boundaries.
13757         * lto-streamer-out.c (tree_is_indexable): Results decls and
13758         parm decls are not indexable.
13759         (DFS_write_tree_body): Do not follow args and results.
13760         (hash_tree): Likewise.
13761         (output_functions): Rearrange so struct function is needed
13762         only when real body is output; be able to also ouptut abstract
13763         functions; output DECL_ARGUMENTS and DECL_RESULT.
13764         (lto_output): When not in WPA, ale store abstract functions.
13765         (write_symbol): Do not care about RESULT_DECL.
13766         (output_symbol_p): Handle correctly sbtract decls.
13767         * lto-streamer-in.c (input_function): Rearrange so struct
13768         function can be NULL at entry; allow streaming of
13769         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
13770         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
13771         sanity check during LTO.
13772         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
13773         RESULT_DECl and DECL_ARGUMENTS.
13774         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
13775         Likewise.
13777 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13779         * pretty-print.h (pp_underscore): New.
13780         (pp_comma): Tidy.
13781         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
13782         printer functions instead of pp_character.
13783         (dump_binary_rhs): Likewise.
13784         (dump_ternary_rhs): Likewise.
13785         (dump_gimple_call_args): Likewise.
13786         (pp_points_to_solution): Likewise.
13787         (dump_gimple_call): Likewise.
13788         (dump_gimple_switch): Likewise.
13789         (dump_gimple_cond): Likewise.
13790         (dump_gimple_bind): Likewise.
13791         (dump_gimple_try): Likewise.
13792         (dump_gimple_omp_for): Likewise.
13793         (dump_gimple_omp_continue): Likewise.
13794         (dump_gimple_omp_single): Likewise.
13795         (dump_gimple_omp_sections): Likewise.
13796         (dump_gimple_omp_block): Likewise.
13797         (dump_gimple_omp_critical): Likewise.
13798         (dump_gimple_transaction): Likewise.
13799         (dump_gimple_asm): Likewise.
13800         (dump_gimple_phi): Likewise.
13801         (dump_gimple_omp_parallel): Likewise.
13802         (dump_gimple_omp_task): Likewise.
13803         (dump_gimple_omp_atomic_load): Likewise.
13804         (dump_gimple_omp_atomic_store): Likewise.
13805         (dump_gimple_mem_ops): Likewise.
13806         (pp_gimple_stmt_1): Likewise.
13807         (pp_cfg_jump): Likewise.
13808         (dump_implicit_edges): Likewise.
13809         (gimple_dump_bb_for_graph): Likewise.
13810         * graph.c (draw_cfg_node): Likewise.
13811         * langhooks.c (lhd_print_error_function): Likewise.
13812         * sched-vis.c (print_exp): Likewise.
13813         (print_value): Likewise.
13814         (print_pattern): Likewise.
13815         (print_insn): Likewise.
13816         (rtl_dump_bb_for_graph): Likewise.
13817         * tree-pretty-print.c (dump_function_declaration): Likewise.
13818         (dump_array_domain): Likewise.
13819         (dump_omp_clause): Likewise.
13820         (dump_location): Likewise.
13821         (dump_generic_node): Likewise.
13822         (print_struct_decl): Likewise.
13823         * diagnostic.c (diagnostic_show_locus): Use pp_space.
13825 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
13827         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
13828         candidate table when replacing a candidate statement.
13829         (replace_rhs_if_not_dup): Likewise.
13830         (replace_one_candidate): Likewise.
13832 2013-08-02  Jan Hubicka  <jh@suse.cz>
13833             Martin Liska  <marxin.liska@gmail.com>
13835         * cgraphunit.c (add_new_function): Fix logic when adding from
13836         late IPA pass.
13837         (assemble_thunk): Rename to ...
13838         (expand_thunk); .. this one; export; get it working with
13839         general functions; make produced gimple valid.
13840         * cgraph.h (expand_thunk): Declare.
13842 2013-08-02  Jan Hubicka  <jh@suse.cz>
13844         * ipa-cp.c (gather_context_independent_values): Use
13845         ipa_get_param_move_cost.
13846         (get_replacement_map): Remove PARAM; move parameter folding
13847         into tree-inline.c
13848         (create_specialized_node): Update.
13849         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
13850         assert that we have gimple body; update move_cost.
13851         (count_formal_params): Assert that we have gimple body.
13852         (ipa_dump_param): New function.
13853         (ipa_alloc_node_params): Break out from ...
13854         (ipa_initialize_node_params): ... here.
13855         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
13856         (ipa_write_node_info): Stream move costs.
13857         (ipa_read_node_info): Read move costs.
13858         (ipa_update_after_lto_read): Do not recompute node params.
13859         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
13860         (ipa_get_param): Check we are not in WPA.
13861         (ipa_get_param_move_cost): New.
13862         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
13863         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
13864         parm numbers to be present.
13866 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
13868         PR rtl-optimization/58048
13869         * lra-constraints.c (process_alt_operands): Don't check asm
13870         operand on register.
13872 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
13874         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
13875         the implied StoreLoad barrier for atomic operations if before.
13877 2013-08-02  Jan Hubicka  <jh@suse.cz>
13878             Martin Liska  <marxin.liska@gmail.com>
13880         * cgraph.c (cgraph_function_body_availability): Do not check
13881         cgraph flags.
13882         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
13883         symtab_node_availability): Declare.
13884         * ipa.c (can_replace_by_local_alias): New.
13885         (function_and_variable_visibility): Use it.
13886         * symtab.c (symtab_for_node_and_aliases,
13887         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
13889 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
13891         PR rtl-optimization/57963
13892         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
13893         (lra_constraints): Use them.
13895 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
13897         * config/arm/types.md (define_attr "type"): Add "load_acq"
13898         and "store_rel".
13899         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
13900         changes.
13901         (cortex_a53_store1): Likewise.
13903 2013-08-01  Jan Hubicka  <jh@suse.cz>
13905         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
13906         partitions are not needed.
13908 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
13910         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
13911         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
13912         MAYBE_NON_Q_CLASS_P where appropriate.
13914 2013-08-01  Jan Hubicka  <jh@suse.cz>
13916         * cgraph.h (release_function_body): Declare.
13917         * tree.c (free_lang_data_in_decl): Free, parameters and return values
13918         of unused delcarations.
13920 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13922         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
13923         RTL form when subtracting a constant.
13925 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13927         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
13928         Generate canonical plus rtx with negated immediate instead of minus
13929         where appropriate.
13930         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
13932 2013-08-01  Jan Hubicka  <jh@suse.cz>
13934         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
13935         (cgraph_release_function_body): Likewise.
13936         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
13937         * cgraph.h (cgrpah_node): Rename abstract_and_needed
13938         to used_as_abstract_origin.
13939         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
13940         symbols used as abstract origins.
13941         * cgraphunit.c (analyze_functions): Update.
13942         * ipa.c (symtab_remove_unreachable_nodes): Recompute
13943         used_as_abstract_origin.
13944         * tree-inline.c (tree_function_versioning): Update
13945         used_as_abstract_origin; be ready for DECL_RESULT and
13946         DECL_ARGUMENTS to be NULL.
13948         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
13949         for abstract functions.
13950         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
13951         real symbols.
13953 2013-08-01  Jan Hubicka  <jh@suse.cz>
13955         * profile.c (compute_value_histograms): Fix thinko.
13957 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
13959         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
13960         aarch-common-protos.h to extra_headers.
13961         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
13962         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
13963         * config/aarch64/t-aarch64 (aarch-common.o): Define.
13965 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
13967         * config/aarch64/aarch64.md (define_attr "type"): Delete.
13968         Include "../arm/types.md".  Define "type" attribute for all patterns.
13969         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
13970         attribute changes.
13972 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
13974         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
13975         to support power8 load fusion.
13976         (fusion_gpr_mem_load): Likewise.
13978         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
13980         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
13981         declarations for power8 load fusion.
13982         (emit_fusion_gpr_load): Likewise.
13984         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
13985         tuning for power8, turn on fusion mode by default.  Turn on sign
13986         extending fusion mode if normal fusion mode is on, and we are at
13987         -O2 or -O3.
13988         (fusion_gpr_load_p): New function, return true if we can fuse an
13989         addis instruction with a dependent load to a GPR.
13990         (emit_fusion_gpr_load): Emit the instructions for power8 load
13991         fusion to GPRs.
13993         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
13994         (VSX load fusion peepholes): New peepholes to fuse together an
13995         addi instruction with a VSX load instruction.
13997         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
13998         peepholes to fuse an addis instruction with a load to a GPR base
13999         register.  If we are supporting sign extending fusions, convert
14000         sign extending loads to zero extending loads and add an explicit
14001         sign extension.
14003 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14005         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
14006         aarch-common-protos.h to extra_headers.
14007         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14008         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
14009         (arm_early_store_addr_dep): Likewise.
14010         (arm_no_early_alu_shift_dep): Likewise.
14011         (arm_no_early_alu_shift_value_dep): Likewise.
14012         (arm_no_early_mul_dep): Likewise.
14013         (arm_no_early_store_addr_dep): Likewise.
14014         (arm_mac_accumulator_is_mul_result): Likewise.
14015         (arm_mac_accumulator_is_result): Likewise.
14016         * config/arm/aarch-common.c: ... here.  New file.
14017         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
14018         here to ...
14019         (arm_early_store_addr_dep): Likewise.
14020         (arm_no_early_alu_shift_dep): Likewise.
14021         (arm_no_early_alu_shift_value_dep): Likewise.
14022         (arm_no_early_mul_dep): Likewise.
14023         (arm_no_early_store_addr_dep): Likewise.
14024         (arm_mac_accumulator_is_mul_result): Likewise.
14025         (arm_mac_accumulator_is_result): Likewise.
14026         * config/arm/aarch-common-protos.h: ... here.  New file.
14027         * config/arm/t-arm (aarch-common.o): Define.
14029 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14031         * config/arm/arm.md: Include new file "types.md".
14032         (define_attr "type"): Move from here to ...
14033         (define_attr "mul32"): Likewise.
14034         (define_attr "mul64"): Likewise.
14035         * config/arm/types.md: ... here.  New file.
14037 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14039         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
14040         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
14042 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
14044         * gen-pass-instances.awk: Fix offset of substr().
14046 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
14048         * Makefile.in (pass-instances.def): New.
14049         (passes.o): Replace dependency on passes.def with one on
14050         pass-instances.def
14052         * gen-pass-instances.awk: New.
14054         * passes.c (pass_manager::pass_manager): Use pass-instances.def
14055         rather than passes.def, updating local definition of NEXT_PASS
14056         macro to add an extra NUM parameter (currently unused).
14058 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
14060         * Makefile.in (PASS_MANAGER_H): New.
14061         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
14062         (passes.o): Likewise.
14063         (statistics.o): Likewise.
14064         (cgraphunit.o): Likewise.
14065         (context.o): Depend on PASS_MANAGER_H.
14067         * pass_manager.h: New.
14069         * cgraphunit.c (cgraph_add_new_function): Update for moves
14070         of globals to fields of pass_manager.
14071         (analyze_function): Likewise.
14072         (expand_function): Likewise.
14073         (ipa_passes): Likewise.
14074         (compile): Likewise.
14076         * context.c (context::context): New.
14077         * context.h  (context::context): New.
14078         (context::get_passes): New.
14079         (context::passes_): New.
14081         * lto-cgraph.c (input_node): Update for moves of globals to
14082         fields of pass_manager.
14084         * passes.c (all_passes): Remove, in favor of a field of the
14085         same name within the new class pass_manager.
14086         (all_small_ipa_passes): Likewise.
14087         (all_lowering_passes): Likewise.
14088         (all_regular_ipa_passes): Likewise.
14089         (all_late_ipa_passes): Likewise.
14090         (all_lto_gen_passes): Likewise.
14091         (passes_by_id): Likewise.
14092         (passes_by_id_size): Likewise.
14093         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14094         the new class pass_manager.
14095         (set_pass_for_id): Convert to...
14096         (pass_manager::set_pass_for_id): ...method.
14097         (get_pass_for_id): Convert to...
14098         (pass_manager::get_pass_for_id): ...method.
14099         (register_one_dump_file): Move body of implementation into...
14100         (pass_manager::register_one_dump_file): ...here.
14101         (register_dump_files_1): Convert to...
14102         (pass_manager::register_dump_files_1): ...method.
14103         (register_dump_files): Convert to...
14104         (pass_manager::register_dump_files): ...method.
14105         (create_pass_tab): Update for moves of globals to fields of
14106         pass_manager.
14107         (dump_passes): Move body of implementation into...
14108         (pass_manager::dump_passes): ...here.
14109         (register_pass): Move body of implementation into...
14110         (pass_manager::register_pass): ...here.
14111         (init_optimization_passes): Convert into...
14112         (pass_manager::pass_manager): ...constructor for new
14113         pass_manager class, and initialize the pass_lists array.
14114         (check_profile_consistency): Update for moves of globals to
14115         fields of pass_manager.
14116         (dump_profile_report): Move body of implementation into...
14117         (pass_manager::dump_profile_report): ...here.
14118         (ipa_write_summaries_1): Update for moves of pass lists from
14119         being globals to fields of pass_manager.
14120         (ipa_write_optimization_summaries): Likewise.
14121         (ipa_read_summaries):  Likewise.
14122         (ipa_read_optimization_summaries): Likewise.
14123         (execute_all_ipa_stmt_fixups): Likewise.
14125         * statistics.c (statistics_fini): Update for moves of globals to
14126         fields of pass_manager.
14128         * toplev.c (general_init): Replace call to
14129         init_optimization_passes with construction of the pass_manager
14130         instance.
14132         * tree-pass.h (all_passes): Remove, in favor of a field of the
14133         same name within the new class pass_manager.
14134         (all_small_ipa_passes): Likewise.
14135         (all_lowering_passes): Likewise.
14136         (all_regular_ipa_passes): Likewise.
14137         (all_lto_gen_passes): Likewise.
14138         (all_late_ipa_passes): Likewise.
14139         (passes_by_id): Likewise.
14140         (passes_by_id_size): Likewise.
14141         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14142         the new class pass_manager.
14143         (get_pass_for_id): Remove.
14145 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14147         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
14148         configs.
14150 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14152         * arm.md (mulhi3): New expand pattern.
14154 2013-07-30  Jan Hubicka  <jh@suse.cz>
14155             Martin Liska  <marxin.liska@gmail.com>
14157         * profile.c (compute_value_histograms): Do not ICE when
14158         there is mismatch only on some counters.
14160 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
14162         PR rtl-optimization/57637
14163         * function.c (move_insn_for_shrink_wrap): Also check the
14164         GEN set of the LIVE problem for the liveness analysis
14165         if it exists, otherwise give up.
14167 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14169         PR tree-optimization/57993
14170         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
14171         replaced statement in the candidate table.
14172         (phi_add_costs): Return infinite cost when the hidden basis does
14173         not dominate all phis on which the candidate is dependent.
14174         (replace_one_candidate): Record replaced statement in the
14175         candidate table.
14177 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
14179         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
14180         (ashlv2si3): New expander.
14181         (*ashlv2si3_i): New define_insn_and_split.
14182         * predicates.md (float_operation): Allow patterns with three
14183         basic sub-patterns.
14185         PR rtl-optimization/58021
14186         * mode-switching.c (create_pre_exit): Always split off preceding
14187         insns if we are not at the basic block head.
14189 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
14191         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
14192         (UCLIBC_DYNAMIC_LINKER): New macro.
14193         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
14194         `-mnan=2008'.
14195         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
14196         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
14197         `-mnan=2008'.
14198         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
14199         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
14200         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
14201         for SF and DF modes.  Use ieee_quad_format for TF mode.
14202         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
14203         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
14204         (mips_option_override): Handle `-mnan=legacy'.
14205         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
14206         `-mabs=2008' and `-mnan=2008'.
14207         (OPTION_DEFAULT_SPECS): Add "nan" default.
14208         (ASM_SPEC): Handle `-mnan='.
14209         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
14210         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
14211         comment accordingly.
14212         (neg<mode>2): Likewise.
14213         * config/mips/mips.opt (mabs, mnan): New options.
14214         * doc/install.texi (Configuration): Document `--with-nan=' option.
14215         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
14216         `-mnan=' options.
14217         (MIPS Options): Document them.
14218         * config.gcc <mips*-*-*>: Handle `--with-nan='.
14219         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
14220         * configure: Regenerate.
14221         * config.in: Regenerate.
14223 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14225         * config/i386/i386.md (float post-reload splitters): Do not check
14226         for subregs of SSE registers.
14228 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14229             H.J. Lu  <hongjiu.lu@intel.com>
14231         PR target/57954
14232         PR target/57988
14233         * config/i386/i386.md (post-reload splitter
14234         to avoid partial SSE reg dependency stalls): New pattern.
14236 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
14238         * config/s390/s390.md ("movcc"): Swap load and store instructions.
14240 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
14242         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
14243         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
14245 2013-07-26  Cary Coutant  <ccoutant@google.com>
14247         * dwarf2out.c (die_checksum_ordered): Don't include template
14248         instantiations in signature.
14249         (is_template_parameter): New function.
14250         (is_template_instantiation): New function.
14251         (generate_skeleton_bottom_up): Don't include template instantiations
14252         in type unit DIE.
14253         (generate_skeleton): Likewise.
14254         (break_out_comdat_types): Move recursive call to break out nested
14255         types earlier.
14256         (prune_unused_types_mark_generic_parms_dies): Call
14257         is_template_parameter.
14259 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
14261         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
14262         uses vector registers.
14263         * config/aarch64/iterators.md: Add attributes rtn and vas.
14265 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14266             Richard Earnshaw  <richard.earnshaw@arm.com>
14268         * combine.c (simplify_comparison): Re-canonicalize operands
14269         where appropriate.
14270         * config/arm/arm.md (movcond_addsi): New splitter.
14272 2013-07-25  Sterling Augustine  <saugustine@google.com>
14274         * dwarf2out.c (size_of_pubnames): Move code to...
14275         (include_pubname_in_output): ...here.  New.
14276         (want_pubnames): Rearrange.
14277         (output_pubnames): Call include_pubname_in_output.  Move assertion.
14279 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14281         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
14283 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14285         PR target/38836
14286         * doc/extend.texi: Remove obsolete builtins. Fix
14287         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
14289 2013-07-25  Jan Hubicka  <jh@suse.cz>
14291         * cgraph.c (release_function_body): Break out from ...
14292         (cgraph_release_function_body): ... this one; also release DECL_RESULT
14293         and DECL_ARGUMENTS.
14294         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
14295         old_tree in the map.
14296         (create_specialized_node): Update.
14297         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
14298         into index.
14299         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
14300         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
14301         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
14302         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
14303         DECL_RESULT.
14305 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14307         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
14308         addsi3_carryin_alt2_<optab>): Correct output template.
14310 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14312         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
14313         Adjust for arm_restrict_it.
14314         Remove trailing whitespace.
14316 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
14318         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
14319         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
14321         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
14323 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
14325         PR rtl-optimization/57960
14326         * lra-constraints.c (process_alt_operands): Use the right mode
14327         when checking strict_low.
14329 2013-07-25  Jan Hubicka  <jh@suse.cz>
14331         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
14332         * cgraph.c (cgraph_remove_node): Do not release function body
14333         when in cgraph streaming.
14334         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
14335         in other partitions are not considered reachable; fix handling of
14336         clones.
14338 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14340         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
14342 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14344         PR target/19599
14345         PR target/57731
14346         PR target/57837
14347         * config/arm/arm.md ("*sibcall_insn): Replace use of
14348         Ss with US. Adjust output for v5 and v4t.
14349         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
14351         * config/arm/constraints.md ("Ss"): Rename to US.
14353 2013-07-25  Terry Guo  <terry.guo@arm.com>
14355         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
14356         shift_add/shift_sub0/shift_sub1 RTXs.
14358 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
14359             Anton Blanchard  <anton@au1.ibm.com>
14361         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
14362         (altivec_vpks<VI_char>ss): Likewise.
14363         (altivec_vpks<VI_char>us): Likewise.
14364         (altivec_vpku<VI_char>us): Likewise.
14365         (altivec_vpku<VI_char>um): Likewise.
14367 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
14369         Introduce context class.
14371         * Makefile.in (CONTEXT_H): New.
14372         (OBJS): Add context.o.
14373         (toplev.o): Add CONTEXT_H to dependencies.
14374         (context.o): New.
14376         * toplev.c (general_init): Create the singleton gcc::context instance.
14378         * context.c: New.
14380         * context.h: New.
14382 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
14384         PR rtl-optimization/57968
14385         * mode-switching.c (create_pre_exit): Allow instructions that
14386         don't set a return register to need a non-exit mode.
14388 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14389             Anton Blanchard  <anton@au1.ibm.com>
14391         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
14392         operands to vperm for little endian.
14393         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
14394         of lvsl to create the control mask for a vperm for little endian.
14396 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14397             Anton Blanchard  <anton@au1.ibm.com>
14399         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
14400         two operands for little-endian.
14402 2013-07-23  Steve Ellcey  <sellcey@mips.com>
14404         * config/mips/mips.c (mips_case_values_threshold): New.
14405         (TARGET_CASE_VALUES_THRESHOLD): Define.
14407 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14408             Anton Blanchard  <anton@au1.ibm.com>
14410         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
14411         selection of field for vector splat in little endian mode.
14413 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
14415         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
14416         expanders to rs6000.md.
14417         (ior<mode>3): Likewise.
14418         (and<mode>3): Likewise.
14419         (one_cmpl<mode>2): Likewise.
14420         (nor<mode>3): Likewise.
14421         (andc<mode>3): Likewise.
14422         (eqv<mode>3): Likewise.
14423         (nand<mode>3): Likewise.
14424         (orc<mode>3): Likewise.
14426         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
14427         declaration.
14429         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
14430         to split multi-word logical operations.
14431         (rs6000_split_logical_di): Likewise.
14432         (rs6000_split_logical): Likewise.
14434         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
14435         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
14436         and allow TImode operations in 32-bit.
14437         (vsx_and<mode>3_64bit): Likewise.
14438         (vsx_ior<mode>3_32bit): Likewise.
14439         (vsx_ior<mode>3_64bit): Likewise.
14440         (vsx_xor<mode>3_32bit): Likewise.
14441         (vsx_xor<mode>3_64bit): Likewise.
14442         (vsx_one_cmpl<mode>2_32bit): Likewise.
14443         (vsx_one_cmpl<mode>2_64bit): Likewise.
14444         (vsx_nor<mode>3_32bit): Likewise.
14445         (vsx_nor<mode>3_64bit): Likewise.
14446         (vsx_andc<mode>3_32bit): Likewise.
14447         (vsx_andc<mode>3_64bit): Likewise.
14448         (vsx_eqv<mode>3_32bit): Likewise.
14449         (vsx_eqv<mode>3_64bit): Likewise.
14450         (vsx_nand<mode>3_32bit): Likewise.
14451         (vsx_nand<mode>3_64bit): Likewise.
14452         (vsx_orc<mode>3_32bit): Likewise.
14453         (vsx_orc<mode>3_64bit): Likewise.
14455         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
14456         logical types in GPRs.
14458         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
14459         logical insns to rs6000.md, and allow TImode operations in
14460         32-bit.
14461         (altivec_ior<mode>3): Likewise.
14462         (altivec_xor<mode>3): Likewise.
14463         (altivec_one_cmpl<mode>2): Likewise.
14464         (altivec_nor<mode>3): Likewise.
14465         (altivec_andc<mode>3): Likewise.
14467         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
14468         attributes for moving the 128-bit logical operations into
14469         rs6000.md.
14470         (BOOL_REGS_OUTPUT): Likewise.
14471         (BOOL_REGS_OP1): Likewise.
14472         (BOOL_REGS_OP2): Likewise.
14473         (BOOL_REGS_UNARY): Likewise.
14474         (BOOL_REGS_AND_CR0): Likewise.
14475         (one_cmpl<mode>2): Add support for DI logical operations on
14476         32-bit, splitting the operations to 32-bit.
14477         (anddi3): Likewise.
14478         (iordi3): Likewise.
14479         (xordi3): Likewise.
14480         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
14481         changes to combine the 32/64-bit code, allow logical operations on
14482         TI mode in 32-bit, and to use similar match_operator patterns like
14483         scalar mode uses.  Combine the Altivec and VSX code for logical
14484         operations, and move it here.
14485         (ior<mode>3, 128-bit types): Likewise.
14486         (xor<mode>3, 128-bit types): Likewise.
14487         (one_cmpl<mode>3, 128-bit types): Likewise.
14488         (nor<mode>3, 128-bit types): Likewise.
14489         (andc<mode>3, 128-bit types): Likewise.
14490         (eqv<mode>3, 128-bit types): Likewise.
14491         (nand<mode>3, 128-bit types): Likewise.
14492         (orc<mode>3, 128-bit types): Likewise.
14493         (and<mode>3_internal): Likewise.
14494         (bool<mode>3_internal): Likewise.
14495         (boolc<mode>3_internal1): Likewise.
14496         (boolc<mode>3_internal2): Likewise.
14497         (boolcc<mode>3_internal1): Likewise.
14498         (boolcc<mode>3_internal2): Likewise.
14499         (eqv<mode>3_internal1): Likewise.
14500         (eqv<mode>3_internal2): Likewise.
14501         (one_cmpl1<mode>3_internal): Likewise.
14503 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14505         * config/microblaze/microblaze.c (microblaze_expand_prologue):
14506         Rename flag_stack_usage to flag_stack_usage_info.
14508 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14510         * config/microblaze/sync.md: New file.
14511         * config/microblaze/microblaze.md: Include sync.md
14512         * config/microblaze/microblaze.c: Add print_operand 'y'.
14513         * config/microblaze/constraints.md: Add memory_contraint
14514         'Q' which is a single register.
14516 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
14518         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
14520 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14522         * reload.c (find_reloads): Exit loop once we find this operand
14523         cannot be reloaded somehow for this alternative.
14525         * reload.c (find_reloads): Exit loop once we find a hard register.
14527         * rtlanal.c (computed_jump_p): Exit loop once we find label
14528         reference is used.
14530         * i386.c (ix86_pad_returns): Exit loop after setting replace.
14532         * cfgloopmanip.c (remove_path): Exit loop after setting
14533         irred_invalidated.
14535         * gensupport.c (subst_dup): Avoid loop if code is not
14536         MATCH_DUP nor MATCH_OP_DUP.
14538 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
14540         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
14542 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14544         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
14545         true for SP_REGNUM if mode == ptr_mode.
14546         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
14547         with value R0_REGNUM + 31.
14549 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14551         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
14552         pad pointer-typed argument downward.
14554 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14556         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
14557         and __ILP32__ when the ILP32 model is in use.
14559 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14561         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
14562         (aarch64_load_symref_appropriately): In the case of
14563         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
14564         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
14565         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
14566         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
14567         if 'mode' doesn't equal to 'ptr_mode'.
14568         (aarch64_output_mi_thunk): Add an assertion on the alignment of
14569         'vcall_offset'; change to call aarch64_emit_move differently depending
14570         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
14571         to calculate the upper bound of 'vcall_offset'.
14572         (aarch64_cannot_force_const_mem): Change to also return true if
14573         mode != ptr_mode.
14574         (aarch64_legitimize_reload_address): In the case of large
14575         displacements, add new local variable 'xmode' and an assertion
14576         based on it; change to use 'xmode' to generate the new rtx and
14577         reload.
14578         (aarch64_asm_trampoline_template): Change to generate the template
14579         differently depending on TARGET_ILP32 or not; change to use
14580         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
14581         (aarch64_trampoline_size): Removed.
14582         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
14583         and replace immediate literals with it.  Change to use 'ptr_mode'
14584         instead of 'DImode' and call convert_memory_address if the mode
14585         of 'fnaddr' doesn't equal to 'ptr_mode'.
14586         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
14587         to output symbol.
14588         (aarch64_elf_asm_destructor): Likewise.
14589         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
14590         on TARGET_ILP32 instead of aarch64_trampoline_size.
14591         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
14592         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
14593         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
14594         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
14595         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
14596         (storewb_pair<GPI:mode>_<P:mode>): ... this.
14597         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
14598         depending on the value of 'mode'.
14599         (add_losym_<mode>): New.
14600         (ldr_got_small_<mode>): New, based on ldr_got_small.
14601         (ldr_got_small): Remove.
14602         (ldr_got_small_sidi): New.
14603         * config/aarch64/iterators.md (P): New.
14604         (PTR): Change to 'ptr_mode' in the condition.
14606 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14608         * config.gcc (aarch64*-*-*): Support --with-abi.
14609         (aarch64*-*-elf): Support --with-multilib-list.
14610         (aarch64*-*-linux*): Likewise.
14611         (supported_defaults): Add abi to aarch64*-*-*.
14612         * configure.ac: Mention AArch64 for --with-multilib-list.
14613         * configure: Re-generated.
14614         * config/aarch64/biarchilp32.h: New file.
14615         * config/aarch64/biarchlp64.h: New file.
14616         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
14617         (ABI_SPEC): Ditto.
14618         (MULTILIB_DEFAULTS): Ditto.
14619         (DRIVER_SELF_SPECS): Ditto.
14620         (ASM_SPEC): Update to also substitute -mabi.
14621         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
14622         file whose name depends on -mabi= and -mbig-endian.
14623         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
14624         TARGET_ILP32.
14625         (POINTER_SIZE): New define.
14626         (POINTERS_EXTEND_UNSIGNED): Ditto.
14627         (enum aarch64_abi_type): New enumeration tag.
14628         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
14629         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
14630         (TARGET_ILP32): New define.
14631         * config/aarch64/aarch64.opt (mabi): New.
14632         (aarch64_abi): New.
14633         (ilp32, lp64): New values for -mabi.
14634         * config/aarch64/t-aarch64 (comma): New define.
14635         (MULTILIB_OPTIONS): Ditto.
14636         (MULTILIB_DIRNAMES): Ditto.
14637         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
14638         * doc/invoke.texi: Document -mabi for AArch64.
14640 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
14642         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
14644 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14645             Anton Blanchard  <anton@au1.ibm.com>
14647         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
14648         endianness when selecting field to splat.
14650 2013-07-22  Eric Christopher  <echristo@gmail.com>
14652         * dwarf2out.c (die_odr_checksum): New function to use
14653         CHECKSUM_ macros and ULEB128 for DIE tag.
14654         (generate_type_signature): Use.
14656 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
14658         * config.gcc (sparc*-*-*): Accept leon3 processor.
14659         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
14660         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
14661         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
14662         * config/sparc/sparc.opt (enum processor_type): Add leon3.
14663         (mfix-ut699): Adjust comment.
14664         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
14665         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
14666         (CPP_CPU_SPEC): Likewise.
14667         (ASM_CPU_SPEC): Likewise.
14668         * config/sparc/sparc.c (leon3_cost): New constant.
14669         (sparc_option_override): Add leon3 support.
14670         (mem_ref): New function.
14671         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
14672         (sparc_do_work_around_errata): Look into the instruction in the delay
14673         slot and adjust accordingly.  Add fix for the data cache nullify issues
14674         of the UT699.  Change insertion position for the NOP.
14675         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
14676         (leon3_load): New reservation.
14677         (leon_store): Bump latency to 2.
14678         (grfpu): New automaton.
14679         (grfpu_alu): New unit.
14680         (grfpu_ds): Likewise.
14681         (leon_fp_alu): Adjust.
14682         (leon_fp_mult): Delete.
14683         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
14684         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
14685         * config/sparc/sparc.md (cpu): Add leon3.
14686         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
14687         (swapsi): Likewise.
14688         (atomic_test_and_set): Likewise.
14689         (ldstub): Likewise.
14691 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
14693         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
14694         default for R5900 targets.
14695         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
14696         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
14697         * config/mips/mips.c (mips_option_override): Report an error for
14698         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
14699         for -march=r5900 -mhard-float.
14701 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14703         * df-problems.c (can_move_insns_across): Exit loop once we
14704         find a non-fixed, non-global register.
14706         * ipa-pure-const.c (propagate_nothrow): Exit loop after
14707         setting can_throw.
14709         * omega.c (omega_eliminate_red): Break after setting red_found.
14710         (omega_problem_has_red_equations): Similarly after setting found.
14711         (omega_query_variable): Similarly after setting coupled.
14713 2013-07-22  Marek Polacek  <polacek@redhat.com>
14715         * gimplify.c: Don't include gimple.h twice.
14717 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14719         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
14720         instead of TARGET_THUMB1.
14721         (Pz): New constraint.
14722         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
14723         encodings.
14724         (compare_negsi_si): Likewise.
14725         (compare_addsi2_op0): Likewise.
14726         (compare_addsi2_op1): Likewise.
14727         (addsi3_carryin_<optab>): Likewise.
14728         (addsi3_carryin_alt2_<optab>): Likewise.
14729         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
14730         for arm_restrict_it.
14731         (subsi3_carryin): Likewise.
14732         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
14733         (minmax_arithsi): Disable for arm_restrict_it.
14734         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
14735         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
14736         (satsi_<SAT:code>_shift): Likewise.
14737         (arm_shiftsi3): Add alternative for 16-bit encoding.
14738         (arm32_movhf): Disable for arm_restrict_it.
14739         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
14740         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
14742 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
14744         * config/arm/arm.md (attribute "insn"): Delete.
14745         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
14746         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
14747         (not_shiftsi): Update for attribute change.
14748         (not_shiftsi_compare0): Likewise.
14749         (not_shiftsi_compare0_scratch): Likewise.
14750         (arm_one_cmplsi2): Likewise.
14751         (thumb1_one_cmplsi2): Likewise.
14752         (notsi_compare0): Likewise.
14753         (notsi_compare0_scratch): Likewise.
14754         (thumb1_movdi_insn): Likewise.
14755         (arm_movsi_insn): Likewise.
14756         (movhi_insn_arch4): Likewise.
14757         (movhi_bytes): Likewise.
14758         (arm_movqi_insn): Likewise.
14759         (thumb1_movqi_insn): Likewise.
14760         (arm32_movhf): Likewise.
14761         (thumb1_movhf): Likewise.
14762         (arm_movsf_soft_insn): Likewise.
14763         (thumb1_movsf_insn): Likewise.
14764         (thumb_movdf_insn): Likewise.
14765         (movsicc_insn): Likewise.
14766         (movsfcc_soft_insn): Likewise.
14767         (and_scc): Likewise.
14768         (cond_move): Likewise.
14769         (if_move_not): Likewise.
14770         (if_not_move): Likewise.
14771         (if_shift_move): Likewise.
14772         (if_move_shift): Likewise.
14773         (if_shift_shift): Likewise.
14774         (if_not_arith): Likewise.
14775         (if_arith_not): Likewise.
14776         (cond_move_not): Likewise.
14777         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
14778         (neon_mov<mode>): Likewise.
14779         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
14780         (thumb2_movsi_vfp): Likewise.
14781         (movsf_vfp): Likewise.
14782         (thumb2_movsf_vfp): Likewise.
14783         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
14784         change.
14785         (cortexa7_older_only): Likewise.
14786         (cortexa7_younger): Likewise.
14787         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
14788         (1020alu_shift_op): Likewise.
14789         (1020alu_shift_reg_op): Likewise.
14790         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
14791         (alu_shift_op): Likewise.
14792         (alu_shift_reg_op): Likewise.
14793         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
14794         (11_alu_shift_op): Likewise.
14795         (11_alu_shift_reg_op): Likewise.
14796         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
14797         (9_alu_shift_reg_op): Likewise.
14798         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
14799         change.
14800         (cortex_a15_alu_shift): Likewise.
14801         (cortex_a15_alu_shift_reg): Likewise.
14802         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
14803         (cortex_a5_alu_shift): Likewise.
14804         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
14805         change.
14806         (cortex_a53_alu_shift): Likewise.
14807         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
14808         change.
14809         (cortex_a7_alu_reg): Likewise.
14810         (cortex_a7_alu_shift): Likewise.
14811         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
14812         (cortex_a8_alu_shift): Likewise.
14813         (cortex_a8_alu_shift_reg): Likewise.
14814         (cortex_a8_mov): Likewise.
14815         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
14816         (cortex_a9_dp_shift): Likewise.
14817         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
14818         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
14819         (cortex_r4_mov): Likewise.
14820         (cortex_r4_alu_shift): Likewise.
14821         (cortex_r4_alu_shift_reg): Likewise.
14822         * config/arm/fa526.md (526_alu_op): Update for attribute change.
14823         (526_alu_shift_op): Likewise.
14824         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
14825         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
14826         (626te_alu_shift_op): Likewise.
14827         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
14828         (726te_alu_op): Likewise.
14829         (726te_alu_shift_op): Likewise.
14830         (726te_alu_shift_reg_op): Likewise.
14831         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
14832         (mp626_alu_shift_op): Likewise.
14833         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
14834         (pj4_alu_e1_conds): Likewise.
14835         (pj4_alu): Likewise.
14836         (pj4_alu_conds): Likewise.
14837         (pj4_shift): Likewise.
14838         (pj4_shift_conds): Likewise.
14839         (pj4_alu_shift): Likewise.
14840         (pj4_alu_shift_conds): Likewise.
14842 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14844         * config/arm/predicates.md (shiftable_operator_strict_it):
14845         New predicate.
14846         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
14847         Disable cond_exec version for arm_restrict_it.
14848         (thumb2_smaxsi3): Convert to generate cond_exec.
14849         (thumb2_sminsi3): Likewise.
14850         (thumb32_umaxsi3): Likewise.
14851         (thumb2_uminsi3): Likewise.
14852         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
14853         (thumb2_neg_abssi2): Likewise.
14854         (thumb2_mov_scc): Add alternative for 16-bit encoding.
14855         (thumb2_movsicc_insn): Adjust alternatives.
14856         (thumb2_mov_negscc): Disable for arm_restrict_it.
14857         (thumb2_mov_negscc_strict_it): New pattern.
14858         (thumb2_mov_notscc_strict_it): New pattern.
14859         (thumb2_mov_notscc): Disable for arm_restrict_it.
14860         (thumb2_ior_scc): Likewise.
14861         (thumb2_ior_scc_strict_it): New pattern.
14862         (thumb2_cond_move): Adjust for arm_restrict_it.
14863         (thumb2_cond_arith): Disable for arm_restrict_it.
14864         (thumb2_cond_arith_strict_it): New pattern.
14865         (thumb2_cond_sub): Adjust for arm_restrict_it.
14866         (thumb2_movcond): Likewise.
14867         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
14868         (thumb2_zero_extendhisi2_v6): Likewise.
14869         (thumb2_zero_extendqisi2_v6): Likewise.
14870         (orsi_notsi_si): Likewise.
14871         (orsi_not_shiftsi_si): Likewise.
14873 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
14875         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
14876         instruction sequence is 1 byte shorter.
14878 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
14880         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
14881         it is not needed after split.
14883 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
14885         PR target/51784
14886         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
14887         second label for nonlocal goto receivers. Don't output pic base labels
14888         unless we're producing PIC; mark that action unreachable().
14889         (ix86_save_reg): If the function contains a nonlocal label, save the
14890         PIC base reg.
14891         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
14892         * config/darwin.c (emitted_pic_label_num): New GTY.
14893         (update_pic_label_number_if_needed): New.
14894         (machopic_output_function_base_name): Adjust for nonlocal receiver
14895         case.
14896         (machopic_should_output_picbase_label): New.
14897         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
14898         (nonlocal_goto_receiver): New insn and split.
14900 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
14902         * config/aarch64/aarch64-builtins.c
14903         (aarch64_fold_builtin): Fold abs in all modes.
14904         * config/aarch64/aarch64-simd-builtins.def
14905         (abs): Enable for all modes.
14906         * config/aarch64/arm_neon.h
14907         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
14908         (vabs_f64): Add missing intrinsic.
14910 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
14912         * config/aarch64/arm_neon.h (vabs_s64): New function
14914 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
14916         PR target/57516
14917         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
14918         * config/avr/avr.md (adjust_len): Add `round'.
14919         * config/avr/avr-protos.h (avr_out_round): New prototype.
14920         (avr_out_plus): Add `out_label' argument.
14921         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
14922         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
14923         Handle the case where `insn' is just a pattern.
14924         (avr_out_bitop): Handle the case where `insn' is just a pattern.
14925         (avr_out_round): New function.
14926         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
14928 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
14930         * config/microblaze/microblaze.c (microblaze_expand_prologue):
14931         Add check for flag_stack_usage to handle -fstack-usage support
14933 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
14935         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
14936         interaction for new Power8 flags and VSX.
14938 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
14940         PR middle-end/57698
14941         * tree-inline.c (expand_call_inline): Emit errors during
14942         early_inlining only if optimization is not turned on.
14944 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14946         * passes.def: New.
14948         * passes.c (init_optimization_passes): Move the construction of
14949         the pass hierarchy into a new passes.def file.
14951         * Makefile.in (passes.o): Add dependency on passes.def.
14953 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14955         * passes.c (init_optimization_passes): Introduce macros for
14956         constructing the tree of passes (INSERT_PASSES_AFTER,
14957         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
14958         TERMINATE_PASS_LIST).
14960 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
14961             Wei Mi  <wmi@google.com>
14963         PR rtl-optimization/57878
14964         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
14965         top.
14966         (reload_pseudo_compare_func): Check nregs first for reload
14967         pseudos.
14969 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
14971         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
14973 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
14975         * read-rtl.c (validate_const_int): Once an invalid character is
14976         seen, quit the loop.
14978         * gengtype.c (write_roots): Similarly once we find the "deletable"
14979         or "if_marked" option.
14981 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
14983         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
14984         "xtab" and "sat".  Move value "clz" from here to ...
14985         (attriubte "type"): ... here.
14986         (satsi_<SAT:code>): Delete "insn" attribute.
14987         (satsi_<SAT:code>_shift): Likewise.
14988         (arm_zero_extendqisi2addsi): Likewise.
14989         (arm_extendqisi2addsi): Likewise.
14990         (clzsi2): Update for attribute changes.
14991         (rbitsi2): Likewise.
14992         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
14993         attribute.
14994         (arm_usatsihi): Likewise.
14995         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
14997 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
14999         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
15000         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
15001         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
15002         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
15003         in alphabetical order.
15004         (attribute "core_cycles"): Update for attribute changes.
15005         (arm_addsi3): Likewise.
15006         (addsi3_compare0): Likewise.
15007         (addsi3_compare0_scratch): Likewise.
15008         (addsi3_compare_op1): Likewise.
15009         (addsi3_compare_op2): Likewise.
15010         (compare_addsi2_op0): Likewise.
15011         (compare_addsi2_op1): Likewise.
15012         (addsi3_carryin_shift_<optab>): Likewise.
15013         (subsi3_carryin_shift): Likewise.
15014         (rsbsi3_carryin_shift): Likewise.
15015         (arm_subsi3_insn): Likewise.
15016         (subsi3_compare0): Likewise.
15017         (subsi3_compare): Likewise.
15018         (arm_andsi3_insn): Likewise.
15019         (thumb1_andsi3_insn): Likewise.
15020         (andsi3_compare0): Likewise.
15021         (andsi3_compare0_scratch): Likewise.
15022         (zeroextractsi_compare0_scratch
15023         (andsi_not_shiftsi_si): Likewise.
15024         (iorsi3_insn): Likewise.
15025         (iorsi3_compare0): Likewise.
15026         (iorsi3_compare0_scratch): Likewise.
15027         (arm_xorsi3): Likewise.
15028         (thumb1_xorsi3_insn): Likewise.
15029         (xorsi3_compare0): Likewise.
15030         (xorsi3_compare0_scratch): Likewise.
15031         (satsi_<SAT:code>_shift): Likewise.
15032         (rrx): Likewise.
15033         (arm_shiftsi3): Likewise.
15034         (shiftsi3_compare0): Likewise.
15035         (not_shiftsi): Likewise.
15036         (not_shiftsi_compare0): Likewise.
15037         (not_shiftsi_compare0_scratch): Likewise.
15038         (arm_one_cmplsi2): Likewise.
15039         (thumb_one_complsi2): Likewise.
15040         (notsi_compare0): Likewise.
15041         (notsi_compare0_scratch): Likewise.
15042         (thumb1_zero_extendhisi2): Likewise.
15043         (arm_zero_extendhisi2): Likewise.
15044         (arm_zero_extendhisi2_v6): Likewise.
15045         (arm_zero_extendhisi2addsi): Likewise.
15046         (thumb1_zero_extendqisi2): Likewise.
15047         (thumb1_zero_extendqisi2_v6): Likewise.
15048         (arm_zero_extendqisi2): Likewise.
15049         (arm_zero_extendqisi2_v6): Likewise.
15050         (arm_zero_extendqisi2addsi): Likewise.
15051         (thumb1_extendhisi2): Likewise.
15052         (arm_extendhisi2): Likewise.
15053         (arm_extendhisi2_v6): Likewise.
15054         (arm_extendqisi): Likewise.
15055         (arm_extendqisi_v6): Likewise.
15056         (arm_extendqisi2addsi): Likewise.
15057         (thumb1_extendqisi2): Likewise.
15058         (thumb1_movdi_insn): Likewise.
15059         (arm_movsi_insn): Likewise.
15060         (movsi_compare0): Likewise.
15061         (movhi_insn_arch4): Likewise.
15062         (movhi_bytes): Likewise.
15063         (arm_movqi_insn): Likewise.
15064         (thumb1_movqi_insn): Likewise.
15065         (arm32_movhf): Likewise.
15066         (thumb1_movhf): Likewise.
15067         (arm_movsf_soft_insn): Likewise.
15068         (thumb1_movsf_insn): Likewise.
15069         (movdf_soft_insn): Likewise.
15070         (thumb_movdf_insn): Likewise.
15071         (arm_cmpsi_insn): Likewise.
15072         (cmpsi_shiftsi): Likewise.
15073         (cmpsi_shiftsi_swp): Likewise.
15074         (arm_cmpsi_negshiftsi_si): Likewise.
15075         (movsicc_insn): Likewise.
15076         (movsfcc_soft_insn): Likewise.
15077         (arith_shiftsi): Likewise.
15078         (arith_shiftsi_compare0
15079         (arith_shiftsi_compare0_scratch
15080         (sub_shiftsi): Likewise.
15081         (sub_shiftsi_compare0
15082         (sub_shiftsi_compare0_scratch
15083         (and_scc): Likewise.
15084         (cond_move): Likewise.
15085         (if_plus_move): Likewise.
15086         (if_move_plus): Likewise.
15087         (if_move_not): Likewise.
15088         (if_not_move): Likewise.
15089         (if_shift_move): Likewise.
15090         (if_move_shift): Likewise.
15091         (if_shift_shift): Likewise.
15092         (if_not_arith): Likewise.
15093         (if_arith_not): Likewise.
15094         (cond_move_not): Likewise.
15095         (thumb1_ashlsi3): Set type attribute.
15096         (thumb1_ashrsi3): Likewise.
15097         (thumb1_lshrsi3): Likewise.
15098         (thumb1_rotrsi3): Likewise.
15099         (shiftsi3_compare0_scratch): Likewise.
15100         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
15101         (neon_mov<mode>): Likewise.
15102         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
15103         attribute changes.
15104         (thumb2_movsi_insn): Likewise.
15105         (thumb2_cmpsi_neg_shiftsi): Likewise.
15106         (thumb2_extendqisi_v6): Likewise.
15107         (thumb2_zero_extendhisi2_v6): Likewise.
15108         (thumb2_zero_extendqisi2_v6): Likewise.
15109         (thumb2_shiftsi3_short): Likewise.
15110         (thumb2_addsi3_compare0_scratch): Likewise.
15111         (orsi_not_shiftsi_si): Likewise.
15112         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
15113         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
15114         changes.
15115         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
15116         (1020alu_shift_op): Likewise.
15117         (1020alu_shift_reg_op): Likewise.
15118         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
15119         (alu_shift_op): Likewise.
15120         (alu_shift_reg_op): Likewise.
15121         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
15122         (11_alu_shift_op): Likewise.
15123         (11_alu_shift_reg_op): Likewise.
15124         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
15125         (9_alu_shift_reg_op): Likewise.
15126         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15127         changes.
15128         (cortex_a15_alu_shift): Likewise.
15129         (cortex_a15_alu_shift_reg): Likewise.
15130         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
15131         changes.
15132         (cortex_a5_alu_shift): Likewise.
15133         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
15134         changes.
15135         (cortex_a53_alu_shift): Likewise.
15136         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15137         changes.
15138         (cortex_a7_alu_reg): Likewise.
15139         (cortex_a7_alu_shift): Likewise.
15140         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
15141         changes.
15142         (cortex_a8_alu_shift): Likewise.
15143         (cortex_a8_alu_shift_reg): Likewise.
15144         (cortex_a8_mov): Likewise.
15145         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
15146         (cortex_a9_dp_shift): Likewise.
15147         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
15148         changes.
15149         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
15150         changes.
15151         (cortex_r4_mov): Likewise.
15152         (cortex_r4_alu_shift): Likewise.
15153         (cortex_r4_alu_shift_reg): Likewise.
15154         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
15155         (526_alu_shift_op): Likewise.
15156         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
15157         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
15158         (626te_alu_shift_op): Likewise.
15159         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
15160         (726te_alu_op): Likewise.
15161         (726te_alu_shift_op): Likewise.
15162         (726te_alu_shift_reg_op): Likewise.
15163         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
15164         (mp626_alu_shift_op): Likewise.
15165         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
15166         (pj4_alu_e1_conds): Likewise.
15167         (pj4_alu): Likewise.
15168         (pj4_alu_conds): Likewise.
15169         (pj4_shift): Likewise.
15170         (pj4_shift_conds): Likewise.
15171         (pj4_alu_shift): Likewise.
15172         (pj4_alu_shift_conds): Likewise.
15173         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15174         changes.
15175         (cortexa7_older_only): Likewise.
15176         (cortexa7_younger): Likewise.
15178 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15180         * ipa-pure-const.c (generate_summary): Rename to...
15181         (pure_const_generate_summary): ... this.
15183 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
15185         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
15187 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
15189         PR target/57909
15190         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
15191         usage in HI mode.
15193 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15195         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
15196         enabled without -march=zEC12.
15197         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
15198         flags to be set.
15200 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15202         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
15203         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
15204         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
15205         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
15206         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
15207         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
15208         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
15209         ISA_HAS_FP4.
15210         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
15211         and ISA_HAS_NMADD3_NMSUB3.
15212         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
15213         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
15214         (nmsub4<mode>, nmsub3<mode>): Likewise.
15215         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
15217 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15219         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
15220         TARGET_MIPS5400 checking.
15222 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
15223             Peter Bergner  <bergner@vnet.ibm.com>
15225         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
15226         registers in the comment.
15227         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
15228         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
15229         rather than FIRST_PSEUDO_REGISTERS.
15231 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
15233         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15234         enable extra ISA flags with TARGET_HTM.
15236 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15238         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
15239         Fix comment typos.
15241 2013-07-15  Cong Hou  <congh@google.com>
15243         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
15244         in compare function for sorting.
15246 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
15248         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
15249         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
15250         * config/rs6000/rs6000.opt: Add -mhtm option.
15251         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
15252         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
15253         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
15254         __HTM__ if the HTM instructions are available.
15255         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
15256         htm_spr_reg_operand): New define_predicates.
15257         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
15258         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
15259         Include htm.md.
15260         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
15261         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
15262         HTM builtin functions.
15263         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
15264         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
15265         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
15266         (rs6000_builtin_mask_calculate): Likewise.
15267         (rs6000_option_override_internal): Likewise.
15268         (bdesc_htm): Add new HTM builtin support.
15269         (htm_spr_num): New function.
15270         (htm_spr_regno): Likewise.
15271         (rs6000_htm_spr_icode): Likewise.
15272         (htm_expand_builtin): Likewise.
15273         (htm_init_builtins): Likewise.
15274         (rs6000_expand_builtin): Add support for HTM builtin functions.
15275         (rs6000_init_builtins): Likewise.
15276         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
15277         option.
15278         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
15279         (TARGET_HTM, MASK_HTM): Define macros.
15280         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
15281         (FIXED_REGISTERS): Likewise.
15282         (CALL_USED_REGISTERS): Likewise.
15283         (CALL_REALLY_USED_REGISTERS): Likewise.
15284         (REG_ALLOC_ORDER): Likewise.
15285         (enum reg_class): Likewise.
15286         (REG_CLASS_NAMES): Likewise.
15287         (REG_CLASS_CONTENTS): Likewise.
15288         (REGISTER_NAMES): Likewise.
15289         (ADDITIONAL_REGISTER_NAMES): Likewise.
15290         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
15291         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
15292         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
15293         * config/rs6000/htm.md: New file.
15294         * config/rs6000/htmintrin.h: New file.
15295         * config/rs6000/htmxlintrin.h: New file.
15297 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15299         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15300         Define SYMBOL_TINY_GOT, update comment.
15301         * config/aarch64/aarch64.c
15302         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
15303         (aarch64_expand_mov_immediate): Likewise.
15304         (aarch64_print_operand): Likewise.
15305         (aarch64_classify_symbol): Likewise.
15306         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
15307         (ldr_got_tiny): Define.
15309 2013-07-13  Tobias Grosser  <tobias@grosser.es>
15311         PR tree-optimization/54094
15312         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
15313           scheduling dimension for the parallelism check from the polyhedral
15314           information in the AST.
15315         * graphite-dependences.c (carries_deps): Do not assume the schedule is
15316           in 2D + 1 form.
15318 2013-07-13  Jason Merrill  <jason@redhat.com>
15320         * print-tree.c (debug_vec_tree): Use debug_raw.
15321         (debug_raw (vec<tree, va_gc> &)): New.
15322         (debug_raw (vec<tree, va_gc> *)): New.
15323         * tree.h: Declare them.
15325 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
15327         * ifcvt.c (ifcvt_after_combine): New static variable.
15328         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
15329         for size.
15330         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
15331         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
15332         rest_of_handle_if_after_reload): Pass new argument for if_convert.
15334 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
15336         * config/mips/mips.c (mips_expand_call): Remove empty statement.
15338 2013-07-12  Michael Matz  <matz@suse.de>
15340         PR middle-end/55771
15341         * convert.c (convert_to_real): Reject non-float inner types.
15343 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
15345         * config/aarch64/aarch64-protos.h
15346         (aarch64_simd_immediate_valid_for_move): Remove.
15347         * config/aarch64/aarch64.c (simd_immediate_info): New member.
15348         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
15349         cases.
15350         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
15352 2013-07-11  Steve Ellcey  <sellcey@mips.com>
15354         * config/mips/mips.c (mips_conditional_register_usage): Do not
15355         use t[0-7] registers in MIPS16 mode when optimizing for size.
15357 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
15359         * config/i386/i386.c (dispatch_function_versions): Fix array
15360         indexing of function_version_info to match actual_versions.
15362 2013-07-11  Teresa Johnson  <tejohnson@google.com>
15364         * vec.h (struct va_gc): Move release out-of-line.
15365         (va_gc::release): Call ggc_free on released vec.
15367 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15369         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
15370         Require GOT register as additional operand in UNSPEC.
15371         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
15372         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15373         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15374         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
15375         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
15376         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
15377         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
15379 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
15381         PR target/57631
15382         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
15383         name seen by assembler/linker rather if available.
15385 2013-07-11  Andreas Schwab  <schwab@suse.de>
15387         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
15389 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
15391         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
15393 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
15395         * doc/tm.texi.in: Move hook documentation to ....
15396         * target.def: ... here.
15398         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
15399         text on @hook line.
15400         * doc/tm.texi: Regenerate.
15402 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
15404         PR c++/57869
15405         * doc/invoke.texi: Document -Wconditionally-supported.
15407 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15409         PR target/57844
15410         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
15411         of my_fp.
15413 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15415         PR target/57506
15416         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
15417         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
15418         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
15419         Remove duplicate devices.
15420         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
15421         * config/avr/t-multilib: Regenerate.
15422         * config/avr/avr-tables.opt: Regenerate.
15423         * doc/avr-mmcu.texi: Regenerate.
15425 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15427         PR target/56987
15428         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
15430 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
15432         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
15433         the cost of MULT when optimizing for size.
15435 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
15437         * config/cr16/cr16-protos.h: Don't include target.h.
15439 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
15441         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
15442         adjust register size for TDmode and TFmode for VSX registers.
15444 2013-07-08  Kai Tietz  <ktietz@redhat.com>
15446         PR target/56892
15447         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
15448         hook_bool_const_tree_true.
15450 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15452         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
15453         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
15454         * config/s390/s390.md: Define FPR*_REGNUM constants.
15455         Fix FPR2_REGNUM constant (18 -> 17).
15456         ("*trunc<BFP:mode><DFP_ALL:mode>2")
15457         ("*trunc<DFP_ALL:mode><BFP:mode>2")
15458         ("trunc<BFP:mode><DFP_ALL:mode>2")
15459         ("trunc<DFP_ALL:mode><BFP:mode>2")
15460         ("*extend<BFP:mode><DFP_ALL:mode>2")
15461         ("*extend<DFP_ALL:mode><BFP:mode>2")
15462         ("extend<BFP:mode><DFP_ALL:mode>2")
15463         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
15464         FPR4_REGNUM.
15466 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
15468         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
15470 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15472         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
15473         and cfun_fpr_bit_p to cfun_fpr_save_p.
15474         (s390_frame_area, s390_register_info, s390_frame_info)
15475         (s390_emit_prologue, s390_emit_epilogue)
15476         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
15477         register numbers.
15478         * config/s390/s390.h: Define *_REGNUM macros for floating point
15479         register numbers.
15481 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
15483         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
15485 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
15487         PR rtl-optimization/57786
15488         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
15489         and break out of the loop when it is set to false.
15491 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
15493         PR target/57819
15494         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
15495         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
15496         (const_int 63)) 0)).
15497         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
15498         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
15499         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
15501         PR rtl-optimization/57829
15502         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
15503         mask bits outside of mode are just sign-extension from mode to HWI.
15505 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
15507         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
15508         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
15509         adjust_address instead of change_address to keep info about alignment.
15510         (emit_strmov): Remove.
15511         (emit_memmov): New function.
15512         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
15513         (expand_movmem_epilogue): Likewise and return updated rtx for
15514         destination.
15515         (expand_constant_movmem_prologue): Likewise and return updated rtx for
15516         destination and source.
15517         (decide_alignment): Refactor, handle vector_loop.
15518         (ix86_expand_movmem): Likewise.
15519         (ix86_expand_setmem): Likewise.
15520         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
15522 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
15524         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
15525         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
15527 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15529         * config/i386/sse.md (sse_movlhps): Change alternative 3
15530         of operand 2 to "m".
15532 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15534         PR target/57807
15535         * config/i386/sse.md (iptr): New mode attribute.
15536         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
15537         (<sse>_vm<plusminus_insn><mode>3): Ditto.
15538         (<sse>_vmmul<mode>3): Ditto.
15539         (<sse>_vmdiv<mode>3): Ditto.
15540         (sse_vmrcpv4sf2): Ditto.
15541         (<sse>_vmsqrt<mode>2): Ditto.
15542         (sse_vmrsqrtv4sf2): Ditto.
15543         (<sse>_vm<code><mode>3): Ditto.
15544         (avx_vmcmp<mode>3): Ditto.
15545         (<sse>_vmmaskcmp<mode>3): Ditto.
15546         (<sse>_comi): Ditto.
15547         (<sse>_ucomi): Ditto.
15548         (*xop_vmfrcz_<mode>): Ditto.
15549         (*fmai_fmadd_<mode>): Ditto.
15550         (*fmai_fmsub_<mode>): Ditto.
15551         (*fmai_fnmadd_<mode>): Ditto.
15552         (*fmai_fnmsub_<mode>): Ditto.
15553         (*fma4i_vmfmadd_<mode>): Ditto.
15554         (*fma4i_vmfmsub_<mode>): Ditto.
15555         (*fma4i_vmfnmadd_<mode>): Ditto.
15556         (*fma4i_vmfnmsub_<mode>): Ditto.
15557         (*xop_vmfrcz_<mode>): Ditto.
15558         (sse_cvtps2pi): Ditto.
15559         (sse_cvttps2pi): Ditto.
15560         (sse_cvtss2si): Ditto.
15561         (sse_cvtss2si_2): Ditto.
15562         (sse_cvtss2siq_2): Ditto.
15563         (sse_cvttss2si): Ditto.
15564         (sse_cvttss2siq): Ditto.
15565         (sse_cvtsd2si): Ditto.
15566         (sse_cvtsd2si_2): Ditto.
15567         (sse_cvtsd2siq_2): Ditto.
15568         (sse_cvttsd2si): Ditto.
15569         (sse_cvttsd2siq): Ditto.
15570         (sse_cvtsd2ss): Ditto.
15571         (sse_cvtss2sd): Ditto.
15572         (avx2_pbroadcast<mode>): Ditto.
15573         (avx2_pbroadcast<mode>_1): Ditto.
15574         (*avx_vperm_broadcast_v4sf): Ditto.
15576         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
15577         (sse_movlhps): Ditto.
15578         (sse_storehps): Ditto.
15579         (sse_loadhps): Ditto.
15580         (sse_storelps): Ditto.
15581         (sse_loadlps): Ditto.
15582         (*vec_concatv4sf): Ditto.
15583         (*vec_interleave_highv2df): Ditto.
15584         (*vec_interleave_lowv2df): Ditto.
15585         (*vec_extractv2df_1_sse): Ditto.
15586         (*vec_extractv2df_0_sse): Ditto.
15587         (sse2_storelpd): Ditto.
15588         (sse2_loadlpd): Ditto.
15589         (sse2_movsd): Ditto.
15590         (*vec_concatv4si): Ditto.
15591         (vec_concatv2di): Ditto.
15593         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
15594         for Intel asm dialect.
15595         (mmx_punpcklwd): Ditto.
15596         (mmx_punpckldq): Ditto.
15598         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
15599         for intel assembler dialect.
15601 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
15603         PR target/29776
15604         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
15605         for BUILT_IN_C{LZ,LRSB}*.
15606         * tree.h (CASE_INT_FN): Add FN##IMAX case.
15607         * tree-vrp.c (extract_range_basic): Handle
15608         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
15609         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
15610         fall thru to code calling set_value*.
15611         * builtins.c (expand_builtin): Remove *IMAX cases.
15612         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
15613         if width is bigger than 2*HWI.
15615 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
15617         PR rtl-optimization/55342
15618         * lra-int.h (lra_subreg_reload_pseudos): New.
15619         * lra.c: Add undoing optional reloads to the block diagram.
15620         (lra_subreg_reload_pseudos): New.
15621         (lra_optional_reload_pseudos): Change comments.
15622         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
15623         lra_optional_reload_pseudos after undo transformations.
15624         * lra-assigns.c (pseudo_prefix_title): New.
15625         (lra_setup_reg_renumber): Use it.
15626         (spill_for): Ditto.  Check subreg reload pseudos too.
15627         (assign_by_spills): Consider subreg reload pseudos too.
15628         * lra-constraints.c (simplify_operand_subreg): Use
15629         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
15630         (curr_insn_transform): Recognize and do optional reloads.
15631         (undo_optional_reloads): New.
15632         (lra_undo_inheritance): Call undo_optional_reloads.
15634 2013-07-05  Thomas Quinot  <quinot@adacore.com>
15636         * tree-complex.c (expand_complex_operations_1): Fix typo.
15638 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
15640         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
15641         (tune_params): New member 'const vec_costs'.
15642         * config/aarch64/aarch64.c (generic_vector_cost): New.
15643         (generic_tunings): New member 'generic_vector_cost'.
15644         (aarch64_builtin_vectorization_cost): New.
15645         (aarch64_add_stmt_cost): New.
15646         (TARGET_VECTORIZE_ADD_STMT_COST): New.
15647         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
15649 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
15651         PR target/57777
15652         * config/i386/predicates.md (vsib_address_operand): Disallow
15653         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
15655 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
15657         PR middle-end/55030
15658         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
15659         expand_builtin_setjmp_receiver.
15660         (expand_label): Adjust, call expand_builtin_setjmp_receiver
15661         with NULL for the label parameter.
15662         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
15663         the frame-pointer.  Adjust comments.
15664         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
15665         only if LABEL is non-NULL.
15667 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
15669         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
15670         (ARM_ABI_AAPCS64): Ditto.
15671         (arm_abi): Ditto.
15672         (ARM_DEFAULT_ABI): Ditto.
15674 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
15676         * config/aarch64/aarch64-builtins.c
15677         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
15678         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
15679         (st1): Likewise.
15680         * config/aarch64/aarch64-simd.md
15681         (aarch64_ld1<VALL:mode>): New.
15682         (aarch64_st1<VALL:mode>): Likewise.
15683         * config/aarch64/arm_neon.h
15684         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
15686 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
15688         * config/i386/i386.c (gate_insert_vzeroupper): Check if
15689         target ISA is AVX.
15690         (ix86_option_override_internal):Turn on all -mavx target flags by
15691         default as they are dependent on AVX anyway.
15693 2013-07-02  Cary Coutant  <ccoutant@google.com>
15695         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
15696         deterministic hash.
15697         (loc_checksum_ordered): Likewise.
15698         (hash_loc_operands): Remove inline keyword.
15700 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
15702         PR tree-optimization/57741
15703         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
15704         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
15705         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
15706         Allow REAL_CST step_exprs if flag_associative_math.
15707         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
15709 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15711         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
15713 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15715         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
15717 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15719         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
15720         encoding.
15721         (iorsi3_insn): Likewise.
15722         (arm_xorsi3): Likewise.
15724 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
15726         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
15727         to ...
15728         (attribute "type"): ... here, and prefix with "wmmx_".
15729         (attribute "core_cycles"): Update for attribute changes.
15730         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
15731         (tbcstv4hi): Likewise.
15732         (tbcstv2si): Likewise.
15733         (iwmmxt_iordi3): Likewise.
15734         (iwmmxt_xordi3): Likewise.
15735         (iwmmxt_anddi3): Likewise.
15736         (iwmmxt_nanddi3): Likewise.
15737         (iwmmxt_arm_movdi): Likewise.
15738         (iwmmxt_movsi_insn): Likewise.
15739         (mov<mode>_internal): Likewise.
15740         (and<mode>3_iwmmxt): Likewise.
15741         (ior<mode>3_iwmmxt): Likewise.
15742         (xor<mode>3_iwmmxt): Likewise.
15743         (add<mode>3_iwmmxt): Likewise.
15744         (ssaddv8qi3): Likewise.
15745         (ssaddv4hi3): Likewise.
15746         (ssaddv2si3): Likewise.
15747         (usaddv8qi3): Likewise.
15748         (usaddv4hi3): Likewise.
15749         (usaddv2si3): Likewise.
15750         (sub<mode>3_iwmmxt): Likewise.
15751         (sssubv8qi3): Likewise.
15752         (sssubv4hi3): Likewise.
15753         (sssubv2si3): Likewise.
15754         (ussubv8qi3): Likewise.
15755         (ussubv4hi3): Likewise.
15756         (ussubv2si3): Likewise.
15757         (mulv4hi3_iwmmxt): Likewise.
15758         (smulv4hi3_highpart): Likewise.
15759         (umulv4hi3_highpart): Likewise.
15760         (iwmmxt_wmacs): Likewise.
15761         (iwmmxt_wmacsz): Likewise.
15762         (iwmmxt_wmacu): Likewise.
15763         (iwmmxt_wmacuz): Likewise.
15764         (iwmmxt_clrdi): Likewise.
15765         (iwmmxt_clrv8qi): Likewise.
15766         (iwmmxt_clr4hi): Likewise.
15767         (iwmmxt_clr2si): Likewise.
15768         (iwmmxt_uavgrndv8qi3): Likewise.
15769         (iwmmxt_uavgrndv4hi3): Likewise.
15770         (iwmmxt_uavgv8qi3): Likewise.
15771         (iwmmxt_uavgv4hi3): Likewise.
15772         (iwmmxt_tinsrb): Likewise.
15773         (iwmmxt_tinsrh): Likewise.
15774         (iwmmxt_tinsrw): Likewise.
15775         (iwmmxt_textrmub): Likewise.
15776         (iwmmxt_textrmsb): Likewise.
15777         (iwmmxt_textrmuh): Likewise.
15778         (iwmmxt_textrmsh): Likewise.
15779         (iwmmxt_textrmw): Likewise.
15780         (iwmxxt_wshufh): Likewise.
15781         (eqv8qi3): Likewise.
15782         (eqv4hi3): Likewise.
15783         (eqv2si3): Likewise.
15784         (gtuv8qi3): Likewise.
15785         (gtuv4hi3): Likewise.
15786         (gtuv2si3): Likewise.
15787         (gtv8qi3): Likewise.
15788         (gtv4hi3): Likewise.
15789         (gtv2si3): Likewise.
15790         (smax<mode>3_iwmmxt): Likewise.
15791         (umax<mode>3_iwmmxt): Likewise.
15792         (smin<mode>3_iwmmxt): Likewise.
15793         (umin<mode>3_iwmmxt): Likewise.
15794         (iwmmxt_wpackhss): Likewise.
15795         (iwmmxt_wpackwss): Likewise.
15796         (iwmmxt_wpackdss): Likewise.
15797         (iwmmxt_wpackhus): Likewise.
15798         (iwmmxt_wpackwus): Likewise.
15799         (iwmmxt_wpackdus): Likewise.
15800         (iwmmxt_wunpckihb): Likewise.
15801         (iwmmxt_wunpckihh): Likewise.
15802         (iwmmxt_wunpckihw): Likewise.
15803         (iwmmxt_wunpckilb): Likewise.
15804         (iwmmxt_wunpckilh): Likewise.
15805         (iwmmxt_wunpckilw): Likewise.
15806         (iwmmxt_wunpckehub): Likewise.
15807         (iwmmxt_wunpckehuh): Likewise.
15808         (iwmmxt_wunpckehuw): Likewise.
15809         (iwmmxt_wunpckehsb): Likewise.
15810         (iwmmxt_wunpckehsh): Likewise.
15811         (iwmmxt_wunpckehsw): Likewise.
15812         (iwmmxt_wunpckelub): Likewise.
15813         (iwmmxt_wunpckeluh): Likewise.
15814         (iwmmxt_wunpckeluw): Likewise.
15815         (iwmmxt_wunpckelsb): Likewise.
15816         (iwmmxt_wunpckelsh): Likewise.
15817         (iwmmxt_wunpckelsw): Likewise.
15818         (ror<mode>3): Likewise.
15819         (ashr<mode>3_iwmmxt): Likewise.
15820         (lshr<mode>3_iwmmxt): Likewise.
15821         (ashl<mode>3_iwmmxt): Likewise.
15822         (ror<mode>3_di): Likewise.
15823         (ashr<mode>3_di): Likewise.
15824         (lshr<mode>3_di): Likewise.
15825         (ashl<mode>3_di): Likewise.
15826         (iwmmxt_wmadds): Likewise.
15827         (iwmmxt_wmaddu): Likewise.
15828         (iwmmxt_tmia): Likewise.
15829         (iwmmxt_tmiaph): Likewise.
15830         (iwmmxt_tmiabb): Likewise.
15831         (iwmmxt_tmiatb): Likewise.
15832         (iwmmxt_tmiabt): Likewise.
15833         (iwmmxt_tmiatt): Likewise.
15834         (iwmmxt_tmovmskb): Likewise.
15835         (iwmmxt_tmovmskh): Likewise.
15836         (iwmmxt_tmovmskw): Likewise.
15837         (iwmmxt_waccb): Likewise.
15838         (iwmmxt_wacch): Likewise.
15839         (iwmmxt_waccw): Likewise.
15840         (iwmmxt_waligni): Likewise.
15841         (iwmmxt_walignr): Likewise.
15842         (iwmmxt_walignr0): Likewise.
15843         (iwmmxt_walignr1): Likewise.
15844         (iwmmxt_walignr2): Likewise.
15845         (iwmmxt_walignr3): Likewise.
15846         (iwmmxt_wsadb): Likewise.
15847         (iwmmxt_wsadh): Likewise.
15848         (iwmmxt_wsadbz): Likewise.
15849         (iwmmxt_wsadhz): Likewise.
15850         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
15851         (iwmmxt_wabsdiffb): Likewise.
15852         (iwmmxt_wabsdiffh): Likewise.
15853         (iwmmxt_wabsdiffw): Likewise.
15854         (iwmmxt_waddsubhx): Likewise
15855         (iwmmxt_wsubaddhx): Likewise.
15856         (addc<mode>3): Likewise.
15857         (iwmmxt_avg4): Likewise.
15858         (iwmmxt_avg4r): Likewise.
15859         (iwmmxt_wmaddsx): Likewise.
15860         (iwmmxt_wmaddux): Likewise.
15861         (iwmmxt_wmaddsn): Likewise.
15862         (iwmmxt_wmaddun): Likewise.
15863         (iwmmxt_wmulwsm): Likewise.
15864         (iwmmxt_wmulwum): Likewise.
15865         (iwmmxt_wmulsmr): Likewise.
15866         (iwmmxt_wmulumr): Likewise.
15867         (iwmmxt_wmulwsmr): Likewise.
15868         (iwmmxt_wmulwumr): Likewise.
15869         (iwmmxt_wmulwl): Likewise.
15870         (iwmmxt_wqmulm): Likewise.
15871         (iwmmxt_wqmulwm): Likewise.
15872         (iwmmxt_wqmulmr): Likewise.
15873         (iwmmxt_wqmulwmr): Likewise.
15874         (iwmmxt_waddbhusm): Likewise.
15875         (iwmmxt_waddbhusl): Likewise.
15876         (iwmmxt_wqmiabb): Likewise.
15877         (iwmmxt_wqmiabt): Likewise.
15878         (iwmmxt_wqmiatb): Likewise.
15879         (iwmmxt_wqmiatt): Likewise.
15880         (iwmmxt_wqmiabbn): Likewise.
15881         (iwmmxt_wqmiabtn): Likewise.
15882         (iwmmxt_wqmiatbn): Likewise.
15883         (iwmmxt_wqmiattn): Likewise.
15884         (iwmmxt_wmiabb): Likewise.
15885         (iwmmxt_wmiabt): Likewise.
15886         (iwmmxt_wmiatb): Likewise.
15887         (iwmmxt_wmiatt): Likewise.
15888         (iwmmxt_wmiabbn): Likewise.
15889         (iwmmxt_wmiabtn): Likewise.
15890         (iwmmxt_wmiatbn): Likewise.
15891         (iwmmxt_wmiattn): Likewise.
15892         (iwmmxt_wmiawbb): Likewise.
15893         (iwmmxt_wmiawbt): Likewise.
15894         (iwmmxt_wmiawtb): Likewise.
15895         (iwmmxt_wmiawtt): Likewise.
15896         (iwmmxt_wmiawbbn): Likewise.
15897         (iwmmxt_wmiawbtn): Likewise.
15898         (iwmmxt_wmiawtbn): Likewise.
15899         (iwmmxt_wmiawttn): Likewise.
15900         (iwmmxt_wmerge): Likewise.
15901         (iwmmxt_tandc<mode>3): Likewise.
15902         (iwmmxt_torc<mode>3): Likewise.
15903         (iwmmxt_torvsc<mode>3): Likewise.
15904         (iwmmxt_textrc<mode>3): Likewise.
15905         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
15906         (wmmxt_pack): Likewise.
15907         (wmmxt_mult_c1): Likewise.
15908         (wmmxt_mult_c2): Likewise.
15909         (wmmxt_alu_c1): Likewise.
15910         (wmmxt_alu_c2): Likewise.
15911         (wmmxt_alu_c3): Likewise.
15912         (wmmxt_transfer_c1): Likewise.
15913         (wmmxt_transfer_c2): Likewise.
15914         (wmmxt_transfer_c3): Likewise.
15915         (marvell_f_iwmmxt_wstr): Likewise.
15916         (marvell_f_iwmmxt_wldr): Likewise.
15918 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
15920         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
15922 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15924         Revert:
15925         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15926         * lra-constraints.c (need_for_split_p): Check call used hard regs
15927         living through calls.
15929         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
15930         call used regs for call insn.
15932 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
15934         PR target/57736
15935         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
15936         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
15938 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
15940         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
15942 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
15944         * lra-constraints.c (need_for_split_p): Check call used hard regs
15945         living through calls.
15947 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
15949         PR target/57744
15950         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
15951         to tie with any other modes.  Eliminate Altivec vector mode tests,
15952         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
15953         code, to return 0 if testing MODE2 for a condition, if we've
15954         already tested MODE1 for the same condition.
15956 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15958         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
15959         layout.
15961 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15963         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15964         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
15966 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15968         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
15969         Define.
15970         (aarch64_symbolic_constant_p): Remove.
15971         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
15972         static.  Fix line length and white space.
15973         (aarch64_symbolic_constant_p): Remove.
15974         * config/aarch64/predicates.md (aarch64_valid_symref):
15975         Use aarch64_classify_symbol_expression.
15977 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15979         * config/arm/constraints.md (Ts): New constraint.
15980         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
15981         16-bit encodings.
15982         (compare_scc): Use "Ts" constraint for operand 0.
15983         (ior_scc_scc): Likewise.
15984         (and_scc_scc): Likewise.
15985         (and_scc_scc_nodom): Likewise.
15986         (ior_scc_scc_cmp): Likewise for operand 7.
15987         (and_scc_scc_cmp): Likewise.
15988         * config/arm/thumb2.md (thumb2_movsi_insn):
15989         Add alternatives for 16-bit encodings.
15990         (thumb2_movhi_insn): Likewise.
15991         (thumb2_movsicc_insn): Likewise.
15992         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
15993         (thumb2_negscc): Use "Ts" constraint.
15994         Move mvn instruction outside cond_exec block.
15995         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
15996         for 16-bit encodings.
15998 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16000         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
16001         encoding.
16002         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
16003         (mulsi3subsi): Likewise.
16004         (mulsidi3adddi): Likewise.
16005         (mulsidi3_v6): Likewise.
16006         (umulsidi3_v6): Likewise.
16007         (umulsidi3adddi_v6): Likewise.
16008         (smulsi3_highpart_v6): Likewise.
16009         (umulsi3_highpart_v6): Likewise.
16010         (mulhisi3tb): Likewise.
16011         (mulhisi3bt): Likewise.
16012         (mulhisi3tt): Likewise.
16013         (maddhisi4): Likewise.
16014         (maddhisi4tb): Likewise.
16015         (maddhisi4tt): Likewise.
16016         (maddhidi4): Likewise.
16017         (maddhidi4tb): Likewise.
16018         (maddhidi4tt): Likewise.
16019         (zeroextractsi_compare0_scratch): Likewise.
16020         (insv_zero): Likewise.
16021         (insv_t2): Likewise.
16022         (anddi_notzesidi_di): Likewise.
16023         (anddi_notsesidi_di): Likewise.
16024         (andsi_notsi_si): Likewise.
16025         (iordi_zesidi_di): Likewise.
16026         (xordi_zesidi_di): Likewise.
16027         (andsi_iorsi3_notsi): Likewise.
16028         (smax_0): Likewise.
16029         (smax_m1): Likewise.
16030         (smin_0): Likewise.
16031         (not_shiftsi): Likewise.
16032         (unaligned_loadsi): Likewise.
16033         (unaligned_loadhis): Likewise.
16034         (unaligned_loadhiu): Likewise.
16035         (unaligned_storesi): Likewise.
16036         (unaligned_storehi): Likewise.
16037         (extv_reg): Likewise.
16038         (extzv_t2): Likewise.
16039         (divsi3): Likewise.
16040         (udivsi3): Likewise.
16041         (arm_zero_extendhisi2addsi): Likewise.
16042         (arm_zero_extendqisi2addsi): Likewise.
16043         (compareqi_eq0): Likewise.
16044         (arm_extendhisi2_v6): Likewise.
16045         (arm_extendqisi2addsi): Likewise.
16046         (arm_movt): Likewise.
16047         (thumb2_ldrd): Likewise.
16048         (thumb2_ldrd_base): Likewise.
16049         (thumb2_ldrd_base_neg): Likewise.
16050         (thumb2_strd): Likewise.
16051         (thumb2_strd_base): Likewise.
16052         (thumb2_strd_base_neg): Likewise.
16053         (arm_negsi2): Add alternative for 16-bit encoding.
16054         (arm_one_cmplsi2): Likewise.
16056 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16058         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
16059         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
16060         (movdfcc): Likewise.
16061         * config/arm/vfp.md (*thumb2_movsf_vfp):
16062         Disable predication for arm_restrict_it.
16063         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
16064         (*thumb2_movdfcc_vfp): Likewise.
16065         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
16066         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
16067         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
16068         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
16069         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
16070         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
16071         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
16072         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
16073         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
16074         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
16075         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
16076         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
16077         Disable predication for arm_restrict_it.
16079 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
16081         * config/i386/bmiintrin.h (_bextr_u32): New.
16082         (_bextr_u64): Ditto.
16084 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
16086         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
16087         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
16088         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
16089         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
16090         * config/mips/n32-elf.h: ...this new file.
16092 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
16094         PR target/57224
16095         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
16096         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
16098 2013-06-27  Catherine Moore  <clm@codesourcery.com>
16100         * config/mips/mips-tables.opt: Regenerate.
16101         * config/mips/mips-cpus.def: Add m14ke and m14kec.
16102         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
16103         * doc/invoke.texi: Add -m14kc.
16105 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16107         PR target/57623
16108         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
16109         constraints of operand 1 and 2.
16111         PR target/57623
16112         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
16113         to match RTL canonicalization.  Swap predicates and
16114         constraints of operand 1 and 2.
16116 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
16118         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
16119         Process OP_INOUT regs for splitting too.
16121 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16123         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
16124         decl before the loop, initialize to NULL.
16125         (vectorizable_load): Initialize ptr_incr to NULL.
16127 2013-06-27  Martin Jambor  <mjambor@suse.cz>
16129         PR lto/57208
16130         * ipa-ref.h (ipa_maybe_record_reference): Declare.
16131         * ipa-ref.c (ipa_maybe_record_reference): New function.
16132         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
16133         * ipa-cp.c (create_specialized_node): Record potential references from
16134         aggvals.
16135         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
16137 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16139         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
16140         parameter 'mode' of type 'enum machine_mode mode'; change to pass
16141         'mode' to force_reg.
16142         (aarch64_add_offset): Update calls to aarch64_force_temporary.
16143         (aarch64_expand_mov_immediate): Likewise.
16145 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16147         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
16148         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
16150 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16152         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
16153         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16154         (struct machine_function): Add tbegin_p.
16155         (s390_canonicalize_comparison): Fold CC mode compares to
16156         conditional jump if possible.
16157         (s390_emit_jump): Return the emitted jump.
16158         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
16159         Handle CCRAWmode compares.
16160         (s390_option_override): Default to -mhtm if available.
16161         (s390_reg_clobbered_rtx): Handle floating point regs as well.
16162         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
16163         FPRs instead of df_regs_ever_live_p.
16164         (s390_optimize_nonescaping_tx): New function.
16165         (s390_init_frame_layout): Extend clobbered_regs array to cover
16166         FPRs as well.
16167         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
16168         (s390_expand_tbegin): New function.
16169         (enum s390_builtin): New enum definition.
16170         (code_for_builtin): New array definition.
16171         (s390_init_builtins): New function.
16172         (s390_expand_builtin): New function.
16173         (TARGET_INIT_BUILTINS): Define.
16174         (TARGET_EXPAND_BUILTIN): Define.
16175         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
16176         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
16177         (s390_alc_comparison): Likewise.
16178         * config/s390/s390-modes.def: Add CCRAWmode.
16179         * config/s390/s390.h (processor_flags): Add PF_TX.
16180         (TARGET_CPU_HTM): Define macro.
16181         (TARGET_HTM): Define macro.
16182         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
16183         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
16184         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16185         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
16186         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
16187         (TBEGIN_MASK, TBEGINC_MASK): New constants.
16188         ("*cc_to_int"): Move up.
16189         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
16190         constants other than 0.
16191         ("*ccraw_to_int"): New insn and splitter definition.
16192         ("tbegin", "tbegin_nofloat", "tbegin_retry")
16193         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
16194         ("tx_assist"): New expander.
16195         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
16196         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
16197         * config/s390/s390.opt: Add -mhtm option.
16198         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
16199         * config/s390/htmxlintrin.h: New file.
16200         * config/s390/htmintrin.h: New file.
16201         * config/s390/s390intrin.h: New file.
16202         * doc/extend.texi: Document htm builtins.
16203         * config.gcc: Add the new header files to extra_headers.
16205 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
16207         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
16208         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
16210 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
16211             Pat Haugen  <pthaugen@us.ibm.com>
16212             Peter Bergner  <bergner@vnet.ibm.com>
16214         * config/rs6000/power8.md: New.
16215         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
16216         setting for power8 entry.
16217         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
16218         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
16219         test for Power4/Power5 only.
16220         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
16221         support.
16222         (force_new_group): Adjust comment.
16223         * config/rs6000/rs6000.md: Include power8.md.
16225 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
16227         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
16228         * config/arm/arm-protos.h (arm_max_conditional_execute): New
16229         declaration.
16230         (tune_params): Update comment.
16231         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
16232         (arm_max_conditional_execute): New function.
16233         (thumb2_final_prescan_insn): Use max_insn_skipped and
16234         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
16236 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
16238         PR tree-optimization/57705
16239         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
16240         SSA_NAME step, provided that it is not defined inside the loop.
16241         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
16242         (get_initial_def_for_induction): Handle SSA_NAME IV step.
16244 2013-06-25  Martin Jambor  <mjambor@suse.cz>
16246         PR middle-end/57670
16247         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
16248         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
16249         calls in the dump.
16250         (ipa_note_param_call): Initialize member_ptr flag.
16251         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
16252         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
16253         (ipa_write_indirect_edge_info): Stream member_ptr flag.
16254         (ipa_read_indirect_edge_info): Likewise.
16256 2013-06-25  Richard Biener  <rguenther@suse.de>
16258         PR middle-end/56977
16259         * passes.c (init_optimization_passes): Move pass_fold_builtins
16260         and pass_dce earlier with -Og.
16262 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
16264         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
16265         <BIT_FIELD_REF>: Remove trailing TAB.
16266         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
16267         remove blank line.
16269 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16271         PR tree-optimization/57358
16272         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
16273         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
16274         (ipa_analyze_params_uses): Generate pessimistic info when true.
16276 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16278         PR tree-optimization/57539
16279         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
16280         global.inlined_to of the new node to it.  All callers changed.
16281         * ipa-inline-transform.c (clone_inlined_nodes): New variable
16282         inlining_into, pass it to cgraph_clone_node.
16283         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
16284         ipa_free_edge_args_substructures.
16285         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
16286         rdesc linked list.  Do not assert rdesc edges have inlined caller.
16287         Assert we have found an rdesc in the rdesc list.
16289 2013-06-24  Richard Biener  <rguenther@suse.de>
16291         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
16292         (pointer_set_lookup): Declare.
16293         (class pointer_map): New template class implementing a
16294         generic pointer to T map.
16295         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
16296         pointer_map<T>::contains, pointer_map<T>::insert,
16297         pointer_map<T>::traverse): New functions.
16298         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
16299         (pointer_set_lookup): New function.
16300         (pointer_set_contains): Use pointer_set_lookup.
16301         (pointer_set_insert): Likewise.
16302         (insert_aux): Remove.
16303         (struct pointer_map_t): Embed a pointer_set_t.
16304         (pointer_map_create): Adjust.
16305         (pointer_map_destroy): Likewise.
16306         (pointer_map_contains): Likewise.
16307         (pointer_map_insert): Likewise.
16308         (pointer_map_traverse): Likewise.
16309         * tree-streamer.h (struct streamer_tree_cache_d): Use a
16310         pointer_map<unsigned> instead of a pointer_map_t.
16311         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
16312         (streamer_tree_cache_lookup): Likewise.
16313         (streamer_tree_cache_create): Likewise.
16314         (streamer_tree_cache_delete): Likewise.
16315         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
16316         pointer_map<unsigned> instead of a pointer_map_t.
16317         (lto_init_tree_ref_encoder): Adjust.
16318         (lto_destroy_tree_ref_encoder): Likewise.
16319         * lto-section-out.c (lto_output_decl_index): Likewise.
16320         (lto_record_function_out_decl_state): Likewise.
16321         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
16323 2013-06-24  Richard Biener  <rguenther@suse.de>
16325         PR tree-optimization/57488
16326         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
16328 2013-06-24  Alan Modra  <amodra@gmail.com>
16330         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
16331         (gen_easy_altivec_constant): Likewise.
16332         * config/rs6000/predicates.md (easy_vector_constant_add_self,
16333         easy_vector_constant_msb): Likewise.
16335 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
16337         PR target/57688
16338         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
16339         add missing return true.
16341 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
16343         PR target/52483
16344         * config/sh/predicates.md (general_extend_operand): Invoke
16345         general_movsrc_operand for memory operands.
16346         (general_movsrc_operand): Allow reg+reg addressing, do not use
16347         general_operand for memory operands.
16349 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
16351         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
16352         when current target options does not apply.
16353         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
16354         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
16355         * config/i386/bmiintrin.h: Pass appropriate target
16356         attributes to header.
16357         * config/i386/mmintrin.h: Ditto.
16358         * config/i386/nmmintrin.h: Ditto.
16359         * config/i386/avx2intrin.h: Ditto.
16360         * config/i386/fxsrintrin.h: Ditto.
16361         * config/i386/tbmintrin.h: Ditto.
16362         * config/i386/xsaveintrin.h: Ditto.
16363         * config/i386/f16cintrin.h: Ditto.
16364         * config/i386/xtestintrin.h: Ditto.
16365         * config/i386/xsaveoptintrin.h: Ditto.
16366         * config/i386/bmi2intrin.h: Ditto.
16367         * config/i386/lzcntintrin.h: Ditto.
16368         * config/i386/smmintrin.h: Ditto.
16369         * config/i386/wmmintrin.h: Ditto.
16370         * config/i386/x86intrin.h: Remove all header include guards.
16371         * config/i386/prfchwintrin.h: Ditto.
16372         * config/i386/pmmintrin.h: Ditto.
16373         * config/i386/tmmintrin.h: Ditto.
16374         * config/i386/xmmintrin.h: Ditto.
16375         * config/i386/popcntintrin.h: Ditto.
16376         * config/i386/rdseedintrin.h: Ditto.
16377         * config/i386/ammintrin.h: Ditto.
16378         * config/i386/emmintrin.h: Ditto.
16379         * config/i386/immintrin.h: Remove all header include guards.
16380         * config/i386/fma4intrin.h: Ditto.
16381         * config/i386/lwpintrin.h: Ditto.
16382         * config/i386/xopintrin.h: Ditto.
16383         * config/i386/ia32intrin.h: Ditto.
16384         * config/i386/avxintrin.h: Ditto.
16385         * config/i386/rtmintrin.h: Ditto.
16386         * config/i386/fmaintrin.h: Ditto.
16387         * config/i386/mm3dnow.h: Ditto.
16389 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
16391         * common/config/i386/i386-common.c: Handle LZCNT.
16393 2013-06-22  Andi Kleen  <ak@linux.intel.com>
16395         * doc/extend.texi: Use __atomic_store_n instead of
16396         __atomic_store in HLE example.
16398 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
16400         * config/sh/sh.c: Remove <cstdlib> workaround.
16402 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16404         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
16406 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16408         * doc/extend.texi: Document that __atomic_clear and
16409           __atomic_test_and_set should only be used with bool.
16411 2013-06-20  Jan Hubicka  <jh@suse.cz>
16413         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
16414         types_same_for_odr.
16415         * tree.c (decls_same_for_odr): New function.
16416         (same_for_edr): New function.
16417         (types_same_for_odr): New function.
16418         (get_binfo_at_offset): Use it.
16419         * tree.h (types_same_for_odr): Declare.
16421 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
16422             Jason Merrill  <jason@redhat.com>
16424         * system.h: Include <cstdlib> as well as <stdlib.h>.
16426 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
16428         PR target/57655
16429         * config/i386/i386.c (construct_container): Report error if
16430         long double is used with disabled x87 float returns.
16432 2013-06-20  Jan Hubicka  <jh@suse.cz>
16434         * lto-cgraph.c (input_symtab): Do not set cgraph state.
16436 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
16438         PR rtl-optimization/57425
16439         PR rtl-optimization/57569
16440         * alias.c (write_dependence_p): Remove parameters mem_mode and
16441         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
16442         Changed all callers.
16443         (canon_anti_dependence): Get comments and semantics in sync.
16444         Add parameter mem_canonicalized.  Changed all callers.
16445         * rtl.h (canon_anti_dependence): Update prototype.
16447 2013-06-20  Richard Biener  <rguenther@suse.de>
16449         * data-streamer-in.c (streamer_read_uhwi): Optimize single
16450         byte case, inline streamer_read_uchar and defer section
16451         overrun check.
16453 2013-06-20  Richard Biener  <rguenther@suse.de>
16455         PR tree-optimization/57584
16456         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
16457         SSA names into the expanded expression that take part in
16458         abnormal coalescing.
16460 2013-06-19  Sharad Singhai  <singhai@google.com>
16462         * gcov.c (print_usage): Handle new option.
16463         (process_args): Ditto.
16464         (get_gcov_intermediate_filename): New function.
16465         (output_intermediate_file): New function.
16466         (output_gcov_file): New function
16467         (generate_results): Handle new option.
16468         (release_function): Relase demangled name.
16469         (read_graph_file): Handle demangled name.
16470         (output_lines): Ditto.
16471         * doc/gcov.texi: Document gcov intermediate format.
16473 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
16475         PR bootstrap/57604
16476         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
16477         (lra_emit_add): Use the functions.  Add comment about Y as an
16478         address segment.
16480 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
16482         PR driver/57652
16483         * collect2.c (collect_atexit): New.
16484         (collect_exit): Delete.
16485         (main): Register collect_atexit with atexit.
16486         (collect_wait): Change collect_exit to exit.
16487         (do_wait): Same.
16488         * collect2.h (collect_exit): Delete.
16489         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
16491 2013-06-19  Wei Mi  <wmi@google.com>
16493         PR rtl-optimization/57518
16494         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
16495         if regno is used in paradoxical subreg.
16496         (update_equiv_regs): Check pdx_subregs[regno] before
16497         set a reg to be equivalent with a mem.
16499 2013-06-19  Matthias Klose  <doko@ubuntu.com>
16501         PR driver/57651
16502         * file-find.h (find_a_file): Add a mode parameter.
16503         * file-find.c (find_a_file): Likewise.
16504         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
16505         with X_OK for the executables.
16506         * collect2.c (main): Call find_a_file with X_OK.
16508 2013-06-19  Steve Ellcey  <sellcey@mips.com>
16510         PR target/56942
16511         * config/mips/mips.md (casesi_internal_mips16_<mode>):
16512         Use NEXT_INSN instead of next_real_insn.
16514 2013-06-19  Jan Hubicka  <jh@suse.cz>
16516         * cgraph.h (const_value_known_p): Replace by ...
16517         (ctor_for_folding): .. this one.
16518         * cgraphunit.c (process_function_and_variable_attributes): Use it.
16519         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
16520         * expr.c (expand_expr_real_1): Likewise.
16521         (string_constant): Likewise.
16522         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
16523         * ipa.c (process_references): Likewise.
16524         (symtab_remove_unreachable_nodes): Likewise.
16525         * ipa-inline-analysis.c (param_change_prob): Likewise.
16526         * gimple-fold.c (canonicalize_constructor_val): Likewise.
16527         (get_base_constructor): Likwise.
16528         * varpool.c (varpool_remove_node): Likewise.
16529         (varpool_remove_initializer): LIkewise.
16530         (dump_varpool_node): LIkwise.
16531         (const_value_known_p): Rewrite to ...
16532         (ctor_for_folding): ... this one.
16534 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
16536         PR driver/57651
16537         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
16538         PERSONALITY in $PATH derived prefixes.
16540 2013-06-19  Jeff Law  <law@redhat.com>
16542         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
16543         in comment.
16545         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
16546         (simplify_bitwise_binary): Use it to simpify certain binary ops on
16547         booleans.
16549 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
16551         * config/arm/vfp.md: Move VFP instruction classification documentation
16552         to ...
16553         * config/arm/arm.md: ... here.  Update instruction classification
16554         documentation.
16556 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
16558         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
16559         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
16560         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
16562 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
16564         PR target/57609
16565         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16566         with NEXT_INSN.  Use tablejump_p to check for jump table data
16567         insns.
16569 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
16571         PR c++/56544
16572         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
16573         that now in C++ the value is correct per the C++ standards.
16575 2013-06-19  Richard Biener  <rguenther@suse.de>
16577         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
16578         for global context.
16580 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16582         Revert:
16583         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16585         PR target/57609
16586         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16587         with next_active_insn.
16589 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
16591         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
16592         functions are inlined during failures to flag an error.
16593         * tree-inline.c (expand_call_inline): Allow the error to be flagged
16594         in early inline pass.
16596 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
16598         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
16599         in comments.
16601 2013-06-18  Julian Brown  <julian@codesourcery.com>
16603         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
16604         Permit virtual register pre-reload if !strict.
16605         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
16606         change.
16607         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
16608         prototype.
16609         * config/arm/neon.md (movmisalign<mode>): Use
16610         neon_perm_struct_or_reg_operand instead of
16611         neon_struct_or_register_operand.
16612         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
16613         neon_permissive_struct_operand instead of neon_struct_operand.
16614         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
16615         neon_vector_mem_operand.
16616         * config/arm/predicates.md (neon_struct_operand): Adjust call to
16617         neon_vector_mem_operand.
16618         (neon_permissive_struct_operand): New.
16619         (neon_struct_or_register_operand): Rename to...
16620         (neon_perm_struct_or_reg_operand): This. Adjust call to
16621         neon_vector_mem_operand.
16623 2013-06-18  Richard Biener  <rguenther@suse.de>
16625         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
16626         * lto-streamer.h: Include pointer-set.h.
16627         (struct lto_decl_slot): Remove.
16628         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
16629         Remove next_index entry.
16630         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16631         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16632         (lto_init_tree_ref_encoder): Adjust.
16633         (lto_destroy_tree_ref_encoder): Likewise.
16634         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16635         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16636         (lto_output_decl_index): Adjust.
16637         (lto_new_out_decl_state): Likewise.
16638         (lto_record_function_out_decl_state): Likewise.
16639         * lto-streamer-out.c (copy_function): Likewise.
16641 2013-06-18  Richard Biener  <rguenther@suse.de>
16643         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
16644         * cgraphunit.c: Include cfgloop.h.
16645         (init_lowered_empty_function): Initialize the loop tree.
16646         (assemble_thunk): Insert new BBs into loops.
16648 2013-06-18  Richard Biener  <rguenther@suse.de>
16650         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
16651         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
16652         the map from cache entry to cache index optional.
16653         (streamer_tree_cache_replace_tree): Adjust accordingly.
16654         (streamer_tree_cache_append): Likewise.
16655         (streamer_tree_cache_delete): Likewise.
16656         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
16657         streamer cache map from cache entry to cache index.
16658         * lto-streamer-out.c (create_output_block): Adjust.
16660 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
16662         * config/arm/arm.md (attribute "insn"): Move multiplication and
16663         division attributes to...
16664         (attribute "type"): ... here.  Remove mult.
16665         (attribute "mul32"): New attribute.
16666         (attribute "mul64"): Add umaal.
16667         (*arm_mulsi3): Update attributes.
16668         (*arm_mulsi3_v6): Likewise.
16669         (*thumb_mulsi3): Likewise.
16670         (*thumb_mulsi3_v6): Likewise.
16671         (*mulsi3_compare0): Likewise.
16672         (*mulsi3_compare0_v6): Likewise.
16673         (*mulsi_compare0_scratch): Likewise.
16674         (*mulsi_compare0_scratch_v6): Likewise.
16675         (*mulsi3addsi): Likewise.
16676         (*mulsi3addsi_v6): Likewise.
16677         (*mulsi3addsi_compare0): Likewise.
16678         (*mulsi3addsi_compare0_v6): Likewise.
16679         (*mulsi3addsi_compare0_scratch): Likewise.
16680         (*mulsi3addsi_compare0_scratch_v6): Likewise.
16681         (*mulsi3subsi): Likewise.
16682         (*mulsidi3adddi): Likewise.
16683         (*mulsi3addsi_v6): Likewise.
16684         (*mulsidi3adddi_v6): Likewise.
16685         (*mulsidi3_nov6): Likewise.
16686         (*mulsidi3_v6): Likewise.
16687         (*umulsidi3_nov6): Likewise.
16688         (*umulsidi3_v6): Likewise.
16689         (*umulsidi3adddi): Likewise.
16690         (*umulsidi3adddi_v6): Likewise.
16691         (*smulsi3_highpart_nov6): Likewise.
16692         (*smulsi3_highpart_v6): Likewise.
16693         (*umulsi3_highpart_nov6): Likewise.
16694         (*umulsi3_highpart_v6): Likewise.
16695         (mulhisi3): Likewise.
16696         (*mulhisi3tb): Likewise.
16697         (*mulhisi3bt): Likewise.
16698         (*mulhisi3tt): Likewise.
16699         (maddhisi4): Likewise.
16700         (*maddhisi4tb): Likewise.
16701         (*maddhisi4tt): Likewise.
16702         (maddhidi4): Likewise.
16703         (*maddhidi4tb): Likewise.
16704         (*maddhidi4tt): Likewise.
16705         (divsi3): Likewise.
16706         (udivsi3): Likewise.
16707         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
16708         (thumb2_mulsi_short_compare0): Likewise.
16709         (thumb2_mulsi_short_compare0_scratch): Likewise.
16710         * config/arm/arm1020e.md (1020mult1): Update attribute change.
16711         (1020mult2): Likewise.
16712         (1020mult3): Likewise.
16713         (1020mult4): Likewise.
16714         (1020mult5): Likewise.
16715         (1020mult6): Likewise.
16716         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
16717         change.
16718         (cortex_a15_mult64): Likewise.
16719         (cortex_a15_sdiv): Likewise.
16720         (cortex_a15_udiv): Likewise.
16721         * config/arm/arm1026ejs.md (mult1): Update attribute change.
16722         (mult2): Likewise.
16723         (mult3): Likewise.
16724         (mult4): Likewise.
16725         (mult5): Likewise.
16726         (mult6): Likewise.
16727         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
16728         (pj4_ir_div): Likewise.
16729         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
16730         (11_mult2): Likewise.
16731         (11_mult3): Likewise.
16732         (11_mult4): Likewise.
16733         (11_mult5): Likewise.
16734         (11_mult6): Likewise.
16735         (11_mult7): Likewise.
16736         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
16737         (cortex_a8_mla): Likewise.
16738         (cortex_a8_mull): Likewise.
16739         (cortex_a8_smulwy): Likewise.
16740         (cortex_a8_smlald): Likewise.
16741         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
16742         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
16743         (cortex_r4_mul_3): Likewise.
16744         (cortex_r4_mla_4): Likewise.
16745         (cortex_r4_mla_3): Likewise.
16746         (cortex_r4_smlald): Likewise.
16747         (cortex_r4_mull): Likewise.
16748         (cortex_r4_sdiv): Likewise.
16749         (cortex_r4_udiv): Likewise.
16750         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
16751         (cortex_a7_idiv): Likewise.
16752         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
16753         (9_mult2): Likewise.
16754         (9_mult3): Likewise.
16755         (9_mult4): Likewise.
16756         (9_mult5): Likewise.
16757         (9_mult6): Likewise.
16758         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
16759         (cortex_a53_sdiv): Likewise.
16760         (cortex_a53_udiv): Likewise.
16761         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
16762         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
16763         (mp626_mult2): Likewise.
16764         (mp626_mult3): Likewise.
16765         (mp626_mult4): Likewise.
16766         * config/arm/fa526.md (526_mult1): Update attribute change.
16767         (526_mult2): Likewise.
16768         * config/arm/arm-generic.md (mult): Update attribute change.
16769         (mult_ldsched_strongarm): Likewise.
16770         (mult_ldsched): Likewise.
16771         (multi_cycle): Likewise.
16772         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
16773         * config/arm/fa606te.md (606te_mult1): Update attribute change.
16774         (606te_mult2): Likewise.
16775         (606te_mult3): Likewise.
16776         (606te_mult4): Likewise.
16777         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
16778         (cortex_a9_mac16): Likewise.
16779         (cortex_a9_multiply): Likewise.
16780         (cortex_a9_mac): Likewise.
16781         (cortex_a9_multiply_long): Likewise.
16782         * config/arm/fa626te.md (626te_mult1): Update attribute change.
16783         (626te_mult2): Likewise.
16784         (626te_mult3): Likewise.
16785         (626te_mult4): Likewise.
16787 2013-06-18  Richard Biener  <rguenther@suse.de>
16789         PR lto/57334
16790         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
16792 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16794         PR target/57609
16795         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16796         with next_active_insn.
16798 2013-06-18  Alan Modra  <amodra@gmail.com>
16800         * config/rs6000/rs6000.h (enum data_align): New.
16801         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
16802         (DATA_ABI_ALIGNMENT): Define.
16803         (CONSTANT_ALIGNMENT): Correct comment.
16804         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
16805         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
16807 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
16809         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
16810         ATTRIBUTE_UNUSED marking.
16812 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
16814         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
16815         alternative and update.
16816         (aarch64_dup_lanedi): Delete.
16817         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
16818         * config/aarch64/aarch64-simd-builtins.def: Update.
16820 2013-06-17  Richard Biener  <rguenther@suse.de>
16822         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
16823         (lto_input_scc): Declare.
16824         (lto_input_tree_1): Likewise.
16825         (struct lto_stats_d): Add num_tree_bodies_output and
16826         num_pickle_refs_output.
16827         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
16828         (lto_read_tree_1): Split out from ...
16829         (lto_read_tree): ... this.
16830         (lto_input_scc): New function.
16831         (lto_input_tree_1): Split out from ...
16832         (lto_input_tree): ... this.  Handle LTO_tree_scc.
16833         (lto_data_in_create): Create the streamer cache without hashes.
16834         * lto-streamer-out.c (create_output_block): Create the streamer
16835         cache with hashes when not doing WPA.
16836         (lto_write_tree_1): Split out from ...
16837         (lto_write_tree): ... this.
16838         (get_symbol_initial_value): New function.
16839         (lto_output_tree_1): Split out from ...
16840         (lto_output_tree): ... this.  Write trees as series of SCCs
16841         using a DFS walk via DFS_write_tree.
16842         (struct sccs, struct scc_entry): New types.
16843         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
16844         (DFS_write_tree_body): New function.
16845         (DFS_write_tree): Likewise.
16846         (hash_tree): Likewise.
16847         (scc_entry_compare): Likewise.
16848         (hash_scc): Likewise.
16849         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
16850         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
16851         TREE_CHAIN as regular reference.
16852         (streamer_read_integer_cst): Remove.
16853         (streamer_get_pickled_tree): Adjust.
16854         * tree-streamer-out.c (streamer_write_chain): Disable streaming
16855         of DECL_EXTERNALs in BLOCK_VARS for now.
16856         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
16857         reference.
16858         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
16859         Add hash value argument and record that if hashes are recorded
16860         in the cache.
16861         (streamer_tree_cache_insert_1): Adjust.
16862         (streamer_tree_cache_insert): Likewise.
16863         (streamer_tree_cache_insert_at): Rename to ...
16864         (streamer_tree_cache_replace_tree): ... this and adjust.
16865         (streamer_tree_cache_append): Adjust.
16866         (record_common_node): Likewise.
16867         (streamer_tree_cache_create): Add argument whether to
16868         record hash values together with trees.
16869         (streamer_tree_cache_delete): Adjust.
16870         * tree-streamer.h (struct streamer_tree_cache_d): Add
16871         vector of hashes.
16872         (streamer_read_integer_cst): Remove.
16873         (streamer_tree_cache_insert): Adjust.
16874         (streamer_tree_cache_append): Likewise.
16875         (streamer_tree_cache_insert_at): Rename to ...
16876         (streamer_tree_cache_replace_tree): ... this and adjust.
16877         (streamer_tree_cache_create): Add argument whether to record hashes.
16878         (streamer_tree_cache_get): Rename to ...
16879         (streamer_tree_cache_get_tree): ... this.
16880         (streamer_tree_cache_get_hash): New function.
16881         * tree.c (cache_integer_cst): New function.
16882         * tree.h (cache_integer_cst): Declare.
16883         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
16884         * lto-symtab.c (lto_varpool_replace_node): Only release
16885         DECL_INITIAL of non-prevailing decls.
16886         * varpool.c (varpool_remove_initializer): Do not release
16887         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
16889 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
16891         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
16892         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
16893         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
16894         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
16895         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
16896         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
16897         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
16898         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
16899         instead of TARGET_64BIT.
16900         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
16901         Require ISA_HAS_<D>DIV.
16903 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
16905         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
16906         (mips*-*-linux*): Move default with_llsc setting to where other
16907         defaults are set.
16908         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
16909         with_arch block.
16910         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
16911         Likewise.  Remove default with_tune setting.  Move default float
16912         setting to its own block.  Handle with_llsc in the same block as above.
16914 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
16916         PR rtl-optimization/57425
16917         PR rtl-optimization/57569
16918         * alias.c (write_dependence_p): Add new parameters mem_mode,
16919         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
16920         Changed all callers.
16921         (canon_anti_dependence): New function.
16922         * cse.c (check_dependence): Use canon_anti_dependence.
16923         * cselib.c (cselib_invalidate_mem): Likewise.
16924         * rtl.h (canon_anti_dependence): Declare.
16926 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
16928         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
16929         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
16930         ".set mips3" for 64-bit targets.
16932 2013-06-15  Dehao Chen  <dehao@google.com>
16934         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
16935         * gimple-low.c (gimple_check_call_matching_types): Likewise.
16936         (gimple_check_call_args): Likewise.
16937         * value-prof.c (check_ic_target): Likewise.
16938         * ipa-inline.c (early_inliner): Likewise.
16939         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
16940         * cgraph.c (cgraph_create_edge_1): Likewise.
16941         (cgraph_make_edge_direct): Likewise.
16943 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
16945         PR target/57615
16946         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
16947         rs6000_output_move_128bit to handle emitting quad memory
16948         operations.  Set attribute length to 8 bytes.
16950 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
16952         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
16953         New pattern.
16954         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
16955         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
16956         (aarch64_<su>mlsl<mode>): Likewise.
16958 2013-06-14  Mike Stump  <mikestump@comcast.net>
16960         * Makefile.in (TARGET_H): Add insn-codes.h.
16962 2013-06-14  Alan Modra  <amodra@gmail.com>
16964         PR middle-end/57134
16965         PR middle-end/57586
16966         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
16967         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
16968         bitfield expansion when EXPAND_MEMORY.
16969         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
16971 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
16973         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
16974         test for clearing quad memory on 32-bit later.
16976 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
16978         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
16979         (fold_negate_expr): Likewise.
16980         (fold_real_zero_addition_p): Handle vectors.
16981         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
16983 2013-06-14  Alan Modra  <amodra@gmail.com>
16985         * varasm.c (force_const_mem): Revert 2013-06-07 change.
16987 2013-06-13  Jan Hubicka  <jh@suse.cz>
16989         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
16990         Local comdats are not externally visible.
16991         * symtab.c (dump_symtab_base): Dump externally visible.
16992         (verify_symtab_base): Verify back links in the symtab hash.
16994 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
16996         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
16997         CONVERT_EXPR as equal nodes.
16999 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17001         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
17003 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17005         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
17006         Generalize to complex and vector.
17007         * tree.c (build_all_ones_cst): New function.
17008         * tree.h (build_all_ones_cst): Declare it.
17010 2013-06-13  Alan Modra  <amodra@gmail.com>
17012         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
17013         * config/rs6000/rs6000.md (signbittf2): New insn.
17014         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
17015         (abstf2_internal, cmptf_internal2): Likewise.
17016         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
17018 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
17019             Pat Haugen  <pthaugen@us.ibm.com>
17020             Peter Bergner  <bergner@vnet.ibm.com>
17022         * config/rs6000/rs6000.c (emit_load_locked): Add support for
17023         power8 byte, half-word, and quad-word atomic instructions.
17024         (emit_store_conditional): Likewise.
17025         (rs6000_expand_atomic_compare_and_swap): Likewise.
17026         (rs6000_expand_atomic_op): Likewise.
17028         * config/rs6000/sync.md (larx): Add new modes for power8.
17029         (stcx): Likewise.
17030         (AINT): New mode iterator to include TImode as well as normal
17031         integer modes on power8.
17032         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
17033         that VSX registers are not considered.  Use AINT mode iterator
17034         instead of INT1 to allow inclusion of quad word atomic operations
17035         on power8.
17036         (load_locked<mode>): Likewise.
17037         (store_conditional<mode>): Likewise.
17038         (atomic_compare_and_swap<mode>): Likewise.
17039         (atomic_exchange<mode>): Likewise.
17040         (atomic_nand<mode>): Likewise.
17041         (atomic_fetch_<fetchop_name><mode>): Likewise.
17042         (atomic_nand_fetch<mode>): Likewise.
17043         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
17044         each type.
17045         (ATOMIC): On power8, add QImode, HImode modes.
17046         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
17047         modes that promote to SImode.
17048         (load_lockedti): Convert TImode arguments to PTImode, so that we
17049         get a guaranteed even/odd register pair.
17050         (load_lockedpti): Likewise.
17051         (store_conditionalti): Likewise.
17052         (store_conditionalpti): Likewise.
17054         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
17055         atomic load/store instructions.
17056         (HSI): Likewise.
17058 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
17060         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
17061         loads.
17062         (insn_count): New attribute, with most cases extracted from...
17063         (length): ...here.  Redefine most cases in terms of insn_count.
17064         (single_insn): Delete.
17065         (can_delay): Use insn_count to check for single instructions.
17066         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
17067         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
17068         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
17069         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
17070         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
17071         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
17072         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
17073         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
17074         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
17075         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
17076         rather than "length".
17077         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
17078         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
17079         Use "insn_count" rather than "length".
17080         * config/mips/mips-dsp.md
17081         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
17082         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
17083         length attributes.
17085 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
17087         PR tree-optimization/57361
17088         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
17090 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
17092         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
17093         to split.
17094         (aarch64_simd_combine<mode>): New instruction expansion.
17095         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
17096         function prototype.
17097         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
17098         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
17100 2013-06-12  Jan Hubicka  <jh@suse.cz>
17102         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
17103         decl has when in streaming stage.
17104         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
17105         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
17107 2013-06-12  Roland Stigge  <stigge@antcom.de>
17109         PR target/57578
17110         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
17112 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17114         PR tree-optimization/57537
17115         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
17116         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
17118 2013-06-12  Richard Biener  <rguenther@suse.de>
17120         * data-streamer.h (streamer_write_char_stream): CSE
17121         obs->current_pointer.
17122         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
17123         streamer_write_char_stream manually and optimize the resulting loop.
17124         (streamer_write_hwi_stream): Likewise.
17126 2013-06-12  Jan Hubicka  <jh@suse.cz>
17128         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
17129         * cgraph.h (varpool_create_empty_node): Declare.
17130         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
17131         duplicated nodes.
17132         * symtab.c (symtab_unregister_node): Be lax about missin entries
17133         in node hash.
17134         (symtab_get_node): Update comment.
17135         * varpool.c (varpool_create_empty_node): Break out from ...
17136         (varpool_node_for_decl): ... here.
17137         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
17139 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
17141         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
17142         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
17143         part.  Use straight-line flow at the end.
17144         <COMPONENT_REF>: Remove superfluous else.
17145         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
17147 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17149         PR target/56564
17150         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
17151         target hook even for !TREE_PUBLIC decls.  If no resolution info
17152         is available, return false for common and external decls.
17154 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17156         * config/rl78/constraints.md (U): New constraint.
17157         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
17158         valloc attribute.
17160 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17162         PR target/57589
17163         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
17164         to allow returning address to AT_PLATFORM name.
17166 2013-06-11  Jan Hubicka  <jh@suse.cz>
17168         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
17169         * cgraph.h (symtab_node_base): Add weakref flag.
17170         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
17171         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
17172         (output_weakrefs): Use weakref flag.
17173         * fold-const.c (simple_operand_p): Handle WEAK.
17174         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
17175         * ipa.c (varpool_externally_visible_p): Drop weakref.
17176         (function_and_variable_visibility): Update comment; fix weakref
17177         sanity checks; do not clear DECL_WEAK on them.
17178         * lto-cgraph.c (lto_output_node): update.
17179         (lto_output_varpool_node): Update.
17180         (input_overwrite_node): Update.
17181         (input_node): Update.
17182         (input_varpool_node): Update.
17183         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
17184         (lto_symtab_merge_symbols): Add sanity check.
17185         (lto_symtab_prevailing_decl): Do not special case weakrefs.
17186         * passes.c (rest_of_decl_compilation): Set static flag, too.
17187         * symtab.c (dump_symtab_base): Dump weakref.
17188         (verify_symtab_base): Sanity check weakrefs.
17189         (symtab_make_decl_local): Remove duplicated code.
17190         (symtab_alias_ultimate_target): Simplify.
17191         * varpool.c (varpool_create_variable_alias): Set weakref flag.
17193 2013-06-11  Tom de Vries  <tom@codesourcery.com>
17195         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
17196         sequence_vect == NULL.
17198 2013-06-11  DJ Delorie  <dj@redhat.com>
17200         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
17201         (rl78_unwind_word_mode): New.
17203 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17205         * final.c (debug_prefix_maps): Make static.
17207 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17209         * function.c (initial_trampoline): Remove stray copy.
17211 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
17213         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
17215 2013-06-11  Martin Jambor  <mjambor@suse.cz>
17217         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
17218         within bounds at the beginning of the function.
17220 2013-06-11  Alan Modra  <amodra@gmail.com>
17222         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
17223         reporting.
17224         (get_named_section): Don't NULL !DECL_P decl.
17226 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
17228         * doc/invoke.texi (core-avx2): Document.
17229         (slm): Likewise.
17230         (atom): Updated with MOVBE.
17232 2013-06-11  Richard Biener  <rguenther@suse.de>
17234         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
17236 2013-06-11  Anton Blanchard  <anton@samba.org>
17238         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
17239         correct shift value in little-endian mode.
17241 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
17243         PR target/56564
17244         * varasm.c (get_variable_align): Move #endif to the right place.
17246 2013-06-10  Cary Coutant  <ccoutant@google.com>
17248         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
17249         for hash so that hash table traversal order is deterministic.
17251 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
17252             Pat Haugen  <pthaugen@us.ibm.com>
17253             Peter Bergner  <bergner@vnet.ibm.com>
17255         * config/rs6000/vector.md (GPR move splitter): Do not split moves
17256         of vectors in GPRS if they are direct moves or quad word load or
17257         store moves.
17259         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
17260         declaration.
17261         (direct_move_p): Likewise.
17262         (quad_load_store_p): Likewise.
17264         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
17265         classes into bins based on the physical register type.
17266         (reg_class_to_reg_type): Likewise.
17267         (IS_STD_REG_TYPE): Likewise.
17268         (IS_FP_VECT_REG_TYPE): Likewise.
17269         (reload_fpr_gpr): Arrays to determine what insn to use if we can
17270         use direct move instructions.
17271         (reload_gpr_vsx): Likewise.
17272         (reload_vsx_gpr): Likewise.
17273         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
17274         information that is a simplification of register classes.  Also
17275         precalculate direct move reload helpers.
17276         (direct_move_p): New function to return true if the operation can
17277         be done as a direct move instruciton.
17278         (quad_load_store_p): New function to return true if the operation
17279         is a quad memory operation.
17280         (rs6000_legitimize_address): If quad memory, only allow register
17281         indirect for TImode addresses.
17282         (rs6000_legitimate_address_p): Likewise.
17283         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
17284         (rs6000_reload_register_type): Likewise.
17285         (register_to_reg_type): Return register type.
17286         (rs6000_secondary_reload_simple_move): New helper function for
17287         secondary reload and secondary memory needed to identify anything
17288         that is a simple move, and does not need reloading.
17289         (rs6000_secondary_reload_direct_move): New helper function for
17290         secondary reload to identify cases that can be done with several
17291         instructions via the direct move instructions.
17292         (rs6000_secondary_reload_move): New helper function for secondary
17293         reload to identify moves between register types that can be done.
17294         (rs6000_secondary_reload): Add support for quad memory operations
17295         and for direct move.
17296         (rs6000_secondary_memory_needed): Likewise.
17297         (rs6000_debug_secondary_memory_needed): Change argument names.
17298         (rs6000_output_move_128bit): New function to return the move to
17299         use for 128-bit moves, including knowing about the various
17300         limitations of quad memory operations.
17302         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
17303         memory operations.  call rs6000_output_move_128bit for the actual
17304         instruciton(s) to generate.
17305         (vsx_movti_64bit): Likewise.
17307         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
17308         (UNSPEC_P8V_MTVSRWZ): Likewise.
17309         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
17310         (UNSPEC_P8V_MTVSRD): Likewise.
17311         (UNSPEC_P8V_XXPERMDI): Likewise.
17312         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
17313         (UNSPEC_FUSION_GPR): Likewise.
17314         (FMOVE128_GPR): New iterator for direct move.
17315         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
17316         (f32_sv): Likewise.
17317         (f32_dm): Likewise.
17318         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
17319         loads and direct move instructions.
17320         (zero_extendsidi2_lfiwzx): Likewise.
17321         (extendsidi2_lfiwax): Likewise.
17322         (extendsidi2_nocell): Likewise.
17323         (floatsi<mode>2_lfiwax): Likewise.
17324         (lfiwax): Likewise.
17325         (floatunssi<mode>2_lfiwzx): Likewise.
17326         (lfiwzx): Likewise.
17327         (fix_trunc<mode>_stfiwx): Likewise.
17328         (fixuns_trunc<mode>_stfiwx): Likewise.
17329         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
17330         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
17331         (parity<mode>2_cmpb): Set length/type attr.
17332         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
17333         for 'mr.' to fast_compare.
17334         (bpermd_<mode>): Change type attr to popcnt.
17335         (p8_fmrgow_<mode>): New insns for power8 direct move support.
17336         (p8_mtvsrwz_1): Likewise.
17337         (p8_mtvsrwz_2): Likewise.
17338         (reload_fpr_from_gpr<mode>): Likewise.
17339         (p8_mtvsrd_1): Likewise.
17340         (p8_mtvsrd_2): Likewise.
17341         (p8_xxpermdi_<mode>): Likewise.
17342         (reload_vsx_from_gpr<mode>): Likewise.
17343         (reload_vsx_from_gprsf): Likewise.
17344         (p8_mfvsrd_3_<mode>): LIkewise.
17345         (reload_gpr_from_vsx<mode>): Likewise.
17346         (reload_gpr_from_vsxsf): Likewise.
17347         (p8_mfvsrd_4_disf): Likewise.
17348         (multi-word GPR splits): Do not split direct moves or quad memory
17349         operations.
17351 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
17353         * tree-into-ssa.c (interesting_blocks): Make static.
17355 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
17357         PR target/56564
17358         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
17359         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
17360         Use DATA_ABI_ALIGNMENT for that case instead if defined.
17361         (get_variable_align): New function.
17362         (get_variable_section, emit_bss, emit_common,
17363         assemble_variable_contents, place_block_symbol): Use
17364         get_variable_align instead of DECL_ALIGN.
17365         (assemble_noswitch_variable): Add align argument, use it
17366         instead of DECL_ALIGN.
17367         (assemble_variable): Adjust caller.  Use get_variable_align
17368         instead of DECL_ALIGN.
17369         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
17370         caller.
17371         (DATA_ABI_ALIGNMENT): Define.
17372         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
17373         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
17374         opt is false, only return the psABI mandated alignment increase.
17375         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
17376         (DATA_ABI_ALIGNMENT): ... this.
17377         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
17378         (DATA_ABI_ALIGNMENT): ... this.
17379         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
17380         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
17381         (DATA_ABI_ALIGNMENT): ... this.
17382         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
17383         * doc/tm.texi: Regenerated.
17385 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
17387         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
17388         cmp_code to construct REG_EQUAL note.
17390 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
17392         PR target/57568
17393         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
17394         that operands[2] doesn't overlap with operands[0].
17396 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
17397             Jan Hubicka  <jh@suse.cz>
17399         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
17400         hack to mark symbols as used.
17402 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
17404         PR rtl-optimization/57559
17405         * lra-constraints.c (process_alt_operands): Don't discourage
17406         memory with known offset for offsetable memory constraint.
17407         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
17409 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
17411         * varasm.c (struct oc_local_state): Reorder fields.
17412         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
17413         and adjust accordingly.
17414         (output_constructor): Reorder initialization code and adjust call to
17415         output_constructor_bitfield.
17417 2013-06-07  Jan Hubicka  <jh@suse.cz>
17419         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
17421 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
17423         * tree-object-size.c (unknown): Make const.
17425 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17427         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
17428         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
17429         for last alternative in the cpu_facility attribute.
17431 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17433         PR target/56315
17434         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
17435         (xordi3): Change operand 2 constraint to arm_xordi_operand.
17436         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
17437         * config/arm/constraints.md (Dg): New constraint.
17438         * config/arm/neon.md (xordi3_neon): Remove.
17439         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
17440         * config/arm/predicates.md (arm_xordi_operand): New predicate.
17442 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17444         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
17445         Clean up alternatives.
17447 2013-06-07  Alan Modra  <amodra@gmail.com>
17449         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
17450         va_list_gpr_size.
17452 2013-06-07  Alan Modra  <amodra@gmail.com>
17454         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
17456 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17458         * config/arm/constraints.md (Df): New constraint.
17459         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
17460         Correct length attribute for last two alternatives.
17462 2013-06-07  Alan Modra  <amodra@gmail.com>
17464         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
17465         override user -mfp-in-toc.
17466         (offsettable_ok_by_alignment): Consider just the current access
17467         rather than the whole object, unless BLKmode.  Handle
17468         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
17469         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
17470         for -mcmodel=medium.
17471         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
17472         override user -mfp-in-toc or -msum-in-toc.  Default to
17473         -mno-fp-in-toc for -mcmodel=medium.
17475 2013-06-06  DJ Delorie  <dj@redhat.com>
17477         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
17478         TARGET_VALID_POINTER_MODE.
17480 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
17481             Pat Haugen  <pthaugen@us.ibm.com>
17482             Peter Bergner  <bergner@vnet.ibm.com>
17484         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
17485         Document new power8 builtins.
17487         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
17488         condition code register, to allow 128-bit logical operations to be
17489         done in the VSX or GPR registers.
17490         (nor<mode>3): Use the canonical form for nor.
17491         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
17492         vclz*, and vpopcnt* vector instructions.
17493         (nand<mode>3): Likewise.
17494         (orc<mode>3): Likewise.
17495         (clz<mode>2): LIkewise.
17496         (popcount<mode>2): Likewise.
17498         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
17499         that only the GPRs are recognized.
17501         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17502         support for new power8 builtins.
17504         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
17505         builtin functions.
17506         (xscvdpspn): Likewise.
17507         (vclz): Likewise.
17508         (vclzb): Likewise.
17509         (vclzh): Likewise.
17510         (vclzw): Likewise.
17511         (vclzd): Likewise.
17512         (vpopcnt): Likewise.
17513         (vpopcntb): Likewise.
17514         (vpopcnth): Likewise.
17515         (vpopcntw): Likewise.
17516         (vpopcntd): Likewise.
17517         (vgbbd): Likewise.
17518         (vmrgew): Likewise.
17519         (vmrgow): Likewise.
17520         (eqv): Likewise.
17521         (eqv_v16qi3): Likewise.
17522         (eqv_v8hi3): Likewise.
17523         (eqv_v4si3): Likewise.
17524         (eqv_v2di3): Likewise.
17525         (eqv_v4sf3): Likewise.
17526         (eqv_v2df3): Likewise.
17527         (nand): Likewise.
17528         (nand_v16qi3): Likewise.
17529         (nand_v8hi3): Likewise.
17530         (nand_v4si3): Likewise.
17531         (nand_v2di3): Likewise.
17532         (nand_v4sf3): Likewise.
17533         (nand_v2df3): Likewise.
17534         (orc): Likewise.
17535         (orc_v16qi3): Likewise.
17536         (orc_v8hi3): Likewise.
17537         (orc_v4si3): Likewise.
17538         (orc_v2di3): Likewise.
17539         (orc_v4sf3): Likewise.
17540         (orc_v2df3): Likewise.
17542         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
17543         allow power8 quad mode in 64-bit.
17544         (rs6000_builtin_vectorized_function): Add support to vectorize
17545         ISA 2.07 count leading zeros, population count builtins.
17546         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
17547         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
17548         (builtin_function_type): Add vgbbd builtin function which takes an
17549         unsigned argument.
17550         (altivec_expand_vec_perm_const): Add support for new power8 merge
17551         instructions.
17553         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
17554         that does not include TImdoe for use with 32-bit.
17555         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
17556         instructions.
17557         (UNSPEC_VSX_CVDPSPN): Likewise.
17558         (vsx_xscvdpspn): Likewise.
17559         (vsx_xscvspdpn): Likewise.
17560         (vsx_xscvdpspn_scalar): Likewise.
17561         (vsx_xscvspdpn_directmove): Likewise.
17562         (vsx_and<mode>3): Split logical operations into 32-bit and
17563         64-bit. Add support to do logical operations on TImode as well as
17564         VSX vector types.  Allow logical operations to be done in either
17565         VSX registers or in general purpose registers in 64-bit mode.  Add
17566         splitters if GPRs were used. For AND, add clobber of CCmode to
17567         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
17568         encoding.
17569         (vsx_and<mode>3_32bit): Likewise.
17570         (vsx_and<mode>3_64bit): Likewise.
17571         (vsx_ior<mode>3): Likewise.
17572         (vsx_ior<mode>3_32bit): Likewise.
17573         (vsx_ior<mode>3_64bit): Likewise.
17574         (vsx_xor<mode>3): Likewise.
17575         (vsx_xor<mode>3_32bit): Likewise.
17576         (vsx_xor<mode>3_64bit): Likewise.
17577         (vsx_one_cmpl<mode>2): Likewise.
17578         (vsx_one_cmpl<mode>2_32bit): Likewise.
17579         (vsx_one_cmpl<mode>2_64bit): Likewise.
17580         (vsx_nor<mode>3): Likewise.
17581         (vsx_nor<mode>3_32bit): Likewise.
17582         (vsx_nor<mode>3_64bit): Likewise.
17583         (vsx_andc<mode>3): Likewise.
17584         (vsx_andc<mode>3_32bit): Likewise.
17585         (vsx_andc<mode>3_64bit): Likewise.
17586         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
17587         and xxlorc instructions.
17588         (vsx_eqv<mode>3_64bit): Likewise.
17589         (vsx_nand<mode>3_32bit): Likewise.
17590         (vsx_nand<mode>3_64bit): Likewise.
17591         (vsx_orc<mode>3_32bit): Likewise.
17592         (vsx_orc<mode>3_64bit): Likewise.
17594         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
17596         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
17597         instruction.
17598         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
17599         (p8_vmrgow): Likewise.
17600         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
17601         GPRs to be split under VSX.
17602         (p8v_clz<mode>2): Add power8 count leading zero support.
17603         (p8v_popcount<mode>2): Add power8 population count support.
17604         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
17605         support.
17607         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
17608         instruction.
17610         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
17611         builtin functions.
17612         (vec_nand): Likewise.
17613         (vec_vclz): Likewise.
17614         (vec_vclzb): Likewise.
17615         (vec_vclzd): Likewise.
17616         (vec_vclzh): Likewise.
17617         (vec_vclzw): Likewise.
17618         (vec_vgbbd): Likewise.
17619         (vec_vmrgew): Likewise.
17620         (vec_vmrgow): Likewise.
17621         (vec_vpopcnt): Likewise.
17622         (vec_vpopcntb): Likewise.
17623         (vec_vpopcntd): Likewise.
17624         (vec_vpopcnth): Likewise.
17625         (vec_vpopcntw): Likewise.
17627 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17629         PR rtl-optimization/57468
17630         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
17631         spilled pseudos.
17633 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17635         PR rtl-optimization/57459
17636         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
17637         type when setting live regs.
17639 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17641         * config/s390/s390.opt (mlra): New option.
17642         * config/s390/s390.c (s390_decompose_address): Check displacement
17643         for all registers for LRA.
17644         (s390_secondary_reload): Don't used secondary reloads for LRA.
17645         (s390_lra_p): New function.
17646         (TARGET_LRA_P): Define.
17647         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
17648         of attribute cpu_facility to zarch for the last alternative.
17649         (*cmpmem_short): Ditto.
17651 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
17653         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
17654         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
17655         (arm_expand_prologue): Likewise.
17657 2013-06-06  Teresa Johnson  <tejohnson@google.com>
17659         PR c++/53743
17660         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
17661         as this is now done by redirect_edge_and_branch_force.
17662         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
17663         barriers, and fix interaction with splitting.
17664         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
17665         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
17666         reflect changes made in the routine.
17667         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
17668         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
17669         since this is called in cfglayout mode, and replace partition fixup
17670         with assert as that is now done by force_nonfallthru_and_redirect.
17671         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
17672         already be marked with region crossing note.
17673         (insert_section_boundary_note): Make non-static, gate on flag
17674         has_bb_partition, rewrite to also check for multiple partitions.
17675         (rest_of_handle_reorder_blocks): Remove call to
17676         insert_section_boundary_note, now done later during free_cfg.
17677         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
17678         * bb-reorder.h (insert_section_boundary_note): Declare.
17679         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
17680         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
17681         invoke insert_section_boundary_note.
17682         (try_redirect_by_replacing_jump): Remove unnecessary
17683         check for region crossing note.
17684         (fixup_partition_crossing): New function.
17685         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
17686         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
17687         in non-cfglayout mode.
17688         (force_nonfallthru_and_redirect): Fixup partition boundaries,
17689         remove old code that tried to do this. Emit barrier correctly
17690         when we are in cfglayout mode.
17691         (last_bb_in_partition): New function.
17692         (rtl_split_edge): Correctly fixup partition boundaries.
17693         (commit_one_edge_insertion): Remove old code that tried to
17694         fixup region crossing edge since this is now handled in
17695         split_block, and set up insertion point correctly since
17696         block may now end in a jump.
17697         (verify_hot_cold_block_grouping): Guard against checking when not in
17698         linearized RTL mode.
17699         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
17700         notes.
17701         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
17702         rtl_verify_flow_info, so not called in cfglayout mode.
17703         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
17704         (fixup_reorder_chain): Remove old code that attempted to fixup region
17705         crossing note as this is now handled in force_nonfallthru_and_redirect.
17706         (duplicate_insn_chain): Don't duplicate switch section notes.
17707         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
17708         note.
17709         * basic-block.h (emit_barrier_after_bb): Declare.
17711 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17713         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
17714         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
17715         arm_usatsihi): Adjust alternatives for arm_restrict_it.
17717 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17719         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
17720         where appropriate.
17721         * config/arm/ldmstm.md: Regenerate.
17723 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17725         * config/arm/sync.md (atomic_loaddi_1):
17726         Disable predication for arm_restrict_it.
17727         (arm_load_exclusive<mode>): Likewise.
17728         (arm_load_exclusivesi): Likewise.
17729         (arm_load_exclusivedi): Likewise.
17730         (arm_load_acquire_exclusive<mode>): Likewise.
17731         (arm_load_acquire_exclusivesi): Likewise.
17732         (arm_load_acquire_exclusivedi): Likewise.
17733         (arm_store_exclusive<mode>): Likewise.
17734         (arm_store_exclusive<mode>): Likewise.
17735         (arm_store_release_exclusivedi): Likewise.
17736         (arm_store_release_exclusive<mode>): Likewise.
17738 2013-06-06  Richard Biener  <rguenther@suse.de>
17740         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
17741         after LTO_null.
17742         (lto_tag_is_tree_code_p): Adjust.
17743         (lto_tag_is_gimple_code_p): Likewise.
17744         (lto_gimple_code_to_tag): Likewise.
17745         (lto_tag_to_gimple_code): Likewise.
17746         (lto_tree_code_to_tag): Likewise.
17747         (lto_tag_to_tree_code): Likewise.
17748         * data-streamer.h (streamer_write_hwi_in_range): Use
17749         uhwi streaming to stream the normalized range.
17750         (streamer_read_hwi_in_range): Likewise.
17752 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17754         * config/arm/arm.md (enabled_for_depr_it): New attribute.
17755         (predicable_short_it): Likewise.
17756         (predicated): Likewise.
17757         (enabled): Handle above.
17758         (define_cond_exec): Set predicated attribute to yes.
17760 2013-06-05  Mike Stump  <mikestump@comcast.net>
17762         * gdbinit.in (__FUNCTION__): Add.
17764 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
17766         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
17767         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
17769 2013-06-05  Jan Hubicka  <jh@suse.cz>
17771         * varasm.c (mark_decl_referenced): Revert the removal until targets
17772         are fixed.
17774 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
17776         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
17777         instead of mark_decl_referenced.
17779 2013-06-05  Jan Hubicka  <jh@suse.cz>
17781         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
17782         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
17783         and symtab_used_from_object_file_p.
17784         (cgraph_make_node_local_1): Clear forced_by_abi.
17785         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
17786         * cgraph.h (symtab_node_base): Add forced_by_abi.
17787         (decide_is_variable_needed): Remove.
17788         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
17789         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
17790         (decide_is_symbol_needed): ... this one; handle symbols in general;
17791         always analyze virtuals; honnor forced_by_abi.
17792         (cgraph_finalize_function): Update.
17793         (varpool_finalize_decl): Update.
17794         (symbol_defined_and_needed): Remove.
17795         (analyze_functions): Update.
17796         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17797         output_refs, input_overwrite_node): Handle forced_by_abi.
17798         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
17799         (address_taken_from_non_vtable_p): ... this one.
17800         (comdat_can_be_unshared_p_1): New function.
17801         (cgraph_comdat_can_be_unshared_p): Rename to ...
17802         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
17803         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
17804         (function_and_variable_visibility): Clear forced_by_abi as needed.
17805         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
17806         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
17807         * symtab.c (dump_symtab_base): Dump forced_by_abi.
17808         * varpool.c (decide_is_variable_needed): Remove.
17810 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17812         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
17813         (arm_option_override): Override arm_restrict_it where appropriate.
17814         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
17815         * config/arm/arm.opt (mrestrict-it): New command-line option.
17816         * doc/invoke.texi: Document -mrestrict-it.
17818 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
17820         * tsan.c (tsan_atomic_table): Make const.
17822 2013-06-05  Richard Biener  <rguenther@suse.de>
17824         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
17825         index associated with the tree we are supposed to replace.
17826         * tree-streamer-out.c (pack_ts_base_value_fields): Output
17827         TREE_ASM_WRITTEN as zero for everything but SSA names.
17829 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
17831         * tree-ssa-structalias.c (call_stmt_vars): Make static.
17833 2013-06-04  Jan Hubicka  <jh@suse.cz>
17835         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
17836         (input_node, input_varpool_node): Handle correctly external same
17837         body aliases.
17838         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
17839         nodes at ltrans stage.
17841 2013-06-04  Jan Hubicka  <jh@suse.cz>
17843         * ipa-inline.c (update_caller_keys): Fix availability test.
17844         (update_callee_keys): Likewise.
17845         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
17846         to follow ELF standard.
17848 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
17850         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
17851         (mips64r5900el-*-elf*): New configurations.
17852         * config/mips/mips-cpus.def (r5900): New processor.
17853         * config/mips/mips-tables.opt: Regenerate.
17854         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
17855         (mips_issue_rate): Handle PROCESSOR_R5900.
17856         (mips_reorg_process_insns): Force reorder mode for the R5900.
17857         * config/mips/mips.h (TARGET_MIPS5900): Define.
17858         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
17859         TARGET_MIPS5900.
17860         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
17861         TARGET_MIPS5900.
17862         * config/mips/mips.md (processor): Add r5900.
17863         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
17865 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17867         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
17868         into function to generate MOVI instruction.
17869         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
17870         (aarch64_preferred_simd_mode): Turn into wrapper.
17871         (aarch64_output_scalar_simd_mov_immediate): New function.
17872         * config/aarch64/aarch64-protos.h: Add prototype for above.
17874 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17876         * config/aarch64/aarch64.c (simd_immediate_info): Remove
17877         element_char member.
17878         (sizetochar): Return signed char.
17879         (aarch64_simd_valid_immediate): Remove elchar and other
17880         unnecessary variables.
17881         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
17882         Calculate element_char as required.
17883         * config/aarch64/aarch64-protos.h: Update and move prototype
17884         for aarch64_output_simd_mov_immediate.
17885         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
17886         Update arguments.
17888 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17890         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
17891         information completed by aarch64_simd_valid_immediate.
17892         (aarch64_legitimate_constant_p): Update arguments.
17893         (aarch64_simd_valid_immediate): Work with struct rather than many
17894         pointers.
17895         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
17896         (aarch64_simd_make_constant): Update arguments.
17897         (aarch64_output_simd_mov_immediate): Work with struct rather than
17898         many pointers.  Output immediate directly rather than as operand.
17899         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
17900         Update prototype.
17901         * config/aarch64/constraints.md (Dn): Update arguments.
17903 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17905         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
17906         longer static.
17907         (aarch64_simd_immediate_valid_for_move): Remove.
17908         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
17909         (aarch64_simd_make_constant): Update call.
17910         (aarch64_output_simd_mov_immediate): Update call.
17911         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
17912         Add prototype.
17913         * config/aarch64/constraints.md (Dn): Update call.
17915 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
17917         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
17918         return type to bool for prototype.
17919         (aarch64_legitimate_constant_p): Check for true instead of not -1.
17920         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
17921         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
17923 2013-06-04  Catherine Moore  <clm@codesourcery.com>
17925         * config/mips/mips.opt (meva): New.
17926         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
17927         (ASM_SPEC): Handle -meva.
17928         * doc/invoke.texi (meva):  Document.
17930 2013-06-04  Alan Modra  <amodra@gmail.com>
17932         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
17933         constant output.
17935 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17937         * rtl.def: Add extra fourth optional field to define_cond_exec.
17938         * gensupport.c (process_one_cond_exec): Process attributes from
17939         define_cond_exec.
17940         * doc/md.texi: Document fourth field in define_cond_exec.
17942 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
17944         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
17945         out the processing order as in store_bit_field_1.
17947 2013-06-04  Jan Hubicka  <jh@suse.cz>
17949         PR middle-end/57500
17950         * cgraphunit.c (cgraph_process_same_body_aliases): Create
17951         non-VAR_DECL node if it does not exist yet.
17953 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
17955         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
17956         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
17957         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
17958         target_cpu_default setting.
17960 2013-06-03  Teresa Johnson  <tejohnson@google.com>
17962         * dumpfile.c (opt_info_switch_p): Change -fopt-info
17963         default to -fopt-info=optimized instead of all.
17964         * doc/invoke.texi: Ditto.
17965         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
17966         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
17967         (execute_vect_slp): Emit BB vectorization success under
17968         MSG_OPTIMIZED_LOCATIONS.
17969         * tree-vect-slp.c (vect_slp_transform_bb): Change
17970         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
17971         * tree-vect-loop.c (vect_transform_loop): Ditto.
17973 2013-06-03  Jason Merrill  <jason@redhat.com>
17975         PR c++/57415
17976         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17977         Use TARGET_EXPR for C++.
17979 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
17981         PR rtl-optimization/57268
17982         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
17983         if DEBUG_INSN_P (insn).
17985         Reapply
17986         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
17988         PR rtl-optimization/57268
17989         * sched-deps.c (sched_analyze_2): Flush dependence lists if
17990         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
17992 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
17994         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
17995         (ix86_avoid_lea_for_addr): Likewise.
17996         (exact_dependency_1): Likewise.
17997         (ix86_adjust_cost): Likewise.
17998         (swap_top_of_ready_list): Fix formatting and !reload_completed check
17999         removed.
18000         (do_reorder_for_imul): Fix typo, formatting and
18001         !reload_completed check removed.
18002         (ix86_sched_reorder): Fix typo and formatting.
18003         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
18004         list.
18006 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
18008         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
18010 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
18012         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
18013         <STRING_CST>: Likewise.
18014         <VECTOR_CST>: Likewise.
18016 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
18017             Mikael Morin  <mikael@gcc.gnu.org>
18019         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
18020         * config.in: Regenerated.
18021         * configure: Regenerated.
18023 2013-06-01  Jan Hubicka  <jh@suse.cz>
18025         PR middle-end/57366
18026         * cgraphunit.c (compile): When weakref is not supported,
18027         set up transparent aliases before final output pass.
18028         * varasm.c (assemble_alias): Do not try to do it here.
18030 2013-06-01  Jan Hubicka  <jh@suse.cz>
18032         PR middle-end/57467
18033         * passes.c (for_per_function): Skip unanalyzed functions.
18035 2013-06-01  Jan Hubicka  <jh@suse.cz>
18037         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
18038         (lto_symtab_merge_symbols_1): ... this one.
18039         (lto_symtab_merge_cgraph_nodes): Rename to ...
18040         (lto_symtab_merge_symbols): ... this one; simplify.
18041         * cgraph.c (same_body_aliases_done): Rename to ...
18042         (cpp_implicit_aliases_done): ... this one.
18043         (cgraph_create_function_alias): Update.
18044         (cgraph_same_body_alias): Update.
18045         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
18046         (verify_edge_corresponds_to_fndecl): Simplify.
18047         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
18048         (cgraph_node): Remove same_body_alias.
18049         (varpool_node): Remove alias_of and extra_name_alias.
18050         (same_body_aliases_done): Rename to ..
18051         (cpp_implicit_aliases_done): ... this one.
18052         (symtab_alias_ultimate_target): Add default parameter.
18053         (symtab_resolve_alias): New function.
18054         (fixup_same_cpp_alias_visibility): Declare.
18055         (cgraph_function_node): Add default parameter.
18056         (cgraph_node_asm_name): Likewise.
18057         (cgraph_function_or_thunk_node): Add default parameter; do
18058         not ICE when it is NULL.
18059         (varpool_variable_node): Likewise.
18060         * tree-emutls.c (create_emultls_var): Update.
18061         (ipa_lower_emutls): Update.
18062         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
18063         (cgraph_reset_node): Reset alias info.
18064         (cgraph_finalize_function): Update.
18065         (fixup_same_cpp_alias_visibility): Move to symtab.c.
18066         (analyze_function): Simplify.
18067         (cgraph_process_same_body_aliases): Simplify.
18068         (analyze_functions): Fixup same body aliases.
18069         (handle_alias_pairs): Simplify.
18070         (assemble_thunk): Update.
18071         (assemble_thunks_and_aliases): Update.
18072         (output_weakrefs): Rewrite.
18073         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
18074         (lto_output_varpool_node): Likewise.
18075         (compute_ltrans_boundary): Remve assert.
18076         (get_alias_symbol): New functoin.
18077         (input_node): Rewrite alias handling.
18078         (input_varpool_node): Likewise.
18079         * ipa-pure-const.c (propagate_pure_const): Fix formating.
18080         * ipa.c (process_references): Handle weakrefs correctly.
18081         (symtab_remove_unreachable_nodes): Likewise.
18082         * trans-mem.c (get_cg_data): Update.
18083         (ipa_tm_create_version_alias): Update.
18084         (ipa_tm_execute): Update.
18085         * symtab.c (dump_symtab_base): Dump aliases.
18086         (verify_symtab_base): Verify aliases.
18087         (symtab_node_availability): New function.
18088         (symtab_alias_ultimate_target): Simplify.
18089         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
18090         handle all the fixup cases.
18091         (symtab_resolve_alias): New function.
18092         * passes.c (ipa_write_summaries): Handle weakrefs.
18093         * varpool.c (varpool_analyze_node): Simplify.
18094         (assemble_aliases): Update.
18095         (varpool_create_variable_alias): Simplify.
18096         (varpool_extra_name_alias): Simplify.
18097         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
18098         (lto_symtab_merge_symbols): ... this one.
18100 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
18102         Revert
18103         PR rtl-optimization/57268
18104         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18105         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18107 2013-06-01  Tobias Burnus  <burnus@net-b.de>
18109         Partially reverted:
18110         2013-05-31  Tobias Burnus  <burnus@net-b.de>
18112         PR middle-end/57073
18113         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
18114         further up.
18116 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18118         PR rtl-optimization/57268
18119         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18120         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18122 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18124         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
18125         unordered comparison operators when -fno-trapping-math is in effect
18126         on the e500.
18127         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
18128         and implement unordered comparison operators properly on the e500.
18130 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18132         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
18133         for constant scalar integers.
18134         (simplify_relational_operation_1): Likewise.
18136 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
18138         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
18139         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
18140         Fix comment.
18142 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18143             Igor Zamyatin  <igor.zamyatin@intel.com>
18145         Silvermont (SLM) architecture performance tuning.
18146         * config/i386/i386.h (enum ix86_tune_indices): Add
18147         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
18148         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
18150         * config/i386/i386.c (initial_ix86_tune_features)
18151         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
18152         (ix86_lea_outperforms): Handle Silvermont tuning.
18153         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
18154         call.
18155         (ix86_use_lea_for_mov): Likewise.
18156         (ix86_avoid_lea_for_addr): Likewise.
18157         (ix86_lea_for_add_ok): Likewise.
18158         (exact_dependency_1): New function.
18159         (exact_store_load_dependency): Likewise.
18160         (ix86_adjust_cost): Handle Silvermont tuning.
18161         (do_reoder_for_imul): Likewise.
18162         (swap_top_of_ready_list): New function.
18163         (ix86_sched_reorder): Changed to handle Silvermont tuning.
18165         * config/i386/i386.md (peepholes that split memory operand in fp
18166         converts): New.
18168 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18170         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
18171         Remove un-necessary braces.
18173 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18175         * config/aarch64/aarch64.c (aarch64_classify_symbol):
18176         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
18178 2013-05-31  Tobias Burnus  <burnus@net-b.de>
18180         PR middle-end/57073
18181         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
18183 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18185         PR target/56315
18186         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
18187         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
18188         * config/arm/neon.md (iordi3_neon): Remove.
18189         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
18190         * config/arm/predicates.md (imm_for_neon_logic_operand):
18191         Move to earlier in the file.
18192         (neon_logic_op2): Likewise.
18193         (arm_iordi_operand_neon): New predicate.
18195 2013-05-31  Richard Biener  <rguenther@suse.de>
18197         PR tree-optimization/57478
18198         PR tree-optimization/57453
18199         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
18200         are life as well.
18202 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
18204         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
18205         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
18207 2013-05-30  Tobias Burnus  <burnus@net-b.de>
18208             Thomas Koenig  <tkoenig@gcc.gnu.org>
18210         PR middle-end/57073
18211         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
18212         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
18214 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
18216         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
18218 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
18220         * target.def (register_usage_leveling_p): New hook.
18221         * targhooks.c (default_register_usage_leveling_p): New.
18222         * targhooks.h (default_register_usage_leveling_p): New prototype.
18223         * lra-assigns.c (register_usage_leveling_p): Use the hook.
18224         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
18225         * doc/tm.texi: Update.
18226         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
18228 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
18230         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
18231         (*insv_reg<mode>): New define_insn.
18233 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
18235         PR rtl-optimization/57439
18236         * postreload.c (move2add_valid_value_p): Check that we have
18237         a zero subreg_regno_offset when accessing the register in
18238         the requested mode.
18240 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18241             Igor Zamyatin  <igor.zamyatin@intel.com>
18243         Silvermont (SLM) architecture pipeline model, tuning and
18244         insn selection.
18245         * config.gcc: Add slm config options and target.
18247         * config/i386/slm.md: New.
18249         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
18251         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
18252         PROCESSOR_SLM.
18253         (ix86_target_macros_internal): Likewise.
18255         * config/i386/i386.c (slm_cost): New cost.
18256         (m_SLM): New macro flag.
18257         (initial_ix86_tune_features): Set m_SLM.
18258         (x86_accumulate_outgoing_args): Likewise.
18259         (x86_arch_always_fancy_math_387): Likewise.
18260         (processor_target_table): Add slm cost.
18261         (cpu_names): Add slm cpu name.
18262         (x86_option_override_internal): Set SLM ISA.
18263         (ix86_issue_rate): New case PROCESSOR_SLM.
18264         (ia32_multipass_dfa_lookahead): Likewise.
18265         (fold_builtin_cpu): Add slm.
18267         * config/i386/i386.h (TARGET_SLM): New target macro.
18268         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
18269         (processor_type): Add PROCESSOR_SLM.
18271         * config/i386/i386.md (cpu): Add new value "slm".
18272         (slm.md): Include slm.md.
18274 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
18275             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18277         * config/arm/arm-protos.h: Add and update function protos.
18278         * config/arm/arm.c (use_simple_return_p): New added.
18279         (thumb2_expand_return): Check simple_return flag.
18280         * config/arm/arm.md: Add simple_return and conditional simple_return.
18281         * config/arm/iterators.md: Add iterator for return and simple_return.
18283 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18285         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
18286         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
18287         (arm_emit_vfp_multi_reg_pop): Likewise.
18288         (thumb2_emit_ldrd_pop): Likewise.
18289         (arm_expand_epilogue): Add misc REG_CFA notes.
18290         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
18292 2013-05-29  Lawrence Crowl  <crowl@google.com>
18294         * config/arm/t-arm: Update for below.
18296         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
18297         Change type to hash_table.  Update dependent calls and types.
18299         * config/i386/t-cygming: Update for below.
18301         * config/i386/t-interix: Update for below.
18303         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
18304         Change type to hash_table.  Update dependent calls and types.
18305         (i386_find_on_wrapper_list::wrappers): Likewise.
18307         * config/ia64/t-ia64: Update for below.
18309         * config/ia64/ia64.c (bundle_state_table):
18310         Change type to hash_table.  Update dependent calls and types.
18312         * config/mips/mips.c (mips_reorg_process_insns::htab):
18313         Change type to hash_table.  Update dependent calls and types.
18315         * config/sol2.c (solaris_comdat_htab):
18316         Change type to hash_table.  Update dependent calls and types.
18318         * config/t-sol2: Update for above.
18320 2013-05-29  Teresa Johnson  <tejohnson@google.com>
18322         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
18323         functions are not yet marked as defined.
18325 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
18326             Pat Haugen  <pthaugen@us.ibm.com>
18327             Peter Bergner  <bergner@vnet.ibm.com>
18329         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
18330         instructions.
18331         (VEC_A): Likewise.
18332         (VEC_C): Likewise.
18333         (vrotl<mode>3): Likewise.
18334         (vashl<mode>3): Likewise.
18335         (vlshr<mode>3): Likewise.
18336         (vashr<mode>3): Likewise.
18338         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18339         support for power8 V2DI builtins.
18341         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
18342         power8 V2DI builtins.
18343         (vupkhsw): Likewise.
18344         (vupklsw): Likewise.
18345         (vaddudm): Likewise.
18346         (vminsd): Likewise.
18347         (vmaxsd): Likewise.
18348         (vminud): Likewise.
18349         (vmaxud): Likewise.
18350         (vpkudum): Likewise.
18351         (vpksdss): Likewise.
18352         (vpkudus): Likewise.
18353         (vpksdus): Likewise.
18354         (vrld): Likewise.
18355         (vsld): Likewise.
18356         (vsrd): Likewise.
18357         (vsrad): Likewise.
18358         (vsubudm): Likewise.
18359         (vcmpequd): Likewise.
18360         (vcmpgtsd): Likewise.
18361         (vcmpgtud): Likewise.
18362         (vcmpequd_p): Likewise.
18363         (vcmpgtsd_p): Likewise.
18364         (vcmpgtud_p): Likewise.
18365         (vupkhsw): Likewise.
18366         (vupklsw): Likewise.
18367         (vaddudm): Likewise.
18368         (vmaxsd): Likewise.
18369         (vmaxud): Likewise.
18370         (vminsd): Likewise.
18371         (vminud): Likewise.
18372         (vpksdss): Likewise.
18373         (vpksdus): Likewise.
18374         (vpkudum): Likewise.
18375         (vpkudus): Likewise.
18376         (vrld): Likewise.
18377         (vsld): Likewise.
18378         (vsrad): Likewise.
18379         (vsrd): Likewise.
18380         (vsubudm): Likewise.
18382         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
18383         support for power8 V2DI instructions.
18385         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
18386         power8 V2DI instructions.  Combine pack and unpack insns to use an
18387         iterator for each mode.  Check whether a particular mode supports
18388         Altivec instructions instead of just checking TARGET_ALTIVEC.
18389         (UNSPEC_VPKUWUM): Likewise.
18390         (UNSPEC_VPKSHSS): Likewise.
18391         (UNSPEC_VPKSWSS): Likewise.
18392         (UNSPEC_VPKUHUS): Likewise.
18393         (UNSPEC_VPKSHUS): Likewise.
18394         (UNSPEC_VPKUWUS): Likewise.
18395         (UNSPEC_VPKSWUS): Likewise.
18396         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
18397         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
18398         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
18399         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
18400         (UNSPEC_VUPKHSB): Likewise.
18401         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
18402         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
18403         (UNSPEC_VUPKHSH): Likewise.
18404         (UNSPEC_VUPKLSB): Likewise.
18405         (UNSPEC_VUPKLSH): Likewise.
18406         (VI2): Likewise.
18407         (VI_char): Likewise.
18408         (VI_scalar): Likewise.
18409         (VI_unit): Likewise.
18410         (VP): Likewise.
18411         (VP_small): Likewise.
18412         (VP_small_lc): Likewise.
18413         (VU_char): Likewise.
18414         (add<mode>3): Likewise.
18415         (altivec_vaddcuw): Likewise.
18416         (altivec_vaddu<VI_char>s): Likewise.
18417         (altivec_vadds<VI_char>s): Likewise.
18418         (sub<mode>3): Likewise.
18419         (altivec_vsubcuw): Likewise.
18420         (altivec_vsubu<VI_char>s): Likewise.
18421         (altivec_vsubs<VI_char>s): Likewise.
18422         (altivec_vavgs<VI_char>): Likewise.
18423         (altivec_vcmpbfp): Likewise.
18424         (altivec_eq<mode>): Likewise.
18425         (altivec_gt<mode>): Likewise.
18426         (altivec_gtu<mode>): Likewise.
18427         (umax<mode>3): Likewise.
18428         (smax<mode>3): Likewise.
18429         (umin<mode>3): Likewise.
18430         (smin<mode>3): Likewise.
18431         (altivec_vpkuhum): Likewise.
18432         (altivec_vpkuwum): Likewise.
18433         (altivec_vpkshss): Likewise.
18434         (altivec_vpkswss): Likewise.
18435         (altivec_vpkuhus): Likewise.
18436         (altivec_vpkshus): Likewise.
18437         (altivec_vpkuwus): Likewise.
18438         (altivec_vpkswus): Likewise.
18439         (altivec_vpks<VI_char>ss): Likewise.
18440         (altivec_vpks<VI_char>us): Likewise.
18441         (altivec_vpku<VI_char>us): Likewise.
18442         (altivec_vpku<VI_char>um): Likewise.
18443         (altivec_vrl<VI_char>): Likewise.
18444         (altivec_vsl<VI_char>): Likewise.
18445         (altivec_vsr<VI_char>): Likewise.
18446         (altivec_vsra<VI_char>): Likewise.
18447         (altivec_vsldoi_<mode>): Likewise.
18448         (altivec_vupkhsb): Likewise.
18449         (altivec_vupkhs<VU_char>): Likewise.
18450         (altivec_vupkls<VU_char>): Likewise.
18451         (altivec_vupkhsh): Likewise.
18452         (altivec_vupklsb): Likewise.
18453         (altivec_vupklsh): Likewise.
18454         (altivec_vcmpequ<VI_char>_p): Likewise.
18455         (altivec_vcmpgts<VI_char>_p): Likewise.
18456         (altivec_vcmpgtu<VI_char>_p): Likewise.
18457         (abs<mode>2): Likewise.
18458         (vec_unpacks_hi_v16qi): Likewise.
18459         (vec_unpacks_hi_v8hi): Likewise.
18460         (vec_unpacks_lo_v16qi): Likewise.
18461         (vec_unpacks_hi_<VP_small_lc>): Likewise.
18462         (vec_unpacks_lo_v8hi): Likewise.
18463         (vec_unpacks_lo_<VP_small_lc>): Likewise.
18464         (vec_pack_trunc_v8h): Likewise.
18465         (vec_pack_trunc_v4si): Likewise.
18466         (vec_pack_trunc_<mode>): Likewise.
18468         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
18469         V2DI builtins.
18470         (vec_vmaxsd): Likewise.
18471         (vec_vmaxud): Likewise.
18472         (vec_vminsd): Likewise.
18473         (vec_vminud): Likewise.
18474         (vec_vpksdss): Likewise.
18475         (vec_vpksdus): Likewise.
18476         (vec_vpkudum): Likewise.
18477         (vec_vpkudus): Likewise.
18478         (vec_vrld): Likewise.
18479         (vec_vsld): Likewise.
18480         (vec_vsrad): Likewise.
18481         (vec_vsrd): Likewise.
18482         (vec_vsubudm): Likewise.
18483         (vec_vupkhsw): Likewise.
18484         (vec_vupklsw): Likewise.
18486 2013-05-29  Jan Hubicka  <jh@suse.cz>
18488         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
18489         flags; reorder rest of fields in more consistent way.
18490         (varpool_node): Remove analyzed, finalized and alias.
18491         (cgraph_ndoe): Likewise.
18492         (symtab_alias_ultimate_target): New function.
18493         (cgraph_function_node): Move offline.
18494         (cgraph_reset_node): Declare.
18495         (cgraph_comdat_can_be_unshared_p): Remove.
18496         (varpool_remove_initializer): Declare.
18497         (varpool_first_defined_variable, varpool_next_defined_variable
18498         cgraph_first_defined_function, cgraph_next_defined_function): Update.
18499         (cgraph_function_with_gimple_body_p): Update.
18500         (varpool_all_refs_explicit_p): Update.
18501         (symtab_alias_target): New function.
18502         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
18503         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
18504         (cgraph_function_or_thunk_node): Simplify using
18505         symtab_alias_ultimate_target.
18506         (varpool_variable_node): Likewise.
18507         * cgraph.c (cgraph_create_function_alias): Update.
18508         (cgraph_add_thunk): Update.
18509         (cgraph_remove_node): Update.
18510         (dump_cgraph_node): Do not dump removed flags.
18511         (cgraph_function_body_availability): Update.
18512         (cgraph_propagate_frequency): Update.
18513         (verify_cgraph_node): Check sanity of local flag.
18514         (cgraph_function_node): Move here from cgraph.h; revamp for
18515         cgraph_function_or_thunk_node.
18516         * lto-symtab.c (lto_varpool_replace_node): Update.
18517         (lto_symtab_resolve_can_prevail_p): Update.
18518         (lto_symtab_merge_cgraph_nodes): Update.
18519         * ipa-cp.c (determine_versionability, initialize_node_lattices,
18520         propagate_constants_accross_call, devirtualization_time_bonus,
18521         ipcp_propagate_stage): Update.
18522         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
18523         * ipa-inline-transform.c (clone_inlined_nodes,
18524         preserve_function_body_p): Update.
18525         * ipa-reference.c (propagate): Update.
18526         (write_node_summary_p): Update.
18527         * toplev.c (wrapup_global_declaration_2): Update.
18528         * cgraphunit.c (cgraph_analyze_function): Rename to ...
18529         (analyze_function) ... this one.
18530         (cgraph_process_new_functions): Update.
18531         (cgraph_reset_node): Export.
18532         (cgraph_finalize_function): Update.
18533         (cgraph_add_new_function): Update.
18534         (process_function_and_variable_attributes): Update.
18535         (varpool_finalize_decl): Update.
18536         (symbol_finalized): Remove.
18537         (symbol_finalized_and_needed): Rename to ...
18538         (symbol_defined_and_needed): ... update.
18539         (cgraph_analyze_functions): Update.
18540         (handle_alias_pairs): Update.
18541         (mark_functions_to_output): Update.
18542         (assemble_thunk): Update.
18543         (output_in_order): Update.
18544         (output_weakrefs): Update.
18545         (finalize_compilation_unit): Update.
18546         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
18547         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
18548         input_node, input_varpool_node): Update.
18549         * dbxout.c (dbxout_expand_expr): Update.
18550         * cgraphclones.c (cgraph_clone_node): Update.
18551         (cgraph_copy_node_for_versioning): Update.
18552         (cgraph_materialize_clone): Update.
18553         (cgraph_materialize_all_clones): Update.
18554         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
18555         propagate_pure_const, propagate_nothrow): Update.
18556         * lto-streamer-out.c (lto_output, write_symbol): Update.
18557         * ipa-utils.c (ipa_reverse_postorder): Update.
18558         * ipa-inline.c (can_inline_edge_p): Update.
18559         (update_caller_keys, ipa_inline): Update.
18560         * dwarf2out.c (reference_to_unused,
18561         premark_types_used_by_global_vars_helper): Update.
18562         * tree-eh.c (tree_could_trap_p): Update.
18563         * ipa-split.c (consider_split, execute_split_functions): Update.
18564         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
18565         has_addr_references_p): Update; move ahead in file for better
18566         readability.
18567         (process_references): Simplify.
18568         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
18569         bodies are removed.
18570         (cgraph_comdat_can_be_unshared_p): Make static.
18571         (cgraph_externally_visible_p): Update.
18572         (varpool_externally_visible_p): Update.
18573         (function_and_variable_visibility): Update.
18574         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
18575         ipa_tm_mark_force_output_node): Update.
18576         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
18577         estimate_edge_devirt_benefit, inline_generate_summary,
18578         inline_write_summary): Update.
18579         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
18580         * ipa-prop.c (ipa_compute_jump_functions): Update.
18581         (ipa_print_node_params, ipa_prop_read_section,
18582         ipa_update_after_lto_read, read_replacements_section): Update.
18583         * varasm.c (mark_decl_referenced): Update.
18584         (assemble_alias, dump_tm_clone_pairs): Update.
18585         * tree-inline.c (copy_bb): Update.
18586         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
18587         Update.
18588         * symtab.c (dump_symtab_base): Print new flags.
18589         (verify_symtab_base): Verify new flags.
18590         (symtab_alias_ultimate_target): New function.
18591         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
18592         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
18593         Update.
18594         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
18595         Update.
18596         * i386.c (ix86_get_function_versions_dispatcher,
18597         ix86_generate_version_dispatcher_body): Update.
18598         (fold_builtin_cpu): Use varpool_add_new_variable.
18599         * varpool.c (varpool_remove_initializer): Break out from ...
18600         (varpool_remove_node): ... this one.
18601         (dump_varpool_node, varpool_node_for_asm,
18602         cgraph_variable_initializer_availability, varpool_analyze_node,
18603         varpool_assemble_decl, varpool_remove_unreferenced_decls,
18604         varpool_finalize_named_section_flags, varpool_create_variable_alias):
18605         Update.
18607 2013-05-29  Jan Hubicka  <jh@suse.cz>
18609         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
18611 2013-05-29  Easwaran Raman  <eraman@google.com>
18613         PR tree-optimization/57442
18614         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
18615         when control exits the main loop.
18617 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
18619         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
18620         and RX600.
18621         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
18622         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
18623         * rx/t-rx: Add rx100 under multi library matches option for nofpu
18624         option.
18626 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18628         PR tree-optimization/57441
18629         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
18630         Don't limit size of incr_vec to number of candidates.
18632 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
18634         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
18635         and mips16 directories.
18636         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
18637         (MULTILIB_DIRNAMES): Ditto.
18638         (MULTILIB_EXCEPTIONS): Add new exceptions.
18639         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
18640         (MULTILIB_DIRNAMES): Ditto.
18641         (MULTILIB_EXCEPTIONS): Add new exceptions.
18643 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18644             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18646         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
18647         SYMBOL_TINY_ABSOLUTE.
18648         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
18649         SYMBOL_TINY_ABSOLUTE.
18650         (aarch64_expand_mov_immediate): Likewise.
18651         (aarch64_classify_symbol): Likewise.
18652         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
18653         Permit SYMBOL_TINY_ABSOLUTE.
18654         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
18656 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18657             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18659         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
18660         Refactor if/switch.  Replace gcc_assert with if.
18662 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
18664         * config/i386/i386.c (initial_ix86_tune_features): Enable
18665         FP Reassociation for AMD bdver1 and bdver2.
18667 2013-05-29  Martin Jambor  <mjambor@suse.cz>
18669         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
18670         and IMAGPART_EXPR do not occur within other handled_components.
18672 2013-05-29  Richard Biener  <rguenther@suse.de>
18674         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
18675         access on whether the use is in the BB we currently try to
18676         vectorize.
18677         (vect_bb_vectorization_profitable_p): Pass the BB we currently
18678         vectorize to vect_bb_slp_scalar_cost.
18680 2013-05-29  Richard Biener  <rguenther@suse.de>
18682         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
18683         computing scalar cost offsetted by stmts that are kept live
18684         by scalar uses.
18685         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
18686         for computation of scalar cost.
18688 2013-05-28  Steve Ellcey  <sellcey@mips.com>
18690         * config/mips/mips-cpus.def (mips32r2): Change processor type.
18692 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
18694         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
18695         array notation built-in reduction functions.
18696         * doc/passes.texi (Passes): Added documentation about changes done
18697         for Cilk Plus.
18698         * doc/invoke.texi (C Dialect Options): Added documentation about
18699         the -fcilkplus flag.
18700         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
18701         (BUILTINS_DEF): Depend on cilkplus.def.
18702         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
18703         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
18704         * cilkplus.def: New file.
18706 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
18708         PR rtl-optimization/57439
18709         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
18711 2013-05-28  Easwaran Raman  <eraman@google.com>
18713         PR tree-optimization/57337
18714         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
18715         (find_insert_point): Correctly identify the insertion point
18716         when two statements with the same UID is compared.
18718 2013-05-28  Richard Biener  <rguenther@suse.de>
18720         PR tree-optimization/56787
18721         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
18722         from the list of data references.
18723         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
18724         clobbers.
18725         (vect_analyze_loop_operations): Likewise.
18726         (vect_transform_loop): Remove clobbers.
18728 2013-05-28  Martin Jambor  <mjambor@suse.cz>
18730         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
18731         and REALPART_EXPRs have scalar type.
18733 2013-05-28  Richard Biener  <rguenther@suse.de>
18735         PR tree-optimization/57411
18736         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
18737         virtual operands.
18738         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
18739         virtual operand propagation.
18741 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18743         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
18744         destination register for bmasksi_vis.
18745         (vector_init_bshuffle): Likewise.
18746         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
18748 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18750         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
18751         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
18752         mode if the instruction isn't available in the original mode.
18753         * config/sparc/sparc.opt (mfix-ut699): New option.
18754         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
18755         (divdf3): Turn into expander.
18756         (divdf3_nofix): New insn.
18757         (divdf3_fix): Likewise.
18758         (divsf3): Disable if -mfix-ut699.
18759         (sqrtdf2): Turn into expander.
18760         (sqrtdf2_nofix): New insn.
18761         (sqrtdf2_fix): Likewise.
18762         (sqrtsf2): Disable if -mfix-ut699.
18764 2013-05-27  Richard Biener  <rguenther@suse.de>
18766         PR middle-end/57412
18767         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
18768         block for the new loop.
18770 2013-05-27  Richard Biener  <rguenther@suse.de>
18772         PR tree-optimization/57343
18773         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
18774         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
18775         (number_of_iterations_cond): Do not build the folded tree.
18777 2013-05-27  Richard Biener  <rguenther@suse.de>
18779         Revert
18780         PR middle-end/57381
18781         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
18782         OEP_CONSTANT_ADDRESS_OF retained.
18784         PR tree-optimization/57417
18785         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
18786         for unchanged base.
18787         (set_ssa_val_to): Compare addresses using
18788         get_addr_base_and_unit_offset.
18790 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
18792         PR rtl-optimization/56833
18793         * postreload.c (move2add_record_mode): New function.
18794         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
18795         (move2add_use_add2_insn): Use move2add_record_sym_value.
18796         (move2add_use_add3_insn): Likewise.
18797         (reload_cse_move2add): Use move2add_valid_value_p and
18798         move2add_record_mode.  Invalidate call-clobbered and REG_INC
18799         affected regs by setting reg_mode to VOIDmode.
18800         (move2add_note_store): Don't pretend the inside of a SUBREG is
18801         the actual destination.  Invalidate single/leading registers by
18802         setting reg_mode to VOIDmode.
18803         Use move2add_record_sym_value, move2add_valid_value_p and
18804         move2add_record_mode.
18806 2013-05-27  Richard Biener  <rguenther@suse.de>
18808         PR tree-optimization/57396
18809         * tree-affine.c (double_int_constant_multiple_p): Properly
18810         return false for val == 0 and div != 0.
18812 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18814         * config/mips/mips.h: Use #elif in preprocessor conditions.
18816 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18818         PR target/53916
18819         * config/mips/constraints.md (kl): New constraint.
18820         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
18821         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
18822         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
18823         from LO for MIPS16.
18824         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
18826 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
18828         PR target/55777
18829         * config/mips/mips.c (mips_can_inline_p): New function.
18830         (TARGET_CAN_INLINE_P): Define.
18832 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
18834         * sched-int.h (ds_t, dw_t): Make unsigned int.
18835         Fix documentation that describes how all the ds_t bits are used.
18836         Reserve the last bit for delayed-branch scheduling.
18837         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
18838         (BITS_PER_DEP_WEAK): Fix definition and documentation.
18839         (gen_dep_weak_1): Remove prototype.
18840         * sched-deps.c (get_dep_weak_1): Make static.
18841         * target.def (speculate_insn, needs_block_p, gen_spec_check,
18842         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
18843         * doc/tm.texi: Regenerate.
18844         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
18846 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
18848         PR debug/56950
18849         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
18851 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
18852             Sandra Loosemore  <sandra@codesourcery.com>
18854         * config.gcc (powerpc-*): Allow native for with-cpu.
18856 2013-05-24  Jeff Law  <law@redhat.com>
18858         PR tree-optimization/57124
18859         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
18860         conversion feeding a condition if the range has an overflow
18861         if -fstrict-overflow.  Add warnings for when we do make the
18862         transformation.
18864 2013-05-24  Dehao Chen  <dehao@google.com>
18866         * tree-cfg.c (locus_discrim_map): Fix the typo.
18867         (locus_discrim_hasher): Likewise.
18868         (locus_discrim_hasher::hash): Likewise.
18869         (locus_discrim_hasher::equal): Likewise.
18871 2013-05-24  Martin Jambor  <mjambor@suse.cz>
18873         PR tree-optimization/57294
18874         * cgraph.h (ipa_record_stmt_references): Declare.
18875         * cgraphbuild.c (ipa_record_stmt_references): New function.
18876         (build_cgraph_edges): Use ipa_record_stmt_references.
18877         (rebuild_cgraph_edges): Likewise.
18878         (cgraph_rebuild_references): Likewise.
18879         * ipa-prop.c (ipa_modify_call_arguments): Discard references
18880         associated with the old statement and build references from the
18881         newly built statements.
18882         * ipa-ref.c (ipa_remove_stmt_references): New function.
18883         * ipa-ref.h (ipa_remove_stmt_references): Declare.
18885 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
18887         * lra-constraints.c (emit_spill_move): Use smaller mode for
18888         mem-mem moves.
18889         (check_and_process_move): Consider mem-reg moves for secondary
18890         too.
18891         (curr_insn_transform): Don't lose insns emitted before for
18892         secondary memory moves.
18893         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
18894         reg set up in the current insn.
18896 2013-05-24  Dehao Chen  <dehao@google.com>
18898         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
18899         hash function.
18900         (locus_descrim_hasher::equal): Likewise.
18901         (build_gimple_cfg): New discriminator assignment algorithm.
18902         (make_edges): Likewise.
18903         (next_discriminator_for_locus): Likewise.
18904         (same_line_p): Likewise.
18905         (assign_discriminators): Likewise.
18906         (make_cond_expr_edges): Likewise.
18907         (make_gimple_switch_edges): Likewise.
18908         (make_goto_expr_edges): Likewise.
18909         (make_gimple_asm_edges): Likewise.
18911 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
18913         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
18914         X format specifier to only display bottom 16 bits.
18915         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
18916         immediate to match for operand 2, since it will be masked.
18918 2013-05-24  Richard Biener  <rguenther@suse.de>
18920         PR tree-optimization/57287
18921         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
18922         all SSA names that occur in abnormal PHIs.
18924 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
18926         PR tree-ssa/57385
18927         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
18928         that index is not negative.
18930 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
18932         PR rtl-optimization/55177
18933         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
18934         (simplify_byte_swapping_operation): New.
18935         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
18936         (simplify_relational_operation_1): Deal with BSWAP.
18938 2013-05-23  Richard Henderson  <rth@redhat.com>
18940         PR target/56742
18941         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
18942         (ix86_reorg): Call it.
18944 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
18946         PR target/57379
18947         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
18948         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
18949         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
18951 2013-05-23  Christian Bruel  <christian.bruel@st.com>
18953         PR debug/57351
18954         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
18956 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18957             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18959         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
18960         * config/aarch64/constraints.md (Usa): Remove.
18961         * doc/md.texi (AArch64 Usa): Remove.
18963 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18964             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18966         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
18967         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
18968         * config/aarch64/predicates.md (aarch64_const_address): Remove.
18969         (aarch64_mov_operand): Use aarch64_mov_operand_p.
18971 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
18973         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
18974         instruction (AdvSIMD).
18975         * config/aarch64/aarch64-builtins.c
18976         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
18977         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
18979 2013-05-23  Martin Jambor  <mjambor@suse.cz>
18981         PR middle-end/57347
18982         * tree.h (contains_bitfld_component_ref_p): Declare.
18983         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
18984         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
18985         caller.
18986         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
18987         not access a bit-field.  Assert all final offsets are byte-aligned.
18989 2013-05-23  Richard Biener  <rguenther@suse.de>
18991         PR tree-optimization/57380
18992         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
18993         least one invariant or re-used load.
18994         * passes.c (init_optimization_passes): Move pass_phiprop before
18995         pass_forwprop.
18997 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
18999         * config/aarch64/aarch64-simd.md
19000         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
19002 2013-05-23  Richard Biener  <rguenther@suse.de>
19004         PR middle-end/57381
19005         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19006         OEP_CONSTANT_ADDRESS_OF retained.
19008 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
19010         PR middle-end/57344
19011         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
19012         don't lower unit.  Handle unit not being always BITS_PER_WORD.
19014 2013-05-23  Richard Biener  <rguenther@suse.de>
19016         PR rtl-optimization/57341
19017         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
19018         instead of true_dependence.
19020 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
19022         * bb-reorder.c (branch_threshold): Make const.
19023         (exec_threshold): Ditto.
19025 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19026             Pat Haugen  <pthaugen@us.ibm.com>
19027             Peter Bergner  <bergner@vnet.ibm.com>
19029         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
19030         documentation for the power8 crypto builtins.
19032         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
19034         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
19035         macros for defining power8 builtin functions.
19036         (BU_P8V_AV_2): Likewise.
19037         (BU_P8V_AV_P): Likewise.
19038         (BU_P8V_VSX_1): Likewise.
19039         (BU_P8V_OVERLOAD_1): Likewise.
19040         (BU_P8V_OVERLOAD_2): Likewise.
19041         (BU_CRYPTO_1): Likewise.
19042         (BU_CRYPTO_2): Likewise.
19043         (BU_CRYPTO_3): Likewise.
19044         (BU_CRYPTO_OVERLOAD_1): Likewise.
19045         (BU_CRYPTO_OVERLOAD_2): Likewise.
19046         (XSCVSPDP): Fix typo, point to the correct instruction.
19047         (VCIPHER): Add power8 crypto builtins.
19048         (VCIPHERLAST): Likewise.
19049         (VNCIPHER): Likewise.
19050         (VNCIPHERLAST): Likewise.
19051         (VPMSUMB): Likewise.
19052         (VPMSUMH): Likewise.
19053         (VPMSUMW): Likewise.
19054         (VPERMXOR_V2DI): Likewise.
19055         (VPERMXOR_V4SI: Likewise.
19056         (VPERMXOR_V8HI: Likewise.
19057         (VPERMXOR_V16QI: Likewise.
19058         (VSHASIGMAW): Likewise.
19059         (VSHASIGMAD): Likewise.
19060         (VPMSUM): Likewise.
19061         (VPERMXOR): Likewise.
19062         (VSHASIGMA): Likewise.
19064         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
19065         __CRYPTO__ if the crypto instructions are available.
19066         (altivec_overloaded_builtins): Add support for overloaded power8
19067         builtins.
19069         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19070         support for power8 crypto builtins.
19071         (builtin_function_type): Likewise.
19072         (altivec_init_builtins): Add support for builtins that take vector
19073         long long (V2DI) arguments.
19075         * config/rs6000/crypto.md: New file, define power8 crypto
19076         instructions.
19078 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19079             Pat Haugen  <pthaugen@us.ibm.com>
19080             Peter Bergner  <bergner@vnet.ibm.com>
19082         * doc/invoke.texi (Option Summary): Add power8 options.
19083         (RS/6000 and PowerPC Options): Likewise.
19085         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
19086         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
19087         wm, wn, wr documentation.
19089         * config/rs6000/constraints.md (wm): New constraint for VSX
19090         registers if direct move instructions are enabled.
19091         (wn): New constraint for no registers.
19092         (wq): New constraint for quad word even GPR registers.
19093         (wr): New constraint if 64-bit instructions are enabled.
19094         (wv): New constraint if power8 vector instructions are enabled.
19095         (wQ): New constraint for quad word memory locations.
19097         * config/rs6000/predicates.md (const_0_to_15_operand): New
19098         constraint for 0..15 for crypto instructions.
19099         (gpc_reg_operand): If VSX allow registers in VSX registers as well
19100         as GPR and floating point registers.
19101         (int_reg_operand): New predicate to match only GPR registers.
19102         (base_reg_operand): New predicate to match base registers.
19103         (quad_int_reg_operand): New predicate to match even GPR registers
19104         for quad memory operations.
19105         (vsx_reg_or_cint_operand): New predicate to allow vector logical
19106         operations in both GPR and VSX registers.
19107         (quad_memory_operand): New predicate for quad memory operations.
19108         (reg_or_indexed_operand): New predicate for direct move support.
19110         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
19111         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
19112         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
19113         (POWERPC_MASKS): Add power8 options.
19114         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
19115         various options.
19117         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
19118         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
19120         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
19121         (-mpower8-fusion): New power8 options.
19122         (-mpower8-fusion-sign): Likewise.
19123         (-mpower8-vector): Likewise.
19124         (-mcrypto): Likewise.
19125         (-mdirect-move): Likewise.
19126         (-mquad-memory): Likewise.
19128         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
19129         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
19130         registers.
19131         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
19132         (rs6000_debug_vector_unit): Add p8_vector.
19133         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
19134         definitions.  Also print fusion state.
19135         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
19136         (rs6000_builtin_mask_calculate): Add power8 builtin support.
19137         (rs6000_option_override_internal): Add support for power8.
19138         (rs6000_common_init_builtins): Add debugging for skipped builtins
19139         if -mdebug=builtin.
19140         (rs6000_adjust_cost): Add power8 support.
19141         (rs6000_issue_rate): Likewise.
19142         (insn_must_be_first_in_group): Likewise.
19143         (insn_must_be_last_in_group): Likewise.
19144         (force_new_group): Likewise.
19145         (rs6000_register_move_cost): Likewise.
19146         (rs6000_opt_masks): Likewise.
19148         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
19149         power8 capable assembler, default to power7 options.
19150         (TARGET_DIRECT_MOVE): Likewise.
19151         (TARGET_CRYPTO): Likewise.
19152         (TARGET_P8_VECTOR): Likewise.
19153         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
19154         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
19155         (VECTOR_MEM_P8_VECTOR_P): Likewise.
19156         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
19157         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
19158         (TARGET_XSCVDPSPN): Likewise.
19159         (TARGET_XSCVSPDPN): Likewsie.
19160         (TARGET_SYNC_HI_QI): Likewise.
19161         (TARGET_SYNC_TI): Likewise.
19162         (MASK_CRYPTO): Likewise.
19163         (MASK_DIRECT_MOVE): Likewise.
19164         (MASK_P8_FUSION): Likewise.
19165         (MASK_P8_VECTOR): Likewise.
19166         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
19167         temporary used by some of the direct move instructions to get two FP
19168         temporary registers does not force creation of a stack frame.
19169         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
19170         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
19171         that any VSX registers are tieable, even if they are also an
19172         Altivec vector mode.
19173         (r6000_reg_class_enum): Add wm, wr, wv constraints.
19174         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
19175         (RS6000_BTM_CRYPTO): Likewise.
19176         (RS6000_BTM_COMMON): Likewise.
19178         * config/rs6000/rs6000.md (cpu attribute): Add power8.
19179         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
19180         (enum rs6000_vector): Add power8 vector support.
19182 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19184         PR target/19599
19185         PR target/57340
19186         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
19187         (any_sibcall_could_use_r3): this and handle indirect calls.
19188         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
19190 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19192         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
19194 2013-05-22  Richard Biener  <rguenther@suse.de>
19196         PR middle-end/57349
19197         * profile.c (branch_prob): Do not split blocks that are
19198         abnormally receiving from ECF_RETURNS_TWICE functions.
19200 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
19202         * recog.c (offsettable_address_addr_space_p): Fix calculation of
19203         address mode.  Move pointer mode initialization to the same place.
19205 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
19207         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
19208         while it has any effect.
19210 2013-05-21  Easwaran Raman  <eraman@google.com>
19212         PR tree-optimization/57322
19213         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
19214         UID of the statement added to the BB to be 1.
19216 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
19218         PR tree-optimization/57331
19219         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
19220         of conversion from pointer type to integral type with integer.
19222 2013-05-21  Martin Jambor  <mjambor@suse.cz>
19224         PR lto/57289
19225         * ipa-prop.c (ipa_read_node_info): Process param_used and
19226         controlled_uses in the same order as when writing.
19228 2013-05-21  Magnus Granberg  <baldrick@free.fr>
19230         PR plugins/56754
19231         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
19233 2013-05-21  Richard Biener  <rguenther@suse.de>
19235         PR tree-optimization/57318
19236         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
19237         estimate stmts with side-effects as likely eliminated.
19239 2013-05-21  Richard Biener  <rguenther@suse.de>
19241         PR tree-optimization/57330
19242         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
19243         preserve the call stmts fntype.
19245 2013-05-21  Richard Biener  <rguenther@suse.de>
19247         PR tree-optimization/57303
19248         * tree-ssa-sink.c (statement_sink_location): Improve killing
19249         stmt detection and properly handle self-assignments.
19251 2013-05-21  Christian Bruel  <christian.bruel@st.com>
19253         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
19254         spanning registers. LEAF_REG_REMAP is supported only for contiguous
19255         registers. Set register size out of the PARALLEL loop.
19257 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
19259         PR target/56547
19260         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
19261         (*fmasf4, *fmasf4_media): New insns.
19263 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19265         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
19266         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
19267         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
19268         (mips_idiv_insns): Update the comments to say that the returned
19269         instruction counts are in units of BASE_INSN_LENGTH.
19270         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
19271         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
19272         using 2 rather than 4 as the length of indirect MIPS16 and
19273         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
19274         length of a NOP.  Don't divide MIPS16 lengths by 2.
19275         (mips16_split_long_branches): Assume a branch is long if the
19276         length is greater than 4 rather than 8.
19277         * config/mips/mips.md (length): Give MIPS16 lengths directly,
19278         rather than multiplying them by 2.  Multiply instruction counts
19279         by BASE_INSN_LENGTH rather than 4.
19280         (*jump_mips16, tls_get_tp_mips16_<mode>)
19281         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
19283 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19285         * config/mips/mips.md (extended_mips16): Remove branch case.
19286         (length): Remove duplicated extended_mips16 test.
19288 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19290         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
19292 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
19294         * recog.h (Recog_data): Rename to...
19295         (recog_data_d): ...this.
19296         (recog_data): Update accordingly.
19297         * recog.c (recog_data): Likewise.
19298         * reload.c (save_recog_data): Likewise.
19299         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
19300         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
19302 2013-05-17  Julian Brown  <julian@codesourcery.com>
19304         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
19305         found in a REG_EQUAL note, invalidate it.
19307 2013-05-17  Easwaran Raman  <eraman@google.com>
19309         * tree-ssa-reassoc.c (find_insert_point): New function.
19310         (insert_stmt_after): Likewise.
19311         (get_def_stmt): Likewise.
19312         (ensure_ops_are_available): Likewise.
19313         (not_dominated_by): Likewise.
19314         (rewrite_expr_tree): Do not move statements beyond what is
19315         necessary. Remove call to swap_ops_for_binary_stmt...
19316         (reassociate_bb): ... and move it here.
19317         (build_and_add_sum): Assign UIDs for new statements.
19318         (linearize_expr): Likewise.
19319         (do_reassoc): Renumber gimple statement UIDs.
19321 2013-05-17  Jan Hubicka  <jh@suse.cz>
19323         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
19324         weakrefs.
19325         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
19326         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
19327         weakrefs.
19328         (output_weakrefs): Update.
19330 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
19331             Martin Jambor  <mjambor@suse.cz>
19333         PR middle-end/57276
19334         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
19335         value that corresponds to the given aggval is found in values vector.
19337 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
19339         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
19340         sse, sse2, sse3, ssse3 and sse4a flags to options.
19342 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
19344         * gengtype-state.c: (s_expr_writer): New class, to handle
19345         prettifying of output layout of s-expressions.
19346         (state_writer): New class, to write out gtype.state.
19347         (state_written_type_count): Move this variable into member data of
19348         state_writer.
19349         (s_expr_writer::s_expr_writer): New code: constructor for new class
19350         (state_writer::state_writer(): ditto
19351         (s_expr_writer::write_new_line): New function
19352         (s_expr_writer::write_any_indent): ditto
19353         (s_expr_writer::begin_s_expr): ditto
19354         (s_expr_writer::end_s_expr): ditto
19355         (write_state_fileloc): convert to method of state_writer...
19356         (state_writer:: write_state_fileloc): ...and use methods of
19357         s_expr_writer to write indentation into the gtype.state output file
19358         to visually represent the hierarchical structure of the list
19359         structures
19360         (write_state_fields): ditto, renaming to...
19361         (state_writer::write_state_fields)
19362         (write_state_a_string): ditto, renaming to...
19363         (state_writer::write_state_a_string)
19364         (write_state_string_option): ditto, renaming to...
19365         (state_writer::write_state_string_option)
19366         (write_state_type_option): ditto, renaming to...
19367         (state_writer::write_state_type_option)
19368         (write_state_nested_option): ditto, renaming to...
19369         (state_writer::write_state_nested_option)
19370         (write_state_option): ditto, renaming to...
19371         (state_writer::write_state_option)
19372         (write_state_options): ditto, renaming to...
19373         (state_writer::write_state_options)
19374         (write_state_lang_bitmap): ditto, renaming to...
19375         (state_writer::write_state_lang_bitmap)
19376         (write_state_version): ditto, renaming to...
19377         (state_writer::write_state_version)
19378         (write_state_scalar_type): ditto, renaming to...
19379         (state_writer::write_state_scalar_type)
19380         (write_state_string_type): ditto, renaming to...
19381         (state_writer::write_state_string_type)
19382         (write_state_undefined_type): ditto, renaming to...
19383         (state_writer::write_state_undefined_type)
19384         (write_state_struct_union_type): ditto, renaming to...
19385         (state_writer::write_state_struct_union_type)
19386         (write_state_struct_type): ditto, renaming to...
19387         (state_writer::write_state_struct_type)
19388         (write_state_user_struct_type): ditto, renaming to...
19389         (state_writer::write_state_user_struct_type)
19390         (write_state_lang_struct_type): ditto, renaming to...
19391         (state_writer::write_state_lang_struct_type)
19392         (write_state_param_struct_type): ditto, renaming to...
19393         (state_writer::write_state_param_struct_type)
19394         (write_state_pointer_type): ditto, renaming to...
19395         (state_writer::write_state_pointer_type)
19396         (write_state_array_type): ditto, renaming to...
19397         (state_writer::write_state_array_type)
19398         (write_state_gc_used): ditto, renaming to...
19399         (state_writer::write_state_gc_used)
19400         (write_state_common_type_content): ditto, renaming to...
19401         (state_writer::write_state_common_type_content)
19402         (write_state_type): ditto, renaming to...
19403         (state_writer::write_state_type)
19404         (write_state_pair_list): ditto, renaming to...
19405         (state_writer::write_state_pair_list)
19406         (write_state_pair): ditto, renaming to...
19407         (state_writer::write_state_pair)
19408         (write_state_typedefs): ditto, renaming to...
19409         (state_writer::write_state_typedefs)
19410         (write_state_structures): ditto, renaming to...
19411         (state_writer::write_state_structures)
19412         (write_state_param_structs): ditto, renaming to...
19413         (state_writer::write_state_param_structs)
19414         (write_state_variables): ditto, renaming to...
19415         (state_writer::write_state_variables)
19416         (write_state_srcdir): ditto, renaming to...
19417         (state_writer::write_state_srcdir)
19418         (write_state_files_list): ditto, renaming to...
19419         (state_writer::write_state_files_list)
19420         (write_state_languages): ditto, renaming to...
19421         (state_writer::write_state_languages)
19422         (write_state): create a state_writer instance and use it when
19423         writing out the state file
19425 2013-05-17  Mike Stump  <mikestump@comcast.net>
19427         PR rtl-optimization/57304
19428         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
19429         accessing DF_REF_REAL_LOC.
19431 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19433         PR rtl-optimization/57281
19434         PR rtl-optimization/57300
19435         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
19436         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
19437         what the other splitter did if the registers are dead.
19439 2013-05-17  Richard Biener  <rguenther@suse.de>
19441         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
19442         MEM_REF offsets.
19444 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19446         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
19447         linking.
19449 2013-05-17  Marek Polacek  <polacek@redhat.com>
19451         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
19452         length when doing non-zero store of storing '\0' to '\0'.
19454 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19456         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
19457         vect_external_def oprnd1 with loop_vinfo, try to emit
19458         optional cast, negation and and stmts on the loop preheader
19459         edge instead of into the pattern def seq.
19461         PR tree-optimization/57051
19462         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
19463         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
19465 2013-05-16  Nick Clifton  <nickc@redhat.com>
19467         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
19468         (rl78_is_naked_func): New function.
19469         (rl78_expand_prologue): Skip prologue generation for naked functions.
19470         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
19471         * doc/extend.texi (naked): Add RL78 to the list of processors
19472         that supports this attribute.
19474 2013-05-16  Jeff Law  <law@redhat.com>
19476         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
19478 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19480         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
19481         cache parameters using detect_caches_amd also for CYRIX,
19482         NSC and TM2 signatures.
19484 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19485             Dzianis Kahanovich  <mahatma@eu.by>
19487         PR target/45359
19488         PR target/46396
19489         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19490         VIA/Centaur processors and determine their cache parameters
19491         using detect_caches_amd.
19493 2013-05-16  Teresa Johnson  <tejohnson@google.com>
19495         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
19496         (rtl_verify_edges): New function.
19497         (rtl_verify_bb_insns): Ditto.
19498         (rtl_verify_bb_pointers): Ditto.
19499         (rtl_verify_bb_insn_chain): Ditto.
19500         (rtl_verify_fallthru): Ditto.
19501         (rtl_verify_bb_layout): Ditto.
19502         (rtl_verify_flow_info_1): Outline checks into new functions.
19503         (rtl_verify_flow_info): Ditto.
19505 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
19507         * cfghooks.c (copy_bbs): Add update_dominance argument.
19508         * cfghooks.h (copy_bbs): Update prototype.
19509         * tree-cfg.c (gimple_duplicate_sese_region):
19510         Add update_dominance argument.
19511         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
19512         * tree-ssa-loop-ch.c (copy_loop_headers): Update
19513         gimple_duplicate_sese_region call.
19514         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
19515         Update copy_bbs call.
19516         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
19517         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
19519 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19521         * tree-vectorizer.h (NUM_PATTERNS): Increment.
19522         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
19523         vect_recog_rotate_pattern.
19524         (vect_recog_rotate_pattern): New function.
19526 2013-05-16  Jason Merrill  <jason@redhat.com>
19528         * Makefile.in (LLINKER): New variable.
19529         (mostlyclean): Remove link mutex.
19530         * configure.ac: Handle --enable-link-mutex.
19531         * lock-and-run.sh: New script.
19533 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19535         PR target/19599
19536         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
19537         for NULL decl.
19539 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19541         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
19543 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
19545         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
19546         * config/arm/arm.c (next_consecutive_mem): New function.
19547         (gen_movmem_ldrd_strd): Likewise.
19548         * config/arm/arm.md (movmemqi): Update condition and code.
19549         (unaligned_loaddi, unaligned_storedi): New patterns.
19551 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19553         * config.gcc: Obsolete *-*-solaris2.9*.
19554         * doc/install.texi (Specific, *-*-solaris2*): Document it.
19556 2013-05-16  Richard Biener  <rguenther@suse.de>
19558         * passes.c (init_optimization_passes): Move pass_parallelize_loops
19559         earlier, after GRAPHITE transforms and IV canonicalization.
19561 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19563         * omp-low.c (extract_omp_for_data): For collapsed loops,
19564         if at least one of the loops is known at compile time to
19565         iterate zero times, set count to 0.
19566         (expand_omp_regimplify_p): New function.
19567         (expand_omp_for_generic): For collapsed loops, if at least
19568         one of the loops isn't known to iterate at least once,
19569         add runtime check with setting count to 0.
19570         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
19571         For unsigned types if it isn't known at compile time that
19572         the loop will iterate at least once, add runtime check to bypass
19573         the whole loop if initial condition isn't true.
19575 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
19577         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
19579 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
19581         PR middle-end/57286
19582         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
19583         transformations to avoid an infinite loop.
19585 2013-05-16  Marek Polacek  <polacek@redhat.com>
19587         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
19589 2013-05-15  Leif Ekblad  <leif@rdos.net>
19591         * config/i386/i386.c (ix86_decompose_address): Use
19592         DEFAULT_TLS_SEG_REG to access TLS segment register.
19593         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
19594         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
19595         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
19597 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
19599         PR target/57260
19600         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
19601         sibling calls to functions that would normally be lazily bound,
19602         unless $gp is call-clobbered.
19604 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
19606         * config/i386/i386.c (ix86_option_override_internal): Update
19607         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
19608         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
19609         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
19610         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
19611         of TARGET_3DNOW.
19612         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
19614 2013-05-15  Andreas Schwab  <schwab@suse.de>
19616         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
19617         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
19618         third operand.
19620 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19622         * loop-unroll.c (report_unroll_peel): Check decision before
19623         emitting unroll/peel message.
19625 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19627         * function.h (has_bb_partition): New rtl_data flag.
19628         (bb_reorder_complete): Ditto.
19629         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
19630         instead of flag_reorder_blocks_and_partition.
19631         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
19632         with some enhancements.
19633         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
19634         * bb-reorder.c (connect_traces): Check for has_bb_partition
19635         instead of flag_reorder_blocks_and_partition.
19636         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
19637         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
19638         verify_hot_cold_block_grouping.
19639         (partition_hot_cold_basic_blocks): Set has_bb_partition.
19641 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19643         PR target/19599
19644         * config/arm/predicates.md (call_insn_operand): New predicate.
19645         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
19646         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
19647         if insn is not a tail call.
19648         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
19649         registers.
19650         * config/arm/arm.h (enum reg_class): New caller save register class.
19651         (REG_CLASS_NAMES): Likewise.
19652         (REG_CLASS_CONTENTS): Likewise.
19653         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
19654         without decls.
19656 2013-05-15  Richard Biener  <rguenther@suse.de>
19658         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
19659         of MSG_OPTIMIZED_LOCATIONS.
19660         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
19661         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
19662         message.
19663         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
19664         of MSG_OPTIMIZED_LOCATIONS.
19665         (execute_vect_slp): Likewise.
19666         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
19667         (vect_create_cond_for_alias_checks): Likewise.
19668         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
19669         (vect_recog_widen_mult_pattern): Likewise.
19670         (vect_recog_widen_sum_pattern): Likewise.
19671         (vect_recog_over_widening_pattern): Likewise.
19672         (vect_recog_widen_shift_pattern): Likewise.
19673         (vect_recog_vector_vector_shift_pattern): Likewise.
19674         (vect_recog_divmod_pattern): Likewise.
19675         (vect_recog_mixed_size_cond_pattern): Likewise.
19676         (vect_recog_bool_pattern): Likewise.
19677         (vect_pattern_recog_1): Likewise.
19679 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19681         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
19682         non-functions to builtin_unreachable.
19683         * ipa-inline-transform.c (inline_call): Do not assert estimates were
19684         correct when new direct edges were discovered.
19686 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19688         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
19689         header, print symbol order instead of node uid, print more information
19690         about indirect edge targets.
19691         (ipa_make_edge_direct_to_target): Print symbol order instead of node
19692         uids.
19693         (ipa_make_edge_direct_to_target): Likewise.
19694         (remove_described_reference): Likewise.
19695         (propagate_controlled_uses): Likewise.
19696         (ipa_print_node_params): Also print symbol order.
19697         (ipcp_transform_function): Print symbol order instead of node uids.
19698         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
19699         (cgraph_get_create_real_symbol_node): Likewise.
19700         * ipa-cp.c (print_lattice): Likewise.
19701         (print_all_lattices): Likewise.
19702         (determine_versionability): Likewise.
19703         (initialize_node_lattices): Likewise.
19704         (estimate_local_effects): Likewise.
19705         (update_profiling_info): Likewise.
19706         (create_specialized_node): Likewise.
19707         (perhaps_add_new_callers): Likewise.
19708         (decide_about_value): Likewise.
19709         (decide_whether_version_node): Likewise.
19710         (identify_dead_nodes): Likewise.
19711         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
19712         (dump_inline_summary): Likewise.
19713         (estimate_node_size_and_time): Likewise.
19714         (inline_analyze_function): Likewise.
19715         * ipa-inline.c (report_inline_failed_reason): Likewise.
19716         (want_early_inline_function_p): Likewise.
19717         (edge_badness): Likewise.
19718         (update_edge_key): Likewise.
19719         (inline_small_functions): Likewise.  Add dumping of order to two other
19720         dumps.
19721         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
19722         instead of node uids.
19723         (propagate_pure_const): Likewise.
19724         (propagate_pure_const): Likewise.
19725         * ipa-utils.c (dump_cgraph_node_set): Likewise.
19726         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
19727         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
19728         of node uids.
19729         * tree-pretty-print.c (dump_function_header): Likewise.
19730         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
19731         Print symbol order instead of node uids.
19733 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19735         * config/s390/s390.c (s390_register_move_cost): Don't impose the
19736         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
19738 2013-05-15  Richard Biener  <rguenther@suse.de>
19740         PR tree-optimization/57275
19741         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
19742         return value for fail to do runtime alias checks for gather loads.
19744 2013-05-15  Jan Hubicka  <jh@suse.cz>
19746         PR lto/57038
19747         PR lto/47375
19748         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
19749         weakrefs are not external.
19750         (lto_symtab_merge_decls): Fix thinko when dealing with
19751         non-lto_symtab decls.
19752         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
19753         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
19754         * varpool.c (dump_varpool_node): Dump more flags.
19756 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19758         * config/i386/i386.c (processor_alias_table): Add instruction
19759         FSGSBASE for AMD bdver3 architecture.
19761 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19763         * tree.c (warn_deprecated_use): Print file:line using locus color.
19764         * diagnostic.c (diagnostic_report_current_module): Print file:line
19765         and file:line:column using locus color.
19767 2013-05-14  Mike Stump  <mikestump@comcast.net>
19769         * gdbinit.in: Add __null.
19771 2013-05-14  Mike Stump  <mikestump@comcast.net>
19773         * recog.h: Rename struct recog_data to Recog_data.
19774         * recog.c: Likewise.
19775         * reload.c (can_reload_into): Likewise.
19776         * config/picochip/picochip.c: Likewise.
19778 2013-05-14  Mike Stump  <mikestump@comcast.net>
19780         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
19782 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
19784         * resource.h (struct resources): Remove unch_memory member.
19785         (CLEAR_RESOURCE): Don't clear unch_memory.
19786         * resource.c (mark_referenced_resources): Don't set it.
19787         (mark_set_resources): Likewise.
19788         (mark_target_live_regs): Don't clear it.
19789         (init_resource_info): Likewise.
19790         * reorg.c (resource_conflicts_p): Don't compare it.
19791         (redundant_insn): Don't set it.
19793         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
19794         Remove prototypes.
19795         * emit-rtl.c (next_label): Remove unused function.
19796         (skip_consecutive_labels, link_cc0_insns): Move to ...
19797         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
19798         only place where these functions are used, and make them static.
19800 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
19802         * fold-const.c (fold_negate_expr): Handle vectors.
19803         (fold_truth_not_expr): Make it static.
19804         (fold_invert_truthvalue): New static function.
19805         (invert_truthvalue_loc): Handle vectors. Do not call
19806         fold_truth_not_expr directly.
19807         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
19808         <TRUTH_NOT_EXPR>: Do not cast to boolean.
19809         (fold_comparison): Handle vector constants.
19810         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
19811         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
19812         * tree.h (fold_truth_not_expr): Remove declaration.
19814 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
19816         * config/aarch64/aarch64-simd.md
19817         (aarch64_vcond_internal<mode>): Rename to...
19818         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
19819         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
19820         float modes. Clarify all iterator modes.
19821         (vcond<mode><mode>): Use new name for vcond expanders.
19822         (vcond<v_cmp_result><mode>): Likewise.
19823         (vcondu<mode><mode>: Likewise.
19824         * config/aarch64/iterators.md (VDQF_COND): New.
19826 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
19828         PR bootstrap/57266
19829         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
19830         variable for the shift amount. Check that we shift by non-negative
19831         amounts.
19833 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
19835         PR target/42017
19836         * config/arm/arm.h (EPILOGUE_USES): Only return true
19837         for LR_REGNUM after epilogue_completed.
19839 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
19841         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
19842         is error_mark_node.
19844 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19846         PR target/57261
19847         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
19848         and Solaris 11+/x86 with gld.
19849         * configure: Regenerate.
19851 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19853         * expmed.c (expand_shift_1): Canonicalize rotates by
19854         constant bitsize / 2 to bitsize - 1.
19855         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
19856         case ROTATERT>: Likewise.
19858         Revert:
19859         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
19861         * config/i386/i386.md (rotateinv): New code attr.
19862         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
19863         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
19864         roll $31, %eax, etc.
19866 2013-05-14  Richard Biener  <rguenther@suse.de>
19868         PR middle-end/57235
19869         * tree-eh.c (sink_clobbers): Give up for successors with
19870         multiple predecessors and no virtual uses.
19872 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
19874         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
19875         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
19877 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19879         PR middle-end/57251
19880         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
19881         the case when both op0 and op1 have VOIDmode.
19883 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
19885         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
19886         in multiply-accumulate mode.
19888 2013-05-13  Guozhi Wei  <carrot@google.com>
19890         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
19892 2013-05-13  Kai Tietz  <ktietz@redhat.com>
19894         PR target/56975
19895         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
19896         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
19897         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
19898         * config/i386/i386.c (ix86_option_override_internal): Likewise.
19899         (ix86_expand_prologue): Likewise.
19900         (ix86_expand_split_stack_prologue): Likewise.
19901         (legitimate_pic_address_disp_p): Likewise.
19902         (legitimize_pic_address): Likewise.
19903         (legitimize_tls_address): Likewise.
19904         (legitimize_pe_coff_symbol): Likewise.
19905         (output_pic_addr_const): Likewise.
19906         (construct_plt_address): Likewise.
19907         (ix86_expand_call): Likewise.
19908         (x86_output_mi_thunk): Likewise.
19909         (x86_function_profiler): Likewise.
19911 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
19913         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
19914         similar switch cases.
19915         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
19916         (aarch64_simd_mov_to_<mode>low): Delete.
19917         (aarch64_simd_mov_to_<mode>high): Delete.
19918         (move_lo_quad_<mode>): Add w<-r alternative.
19919         (aarch64_simd_move_hi_quad_<mode>): Likewise.
19920         (aarch64_simd_mov_from_*): Update type attribute.
19921         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
19922         statement.
19924 2013-05-13  Jan Hubicka  <jh@suse.cz>
19926         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
19927         * config/i386/i386.c (ix86_compute_frame_layout,
19928         ix86_expand_epilogue, emit_i387_cw_initialization,
19929         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
19930         ix86_local_alignment): Fix use of size/speed predicates.
19932 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
19934         PR tree-optimization/45216
19935         PR tree-optimization/57157
19936         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
19937         the (-Y) & (B - 1) variant if OP is |.
19938         * expmed.c (expand_shift_1): For rotations by const0_rtx just
19939         return shifted.  Use (-op1) & (prec - 1) as other_amount
19940         instead of prec - op1.
19942 2013-05-13  Martin Jambor  <mjambor@suse.cz>
19944         PR middle-end/42371
19945         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
19946         (ipa_constant_data): New type.
19947         (ipa_jump_func): Use ipa_constant_data to hold information about
19948         constant jump functions.
19949         (ipa_get_jf_constant): Adjust to jump function type changes.
19950         (ipa_get_jf_constant_rdesc): New function.
19951         (ipa_param_descriptor): New field controlled_uses.
19952         (ipa_get_controlled_uses): New function.
19953         (ipa_set_controlled_uses): Likewise.
19954         * ipa-ref.h (ipa_find_reference): Declare.
19955         * ipa-prop.c (ipa_cst_ref_desc): New type.
19956         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
19957         changes.
19958         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
19959         New parameter cs.  Adjust all callers.
19960         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
19961         (remove_described_reference): New function.
19962         (jfunc_rdesc_usable): Likewise.
19963         (try_make_edge_direct_simple_call): Decrement controlled use count,
19964         attempt to remove reference if it hits zero.
19965         (combine_controlled_uses_counters): New function.
19966         (propagate_controlled_uses): Likewise.
19967         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
19968         (ipa_edge_duplication_hook): Duplicate reference descriptions.
19969         (ipa_print_node_params): Print described use counter.
19970         (ipa_write_jump_function): Adjust to jump function type changes.
19971         (ipa_read_jump_function): New parameter CS, pass it to
19972         ipa_set_jf_constant.  Adjust caller.
19973         (ipa_write_node_info): Stream controlled use count
19974         (ipa_read_node_info): Likewise.
19975         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
19976         asserting.
19977         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
19978         count.  Remove cloning-added reference if it reaches zero.
19979         * ipa-ref.c (ipa_find_reference): New function.
19981 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19983         * config/i386/i386.c (processor_target_table): Modified default
19984         alignment values for AMD BD and BT architectures.
19986 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
19988         * tree-vect-generic.c (uniform_vector_p): Move ...
19989         * tree.c (uniform_vector_p): ... here.
19990         * tree.h (uniform_vector_p): Declare it.
19991         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
19992         into a scalar.
19994 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
19996         PR tree-optimization/57230
19997         * tree-ssa-strlen.c (handle_char_store): Record length for
19998         array store from STRING_CST.
20000         PR tree-optimization/57230
20001         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
20002         check.
20004 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
20006         * config/epiphany/epiphany.c (epiphany_init): Check size of
20007         NUM_MODES_FOR_MODE_SWITCHING.
20008         (epiphany_expand_prologue):
20009         Remove CONFIG_REGNUM initial value handling code.
20010         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
20011         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
20012         (emit_set_fp_mode, epiphany_mode_after): Likewise.
20013         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
20014         Don't return 1 for FP_MODE_NONE.
20015         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
20016         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
20017         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
20018         * config/epiphany/epiphany.md (save_config): New pattern.
20020 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
20022         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
20024 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
20026         * config/i386/i386.md (memory): Handle sseishft1.
20027         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
20028         (*vec_extractv2di_1): Ditto.
20030 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
20032         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
20033         saved registers.
20035 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
20037         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
20038         Add mthumb/march=armv7-a multilib.
20039         Add mthumb/march=armv7-r multilib.
20040         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
20042 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
20044         * config/v850/t-rtems: Add more multilibs.
20046 2013-05-10  Richard Biener  <rguenther@suse.de>
20048         PR tree-optimization/57214
20049         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
20050         not propagate from SSA names that occur in abnormal PHI nodes.
20052 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
20054         * stor-layout.c (element_precision): New function.
20055         * machmode.h (element_precision): Declare it.
20056         * tree.c (build_minus_one_cst): New function.
20057         (element_precision): Likewise.
20058         * tree.h (build_minus_one_cst): Declare new function.
20059         (element_precision): Likewise.
20060         * fold-const.c (operand_equal_p): Use element_precision.
20061         (fold_binary_loc): Handle vector types.
20062         * convert.c (convert_to_integer): Use element_precision.
20063         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
20064         separately.
20066 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
20068         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
20069         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
20070         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
20071         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20072         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
20073         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
20074         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
20075         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20076         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
20077         (Uuw8): New constraints.
20078         (Usb4): Move into alphabetical order.
20079         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
20080         (sd8_operand, ub8_operand, uw8_operand): New predicates.
20081         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
20082         previously unnamed patterns.
20083         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
20084         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
20085         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
20086         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
20087         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
20088         of specific lengths.
20090 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20092         * config/i386/i386.md (rotateinv): New code attr.
20093         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20094         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20095         roll $31, %eax, etc.
20097         PR tree-optimization/45216
20098         PR tree-optimization/57157
20099         * tree-ssa-forwprop.c (simplify_rotate): New function.
20100         (ssa_forward_propagate_and_combine): Call it.
20102 2013-05-10  Richard Biener  <rguenther@suse.de>
20104         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
20105         disable peeling when we version for aliasing.
20106         (vector_alignment_reachable_p): Honor explicit user alignment.
20107         (vect_supportable_dr_alignment): Likewise.
20108         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
20109         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
20110         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
20111         then peeling to arrange for the cost-model check to come first.
20113 2013-05-10  Alan Modra  <amodra@gmail.com>
20115         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
20116         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
20117         * configure: Regenerate.
20119 2013-05-10  Alan Modra  <amodra@gmail.com>
20121         PR target/55033
20122         * varasm.c (default_elf_select_section): Move !DECL_P check..
20123         (get_named_section): ..to here before calling get_section_name.
20124         Adjust assertion.
20125         (default_section_type_flags): Add DECL_P check.
20126         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
20127         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
20129 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
20131         * config/epiphany/epiphany.c (epiphany_expand_prologue):
20132         When using gen_stack_adjust_str with a register offset, add a
20133         REG_FRAME_RELATED_EXPR note.
20135 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
20137         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
20138         (*vec_extractv4si_zext_mem): Ditto.
20139         (*vec_extractv2di): Add 0->x and x->x alternatives.
20140         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
20141         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
20143 2013-05-09  Jason Merrill  <jason@redhat.com>
20145         N3639 C++1y VLA support
20146         * gimplify.c (gimplify_vla_decl): Don't touch an existing
20147         DECL_VALUE_EXPR.
20149         * tree.c (build_constructor_va): New.
20150         * tree.h: Declare it.
20152 2013-05-09  Martin Jambor  <mjambor@suse.cz>
20154         PR lto/57084
20155         * gimple-fold.c (canonicalize_constructor_val): Call
20156         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
20158 2013-05-09  Jan Hubicka  <jh@suse.cz>
20159             Richard Biener  <rguenther@suse.de>
20161         PR lto/54095
20162         * symtab.c (symtab_make_decl_local): Do not add private names.
20164 2013-05-09  Jan Hubicka  <jh@suse.cz>
20166         PR lto/54095
20167         * symtab.c (insert_to_assembler_name_hash): Handle clones.
20168         (unlink_from_assembler_name_hash): Likewise.
20169         (symtab_prevail_in_asm_name_hash, symtab_register_node,
20170         symtab_unregister_node, symtab_initialize_asm_name_hash,
20171         change_decl_assembler_name): Update.
20173 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
20175         * config/aarch64/aarch64.md: New movtf split.
20176         (*movtf_aarch64): Update.
20177         (aarch64_movdi_tilow): Handle TF modes and rename to
20178         aarch64_movdi_<mode>low.
20179         (aarch64_movdi_tihigh): Handle TF modes and rename to
20180         aarch64_movdi_<mode>high
20181         (aarch64_movtihigh_di): Handle TF modes and rename to
20182         aarch64_mov<mode>high_di
20183         (aarch64_movtilow_di): Handle TF modes and rename to
20184         aarch64_mov<mode>low_di
20185         (aarch64_movtilow_tilow): Remove spurious whitespace.
20186         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
20187         splits.
20188         (aarch64_print_operand): Update.
20190 2013-05-09  Alan Modra  <amodra@gmail.com>
20192         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
20193         powerpc64le.
20194         * configure: Regenerate.
20196 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20198         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
20199         splitter preparation statements.
20200         * config/i386/sse.md (*vec_extract* splitters): Ditto.
20201         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
20202         adjust_address_nv.
20204 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20206         * gimple-ssa-strength-reduction.c (count_candidates): Change
20207         return value to int.
20208         (analyze_candidates_and_replace): Change type of length to int.
20210 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20212         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
20213         (*vec_extract<mode>): Use VI12_128 mode iterator.
20214         (*vec_extract<mode>_mem): Ditto.
20215         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
20216         attribute.
20218 2013-05-08  Diego Novillo  <dnovillo@google.com>
20220         PR bootstrap/54659
20222         Revert:
20223         2012-08-17  Diego Novillo  <dnovillo@google.com>
20225         PR bootstrap/54281
20226         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
20227         * config.in: Regenerate.
20228         * configure: Regenerate.
20229         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
20231 2013-05-08  Jan Hubicka  <jh@suse.cz>
20233         PR lto/54095
20234         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
20235         * cgraph.h (symtab_node_base): Add unique_name.
20236         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
20237         input_overwrite_node, input_varpool_node): Stream unique_name.
20238         * cgraphclones.c (cgraph_create_virtual_clone,
20239         cgraph_function_versioning): Set unique_name.
20240         * ipa.c (function_and_variable_visibility): Set unique_name.
20242 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20244         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
20245         (alloc_cand_and_find_basis): Restrict conditional candidate
20246         processing to CAND_MULTs.
20248 2013-05-08  Jan Hubicka  <jh@suse.cz>
20250         PR lto/54095
20251         lto-symtab.c (lto_symtab_symbol_p): New function.
20252         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
20253         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
20254         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
20255         Skip static symbols.
20257 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
20259         PR tree-optimization/57200
20260         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
20261         Only call inform if the preceding warning_at returns true.
20263 2013-05-07  Han Shen  <shenhan@google.com>
20265         * cfgexpand.c (record_or_union_type_has_array_p): New function.
20266         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
20267         * common.opt (fstack-protector-strong): New option.
20268         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
20269         * doc/invoke.texi (Optimization Options): Document
20270         "-fstack-protector-strong".
20271         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
20273 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20275         * config/mips/mips.c (mips_machine_reorg2): Return 0.
20277 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
20279         * ira.c (update_equiv_regs): Add insn having equiv memory even if
20280         it is not lhs of the insn.
20281         (setup_reg_equiv): Remove insn having equiv memory which it is not
20282         lhs of the insn.
20283         * lra-constraints.c (process_address): Try to improve generation
20284         code for address base + disp.
20285         (lra_constraints): Make correct the code for checking insn setting
20286         up backward equivalence.  Remove insn only if it is in the init
20287         insn list.
20288         * lra-eliminations.c (update_reg_eliminate): Change return value.
20289         (lra_eliminate): Use the result.
20291 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
20293         * config/i386/sse.md (ssescalarnummask): New mode attribute.
20294         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
20295         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
20296         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
20297         register target operands.
20298         (*vec_extractv8hi_sse2): New pattern.
20299         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
20300         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
20301         (*vec_extract<mode>_mem): New insn and split pattern.
20303 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
20305         * config/arm/arm.c (arm_asan_shadow_offset): New function.
20306         (TARGET_ASAN_SHADOW_OFFSET): Define.
20307         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
20308         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
20310 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20312         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
20313         (incr_vec_index): Return -1 if increment not found.
20314         (create_add_on_incoming_edge): Assert if increment not found.
20315         (record_increment): Limit number of increments recorded.
20316         (all_phi_incrs_profitable): Return false if an increment not found.
20317         (replace_profitable_candidates): Don't process increments that were
20318         not recorded.
20319         (analyze_candidates_and_replace): Limit size of incr_vec.
20321 2013-05-07  Richard Biener  <rguenther@suse.de>
20323         * calls.c (special_function_p): setjmp-like functions are leaf.
20324         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
20325         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
20327 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
20329         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
20330         (aarch64_simd_mov<mode>): New expander.
20331         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
20332         (aarch64_simd_mov_to_<mode>high): Likewise.
20333         (aarch64_simd_mov_from_<mode>low): Likewise.
20334         (aarch64_simd_mov_from_<mode>high): Likewise.
20335         (aarch64_dup_lane<mode>): Update.
20336         (aarch64_dup_lanedi): New instruction pattern.
20337         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
20338         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
20340 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20342         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
20343         (replace_mult_candidate): Remove unnecessary argument; remove
20344         unnecessary parameter from call to introduce_cast_before_cand.
20345         (replace_unconditional_candidate): Remove unnecessary parameter
20346         from call to replace_mult_candidate.
20347         (replace_conditional_candidate): Likewise.
20348         (insert_initializers): Use make_temp_ssa_name.
20349         (introduce_cast_before_cand): Remove unnecessary argument; use
20350         make_temp_ssa_name.
20351         (replace_one_candidate): Remove unnecessary argument; remove
20352         unnecessary parameter from calls to introduce_cast_before_cand.
20353         (replace_profitable_candidates): Remove unnecessary parameters
20354         from calls to replace_one_candidate.
20356 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20358         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
20359         phi def as possibly hiding a basis for a CAND_ADD whose operands
20360         have been commuted in the analysis.
20361         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
20363 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20365         * config/aarch64/aarch64.md
20366         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
20367         shift value between 0-4.
20369 2013-05-07  Richard Biener  <rguenther@suse.de>
20371         * double-int.h (rshift): New overload.
20372         * double-int.c (rshift): New function.
20373         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
20374         (create_reference_ops_from_ref): Remove.
20375         (vn_reference_insert): Use shared ops for constructing the
20376         reference and copy it.
20378 2013-05-07  Richard Biener  <rguenther@suse.de>
20380         PR middle-end/57190
20381         * tree-eh.c (sink_clobbers): Properly propagate
20382         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
20384 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
20386         PR tree-optimization/57149
20387         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
20388         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
20389         collect_phi_def_edges, execute_late_warn_uninitialized): Use
20390         uninit_undefined_value_p instead of ssa_undefined_value_p.
20392         PR debug/57184
20393         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
20394         for modifier == EXPAND_INITIALIZER.
20396 2013-05-07  Anton Blanchard  <anton@samba.org>
20398         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
20399         for powerpc64 little endian.
20400         * configure: Regenerate.
20402 2013-05-06  Graham Stott  <grahams@btinternet.com>
20404         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
20405         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
20406         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
20407         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
20409 2013-05-06  Graham Stott  <grahams@btinternet.com>
20411         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
20412         codes which allow non-lvalues.
20414 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
20416         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
20417         components are all 1s.
20418         (integer_minus_onep): New function.
20419         * tree.h (integer_minus_onep): Declare it.
20420         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
20421         integer_minus_onep instead of integer_all_onesp.
20423 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20425         PR target/52933
20426         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
20427         variations of these patterns.
20429 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20431         * config/i386/i386.md (isa): Add x64_sse4 member.
20432         (enabled): Handle x64_sse4.
20433         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
20434         instruction for 64bit SSE4_1 targets.  Update insn attributes.
20435         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
20436         instruction for SSE4_1 targets.  Update insn attributes.
20437         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20438         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
20439         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
20440         const_1 selector.
20441         (*vec_extractv4si): Rename from *sse4_1_pextrd.
20442         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
20443         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
20445 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20447         PR target/57108
20448         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
20450 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
20452         * final.c (do_assembler_dialects): Don't handle curly braces and
20453         vertical bar escaped by % as dialect delimiters.
20454         (output_asm_insn): Print curly braces and vertical bar if escaped
20455         by % and ASSEMBLER_DIALECT defined.
20456         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
20457         * doc/tm.texi: Regenerated.
20459 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20461         * config/mips/mips.c: Include tree-pass.h.
20462         (mips_reorg): Split in pre- and post-dbr_schedule parts.
20463         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
20464         (pass_mips_machine_reorg2): New machine specific pass.
20465         (insert_pass_mips_machine_reorg2): New pass plugin definition.
20466         (mips_option_override): Register the new pass.
20467         * rtl.h (cleanup_barriers): Remove prototype.
20468         (dbr_schedule): Likewise.
20469         * jump.c (cleanup_barriers): Make static.
20470         * reorg.c (dbr_schedule): Likewise.
20472 2013-05-06  Richard Biener  <rguenther@suse.de>
20474         PR tree-optimization/57185
20475         * tree-parloops.c (add_field_for_reduction): Handle anonymous
20476         SSA names properly.
20478 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20480         PR target/57106
20481         * config/i386/i386.c (add_parameter_dependencies): Add dependence
20482         between "first_arg" and "insn", not "last" and "insn".
20484 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20486         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
20487         (find_candidates_in_block): Re-enable slsr_process_phi.
20488         (create_phi_basis): Fix double counting of candidate adjustment.
20490 2013-05-06  Richard Biener  <rguenther@suse.de>
20492         PR middle-end/57147
20493         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
20494         the edge is also fallthru, preserve it and just clear the
20495         abnormal flag.
20496         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
20497         also complex, preserve that and just clear the fallthru flag.
20498         * tree-inline.c (update_ssa_across_abnormal_edges): Also
20499         update virtual operands.
20501 2013-05-06  Alan Modra  <amodra@gmail.com>
20503         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
20504         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
20505         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
20506         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
20507         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20508         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
20510 2013-05-06  Alan Modra  <amodra@gmail.com>
20512         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
20513         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
20514         (DEFAULT_ASM_ENDIAN): Define.
20515         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20516         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
20517         Update -K PIC clause from sysv4.h.
20518         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
20519         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
20521 2013-05-06  Alan Modra  <amodra@gmail.com>
20523         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
20524         twice for little-endian.
20525         (ashrdi3_no_power, ashrdi3): Support little-endian.
20527 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20529         PR target/55303
20530         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
20531         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
20532         related expanders.
20533         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
20534         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
20535         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
20536         New predicates.
20538 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
20539             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20541         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
20542         * config/pa/pa.opt: Make mbig-switch a no-op.
20543         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
20544         (CASE_VECTOR_MODE): Always return SImode.
20545         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
20546         for the !TARGET_BIG_SWITCH case.
20547         * config/pa/pa-linux.h: Likewise.
20548         * config/pa/pa-openbsd.h: Likewise.
20549         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
20550         * config/pa/pa.md (short_jump): Remove define_insn.
20551         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
20552         (casesi0): Remove define_insn.
20553         (type): Remove btable_branch.
20554         (pa_combine_type): Likewise.
20555         (in_nullified_branch_delay): Likewise.
20556         (in_call_delay): Likewise.
20557         (define_delay): Likewise.
20558         (define_insn_reservation "Z3"): Likewise.
20559         (define_insn_reservation "Z4"): Likewise.
20560         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
20561         (pa_adjust_insn_length): Remove adjustment for btable branches.
20562         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
20563         and mno-big-switch
20565 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
20567         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20568         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
20569         Add m->r,x alternatives.
20570         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
20571         splitters using SWI48x mode iterator.
20572         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
20573         TARGET_64BIT.  Add m->x alternative.
20574         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
20575         Add o->x alternative.  Enable for TARGET_SSE.
20576         (sse_storeq): Remove expander.
20577         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
20578         with memory input operand.
20579         (*vec_extractv2di_1 splitter): New.
20580         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
20581         * config/i386/i386.md (ssevecmodelower): New mode attribute.
20583 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20585         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
20586         (INT_LOWPART): Delete.
20587         (extract_MB): Adjust.
20588         (extract_ME): Adjust.
20589         (print_operand): Adjust.
20591 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20593         * config/rs6000/predicates.md (reg_or_add_cint_operand,
20594         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20595         (reg_or_logical_cint_operand, easy_fp_constant,
20596         logical_const_operand): Delete "CONST_DOUBLE" case.
20597         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
20598         "HOST_BITS_PER_WIDE_INT == 64" test.
20599         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
20600         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
20601         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
20602         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
20603         test.
20604         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
20605         CONST_DOUBLE DImode/VOIDmode case.
20606         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
20607         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20608         CONST_DOUBLE VOIDmode case.
20609         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20610         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
20611         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
20612         Delete CONST_DOUBLE case.
20613         (splitters for mov FMOVE64 const_double): Delete
20614         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20615         "HOST_BITS_PER_WIDE_INT >= 64" test.
20616         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
20617         case.
20618         (mov DI const_double): Delete.
20620 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
20622         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
20623         on op shows all bits zero in mode of a lowpart subreg, return zero.
20625 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
20627         PR target/57150
20628         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
20629         to save TFmode registers and DImode to save TImode registers for
20630         caller save operations.
20631         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
20632         mark being partially clobbered since they only use the first
20633         double word.
20635         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
20636         and TDmode only use the upper 64-bits of each VSX register.
20638 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20640         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
20641         (find_candidates_in_block): Disable slsr_process_phi.
20643 2013-05-03  Guozhi Wei  <carrot@google.com>
20645         * coverage.c (coverage_obj_init): Move the construction of gcov
20646         constructor to ...
20647         (build_init_ctor): ... here.
20649 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20651         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
20652         (slsr_cand_d): Redefine def_phi.
20653         (stride_status, phi_adjust_status, count_phis_status): New enums.
20654         (find_phi_def): New.
20655         (find_basis_for_base_expr): New.
20656         (find_basis_for_candidate): Handle hidden bases.
20657         (alloc_cand_and_find_basis): Handle phi candidates.
20658         (slsr_process_phi): New.
20659         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
20660         (create_mul_imm_cand): Likewise.
20661         (create_add_ssa_cand): Exclude phi base candidates.
20662         (create_add_imm_cand): Likewise.
20663         (slsr_process_cast): Likewise.
20664         (slsr_process_copy): Likewise.
20665         (find_candidates_in_block): Handle phi candidates.
20666         (dump_candidate): Likewise.
20667         (unconditional_cands): Delete.
20668         (unconditional_cands_with_known_stride_p): Delete.
20669         (phi_dependent_cand_p): New.
20670         (cand_increment): Handle phi-dependent candidates.
20671         (replace_dependent): Delete.
20672         (replace_mult_candidate): New.
20673         (replace_unconditional_candidate): New.
20674         (incr_vec_index): Move to avoid forward reference.
20675         (create_add_on_incoming_edge): New.
20676         (create_phi_basis): New.
20677         (replace_dependents): Delete.
20678         (replace_conditional_candidate): New.
20679         (phi_add_costs): New.
20680         (replace_uncond_cands_and_profitable_phis): New.
20681         (record_increment): Handle phi adjustments.
20682         (record_phi_increments): New.
20683         (record_increments): Handle phi adjustments.
20684         (phi_incr_cost): New.
20685         (lowest_cost_path): Handle phis.
20686         (total_savings): Likewise.
20687         (analyze_increments): Likewise.
20688         (ncd_with_phi): New.
20689         (ncd_of_cand_and_phis): New.
20690         (nearest_common_dominator_for_cands): Handle phi increments.
20691         (all_phi_incrs_profitable): New.
20692         (replace_profitable_candidates): Handle phi-dependent candidates.
20693         (analyze_candidates_and_replace): Likewise.
20695 2013-05-03  Teresa Johnson  <tejohnson@google.com>
20697         PR bootstrap/57154
20698         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
20699         do not exceed REG_BR_PROB_BASE.
20701 2013-05-03  Jeff Law  <law@redhat.com>
20703         PR tree-optimization/57144
20704         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
20705         operand of the condition will bit into the new type when eliminating
20706         a cast feeding a condition.
20708 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
20710         PR rtl-optimization/57130
20711         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
20712         of COMPARE as in_code to the recursive call if needed.
20714 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
20716         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
20717         (enabled): Handle new members.
20718         * config/i386/sse.md (*vec_concatv2si): Merge from
20719         *vec_concatv2si_sse2 and vec_concatv2si_sse.
20720         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
20722 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
20724         PR tree-optimization/57027
20725         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
20726         for fnms opportunity, check we got the prerequisite kind
20727         of tree / gimple before using accessor functions.
20729 2013-05-03  Richard Biener  <rguenther@suse.de>
20731         * double-int.h (lshift): New overload without precision
20732         and arith argument.
20733         (operator *=, operator +=, operator -=): Move ...
20734         * double-int.c (operator *=, operator +=, operator -=): ... here
20735         and implement more efficiently.
20736         (mul_double_with_sign): Remove.
20737         (lshift_double): Adjust to take unsinged shift argument, push
20738         dispatching code to callers.
20739         (mul_double_wide_with_sign): Add early out for callers that
20740         are not interested in high parts or overflow.
20741         (lshift): New function.
20742         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
20743         dispatch code here.
20744         (lrotate, rrotate): Use logical shifts.
20745         * expr.c (get_inner_reference): Use lshift.
20746         * fixed-value.c (do_fixed_divide): Likewise.
20747         * tree-dfa.c (get_ref_base_and_extent): Likewise.
20748         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
20749         (indirect_refs_may_alias_p): Likewise.
20750         (stmt_kills_ref_p_1): Likewise.
20752 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20754         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
20756 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20758         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
20759         scalar form of FABD instruction.
20761 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
20763         * lra-constraints.c (process_alt_operands): Add checking alt
20764         number to choose the best alternative.
20766 2013-05-02  Richard Biener  <rguenther@suse.de>
20768         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
20769         bitmap and its handling.
20770         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
20772 2013-05-02  Richard Biener  <rguenther@suse.de>
20774         PR middle-end/57140
20775         * tree-inline.c (copy_loops): Properly handle removed loops.
20776         (copy_cfg_body): Mark destination loops for fixup if source
20777         loops needed fixup.
20779 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
20781         PR target/56732
20782         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
20783         generating simple_return for naked functions.
20785 2013-05-02  Martin Jambor  <mjambor@suse.cz>
20787         PR middle-end/56988
20788         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
20789         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
20790         flags match.
20791         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
20792         ipa_agg_replacement_value structures.
20793         (known_aggs_to_agg_replacement_list): Likewise.
20794         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
20795         (read_agg_replacement_chain): Likewise.
20796         (ipcp_transform_function): Also check that by_ref flags match.
20798 2013-05-02  Richard Biener  <rguenther@suse.de>
20800         * graphds.h (struct graph): Add obstack member.
20801         * graphds.c (new_graph): Initialize obstack and allocate
20802         vertices from it.
20803         (add_edge): Allocate edge from the obstack.
20804         (free_graph): Free the obstack instead of all edges and vertices.
20806 2013-05-02  Teresa Johnson  <tejohnson@google.com>
20808         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
20809         divides.
20810         * cfg.c (update_bb_profile_for_threading): Ditto.
20811         * tree-inline.c (copy_bb): Ditto.
20812         (copy_edges_for_bb): Ditto.
20813         (initialize_cfun): Ditto.
20814         (copy_cfg_body): Ditto.
20815         (expand_call_inline): Ditto.
20816         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
20817         (estimate_node_size_and_time): Ditto.
20818         (inline_merge_summary): Ditto.
20819         * cgraphclones.c (cgraph_clone_edge): Ditto.
20820         (cgraph_clone_node): Ditto.
20821         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20822         (compute_trg_info): Ditto.
20824 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
20826         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
20827         S reg when fp attribute set.
20828         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
20830 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
20832         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
20833         New pattern.
20834         (*and_one_cmplsi3_compare0_uxtw): Likewise.
20835         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
20836         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
20838 2013-05-02  Richard Biener  <rguenther@suse.de>
20840         * tree-scalar-evolution.c (scev_info_hasher): Remove.
20841         (struct instantiate_cache_entry): New type.
20842         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
20843         (struct instantiate_cache_type): New type.
20844         (set_instantiated_value, get_instantiated_value): Remove.
20845         (get_instantiated_value_entry): New function.
20846         (instantiate_scev_name): Use the new cache and adjust.
20847         (instantiate_scev_poly): Adjust.
20848         (instantiate_scev_binary): Likewise.
20849         (instantiate_array_ref): Likewise.
20850         (instantiate_scev_convert): Likewise.
20851         (instantiate_scev_not): Likewise.
20852         (instantiate_scev_3): Likewise.
20853         (instantiate_scev_2): Likewise.
20854         (instantiate_scev_r): Likewise.
20855         (instantiate_scev): Likewise.
20856         (resolve_mixers): Likewise.
20858 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
20860         PR target/57091
20861         * lra-constraints.c (best_small_class_operands_num): Remove.
20862         (process_alt_operands): Remove small_class_operands_num.  Take
20863         small classes operands into losers and only if the operand is not
20864         matched.  Modify debugging output.
20865         (curr_insn_transform): Remove best_small_class_operands_num.
20866         Print insn name.
20868 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20870         * config/aarch64/aarch64-builtins.c
20871         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
20872         * config/aarch64/aarch64-simd-builtins.def
20873         (reduc_splus_): Add new modes.
20874         (reduc_uplus_): New.
20875         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
20876         (reduc_uplus_v4sf): Likewise.
20877         (reduc_splus_v4sf): Likewise.
20878         (aarch64_addv<mode>): Likewise.
20879         (reduc_uplus_<mode>): Likewise.
20880         (reduc_splus_<mode>): Likewise.
20881         (aarch64_addvv2di): Likewise.
20882         (reduc_uplus_v2di): Likewise.
20883         (reduc_splus_v2di): Likewise.
20884         (aarch64_addvv2si): Likewise.
20885         (reduc_uplus_v2si): Likewise.
20886         (reduc_splus_v2si): Likewise.
20887         (reduc_<sur>plus_<mode>): New.
20888         (reduc_<sur>plus_v2di): Likewise.
20889         (reduc_<sur>plus_v2si): Likewise.
20890         (reduc_<sur>plus_v4sf): Likewise.
20891         (aarch64_addpv4sf): Likewise.
20892         * config/aarch64/arm_neon.h
20893         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
20894         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
20895         add UNSPEC_SADDV, UNSPEC_UADDV.
20896         (SUADDV): New.
20897         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
20899 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20901         * config/aarch64/arm_neon.h
20902         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
20904 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20906         * config/aarch64/aarch64-builtins
20907         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
20909 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20911         * config/aarch64/aarch64-simd-builtins.def
20912         (reduc_smax_): New.
20913         (reduc_smin_): Likewise.
20914         (reduc_umax_): Likewise.
20915         (reduc_umin_): Likewise.
20916         (reduc_smax_nan_): Likewise.
20917         (reduc_smin_nan_): Likewise.
20918         (fmax): Remove.
20919         (fmin): Likewise.
20920         (smax): Update for V2SF, V4SF and V2DF modes.
20921         (smin): Likewise.
20922         (smax_nan): New.
20923         (smin_nan): Likewise.
20924         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
20925         (<su><maxmin><mode>3): ...This, refactor.
20926         (s<maxmin><mode>3): New.
20927         (<maxmin_uns><mode>3): Likewise.
20928         (reduc_<maxmin_uns>_<mode>): Refactor.
20929         (reduc_<maxmin_uns>_v4sf): Likewise.
20930         (reduc_<maxmin_uns>_v2si): Likewise.
20931         (aarch64_<fmaxmin><mode>: Remove.
20932         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
20933         new builtin names.
20934         (vmin<q>_f<32,64>): Likewise.
20935         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
20936         (FMAXMIN): New.
20937         (su): Add mappings for smax, smin, umax, umin.
20938         (maxmin): New.
20939         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
20940         (FMAXMIN): Rename as...
20941         (FMAXMIN_UNS): ...This.
20942         (maxminv): Remove.
20943         (fmaxminv): Likewise.
20944         (fmaxmin): Likewise.
20945         (maxmin_uns): New.
20946         (maxmin_uns_op): Likewise.
20948 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20950         * config/aarch64/arm_neon.h
20951         (vac<ge, gt><sd>_f<32, 64>): Rename to...
20952         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
20953         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
20955 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20957         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
20958         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
20960 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20962         * config/aarch64/aarch64-simd.md
20963         (vcond<mode>_internal): Handle special cases for constant masks.
20964         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
20965         (vcondu<mode><mode>): Likewise.
20966         (vcond<v_cmp_result><mode>): New.
20968 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20970         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
20971         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
20972         * config/aarch64/aarch64-simd-builtins.def
20973         (cmeq): Update to BUILTIN_VALLDI.
20974         (cmgt): Likewise.
20975         (cmge): Likewise.
20976         (cmle): Likewise.
20977         (cmlt): Likewise.
20978         * config/aarch64/arm_neon.h
20979         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
20980         to builtins or C as appropriate.
20982 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
20984         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
20985         (cmgeu): ...This.
20986         (cmhi): Rename to...
20987         (cmgtu): ...This.
20988         * config/aarch64/aarch64-simd.md
20989         (simd_mode): Add SF.
20990         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
20991         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
20992         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
20993         (cstore<mode>_neg): ...This.
20994         * config/aarch64/iterators.md
20995         (VALLF): new.
20996         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
20997         (COMPARISONS): New.
20998         (UCOMPARISONS): Likewise.
20999         (optab): Add missing comparisons.
21000         (n_optab): New.
21001         (cmp_1): Likewise.
21002         (cmp_2): Likewise.
21003         (CMP): Likewise.
21004         (cmp): Remove.
21005         (VCMP_S): Likewise.
21006         (VCMP_U): Likewise.
21007         (V_cmp_result): Add DF, SF modes.
21008         (v_cmp_result): Likewise.
21009         (v): Likewise.
21010         (vmtype): Likewise.
21011         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
21013 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
21015         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
21016         define_insn to define_insn_and_split.
21017         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
21018         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
21019         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
21020         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
21021         (thumb2_negscc): Likewise.
21023 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21025         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
21027 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21029         * config/arm/thumb2.md: Remove trailing whitespaces.
21031 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
21033         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
21034         Use gen_int_mode rather than GEN_INT.
21036 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
21038         * value-prof.c (stream_in_histogram_value): Remove the strayed
21039         debug_gimple_stmt.
21041 2013-04-30  Richard Biener  <rguenther@suse.de>
21043         PR middle-end/57122
21044         * cfghooks.c (split_edge): Properly check for the loop latch edge.
21046 2013-04-30  Richard Biener  <rguenther@suse.de>
21048         PR middle-end/57107
21049         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
21051 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
21053         PR rtl-optimization/56957
21054         PR rtl-optimization/57105
21055         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
21056         variable.  Use just INSN_UID for determining whether an insn
21057         should be only disconnected from the insn stream.
21058         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
21060 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
21062         PR tree-optimization/57104
21063         * tsan.c (instrument_expr): Don't instrument accesses to
21064         DECL_HARD_REGISTER VAR_DECLs.
21066 2013-04-30  Richard Biener  <rguenther@suse.de>
21068         * function.h (loops_for_fn): New inline function.
21069         (set_loops_for_fn): Likewise.
21070         * cfgloop.h (place_new_loop): Add struct function parameter.
21071         (get_loop): Likewise.
21072         (get_loops): Likewise.
21073         (number_of_loops): Likewise.
21074         (fel_next): Adjust.
21075         (fel_init): Likewise.
21076         * cfg.c (get_loop_copy): Adjust.
21077         * cfgloop.c (flow_loops_dump): Likewise.
21078         (record_loop_exits): Likewise.
21079         (verify_loop_structure): Likewise.
21080         * cfgloopanal.c (mark_irreducible_loops): Likewise.
21081         (estimate_reg_pressure_cost): Likewise.
21082         (mark_loop_exit_edges): Likewise.
21083         * cfgloopmanip.c (place_new_loop): Likewise.
21084         (add_loop): Likewise.
21085         (duplicate_loop): Likewise.
21086         * graph.c (draw_cfg_nodes): Likewise.
21087         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
21088         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
21089         (extract_affine_chrec): Likewise.
21090         (build_scop_iteration_domain): Likewise.
21091         * graphite.c (graphite_initialize): Likewise.
21092         * ira-build.c (create_loop_tree_nodes): Likewise.
21093         (more_one_region_p): Likewise.
21094         (rebuild_regno_allocno_maps): Likewise.
21095         (mark_loops_for_removal): Likewise.
21096         (mark_all_loops_for_removal): Likewise.
21097         (remove_unnecessary_regions): Likewise.
21098         (ira_build): Likewise.
21099         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21100         * loop-init.c (fix_loop_structure): Likewise.
21101         (gate_rtl_move_loop_invariants): Likewise.
21102         (gate_rtl_unswitch): Likewise.
21103         (gate_rtl_unroll_and_peel_loops): Likewise.
21104         (rtl_doloop): Likewise.
21105         * lto-streamer-in.c (input_cfg): Likewise.
21106         * lto-streamer-out.c (output_cfg): Likewise.
21107         * modulo-sched.c (sms_schedule): Likewise.
21108         * predict.c (tree_estimate_probability): Likewise.
21109         (tree_estimate_probability_driver): Likewise.
21110         (estimate_loops): Likewise.
21111         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
21112         (move_sese_region_to_fn): Likewise.
21113         (debug_loop_num): Likewise.
21114         * tree-chrec.c (chrec_evaluate): Likewise.
21115         (hide_evolution_in_other_loops_than_loop): Likewise.
21116         (chrec_component_in_loop_num): Likewise.
21117         (reset_evolution_in_loop): Likewise.
21118         (evolution_function_is_invariant_rec_p): Likewise.
21119         * tree-if-conv.c (main_tree_if_conversion): Likewise.
21120         * tree-inline.c (copy_loops): Likewise.
21121         (copy_cfg_body): Likewise.
21122         (tree_function_versioning): Likewise.
21123         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
21124         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
21125         Likewise.
21126         (add_to_evolution_1): Likewise.
21127         (scev_const_prop): Likewise.
21128         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
21129         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
21130         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
21131         (tree_ssa_lim_initialize): Likewise.
21132         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
21133         (verify_loop_closed_ssa): Likewise.
21134         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
21135         (tree_ssa_loop_im): Likewise.
21136         (tree_ssa_loop_unswitch): Likewise.
21137         (tree_vectorize): Likewise.
21138         (check_data_deps): Likewise.
21139         (tree_ssa_loop_ivcanon): Likewise.
21140         (tree_ssa_loop_bounds): Likewise.
21141         (tree_complete_unroll): Likewise.
21142         (tree_complete_unroll_inner): Likewise.
21143         (tree_parallelize_loops): Likewise.
21144         (tree_ssa_loop_prefetch): Likewise.
21145         (tree_ssa_loop_ivopts): Likewise.
21146         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
21147         * tree-vectorizer.c (vectorize_loops): Likewise.
21149 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
21151         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
21152         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
21153         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
21154         with EABI_LINK_SPEC.
21156 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21158         PR target/44578
21159         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
21160         alternative.
21162 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
21164         PR target/57097
21165         * lra-constraints.c (process_alt_operands): Discourage a bit more
21166         using memory for pseudos.  Print cost dump for alternatives.
21167         Modify cost values for conflicts with early clobbers.
21168         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
21170 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21172         PR target/57098
21173         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
21175 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21177         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
21178         from/to S register.
21179         (movdi_aarch64): Support LDR/STR from/to D register.
21181 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21183         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
21184         or higher by default.
21186 2013-04-29  Richard Biener  <rguenther@suse.de>
21188         PR middle-end/57075
21189         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
21190         even if not adding abnormal edges for calls that can make
21191         abnormal gotos.
21193 2013-04-29  Richard Biener  <rguenther@suse.de>
21195         PR middle-end/57103
21196         * tree-cfg.c (move_stmt_op): Fix condition under which to update
21197         TREE_BLOCK.
21198         (move_stmt_r): Remove redundant checking.
21200 2013-04-29  Teresa Johnson  <tejohnson@google.com>
21202         PR bootstrap/57077
21203         * basic-block.h (apply_scale): New function.
21204         (apply_probability): Use apply_scale.
21205         * gimple-streamer-in.c (input_bb): Ditto.
21206         * lto-streamer-in.c (input_cfg): Ditto.
21207         * lto-cgraph.c (merge_profile_summaries): Ditto.
21208         * tree-optimize.c (execute_fixup_cfg): Ditto.
21209         * tree-inline.c (copy_bb): Update comment to use apply_scale.
21210         (copy_edges_for_bb): Ditto.
21211         (copy_cfg_body): Ditto.
21213 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21215         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
21216         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
21217         (tail_merge_optimize): Handle current_loops == NULL.
21219 2013-04-26  Jeff Law  <law@redhat.com>
21221         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
21222         (simplify_cond_using_ranges): Generalize code to simplify
21223         COND_EXPRs where one argument is a constant and the other
21224         is an SSA_NAME created by an integral type conversion.
21226 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21228         * config/arm/arm.md (store_minmaxsi): Use only when
21229         optimize_insn_for_size_p.
21231 2013-04-29  Christian Bruel  <christian.bruel@st.com>
21233         PR target/57108
21234         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
21236 2013-04-29  Richard Biener  <rguenther@suse.de>
21238         PR middle-end/57089
21239         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
21240         loop tree make sure to schedule a fixup for the child as well.
21241         (expand_omp_for_generic): Properly add loops.
21242         (expand_omp_for_static_nochunk): Likewise.
21243         (expand_omp_for_static_chunk): Likewise.
21244         (expand_omp_for): For the degenerate case fixup loops.
21245         (expand_omp_sections): Fix default bb placement in loops.
21246         (expand_omp_atomic_pipeline): Properly add loops.
21248 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21250         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
21252 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21254         * tree-ssa-tail-merge.c: Update header comment.
21256 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21258         * config/aarch64/arm_neon.h
21259         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
21260         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
21261         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
21262         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
21263         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
21264         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
21265         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
21266         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
21268 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21270         * config/aarch64/aarch64-simd.md
21271         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
21272         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
21273         fix_trunc, fixuns_trunc.
21274         (ftrunc<VDQF:mode>2): New.
21275         * config/aarch64/iterators.md (optab): Add fix, fixuns.
21276         (fix_trunc_optab): New.
21278 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21280         * config/aarch64/aarch64-builtins.c
21281         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
21282         iceilf, lround, iroundf.
21284 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21286         PR target/54349
21287         * config/i386/i386.h (enum ix86_tune_indices)
21288         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
21289         New, split from X86_TUNE_INTER_UNIT_MOVES.
21290         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
21291         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
21292         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
21293         (TARGET_INTER_UNIT_MOVES): Remove.
21294         * config/i386/i386.c (initial_ix86_tune_features): Update.
21295         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
21296         (ix86_expand_convert_uns_didf_sse): Use
21297         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21298         (ix86_expand_vector_init_one_nonzero): Ditto.
21299         (ix86_expand_vector_init_interleave): Ditto.
21300         (inline_secondary_memory_needed): Return true for moves from SSE class
21301         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
21302         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21303         * config/i386/constraints.md (Yi, Ym): Depend on
21304         TARGET_INTER_UNIT_MOVES_TO_VEC.
21305         (Yj, Yn): New constraints.
21306         * config/i386/i386.md (*movdi_internal): Change constraints of
21307         operand 1 from Yi to Yj and from Ym to Yn.
21308         (*movsi_internal): Ditto.
21309         (*movdf_internal): Ditto.
21310         (*movsf_internal): Ditto.
21311         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
21312         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21313         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
21314         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
21315         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
21316         * config/i386/sse.md (movdi_to_sse): Ditto.
21317         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
21318         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
21319         TARGET_INTER_UNIT_MOVES.
21320         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
21321         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
21322         instead of TARGET_INTER_UNIT_MOVES.
21323         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
21324         operand 1 from Yi to Yj and from Ym to Yn.
21326 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21328         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
21329         (float_truncate_hi_): Likewise.
21330         (float_extend_lo_): Likewise.
21331         (float_truncate_lo_): Likewise.
21332         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
21333         (aarch64_float_extend_lo_v2df): Likewise.
21334         (vec_unpacks_hi_v4sf): Likewise.
21335         (aarch64_float_truncate_lo_v2sf): Likewise.
21336         (aarch64_float_truncate_hi_v4sf): Likewise.
21337         (vec_pack_trunc_v2df): Likewise.
21338         (vec_pack_trunc_df): Likewise.
21340 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21342         * config/aarch64/aarch64-builtins.c
21343         (aarch64_fold_builtin): Fold float conversions.
21344         * config/aarch64/aarch64-simd-builtins.def
21345         (floatv2si, floatv4si, floatv2di): New.
21346         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
21347         * config/aarch64/aarch64-simd.md
21348         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
21349         * config/aarch64/iterators.md (FLOATUORS): New.
21350         (optab): Add float, floatuns.
21351         (su_optab): Likewise.
21353 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21355         * config/aarch64/aarch64-builtins.c
21356         (aarch64_builtin_vectorized_function): Use new names for
21357         fcvt builtins.
21358         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
21359         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
21360         (fcvtzu): Split as...
21361         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
21362         (fcvtas): Split as...
21363         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
21364         (fcvtau): Split as...
21365         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
21366         (fcvtps): Split as...
21367         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
21368         (fcvtpu): Split as...
21369         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
21370         (fcvtms): Split as...
21371         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
21372         (fcvtmu): Split as...
21373         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
21374         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
21375         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
21376         (lfrintnusf, lfrintnudf): Likewise.
21377         * config/aarch64/aarch64-simd.md
21378         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
21379         define_insn.
21380         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
21381         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
21382         (fcvt_pattern): Likewise.
21384 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21386         * config/aarch64/aarch64-simd.md
21387         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
21388         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
21390 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21392         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
21393         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
21394         (vrnd<a,m,n,p>_f32): Implement using builtins.
21395         (vrnd<i,x><q>_f<32, 64>): New.
21397 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21399         * config/aarch64/aarch64-builtins.c
21400         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
21401         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
21402         (frintz): Rename to...
21403         (btrunc): ...this.
21404         (frintp): Rename to...
21405         (ceil): ...this.
21406         (frintm): Rename to...
21407         (floor): ...this.
21408         (frinti): Rename to...
21409         (nearbyint): ...this.
21410         (frintx): Rename to...
21411         (rint): ...this.
21412         (frinta): Rename to...
21413         (round): ...this.
21414         * config/aarch64/aarch64-simd.md
21415         (aarch64_frint<frint_suffix><mode>): Delete.
21416         (<frint_pattern><mode>2): Convert to insn.
21417         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
21418         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
21419         (frint_pattern): Likewise.
21420         (frint_suffix): Likewise.
21422 2013-04-29  Richard Biener  <rguenther@suse.de>
21424         PR tree-optimization/57081
21425         * loop-init.c: Include tree-flow.h.
21426         (loop_optimizer_finalize): Free number of iteration estimates.
21427         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
21429 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
21431         PR tree-optimization/57083
21432         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
21433         non-singleton shift count range, zero extend low_bound for uns case.
21435         * config/i386/predicates.md (general_vector_operand): New predicate.
21436         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
21437         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
21438         if they aren't nonimmediate operands.  If their original values
21439         satisfy const_vector_equal_evenodd_p, don't shift them.
21440         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
21441         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
21442         if not nonimmediate_operand.
21443         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
21444         instead of register_operand.
21445         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
21447 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
21449         * stor-layout.c (finalize_size_functions): Allocate a structure and
21450         reset cfun before dumping the functions.
21452 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
21454         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
21456         PR target/56866
21457         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
21458         use xop_pmacsdqh if uns_p.
21459         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
21460         the immediate rotate count.
21462 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
21464         * rtl.h (struct rtx_def): Add comment for field jump.
21465         (LRA_SUBREG_P): New macro.
21466         * recog.c (register_operand): Check LRA_SUBREG_P.
21467         * lra.c (lra): Add note at the end of RTL code. Align non-empty
21468         stack frame.
21469         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
21470         (lra_final_code_change): Skip subreg change for operators.
21471         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
21472         if there are no operand changes.
21473         * lra-constraints.c (curr_insn_set): New.
21474         (match_reload): Set LRA_SUBREG_P.
21475         (emit_spill_move): Ditto.
21476         (check_and_process_move): Use curr_insn_set. Process only single
21477         set insns.  Don't initialize sec_mem_p and change_p.
21478         (simplify_operand_subreg): Use LRA_SUBREG_P.
21479         (reg_in_class_p): New function.
21480         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
21481         of #ifdef.  Add code to remove cycling.
21482         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
21483         non-null disp.  Reload inner instead of disp when base and index
21484         are null.  Try to put lo_sum into register.
21485         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
21486         (check_and_process_move): Move code for move cost check to
21487         simple_move_p.  Remove equiv_substitution.
21488         (simple_move_p): New function.
21489         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
21490         curr_insn_set.  Call check_and_process_move only for single set
21491         insns.  Use the new function.  Move call of check_and_process_move
21492         after operand equiv substitution and address process.
21494 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21496         PR go/57045
21497         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
21498         with nonlocal goto receivers or returns twice calls, ignore
21499         unininitialized values from abnormal edges to nl goto receiver
21500         or returns twice call.
21502 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21504         PR tree-optimization/57051
21505         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
21506         and VEC_RSHIFT_EXPR if shift count is a multiple of element
21507         bitsize.
21509 2013-04-26  Richard Biener  <rguenther@suse.de>
21511         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
21512         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
21513         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
21514         (fixup_loop_arrays_after_move): New function.
21515         (move_sese_region_to_fn): Properly outline the loop tree parts
21516         of the SESE region.
21518 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
21520         * config/i386/i386.md (type, unit): Fix long lines.
21522 2013-04-26  Richard Biener  <rguenther@suse.de>
21524         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
21525         (lto-streamer-out.o): Likewise.
21526         * cfgloop.c (init_loops_structure): Export, add struct function
21527         argument and adjust.
21528         (flow_loops_find): Adjust.
21529         * cfgloop.h (enum loop_estimation): Add EST_LAST.
21530         (init_loops_structure): Declare.
21531         * lto-streamer-in.c: Include cfgloop.h.
21532         (input_cfg): Input the loop tree.
21533         * lto-streamer-out.c: Include cfgloop.h.
21534         (output_cfg): Output the loop tree.
21535         (output_struct_function_base): Do not drop PROP_loops.
21537 2013-03-26  Richard Biener  <rguenther@suse.de>
21539         * tree-cfg.c (execute_build_cfg): Build the loop tree.
21540         (pass_build_cfg): Provide PROP_loops.
21541         (move_sese_region_to_fn): Remove loops that are outlined into fn
21542         for now.
21543         * tree-inline.c: Include cfgloop.h.
21544         (initialize_cfun): Do not drop PROP_loops.
21545         (copy_loops): New function.
21546         (copy_cfg_body): Copy loop structure.
21547         (tree_function_versioning): Initialize destination loop tree.
21548         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
21549         (pass_parallelize_loops): Do IL verification.
21550         * loop-init.c (loop_optimizer_init): Fixup loops if required.
21551         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
21552         the CFG make sure we fixup loops as well.
21553         * tree-ssa-tail-merge.c: Include cfgloop.h.
21554         (replace_block_by): When merging loop latches mark loops for fixup.
21555         * lto-streamer-out.c (output_struct_function_base): Drop
21556         PROP_loops for now.
21557         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
21558         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
21559         * ipa-split.c: Include cfgloop.h.
21560         (split_function): Add the new return block to the loop tree root.
21561         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
21562         whether we have removed the forwarder block.
21563         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
21564         * cfgloop.h (place_new_loop): Declare.
21565         * cfgloopmanip.c (place_new_loop): Export.
21566         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
21567         (tree-switch-conversion.o): Likewise.
21568         (tree-complex.o): Likewise.
21569         (tree-inline.o): Likewise.
21570         (tree-ssa-tailmerge.o): Likewise.
21571         (ipa-split.o): Likewise.
21572         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
21573         (tree-ssa-copy.o): Likewise.
21574         * tree-switch-conversion.c: Include cfgloop.h
21575         (process_switch): If we emit a bit-test cascade, schedule loops
21576         for fixup.
21577         * tree-complex.c: Include cfgloop.h.
21578         (expand_complex_div_wide): Properly add new basic-blocks to loops.
21579         * asan.c: Include cfgloop.h.
21580         (create_cond_insert_point): Properly add new basic-blocks to
21581         loops, schedule loop fixup.
21582         * cfgloop.c (verify_loop_structure): Check that looks are not
21583         marked for fixup.
21584         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
21585         to loops.
21586         (expand_omp_for_generic): Likewise.
21587         (expand_omp_sections): Likewise.
21588         (expand_omp_atomic_pipeline): Schedule loops for fixup.
21589         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
21590         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
21591         is initialized, not when loops are present.
21592         * tree-parloops.c (parallelize_loops): Remove checking here.
21593         * passes.c (init_optimization_passes): Schedule a copy-propagation
21594         pass before complete unrolling of inner loops.
21596 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21598         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
21599         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
21600         (colorize_init): Add argument to _WIN32 version.
21601         * toplev.c: Include diagnostic-color.h.
21602         (process_options): Default to -fdiagnostics-color=auto if
21603         GCC_COLORS env var is in the environment.
21604         * common.opt (fdiagnostics-color=): Add Var and Init.
21605         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
21606         env var is in the environment, the default is auto rather than never.
21608         * diagnostic.h (file_name_as_prefix): Add context argument.
21609         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
21610         the string as locus.
21611         * langhooks.c (lhd_print_error_function): Adjust caller.
21613 2013-04-25  Lawrence Crowl  <crowl@google.com>
21615         * var-tracking.c (shared_hash_def::htab):
21616         Change type to hash_table.  Update dependent calls and types.
21618 2013-04-25  Lawrence Crowl  <crowl@google.com>
21620         * Makefile.in: Update as needed below.
21622         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
21623         Move declaration to after the type's method definitons.
21625         * attribs.c (htab_t scoped_attributes::attribute_hash):
21626         Change type to hash_table.  Update dependent calls and types.
21628         * bitmap.c (htab_t bitmap_desc_hash):
21629         Change type to hash_table.  Update dependent calls and types.
21631         * cselib.c (htab_t cselib_hash_table):
21632         Change type to hash_table.  Update dependent calls and types.
21634         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
21635         (hash_string_slot_node): Move implementation into lto-streamer.h
21636         struct string_slot_hasher.
21637         (eq_string_slot_node): Likewise.
21639         * data-streamer-out.c: Update output_block::string_hash_table
21640         dependent calls and types.
21642         * dwarf2cfi.c (htab_t trace_index):
21643         Change type to hash_table.  Update dependent calls and types.
21645         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
21646         Change type to hash_table.  Update dependent calls and types.
21647         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
21648         (htab_t optimize_external_refs::map): Likewise.
21649         (htab_t output_comp_unit::extern_map): Likewise.
21650         (htab_t output_comdat_type_unit::extern_map): Likewise.
21651         (htab_t output_macinfo::macinfo_htab): Likewise.
21652         (htab_t optimize_location_lists::htab): Likewise.
21653         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
21655         * except.c (htab_t ehspec_hash_type):
21656         Change type to hash_table.  Update dependent calls and types.
21657         (assign_filter_values::ttypes): Likewise.
21658         (assign_filter_values::ehspec): Likewise.
21659         (sjlj_assign_call_site_values::ar_hash): Likewise.
21660         (convert_to_eh_region_ranges::ar_hash): Likewise.
21662         * gcse.c (htab_t pre_ldst_table):
21663         Change type to hash_table.  Update dependent calls and types.
21665         * ggc-common.c (htab_t saving_htab):
21666         Change type to hash_table.  Update dependent calls and types.
21667         (htab_t loc_hash): Likewise.
21668         (htab_t ptr_hash): Likewise.
21669         (call_count): Rename ggc_call_count.
21670         (call_alloc): Rename ggc_call_alloc.
21671         (loc_descriptor): Rename make_loc_descriptor.
21672         (add_statistics): Rename ggc_add_statistics.
21674         * ggc-common.c (saving_htab):
21675         Change type to hash_table.  Update dependent calls and types.
21677         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
21678         (push_gimplify_context): Likewise.
21679         (pop_gimplify_context): Likewise.
21680         (struct gimple_temp_hash_elt): Added.
21681         (struct gimplify_hasher): Likewise.
21682         (struct gimplify_ctx.temp_htab):
21683         Change type to hash_table.  Update dependent calls and types.
21685         * gimple-fold.c: Include gimplify-ctx.h.
21687         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
21688         Change type to hash_table.  Update dependent calls and types.
21689         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
21690         avoid potential global name collision.
21692         * gimplify.c: Include gimplify-ctx.h.
21693         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
21694         (htab_t gimplify_ctx::temp_htab):
21695         Update dependent calls and types for new type hash_table.
21696         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
21697         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
21699         * gimplify-ctx.h: New.
21700         (struct gimple_temp_hash_elt): Move from gimplify.c.
21701         (class gimplify_hasher): New.
21702         (struct gimplify_ctx): Move from gimple.h.
21703         (htab_t gimplify_ctx::temp_htab):
21704         Change type to hash_table.  Update dependent calls and types.
21706         * graphite-clast-to-gimple.c: Include graphite-htab.h.
21707         (htab_t ivs_params::newivs_index):
21708         Change type to hash_table.  Update dependent calls and types.
21709         (htab_t ivs_params::params_index): Likewise.
21710         (htab_t print_generated_program::params_index): Likewise.
21711         (htab_t gloog::newivs_index): Likewise.
21712         (htab_t gloog::params_index): Likewise.
21714         * graphite.c: Include graphite-htab.h.
21715         4htab_t graphite_transform_loops::bb_pbb_mapping):
21716         Change type to hash_table.  Update dependent calls and types.
21718         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
21719         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
21720         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
21722         * graphite-dependences.c: Include graphite-htab.h.
21723         (loop_is_parallel_p): Change hash table type of parameter.
21725         * graphite-htab.h: New.
21726         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
21727         (extern find_pbb_via_hash): Move from graphite-poly.h.
21728         (extern loop_is_parallel_p): Move from graphite-poly.h.
21729         (extern get_loop_body_pbbs): Move from graphite-poly.h.
21731         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
21732         (extern loop_is_parallel_p): Move to graphite-htab.h.
21733         (extern get_loop_body_pbbs): Move to graphite-htab.h.
21735         * haifa-sched.c (htab_t delay_htab):
21736         Change type to hash_table.  Update dependent calls and types.
21737         (htab_t delay_htab_i2): Likewise.
21739         * ira-color.c (htab_t allocno_hard_regs_htab):
21740         Change type to hash_table.  Update dependent calls and types.
21742         * ira-costs.c (htab_t cost_classes_htab):
21743         Change type to hash_table.  Update dependent calls and types.
21745         * loop-invariant.c (htab_t merge_identical_invariants::eq):
21746         Change type to hash_table.  Update dependent calls and types.
21748         * loop-iv.c (htab_t bivs):
21749         Change type to hash_table.  Update dependent calls and types.
21751         * loop-unroll.c (htab_t opt_info::insns_to_split):
21752         Change type to hash_table.  Update dependent calls and types.
21753         (htab_t opt_info::insns_with_var_to_expand): Likewise.
21755         * lto-streamer.h (struct string_slot): Move from data-streamer.h
21756         (struct string_slot_hasher): New.
21757         (htab_t output_block::string_hash_table):
21758         Change type to hash_table.  Update dependent calls and types.
21760         * lto-streamer-in.c (freeing_string_slot_hasher): New.
21761         (htab_t file_name_hash_table):
21762         Change type to hash_table.  Update dependent calls and types.
21764         * lto-streamer-out.c: Update output_block::string_hash_table dependent
21765         calls and types.
21767         * lto-streamer.c (htab_t tree_htab):
21768         Change type to hash_table.  Update dependent calls and types.
21770         * omp-low.c: Include gimplify-ctx.h.
21772         * passes.c (htab_t name_to_pass_map):
21773         Change type to hash_table.  Update dependent calls and types.
21774         (pass_traverse): Rename to passes_pass_traverse.
21776         * plugin.c (htab_t event_tab):
21777         Change type to hash_table.  Update dependent calls and types.
21779         * postreload-gcse.c (htab_t expr_table):
21780         Change type to hash_table.  Update dependent calls and types.
21781         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
21783         * sese.c (debug_rename_map_1): Make extern.
21784         (htab_t copy_bb_and_scalar_dependences::rename_map):
21785         Change type to hash_table.  Update dependent calls and types.
21787         * sese.h (extern debug_rename_map): Move to .c file.
21789         * store-motion.c (htab_t store_motion_mems_table):
21790         Change type to hash_table.  Update dependent calls and types.
21792         * trans-mem.c (htab_t tm_new_mem_hash):
21793         Change type to hash_table.  Update dependent calls and types.
21795         * tree-browser.c (htab_t TB_up_ht):
21796         Change type to hash_table.  Update dependent calls and types.
21798         * tree-cfg.c (htab_t discriminator_per_locus):
21799         Change type to hash_table.  Update dependent calls and types.
21801         * tree-complex.c: Include tree-hasher.h
21802         (htab_t complex_variable_components):
21803         Change type to hash_table.  Update dependent calls and types.
21805         * tree-eh.c (htab_t finally_tree):
21806         Change type to hash_table.  Update dependent calls and types.
21808         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
21809         struct int_tree_hasher.
21810         (extern int_tree_map_eq): Likewise.
21811         (uid_decl_map_hash): Removed.
21812         (extern decl_tree_map_eq): Likewise.
21814         * tree-hasher.h: New.
21815         (struct int_tree_hasher): New.
21816         (typedef int_tree_htab_type): New.
21818         * tree-inline.c: Include gimplify-ctx.h.
21820         * tree-mudflap.c: Include gimplify-ctx.h.
21822         * tree-parloops.c: Include tree-hasher.h.
21823         (htab_t eliminate_local_variables_stmt::decl_address):
21824         Change type to hash_table.  Update dependent calls and types.
21825         (htab_t separate_decls_in_region::decl_copies): Likewise.
21827         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
21828         Change type to hash_table.  Update dependent calls and types.
21830         * tree-sra.c (candidates):
21831         Change type to hash_table.  Update dependent calls and types.
21833         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
21834         in tree-flow.h.
21835         (int_tree_map_hash): Likewise.
21837         * tree-ssa-dom.c (htab_t avail_exprs):
21838         Change type to hash_table.  Update dependent calls and types.
21840         * tree-ssa-live.c (var_map_base_init::tree_to_index):
21841         Change type to hash_table.  Update dependent calls and types.
21843         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
21844         Change type to hash_table.  Update dependent calls and types.
21846         * tree-ssa-phiopt.c (seen_ssa_names):
21847         Change type to hash_table.  Update dependent calls and types.
21849         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
21850         Change type to hash_table.  Update dependent calls and types.
21852         * tree-ssa-uncprop.c (equiv):
21853         Change type to hash_table.  Update dependent calls and types.
21855 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
21857         PR rtl-optimization/57003
21858         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
21859         call note_stores with kill_clobbered_value callback again after
21860         killing regs_invalidated_by_call.
21862 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21864         * config/aarch64/aarch64-simd.md
21865         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
21866         (aarch64_simd_bsl<mode>): Likewise.
21867         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
21869 2013-04-25  Marek Polacek  <polacek@redhat.com>
21871         PR tree-optimization/57066
21872         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
21874 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21876         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
21878 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21880         * config/aarch64/aarch64-builtins.c
21881         (aarch64_fold_builtin): New.
21882         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
21883         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
21884         * config/aarch64/aarch64-simd-builtins.def (abs): New.
21885         * config/aarch64/arm_neon.h
21886         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
21888 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21889             Tejas Belagod  <tejas.belagod@arm.com>
21891         * config/aarch64/aarch64-builtins.c
21892         (aarch64_gimple_fold_builtin): New.
21893         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
21894         * config/aarch64/aarch64-simd-builtins.def (addv): New.
21895         * config/aarch64/aarch64-simd.md (addpv4sf): New.
21896         (addvv4sf): Update.
21897         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
21899 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21901         * config/aarch64/aarch64.md
21902         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
21904 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21906         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
21907         (*ngcsi_uxtw): New pattern.
21909 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21910             Julian Brown  <julian@codesourcery.com>
21912         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
21913         (TB_DREG): Add T_V4HF.
21914         (v4hf_UP): New macro.
21915         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21916         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21917         Handle initialisation of V4HF. Adjust initialisation of reinterpret
21918         built-ins.
21919         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
21920         (arm_vector_mode_supported_p): Handle V4HF.
21921         (arm_mangle_map): Handle V4HFmode.
21922         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
21923         * config/arm/arm_neon_builtins.def: Add entries for
21924         vcvtv4hfv4sf, vcvtv4sfv4hf.
21925         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
21926         (neon_vcvtv4hfv4sf): Likewise.
21927         * config/arm/neon-gen.ml: Handle half-precision floating point
21928         features.
21929         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
21930         * config/arm/arm_neon.h: Regenerate.
21931         * config/arm/neon.ml (type elts): Add F16.
21932         (type vectype): Add T_float16x4, T_floatHF.
21933         (type vecmode): Add V4HF.
21934         (type features): Add Requires_FP_bit feature.
21935         (elt_width): Handle F16.
21936         (elt_class): Likewise.
21937         (elt_of_class_width): Likewise.
21938         (mode_of_elt): Refactor.
21939         (type_for_elt): Handle F16, fix error messages.
21940         (vectype_size): Handle T_float16x4.
21941         (vcvt_sh): New function.
21942         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
21943         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
21944         (string_of_mode): Handle V4HF.
21945         * doc/arm-neon-intrinsics.texi: Regenerate.
21947 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
21949         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
21950         format specifier in 'X' case.
21952 2013-04-25  Alan Modra  <amodra@gmail.com>
21954         PR target/57052
21955         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
21956         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
21957         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
21958         Repeat for many other rotate/shift and mask patterns using subregs.
21959         Name lshiftrt insns.
21960         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
21961         on WORDS_BIG_ENDIAN.
21963 2013-04-25  Alan Modra  <amodra@gmail.com>
21965         * config.gcc: Support little-endian powerpc-linux targets.
21966         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
21967         (LINK_OS_LINUX_SPEC): Define.
21968         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
21969         Preserve MASK_LITTLE_ENDIAN.
21970         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
21971         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
21972         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
21973         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
21974         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
21975         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
21976         Correct fp word order for little-endian.  Don't shift toc entries
21977         smaller than a word for little-endian.
21978         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
21979         (bswapdi2 splits): Correct low-part subreg for little-endian.
21980         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
21981         low/high where such is correct only for be.
21982         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
21983         little-endian for -mcall-aixdesc.
21985 2013-04-25  Alan Modra  <amodra@gmail.com>
21987         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
21988         replace_equiv_address_nv.
21990 2013-04-25  Alan Modra  <amodra@gmail.com>
21992         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
21994 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21996         Revert:
21997         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
21998         * rtl.h (struct rtx_def): ...
22000 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22002         PR rtl-optimizations/57046
22003         * lra-constraints (split_reg): Set up lra_risky_transformations_p
22004         for multi-reg splits.
22006 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
22008         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
22010 2013-04-24  Sterling Augustine  <saugustine@google.com>
22012         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
22013         (comp_dir_string, debug_str_dwo_section): New.
22014         (DEBUG_STR_DWO_SECTION): Rename to ...
22015         (DEBUG_DWO_STR_SECTION): ... this.
22016         (DEBUG_NORM_STR_SECTION): Delete.
22017         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
22018         (DEBUG_STR_DWO_SECTION_FLAGS): New.
22019         (find_AT_string): Move most logic to ...
22020         (find_AT_string_in_table): ... here.  New.
22021         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
22022         add_skeleton_AT_string.  Delete logic.
22023         (output_skeleton_debug_sections): Remove call to
22024         add_top_level_skeleton_die_attrs.
22025         (add_comp_dir_attribute): Move logic to comp_dir_string.
22026         (dwarf2out_init): Initialize debug_str_dwo_section.
22027         (output_indirect_string): Call find_string_form.
22028         (output_indirect_strings): Rewrite.
22029         (prune_unused_types): Empty skeleton_debug_str_hash.
22030         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
22031         (dwarf2out_finish):  Call output_indirect_strings.
22033 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22035         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
22037 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22039         * rtl.h (struct rtx_def): Add comment for field jump.
22040         (LRA_SUBREG_P): New macro.
22041         * recog.c (register_operand): Check LRA_SUBREG_P.
22042         * lra.c (lra): Add note at the end of RTL code. Align non-empty
22043         stack frame.
22044         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
22045         (lra_final_code_change): Skip subreg change for operators.
22046         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
22047         if there are no operand changes.
22048         * lra-constraints.c (curr_insn_set): New.
22049         (match_reload): Set LRA_SUBREG_P.
22050         (emit_spill_move): Ditto.
22051         (check_and_process_move): Use curr_insn_set. Process only single
22052         set insns.  Don't initialize sec_mem_p and change_p.
22053         (simplify_operand_subreg): Use LRA_SUBREG_P.
22054         (reg_in_class_p): New function.
22055         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
22056         of #ifdef.  Add code to remove cycling.
22057         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
22058         non-null disp.  Reload inner instead of disp when base and index
22059         are null.  Try to put lo_sum into register.
22060         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
22061         (check_and_process_move): Move code for move cost check to
22062         simple_move_p.  Remove equiv_substitution.
22063         (simple_move_p): New function.
22064         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
22065         curr_insn_set.  Call check_and_process_move only for single set
22066         insns.  Use the new function.  Move call of check_and_process_move
22067         after operand equiv substitution and address process.
22069 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
22071         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
22072         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
22073         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
22075 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22077         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
22079 2013-04-24  Marek Polacek  <polacek@redhat.com>
22081         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
22082         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
22083         (select_loops_exit_conditions): Likewise.
22084         (number_of_iterations_for_all_loops): Likewise.
22085         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
22086         (scev_analysis): Likewise.
22088 2013-04-02  Catherine Moore  <clm@codesourcery.com>
22089             Chao-ying Fu  <fu@mips.com>
22091         * config/mips/micromips.md (jraddiusp): New pattern.
22092         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
22093         instruction if possible.
22095 2013-04-24  Alan Modra  <amodra@gmail.com>
22097         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
22099 2013-04-24  Julian Brown  <julian@codesourcery.com>
22100             Chung-Lin Tang  <cltang@codesourcery.com>
22102         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
22103         dependency behavior in enumeration type DIE generation. Add TODO note
22104         to comments about future DW_FORM_sdata/udata re-work of related code.
22106 2013-04-23  Lawrence Crowl  <crowl@google.com>
22108         * Makefile.in: Update as needed below.
22110         * hash-table.h (class hash_table):
22111         Correct many methods with parameter types compare_type to the correct
22112         value_type.  (Correct code was unlikely to notice the change.)
22113         (hash_table::elements_with_deleted) New.
22114         (class hashtable::iterator): New.
22115         (hashtable::begin()): New.
22116         (hashtable::end()): New.
22117         (FOR_EACH_HASH_TABLE_ELEMENT): New.
22119         * statistics.c (statistics_hashes):
22120         Change type to hash_table.  Update dependent calls and types.
22122         * tree-into-ssa.c (var_infos):
22123         Change type to hash_table.  Update dependent calls and types.
22125         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
22126         Change type to hash_table.  Update dependent calls and types.
22128         * tree-ssa-loop-im.c (struct mem_ref.refs):
22129         Change type to hash_table.  Update dependent calls and types.
22131         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
22132         Change type to hash_table.  Update dependent calls and types.
22134         * tree-ssa-sccvn.c (vn_tables_s::nary):
22135         Change type to hash_table.  Update dependent calls and types.
22136         (vn_tables_s::phis): Likewise.
22137         (vn_tables_s::references): Likewise.
22139         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
22140         (vn_reference_eq): Update parameter and return types.
22142         * tree-ssa-structalias.c (pointer_equiv_class_table):
22143         Change type to hash_table.  Update dependent calls and types.
22144         (location_equiv_class_table): Likewise.
22146         * tree-vect-data-refs.c: Consequential changes for making
22147         peeling a hash_table.
22149         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
22150         (destroy_loop_vec_info): Dependent hash_table update.
22152         * tree-vectorizer.h (peeling_htab):
22153         Change type to hash_table.  Update dependent calls and types.
22155 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
22157         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
22158         to check the register content is equal or not.
22159         * lra-constraints.c (match_reload): Use lra_assign_reg_val
22160         to assign register content record.
22161         * lra-eliminations.c (update_reg_eliminate): Use
22162         lra_update_reg_val_offset to update register content offset.
22163         * lra-int.h (struct lra_reg): Add offset member.
22164         (lra_reg_val_equal_p): New static inline function.
22165         (lra_update_reg_val_offset): New static inline function.
22166         (lra_assign_reg_val): New static inline function.
22167         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
22168         to assign register content record.
22169         (initialize_lra_reg_info_element): Initial offset to zero.
22171 2013-04-23  Catherine Moore  <clm@codesourcery.com>
22173         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
22174         operands.  Record compression.
22176 2013-04-23  Xinliang David Li  <davidxl@google.com>
22178         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
22180 2013-04-23  Richard Biener  <rguenther@suse.de>
22182         PR middle-end/57036
22183         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
22184         parameter, only add abnormal goto edges from the copied body
22185         if the call could perform abnormal gotos.
22186         (copy_cfg_body): Adjust.
22188 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
22190         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
22192 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
22194         * coretypes.h (gimple_stmt_iterator): Add struct to make
22195         compatible with C.
22197 2013-04-23  Richard Biener  <rguenther@suse.de>
22199         PR tree-optimization/57026
22200         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
22201         from SSA names occuring in abnormal PHI nodes.
22203 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22205         * lto/lto.c (print_lto_report_1): Fix LTO report names.
22207 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22209         * lto/lto.c (print_lto_report_1): Declare early.
22210         (read_cgraph_and_symbols): Call print_lto_report_1 early.
22212 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22214         * common.opt (-flto-report-wpa): Add.
22215         * doc/invoke.texi (-flto-report-wpa): Add.
22216         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
22217         (lto_main): dito.
22219 2013-04-22  Xinliang David Li  <davidxl@google.com>
22221         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
22222         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
22223         * Makefile.in: New dependency
22225             David Daney <ddaney.cavm@gmail.com>
22227         * configure.ac (gcc_cv_as_micromips_support):  Use the
22228         --fatal-warnings option.
22229         * configure: Regenerate.
22231 2013-04-22  Marek Polacek  <polacek@redhat.com>
22233         PR sanitizer/56990
22234         * tsan.c (instrument_expr): Don't instrument expression
22235         in case its size is zero.
22237 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
22239         PR target/57032
22240         Revert:
22241         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22243         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22245 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22247         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
22248         (gimple_stmt_iterator): New typedef.
22249         * gimple.h (gimple_stmt_iterator): Rename to...
22250         (gimple_stmt_iterator_d): ... This.
22251         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
22252         trees be valid for GIMPLE and GENERIC.
22253         (TARGET_GIMPLE_FOLD_BUILTIN): New.
22254         * gimple-fold.c (gimple_fold_call): Call target hook
22255         gimple_fold_builtin.
22256         * hooks.c (hook_bool_gsiptr_false): New.
22257         * hooks.h (hook_bool_gsiptr_false): New.
22258         * target.def (fold_stmt): New.
22259         * doc/tm.texi: Regenerate.
22261 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
22263         PR target/57018
22264         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
22265         a set sp if no stack realignment.
22267 2013-04-22  Nick Clifton  <nickc@redhat.com>
22269         * config.gcc (tilegx-linux): Extend extra_objs rather than
22270         overwriting it.
22271         (tilepro-linux): Likewise.
22273 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22275         * config/aarch64/aarch64-builtins.c
22276         (CF): Remove.
22277         (CF0, CF1, CF2, CF3, CF4, CF10): New.
22278         (VAR<1-12>): Add MAP parameter.
22279         (BUILTIN_*): Likewise.
22280         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
22281         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
22282         (aarch64_ushl_n<mode>): Likewise.
22283         (aarch64_sshr_n<mode>): Likewise.
22284         (aarch64_ushr_n<mode>): Likewise.
22285         (aarch64_<maxmin><mode>): Likewise.
22286         (aarch64_sqrt<mode>): Likewise.
22287         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
22288         (vshr<q>_n_*): Likewise.
22290 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22292         * config/aarch64/aarch64-builtins.c
22293         (aarch64_simd_builtin_type_mode): Handle SF types.
22294         (sf_UP): Define.
22295         (BUILTIN_GPF): Define.
22296         (aarch64_init_simd_builtins): Handle SF types.
22297         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
22298         (frecps): Likewise.
22299         (frecpx): Likewise.
22300         * config/aarch64/aarch64-simd.md
22301         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
22302         (aarch64_frecpe<mode>): New.
22303         (aarch64_frecps<mode>): Likewise.
22304         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
22305         (v8type): Add frecp<esx>.
22306         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
22307         (aarch64_frecps<mode>): Likewise.
22308         * config/aarch64/iterators.md (FRECP): New.
22309         (frecp_suffix): Likewise.
22310         * config/aarch64/arm_neon.h
22311         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
22313 2013-04-22  Christian Bruel  <christian.bruel@st.com>
22315         PR target/56995
22316         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
22317         (REG_CLASS_NAMES): Idem.
22318         (REG_CLASS_CONTENTS): Idem.
22319         (REGCLASS_HAS_FP_REG): Idem.
22320         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
22321         (sh_conditional_register_usage): Idem.
22323 2013-04-21  Jeff Law  <law@redhat.com>
22325         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
22326         (ssa_forward_propagate_and_combine): Use it.
22328 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22330         * lra.c: Update the flow chart diagram.
22332 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22334         PR rtl-optimization/56847
22335         * lra-constraints.c (process_alt_operands): Discourage alternative
22336         with non-matche doffsettable memory constraint fro memory with
22337         known offset.
22339 2013-04-19  Richard Biener  <rguenther@suse.de>
22341         PR tree-optimization/56982
22342         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
22343         function.
22344         * gimplify.c (gimplify_call_expr): Notice special calls.
22345         (gimplify_modify_expr): Likewise.
22346         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
22347         abnormal control flow receivers.
22348         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
22349         in the same way as cfun->has_nonlocal_labels.
22350         (gimple_purge_dead_abnormal_call_edges): Likewise.
22351         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
22352         receivers start a basic-block.
22354 2013-04-19  Richard Biener  <rguenther@suse.de>
22356         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
22357         member ...
22358         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
22359         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
22360         (SLP_TREE_LOAD_PERMUTATION): Add.
22361         (vect_transform_slp_perm_load): Adjust prototype.
22362         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
22363         (vect_free_slp_instance): Likewise.
22364         (vect_create_new_slp_node): Likewise.
22365         (vect_supported_slp_permutation_p): Remove.
22366         (vect_slp_rearrange_stmts): Adjust.
22367         (vect_supported_load_permutation_p): Likewise.  Inline
22368         vect_supported_slp_permutation_p here.
22369         (vect_analyze_slp_instance): Compute load permutations per
22370         slp node instead of per instance.
22371         (vect_get_slp_defs): Adjust.
22372         (vect_transform_slp_perm_load): Likewise.
22373         (vect_schedule_slp_instance): Remove redundant code.
22374         (vect_schedule_slp): Remove hack for PR56270, add it ...
22375         * tree-vect-stmts.c (vectorizable_load): ... here, do not
22376         CSE loads for SLP.  Adjust.
22378 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22380         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
22381         spelling in two comments.
22383 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22385         PR target/56797
22386         * config/arm/arm.c (load_multiple_sequence): Require SP
22387         as base register for loads if SP is in the register list.
22389 2013-04-19  Martin Jambor  <mjambor@suse.cz>
22391         PR tree-optimization/56718
22392         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
22393         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
22394         and made public.  Adjusted all callers.
22395         (ipa_intraprocedural_devirtualization): New function.
22396         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
22397         (ipa_intraprocedural_devirtualization): Likewise.
22398         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
22400 2013-04-19  Richard Biener  <rguenther@suse.de>
22402         PR tree-optimization/57000
22403         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
22405 2013-04-19  Terry Guo  <terry.guo@arm.com>
22407         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
22408         Replace with ...
22409         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
22410         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
22411         (cortex_m4_fmacs): Use new reservations.
22412         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
22414 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
22416         PR rtl-optimization/56999
22417         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
22418         related code.
22419         (lra_coalesce): Remove split_origin_bitmap and related code.
22420         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
22421         ranges if necessary.
22423 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
22425         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
22426         New array.
22427         (ix86_expand_call): Remove clobbered_registers array and use
22428         x86_64_ms_sysv_extra_clobbered_registers instead.
22429         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
22430         Declare here.
22431         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
22432         predicate.
22433         * config/i386/i386.md (*call_rex64_ms_sysv): Use
22434         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
22435         (*call_value_rex64_ms_sysv): Ditto.
22437 2013-04-18  Cary Coutant  <ccoutant@google.com>
22439         * dwarf2out.c (output_pubnames): Check die_perennial_p of
22440         parent instead of die_mark.
22442 2013-04-18  Diego Novillo  <dnovillo@google.com>
22444         * gimple.c (create_gimple_tmp): New.
22445         (get_expr_type): New.
22446         (build_assign): New.
22447         (build_type_cast): New.
22448         * gimple.h (enum ssa_mode): Define.
22449         (gimple_seq_set_location): New.
22450         * asan.c (build_check_stmt): Change some gimple_build_* calls
22451         to use build_assign and build_type_cast.
22453 2013-04-18  Richard Biener  <rguenther@suse.de>
22455         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
22456         handle negative step.  Remove redundant checks.
22457         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
22458         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
22459         for negative step and grouped loads fail to vectorize.
22461 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22463         * emit-rtl.c (reset_insn_used_flags): New function.
22464         (reset_all_used_flags): Use it.
22465         (verify_insn_sharing): New function.
22466         (verify_rtl_sharing): Fix verification for SEQUENCEs.
22468 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
22470         PR tree-optimization/56984
22471         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
22472         and (x >> M) >= N don't register any assertion if N << M is the
22473         minimum value.
22475 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22477         * lower-subreg.c (resolve_simple_move): If called self-recursive,
22478         do not delete_insn insns that have not yet been emitted, only
22479         unlink them with remove_insn.
22480         * df-scan.c (df_insn_delete): Revert r197492.
22482 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
22484         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
22485         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
22487 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22489         * config/arm/arm.md (movsicc_insn): Convert define_insn into
22490         define_insn_and_split.
22491         (and_scc,ior_scc,negscc): Likewise.
22492         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
22494 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22496         * config/arm/arm.c (use_return_insn): Return 0 for targets that
22497         can benefit from using a sequence of LDRD instructions in epilogue
22498         instead of a single LDM instruction.
22500 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
22502         PR 45688
22503         * doc/extend.texi: Fix typo.
22505 2013-04-17  Richard Biener  <rguenther@suse.de>
22507         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
22508         (vect_build_slp_tree): ... here.
22509         (vect_build_slp_tree_1): Compute which stmts of the SLP group
22510         match.  Remove special-casing of mismatched complex loads.
22511         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
22512         re-try the match with swapped commutative operands.
22513         (vect_supported_load_permutation_p): Remove special-casing of
22514         mismatched complex loads.
22515         (vect_analyze_slp_instance): Adjust.
22517 2013-04-17  Richard Biener  <rguenther@suse.de>
22519         PR rtl-optimization/56921
22520         * cfgloop.h (struct loop): Add simple_loop_desc member.
22521         (struct niter_desc): Mark with GTY(()).
22522         (simple_loop_desc): Do not use aux field but simple_loop_desc.
22523         * loop-iv.c (get_simple_loop_desc): Likewise.
22524         (free_simple_loop_desc): Likewise.
22526         Revert
22527         2013-04-16  Richard Biener  <rguenther@suse.de>
22529         PR rtl-optimization/56921
22530         * loop-init.c (pass_rtl_move_loop_invariants): Add
22531         TODO_do_not_ggc_collect to todo_flags_finish.
22532         (pass_rtl_unswitch): Same.
22533         (pass_rtl_unroll_and_peel_loops): Same.
22534         (pass_rtl_doloop): Same.
22536 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
22538         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
22539         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
22540         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
22541         references.
22542         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
22543         * tree-streamer.c (record_common_node): Adjust reference in comment.
22545 2013-04-17  Terry Guo  <terry.guo@arm.com>
22547         * config/arm/cortex-m4.md: Add a new bypass.
22549 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22551         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
22552         New pattern.
22553         (*subs_<optab><mode>_multp2): New pattern.
22554         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22555         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22557 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22559         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
22560         (*subs_mul_imm_<mode>): New pattern.
22562 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
22564         PR target/56948
22565         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
22566         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
22567         (vsx_movti_32bit): Change j->wa to O->wa.
22569 2013-04-16  Richard Biener  <rguenther@suse.de>
22571         PR rtl-optimization/56921
22572         * loop-init.c (pass_rtl_move_loop_invariants): Add
22573         TODO_do_not_ggc_collect to todo_flags_finish.
22574         (pass_rtl_unswitch): Same.
22575         (pass_rtl_unroll_and_peel_loops): Same.
22576         (pass_rtl_doloop): Same.
22578 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
22580         * config/arm/arm.c (emit_multi_reg_push): New declaration
22581         for an existing function.
22582         (arm_emit_strd_push): New function.
22583         (arm_expand_prologue): Used here.
22584         (arm_emit_ldrd_pop): New function.
22585         (arm_expand_epilogue): Used here.
22586         (arm_get_frame_offsets): Update condition.
22587         (arm_emit_multi_reg_pop): Add a special case for load of a single
22588         register with writeback.
22590 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
22592         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
22593         description.
22595 2013-04-16  Richard Biener  <rguenther@suse.de>
22597         PR tree-optimization/56756
22598         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
22599         (first_mem_ref_loc): New.
22600         (execute_sm): Place the load temporarily before a previous
22601         access instead of in the latch edge to ensure its SSA dependencies
22602         are defined at points dominating the load.
22604 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
22606         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
22607         correct fix by moving header and footer insn to the footer of
22608         the merged basic block.  Clear BB_END of the merged-away block.
22610         PR middle-end/43631
22611         * emit-rtl.c (make_note_raw): New function.
22612         (link_insn_into_chain): New static inline function.
22613         (add_insn): Use it.
22614         (add_insn_before, add_insn_after): Factor insn chain linking code...
22615         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
22616         using link_insn_into_chain.
22617         (note_outside_basic_block_p): New helper function for emit_note_after
22618         and emit_note_before.
22619         (emit_note_after): Use nobb variant of add_insn_after if the note
22620         should not be contained in a basic block.
22621         (emit_note_before): Use nobb variant of add_insn_before if the note
22622         should not be contained in a basic block.
22623         (emit_note_copy): Use make_note_raw.
22624         (emit_note): Likewise.
22625         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
22626         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
22627         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
22628         the moved barrier the tail of the basic block it follows.
22629         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
22631 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
22633         PR tree-optimization/56962
22634         * gimple-ssa-strength-reduction.c (record_increment): Only set
22635         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
22636         either rhs1 or rhs2 is equal to c->base_expr.
22638 2013-04-15  Richard Biener  <rguenther@suse.de>
22640         PR tree-optimization/56933
22641         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
22642         member.
22643         (GROUP_READ_WRITE_DEPENDENCE): Remove.
22644         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
22645         * tree-vect-data-refs.c (vect_analyze_group_access): Move
22646         dependence check ...
22647         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
22648         ... here.
22649         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
22650         GROUP_READ_WRITE_DEPENDENCE.
22652 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22654         * emit-rtl.c (reset_all_used_flags): New function.
22655         (verify_rtl_sharing): Call reset_all_used_flags before and after
22656         performing the checks.
22658 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22660         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
22661         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
22662         * config/arm/constraints.md (De): New constraint.
22663         * config/arm/neon.md (anddi3_neon): Delete.
22664         (neon_vand<mode>): Expand to standard anddi3 pattern.
22665         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
22666         Move earlier in the file.
22667         (neon_inv_logic_op2): Likewise.
22668         (arm_anddi_operand_neon): New predicate.
22670 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22672         * configure.ac (gcc_cv_ld_as_needed): Set
22673         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
22674         Use -z ignore, -z record on *-*-solaris2*.
22675         (HAVE_LD_AS_NEEDED): Update comment.
22676         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
22677         * configure: Regenerate.
22678         * config.in: Regenerate.
22679         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
22680         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
22681         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
22682         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
22683         equivalents.  Fix markup.
22684         * doc/tm.texi: Regenerate.
22686 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
22688         * config/i386/i386.opt: New option mstack-protector-guard=.
22689         * config/i386/i386-opts.h: Add enum stack_protector_guard.
22690         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
22691         TARGET_SSP_TLS_GUARD.
22692         * config/i386/i386.c (ix86_option_override_internal): Set
22693         ix86_stack_protector_guard.
22694         * config/i386/i386.md (stack_protect_set): Enable for
22695         TARGET_SSP_TLS_GUARD only.
22696         (stack_protect_set_<mode>): Ditto.
22697         (stack_protect_test): Ditto.
22698         (stack_protect_test_<mode>): Ditto.
22699         * doc/invoke.texi (i386 Option): Document.
22701 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
22703         PR target/56890
22704         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
22705         (S_MODES): Set H_MODE bit.
22706         (SF_MODES): Set only S_MODE and SF_MODE bits.
22707         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
22708         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
22709         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
22710         <MODE_FLOAT>: Likewise.
22712 2013-04-15  Joey Ye  <joey.ye@arm.com>
22714         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
22716 2013-04-15  Joey Ye  <joey.ye@arm.com>
22718         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
22719         for real far jump.
22720         (thumb_far_jump_used_p): Count instruction size and set
22721         far_jump_used.
22723 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
22725         * reorg.c (fill_simple_delay_slots): Reindent block of code.
22726         * resource.c (mark_target_live_regs): Reformat conditional block.
22728 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22730         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
22731         notes, they are emitted only just before final.
22732         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
22734 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22736         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
22737         * cfgrtl.c (delete_insn): Call it here instead.
22738         * lra-spills.c (lra_final_code_change): Use delete_insn.
22739         * haifa-sched.c (sched_remove_insn): Likewise.
22740         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
22741         returning to the nop pool.
22742         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
22743         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
22745 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22747         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
22748         * doc/tm.texi: Regenerated.
22750 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
22752         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
22753         QImode checks.
22755 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22757         * df-core.c (df_find_def): Compare register numbers.
22758         (df_find_use): Likewise.
22760 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
22762         PR target/56903
22763         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
22764         lra_in_progress for return.
22766 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
22768         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
22769         define_insn into define_insn_and_split and emit movsicc patterns.
22771 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
22773         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
22775 2013-04-12  Richard Biener  <rguenther@suse.de>
22777         * tree-pass.h (TODO_do_not_ggc_collect): New.
22778         * passes.c (execute_one_ipa_transform_pass): Honor
22779         TODO_do_not_ggc_collect.
22780         (execute_one_pass): Likewise.
22782         Revert
22783         2013-04-10  Richard Biener  <rguenther@suse.de>
22785         * passes.c (init_optimization_passes): Remove reload pass.
22786         * ira.c (do_reload): Merge into ...
22787         (ira): ... this.
22788         (rest_of_handle_reload): Remove.
22789         (pass_reload): Likewise.
22790         * config/i386/i386.c (ix86_option_override): Refer to ira instead
22791         of reload for vzeroupper pass placement.
22793 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
22795         PR tree-optimization/56918
22796         PR tree-optimization/56920
22797         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
22798         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
22799         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
22800         use wide_mul_with_sign method.
22802 2013-04-12  Richard Biener  <rguenther@suse.de>
22804         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
22805         not be considered a gimple constant.
22807 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
22809         * fold-const.c (const_binop): Handle vector shifts by a scalar.
22810         (fold_binary_loc): Call const_binop also for mixed vector-scalar
22811         operations.
22813 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
22814             Jakub Jelinek  <jakub@redhat.com>
22816         * opts.c: Include diagnostic-color.h.
22817         (common_handle_option): Handle OPT_fdiagnostics_color_.
22818         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
22819         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
22820         (diagnostic-color.o): New.
22821         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
22822         (diagnostic_color_rule): New enum.
22823         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
22824         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
22825         the location string.
22826         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
22827         either NULL, or color kind.
22828         * diagnostic-color.c: New file.
22829         * diagnostic-color.h: New file.
22830         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
22831         arguments.
22832         * doc/invoke.texi (-fdiagnostics-color): Document.
22833         * pretty-print.h (pp_show_color): Define.
22834         (struct pretty_print_info): Add show_color field.
22835         * diagnostic.c: Include diagnostic-color.h.
22836         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
22837         macros.  Colorize error:, warning: etc. strings and also the location
22838         string.
22839         (diagnostic_show_locus): Colorize the caret line.
22840         * pretty-print.c: Include diagnostic-color.h.
22841         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
22842         inside of %< %> quotes or quoted through q format modifier.
22844 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22846         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
22848 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22850         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
22851         code in CC_NZ mode.
22852         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
22853         pattern.
22855 2013-04-11  Marek Polacek  <polacek@redhat.com>
22857         PR tree-optimization/48184
22858         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
22860 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
22862         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
22863         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
22864         (skip_simple_arithmetic): Tidy up.
22865         * tree.h (skip_simple_constant_arithmetic): Declare.
22867 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22869         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
22871 2013-04-11  Richard Biener  <rguenther@suse.de>
22873         * tree-vect-loop.c (get_initial_def_for_induction): Properly
22874         generate vector constants.
22876 2013-04-11  Richard Biener  <rguenther@suse.de>
22878         PR tree-optimization/56878
22879         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
22880         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
22881         New function.
22882         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22883         Prefer to align the DR with the most invariant base address.
22885 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
22887         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
22888         comment.
22890 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
22892         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
22893         floating-point vector comparisons against 0.
22895 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
22897         PR tree-optimization/56899
22898         * fold-const.c (extract_muldiv_1): Apply distributive law
22899         only if TYPE_OVERFLOW_WRAPS (ctype).
22901 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
22903         PR target/56124
22904         * ira-costs.c (scan_one_insn): Check whether the source rtx of
22905         loading has side effect.
22907 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
22909         * config/sparc/sparc.c: Include tree-pass.h.
22910         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
22911         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
22912         head of file.  Change return type.  Split off gate function.
22913         (sparc_gate_work_around_errata): New function.
22914         (pass_work_around_errata): New pass definition.
22915         (insert_pass_work_around_errata) New pass insert definition to
22916         insert pass_work_around_errata just after delayed-branch scheduling.
22917         (sparc_option_override): Insert the pass.
22918         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
22920 2013-04-10  David S. Miller  <davem@davemloft.net>
22922         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
22923         or -mcpu=hypersparc.
22925         * target.def (cstore_mode): New hook.
22926         * target.h: Include insn-codes.h
22927         * targhooks.c: Likewise.
22928         (default_cstore_mode): New function.
22929         * targhooks.h: Declare it.
22930         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
22931         * doc/tm.texi: Rebuild.
22932         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
22933         target hook, rather than inspecting the insn_data.
22934         * config/sparc/sparc.c (sparc_cstore_mode): New function.
22935         (TARGET_CSTORE_MODE): Redefine.
22936         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
22937         result patterns.
22938         * config/sparc/predicates.md (cstore_result_operand): New special
22939         predicate.
22940         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
22941         Use it for operand 0.
22942         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
22943         (*snesi_special): Likewise.
22944         (*snesi_zero): Likewise.
22945         (*seqsi_zero): Likewise.
22946         (*sltu_insn): Likewise.
22947         (*sgeu_insn): Likewise.
22948         (*seqdi_special): Make operand 0 and comparison operation be of
22949         DImode.
22950         (*snedi_special): Likewise.
22951         (*snedi_special_vis3): Likewise.
22952         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
22953         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
22954         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
22955         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
22956         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
22957         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
22958         (*sltu_extend_sp64): Likewise.
22959         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
22960         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
22961         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
22962         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
22963         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
22965 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
22967         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
22968         (aarch64_start_file): Use the new function.
22970 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
22971             Jason Merrill  <jason@redhat.com>
22973         * common.opt: Add -gdwarf.
22974         * opts.c (common_handle_option): Handle it.
22975         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
22977 2013-04-10  Richard Biener  <rguenther@suse.de>
22979         * passes.c (execute_todo): Do not call ggc_collect conditional here.
22980         (execute_one_ipa_transform_pass): But unconditionally here.
22981         (execute_one_pass): And here.
22982         (init_optimization_passes): Remove reload pass.
22983         * tree-pass.h (TODO_ggc_collect): Remove.
22984         (pass_reload): Likewise.
22985         * ira.c (do_reload): Merge into ...
22986         (ira): ... this.
22987         (rest_of_handle_reload): Remove.
22988         (pass_reload): Likewise.
22989         * config/i386/i386.c (ix86_option_override): Refer to ira instead
22990         of reload for vzeroupper pass placement.
22991         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
22992         and todo_flags_finish of all passes.
22994 2013-04-10  Richard Biener  <rguenther@suse.de>
22996         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
22997         first_const_oprnd field, rename first_def_type to first_op_type.
22998         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
22999         (vect_get_and_check_slp_defs): Always use the type of the
23000         operand.  Allow mixed vect_external_def, vect_constant_def types.
23001         (vect_get_constant_vectors): Handle mixed vect_external_def,
23002         vect_constant_def types.
23004 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
23006         PR tree-optimization/55524
23007         * tree-ssa-math-opts.c
23008         (convert_mult_to_fma): Don't use an fms construct
23009         when we don't have an fms operation, but fnma, and it looks
23010         likely that we'll be able to use the latter.
23012 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
23014         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
23015         function.
23016         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
23017         inline fail caused by overwritable functions.
23019 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
23021         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
23022         unnecessary bits in the constant power of two case.
23024 2013-04-10  Richard Biener  <rguenther@suse.de>
23026         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
23027         broken code swapping operands.
23028         (vect_build_slp_tree): Do not compute load permutations here.
23029         (vect_analyze_slp_instance): Compute load permutations here,
23030         after building the SLP tree.
23032 2013-04-09  Christian Bruel  <christian.bruel@st.com>
23034         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
23035         of next/prev_real_insn.
23037 2013-04-09  Jan Hubicka  <jh@suse.cz>
23039         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
23040         Drop aliased parameter.
23041         (function_and_variable_visibility): Do not handle alias pairs.
23042         * cgraph.c (varpool_externally_visible_p): Update prototype.
23043         * varpool.c (varpool_add_new_variable): Update.
23045 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23047         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
23049 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
23051         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
23053         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
23055 2013-04-09  Marek Polacek  <polacek@redhat.com>
23057         PR tree-optimization/48762
23058         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
23060 2013-04-09  Richard Biener  <rguenther@suse.de>
23062         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
23063         dealing with cost.
23064         (vect_build_slp_tree): Likewise.
23065         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
23066         calculating the cost of a SLP instance.
23067         (vect_analyze_slp_instance): Use it from here, after building
23068         the SLP tree.
23070 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23072         PR middle-end/56883
23073         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
23074         expand_omp_for_static_chunk): Use simple_p = true in
23075         force_gimple_operand_gsi calls when assigning to addressable decls.
23077 2013-04-09  Jeff Law  <law@redhat.com>
23079         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
23080         when the boolean was created by converting a wider object which
23081         had a boolean range.
23083 2013-04-09  Richard Biener  <rguenther@suse.de>
23085         * tree-vectorizer.h (slp_void_p): Remove.
23086         (slp_tree): Typedef before _slp_tree declaration.
23087         (struct _slp_tree): Use a vector of slp_tree as children.
23088         (vect_get_place_in_interleaving_chain): Remove.
23089         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
23090         Move ...
23091         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
23092         and make static.
23093         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
23094         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
23095         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
23096         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
23097         Use slp_node instead of slp_void_p and adjust.
23099 2013-04-09  Richard Biener  <rguenther@suse.de>
23101         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
23102         work that is not necessary.
23104 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23106         PR tree-optimization/56854
23107         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
23108         forward into clobber stmts if it would change MEM_REF lhs into
23109         non-MEM_REF.
23111 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
23113         * tree.c (type_hash_lookup, type_hash_add): Make static.
23114         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
23116 2013-04-09  Richard Biener  <rguenther@suse.de>
23118         * tree.h (unsave_expr_now): Remove.
23119         * tree-inline.c (mark_local_for_remap_r): Remove.
23120         (unsave_expr_1): Likewise.
23121         (unsave_r): Likewise.
23122         (unsave_expr_now): Likewise.
23123         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
23124         (propagate_tree_value): Likewise.
23126 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
23128         * doc/rtl.texi (sequence): Rewrite documentation to match the
23129         current use of SEQUENCE rtl objects.
23130         * rtl.def (SEQUENCE): Likewise.
23132         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
23133         Update documentation.
23134         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
23135         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
23137         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
23139 2013-04-08  Teresa Johnson  <tejohnson@google.com>
23141         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
23142         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
23143         methods.
23144         (estimate_edge_size_and_time): Add comment to suggest using rounding
23145         methods.
23146         (estimate_node_size_and_time): Ditto.
23147         (remap_edge_change_prob): Use helper rounding divide methods.
23148         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
23149         (gimple_mod_pow2_value_transform): Ditto.
23150         (gimple_mod_subtract_transform): Ditto.
23151         (gimple_ic_transform): Ditto.
23152         (gimple_stringops_transform): Ditto.
23153         * stmt.c (conditional_probability): Ditto.
23154         (emit_case_dispatch_table): Ditto.
23155         * lto-cgraph.c (merge_profile_summaries): Ditto.
23156         * tree-optimize.c (execute_fixup_cfg): Ditto.
23157         * cfgcleanup.c (try_forward_edges): Ditto.
23158         * cfgloopmanip.c (scale_loop_profile): Ditto.
23159         (loopify): Ditto.
23160         (duplicate_loop_to_header_edge): Ditto.
23161         (lv_adjust_loop_entry_edge): Ditto.
23162         * tree-vect-loop.c (vect_transform_loop): Ditto.
23163         * profile.c (compute_branch_probabilities): Ditto.
23164         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
23165         * lto-streamer-in.c (input_cfg): Ditto.
23166         * gimple-streamer-in.c (input_bb): Ditto.
23167         * ipa-cp.c (update_profiling_info): Ditto.
23168         (update_specialized_profile): Ditto.
23169         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
23170         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
23171         rounding methods.
23172         * sched-rgn.c (compute_dom_prob_ps): Ditto.
23173         (compute_trg_info): Ditto.
23174         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
23175         (purge_dead_edges): Ditto.
23176         * loop-unswitch.c (unswitch_loop): Ditto.
23177         * cgraphclones.c (cgraph_clone_edge): Ditto.
23178         (cgraph_clone_node): Ditto.
23179         * tree-inline.c (copy_bb): Ditto.
23180         (copy_edges_for_bb): Ditto.
23181         (initialize_cfun): Ditto.
23182         (copy_cfg_body): Ditto.
23183         (expand_call_inline): Ditto.
23185 2013-04-08  Kai Tietz  <ktietz@redhat.com>
23187         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
23188         TARGET_CYGWIN64 by TARGET_64BIT.
23190 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23192         * config/epiphany/epiphany.md (GPR_1): New constant.
23193         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
23194         * config/epiphany/epiphany.c (gen_compare_reg):
23195         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
23196         is already in place.
23197         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
23198         Don't require being called during rtl expansion; If y operlaps r0,
23199         return 0.
23200         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
23201         (epiphany_expand_epilogue): Likewise.
23203         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
23204         Don't use CC_FPmode for ORDERED / UNORDERED.
23205         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
23207         * config/epiphany/constraints.md (CnL): New constraint.
23208         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
23209         * config/epiphany/predicates.md (add_operand): Allow 1024.
23211         * config/epiphany/epiphany.md (logical_op): New code iterator.
23212         (op_mnc): New code attribute.
23213         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
23214         (mov_f+1, mov_f+2): New peephole2 patterns.
23216         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
23217         (cstoresi4): Also allow re-use of zero result when doing a NE
23218         comparison to a non-zero operand.
23219         Use (clobber (scratch)) for first insn if the gpr output is not needed.
23221         * config/epiphany/epiphany.md (<insn_opname>v2si3):
23222         Use gen_addsi3_i / gen_subsi3_i.
23224 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23226         PR c++/34949
23227         PR c++/50243
23228         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
23229         contain anything but clobbers, at most one __builtin_stack_restore,
23230         optionally debug stmts and final resx, and if it has at least one
23231         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
23232         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
23233         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
23234         which isn't defaut definition, remove them.
23235         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
23236         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
23237         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
23238         with MEM_REF LHS with SSA_NAME address.
23240 2013-04-08  Jeff Law  <law@redhat.com>
23242         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
23244 2013-04-08  Richard Biener  <rguenther@suse.de>
23246         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
23247         extra newline.
23248         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
23249         determined vector type.
23250         (vect_analyze_data_refs): Likewise.
23251         (vect_get_new_vect_var): Adjust.
23252         (vect_create_destination_var): Preserve SSA name versions.
23253         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
23254         not dump anything here.
23256 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23258         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
23259         Add member lr_slot_known.
23260         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
23261         if necessary.
23262         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
23263         Remove code that sets lr_slot_offset according to what a previous
23264         version of epiphany_emit_save_restore used to do.
23265         (epiphany_emit_save_restore): When doing an lr save or restore,
23266         set/verify lr_slot_known and lr_slot_offset.
23268 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
23270         PR target/54338
23271         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
23272         in ALL_REGS.
23274 2013-04-08  Richard Biener  <rguenther@suse.de>
23276         * alias.c (find_base_term): Fix thinko in previous change.
23278 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23280         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
23281         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
23282         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
23283         if possible to compute val.
23284         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
23285         For QImode integers don't require anything about precision.  Use
23286         const_with_all_bytes_same to find out if the constant doesn't have
23287         repeated bytes in it.
23289 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23291         * config/s390/s390.c (s390_expand_insv): Only accept insertions
23292         within mode size.
23294 2013-04-08  Marek Polacek  <polacek@redhat.com>
23296         PR rtl-optimization/48182
23297         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
23298         value to 1.
23300 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23302         PR target/55487
23303         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
23304         nuses, make sure we have a label.
23306 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23308         PR target/56843
23309         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
23310         (rs6000_emit_swdiv_low_precision): Remove.
23311         (rs6000_emit_swdiv): Rewrite to handle between one and four
23312         iterations of Newton-Raphson generally; modify required number of
23313         iterations for some cases.
23314         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
23316 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
23318         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
23319         set-but-unused variable.
23321         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
23322         basic blocks of released function bodies garbage-collectable.
23324         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
23325         (struct rtl_opt_pass): Add TODO_df_finish.
23327         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
23329 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23331         * config/arm/constraints.md (q): New constraint.
23332         * config/arm/ldrdstrd.md: New file.
23333         * config/arm/arm.md (ldrdstrd.md) New include.
23334         (arm_movdi): Use "q" instead of "r" constraint
23335         for double-word memory access.
23336         (movdf_soft_insn): Likewise.
23337         * config/arm/vfp.md (movdi_vfp): Likewise.
23338         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
23339         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
23340         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
23341         (mem_ok_for_ldrd_strd): Likewise.
23342         (output_move_double): Update assertion.
23344 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23346         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
23348 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23350         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
23351         define_insn_and_split.
23352         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
23354 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23356         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
23357         define_insn_and_split.
23358         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
23359         (shiftsi3_compare): New pattern.
23360         (rrx): New pattern.
23361         * config/arm/unspecs.md (UNSPEC_RRX): New.
23363 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23365         * config/arm/arm.md (negdi_extendsidi): New pattern.
23366         (negdi_zero_extendsidi): Likewise.
23368 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23370         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
23371         define_insn_and_split.
23372         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
23373         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
23375 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23377         * config/arm/arm.md (arm_subdi3): Convert define_insn into
23378         define_insn_and_split.
23379         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
23380         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
23382 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23384         * config/arm/arm.md (subsi3_carryin): New pattern.
23385         (subsi3_carryin_const): Likewise.
23386         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
23387         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
23389 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23391         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
23393 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23395         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
23396         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
23398 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23400         * config/arm/arm.c (arm_expand_builtin): Change fcode
23401         type to unsigned int.
23403 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23405         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
23407 2013-04-04  Ian Lance Taylor  <iant@google.com>
23409         * doc/standards.texi (Standards): The Go frontend supports the Go 1
23410         language standard.
23412 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
23414         PR middle-end/56729
23415         * df-scan.c (df_insn_delete): Disable failing assert.
23417 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23419         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
23420         New function prototype.
23421         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
23422         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
23423         (arm_builtin_vectorized_function): New function.
23425 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23427         * config/arm/arm_neon_builtins.def: New file.
23428         * config/arm/arm.c (neon_builtin_data): Move contents to
23429         arm_neon_builtins.def.
23430         (enum arm_builtins): Include neon builtin definitions.
23431         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
23432         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
23434 2013-04-04  Marek Polacek  <polacek@redhat.com>
23436         PR tree-optimization/48186
23437         * predict.c (maybe_hot_frequency_p): Return false if
23438         HOT_BB_FREQUENCY_FRACTION is 0.
23439         (cgraph_maybe_hot_edge_p): Likewise.
23441 2013-04-04  Richard Biener  <rguenther@suse.de>
23443         PR tree-optimization/56826
23444         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
23445         more accurately.
23447 2013-04-04  Richard Biener  <rguenther@suse.de>
23449         PR tree-optimization/56213
23450         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
23451         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
23453 2013-04-04  Richard Biener  <rguenther@suse.de>
23455         PR tree-optimization/56837
23456         * tree-loop-distribution.c (classify_partition): For non-zero
23457         values require that the value has the same precision as its
23458         mode to be useful as memset value.
23460 2013-04-03  Nick Clifton  <nickc@redhat.com>
23462         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
23463         (fmssf4): Use fmsf.s on E3V5 architectures.
23464         (fnmasf4): Use fnmaf.s on E3V5 architectures.
23465         (fnmssf4): Use fnmsf.s on E3V5 architectures.
23467 2013-04-03  Jeff Law  <law@redhat.com>
23469         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
23470         (lra-eliminations.o): Likewise.
23472 2013-04-03  Teresa Johnson  <tejohnson@google.com>
23474         * gcov-io.c (compute_working_sets): Moved most of body of old
23475         compute_working_sets here from profile.c.
23476         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
23477         (gcov_working_set_t): Moved typedef here from basic-block.h
23478         (compute_working_set): Declare.
23479         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
23480         (get_working_sets): Renamed from compute_working_set,
23481         replace most of body with call to new compute_working_sets.
23482         (get_exec_counts): Replace call to compute_working_sets
23483         to get_working_sets.
23484         * profile.h (get_working_sets): Renamed from compute_working_set.
23485         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
23486         to get_working_sets.
23487         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
23488         * gcov-dump.c (dump_working_sets): New function.
23490 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
23492         * hwint.c (sext_hwi, zext_hwi): New functions.
23493         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
23494         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
23495         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
23496         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
23497         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
23498         (sext_hwi, zext_hwi): New functions.
23500 2013-04-03  Jeff Law  <law@redhat.com>
23502         PR tree-optimization/56799
23503         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
23504         back test for widening conversion erroneously dropped in prior change.
23506 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23508         PR target/56809
23509         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
23510         instead of next_real_insn.
23512 2013-04-03  Marek Polacek  <polacek@redhat.com>
23514         PR sanitizer/55702
23515         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
23517 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23519         PR target/56809
23520         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
23521         next_real_insn.
23522         (thumb1_output_casesi): Likewise.
23523         (thumb2_output_casesi): Likewise.
23525 2013-04-03  Richard Biener  <rguenther@suse.de>
23527         PR tree-optimization/56817
23528         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
23529         Split out ...
23530         (tree_unroll_loops_completely_1): ... new function to manually
23531         walk the loop tree, properly defering outer loops of unrolled
23532         loops to later iterations.
23534 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23536         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
23537         (vectorizable_load): Likewise.
23538         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
23539         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
23541 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23543         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
23544         BIT_FIELD_REF.
23546 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
23548         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
23550 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
23552         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
23554 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23556         PR tree-optimization/56790
23557         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
23558         folding.
23560 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23562         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
23563         Handle VEC_MERGE.
23564         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
23565         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
23566         equal arguments.
23568 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
23570         PR c/19449
23571         * tree.h (force_folding_builtin_constant_p): New decl.
23572         * builtins.c (force_folding_builtin_constant_p): New variable.
23573         (fold_builtin_constant_p): Fold immediately also if
23574         force_folding_builtin_constant_p.
23576 2013-04-03  Richard Biener  <rguenther@suse.de>
23578         PR tree-optimization/56812
23579         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
23580         DRs of the same interleaving chain are independent.
23582 2013-04-02  Jason Merrill  <jason@redhat.com>
23584         * gdbinit.in (pbb): Use debug fn.
23586 2013-04-02  Lawrence Crowl  <crowl@google.com>
23588         * sese.h (struct ivtype_map_elt_s): Remove unused.
23589         (extern debug_ivtype_map): Remove unused.
23590         (extern eq_ivtype_map_elts): Remove unused.
23591         * sese.c (debug_ivtype_map): Removed unused.
23592         (debug_ivtype_map_1): Removed unused.
23593         (debug_ivtype_elt): Remove unused.
23594         (eq_ivtype_map_elts): Remove unused.
23597 2013-04-02  Kai Tietz  <ktietz@redhat.com>
23599         PR target/52790
23600         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
23601         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
23602         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
23603         function.
23604         (legitimize_pe_coff_symbol): Likewise.
23605         (is_imported_p): New helper-function.
23606         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
23607         for Windows x64 targets.
23608         (ix86_expand_prologue): Optimize for pe-coff targets.
23609         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
23610         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
23611         medium/large code-model.
23612         (legitimize_pic_address): Likewise.
23613         (legitimize_tls_address): Likewise.
23614         (ix86_expand_call): Likewise.
23615         (x86_output_mi_thunk): Likewise.
23616         (get_dllimport_decl): Add new beimport argument.
23617         (construct_plt_address): Don't assert for x64 pe-coff targets.
23618         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
23619         targets.
23620         (SYMBOL_FLAG_STUBVAR): New macro.
23621         (SYMBOL_REF_STUBVAR_P): Likewise.
23622         * config/i386/winnt.c (stub_list): New structure.
23623         (stub_head): New local variable.
23624         (i386_pe_record_stub): New function.
23625         (i386_pe_file_end): Emit refptr-stubs.
23627 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
23629         PR rtl-optimization/56745
23630         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
23631         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
23633         PR c++/34949
23634         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
23635         and both of them are MEM_REFs, just compare first argument for
23636         equality and attempt to deal even with differing offsets.
23638         PR c++/34949
23639         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
23640         of gimple_clobber_p to be MEM_REF.
23641         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
23642         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
23643         after gimplification.
23644         * asan.c (get_mem_ref_of_assignment): Don't instrument
23645         gimple_clobber_p stmts.
23646         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
23647         gimple_clobber_p stmt if they have MEM_REF lhs and
23648         are dead because of another gimple_clobber_p stmt.
23649         * tree-ssa-live.c (clear_unused_block_pointer): Treat
23650         gimple_clobber_p stmts like debug stmts.
23651         (remove_unused_locals): Remove clobbers with MEM_REF lhs
23652         that refer to unused VAR_DECLs or uninitialized values.
23653         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
23654         gimple_clobber_p stmts if they refer to removed parameters.
23655         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
23656         formatting.
23658 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
23660         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
23661         using SWI48 mode attribute.
23663 2013-04-02  Wei Mi  <wmi@google.com>
23665         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
23666         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
23667         *<rotate_insn><mode>3_mask in i386.md.
23669 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
23671         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
23673 2013-04-02  Richard Biener  <rguenther@suse.de>
23675         PR tree-optimization/56778
23676         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
23677         Runtime alias tests are not supported for gather loads.
23678         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
23679         stmts referenced from SSA operands before updating SSA form.
23681 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
23682             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23684         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
23685         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
23686         * config/arm/cortex-a53.md: New file.
23687         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
23688         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
23689         * config/arm/arm.c (arm_issue_rate): Likewise.
23690         * config/arm/arm-tune.md: Regenerate
23691         * config/arm/arm-tables.opt: Regenerate.
23692         * config/arm/arm-cores.def: Add cortex-a53.
23694 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
23696         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
23697         non-static link.
23699 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23701         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
23702         scalar load/store operations using B/H registers.
23703         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
23705 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23707         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
23708         scalar move.
23709         * config/aarch64/aarch64.c
23710         (aarch64_simd_scalar_immediate_valid_for_move): New.
23711         * config/aarch64/aarch64-protos.h
23712         (aarch64_simd_scalar_immediate_valid_for_move): New.
23713         * config/aarch64/constraints.md (Dh, Dq): New.
23714         * config/aarch64/iterators.md (hq): New.
23716 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
23718         * reorg.c (get_branch_condition): Deal with conditional returns.
23719         (fill_simple_delay_slots): Remove dead code dealing with jumps.
23721 2013-04-01  Wei Mi  <wmi@google.com>
23723         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
23724         Truncate operand 2 using %b asm operand modifier.
23725         (*<shift_insn><mode>3_mask): Ditto.
23726         (*<rotate_insn><mode>3_mask): Ditto.
23728 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
23730         PR middle-end/56798
23731         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
23733 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
23735         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
23736         of next_real_insn.
23737         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
23739 2013-03-30  Lawrence Crowl  <crowl@google.com>
23741         * dse.c (clear_alias_sets): Remove never set.
23742         (disqualified_clear_alias_sets): Remove never set.
23743         (clear_alias_mode_pool): Remove never set.
23744         (dse_step0): Remove condition that is never true.
23745         (canon_address): Remove condition that is never true.
23746         (dse_step7): Remove condition that is never true.
23747         (rest_of_handle_dse): Remove condition that is never true.
23748         (rest_of_handle_dse::did_global): Remove never read from above.
23749         (dse_step2_spill): Remove never called from above.
23750         (dse_step5_spill): Remove never called from above.
23752 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
23754         * doc/md.texi (Standard Names) <casesi>: Update documentation for
23755         JUMP_TABLE_DATA changes.
23756         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
23757         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
23758         (Insns) <jump_table_data>: New entry.
23759         * doc/tm.texi: Regenerate.
23761         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
23763         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
23764         for table jump at the end of a basic block using tablejump_p.
23765         * targhooks.c (default_invalid_within_doloop): Likewise.
23766         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
23767         target hook implementation that is identical to the default hook.
23768         (rs6000_invalid_within_doloop): Remove.
23770         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
23771         unused variable from tablejump_p call.
23773         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
23774         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
23775         (INSN_DELETED_P): Likewise.
23776         (emit_jump_table_data): New prototype.
23777         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
23778         after 4th as unused.
23779         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
23780         * sched-vis.c (print_insn): Likewise.
23781         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
23782         insn for compatibility with back ends that use next_active_insn to
23783         identify jump table data.
23784         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
23785         (remove_insn): Likewise.
23786         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
23787         to be emitted.
23788         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
23789         (emit_jump_table_data): New function.
23791         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
23792         basic block, a JUMP_TABLE_DATA never is.
23793         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
23794         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
23795         off from code handling real insns.
23796         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
23797         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
23798         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
23799         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
23800         is not a NONDEBUG_INSN_P.
23801         * ira-costs.c (scan_one_insn): Likewise.
23802         * jump.c (mark_all_labels): Likewise.
23803         (mark_jump_label_1): Likewise.
23804         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
23805         * lra.c (get_insn_freq): Expect all insns reaching here to be in
23806         a basic block.
23807         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
23808         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
23809         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
23810         JUMP_TABLE_DATA_P insns.
23811         (calculate_elim_costs_all_insns): Likewise.
23812         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
23813         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
23814         (delete_output_reload): Code style fixups.
23815         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
23816         insn flags on this non-insn.
23817         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
23818         as scheduling barriers, for pre-change compatibility.
23819         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
23820         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
23822         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
23823         redundant JUMP_TABLE_DATA_P test.
23824         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
23825         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
23826         (frv_for_each_packet): Likewise.
23827         * config/i386/i386.c (min_insn_size): Likewise.
23828         (ix86_avoid_jump_mispredicts): Likewise.
23829         * config/m32r/m32r.c (m32r_is_insn): Likewise.
23830         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
23831         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
23832         (mips16_insn_length): Robustify.
23833         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
23834         (mips16_split_long_branches): Likewise.
23835         * config/pa/pa.c (pa_combine_instructions): Likewise.
23836         * config/rs6000/rs6000.c (get_next_active_insn): Treat
23837         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
23838         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
23839         as contributing to pool range lengths.
23840         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
23841         Remove redundant JUMP_TABLE_DATA_P test.
23842         (sh_loop_align): Likewise.
23843         (split_branches): Likewise.
23844         (sh_insn_length_adjustment): Likewise.
23845         * config/spu/spu.c (get_branch_target): Likewise.
23847 2013-03-29  Jan Hubicka  <jh@suse.cz>
23849         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
23850         gcov streaming; stream hot bb threshold to ltrans.
23851         * predict.c (get_hot_bb_threshold): Break out from ....
23852         (maybe_hot_count_p): ... here.
23853         (set_hot_bb_threshold): New function.
23854         * lto-section-in.c (lto_section_name): Add profile.
23855         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
23856         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
23857         and data-streamer.h
23858         (histogram_entry): New structure.
23859         (histogram, histogram_pool): New global vars.
23860         (histogram_hash): New structure.
23861         (histogram_hash::hash): New method.
23862         (histogram_hash::equal): Likewise.
23863         (account_time_size): New function.
23864         (cmp_counts): New function.
23865         (dump_histogram): New function.
23866         (ipa_profile_generate_summary): New function.
23867         (ipa_profile_write_summary): New function.
23868         (ipa_profile_read_summary): New function.
23869         (ipa_profile): Decide on threshold.
23870         (pass_ipa_profile): Add ipa_profile_write_summary and
23871         ipa_profile_read_summary.
23872         * Makefile.in (ipa.o): Update dependencies.
23873         * lto-streamer.h (LTO_section_ipa_profile): New section.
23875 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
23877         * tree.h (VAR_P): New.
23879 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
23881         PR lto/56777
23882         * doc/invoke.texi ([-fwhole-program]): Fix typo.
23884 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
23886         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
23887         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
23888         (control_flow_insn_p): Likewise.
23889         * cfgrtl.c (duplicate_insn_chain): Likewise.
23890         * final.c (get_attr_length_1): Likewise.
23891         (shorten_branches): Likewise.
23892         (final_scan_insn): Likewise.
23893         * function.c (instantiate_virtual_regs): Likewise.
23894         * gcse.c (insert_insn_end_basic_block): Likewise.
23895         * ira-costs.c (scan_one_insn): Likewise.
23896         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
23897         * lra.c (check_rtl): Likewise.
23898         * reload1.c (elimination_costs_in_insn): Likewise.
23899         * reorg.c (follow_jumps): Likewise.
23901         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
23902         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
23903         (thumb_far_jump_used_p): Likewise.
23904         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
23905         (workaround_speculation): Likewise.
23906         (add_sched_insns_for_speculation): Likewise.
23907         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
23908         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
23909         (frv_for_each_packet): Likewise.
23910         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
23911         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
23912         (final_emit_insn_group_barriers): Likewise.
23913         * config/m32r/m32r.c (m32r_is_insn): Likewise.
23914         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
23915         (mips16_insn_length): Likewise.
23916         * config/pa/pa.c (pa_reorg): Likewise.
23917         (pa_combine_instructions): Likewise.
23918         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
23919         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
23920         (sh_reorg): Likewise.
23921         (split_branches): Likewise.
23922         * config/spu/spu.c (get_branch_target): Likewise.
23924         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
23925         JUMP_TABLE_DATA_P.
23927 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
23929         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
23930         Fix declaration name.
23932 2013-03-28  Lawrence Crowl  <crowl@google.com>
23934         * graphds.h (struct graph.indicies): Remove unused.
23935         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
23936         (SCOP_ORIGINAL_PDDRS): Remove unused.
23937         * sese.h (extern insert_loop_close_phis): Removed unused.
23938         (extern insert_guard_phis): Removed unused.
23939         (extern ivtype_map_elt_info): Removed unused.
23940         (new_ivtype_map_elt): Removed unused.
23941         * sese.c (ivtype_map_elt_info): Removed unused.
23943 2013-03-28  Lawrence Crowl  <crowl@google.com>
23945         * Makefile.in: Add several missing include dependences.
23946         (DUMPFILE_H): New.
23947         (test-dump.o): New.  This object is not added to any executable,
23948         but is present for ad-hoc testing.
23949         * bitmap.c
23950         (debug (const bitmap_head_def &)): New.
23951         (debug (const bitmap_head_def *)): New.
23952         * bitmap.h
23953         (extern debug (const bitmap_head_def &)): New.
23954         (extern debug (const bitmap_head_def *)): New.
23955         * cfg.c
23956         (debug (edge_def &)): New.
23957         (debug (edge_def *)): New.
23958         * cfghooks.c
23959         (debug (basic_block_def &)): New.
23960         (debug (basic_block_def *)): New.
23961         * dumpfile.h
23962         (dump_node (const_tree, int, FILE *)): Correct source file.
23963         * dwarf2out.c
23964         (debug (die_struct &)): New.
23965         (debug (die_struct *)): New.
23966         * dwarf2out.h
23967         (extern debug (die_struct &)): New.
23968         (extern debug (die_struct *)): New.
23969         * gimple-pretty-print.c
23970         (debug (gimple_statement_d &)): New.
23971         (debug (gimple_statement_d *)): New.
23972         * gimple-pretty-print.h
23973         (extern debug (gimple_statement_d &)): New.
23974         (extern debug (gimple_statement_d *)): New.
23975         * ira-build.c
23976         (debug (ira_allocno_copy &)): New.
23977         (debug (ira_allocno_copy *)): New.
23978         (debug (ira_allocno &)): New.
23979         (debug (ira_allocno *)): New.
23980         * ira-int.h
23981         (extern debug (ira_allocno_copy &)): New.
23982         (extern debug (ira_allocno_copy *)): New.
23983         (extern debug (ira_allocno &)): New.
23984         (extern debug (ira_allocno *)): New.
23985         * ira-lives.c
23986         (debug (live_range &)): New.
23987         (debug (live_range *)): New.
23988         * lra-int.h
23989         (debug (lra_live_range &)): New.
23990         (debug (lra_live_range *)): New.
23991         * lra-lives.c
23992         (debug (lra_live_range &)): New.
23993         (debug (lra_live_range *)): New.
23994         * omega.c
23995         (debug (omega_pb_d &)): New.
23996         (debug (omega_pb_d *)): New.
23997         * omega.h
23998         (extern debug (omega_pb_d &)): New.
23999         (extern debug (omega_pb_d *)): New.
24000         * print-rtl.c
24001         (debug (const rtx_def &)): New.
24002         (debug (const rtx_def *)): New.
24003         * print-tree.c
24004         (debug_tree (tree): Move within file.
24005         (debug_raw (const tree_node &)): New.
24006         (debug_raw (const tree_node *)): New.
24007         (dump_tree_via_hooks (const tree_node *, int)): New.
24008         (debug (const tree_node &)): New.
24009         (debug (const tree_node *)): New.
24010         (debug_verbose (const tree_node &)): New.
24011         (debug_verbose (const tree_node *)): New.
24012         (debug_head (const tree_node &)): New.
24013         (debug_head (const tree_node *)): New.
24014         (debug_body (const tree_node &)): New.
24015         (debug_body (const tree_node *)): New.
24016         (debug_vec_tree (tree): Move and reimplement in terms of dump.
24017         (debug (vec<tree, va_gc> &)): New.
24018         (debug (vec<tree, va_gc> *)): New.
24019         * rtl.h
24020         (extern debug (const rtx_def &)): New.
24021         (extern debug (const rtx_def *)): New.
24022         * sbitmap.c
24023         (debug_raw (simple_bitmap_def &)): New.
24024         (debug_raw (simple_bitmap_def *)): New.
24025         (debug (simple_bitmap_def &)): New.
24026         (debug (simple_bitmap_def *)): New.
24027         * sbitmap.h
24028         (extern debug (simple_bitmap_def &)): New.
24029         (extern debug (simple_bitmap_def *)): New.
24030         (extern debug_raw (simple_bitmap_def &)): New.
24031         (extern debug_raw (simple_bitmap_def *)): New.
24032         * sel-sched-dump.c
24033         (debug (vinsn_def &)): New.
24034         (debug (vinsn_def *)): New.
24035         (debug_verbose (vinsn_def &)): New.
24036         (debug_verbose (vinsn_def *)): New.
24037         (debug (expr_def &)): New.
24038         (debug (expr_def *)): New.
24039         (debug_verbose (expr_def &)): New.
24040         (debug_verbose (expr_def *)): New.
24041         (debug (vec<rtx> &)): New.
24042         (debug (vec<rtx> *)): New.
24043         * sel-sched-dump.h
24044         (extern debug (vinsn_def &)): New.
24045         (extern debug (vinsn_def *)): New.
24046         (extern debug_verbose (vinsn_def &)): New.
24047         (extern debug_verbose (vinsn_def *)): New.
24048         (extern debug (expr_def &)): New.
24049         (extern debug (expr_def *)): New.
24050         (extern debug_verbose (expr_def &)): New.
24051         (extern debug_verbose (expr_def *)): New.
24052         (extern debug (vec<rtx> &)): New.
24053         (extern debug (vec<rtx> *)): New.
24054         * sel-sched-ir.h
24055         (_list_iter_cond_expr): Make inline instead of static.
24056         * sreal.c
24057         (debug (sreal &)): New.
24058         (debug (sreal *)): New.
24059         * sreal.h
24060         (extern debug (sreal &)): New.
24061         (extern debug (sreal *)): New.
24062         * tree.h
24063         (extern debug_raw (const tree_node &)): New.
24064         (extern debug_raw (const tree_node *)): New.
24065         (extern debug (const tree_node &)): New.
24066         (extern debug (const tree_node *)): New.
24067         (extern debug_verbose (const tree_node &)): New.
24068         (extern debug_verbose (const tree_node *)): New.
24069         (extern debug_head (const tree_node &)): New.
24070         (extern debug_head (const tree_node *)): New.
24071         (extern debug_body (const tree_node &)): New.
24072         (extern debug_body (const tree_node *)): New.
24073         (extern debug (vec<tree, va_gc> &)): New.
24074         (extern debug (vec<tree, va_gc> *)): New.
24075         * tree-cfg.c
24076         (debug (struct loop &)): New.
24077         (debug (struct loop *)): New.
24078         (debug_verbose (struct loop &)): New.
24079         (debug_verbose (struct loop *)): New.
24080         * tree-dump.c: Add header dependence.
24081         * tree-flow.h
24082         (extern debug (struct loop &)): New.
24083         (extern debug (struct loop *)): New.
24084         (extern debug_verbose (struct loop &)): New.
24085         (extern debug_verbose (struct loop *)): New.
24086         * tree-data-ref.c
24087         (debug (data_reference &)): New.
24088         (debug (data_reference *)): New.
24089         (debug (vec<data_reference_p> &)): New.
24090         (debug (vec<data_reference_p> *)): New.
24091         (debug (vec<ddr_p> &)): New.
24092         (debug (vec<ddr_p> *)): New.
24093         * tree-data-ref.h
24094         (extern debug (data_reference &)): New.
24095         (extern debug (data_reference *)): New.
24096         (extern debug (vec<data_reference_p> &)): New.
24097         (extern debug (vec<data_reference_p> *)): New.
24098         (extern debug (vec<ddr_p> &)): New.
24099         (extern debug (vec<ddr_p> *)): New.
24100         * tree-ssa-alias.c
24101         (debug (pt_solution &)): New.
24102         (debug (pt_solution *)): New.
24103         * tree-ssa-alias.h
24104         (extern debug (pt_solution &)): New.
24105         (extern debug (pt_solution *)): New.
24106         * tree-ssa-alias.c
24107         (debug (_var_map &)): New.
24108         (debug (_var_map *)): New.
24109         (debug (tree_live_info_d &)): New.
24110         (debug (tree_live_info_d *)): New.
24111         * tree-ssa-alias.h
24112         (extern debug (_var_map &)): New.
24113         (extern debug (_var_map *)): New.
24114         (extern debug (tree_live_info_d &)): New.
24115         (extern debug (tree_live_info_d *)): New.
24117 2013-03-28  Jan Hubicka  <jh@suse.cz>
24119         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
24121 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
24123         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
24124         record only when desired or required.
24126 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
24128         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
24129         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
24131 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24133         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
24134         (*andsi3_compare0_uxtw): New pattern.
24135         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
24136         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
24138 2013-03-28  Jan Hubicka  <jh@suse.cz>
24140         * data-streamer-in.c (streamer_read_gcov_count): New function.
24141         * gimple-streamer-out.c: Include value-prof.h.
24142         (output_gimple_stmt): Output histogram.
24143         (output_bb): Use streamer_write_gcov_count.
24144         * value-prof.c: Include data-streamer.h
24145         (dump_histogram_value): Add HIST_TYPE_MAX.
24146         (stream_out_histogram_value): New function.
24147         (stream_in_histogram_value): New function.
24148         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
24149         (stream_out_histogram_value, stream_in_histogram_value): Declare.
24150         * data-streamer-out.c (streamer_write_gcov_count): New function.
24151         (streamer_write_gcov_count_stream): New function.
24152         * lto-cgraph.c (lto_output_edge): Update counter streaming.
24153         (lto_output_node): Likewise.
24154         (input_node, input_edge): Likewise.
24155         * lto-streamer-out.c (output_cfg): Update streaming.
24156         * lto-streamer-in.c (input_cfg): Likewise.
24157         * data-streamer.h (streamer_write_gcov_count,
24158         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
24159         * gimple-streamer-in.c: Include value-prof.h
24160         (input_gimple_stmt): Input histograms.
24161         (input_bb): Update profile streaming.
24163 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24165         * genmodes.c (emit_max_int): New function.
24166         (emit_insn_modes_h): Added call to emit_max_function.
24167         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
24168         Added doc.
24169         * machmode.def: Fixed comment.
24171 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24173         * combine.c (try_combine): Removed useless assert.
24174         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
24176 2013-03-28  Marek Polacek  <polacek@redhat.com>
24177             Richard Biener  <rguenther@suse.de>
24179         PR tree-optimization/56695
24180         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
24181         build signed result of a vector comparison.
24182         * tree-cfg.c (verify_gimple_comparison): Check that a result
24183         of a vector comparison has signed type.
24185 2013-03-28  Richard Biener  <rguenther@suse.de>
24187         PR tree-optimization/37021
24188         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
24189         do not restrict gaps between groups.
24190         * tree-vect-stmts.c (vectorizable_load): Properly account for
24191         a gap between groups.
24193 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
24195         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
24196         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
24197         is not enabled.
24199 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
24201         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
24202         * doc/extend.texi (Named Address Spaces): Ditto.
24203         (Variable Attributes): Ditto.
24205 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24207         * config.build: Add support for cygwin x64 target.
24208         * config.gcc: Likewise.
24209         * config.host: Likewise.
24210         * configure.ac: Likewise
24211         * configure: Regenerated.
24213 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24215         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
24216         * config/i386/t-cygwin-w64: New file.
24217         * config/i386/cygwin-w64.h: New file.
24218         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
24219         and add support for x64-cygwin target.
24220         (CPP_SPEC): Likewise.
24221         (CXX_WRAP_SPEC_LIST): Undefine before define.
24222         (LIBGCJ_SONAME): Use 15 as version.
24224 2013-03-27  Richard Biener  <rguenther@suse.de>
24226         PR tree-optimization/56716
24227         * tree-ssa-structalias.c (perform_var_substitution): Adjust
24228         dumping for ref nodes.
24230 2013-03-27  Martin Jambor  <mjambor@suse.cz>
24232         PR tree-optimization/55334
24233         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
24234         restricted pointers to arrays.
24236 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24238         * Makefile.in (.SUFFIXES): Add .cc.
24239         (.c.o): Apply same recipe for implicit rule .cc.o.
24241 2013-03-27  Richard Biener  <rguenther@suse.de>
24243         PR tree-optimization/37021
24244         * tree-vect-data-refs.c (vect_check_strided_load): Allow
24245         REALPART/IMAGPART_EXPRs around the supported refs.
24246         * tree-ssa-structalias.c (find_func_aliases): Assume that
24247         floating-point values are not used to transfer pointers.
24249 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
24251         * target.def (TARGET_HAS_IFUNC_P): New target hook.
24252         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
24253         * doc/tm.texi: Regenerate.
24254         * targhooks.h (default_has_ifunc_p): New.
24255         * targhooks.c (default_has_ifunc_p): Ditto.
24256         * config/linux-protos.h: New file.
24257         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
24258         hook for linux which disables support of indirect functions in android.
24259         * config/linux-android.c: New file.
24260         * config/t-linux-android.c: Ditto.
24261         * config.gcc: Added new object file linux-android.o.
24262         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
24263         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
24264         * varasm.c (do_assemble_alias): Likewise.
24265         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
24266         doesn't support indirect functions.
24267         * configure: Regenerate.
24269 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
24271         PR target/56102
24272         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
24273         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
24274         mult-word mode.
24276 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24278         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
24280 2013-03-27  Terry Guo  <terry.guo@arm.com>
24282         * config/arm/arm-cores.def: Added core cortex-r7.
24283         * config/arm/arm-tune.md: Regenerated.
24284         * config/arm/arm-tables.opt: Regenerated.
24285         * doc/invoke.texi: Added entry for core cortex-r7.
24287 2013-03-27  Walter Lee  <walt@tilera.com>
24289         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
24290         double-decrement of next_scratch_regno.
24292 2013-03-27  Walter Lee  <walt@tilera.com>
24294         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
24295         input operands.
24296         (insn_v1mulus): Ditto.
24297         (insn_v2muls): Ditto.
24299 2013-03-27  Walter Lee  <walt@tilera.com>
24301         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
24302         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
24304 2013-03-27  Walter Lee  <walt@tilera.com>
24306         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
24307         (*sibcall_value): Ditto.
24309 2013-03-27  Walter Lee  <walt@tilera.com>
24311         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
24312         (insn_mnz_v8qi): ... this ...
24313         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
24314         vector equivalent.
24315         (insn_v<n>mnz): Replaced by ...
24316         (insn_v1mnz): ... this ...
24317         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
24318         equivalent.
24319         (insn_mz_<mode>): Replaced by ...
24320         (insn_mz_v8qi): ... this ...
24321         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
24322         vector equivalent.
24323         (insn_v<n>mz): Replaced by ...
24324         (insn_v1mz): ... this ...
24325         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
24326         equivalent.
24328 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
24330         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
24332 2013-03-26  Roland McGrath  <mcgrathr@google.com>
24334         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
24335         than fprintf with a non-constant, non-format string.
24337 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24339         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
24340         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
24341         operand 0 predicate.
24342         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
24343         attribute.  Use general_x64nomem_operand as operand 1 predicate.
24344         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
24345         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24346         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
24347         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24348         (mov<mode>_insv_1): Remove expander.  Merge insn with
24349         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
24350         Use general_x64nomem_operand as operand 1 predicate.
24351         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
24352         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
24353         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24354         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
24355         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
24356         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
24357         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24358         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
24359         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
24360         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
24361         (general_x64nomem_operand): Ditto.
24363 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
24365         * config/rtems.opt: Add -pthread option.
24367 2013-03-26  Richard Biener  <rguenther@suse.de>
24369         * alias.c (find_base_term): Avoid redundant and not used recursion.
24370         (base_alias_check): Get the initial base term from the caller.
24371         (true_dependence_1): Compute and pass base terms to base_alias_check.
24372         (write_dependence_p): Likewise.
24373         (may_alias_p): Likewise.
24375 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
24377         * config/aarch64/aarch64.c (aarch64_classify_address): Support
24378         PC-relative load in SI modes and above only.
24380 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
24382         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
24383         * config/arm/iwmmxt.md (WCGR0): Update.
24384         (WCGR1, WCGR2, WCGR3): Likewise.
24386 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24388         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
24389         Use x64 and nox64 isa attributes.
24391 2013-03-26  Richard Biener  <rguenther@suse.de>
24393         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
24394         alignment computations and rely on get_object_alignment_1
24395         for the !TYPE_P case.
24396         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
24398 2013-03-26  Walter Lee  <walt@tilera.com>
24400         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
24401         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
24403 2013-03-25  Jeff Law  <law@redhat.com>
24405         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
24406         check for INTEGRAL_TYPE_P that was missing due to checking in
24407         wrong version of prior patch.
24409 2013-03-25  Walter Lee  <walt@tilera.com>
24411         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
24412         TILEGX_INSN_SHUFFLEBYTES1.
24413         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
24414         shufflebytes1.
24415         (tilegx_builtins): Ditto.
24416         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
24418 2013-03-25  Walter Lee  <walt@tilera.com>
24420         * config/tilegx/tilegx.md (floatsisf2): New pattern.
24421         (floatunssisf2): New pattern.
24422         (floatsidf2): New pattern.
24423         (floatunssidf2): New pattern.
24425 2013-03-25  Walter Lee  <walt@tilera.com>
24427         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
24428         tests for constraint J, K, N, P.
24430 2013-03-25  Walter Lee  <walt@tilera.com>
24432         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
24433         Use indirect/pcrel encoding.
24434         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
24435         Ditto.
24437 2013-03-25  Steve Ellcey  <sellcey@mips.com>
24439         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
24440         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
24441         * config/mips/mips.c (mips_option_override): Set IMADD default.
24442         * config/mips/mips.h (PTF_AVOID_IMADD): New.
24443         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
24444         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
24445         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
24446         * doc/invoke.texi (-mimadd/-mno-imadd): New.
24448 2013-03-25  Jeff Law  <law@redhat.com>
24450         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
24451         slightly to avoid creating and folding useless trees.  Simplify
24452         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
24454 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
24456         * config/i386/i386.md (*zero_extendsidi2): Merge with
24457         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
24458         * config/i386/predicates.md (x86_64_zext_operand): Rename from
24459         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
24460         targets.  Clarify comment.
24462 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24464         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
24465         pass-through jump functions differently.
24466         (ipa_read_jump_function): Likewise.  Also use setter functions to set
24467         up jump functions.
24469 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24471         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
24472         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
24473         process it.
24474         (ipa_get_indirect_edge_target): New function.
24475         (devirtualization_time_bonus): New parameter known_aggs, pass it to
24476         ipa_get_indirect_edge_target.  Update all callers.
24477         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
24478         ipa_get_indirect_edge_target_1 instead of calling
24479         ipa_get_indirect_edge_target.
24480         (create_specialized_node): Pass aggvlas to
24481         ipcp_discover_new_direct_edges.
24483 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24485         * config/arm/arm.md (f_sels, f_seld): New types.
24486         (*cmov<mode>): New pattern.
24487         * config/arm/predicates.md (arm_vsel_comparison_operator): New
24488         predicate.
24490 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24492         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
24493         POSIX-printf for mingw-hosted  builds.
24495 2013-03-25  Richard Biener  <rguenther@suse.de>
24497         PR middle-end/56694
24498         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
24499         must-not-throw stmt location.
24501 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24503         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
24504         Emit load-acquire versions when acq is true.
24505         (arm_emit_store_exclusive): Add rel parameter.
24506         Emit store-release versions when rel is true.
24507         (arm_split_compare_and_swap): Use acquire-release instructions
24508         instead.
24509         of barriers when appropriate.
24510         (arm_split_atomic_op): Likewise.
24511         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
24512         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
24513         (VUNSPEC_SLX): Likewise.
24514         (VUNSPEC_LDA): Likewise.
24515         (VUNSPEC_STL): Likewise.
24516         * config/arm/sync.md (atomic_load<mode>): New pattern.
24517         (atomic_store<mode>): Likewise.
24518         (arm_load_acquire_exclusive<mode>): Likewise.
24519         (arm_load_acquire_exclusivesi): Likewise.
24520         (arm_load_acquire_exclusivedi): Likewise.
24521         (arm_store_release_exclusive<mode>): Likewise.
24523 2013-03-25  Catherine Moore  <clm@codesourcery.com>
24525         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
24526         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
24527         * config/mip/predicates.md (lwsp_swsp_operand,
24528         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
24529         sb16_operand, db4_operand, db7_operand, ib3_operand,
24530         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
24531         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
24532         andi16_operand): New predicates.
24533         * config/mips/mips.md (compression): New attribute.
24534         (enabled): New attribute.
24535         (length): Consider compression in computing length.
24536         (shift_compression): New code attribute.
24537         (*add<mode>3): New operands. Record compression.
24538         (sub<mode>3): Likewise.
24539         (one_cmpl<mode>2): Likewise.
24540         (*and<mode>3): Likewise.
24541         (*ior<mode>3): Likewise.
24542         (unnamed pattern for xor): Likewise.
24543         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
24544         (*<optab><mode>3): Likewise.
24545         (*mov<mode>_internal: Likewise.
24546         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
24547         (mips_unsigned_immediate_p): New.
24548         (umips_lwsp_swsp_address_p): New.
24549         (m16_based_address_p): New.
24550         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
24551         (mips_unsigned_immediate_p): New prototype.
24552         (lwsp_swsp_address_p): New prototype.
24553         (m16_based_address_p): New prototype.
24554         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
24555         (mips_signed_immediate_p): New function.
24556         (m16_based_address_p): New function.
24557         (lwsp_swsp_address_p): New function.
24558         (mips_print_operand_punctuation): Recognize short delay slot insns
24559         for microMIPS.add<mode>3"
24561 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24563         PR target/56720
24564         * config/arm/iterators.md (v_cmp_result): New mode attribute.
24565         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
24567 2013-03-25  Richard Biener  <rguenther@suse.de>
24569         PR tree-optimization/56689
24570         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
24571         any edge.
24573 2013-03-25  Richard Biener  <rguenther@suse.de>
24575         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
24576         of bitmap.
24577         (memory_references): Likewise.
24578         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
24579         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
24580         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
24581         (gather_mem_refs_in_loops): Fold into ...
24582         (analyze_memory_references): ... this.  Move initialization
24583         to tree_ssa_lim_initialize.
24584         (fill_always_executed_in): Rename to ...
24585         (fill_always_executed_in_1): ... this.
24586         (fill_always_executed_in): Move contains_call computation to
24587         this new function from ...
24588         (tree_ssa_lim_initialize): ... here.
24589         (tree_ssa_lim): Call fill_always_executed_in.
24591 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24593         * postreload.c (reload_combine): Fix code detecting returns.
24595 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24597         * function.c (emit_use_return_register_into_block): On cc0 targets,
24598         do not emit the sequence between cc0 setter and user.
24600 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24602         * config/i386/predicates.md (local_symbolic_operand): Interpret
24603         dll-imported symbols as none-local.
24605 2013-03-25  Richard Biener  <rguenther@suse.de>
24607         * tree-ssa-loop-im.c (struct depend): Remove.
24608         (struct lim_aux_data): Make depends a vec of gimples.
24609         (free_lim_aux_data): Adjust.
24610         (add_dependency): Likewise.
24611         (set_level): Likewise.
24613 2013-03-25  Richard Biener  <rguenther@suse.de>
24615         PR middle-end/56434
24616         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
24617         the pointer returned by calls with ECF_MALLOC set.
24619 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24621         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
24623 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24625         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
24626         using MMXMODE mode iterator.
24627         (*move<mode>_internal): Merge with *movv2sf_internal and
24628         *movv2sf_internal_rex64 using MMXMODE mode iterator.
24630 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24632         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
24633         (record_last_mem_set_info): Likewise.
24635         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
24636         of XNEWVEC followed by memset.
24637         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
24639 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24641         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
24642         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
24643         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
24644         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
24645         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
24646         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
24647         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
24648         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
24649         BARRIER_P instead of GET_CODE.
24651 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24653         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
24654         inaccuracy in the probing code.
24656         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
24657         (ctrapdi4): Likewise.
24659 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24661         * calls.c (expand_call): Add missing guard to code handling return
24662         of non-BLKmode structures in MSB.
24663         * function.c (expand_function_end): Likewise.
24665 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24667         * combine.c (try_combine): Adjust comment.  Do not add the set of
24668         insn #0 if the destination indirectly is set or dies in insn #2.
24669         Tidy up code to distribute a new note.
24671 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24673         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
24674         also for alternatives 16 and 17.
24676 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24678         * config/i386/sse.md (*mov<mode>_internal): Merge with
24679         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
24680         Emit insn template depending on type attribute.  Use
24681         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24682         movd instead of movq mnemonic for interunit moves.  Rewrite mode
24683         attribute calculation.  Remove unit attribute calculation.
24684         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
24685         Set prefix_data16 attribute for DImode ssemov types.
24686         Use Ym instead of y for SSE-MMX conversion alternatives.
24687         Reorder operand constraints.
24689 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
24691         * df.h (df_insn_delete): Adjust prototype.
24692         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
24693         and let it decide whether mark the basic block dirty.
24694         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
24695         * df-scan.c (df_insn_info_delete): New helper function, split
24696         off from df_insn_delete.
24697         (df_scan_free_bb_info): Use it.
24698         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
24699         Likewise.
24700         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
24701         that the insn is actually an insn and it has a non-NULL basic block.
24702         Do not mark basic block dirty if only deleting a DEBUG_INSN.
24704 2013-03-22  Richard Biener  <rguenther@suse.de>
24706         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
24707         dep_ref members.
24708         (mem_ref_alloc): Do not allocate them.
24709         (refs_independent_p): Do not query or maintain a cache.
24711 2013-03-22  Richard Biener  <rguenther@suse.de>
24713         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
24714         (gather_mem_refs_in_loops): Do not compute it.
24715         (analyze_memory_references): Do not allocate it.
24716         (tree_ssa_lim_finalize): Do not free it.
24717         (for_all_locs_in_loop): Do not query all_refs_in_loop.
24719 2013-03-22  Richard Biener  <rguenther@suse.de>
24721         * is-a.h (as_a): Use gcc_checking_assert.
24723 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
24725         * config/aarch64/aarch64.c (aarch64_print_operand): New
24726         format specifier for printing a constant in hex.
24727         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
24728         format specifier for printing second operand.
24730 2013-03-22  Richard Biener  <rguenther@suse.de>
24732         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
24733         bitmaps.
24734         (gather_mem_refs_in_loops): Perform store accumulation here.
24735         (create_vop_ref_mapping_loop): Remove.
24736         (create_vop_ref_mapping): Likewise.
24737         (analyze_memory_references): Initialize refs_stored_in_loop.
24738         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
24739         (record_indep_loop): Remove.
24740         (record_dep_loop): New function.
24741         (ref_indep_loop_p_1): Adjust to only walk over references
24742         in the loop, not its subloops.
24743         (ref_indep_loop_p): Rename to ...
24744         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
24745         maintaining a more fine-grained cache.
24746         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
24747         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
24749 2013-03-22  Richard Biener  <rguenther@suse.de>
24751         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
24752         (struct mem_ref): Make accesses_in_loop a vec of a vec of
24753         aggregate mem_ref_loc.
24754         (free_mem_ref_locs): Inline into ...
24755         (memref_free): ... this and adjust.
24756         (mem_ref_alloc): Adjust.
24757         (mem_ref_locs_alloc): Remove.
24758         (record_mem_ref_loc): Adjust.
24759         (get_all_locs_in_loop): Rewrite into ...
24760         (for_all_locs_in_loop): ... this iterator.
24761         (rewrite_mem_ref_loc): New functor.
24762         (rewrite_mem_refs): Use for_all_locs_in_loop.
24763         (sm_set_flag_if_changed): New functor.
24764         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
24765         (ref_always_accessed): New functor.
24766         (ref_always_accessed_p): Use for_all_locs_in_loop.
24768 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24770         * tree-pass.h (PROP_gimple_lvec): New.
24771         * passes.c (dump_properties): Handle PROP_gimple_lvec.
24772         (init_optimization_passes): Move pass_lower_vector.
24773         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
24774         PROP_gimple_lvec.
24775         (pass_lower_vector): Provide PROP_gimple_lvec.
24776         (pass_lower_vector_ssa): Likewise.
24777         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
24779 2013-03-21  Mark Wielaard  <mjw@redhat.com>
24781         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
24783 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
24785         * config/i386/i386.md (*movdi_internal): Disparage slightly
24786         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
24787         conversion alternatives.
24789 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
24791         PR middle-end/48087
24792         * diagnostic.def (DK_WERROR): New kind.
24793         * diagnostic.h (werrorcount): Define.
24794         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
24795         promoted to DK_ERROR, increment DK_WERROR counter instead of
24796         DK_ERROR counter.
24797         * toplev.c (toplev_main): Call print_ignored_options even if
24798         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
24799         even if just werrorcount is non-zero.
24801         PR debug/55608
24802         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
24803         on failure.
24804         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
24805         (string_cst_pool_decl): New function.
24806         (optimize_one_addr_into_implicit_ptr): New function.
24807         (resolve_addr_in_expr): Optimize DWARF location expression
24808         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
24809         which doesn't live in memory, but has DW_AT_location or
24810         DW_AT_const_value, or refers to a string literal, into
24811         DW_OP_GNU_implicit_pointer.
24812         (optimize_location_into_implicit_ptr): New function.
24813         (resolve_addr): If removing DW_AT_location of a variable because
24814         it was DW_OP_addr of address of the variable, but the variable doesn't
24815         live in memory, try to emit const value attribute for the initializer.
24817 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24819         * tree.h (VECTOR_TYPE_P): New macro.
24820         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
24821         TYPE_MODE): Use it.
24822         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
24823         VEC_COND_EXPR cannot be lvalues.
24824         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
24826 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
24828         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
24829         Restrict the transformation to equal modes.
24831 2013-03-21  Richard Biener  <rguenther@suse.de>
24833         PR tree-optimization/39326
24834         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
24835         (MEM_ANALYZABLE): Adjust.
24836         (record_mem_ref_loc): Move bitmap ops ...
24837         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
24838         unanalyzable refs, do not record locations for it.
24839         (analyze_memory_references): Allocate ref zero as shared
24840         unanalyzable ref.
24841         (refs_independent_p): Do not test for unanalyzed mems here.
24842         (ref_indep_loop_p_1): Special-case disambiguation against
24843         the unanalyzed ref.
24844         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
24846 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
24848         * config/arm/arm-protos.h (tune_params): Add
24849         prefer_neon_for_64bits field.
24850         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
24851         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
24852         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
24853         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
24854         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
24855         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
24856         (arm_option_override): Handle -mneon-for-64bits new option.
24857         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
24858         (prefer_neon_for_64bits): Declare new variable.
24859         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
24860         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
24861         (arch_enabled): Handle new arch types. Remove support for onlya8
24862         and nota8.
24863         (one_cmpldi2): Use new arch names.
24864         (zero_extend<mode>di2, extend<mode>di2): Ditto.
24865         * config/arm/arm.opt (mneon-for-64bits): Add option.
24866         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
24867         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
24868         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
24869         of onlya8.
24870         * doc/invoke.texi (-mneon-for-64bits): Document.
24872 2013-03-21  Richard Biener  <rguenther@suse.de>
24874         PR tree-optimization/39326
24875         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
24876         (sort_bbs_in_loop_postorder_cmp): New function.
24877         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
24879 2013-03-21  Richard Biener  <rguenther@suse.de>
24881         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
24882         (vect_insert_into_interleaving_chain): Likewise.
24883         (vect_drs_dependent_in_basic_block): Inline ...
24884         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
24885         split out from ...
24886         (vect_analyze_data_ref_dependence): ... here.  Simplify.
24887         (vect_check_interleaving): Simplify.
24888         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
24889         (vect_slp_analyze_data_ref_dependences): ... this new function.
24890         (dr_group_sort_cmp): New function.
24891         (vect_analyze_data_ref_accesses): Compute data-reference groups
24892         here instead of in vect_analyze_data_ref_dependence.  Use
24893         a more efficient algorithm.
24894         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
24895         vect_slp_analyze_data_ref_dependences.  Call
24896         vect_analyze_data_ref_accesses earlier.
24897         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
24898         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
24899         (vect_slp_analyze_data_ref_dependences): New prototype.
24901 2013-03-21  Richard Biener  <rguenther@suse.de>
24903         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
24904         ref is stored in the loop.
24905         (find_refs_for_sm): Walk only over all stores.
24906         (store_motion_loop): Allocate from lim_bitmap_obstack.
24907         (store_motion): Likewise.
24909 2013-03-21  Richard Biener  <rguenther@suse.de>
24911         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
24912         Update virtual SSA form.
24914 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
24916         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
24917         * configure: Regenerate.
24918         * config.in: Regenerate.
24919         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
24920         if !HAVE_LD_EH_FRAME_CIEV3.
24922 2013-03-21  Richard Biener  <rguenther@suse.de>
24924         * tree-cfg.c (verify_expr_no_block): New function.
24925         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
24926         nor DECL_VALUE_EXPR have locations with associated blocks.
24927         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
24928         (clear_unused_block_pointer): Remove code dealing with
24929         blocks in DECL_DEBUG_EXPR locations.
24931 2013-03-21  Richard Biener  <rguenther@suse.de>
24933         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
24934         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
24935         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
24936         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
24937         instead of DECL_DEBUG_EXPR_IS_FROM.
24938         * gimplify.c (gimplify_modify_expr): Likewise.
24939         * tree-cfg.c (verify_expr_location_1): Likewise.
24940         * tree-complex.c (create_one_component_var): Likewise.
24941         * tree-sra.c (create_access_replacement): Likewise.
24942         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
24943         (clear_unused_block_pointer): Likewise.
24944         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
24945         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
24946         * var-tracking.c (var_debug_decl): Likewise.
24947         (track_expr_p): Likewise.
24948         * tree-inline.c (add_local_variables): Likewise.  Set
24949         DECL_HAS_DEBUG_EXPR_P after copying it.
24950         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
24951         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
24953 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
24955         PR bootstrap/56656
24956         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
24957         * configure: Regenerate.
24958         * config.in: Regenerate.
24959         * config/i386/i386.md (*movdf_internal): Use
24960         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24961         movd instead of movq mnemonic for interunit moves.
24962         (*movdi_internal): Ditto.
24964 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24966         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
24967         (abd<mode>_3): New pattern.
24968         (aba<mode>_3): New pattern.
24969         (fabd<mode>_3): New pattern.
24971 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24973         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
24974         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
24975         occurrence of REGISTER_PREFIX as its empty string.
24977 2013-03-20  Jeff Law  <law@redhat.com>
24979         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
24980         addititional equivalences for equality comparisons between an SSA_NAME
24981         and a constant where the SSA_NAME was set from a widening conversion.
24983 2013-03-20  Walter Lee  <walt@tilera.com>
24985         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
24987 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
24989         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
24990         depending on type attribute.
24991         (*movti_internal): Ditto.
24992         (*movtf_internal): Ditto.
24993         (*movxf_internal): Ditto.
24994         (*movdf_internal): Ditto.
24995         (*movsf_internal): Ditto.
24997 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
24999         * config/i386/i386.md (*movti_internal): Set prefix attribute to
25000         maybe_vex for sselog1 and ssemov types.
25001         (*movdi_internal): Reorder operand constraints.
25002         (*movsi_internal): Ditto.  Set prefix attribute to
25003         maybe_vex for sselog1 and ssemov types.
25004         (*movtf_internal): Set prefix attribute to maybe_vex
25005         for sselog1 and ssemov types.
25006         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
25007         DImode ssemov types.  Reorder operand constraints.
25008         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
25009         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
25010         attribute for SImode ssemov types.  Reorder operand constraints.
25012 2013-03-20  Martin Jambor  <mjambor@suse.cz>
25014         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
25015         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
25017 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
25019         * config/rs6000/predicates.md (indexed_address, update_address_mem
25020         update_indexed_address_mem): New predicates.
25021         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
25022         attribute for load/store instructions.
25023         * config/rs6000/dfp.md (movsd_store): Likewise.
25024         (movsd_load): Likewise.
25025         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
25026         (unnamed HI->DI extend define_insn): Likewise.
25027         (unnamed SI->DI extend define_insn): Likewise.
25028         (unnamed QI->SI extend define_insn): Likewise.
25029         (unnamed QI->HI extend define_insn): Likewise.
25030         (unnamed HI->SI extend define_insn): Likewise.
25031         (unnamed HI->SI extend define_insn): Likewise.
25032         (extendsfdf2_fpr): Likewise.
25033         (movsi_internal1): Likewise.
25034         (movsi_internal1_single): Likewise.
25035         (movhi_internal): Likewise.
25036         (movqi_internal): Likewise.
25037         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
25038         attribute for load/store instructions.
25039         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
25040         instructions.
25041         (mov<mode>_softfloat): Likewise.
25042         (mov<mode>_hardfloat32): Likewise.
25043         (mov<mode>_hardfloat64): Likewise.
25044         (mov<mode>_softfloat64): Likewise.
25045         (movdi_internal32): Likewise.
25046         (movdi_internal64): Likewise.
25047         (probe_stack_<mode>): Likewise.
25049 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
25051         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
25052         floating point, and decimal floating point to reload iterator.
25054         * config/rs6000/constraints.md (wl constraint): New constraints to
25055         return FLOAT_REGS if certain options are used to reduce the number
25056         of separate patterns that exist in the file.
25057         (wx constraint): Likewise.
25058         (wz constraint): Likewise.
25060         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25061         -mdebug=reg, print wg, wl, wx, and wz constraints.
25062         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25063         Initialize the reload functions for 64-bit binary/decimal floating
25064         point types.
25065         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25066         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25067         create the buffer on the stack to overcome not having a 32-bit
25068         load and store.
25069         (rs6000_emit_move): Likewise.
25070         (rs6000_secondary_memory_needed_rtx): Likewise.
25071         (rs6000_alloc_sdmode_stack_slot): Likewise.
25072         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25073         via xxlxor, just like DFmode 0.0.
25075         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
25076         define as 1 if we are running on a power7 or newer.
25077         (enum r6000_reg_class_enum): Add new constraints.
25079         * config/rs6000/dfp.md (movsd): Delete, combine with binary
25080         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25081         with other moves by using conditional constraits (wg).  Use LFIWZX
25082         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
25083         (movsd splitter): Likewise.
25084         (movsd_hardfloat): Likewise.
25085         (movsd_softfloat): Likewise.
25087         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
25088         binary and decimal floating point moves.
25089         (fmove_ok): New attributes to combine binary and decimal floating
25090         point moves, and to combine power6x (mfpgpr) moves along normal
25091         floating moves.
25092         (real_value_to_target): Likewise.
25093         (f32_lr): Likewise.
25094         (f32_lm): Likewise.
25095         (f32_li): Likewise.
25096         (f32_sr): Likewise.
25097         (f32_sm): Likewise.
25098         (f32_si): Likewise.
25099         (movsf): Combine binary and decimal floating point moves.  Combine
25100         power6x (mfpgpr) moves with other moves by using conditional
25101         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
25102         (mov<mode> for SFmode/SDmode); Likewise.
25103         (SFmode/SDmode splitters): Likewise.
25104         (movsf_hardfloat): Likewise.
25105         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
25106         (movsf_softfloat): Likewise.
25107         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
25109         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
25110         wx and wz constraints.
25112         * config/rs6000/constraints.md (wg constraint): New constraint to
25113         return FLOAT_REGS if -mmfpgpr (power6x) was used.
25115         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
25116         constraint.
25118         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25119         -mdebug=reg, print wg, wl, wx, and wz constraints.
25120         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25121         Initialize the reload functions for 64-bit binary/decimal floating
25122         point types.
25123         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25124         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25125         create the buffer on the stack to overcome not having a 32-bit
25126         load and store.
25127         (rs6000_emit_move): Likewise.
25128         (rs6000_secondary_memory_needed_rtx): Likewise.
25129         (rs6000_alloc_sdmode_stack_slot): Likewise.
25130         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25131         via xxlxor, just like DFmode 0.0.
25133         * config/rs6000/dfp.md (movdd): Delete, combine with binary
25134         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25135         with other moves by using conditional constraits (wg).  Use LFIWZX
25136         and STFIWX for loading SDmode on power7.
25137         (movdd splitters): Likewise.
25138         (movdd_hardfloat32): Likewise.
25139         (movdd_softfloat32): Likewise.
25140         (movdd_hardfloat64_mfpgpr): Likewise.
25141         (movdd_hardfloat64): Likewise.
25142         (movdd_softfloat64): Likewise.
25144         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
25145         64-bit binary and decimal floating point moves.
25146         (FMOVE64X): Likewise.
25147         (movdf): Combine 64-bit binary and decimal floating point moves.
25148         Combine power6x (mfpgpr) moves with other moves by using
25149         conditional constraits (wg).
25150         (mov<mode> for DFmode/DDmode): Likewise.
25151         (DFmode/DDmode splitters): Likewise.
25152         (movdf_hardfloat32): Likewise.
25153         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
25154         (movdf_softfloat32): Likewise.
25155         (movdf_hardfloat64_mfpgpr): Likewise.
25156         (movdf_hardfloat64): Likewise.
25157         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
25158         (movdf_softfloat64): Likewise.
25159         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
25160         (reload_<mode>_load): Move to later in the file so they aren't in
25161         the middle of the floating point move insns.
25162         (reload_<mode>_store): Likewise.
25164         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
25165         constraint.
25167         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
25168         constraint if -mdebug=reg.
25169         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
25170         Enable using dd reload support if needed.
25172         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
25173         binary and decimal floating point moves in rs6000.md.
25174         (movtd_internal): Likewise.
25176         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
25177         decimal floating point moves.
25178         (movtf): Likewise.
25179         (movtf_internal): Likewise.
25180         (mov<mode>_internal, TDmode/TFmode): Likewise.
25181         (movtf_softfloat): Likewise.
25182         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
25184         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
25185         movdi_internal64, using wg constraint for move direct operations.
25186         (movdi_internal64): Likewise.
25188         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
25189         MODES_TIEABLE_P for selected modes.  Print the numerical value of
25190         the various virtual registers. Use GPR/FPR first/last values,
25191         instead of hard coding the register numbers.  Print which modes
25192         have reload functions registered.
25193         (rs6000_option_override_internal): If -mdebug=reg, trace the options
25194         settings before/after setting cpu, target and subtarget settings.
25195         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
25196         and for secondary reload failures in rs6000_secondary_reload_inner.
25197         (rs6000_secondary_reload_fail): Likewise.
25198         (rs6000_secondary_reload_inner): Likewise.
25200         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
25201         macros for first/last GPR and FPR registers.
25202         (LAST_GPR_REGNO): Likewise.
25203         (FIRST_FPR_REGNO): Likewise.
25204         (LAST_FPR_REGNO): Likewise.
25206         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
25207         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
25208         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
25209         (vcond<mode><mode>): Likewise.
25210         (vcondu<mode><mode>): Likewise.
25211         (vector_gtu<mode>): Likewise.
25212         (vector_gte<mode>): Likewise.
25213         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
25214         to prevent the compiler from converting DImode operations to TImode.
25215         (ior<mode>3): Likewise.
25216         (and<mode>3): Likewise.
25217         (one_cmpl<mode>2): Likewise.
25218         (nor<mode>3): Likewise.
25219         (andc<mode>3): Likewise.
25221         * config/rs6000/constraints.md (wt constraint): New constraint
25222         that returns VSX_REGS if TImode is allowed in VSX registers.
25224         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
25225         constant under VSX.
25227         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
25228         similar to TImode, but it is restricted to being in the GPRs.
25230         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
25231         TImode to occupy a single VSX register.
25233         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
25234         -mvsx-timode for power7/power8.
25235         (power7 cpu): Likewise.
25236         (power8 cpu): Likewise.
25238         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
25239         sure that TFmode/TDmode take up two registers if they are ever
25240         allowed in the upper VSX registers.
25241         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
25242         registers.
25243         (rs6000_init_hard_regno_mode_ok): Likewise.
25244         (rs6000_debug_reg_global): Add debugging for PTImode and wt
25245         constraint.  Print if LRA is turned on.
25246         (rs6000_option_override_internal): Give an error if -mvsx-timode
25247         and VSX is not enabled.
25248         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
25249         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
25250         to reg+offset addressing.  Use PTImode when checking offset
25251         addresses for validity.
25252         (reg_offset_addressing_ok_p): Likewise.
25253         (rs6000_legitimate_offset_address_p): Likewise.
25254         (rs6000_legitimize_address): Likewise.
25255         (rs6000_legitimize_reload_address): Likewise.
25256         (rs6000_legitimate_address_p): Likewise.
25257         (rs6000_eliminate_indexed_memrefs): Likewise.
25258         (rs6000_emit_move): Likewise.
25259         (rs6000_secondary_reload): Likewise.
25260         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
25261         reloads to fpr registers to continue to use reg+offset addressing,
25262         but 64-bit reloads to altivec registers need reg+reg addressing.
25263         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
25264         it.  Treat LO_SUM like a PLUS operation.
25265         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
25266         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
25267         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
25268         registers to share a register with a smaller sized type, since VSX
25269         puts scalars in the upper 64-bits.
25270         (print_operand): Add support for PTImode.
25271         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
25272         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
25273         registers, but don't have arithmetic support.
25274         (rs6000_memory_move_cost): Add test for VSX.
25275         (rs6000_opt_masks): Add -mvsx-timode.
25277         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
25278         for TImode.
25279         (VSs): Likewise.
25280         (VSr): Use wt constraint for TImode.
25281         (VSv): Drop TImode support.
25282         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
25283         (vsx_movti_64bit): Likewise.
25284         (vsx_movti_32bit): Likewise.
25285         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
25286         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
25287         one '?' on the appropriate output constraint.  Do not allow TImode
25288         logical operations on 32-bit systems.
25289         (vsx_ior<mode>3): Likewise.
25290         (vsx_xor<mode>3): Likewise.
25291         (vsx_one_cmpl<mode>2): Likewise.
25292         (vsx_nor<mode>3): Likewise.
25293         (vsx_andc<mode>3): Likewise.
25294         (vsx_concat_<mode>): Likewise.
25295         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
25297         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
25298         OPTION_MASK_VSX_TIMODE.
25299         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
25300         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
25302         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
25303         (TI2 iterator): New iterator for TImode, PTImode.
25304         (wd mode attribute): Add values for vector types.
25305         (movti_string): Replace TI move operations with operations for TImode
25306         and PTImode.  Add support for TImode being allowed in VSX registers.
25307         (mov<mode>_string, TImode/PTImode): Likewise.
25308         (movti_ppc64): Likewise.
25309         (mov<mode>_ppc64, TImode/PTImode): Likewise.
25310         (TI mode splitters): Likewise.
25312         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
25313         constraint.
25315 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
25317         PR tree-optimization/56355
25318         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
25319         Also handle integers with undefined overflow.
25321 2013-03-20  Catherine Moore  <clm@codesourcery.com>
25322             Maciej W. Rozycki  <macro@codesourcery.com>
25323             Tom de Vries  <tom@codesourcery.com>
25324             Nathan Sidwell  <nathan@codesourcery.com>
25325             Iain Sandoe  <iain@codesourcery.com>
25326             Nathan Froyd  <froydnj@codesourcery.com>
25327             Chao-ying Fu  <fu@mips.com>
25329         * doc/extend.texi: (micromips, nomicromips, nocompression):
25330         Document new function attributes.
25331         * doc/invoke.texi (minterlink-compressed, mmicromips,
25332         m14k, m14ke, m14kec): Document new options.
25333         (minterlink-mips16): Update documentation.
25334         * doc/md.texi (ZC, ZD): Document new constraints.
25335         * configure.ac (gcc_cv_as_micromips): Check if linker
25336         supports the .set micromips directive.
25337         * configure: Regenerate.
25338         * config.in: Regenerate.
25339         * config/mips/mips-tables.opt: Regenerate.
25340         * config/mips/micromips.md: New file.
25341         * constraints.md (ZC, ZD): New constraints.
25342         * config/mips/predicates.md (movep_src_register): New predicate.
25343         (movep_src_operand): New predicate.
25344         (non_volatile_mem_operand): New predicate.
25345         * config/mips/mips.md (multimem): New type.
25346         (length): Differentiate between 17-bit and 18-bit branch offsets.
25347         (MOVEP1, MOVEP2): New mode iterator.
25348         (mov_<load>l): Use ZC constraint.
25349         (mov_<load>r): Likewise.
25350         (mov_<store>l): Likewise.
25351         (mov_<store>r): Likewise.
25352         (*branch_equality<mode>_inverted): Add microMIPS support.
25353         (*branch_equality<mode>): Likewise.
25354         (*jump_absolute): Likewise.
25355         (indirect_jump_<mode>): Likewise.
25356         (tablejump_<mode>): Likewise.
25357         (<optab>_internal): Likewise.
25358         (sibcall_internal): Likewise.
25359         (sibcall_value_internal): Likewise.
25360         (prefetch): Use constraint ZD.
25361         * config/mips/mips.opt (minterlink-compressed): New option.
25362         (minterlink-mips16): Now an alias for minterlink-compressed.
25363         (mmicromips): New option.
25364         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
25365         (compare_and_swap_12): Likewise.
25366         (sync_add<mode>): Likewise.
25367         (sync_<optab>_12): Likewise.
25368         (sync_old_<optab>_12): Likewise.
25369         (sync_new_<optab>_12): Likewise.
25370         (sync_nand_12): Likewise.
25371         (sync_old_nand_12): Likewise.
25372         (sync_new_nand_12): Likewise.
25373         (sync_sub<mode>): Likewise.
25374         (sync_old_add<mode>): Likewise.
25375         (sync_old_sub<mode>): Likewise.
25376         (sync_new_add<mode>): Likewise.
25377         (sync_new_sub<mode>): Likewise.
25378         (sync_<optab><mode>): Likewise.
25379         (sync_old_<optab><mode>): Likewise.
25380         (sync_new_<optab><mode>): Likewise.
25381         (sync_nand<mode>): Likewise.
25382         (sync_old_nand<mode>): Likewise.
25383         (sync_new_nand<mode>): Likewise.
25384         (sync_lock_test_and_set<mode>): Likewise.
25385         (test_and_set_12): Likewise.
25386         (atomic_compare_and_swap<mode>): Likewise.
25387         (atomic_exchange<mode>_llsc): Likewise.
25388         (atomic_fetch_add<mode>_llsc): Likewise.
25389         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
25390         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
25391         (umips_save_restore_pattern_p): Likewise.
25392         (umips_load_store_pair_p): Likewise.
25393         (umips_output_load_store_pair): Likewise.
25394         (umips_movep_target_p): Likewise.
25395         (umips_12bit_offset_address_p): Likewise.
25396         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
25397         (mips_base_mips16): Rename this...
25398         (mips_base_compression_flags): ...to this. Update all uses.
25399         (mips_attribute_table): Add micromips, nomicromips and nocompression.
25400         (mips_mips16_decl_p): Delete.
25401         (mips_nomips16_decl_p): Delete.
25402         (mips_get_compress_on_flags): New function.
25403         (mips_get_compress_off_flags): New function.
25404         (mips_get_compress_mode): New function.
25405         (mips_get_compress_on_name): New function.
25406         (mips_get_compress_off_name): New function.
25407         (mips_insert_attributes): Support multiple compression types.
25408         (mips_merge_decl_attributes): Likewise.
25409         (umips_12bit_offset_address_p): New function.
25410         (mips_start_function_definition): Emit .set micromips directive.
25411         (mips_call_may_need_jalx_p): New function.
25412         (mips_function_ok_for_sibcall): Add microMIPS support.
25413         (mips_print_operand_punctuation): Support short delay slots and
25414         compact jumps.
25415         (umips_swm_mask, umips_swm_encoding): New.
25416         (umips_build_save_restore): New function.
25417         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
25418         (was_mips16_p): Remove.
25419         (old_compression_mode): New.
25420         (mips_set_compression_mode): New function.
25421         (mips_set_current_function): Add microMIPS support.
25422         (mips_option_override): Likewise.
25423         (umips_save_restore_pattern_p): New function.
25424         (umips_output_save_restore): New function.
25425         (umips_load_store_pair_p_1): New function.
25426         (umips_load_store_pair_p): New function.
25427         (umips_output_load_store_pair_1): New function.
25428         (umips_output_load_store_pair): New function.
25429         (umips_movep_target_p) New function.
25430         (mips_prepare_pch_save): Add microMIPS support.
25431         * config/mips/mips.h (TARGET_COMPRESSION): New.
25432         (TARGET_CPU_CPP_BUILTINS): Update macro
25433         to use new compression flags and to support microMIPS.
25434         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
25435         (MIPS_ARCH_FLOAT_SPEC): Likewise.
25436         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
25437         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
25438         (ASM_SPEC): Support mmicromips and mno-micromips.
25439         (M16STORE_REG_P): New macro.
25440         (MIPS_CALL): Support TARGET_MICROMIPS.
25441         (MICROMIPS_J): New macro.
25442         (mips_base_mips16): Rename this...
25443         (mips_base_compression_flags): ...to this.
25444         (UMIPS_12BIT_OFFSET_P): New macro.
25445         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
25446         (MULTILIB_DIRNAMES): Likewise.
25447 2013-03-20  Richard Biener  <rguenther@suse.de>
25449         PR tree-optimization/56661
25450         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
25451         the result does not have to be distinct.
25453 2013-03-20  Richard Biener  <rguenther@suse.de>
25455         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
25456         remap_gimple_op_r.
25458 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
25459             Steven Bosscher  <steven@gcc.gnu.org>
25461         PR rtl-optimization/56605
25462         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
25464 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25466         PR bootstrap/56656
25467         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
25468         that require movd instead of movq.
25470 2013-03-20  Richard Biener  <rguenther@suse.de>
25472         * tree-ssa-structalias.c (struct variable_info): Add pointer
25473         to the first field of an aggregate with sub-vars.  Make
25474         this and the pointer to the next subfield its ID.
25475         (vi_next): New function.
25476         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
25477         storedanything_id, integer_id): Increment by one.
25478         (new_var_info, get_call_vi, lookup_call_clobber_vi,
25479         get_call_clobber_vi): Adjust.
25480         (solution_set_expand): Simplify and speedup.
25481         (solution_set_add): Inline into ...
25482         (set_union_with_increment): ... this.  Adjust accordingly.
25483         (do_sd_constraint): Likewise.
25484         (do_ds_constraint): Likewise.
25485         (do_complex_constraint): Simplify.
25486         (build_pred_graph): Adjust.
25487         (solve_graph): Likewise.  Simplify and speedup.
25488         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
25489         get_constraint_for_component_ref, get_constraint_for_1,
25490         first_vi_for_offset, first_or_preceding_vi_for_offset,
25491         create_function_info_for, create_variable_info_for_1,
25492         create_variable_info_for, intra_create_variable_infos): Adjust.
25493         (init_base_vars): Push NULL for ID zero.
25494         (compute_points_to_sets): Adjust.
25496 2013-03-20  Richard Biener  <rguenther@suse.de>
25498         * cfgloop.c (verify_loop_structure): Streamline and avoid
25499         ICEing on corrupt loop tree.
25500         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
25501         loop tree.
25503 2013-03-20  Richard Biener  <rguenther@suse.de>
25505         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
25506         check whether an SSA update is needed.
25508 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
25510         * config/mips/constraints.md (T): Rename to...
25511         (Yf): ...this.
25512         (U): Rename to...
25513         (Yd): ...this.
25514         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
25515         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
25517 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25519         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
25520         (*subsi3_carryin_uxtw): Likewise.
25522 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25524         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
25525         (*rorsi3_insn_uxtw): Likewise.
25527 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25529         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
25530         (*extrsi5_insn_uxtw): Likewise.
25532 2013-03-19  Richard Biener  <rguenther@suse.de>
25534         PR tree-optimization/56273
25535         * passes.c (init_optimization_passes): Move second VRP after DOM.
25537 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
25539         * config/i386/i386.md (*movti_internal): Merge from
25540         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
25541         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
25542         nox64 isa attributes.
25544 2013-03-18  Richard Biener  <rguenther@suse.de>
25546         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
25547         (unite): Likewise.
25548         (merge_node_constraints): Likewise.
25549         (build_succ_graph): Likewise.
25550         (valid_graph_edge): Inline into single caller.
25551         (unify_nodes): Likewise.  Use bitmap_set_bit return value
25552         and cache varinfo.
25553         (scc_visit): Fix formatting and variable use.
25554         (do_sd_constraint): Use gcc_checking_assert.
25555         (do_ds_constraint): Likewise.
25556         (do_complex_constraint): Likewise.
25557         (condense_visit): Likewise.  Cleanup.
25558         (dump_pred_graph): New function.
25559         (perform_var_substitution): Dump the pred-graph before
25560         variable substitution.
25561         (find_equivalent_node): Use gcc_checking_assert.
25562         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
25564 2013-03-18  Richard Biener  <rguenther@suse.de>
25566         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
25567         Remove cond_expr_stmt_list argument and do not gimplify the
25568         built expression.
25569         (vect_loop_versioning): Adjust.
25570         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
25571         Cleanup to use less temporaries.
25572         (vect_create_data_ref_ptr): Cleanup.
25574 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25576         PR tree-optimization/56635
25577         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
25578         require types_compatible_p types.
25580 2013-03-18  Nick Clifton  <nickc@redhat.com>
25582         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
25583         spurious backslash.
25585         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
25586         Add missing line to comment describing function.
25588 2013-03-18  Richard Biener  <rguenther@suse.de>
25590         PR tree-optimization/56210
25591         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25592         Handle string / character search functions.
25593         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
25595 2013-03-18  Richard Biener  <rguenther@suse.de>
25597         PR middle-end/56483
25598         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
25599         and implement properly.
25600         * gimple.h (gimple_cond_single_var_p): Remove.
25602 2013-03-18  Richard Biener  <rguenther@suse.de>
25604         * tree-data-ref.h (find_data_references_in_loop): Declare.
25605         * tree-data-ref.c (get_references_in_stmt): Use a stack
25606         vector pre-allocated in the callers.
25607         (find_data_references_in_stmt): Adjust.
25608         (graphite_find_data_references_in_stmt): Likewise.
25609         (create_rdg_vertices): Likewise.
25610         (find_data_references_in_loop): Export.
25611         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
25612         Compute dependences here...
25613         (vect_analyze_data_refs): ...not here.  When we encounter
25614         a non-vectorizable data reference in basic-block vectorization
25615         truncate the data reference vector.  Do not bother to
25616         fixup data-dependence information for gather loads.
25617         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
25618         of data references, as reported.
25620 2013-03-18  Richard Biener  <rguenther@suse.de>
25622         PR tree-optimization/3713
25623         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
25624         has_constants and expr.
25625         (stmt_has_constants): Properly valueize SSA names when deciding
25626         whether the stmt has constants.
25628 2013-03-18  Richard Biener  <rguenther@suse.de>
25630         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
25631         whole function when there is nothing to do.
25632         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
25633         * tree-vectorizer.c (vectorize_loops): Update virtual and
25634         loop-closed SSA once.
25635         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
25637 2013-03-18  Richard Biener  <rguenther@suse.de>
25639         PR middle-end/56113
25640         * domwalk.c (bb_postorder): New global static.
25641         (cmp_bb_postorder): New function.
25642         (walk_dominator_tree): Replace scheme imposing an order for
25643         visiting dominator sons by one sorting them at the time they
25644         are pushed on the stack.
25646 2013-03-18  Richard Biener  <rguenther@suse.de>
25648         PR tree-optimization/39326
25649         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
25650         (struct mem_ref): Replace mem member with ao_ref typed member.
25651         (MEM_ANALYZABLE): Adjust.
25652         (memref_eq): Likewise.
25653         (mem_ref_alloc): Likewise.
25654         (gather_mem_refs_stmt): Likewise.
25655         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
25656         (execute_sm_if_changed_flag_set): Adjust.
25657         (execute_sm): Likewise.
25658         (ref_always_accessed_p): Likewise.
25659         (refs_independent_p): Likewise.
25660         (can_sm_ref_p): Likewise.
25662 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25664         PR c/56566
25665         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
25666         return 1 even for !unsignedp.
25668 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25670         * config/i386/i386.md (isa): Add x64 and nox64.
25671         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
25672         (*pushtf): Enable *roF alternative for x64 isa only.
25673         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
25674         mode attribute of integer alternatives to DImode for TARGET_64BIT.
25675         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
25676         (*movtf_internal): Merge from *movtf_internal_rex64 and
25677         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
25678         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
25679         nox64 isa attributes.
25680         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
25681         nox64 isa attributes.
25682         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
25684 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25686         * config/alpha/alpha.c (TARGET_LRA_P): New define.
25688 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
25690         PR target/56640
25691         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
25692         class names.  Remove trailing comma after "ALL_REGS".
25694 2013-03-16  Jan Hubicka  <jh@suse.cz>
25696         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
25697         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
25698         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
25699         of cgraph_get_create_node.
25700         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
25702 2013-03-16  Jason Merrill  <jason@redhat.com>
25704         PR debug/49090
25705         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
25706         with DW_AT_default_value.
25708 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
25710         * BASE-VER: Set to 4.9.0.
25712 2013-03-14  Andi Kleen  <ak@linux.intel.com>
25714         PR target/56619
25715         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
25716         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
25717         Document _x* TSX intrinsics.
25719 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
25720             David Holsgrove  <david.holsgrove@xilinx.com>
25722         * configure.ac: Add MicroBlaze TLS support detection.
25723         * configure: Regenerate.
25724         * config/microblaze/microblaze-protos.h
25725         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
25726         symbol_mentioned_p, label_mentioned_p): Add prototypes.
25727         * config/microblaze/microblaze.c (microblaze_address_type): Add
25728         ADDRESS_TLS and tls_reloc address types.
25729         (microblaze_address_info): Add tls_reloc.
25730         (TARGET_HAVE_TLS): Define.
25731         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
25732          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
25733          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
25734         load_tls_operand,  microblaze_call_tls_get_addr,
25735         microblaze_legitimize_tls_address): New functions.
25736         (microblaze_classify_unspec): Handle UNSPEC_TLS.
25737         (get_base_reg): Use microblaze_tls_symbol_p.
25738         (microblaze_classify_address): Handle TLS.
25739         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
25740         label_mentioned_p and microblaze_tls_referenced_p.
25741         (microblaze_legitimize_address): Handle TLS.
25742         (microblaze_address_insns): Handle ADDRESS_TLS.
25743         (pic_address_needs_scratch): Handle TLS.
25744         (print_operand_address): Handle TLS.
25745         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
25746         (microblaze_expand_move): Handle TLS.
25747         (microblaze_legitimate_constant_p): Check
25748         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
25749         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
25750         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
25751         (PIC_OFFSET_TABLE_REGNUM): Set.
25752         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
25753         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
25754         (addsi3, movsi_internal2, movdf_internal): Update constraints
25755         * config/microblaze/predicates.md (arith_plus_operand): Define
25756         (move_operand): Redefine as move_src_operand,
25757         check microblaze_tls_referenced_p.
25759 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25761         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
25762         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
25764 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25766         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
25767         CC mode for AND.
25769 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
25771         PR tree-optimization/53265
25772         * common.opt (Waggressive-loop-optimizations): New option.
25773         * tree-ssa-loop-niter.c: Include tree-pass.h.
25774         (do_warn_aggressive_loop_optimizations): New function.
25775         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
25776         if number_of_latch_executions returned constant.
25777         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
25778         early.  If number_of_latch_executions returned constant, set
25779         nb_iterations_upper_bound back to it.
25780         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
25781         field.
25782         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
25783         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
25785         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
25786         (MULTILIB_OSDIRNAMES): Set.
25787         * genmultilib: If defaultosdirname doesn't start with :: , set
25788         defaultosdirname2 instead, clear it and emit two . multilib_raw
25789         entries instead of just one.
25791 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
25793         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
25794         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
25795         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
25796         (SUBTARGET_OVERRIDE_OPTIONS): New.
25798 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
25800         PR target/49880
25801         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
25802         (musermode): Convert to Var(TARGET_USERMODE).
25803         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
25804         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
25805         * config/sh/sh.c (sh_option_override): Use
25806         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
25807         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
25808         condition.
25809         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
25810         TARGET_SH4.
25811         (udivsi3_i4_single, divsi3_i4_single): Use
25812         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
25814 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
25816         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
25817         default setting.
25819 2013-03-13  Richard Biener  <rguenther@suse.de>
25821         PR tree-optimization/56608
25822         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
25823         calls when vectorizing basic-blocks.
25825 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
25827         PR plugins/45078
25828         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
25829         tm_file.
25831 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
25833         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
25835 2013-03-11  Jan Hubicka  <jh@suse.cz>
25837         PR lto/56557
25838         * lto-streamer-out.c (output_symbol_p): Skip references from
25839         constructors of external variables.
25841 2013-03-11  Jan Hubicka  <jh@suse.cz>
25843         PR middle-end/56571
25844         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
25845         from pseudos.
25846         * emit-rtl.c (verify_rtx_sharing): Likewise.
25847         (copy_insn_1): Likewise.
25848         * rtl.c (copy_rtx): Likewise.
25850 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
25852         PR target/56591
25853         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
25854         output_operand_lossage message.
25856 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
25858         PR target/56470
25859         * arm.c (shift_op): Validate RTL pattern on the fly.
25860         (arm_print_operand, case 'S'): Don't use shift_operator to validate
25861         the RTL.
25863 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25865         PR target/56347
25866         * config/pa/pa.md (call_value): Check for calls to powf and direct to
25867         new call patterns that clobber %fr12.
25868         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
25869         split and postreload patterns.
25870         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
25871         registers %fr12 and %fr12R as call used.
25873 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
25875         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
25876         (canon_address, record_store, replace_read, check_mem_read_rtx,
25877         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
25878         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
25879         rest_of_handle_dse): Likewise.
25881 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
25883         PR middle-end/56524
25884         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
25885         Add base_optabs.
25886         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
25887         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
25888         (save_optabs_if_changed): Replace with...
25889         (init_tree_optimization_optabs): ...this.
25890         * optabs.c (save_optabs_if_changed): Rename to...
25891         (init_tree_optimization_optabs): ...this.  Take the optimization node
25892         as argument.  Do nothing if the base optabs are already correct.
25893         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
25894         to recompute optabs.
25895         * function.h (function): Remove optabs field.
25896         * function.c (invoke_set_current_function_hook): Call
25897         init_tree_optimization_optabs.  Use the result to initialize
25898         this_fn_optabs.
25900 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
25902         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
25903         if GTMA_HAS_NO_INSTRUMENTATION.
25904         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
25905         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
25906         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
25907         * gimple-pretty-print.c (dump_gimple_transaction): Handle
25908         GTMA_HAS_NO_INSTRUMENTATION.
25910 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
25912         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
25913         libasan_preinit.o.
25915 2013-03-08  Marek Polacek  <polacek@redhat.com>
25916             Jakub Jelinek  <jakub@redhat.com>
25918         PR tree-optimization/56478
25919         * predict.c (is_comparison_with_loop_invariant_p): Change the
25920         type of loop_step to tree.
25921         (predict_loops): Adjust.
25922         (predict_iv_comparison): Perform the computations on double_ints.
25924 2013-03-08  Richard Biener  <rguenther@suse.de>
25926         PR tree-optimization/56570
25927         * tree-cfg.c (verify_expr_location_1): Verify locations for
25928         DECL_DEBUG_EXPR.
25929         * tree-sra.c (create_access_replacement): Strip locations
25930         from DECL_DEBUG_EXPRs.
25932 2013-03-08  Richard Biener  <rguenther@suse.de>
25934         * tree-inline.c (expand_call_inline): Do not associate
25935         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
25936         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
25938 2013-03-08  Richard Biener  <rguenther@suse.de>
25940         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
25941         or block changes with -Og.  Fix for location / block encoding
25942         changes and PHI arguments with locations.
25944 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
25946         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
25947         for all counters.
25948         (struct output_info): Likewise.
25949         (register_overhead): Remove bad gcc_assert.
25950         (bitmap_find_bit): If there is only a single bitmap element, do not
25951         count a miss as a search.
25952         (print_statistics): Update for counter type changes.
25953         (dump_bitmap_statistics): Likewise.  Print headers such that they
25954         are properly lined up with the printed counters.
25956 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
25958         PR tree-optimization/56559
25959         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
25960         check that it has only a single use.
25962 2013-03-07  Richard Biener  <rguenther@suse.de>
25964         * doc/invoke.texi (fwhole-program): Discourage use in combination
25965         with -flto.
25967 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
25969         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
25971         PR tree-optimization/56539
25972         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
25973         instead of GSI_CONTINUE_LINKING as last argument to
25974         force_gimple_operand_gsi.  Adjust function comment.
25976         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
25977         aarch64-cores.def.
25979         PR middle-end/56548
25980         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
25981         promoted mode, convert the result back to the original mode.
25983 2013-03-06  Richard Biener  <rguenther@suse.de>
25985         PR middle-end/56294
25986         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
25987         (insert_updated_phi_nodes_compare_uids): New function.
25988         (update_ssa): Sort symbols_to_rename after UID before
25989         traversing it to insert PHI nodes.
25991 2013-03-06  Richard Biener  <rguenther@suse.de>
25993         PR middle-end/50494
25994         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
25995         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
25997         Revert
25998         2013-02-13  Richard Biener  <rguenther@suse.de>
26000         PR lto/50494
26001         * varasm.c (output_constant_def_1): Get the decl representing
26002         the constant as argument.
26003         (output_constant_def): Wrap output_constant_def_1.
26004         (make_decl_rtl): Use output_constant_def_1 with the decl
26005         representing the constant.
26006         (build_constant_desc): Optionally re-use a decl already
26007         representing the constant.
26008         (tree_output_constant_def): Adjust.
26010 2013-03-06  Joey Ye  <joey.ye@arm.com>
26012         PR lto/50293
26013         * gcc.c (convert_white_space): New function.
26014         (main): Handles white space in function name.
26016 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
26018         PR target/56529
26019         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
26020         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
26021         to SH_DIV_CALL_TABLE for TARGET_SH2.
26022         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
26023         list.
26024         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
26025         call-table options.
26027 2013-03-05  Sterling Augustine  <saugustine@google.com>
26028             Cary Coutant  <ccoutant@google.com>
26030         PR debug/55364
26031         * dwarf2out.c (resolve_addr): Don't call
26032         remove_loc_list_addr_table_entries a second time for the same
26033         expression.
26035 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26037         PR debug/56510
26038         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
26039         (avoid_complex_debug_insns): New function.
26040         (expand_debug_locations): Call it.
26042         PR rtl-optimization/56484
26043         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
26044         lifetimes of hard registers on small register class machines.
26046 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26048         * config/microblaze/microblaze-protos.h: Rename
26049         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
26050         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
26051         fast_interrupt.
26052         (microblaze_fast_interrupt_function_p): New function.
26053         (microblaze_is_interrupt_handler): Rename to
26054         microblaze_is_interrupt_variant and add fast_interrupt check.
26055         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
26056         (save_restore_insns): Likewise.
26057         (compute_frame_size): Likewise.
26058         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
26059         (microblaze_globalize_label): Likewise.
26060         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
26061         * config/microblaze/microblaze.md: Use wrapper
26062         microblaze_is_interrupt_variant.
26064 2013-03-05  Kai Tietz  <ktietz@redhat.com>
26066         * sdbout.c (sdbout_one_type): Switch to current function's section
26067         supporting cold/hot.
26069 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26071         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
26072         -mxl-reorder.
26074 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26076         PR middle-end/56461
26077         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
26078         if VALGRIND_GET_VBITS is defined, temporarily make object
26079         memory all defined, and restore previous valgrind addressability
26080         and definability afterwards.  Free this_object at the end.
26082         PR middle-end/56461
26083         * lra.c (lra): Call lra_clear_live_ranges if live_p,
26084         right before calling lra_create_live_ranges, also call it
26085         when clearing live_p.  Only call lra_clear_live_ranges
26086         at the end if live_p.
26088         PR middle-end/56461
26089         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
26091 2013-03-05  Richard Biener  <rguenther@suse.de>
26093         PR tree-optimization/56521
26094         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
26095         value-id.
26097 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
26099         PR c++/55135
26100         * except.h (remove_unreachable_eh_regions): New prototype.
26101         * except.c (remove_eh_handler_splicer): New function, split out
26102         of remove_eh_handler.
26103         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
26104         warning about running it on many EH regions one at a time.
26105         (remove_unreachable_eh_regions_worker): New function, walk the
26106         EH tree in depth-first order and remove non-marked regions.
26107         (remove_unreachable_eh_regions): New function.
26108         * tree-eh.c (mark_reachable_handlers): New function, split out
26109         from remove_unreachable_handlers.
26110         (remove_unreachable_handlers): Use mark_reachable_handlers and
26111         remove_unreachable_eh_regions.
26112         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
26113         and remove_unreachable_eh_regions.
26115 2013-03-05  Richard Biener  <rguenther@suse.de>
26117         PR middle-end/56525
26118         * loop-init.c (fix_loop_structure): Remove loops in two stages,
26119         not freeing them until the end.
26121 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26123         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
26125 2013-03-05  Richard Biener  <rguenther@suse.de>
26127         PR tree-optimization/56270
26128         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
26129         of loads after scheduling an SLP instance.
26131 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26133         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
26134         tic6x.exp.
26135         (check_gcc_parallelize): Run guality.exp as a separate job from
26136         vect.exp with unsorted.exp and $(dg_target_exps) separately from
26137         struct-layout-1.exp with stackalign.exp.
26139         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
26141         PR middle-end/56461
26142         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
26143         load_index sbitmap even if some bit in it isn't set.
26145         PR middle-end/56461
26146         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
26147         (discover_iteration_bound_by_body_walk): Change queues to
26148         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
26149         spelling in comment.  Call safe_push on queues[bound_index] directly.
26150         Release queues[queue_index] in every iteration unconditionally.
26151         Release bounds vector.
26153         PR middle-end/56461
26154         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
26155         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
26156         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
26157         inner_phis vector.
26159 2013-03-05  Richard Biener  <rguenther@suse.de>
26161         PR lto/56515
26162         * tree-inline.c (remap_blocks_to_null): New function.
26163         (expand_call_inline): When expanding a call stmt without
26164         an associated block inline remap all callee blocks to NULL.
26166 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26168         PR rtl-optimization/56494
26169         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
26170         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
26171         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
26173         PR middle-end/56461
26174         * sel-sched-ir.c (free_sched_pools): Release
26175         succs_info_pool.stack[succs_info_pool.max_top] vectors too
26176         if succs_info_pool.max_top isn't -1.
26178         PR bootstrap/56509
26179         * opts.c (opts_obstack, opts_concat): Moved to...
26180         * opts-common.c (opts_obstack, opts_concat): ... here.
26182 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26184         PR middle-end/56461
26185         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
26187 2013-03-04  Martin Jambor  <mjambor@suse.cz>
26189         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
26190         all appropriate places.
26192 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
26194         PR tree-optimization/56424
26195         * ipa-split.c (split_function): Do not set the RSO flag if result is
26196         not by reference and its type is a register type.
26198 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
26200         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
26201         (microblaze_legitimate_pic_operand): Likewise
26202         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
26203         new function microblaze_legitimate_pic_operand
26204         * config/microblaze/microblaze-protos.h
26205         (microblaze_legitimate_pic_operand): Declare.
26207 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26209         * config/microblaze/predicates.md (call_insn_simple_operand):
26210         New predicate for supported rtx code types.
26211         * config/microblaze/microblaze.md (call_internal1): Use
26212         call_insn_simple_operand predicate.
26214 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26216         PR middle-end/56461
26217         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
26218         partitions.ordered_remove.
26220         PR middle-end/56461
26221         * tree-vect-stmts.c (vectorizable_conversion): Don't call
26222         vec_oprnds0.create (1) for modifier == NONE.
26224         PR middle-end/56461
26225         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
26226         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
26227         vec_oprnds1 right before pushing anything to it for
26228         scalar_shift_arg.
26230         PR middle-end/56461
26231         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
26232         set nbbs to 0 instead of having separate code path.
26233         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
26234         instead of false as last argument if returning NULL.
26236 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
26238         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
26239         the attribute is now called "target" instead of "option".
26240         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
26241         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
26242         attribute/pragma name for TARGET_OPTION_VALID_P and
26243         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
26244         * doc/tm.texi: Regenerated.
26246 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
26248         * config/microblaze/microblaze.c:
26249         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
26250         * config/microblaze/microblaze.h: Add -mxl-reorder to
26251         DRIVER_SELF_SPECS.
26252         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
26253         instructions emitted if TARGET_REORDER.
26254         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
26255         or 0 for -m/-mno case, but initialises as 2 to detect default use case
26256         separately.
26258 2013-03-01  Xinliang David Li  <davidxl@google.com>
26260         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
26261         walk length.
26263 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
26265         PR middle-end/56461
26266         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
26267         vector even when returning true.  Fix up function comment formatting.
26269         PR middle-end/56461
26270         * ira-build.c (ira_loop_nodes_count): New variable.
26271         (create_loop_tree_nodes): Initialize it.
26272         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
26274         PR middle-end/56461
26275         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
26276         method on dr_chain and result_chain.
26277         * tree-vect-stmts.c (vectorizable_store): Only call
26278         result_chain.create if j == 0.
26280         PR middle-end/56461
26281         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
26282         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
26283         before overwriting it.
26285 2013-03-01  Tobias Burnus  <burnus@net-b.de>
26287         * doc/extended.texi (C Extensions): Change order in @menu
26288         to match @node.
26289         (Other MIPS Built-in Functions): Move last MIPS entry before
26290         "picoChip Built-in Functions".
26291         (SH Built-in Functions): Move after RX Built-in Functions.
26292         * doc/gcc.texi (Introduction): Change order in @menu
26293         to match @node.
26294         * doc/md.texi (Constraints): Ditto.
26295         * gty.texi (Type Information): Ditto.
26296         (User-provided marking routines for template types): Make
26297         subsection.
26298         * doc/invoke.texi (AArch64 Options): Move before
26299         "Adapteva Epiphany Options".
26301 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
26302             Jakub Jelinek  <jakub@redhat.com>
26304         PR sanitizer/56454
26305         * asan.c (gate_asan): Lookup no_sanitize_address instead of
26306         no_address_safety_analysis attribute.
26307         * doc/extend.texi (no_address_safety_attribute): Rename to
26308         no_sanitize_address attribute, mention no_address_safety_analysis
26309         attribute as deprecated alias.
26311 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
26313         PR middle-end/56461
26314         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
26315         type to vec<vec<tree> > *.
26316         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
26317         to be vec<tree> instead of vec<tree> *, set vec_defs
26318         to vNULL and call vec_defs.create (number_of_vects), adjust other
26319         uses of vec_defs.
26320         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
26321         vectorizable_condition): Adjust vect_get_slp_defs callers.
26323 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26325         * config/aarch64/aarch64.c
26326         (aarch64_float_const_representable): Remove unused variable.
26328 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26330         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
26332 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26334         * config/aarch64/aarch64-builtins.c
26335         (aarch64_init_simd_builtins): Make static.
26337 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26339         * config/aarch64/aarch64.c
26340         (aarch64_simd_make_constant): Make static.
26342 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26344         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
26345         with no initialization to the RHS of debug statements.
26347 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26349         PR tree-optimization/56294
26350         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
26351         Adjust dumping.
26352         (get_access_replacement): Do not call create_access_replacement.
26353         Assert a replacement exists.
26354         (get_repl_default_def_ssa_name): Create the replacement declaration
26355         itself.
26357 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26359         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
26360         final_end_function.
26362 2013-02-28  Marek Polacek  <polacek@redhat.com>
26364         PR rtl-optimization/56466
26365         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
26366         if we're changing a loop.
26367         (peel_loops_completely): Likewise.
26369 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
26371         PR c++/55813
26372         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
26374 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26376         PR target/56445
26377         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
26378         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
26379         INTX_FTYPE_FX, FX_FTYPE_INTX.
26380         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
26382 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26384         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
26385         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
26386         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
26387         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
26388         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
26389         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
26390         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
26391         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
26392         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
26393         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
26394         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
26395         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
26396         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
26397         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
26398         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
26399         (avrxmega6): Increase max flash segments from 5 to 6.
26400         * config/avr/t-multilib: Regenerate.
26401         * config/avr/avr-tables.opt: Regenerate.
26402         * doc/avr-mmcu.texi: Regenerate.
26404 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26406         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
26407         (avr_device_to_arch): Rename to avr_device_to_ld.
26408         (avr_device_to_as): New prototype.
26409         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
26410         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
26411         * config/avr/driver-avr.c (avr_device_to_as): New.
26412         (avr_device_to_arch): Rename to avr_device_to_ld.
26414 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26416         PR middle-end/56461
26417         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
26418         method on dr_chain and result_chain.
26420         PR middle-end/56461
26421         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
26422         pointer_set_destroy on not_executed_last_iteration.
26424         PR middle-end/56461
26425         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
26427         PR middle-end/56461
26428         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
26429         FOR_EACH_DEFINED_FUNCTION when freeing state.
26431         PR middle-end/56461
26432         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
26433         pool_free.
26434         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
26435         overwriting it.
26437         PR middle-end/56461
26438         * ipa-cp.c (decide_whether_version_node): Call vec_free on
26439         known_aggs[i].items and release known_aggs vector.
26441         PR middle-end/56461
26442         * ipa-reference.c (propagate): Free node_info even for alias nodes.
26444 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26446         * config/microblaze/microblaze.c (microblaze_emit_compare):
26447         Use xor for EQ/NE comparisions.
26448         * config/microblaze/microblaze.md (cstoresf4): Add constraints
26449         (cbranchsf4): Adjust operator to comparison_operator.
26451 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26453         PR middle-end/56461
26454         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
26455         vector.
26456         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
26457         vec_safe_push, always update *slot.
26458         (redirect_edge_var_map_clear): Use vec_free.
26459         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
26460         (free_var_map_entry): Use vec_free.
26461         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
26462         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
26464 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
26466         PR middle-end/45472
26467         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
26468         when the may_trap_p bit of the exprs being merged differs.
26469         Reorder tests for speculativeness in the logical and operator.
26471 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26473         * incpath.c (add_standard_paths): Use reconcat instead of concat
26474         where appropriate and avoid leaking memory.
26476         * opts.h: Include obstack.h.
26477         (opts_concat): New prototype.
26478         (opts_obstack): New declaration.
26479         * opts.c (opts_concat): New function.
26480         (opts_obstack): New variable.
26481         (init_options_struct): Call gcc_init_obstack on opts_obstack.
26482         (finish_options): Use opts_concat instead of concat
26483         and XOBNEWVEC instead of XNEWVEC.
26484         * opts-common.c (generate_canonical_option, decode_cmdline_option,
26485         generate_option): Likewise.
26486         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
26487         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
26489         PR target/56455
26490         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
26491         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
26493 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26495         PR middle-end/56461
26496         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
26498 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26500         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
26501         (arm_block_move_unaligned_straight): Likewise.
26502         (arm_adjust_block_mem): Likewise.
26504 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26506         PR target/48901
26507         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
26508         temp, cond and label.
26509         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
26511         PR target/52500
26512         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
26513         * config/c6x/c6x.h (dbx_register_map): Update declaration.
26515         PR target/52501
26516         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
26517         of prologue/epilogue functions.
26519         PR target/52550
26520         * config/tilegx/tilegx.c (tilegx_expand_prologue):
26521         Remove unused variable cfa_offset.
26522         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
26524         PR target/54639
26525         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
26526         type promotion to unsigned.
26528         PR target/54640
26529         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
26530         for HOST_WIDE_INT of 32 bit / same size as int.
26531         (arm_block_move_unaligned_straight): Likewise.
26532         (arm_adjust_block_mem): Likewise.
26534         PR target/54662
26535         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
26536         ALL_CFLAGS.
26538 2013-02-26  Marek Polacek  <polacek@redhat.com>
26540         PR tree-optimization/56426
26541         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
26543 2013-02-26  Richard Biener  <rguenther@suse.de>
26545         PR target/56444
26546         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
26547         unused variable loops.
26549 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26551         PR tree-optimization/56448
26552         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
26553         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
26554         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
26555         later operands of the references, or even first operand for
26556         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
26558         PR tree-optimization/56443
26559         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
26560         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
26561         to type_for_mode langhook.
26563 2013-02-25  Matt Turner  <mattst88@gmail.com>
26565         * doc/invoke.texi: Document r4700.
26567 2013-02-25  Richard Biener  <rguenther@suse.de>
26569         PR tree-optimization/56175
26570         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
26571         split out from ...
26572         (simplify_bitwise_binary): ... here.  Also guard the conversion
26573         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
26575 2013-02-25  Catherine Moore  <clm@codesourcery.com>
26577         Revert:
26578         2013-02-24  Catherine Moore  <clm@codesourcery.com>
26579             Maciej W. Rozycki  <macro@codesourcery.com>
26580             Tom de Vries  <tom@codesourcery.com>
26581             Nathan Sidwell  <nathan@codesourcery.com>
26582             Iain Sandoe  <iain@codesourcery.com>
26583             Nathan Froyd  <froydnj@codesourcery.com>
26584             Chao-ying Fu  <fu@mips.com>
26586         * doc/extend.texi: (micromips, nomicromips, nocompression):
26587         Document new function attributes.
26588         * doc/invoke.texi (minterlink-compressed, mmicromips,
26589         m14k, m14ke, m14kec): Document new options.
26590         (minterlink-mips16): Update documentation.
26591         * doc/md.texi (ZC, ZD): Document new constraints.
26592         * configure.ac (gcc_cv_as_micromips): Check if linker
26593         supports the .set micromips directive.
26594         * configure: Regenerate.
26595         * config.in: Regenerate.
26596         * config/mips/mips-tables.opt: Regenerate.
26597         * config/mips/micromips.md: New file.
26598         * constraints.md (ZC, AD): New constraints.
26599         * config/mips/predicates.md (movep_src_register): New predicate.
26600         (movep_src_operand): New predicate.
26601         (non_volatile_mem_operand): New predicate.
26602         * config/mips/mips.md (multimem): New type.
26603         (length): Differentiate between 17-bit and 18-bit branch offsets.
26604         (MOVEP1, MOVEP2): New mode iterator.
26605         (mov_<load>l): Use ZC constraint.
26606         (mov_<load>r): Likewise.
26607         (mov_<store>l): Likewise.
26608         (mov_<store>r): Likewise.
26609         (*branch_equality<mode>_inverted): Add microMIPS support.
26610         (*branch_equality<mode>): Likewise.
26611         (*jump_absolute): Likewise.
26612         (indirect_jump_<mode>): Likewise.
26613         (tablejump_<mode>): Likewise.
26614         (<optab>_internal): Likewise.
26615         (sibcall_internal): Likewise.
26616         (sibcall_value_internal): Likewise.
26617         (prefetch): Use constraint ZD.
26618         * config/mips/mips.opt (minterlink-compressed): New option.
26619         (minterlink-mips16): Now an alias for minterlink-compressed.
26620         (mmicromips): New option.
26621         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26622         (compare_and_swap_12): Likewise.
26623         (sync_add<mode>): Likewise.
26624         (sync_<optab>_12): Likewise.
26625         (sync_old_<optab>_12): Likewise.
26626         (sync_new_<optab>_12): Likewise.
26627         (sync_nand_12): Likewise.
26628         (sync_old_nand_12): Likewise.
26629         (sync_new_nand_12): Likewise.
26630         (sync_sub<mode>): Likewise.
26631         (sync_old_add<mode>): Likewise.
26632         (sync_old_sub<mode>): Likewise.
26633         (sync_new_add<mode>): Likewise.
26634         (sync_new_sub<mode>): Likewise.
26635         (sync_<optab><mode>): Likewise.
26636         (sync_old_<optab><mode>): Likewise.
26637         (sync_new_<optab><mode>): Likewise.
26638         (sync_nand<mode>): Likewise.
26639         (sync_old_nand<mode>): Likewise.
26640         (sync_new_nand<mode>): Likewise.
26641         (sync_lock_test_and_set<mode>): Likewise.
26642         (test_and_set_12): Likewise.
26643         (atomic_compare_and_swap<mode>): Likewise.
26644         (atomic_exchange<mode>_llsc): Likewise.
26645         (atomic_fetch_add<mode>_llsc): Likewise.
26646         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26647         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26648         (umips_save_restore_pattern_p): Likewise.
26649         (umips_load_store_pair_p): Likewise.
26650         (umips_output_load_store_pair): Likewise.
26651         (umips_movep_target_p): Likewise.
26652         (umips_12bit_offset_address_p): Likewise.
26653         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26654         (mips_base_mips16): Rename this...
26655         (mips_base_compression_flags): ...to this. Update all uses.
26656         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26657         (mips_mips16_decl_p): Delete.
26658         (mips_nomips16_decl_p): Delete.
26659         (mips_get_compress_on_flags): New function.
26660         (mips_get_compress_off_flags): New function.
26661         (mips_get_compress_mode): New function.
26662         (mips_get_compress_on_name): New function.
26663         (mips_get_compress_off_name): New function.
26664         (mips_insert_attributes): Support multiple compression types.
26665         (mips_merge_decl_attributes): Likewise.
26666         (umips_12bit_offset_address_p): New function.
26667         (mips_start_function_definition): Emit .set micromips directive.
26668         (mips_call_may_need_jalx_p): New function.
26669         (mips_function_ok_for_sibcall): Add microMIPS support.
26670         (mips_print_operand_punctuation): Support short delay slots and
26671         compact jumps.
26672         (umips_swm_mask, umips_swm_encoding): New.
26673         (umips_build_save_restore): New function.
26674         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26675         (was_mips16_p): Remove.
26676         (old_compression_mode): New.
26677         (mips_set_compression_mode): New function.
26678         (mips_set_current_function): Add microMIPS support.
26679         (mips_option_override): Likewise.
26680         (umips_save_restore_pattern_p): New function.
26681         (umips_output_save_restore): New function.
26682         (umips_load_store_pair_p_1): New function.
26683         (umips_load_store_pair_p): New function.
26684         (umips_output_load_store_pair_1): New function.
26685         (umips_output_load_store_pair): New function.
26686         (umips_movep_target_p) New function.
26687         (mips_prepare_pch_save): Add microMIPS support.
26688         * config/mips/mips.h (TARGET_COMPRESSION): New.
26689         (TARGET_CPU_CPP_BUILTINS): Update macro
26690         to use new compression flags and to support microMIPS.
26691         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
26692         (MIPS_ARCH_FLOAT_SPEC): Likewise.
26693         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
26694         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
26695         (ASM_SPEC): Support mmicromips and mno-micromips.
26696         (M16STORE_REG_P): New macro.
26697         (MIPS_CALL): Support TARGET_MICROMIPS.
26698         (MICROMIPS_J): New macro.
26699         (mips_base_mips16): Rename this...
26700         (mips_base_compression_flags): ...to this.
26701         (UMIPS_12BIT_OFFSET_P): New macro.
26702         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
26703         (MULTILIB_DIRNAMES): Likewise.
26705 2013-02-25  Tom de Vries  <tom@codesourcery.com>
26707         PR rtl-optimization/56131
26708         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
26709         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
26710         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
26712 2013-02-25  Tobias Burnus  <burnus@net-b.de>
26714         * doc/invoke.texi (-fsanitize=): Move from optimization
26715         to debugging options.
26717 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26719         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
26721 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26722             Alexander Monakov  <amonakov@ispras.ru>
26724         PR middle-end/56077
26725         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
26726         flush pending lists also on non-jumps.  Adjust comment.
26728 2013-02-24  Catherine Moore  <clm@codesourcery.com>
26729             Maciej W. Rozycki  <macro@codesourcery.com>
26730             Tom de Vries  <tom@codesourcery.com>
26731             Nathan Sidwell  <nathan@codesourcery.com>
26732             Iain Sandoe  <iain@codesourcery.com>
26733             Nathan Froyd  <froydnj@codesourcery.com>
26734             Chao-ying Fu  <fu@mips.com>
26736         * doc/extend.texi: (micromips, nomicromips, nocompression):
26737         Document new function attributes.
26738         * doc/invoke.texi (minterlink-compressed, mmicromips,
26739         m14k, m14ke, m14kec): Document new options.
26740         (minterlink-mips16): Update documentation.
26741         * doc/md.texi (ZC, ZD): Document new constraints.
26742         * configure.ac (gcc_cv_as_micromips): Check if linker
26743         supports the .set micromips directive.
26744         * configure: Regenerate.
26745         * config.in: Regenerate.
26746         * config/mips/mips-tables.opt: Regenerate.
26747         * config/mips/micromips.md: New file.
26748         * constraints.md (ZC, AD): New constraints.
26749         * config/mips/predicates.md (movep_src_register): New predicate.
26750         (movep_src_operand): New predicate.
26751         (non_volatile_mem_operand): New predicate.
26752         * config/mips/mips.md (multimem): New type.
26753         (length): Differentiate between 17-bit and 18-bit branch offsets.
26754         (MOVEP1, MOVEP2): New mode iterator.
26755         (mov_<load>l): Use ZC constraint.
26756         (mov_<load>r): Likewise.
26757         (mov_<store>l): Likewise.
26758         (mov_<store>r): Likewise.
26759         (*branch_equality<mode>_inverted): Add microMIPS support.
26760         (*branch_equality<mode>): Likewise.
26761         (*jump_absolute): Likewise.
26762         (indirect_jump_<mode>): Likewise.
26763         (tablejump_<mode>): Likewise.
26764         (<optab>_internal): Likewise.
26765         (sibcall_internal): Likewise.
26766         (sibcall_value_internal): Likewise.
26767         (prefetch): Use constraint ZD.
26768         * config/mips/mips.opt (minterlink-compressed): New option.
26769         (minterlink-mips16): Now an alias for minterlink-compressed.
26770         (mmicromips): New option.
26771         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26772         (compare_and_swap_12): Likewise.
26773         (sync_add<mode>): Likewise.
26774         (sync_<optab>_12): Likewise.
26775         (sync_old_<optab>_12): Likewise.
26776         (sync_new_<optab>_12): Likewise.
26777         (sync_nand_12): Likewise.
26778         (sync_old_nand_12): Likewise.
26779         (sync_new_nand_12): Likewise.
26780         (sync_sub<mode>): Likewise.
26781         (sync_old_add<mode>): Likewise.
26782         (sync_old_sub<mode>): Likewise.
26783         (sync_new_add<mode>): Likewise.
26784         (sync_new_sub<mode>): Likewise.
26785         (sync_<optab><mode>): Likewise.
26786         (sync_old_<optab><mode>): Likewise.
26787         (sync_new_<optab><mode>): Likewise.
26788         (sync_nand<mode>): Likewise.
26789         (sync_old_nand<mode>): Likewise.
26790         (sync_new_nand<mode>): Likewise.
26791         (sync_lock_test_and_set<mode>): Likewise.
26792         (test_and_set_12): Likewise.
26793         (atomic_compare_and_swap<mode>): Likewise.
26794         (atomic_exchange<mode>_llsc): Likewise.
26795         (atomic_fetch_add<mode>_llsc): Likewise.
26796         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26797         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26798         (umips_save_restore_pattern_p): Likewise.
26799         (umips_load_store_pair_p): Likewise.
26800         (umips_output_load_store_pair): Likewise.
26801         (umips_movep_target_p): Likewise.
26802         (umips_12bit_offset_address_p): Likewise.
26803         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26804         (mips_base_mips16): Rename this...
26805         (mips_base_compression_flags): ...to this. Update all uses.
26806         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26807         (mips_mips16_decl_p): Delete.
26808         (mips_nomips16_decl_p): Delete.
26809         (mips_get_compress_on_flags): New function.
26810         (mips_get_compress_off_flags): New function.
26811         (mips_get_compress_mode): New function.
26812         (mips_get_compress_on_name): New function.
26813         (mips_get_compress_off_name): New function.
26814         (mips_insert_attributes): Support multiple compression types.
26815         (mips_merge_decl_attributes): Likewise.
26816         (umips_12bit_offset_address_p): New function.
26817         (mips_start_function_definition): Emit .set micromips directive.
26818         (mips_call_may_need_jalx_p): New function.
26819         (mips_function_ok_for_sibcall): Add microMIPS support.
26820         (mips_print_operand_punctuation): Support short delay slots and
26821         compact jumps.
26822         (umips_swm_mask, umips_swm_encoding): New.
26823         (umips_build_save_restore): New function.
26824         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26825         (was_mips16_p): Remove.
26826         (old_compression_mode): New.
26827         (mips_set_compression_mode): New function.
26828         (mips_set_current_function): Add microMIPS support.
26829         (mips_option_override): Likewise.
26830         (umips_save_restore_pattern_p): New function.
26831         (umips_output_save_restore): New function.
26832         (umips_load_store_pair_p_1): New function.
26833         (umips_load_store_pair_p): New function.
26834         (umips_output_load_store_pair_1): New function.
26835         (umips_output_load_store_pair): New function.
26836         (umips_movep_target_p) New function.
26837         (mips_prepare_pch_save): Add microMIPS support.
26838         * config/mips/mips.h (TARGET_COMPRESSION): New.
26839         (TARGET_CPU_CPP_BUILTINS): Update macro
26840         to use new compression flags and to support microMIPS.
26841         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
26842         (MIPS_ARCH_FLOAT_SPEC): Likewise.
26843         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
26844         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
26845         (ASM_SPEC): Support mmicromips and mno-micromips.
26846         (M16STORE_REG_P): New macro.
26847         (MIPS_CALL): Support TARGET_MICROMIPS.
26848         (MICROMIPS_J): New macro.
26849         (mips_base_mips16): Rename this...
26850         (mips_base_compression_flags): ...to this.
26851         (UMIPS_12BIT_OFFSET_P): New macro.
26852         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
26853         (MULTILIB_DIRNAMES): Likewise.
26855 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
26857         PR target/52555
26858         * target-globals.c (save_target_globals): For init_reg_sets and
26859         target_reinit remporarily set this_fn_optabs to this_target_optabs.
26861 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
26863         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
26864         * config/aarch64/t-aarch64
26865         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
26867 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
26869         PR inline-asm/56148
26870         * lra-constraints.c (process_alt_operands): Reload operand
26871         conflicting with earlier clobber only if no more other conflicting
26872         operands.
26874 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
26876         PR sanitizer/56393
26877         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
26878         if not linking a shared library.
26880 2013-02-22  Seth LaForge  <sethml@google.com>
26882         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
26884 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
26886         * config/arm/arm.md (split for extendsidi): Update condition.
26887         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
26888         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
26889         (qhs_zextenddi_cstr): Likewise.
26891 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
26893         PR middle-end/56420
26894         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
26895         avoid signed wrapping.
26896         (expand_mult): Handle properly multiplication by
26897         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
26898         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
26899         in the compiler if coeff is HOST_WIDE_INT_MIN.
26900         (expand_divmod): Don't make ext_op1 static, change it's type to
26901         uhwi.  Avoid undefined behavior in -INTVAL (op1).
26903         PR rtl-optimization/50339
26904         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
26905         field.
26906         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
26907         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
26908         into splitting_ashiftrt field.
26909         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
26910         ASHIFTRT.
26911         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
26912         choices.
26914 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
26916         PR middle-end/56108
26917         * trans-mem.c (execute_tm_mark): Do not expand transactions that
26918         are sure to go irrevocable.
26920 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
26922         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
26923         scalars are valid operands.
26925 2013-02-21  Martin Jambor  <mjambor@suse.cz>
26927         PR tree-optimization/56310
26928         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
26929         only matching indices and non-negative final offsets.
26930         (intersect_aggregates_with_edge): Pass src_idx to
26931         agg_replacements_to_vector.  Pass src_idx insstead of index to
26932         intersect_with_agg_replacements.
26934 2013-02-21  Martin Jambor  <mjambor@suse.cz>
26936         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
26937         instead of hard-wired defaults.
26939 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
26941         * doc/invoke.texi (MIPS Options): Update documentation of the
26942         floating-point multiply-accumulate instruction restrictions.
26944 2013-02-21  Kostya Serebryany  <kcc@google.com>
26946         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
26947         asan_shadow_offset on x86_64 linux.
26949 2013-02-21  Richard Biener  <rguenther@suse.de>
26951         PR tree-optimization/56415
26952         Revert
26953         2013-02-11  Richard Biener  <rguenther@suse.de>
26955         PR tree-optimization/56273
26956         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
26957         first VRP run.
26959 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
26961         PR bootstrap/56258
26962         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
26963         instead of @itemx.
26965         PR inline-asm/56405
26966         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
26967         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
26969 2013-02-20  Jan Hubicka  <jh@suse.cz>
26971         PR tree-optimization/56265
26972         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
26973         when target is referenced for first time.
26975 2013-02-20  Richard Biener  <rguenther@suse.de>
26977         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
26978         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
26979         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
26980         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
26981         not return anything.
26982         (rename_ssa_copies): Do not remove unused locals.
26983         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
26984         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
26985         * passes.c (execute_function_todo): Do not schedule unused locals
26986         removal if cleanup_tree_cfg did something.
26987         * tree-ssa-live.c (remove_unused_locals): Dump statistics
26988         about the number of removed locals.
26990 2013-02-20  Richard Biener  <rguenther@suse.de>
26992         PR tree-optimization/56398
26993         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
26995 2013-02-20  Martin Jambor  <mjambor@suse.cz>
26997         PR tree-optimization/55334
26998         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
26999         restricted pointers to arrays.
27001 2013-02-20  Richard Biener  <rguenther@suse.de>
27002             Jakub Jelinek  <jakub@redhat.com>
27004         PR tree-optimization/56396
27005         * tree-ssa-ccp.c (n_const_val): New static variable.
27006         (get_value): Return NULL for SSA names we don't have a lattice
27007         entry for.
27008         (ccp_initialize): Initialize n_const_val.
27009         * tree-ssa-copy.c (n_copy_of): New static variable.
27010         (init_copy_prop): Initialize n_copy_of.
27011         (get_value): Return NULL_TREE for SSA names we don't have a
27012         lattice entry for.
27014 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27016         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
27018 2013-02-20  Richard Biener  <rguenther@suse.de>
27020         * genpreds.c (write_lookup_constraint): Do not compare first
27021         letter of the constraint again.
27023 2013-02-20  Richard Biener  <rguenther@suse.de>
27025         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
27026         and ceil_log2.
27027         (get_use_iv_cost): Terminate hashtable walk when coming across
27028         an empty entry.
27030 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
27032         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
27033         reassociation for avx2 targets.
27035 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27037         * config/microblaze/microblaze.c: microblaze_has_clz = 0
27038         Add version check for v8.10.a to enable microblaze_has_clz
27039         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
27040         version and TARGET_PATTERN_COMPARE check
27041         * config/microblaze/microblaze.md: New clzsi2 instruction
27043 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27045         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
27046         function before branching.
27048 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27050         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
27051         DUMP_INSN_RTX_UID.
27052         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
27054 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27056         PR middle-end/55889
27057         * sel-sched.c: Include ira.h.
27058         (implicit_clobber_conflict_p): New function.
27059         (moveup_expr): Use it.
27060         * Makefile.in (sel-sched.o): Depend on ira.h.
27062 2013-02-19  Richard Biener  <rguenther@suse.de>
27064         PR tree-optimization/56384
27065         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
27066         (vn_hash_type): Split out from ...
27067         (vn_hash_constant_with_type): ... here.
27068         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
27069         (vn_phi_eq): Compare types from vn_phi_s structure.
27070         (vn_phi_lookup): Populate vn_phi_s type.
27071         (vn_phi_insert): Likewise.
27073 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
27075         PR tree-optimization/56350
27076         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
27077         if haven't found reduction or nested cycle operand, rather than
27078         asserting we must find it.
27080         PR tree-optimization/56381
27081         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
27082         to fold_build3.
27084 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
27085             Jakub Jelinek  <jakub@redhat.com>
27087         PR target/52555
27088         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
27089         (swap_optab_enable): Same.
27090         (init_all_optabs): Use argument instead of global.
27091         * tree.h (struct tree_optimization_option): New field target_optabs.
27092         * expr.h (init_all_optabs): Add argument to prototype.
27093         (TREE_OPTIMIZATION_OPTABS): New.
27094         (save_optabs_if_changed): Protoize.
27095         * optabs.h: Declare this_fn_optabs.
27096         * optabs.c (save_optabs_if_changed): New.
27097         Declare this_fn_optabs.
27098         (init_optabs): Add argument to init_all_optabs() call.
27099         * function.c (invoke_set_current_function_hook): Handle per
27100         function optabs.
27101         * function.h (struct function): New field optabs.
27102         * config/mips/mips.c (mips_set_mips16_mode): Handle when
27103         optimization_current_node has changed.
27104         * target-globals.h (save_target_globals_default_opts): Protoize.
27105         * target-globals.c (save_target_globals_default_opts): New.
27107 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27109         PR target/56347
27110         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
27111         registers %fr12 and %fr12R as call used.
27113         PR target/56214
27114         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
27115         and HImode, require all displacements to be an integer multiple of
27116         their mode size.
27117         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
27118         only allow QImode and HImode when reload is in progress and strict is
27119         true.  Likewise for symbolic addresses.  Use base14_operand to check
27120         displacements in REG+BASE addresses.
27122 2013-02-18  Richard Biener  <rguenther@suse.de>
27124         PR tree-optimization/56366
27125         * tree-vect-loop.c (get_initial_def_for_induction): Properly
27126         handle sign-conversion of outer-loop initial induction value.
27128 2013-02-18  Richard Biener  <rguenther@suse.de>
27130         PR middle-end/56349
27131         * cfghooks.c (merge_blocks): If we merge a latch into another
27132         block adjust references to it.
27133         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
27134         (verify_loop_structure): Verify that a recorded latch is in fact
27135         a latch.
27137 2013-02-18  Richard Biener  <rguenther@suse.de>
27139         PR tree-optimization/56321
27140         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
27141         order SSA name release and virtual operand unlinking.
27143 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27145         * config/microblaze/microblaze.md (save_stack_block): Define.
27146         (restore_stack_block): Likewise.
27148 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27150         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
27151         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
27152         * config/microblaze/microblaze.c (microblaze_option_override):
27153         Bail out early for PIC modes when target does not support PIC.
27155 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27157         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
27158         Replace with a microblaze version.
27159         (microblaze_trampoline_init): Adapt for microblaze.
27160         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
27161         microblaze.
27163 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
27164             Dodji Seketeli  <dodji@redhat.com>
27166         PR asan/56330
27167         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
27168         (instrument_mem_region_access): Do not forget to always put
27169         instrumentation of the of 'base' and 'base + len' in a "if (len !=
27170         0) statement, even for cases where either 'base' or 'base + len'
27171         are not instrumented -- because they have been previously
27172         instrumented.  Simplify the logic by putting all the statements
27173         instrument 'base + len' inside a sequence, and then insert that
27174         sequence right before the current insertion point.  Then, to
27175         instrument 'base + len', just get an iterator on that statement.
27176         And do not forget to update the pointer to iterator the function
27177         received as argument.
27179 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
27181         PR rtl-optimization/56348
27182         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
27184 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
27186         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
27187         (clean_graph_dump_file): Pass base to start_graph_dump.
27189 2013-02-14  Richard Henderson  <rth@redhat.com>
27191         PR target/55941
27192         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
27194 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
27196         * collect2-aix.h: Define F_LOADONLY.
27198 2013-02-14  Richard Biener  <rguenther@suse.de>
27200         PR lto/50494
27201         * varasm.c (output_constant_def_1): Get the decl representing
27202         the constant as argument.
27203         (output_constant_def): Wrap output_constant_def_1.
27204         (make_decl_rtl): Use output_constant_def_1 with the decl
27205         representing the constant.
27206         (build_constant_desc): Optionally re-use a decl already
27207         representing the constant.
27208         (tree_output_constant_def): Adjust.
27210 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
27212         Fix an asan crash
27213         * asan.c (instrument_builtin_call):  Really put the length of the
27214         second source argument into src1_len.
27216 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27218         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
27219         argument.  If it is false, don't create edge from then_bb to
27220         fallthru_bb.
27221         (insert_if_then_before_iter): Pass true to it.
27222         (build_check_stmt): Pass false to it.
27223         (transform_statements): Flush hash table only on extended basic
27224         block boundaries, rather than at the beginning of every bb.
27225         Don't flush hash table on nonfreeing_call_p calls.
27226         * tree-flow.h (nonfreeing_call_p): New prototype.
27227         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
27229 2013-02-13  David S. Miller  <davem@davemloft.net>
27231         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
27233 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
27235         PR target/56184
27236         * ira.c (max_regno_before_ira): Move from ...
27237         (ira): ... here.
27238         (fix_reg_equiv_init): Use max_regno_before_ira instead of
27239         vec_safe_length.
27241 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27243         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
27245 2013-02-13  Richard Biener  <rguenther@suse.de>
27247         PR lto/56295
27248         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
27249         globals in MEM_REFs.
27251 2013-02-13  Richard Biener  <rguenther@suse.de>
27253         * loop-init.c (loop_optimizer_init): Clear loop state when
27254         re-initializing preserved loops.
27255         * loop-unswitch.c (unswitch_single_loop): Return whether
27256         we unswitched the loop.  Do not verify loop state here.
27257         (unswitch_loops): When we unswitched a loop discover new loops.
27259 2013-02-13  Kostya Serebryany  <kcc@google.com>
27261         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
27262         on x86_64 linux.
27263         * sanitizer.def: Rename __asan_init to __asan_init_v1.
27265 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
27267         Avoid instrumenting duplicated memory access in the same basic block
27268         * Makefile.in (asan.o): Add new dependency on hash-table.h
27269         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
27270         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
27271         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
27272         (free_mem_ref_resources, has_mem_ref_been_instrumented)
27273         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
27274         (get_mem_ref_of_assignment): New functions.
27275         (get_mem_refs_of_builtin_call): Extract from
27276         instrument_builtin_call and tweak a little bit to make it fit with
27277         the new signature.
27278         (instrument_builtin_call): Use the new
27279         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
27280         of is_gimple_builtin_call.
27281         (instrument_derefs, instrument_mem_region_access): Insert the
27282         instrumented memory reference into the hash table.
27283         (maybe_instrument_assignment): Renamed instrument_assignment into
27284         this, and change it to advance the iterator when instrumentation
27285         actually happened and return true in that case.  This makes it
27286         homogeneous with maybe_instrument_assignment, and thus give a
27287         chance to callers to be more 'regular'.
27288         (transform_statements): Clear the memory reference hash table
27289         whenever we enter a new BB, when we cross a function call, or when
27290         we are done transforming statements.  Use
27291         maybe_instrument_assignment instead of instrumentation.  No more
27292         need to special case maybe_instrument_assignment and advance the
27293         iterator after calling it; it's now handled just like
27294         maybe_instrument_call.  Update comment.
27296 2013-02-13  Richard Biener  <rguenther@suse.de>
27298         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
27299         Fix loop discovery code.
27301 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
27303         PR inline-asm/56148
27304         * lra-constraints.c (process_alt_operands): Match early clobber
27305         operand with itself.  Check conflicts with earlyclobber only if
27306         the operand is not reloaded.  Prefer to reload conflicting operand
27307         if earlyclobber and matching operands are the same.
27309 2013-02-12  Richard Biener  <rguenther@suse.de>
27311         PR lto/56297
27312         * lto-streamer-out.c (write_symbol): Do not output symbols
27313         for hard register variables.
27315 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
27317         PR target/54222
27318         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
27319         (umulsidi3_insn, mulsidi3_insn): New insns.
27321 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
27323         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
27324         (struct tune_params): Add vec_costs field.
27325         * config/arm/arm.c (arm_builtin_vectorization_cost)
27326         (arm_add_stmt_cost): New functions.
27327         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
27328         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
27329         (arm_default_vec_cost): New struct of type cpu_vec_costs.
27330         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
27331         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
27332         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
27333         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
27335 2013-02-12  Richard Biener  <rguenther@suse.de>
27337         PR lto/56295
27338         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
27339         decls again if possible.
27341 2013-02-12  Richard Biener  <rguenther@suse.de>
27343         PR middle-end/56288
27344         * tree-ssa.c (verify_ssa_name): Fix check, move
27345         SSA_NAME_IN_FREE_LIST check up.
27347 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
27348             Steven Bosscher   <steven@gcc.gnu.org>
27350         PR rtl-optimization/56151
27351         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
27352         equal to op0 or op1, and last_insn pattern is CODE operation
27353         with MEM dest and one of the operands matches that MEM.
27355 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
27357         * doc/extend.texi: Document Function Multiversioning and "default"
27358         parameter string to target attribute.
27359         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
27360         target attribute parameter is "default".
27361         (ix86_compare_version_priority): Remove checks for target attribute.
27362         (ix86_mangle_function_version_assembler_name): Change error to sorry.
27363         Remove check for target attribute equal to NULL. Add assert.
27364         (ix86_generate_version_dispatcher_body): Change error to sorry.
27366 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
27367             Jack Howarth  <howarth@bromo.med.uc.edu>
27368             Patrick Marlier  <patrick.marlier@gmail.com>
27370         PR libitm/55693
27371         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
27372         define ENDFILE_SPEC as TM_DESTRUCTOR.
27373         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
27375 2013-02-11  Alexander Potapenko  <glider@google.com>
27376             Jack Howarth  <howarth@bromo.med.uc.edu>
27377             Jakub Jelinek  <jakub@redhat.com>
27379         PR sanitizer/55617
27380         * config/darwin.c (cdtor_record): Rename ctor_record.
27381         (sort_cdtor_records): Rename sort_ctor_records.
27382         (finalize_dtors): New routine to sort destructors by
27383         priority before use in assemble_integer.
27384         (machopic_asm_out_destructor): Use finalize_dtors if needed.
27386 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
27388         PR rtl-optimization/56275
27389         * simplify-rtx.c (avoid_constant_pool_reference): Check that
27390         offset is non-negative and less than cmode size before
27391         calling simplify_subreg.
27393 2013-02-11  Richard Biener  <rguenther@suse.de>
27395         PR tree-optimization/56264
27396         * cfgloop.h (fix_loop_structure): Adjust prototype.
27397         * loop-init.c (fix_loop_structure): Return the number of
27398         newly discovered loops.
27399         * tree-cfgcleanup.c (repair_loop_structures): When new loops
27400         are discovered, do a full loop-closed SSA rewrite.
27402 2013-02-11  Richard Biener  <rguenther@suse.de>
27404         PR tree-optimization/56273
27405         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27406         first VRP run.
27407         (check_array_ref): Fix missing newline in dumps.
27408         (search_for_addr_array): Likewise.
27410 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
27412         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
27414 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
27416         PR target/56256
27417         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
27419 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
27421         PR rtl-optimization/56246
27422         * lra-constraints.c (simplify_operand_subreg): Try to reuse
27423         reload pseudo.
27424         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
27425         constraints are satisfied.
27427 2013-02-08  Jeff Law  <law@redhat.com>
27429         PR debug/53948
27430         * emit-rtl.c (reg_is_parm_p): New function.
27431         * regs.h (reg_is_parm_p): New prototype.
27432         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
27433         callee-clobbered registers.
27435 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
27437         PR target/56043
27438         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
27439         If there is no implicit builtin declaration, just return NULL.
27441 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27443         * config/i386/sse.md (FMAMODEM): New mode iterator.
27444         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
27445         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
27447 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27449         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
27450         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
27451         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
27453 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27455         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
27456         (microblaze*-*-elf): Likewise.
27457         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
27458         LINK_SPEC.
27459         * config/microblaze/microblaze-c.c: Add builtin defines for
27460         _LITTLE_ENDIAN and _BIG_ENDIAN.
27461         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
27462         add to TARGET_DEFAULT flags.
27463         Expand ASM_SPEC and LINK_SPEC.
27464         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
27465         * config/microblaze/microblaze.md: Update extendsidi2 and
27466         movdi_internal instructions to use low-order / high-order reg
27467         print_operands.
27468         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
27469         options and inversemask / mask of LITTLE_ENDIAN.
27470         * config/microblaze/t-microblaze: Expand multilib options to
27471         include mlittle-endian (le) and update exceptions patterns.
27473 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
27475         PR rtl-optimization/56195
27476         * lra-constraints.c (get_reload_reg): Don't reuse regs
27477         if they have smaller mode than requested, if they have
27478         wider mode than requested, try to return a SUBREG.
27480         PR tree-optimization/56250
27481         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
27482         if type is unsigned and code isn't MULT_EXPR.
27484 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27486         PR tree-optimization/56064
27487         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
27488         bits according to mode.
27489         * fixed-value.h (fixed_from_double_int)
27490         (const_fixed_from_double_int): Adjust comments.
27492 2013-02-08  Richard Biener  <rguenther@suse.de>
27494         PR lto/56231
27495         * lto-streamer.h (struct data_in): Remove current_file, current_line
27496         and current_col members.
27497         * lto-streamer-out.c (lto_output_location): Stream changed bits
27498         en-block for efficiency.
27499         * lto-streamer-in.c (clear_line_info): Remove.
27500         (lto_input_location): Cache current file, line and column
27501         globally via local statics.  Read changed bits en-block.
27502         (input_function): Do not call clear_line_info.
27503         (lto_read_body): Likewise.
27504         (lto_input_toplevel_asms): Likewise.
27506 2013-02-08  Michael Matz  <matz@suse.de>
27508         PR tree-optimization/52448
27509         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
27510         (nt_call_phase): New static.
27511         (add_or_mark_expr): Only mark accesses with newer phase than any
27512         call seen.
27513         (nonfreeing_call_p): New.
27514         (nt_init_block): Update nt_call_phase, mark blocks as visited.
27515         (nt_fini_block): Keep blocks marked as visited.
27516         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
27518 2013-02-08  Richard Biener  <rguenther@suse.de>
27520         * ira.c (ira): Free broken dominator information.
27522 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27524         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
27526 2013-02-08  Marek Polacek  <polacek@redhat.com>
27528         * cfgloop.c (verify_loop_structure): Add more checking of headers.
27530 2013-02-08  Richard Biener  <rguenther@suse.de>
27532         PR middle-end/56181
27533         * cfgloop.h (flow_loops_find): Adjust.
27534         (bb_loop_header_p): Declare.
27535         * cfgloop.c (bb_loop_header_p): New function split out from ...
27536         (flow_loops_find): ... here.  Adjust function signature,
27537         support incremental loop structure update.
27538         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
27539         * cfgloopmanip.c (fix_loop_structure): Move ...
27540         * loop-init.c (fix_loop_structure): ... here.
27541         (apply_loop_flags): Split out from ...
27542         (loop_optimizer_init): ... here.
27543         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
27544         in incremental mode, only remove dead loops here.
27546 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27548         PR target/54222
27549         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
27550         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
27551         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
27552         (*round<mode>3.libgcc): New insns for fixed-modes.
27553         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
27554         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
27555         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
27556         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
27557         implementations.  Define to __builtin_avr_absFX,
27558         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
27559         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
27560         __builtin_avr_countlsFX, respectively.
27561         * config/avr/avr-c.c (target.h): Include it.
27562         (enum avr_builtin_id): New enum.
27563         (avr_resolve_overloaded_builtin): New static function.
27564         (avr_register_target_pragmas): Use it to set
27565         targetm.resolve_overloaded_builtin.
27566         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
27567         tree nodes used by DEF_BUILTIN.
27568         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
27569         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
27570         <AVR_BUILTIN_xxBITS>: Same.
27572 2013-02-08  Richard Biener  <rguenther@suse.de>
27574         * cfgloop.c (verify_loop_structure): Properly handle
27575         a loop exiting to another loop header.
27576         * ira-int.h (ira_loops): Remove.
27577         * ira.c (ira_loops): Remove.
27578         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
27579         (do_reload): Use loop_optimizer_finalize.
27580         * ira-build.c (create_loop_tree_nodes): Use get_loops and
27581         number_of_loops to access the loop tree.
27582         (more_one_region_p): Likewise.
27583         (finish_loop_tree_nodes): Likewise.
27584         (rebuild_regno_allocno_maps): Likewise.
27585         (mark_loops_for_removal): Likewise.
27586         (mark_all_loops_for_removal): Likewise.
27587         (remove_unnecessary_regions): Likewise.
27588         (ira_build): Likewise.
27589         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
27591 2013-02-08  Richard Biener  <rguenther@suse.de>
27593         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
27594         * ipa-pure-const.c (analyze_function): Avoid calling
27595         mark_irreducible_loops twice.
27596         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
27598 2013-02-07  David S. Miller  <davem@davemloft.net>
27600         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
27601         on 'reg'.
27602         * var-tracking.c (vt_add_function_parameter): Test the presence of
27603         HAVE_window_save properly and do not remap argument registers when
27604         we have a leaf function.
27606 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
27608         PR bootstrap/56227
27609         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
27610         instead of "ll".
27611         * config/i386/i386.c (ix86_print_operand): Ditto.
27613 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27615         * lra-constraints.c (process_alt_operands): Fix recently added comment.
27617 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27619         PR rtl-optimization/56225
27620         * lra-constraints.c (process_alt_operands): Check that reload hard
27621         reg can hold value for strict_low_part.
27623 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27625         PR debug/56154
27626         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
27627         dwarf2out_end_function.
27628         (in_first_function_p, maybe_at_text_label_p,
27629         first_loclabel_num_not_at_text_label): New variables.
27630         (dwarf2out_var_location): In the first function find out
27631         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
27632         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
27633         functions.
27635 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
27637         PR rtl-optimization/56178
27638         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
27639         SUBREG of a register.  Tidy up related block of code.
27640         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
27641         note if the source is a register or a SUBREG of a register.
27643 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27645         PR target/56228
27646         * config/rs6000/rs6000.md (ptrm): New mode attr.
27647         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
27648         call_value_indirect_aix<pttrsize>,
27649         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
27650         m in constraints.
27652 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
27654         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
27655         if -bnortl. Convert to strcmp and strncmp.
27657 2013-02-07  Alan Modra  <amodra@gmail.com>
27659         PR target/54009
27660         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
27661         addresses won't wrap when offsetting.
27662         (rs6000_secondary_reload): Provide secondary reloads needed for
27663         wrapping LO_SUM addresses.
27665 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
27667         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
27668         MACH, just __MACH__.
27670 2013-02-06  Richard Biener  <rguenther@suse.de>
27672         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
27673         instead of calling fix_loop_structure.
27675 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
27677         PR middle-end/56217
27678         * omp-low.c (use_pointer_for_field): Return false if
27679         lower_send_shared_vars doesn't generate any copy-out code.
27681 2013-02-06  Tom de Vries  <tom@codesourcery.com>
27683         PR rtl-optimization/56131
27684         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
27685         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
27686         of the label is NULL.  Add comment.
27688 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
27690         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
27692         PR sanitizer/55374
27693         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
27694         (STATIC_LIBTSAN_LIBS): Likewise.
27695         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
27696         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
27697         is defined, don't add anything else beyond that.
27698         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
27699         (LINK_COMMAND_SPEC): Use them.
27701         PR tree-optimization/56205
27702         * tree-stdarg.c (check_all_va_list_escapes): Return true if
27703         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
27704         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
27706 2013-02-05  Richard Biener  <rguenther@suse.de>
27708         PR tree-optimization/53342
27709         PR tree-optimization/53185
27710         * tree-vectorizer.h (vect_check_strided_load): Remove.
27711         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
27712         not disallow peeling for vectorized strided loads.
27713         (vect_check_strided_load): Make static and simplify.
27714         (vect_analyze_data_refs): Adjust.
27715         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
27716         correctly when vectorizing strided loads.
27718 2013-02-05  Richard Biener  <rguenther@suse.de>
27720         * doc/install.texi: Refer to ISL, not PPL.
27722 2013-02-05  Jan Hubicka  <jh@suse.cz>
27724         PR tree-optimization/55789
27725         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
27727 2013-02-05  Jan Hubicka  <jh@suse.cz>
27729         PR tree-optimization/55789
27730         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
27731         the dead call anyway.
27733 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
27735         PR sanitizer/55374
27736         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
27738 2013-02-04  Alexander Potapenko  <glider@google.com>
27739             Jack Howarth  <howarth@bromo.med.uc.edu>
27740             Jakub Jelinek  <jakub@redhat.com>
27742         PR sanitizer/55617
27743         * config/darwin.c (sort_ctor_records): Stabilized qsort
27744         on constructor priority by using original position.
27745         (finalize_ctors): New routine to sort constructors by
27746         priority before use in assemble_integer.
27747         (machopic_asm_out_constructor): Use finalize_ctors if needed.
27749 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
27751         PR libstdc++/54314
27752         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
27753         about visibility on artificial decls.
27754         * config/sol2.c (solaris_assemble_visibility): Likewise.
27756 2013-02-04  Kai Tietz  <ktietz@redhat.com>
27758         PR target/56186
27759         * config/i386/i386.c (function_value_ms_64): Add additional valtype
27760         argument and improve checking of return-argument types for 16-byte
27761         modes.
27762         (ix86_function_value_1): Add additional valtype argument on call
27763         of function_value_64.
27764         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
27765         handling infunction_value_64 function.
27767 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
27769         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
27771 2013-02-04  Richard Biener  <rguenther@suse.de>
27773         PR tree-optimization/56188
27774         * tree-ssa-structalias.c (label_visit): Consider case with
27775         initially non-empty points-to set.
27776         (perform_var_substitution): Dump node mapping and clean up.
27778 2013-02-04  Richard Guenther  <rguenther@suse.de>
27780         PR lto/56168
27781         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
27782         node prevail as last resort.
27783         (lto_symtab_merge_decls): Remove guard on LTRANS here.
27784         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
27786 2013-02-04  Richard Biener  <rguenther@suse.de>
27788         PR tree-optimization/56113
27789         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
27790         Merge into ...
27791         (equiv_class_lookup_or_add): ... this.
27792         (label_visit): Adjust and fix error in previous patch.
27793         (perform_var_substitution): Adjust.
27795 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
27797         * config/sh/divtab.c: Fix formatting and comments throughout the file.
27798         * config/sh/sh4-300.md: Likewise.
27799         * config/sh/sh4a.md: Likewise.
27800         * config/sh/constraints.md: Likewise.
27801         * config/sh/sh.md: Likewise.
27802         * config/sh/netbsd-elf.h: Likewise.
27803         * config/sh/predicates.md: Likewise.
27804         * config/sh/sh-protos.h: Likewise.
27805         * config/sh/ushmedia.h: Likewise.
27806         * config/sh/linux.h: Likewise.
27807         * config/sh/sh.c: Likewise.
27808         * config/sh/superh.h: Likewise.
27809         * config/sh/elf.h: Likewise.
27810         * config/sh/sh4.md: Likewise.
27811         * config/sh/sh.h: Likewise.
27813 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27815         * config/pa/constraints.md: Adjust unused letters.  Change "T"
27816         constraint to match_test floating_point_store_memory_operand().
27817         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
27818         (base14_operand): New.
27819         (floating_point_store_memory_operand): New.
27820         (integer_store_memory_operand): Revise to use base14_operand and
27821         reg_plus_base_memory_operand.
27822         (move_dest_operand): Allow symbolic_memory_operands.
27823         (symbolic_memory_operand): Check for LO_SOM.
27824         (symbolic_operand): Change default case to break.
27825         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
27826         CONST_DOUBLE values to be reloaded by putting them into memory when
27827         the destination is a floating point register.
27828         (movdf): Remove code to handle CONST_DOUBLE.
27829         (movsf): Likewise.
27830         (reload_indf_r1): New.
27831         (reload_insf_r1): New.
27832         Consistently use "Q" and "T" constraints with integer and floating
27833         point move instructions, respectively.
27834         (movdi): Remove FAIL.
27835         Change predicate for source operand unamed DImode move from
27836         general_operand to move_src_operand.
27837         (umulsidi3): Change predicate for destination operand to
27838         register_operand.
27839         Likewise for similar unamed patterns.
27840         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
27841         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
27842         (hppa_legitimize_address): Simplify mask calculation.
27843         (pa_emit_move_sequence): Revised handling of secondary reloads from
27844         REG+D addresses for floating point loads and stores.  Directly handle
27845         loading CONST0_RTX (mode) to a floating point register.
27846         (pa_secondary_reload): Handle reloading DF and SFmode constant values
27847         to floating point registers.  Don't restrict secondary reloads to
27848         floating point registers to integer modes.  Revise some comments and
27849         cleanup some code.
27850         (TARGET_LEGITIMATE_ADDRESS_P): Define.
27851         (pa_legitimate_address_p): New.
27852         (pa_legitimize_reload_address): New.
27853         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
27854         (STRICT_REG_OK_FOR_BASE_P): New.
27855         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
27856         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
27858 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
27859             Andrew Dixie  <andrewd@gentrack.com>
27861         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
27862         flag set.
27864 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
27866         * expmed.c (extract_bit_field_1): Pass the full width of the
27867         structure to get_best_reg_extraction_insn.
27869 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
27871         PR target/54601
27872         * configure.ac (use_cxa_atexit): Add AIX.
27873         * configure: Regenerate.
27875         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
27877 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
27879         PR debug/54793
27880         * final.c (need_profile_function): New variable.
27881         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
27882         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
27883         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
27884         notes, targetm.asm_out.function_prologue doesn't emit anything,
27885         HAVE_prologue and profiler should be emitted before prologue,
27886         set need_profile_function instead of emitting it.
27887         (final_scan_insn): If need_profile_function, emit
27888         profile_function on the first NOTE_INSN_BASIC_BLOCK or
27889         NOTE_INSN_FUNCTION_BEG note.
27891 2013-02-01  Richard Henderson  <rth@redhat.com>
27893         * config/rs6000/rs6000.md (smulditi3): New.
27894         (umulditi3): New.
27896         * config/alpha/alpha.md (umulditi3): New.
27898 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
27900         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
27901         (ASM_OUTPUT_ALIGNED_LOCAL): New.
27903 2013-02-01  Richard Biener  <rguenther@suse.de>
27905         PR tree-optimization/56113
27906         * tree-ssa-structalias.c (label_visit): Reduce work for
27907         single-predecessor nodes.
27909 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
27911         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
27912         range isn't testing for zero.
27914 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
27916         PR middle-end/56113
27917         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
27919 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
27920             Nick Clifton  <nickc@redhat.com>
27922         * config/v850/constraints.md (Q): Define as a memory constraint.
27923         * config/v850/predicates.md (label_ref_operand): New predicate.
27924         (e3v5_shift_operand): New predicate.
27925         (ior_operator): New predicate.
27926         * config/v850/t-v850: Add e3v5 multilib.
27927         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
27928         (v850_gen_movdi): Prototype.
27929         * config/v850/v850.c: Add support for e3v5 architecture.
27930         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
27931         TARGET_V850E_UP.
27932         (construct_save_jarl): Add e3v5 long JARL support.
27933         (v850_adjust_insn_length): New function.  Adjust length of call
27934         insns when using e3v5 instructions.
27935         (v850_gen_movdi): New function: Generate instructions to move a
27936         DImode value.
27937         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
27938         (CPP_SPEC): Define __v850e3v5__ as appropriate.
27939         (TARGET_USE_FPU): Enable for e3v5.
27940         (CONST_OK_FOR_W): New macro.
27941         (ADJUST_INSN_LENGTH): Define.
27942         * config/v850/v850.md (UNSPEC_LOOP): Define.
27943         (attr cpu): Add v850e3v5.
27944         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
27945         (movdi): New pattern.
27946         (movdi_internal): New pattern.
27947         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
27948         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
27949         (cstoresf4): Likewise.
27950         (cstoredf4): Likewise.
27951         (insv): New pattern.
27952         (rotlso3_a): New pattern.
27953         (rotlsi3_b): New pattern
27954         (rotlsi3_v850e3v5): New pattern.
27955         (doloop_begin): New pattern.
27956         (fix_loop_counter): New pattern.
27957         (doloop_end): New pattern.
27958         (branch_normal): Add e3v5 long branch support.
27959         (branch_invert): Likewise.
27960         (branch_z_normal): Likewise.
27961         (branch_z_invert): Likewise.
27962         (branch_nz_normal): Likewise.
27963         (branch_nz_invert): Likewise.
27964         (call_internal_short): Add e3v5 register-indirect JARL support.
27965         (call_internal_long): Likewise.
27966         (call_value_internal_short): Likewise.
27967         (call_value_internal_long): Likewise.
27968         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
27969         (mloop): New option.
27970         * config.gcc: Add support for configuring v840e3v5 target.
27971         * doc/invoke.texi: Document new v850 specific command line options.
27973 2013-01-31  Paul Koning  <ni1d@arrl.net>
27975         PR debug/55059
27976         PR debug/54508
27977         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
27978         children if parent is a class.
27979         (prune_unused_types_prune): Don't add DW_AT_declaration.
27981 2013-01-31  Richard Biener  <rguenther@suse.de>
27983         PR tree-optimization/56157
27984         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
27985         match up operand with SLP child.
27987 2013-01-31  Jason Merrill  <jason@redhat.com>
27989         PR debug/54410
27990         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
27991         parameters the first time.
27992         (gen_scheduled_generic_parms_dies): Check completeness here.
27994 2013-01-31  Richard Biener  <rguenther@suse.de>
27996         PR middle-end/53073
27997         * common.opt (faggressive-loop-optimizations): New flag,
27998         enabled by default.
27999         * doc/invoke.texi (faggressive-loop-optimizations): Document.
28000         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
28001         infer_loop_bounds_from_undefined by it.
28003 2013-01-31  Richard Biener  <rguenther@suse.de>
28005         PR tree-optimization/56150
28006         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
28007         visit virtual operands.
28008         (find_uses_to_rename_bb): Likewise.
28010 2013-01-31  Richard Biener  <rguenther@suse.de>
28012         PR tree-optimization/56150
28013         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
28014         mixed store non-store stmts.
28016 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
28018         PR sanitizer/55374
28019         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
28020         LIBASAN_EARLY_SPEC is defined.
28021         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
28022         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
28023         before %o.
28024         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
28026         PR c++/55742
28027         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
28028         invalid args instead of ICEing on it.
28029         (ix86_valid_target_attribute_tree): Return error_mark_node if
28030         ix86_valid_target_attribute_inner_p failed.
28031         (ix86_valid_target_attribute_p): Return false only if
28032         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
28033         target("default") attribute.
28034         (sorted_attr_string): Change argument from const char * to tree,
28035         merge in all target attribute arguments rather than just one.
28036         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
28037         instead of free.  Avoid using strcat.
28038         (ix86_mangle_function_version_assembler_name): Mangle
28039         target("default") as if no target attribute is present.  Adjust
28040         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
28041         instead of xmalloc and XDELETEVEC instead of free.
28042         (ix86_function_versions): Don't return true if one of the decls
28043         doesn't have target attribute.  If they don't and one of the decls
28044         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
28045         sorted_attr_string caller.  Use XDELETEVEC instead of free.
28046         (ix86_supports_function_versions): Remove.
28047         (make_name): Fix up formatting.
28048         (make_dispatcher_decl): Remove resolver_name and its initialization.
28049         Avoid leaking memory.
28050         (is_function_default_version): Return true if there is
28051         target("default") attribute rather than no target attribute at all.
28052         (make_resolver_func): Avoid leaking memory.
28053         (ix86_generate_version_dispatcher_body): Likewise.
28054         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
28055         * target.def (supports_function_versions): Remove.
28056         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
28057         * doc/tm.texi: Regenerated.
28059 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
28061         PR rtl-optimization/56144
28062         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
28063         for values with side effects.
28065 2013-01-30  Richard Biener  <rguenther@suse.de>
28067         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
28068         (sparseset_pop): Likewise.
28069         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
28070         to be able to use quick_push in the worker loop.
28072 2013-01-30  Marek Polacek  <polacek@redhat.com>
28074         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
28076 2013-01-30  Richard Biener  <rguenther@suse.de>
28078         PR lto/56147
28079         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
28081 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
28083         PR tree-optimization/56064
28084         * fixed-value.c (fixed_from_double_int): New function.
28085         * fixed-value.h (fixed_from_double_int): New prototype.
28086         (const_fixed_from_double_int): New static inline function.
28087         * fold-const.c (native_interpret_fixed): New static function.
28088         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
28089         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
28090         (native_encode_fixed): New static function.
28091         (native_encode_expr) <FIXED_CST>: Use it.
28092         (native_interpret_int): Move double_int worker code to...
28093         * double-int.c (double_int::from_buffer): ...this new static method.
28094         * double-int.h (double_int::from_buffer): Prototype it.
28096 2013-01-30  Richard Biener  <rguenther@suse.de>
28098         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
28099         New pointer-map and obstack.
28100         (init_alias_vars): Allocate pointer-map and obstack.
28101         (delete_points_to_sets): Free them.
28102         (find_what_var_points_to): Cache result.
28103         (find_what_p_points_to): Adjust for changed interface of
28104         find_what_var_points_to.
28105         (compute_points_to_sets): Likewise.
28106         (ipa_pta_execute): Likewise.
28108 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
28110         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
28111         * configure: Regenerate.
28112         * config.in: Regenerate.
28113         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
28114         #nobits/#progbits if supported.
28116 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
28118         PR target/56121
28119         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
28120         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
28121         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
28123 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28125         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
28126         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
28128 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28130         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
28131         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
28133 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28135         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
28136         declaration.
28137         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
28138         * config/arm/cortex-a7.md: New bypasses using
28139         arm_mac_accumulator_is_result.
28141 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28143         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
28144         (cortex_a7_neon_mla): Likewise.
28145         (cortex_a7_fpfmad): New reservation.
28146         (cortex_a7_fpmacs): Use ffmas and update required units.
28147         (cortex_a7_fpmuld): Update required units and latency.
28148         (cortex_a7_fpmacd): Likewise.
28149         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
28150         (cortex_a7_neon). Likewise.
28151         (bypass) Update participating units.
28153 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28155         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
28156         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
28157         from fmac to ffma.
28158         * config/arm/vfp11.md (vfp_farith): Use ffmas.
28159         (vfp_fmul): Use ffmad.
28160         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
28161         (cortex_r4_fmacd): Use ffmad.
28162         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
28163         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
28164         (cortex_a9_fmacd): Use ffmad.
28165         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
28166         (cortex_a8_vfp_macd): Use ffmad.
28167         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
28168         (cortex_a5_fpmacd): Use ffmad.
28169         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
28170         (cortex_a15_vfp_macd): Use ffmad.
28171         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
28173 2013-01-29  Jason Merrill  <jason@redhat.com>
28175         PR libstdc++/54314
28176         * varasm.c (default_assemble_visibility): Don't warn about
28177         visibility on artificial decls.
28179 2013-01-29  Richard Biener  <rguenther@suse.de>
28181         PR tree-optimization/56113
28182         * tree-ssa-structalias.c (equiv_class_lookup): Also return
28183         the bitmap leader.
28184         (label_visit): Free duplicate bitmaps and record the leader instead.
28185         (perform_var_substitution): Adjust.
28187 2013-01-29  Richard Biener  <rguenther@suse.de>
28189         PR tree-optimization/55270
28190         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
28191         the CFG, schedule loops for fixup.
28193 2013-01-29  Nick Clifton  <nickc@redhat.com>
28195         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
28196         SP_REG.
28198 2013-01-28  Leif Ekblad  <leif@rdos.net>
28200         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
28201         * config/i386/i386.h (TARGET_RDOS): New macro.
28202         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
28203         * config/i386/i386.c (ix86_option_override_internal): For 64bit
28204         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
28205         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
28206         DEFAULT_LARGE_SECTION_THRESHOLD.
28207         * config/i386/i386.md (R14_REG, R15_REG): New constants.
28208         * config/i386/rdos.h: New file.
28209         * config/i386/rdos64.h: New file.
28211 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
28213         PR other/54814
28214         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
28215         TEST_HARD_REG_BIT.
28217 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28219         PR rtl-optimization/56117
28220         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
28221         call cselib_lookup_from_insn on the MEM before calling
28222         add_insn_mem_dependence.
28224 2013-01-28  Richard Biener  <rguenther@suse.de>
28226         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
28227         to a stmt that didn't have one.
28228         (copy_phis_for_bb): Likewise for PHI arguments.
28229         (copy_debug_stmt): Likewise for debug stmts.
28231 2013-01-28  Richard Biener  <rguenther@suse.de>
28233         PR tree-optimization/56034
28234         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
28235         (partition_builtin_p): Adjust.
28236         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
28237         it is the last partition.
28238         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
28239         up the vertex for the definition.
28240         (classify_partition): Classify whether a partition is a
28241         PKIND_REDUCTION, thus has uses outside of the loop.
28242         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
28243         Merge all PKIND_REDUCTION partitions into the last partition.
28244         (tree_loop_distribution): Seed partitions from reductions as well.
28246 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28248         PR tree-optimization/56125
28249         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
28250         pow(x,c) into sqrt(x) * powi(x, n/2) or
28251         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
28252         optimizing for size.
28253         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
28254         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
28255         integer.
28257         PR tree-optimization/56094
28258         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
28259         to UNKNOWN_LOCATION while gimplifying expr.
28261 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
28263         PR target/56114
28264         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
28265         operand 0 in movabs insn template for -masm=intel asm alternative.
28266         (*movabs<mode>_2): Ditto for operand 1.
28268 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
28270         PR target/54663
28271         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
28272         of microblaze-c.o
28274 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28276         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
28277         tm_file.
28279 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
28281         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
28282         Undef to avoid warning.
28284 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
28286         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
28287         * configure: Regenerate.
28289 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
28291         PR tree-optimization/56098
28292         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
28293         for stmts with volatile ops.
28294         (cond_store_replacement): Don't optimize if assign has volatile ops.
28295         (cond_if_else_store_replacement_1): Don't optimize if either
28296         then_assign or else_assign have volatile ops.
28297         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
28298         volatile ops.
28300 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28302         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
28304 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28306         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
28307         missing ':' in asm example.
28309 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
28311         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
28312         entries into lane and laneq entries.
28313         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
28314         Remove AdvSIMD scalar modes.
28315         (aarch64_sq<r>dmulh_laneq<mode>): New.
28316         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
28317         modes.
28318         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
28319         builtin implementations to relfect changes in RTL in aarch64-simd.md.
28320         * config/aarch64/iterators.md (VCOND): New.
28321         (VCONQ): New.
28323 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28325         PR target/54222
28326         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
28327         Add NULL LIBNAME argument to existing definitions.
28328         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
28329         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
28330         * config/avr/avr.c (DEF_BUILTIN): Same.
28331         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
28332         (avr_expand_builtin): Expand to a vanilla call if a libgcc
28333         implementation is available (DECL_ASSEMBLER_NAME is set).
28334         (avr_fold_absfx): New static function.
28335         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
28336         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
28337         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
28338         AVR_BUILTIN_ABSLLK.
28339         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
28340         (abshk, absk, abslk, absllk): Provide as static inline functions.
28342 2013-01-25  Marek Polacek  <polacek@redhat.com>
28344         PR tree-optimization/56035
28345         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
28347 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
28349         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
28350         (*movtf_internal_rex64): Add (!o,C) alternative
28351         (*movxf_internal_rex64): Ditto.
28352         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
28354 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
28356         * doc/invoke.texi: fix typo.
28357         * doc/objc.texi: fix typo.
28359 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
28361         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
28362         for the first two alternatives.
28364 2013-01-24  Diego Novillo  <dnovillo@google.com>
28366         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
28367         (ggc-zone.o): Remove.
28368         * configure.ac: Remove option --with-gc.
28369         * configure: Re-generate.
28370         * doc/install.texi: Remove documentation for --with-gc.
28371         * gengtype.c (write_enum_defn): Remove.  Update all users.
28372         (write_Types_process_field): Remove generation of gt_e_* argument.
28373         (output_type_enum): Remove.  Update all users.
28374         (write_enum_defn): Remove.  Update all users.
28375         (enum alloc_zone): Remove.  Update all users.
28376         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
28377         * ggc-common.c (ggc_splay_alloc): Remove first argument.
28378         Update all callers.
28379         (struct ptr_data): Remove field TYPE.  Update all users.
28380         (gt_pch_note_object): Remove argument TYPE.  Update all users.
28381         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
28382         Update all users.
28383         * ggc-none.c (ggc_alloc_typed_stat): Remove.
28384         (struct alloc_zone): Remove.
28385         (ggc_internal_alloc_zone_stat): Remove.
28386         (ggc_internal_cleared_alloc_zone_stat): Remove.
28387         * ggc-page.c (ggc_alloc_typed_stat): Remove.
28388         (ggc_pch_count_object): Remove last argument.  Update all users.
28389         (ggc_pch_alloc_object): Remove last argument.  Update all users.
28390         (struct alloc_zone): Remove.
28391         * ggc-zone.c: Remove.
28392         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
28393         (struct alloc_zone): Remove.
28394         (ggc_alloc_typed_stat): Remove.
28395         (ggc_alloc_typed): Remove.
28396         (ggc_splay_alloc): Remove first argument.
28397         (rtl_zone): Remove.  Update all users.
28398         (tree_zone): Remove.  Update all users.
28399         (tree_id_zone): Remove.  Update all users.
28400         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
28401         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
28402         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
28403         * tree-ssanames.c: Remove references to zone allocator in comments.
28405 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
28407         * config/avr/avr.c (avr_out_fract): Make register numbers that
28408         might be outside of source operand signed.
28410 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
28412         * config/i386/constraints.md (Yf): New constraint.
28413         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
28414         of f constraint to conditionaly disable x87 register preferences.
28415         (*movdf_internal): Ditto.
28416         (*movsf_internal): Ditto.
28418 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
28420         PR inline-asm/55934
28421         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
28422         that have operands with impossible constraints.
28423         Add a FIXME for a speed-up opportunity.
28424         * lra-constraints.c (process_alt_operands): Verify that a class
28425         selected from constraints on asms is valid for the operand mode.
28426         (curr_insn_transform): Remove incorrect comment.
28428 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
28430         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
28431         TOC operand is a valid symbol ref in the constant pool.
28433 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28435         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
28437 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
28439         PR target/54222
28440         * config/avr/stdfix.h: New file.
28441         * t-avr (stdfix-gcc.h): New rule to build it.
28442         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
28444 2013-01-23  Kostya Serebryany  <kcc@google.com>
28446         * config/darwin.h: remove dependency on
28447         CoreFoundation (asan on Mac OS).
28449 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
28451         PR target/49069
28452         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
28453         instead of cmpdi_operand for first comparison operand.
28454         Don't assert that comparison operands aren't both constants.
28456 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28458         * doc/install.texi (Downloading the Source): Update references to
28459         downloading separate components.
28461 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28463         * doc/extend.texi (__int128): Improve grammar.
28465 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
28467         PR target/56028
28468         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
28469         alternative to (o,r).
28470         (*movdi_internal_rex64): Remove (!o,n) alternative.
28471         (DImode immediate->memory splitter): Remove.
28472         (DImode immediate->memory peephole2): Remove.
28473         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
28474         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
28475         alternative to (!o,*r).
28476         (*movtf_internal_sse): New pattern.
28477         (*movxf_internal_rex64): New pattern.
28478         (*movxf_internal): Disable for TARGET_64BIT.
28479         (*movdf_internal_rex64): Remove (!o,F) alternative.
28481 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
28483         PR middle-end/56074
28484         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
28485         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
28486         * tree-vect-loop-manip.c (find_loop_location): Also ignore
28487         stmt locations where LOCATION_LOCUS of the stmt location is
28488         UNKNOWN_LOCATION or BUILTINS_LOCATION.
28490         PR target/55686
28491         * config/i386/i386.md (UNSPEC_STOS): New.
28492         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
28493         *strsetqi_1): Add UNSPEC_STOS.
28495 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
28497         PR c++/56067
28498         * doc/invoke.texi: Remove left over -Wsynth example.
28500 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28502         PR tree-optimization/56051
28503         * fold-const.c (fold_binary_loc): Don't fold
28504         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
28505         a narrowing conversion, or widening conversion from signed
28506         to unsigned.
28508 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28510         PR rtl-optimization/56023
28511         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
28512         dependent on debug instruction.
28514 2013-01-21  Martin Jambor  <mjambor@suse.cz>
28516         PR middle-end/56022
28517         * function.c (allocate_struct_function): Call
28518         invoke_set_current_function_hook earlier.
28520 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28522         * reload1.c (init_reload): Only initialize reload_obstack
28523         during the first call.
28525 2013-01-21  Marek Polacek  <polacek@redhat.com>
28527         * cfgloop.c (verify_loop_structure): Fix up grammar.
28529 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
28531         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
28532         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
28534 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28536         PR target/56058
28537         * config/arm/marvell-pj4.md: Update copyright year.
28538         Fix up use of alu to alu_reg and simple_alu_imm.
28540 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28542         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
28544 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
28546         PR target/55433
28547         * lra-constraints.c (curr_insn_transform): Don't reuse original
28548         insn for secondary memory move when memory mode should be different.
28550 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28552         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
28553         atomic_storedi_1): New patterns.
28555 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
28557         btver2 pipeline descriptions.
28558         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
28559         descriptions.
28560         * config/i386/i386.md (btver2_decode): New type attributes.
28561         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
28562         type attributes.
28563         * config/i386/btver2.md: New file describing btver2 pipelines.
28565 2013-01-19  Andrew Pinski  <apinski@cavium.com>
28567         PR tree-optimization/52631
28568         * tree-ssa-sccvn (visit_use): Before looking up the original
28569         statement, try looking up the simplified expression.
28571 2013-01-19  Anthony Green  <green@moxielogic.com>
28573         * config/moxie/moxie.c (moxie_expand_prologue): Set
28574         current_function_static_stack_size.
28576 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28578         PR tree-optimization/56029
28579         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
28580         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
28582 2013-01-18  Sharad Singhai  <singhai@google.com>
28584         PR tree-optimization/55995
28585         * dumpfile.c (dump_loc): Print location only if available.
28586         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
28588 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
28590         PR target/55433
28591         * lra-constraints.c (curr_insn_transform): Reuse original insn for
28592         secondary memory move.
28593         (inherit_reload_reg): Use rclass instead of cl for
28594         check_secondary_memory_needed_p.
28596 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28598         PR middle-end/56015
28599         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
28600         the case where writing real complex part of target modifies op1.
28602 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
28604         * config/aarch64/aarch64-simd.md
28605         (aarch64_vcond_internal<mode>): Handle unordered cases.
28606         * config/aarch64/iterators.md (v_cmp_result): New.
28608 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
28609             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28611         * config/arm/marvell-pj4.md: New file.
28612         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
28613         * config/arm/arm.md (generic_sched): Add marvell_pj4.
28614         (generic_vfp): Likewise.
28615         * config/arm/arm-cores.def: Add marvell-pj4.
28616         * config/arm/arm-tune.md: Regenerate.
28617         * config/arm/arm-tables.opt: Regenerate.
28618         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
28619         * doc/invoke.texi: Document marvell-pj4.
28621 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
28623         * config/aarch64/arm_neon.h: Map scalar types to standard types.
28625 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28627         PR debug/54114
28628         PR debug/54402
28629         PR debug/49888
28630         * var-tracking.c (negative_power_of_two_p): New.
28631         (global_get_addr_cache, local_get_addr_cache): New.
28632         (get_addr_from_global_cache, get_addr_from_local_cache): New.
28633         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
28634         heading comment.
28635         (vt_stack_offset_p): Remove.
28636         (vt_canon_true_dep): Always canonicalize loc's address.
28637         (clobber_overlapping_mems): Make sure we have a MEM.
28638         (local_get_addr_clear_given_value): New.
28639         (val_reset): Clear local cached entries.
28640         (compute_bb_dataflow): Create and release the local cache.
28641         Disable duplicate MEMs clobbering.
28642         (emit_notes_in_bb): Clobber MEMs likewise.
28643         (vt_emit_notes): Create and release the local cache.
28644         (vt_initialize, vt_finalize): Create and release the global
28645         cache, respectively.
28646         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
28648 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28650         PR libmudflap/53359
28651         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
28652         not found in the symtab.
28654 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28656         PR debug/56006
28657         PR rtl-optimization/55547
28658         PR rtl-optimization/53827
28659         PR debug/53671
28660         PR debug/49888
28661         * alias.c (offset_overlap_p): New, factored out of...
28662         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
28663         the conservative special case for symbolic constants.  Don't
28664         adjust zero sizes on alignment.
28666 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
28668         PR rtl-optimization/52573
28669         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
28670         REG_UNUSED for the same register.
28672 2013-01-17  Richard Biener  <rguenther@suse.de>
28673             Marek Polacek  <polacek@redhat.com>
28675         PR rtl-optimization/55833
28676         * loop-unswitch.c (unswitch_loops): Move loop verification...
28677         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
28678         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
28679         Set it to true when we're removing a loop from hierarchy tree in
28680         an irreducible region.
28681         (fix_bb_placements): Adjust caller.
28682         (fix_loop_placements): Likewise.
28684 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
28686         * config/avr/builtins.def (DEF_BUILTIN): Factor out
28687         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
28688         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
28689         Remove ID.  Adjust comments.
28690         * config/avr/avr-c.c (avr_builtin_name): Remove.
28691         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
28692         * config/avr/avr.c (avr_tolower): New static function.
28693         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
28694         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
28695         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
28696         default expansion.
28698 2013-01-17  Jan Hubicka  <jh@suse.cz>
28700         PR tree-optimization/55273
28701         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
28703 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
28705         PR target/55981
28706         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
28707         store through atomic_store<mode>_1.
28708         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
28710 2013-01-17  Martin Jambor  <mjambor@suse.cz>
28712         PR tree-optimizations/55264
28713         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
28714         for virtual methods.
28715         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
28716         virtual methods before inlining is over.
28717         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
28718         virtual functions.
28719         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
28720         non-virtual.
28722 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
28724         PR rtl-optimization/56005
28725         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
28726         pending reads for prefetch.
28728 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
28730         * config/aarch64/aarch64.md
28731         (*cstoresi_neg_uxtw): New pattern.
28732         (*cmovsi_insn_uxtw): New pattern.
28733         (*<optab>si3_uxtw): New pattern.
28734         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
28735         (*<optab>si3_insn_uxtw): New pattern.
28736         (*bswapsi2_uxtw): New pattern.
28738 2013-01-16  Richard Biener  <rguenther@suse.de>
28740         * tree-inline.c (tree_function_versioning): Remove set but
28741         never used variable.
28743 2013-01-16  Richard Biener  <rguenther@suse.de>
28745         PR tree-optimization/55964
28746         * tree-flow.h (rename_variables_in_loop): Remove.
28747         (rename_variables_in_bb): Likewise.
28748         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
28749         (copy_loop_before): Adjust and delete update-ssa status.
28750         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
28751         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
28752         (rename_variables_in_loop): Remove.
28753         (slpeel_update_phis_for_duplicate_loop): Likewise.
28754         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
28755         use available cfg machinery instead of duplicating it.
28756         Update PHI nodes and perform poor-mans SSA update here.
28757         (slpeel_tree_peel_loop_to_edge): Adjust.
28759 2013-01-16  Richard Biener  <rguenther@suse.de>
28761         PR tree-optimization/54767
28762         PR tree-optimization/53465
28763         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
28764         (vrp_visit_phi_node): For PHI arguments coming via backedges
28765         drop all symbolical range information.
28766         (execute_vrp): Compute backedges.
28768 2013-01-16  Richard Biener  <rguenther@suse.de>
28770         * doc/install.texi: Update CLooG and ISL requirements to
28771         0.18.0 and 0.11.1.
28773 2013-01-16  Christian Bruel  <christian.bruel@st.com>
28775         PR target/55301
28776         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
28777         (broken_move): Handle UNSPECV_SP_SWITCH_B.
28778         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
28780 2013-01-16  DJ Delorie  <dj@redhat.com>
28782         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
28783         (UNSPECV_SP_SWITCH_E): New.
28784         (sp_switch_1): Change to an unspec.
28785         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
28786         replace $r15.
28788 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
28790         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
28791         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
28792         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
28793         (expand_mem_thread_fence): Ditto.
28794         (expand_mem_signal_fence): Ditto.
28795         (expand_atomic_load): Ditto.
28796         (expand_atomic_store): Ditto.
28798 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
28800         PR rtl-optimization/55547
28801         PR rtl-optimization/53827
28802         PR debug/53671
28803         PR debug/49888
28804         * alias.c (memrefs_conflict_p): Set sizes to negative after
28805         AND adjustments.
28807 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28809         PR target/55940
28810         * function.c (thread_prologue_and_epilogue_insns): Always
28811         add crtl->drap_reg to set_up_by_prologue.set, even if
28812         stack_realign_drap is false.
28814 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
28816         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
28817         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
28818         *call): Fix indention.
28820 2013-01-15  Tom de Vries  <tom@codesourcery.com>
28822         PR target/55876
28823         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
28824         Update comment.
28826 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
28828         PR rtl-optimization/55153
28829         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
28831 2013-01-15  Martin Jambor  <mjambor@suse.cz>
28833         PR tree-optimization/55920
28834         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
28835         accesses as grp_to_be_debug_replaced.
28837 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28839         PR tree-optimization/55920
28840         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
28841         there is non-useless type conversion needed from debug rhs to lhs,
28842         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
28844 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
28845             Mikael Pettersson  <mikpe@it.uu.se>
28847         PR target/43961
28848         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
28849         Thumb.
28850         (ASM_OUTPUT_CASE_LABEL): Remove.
28851         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
28852         * final.c (shorten_branches): Update alignment of labels before
28853         jump tables if CASE_VECTOR_SHORTEN_MODE.
28855 2013-01-15  Richard Biener  <rguenther@suse.de>
28857         PR bootstrap/55961
28858         * system.h: Do not include gmp.h for building host tools.
28860 2013-01-15  Richard Biener  <rguenther@suse.de>
28862         PR middle-end/55882
28863         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
28864         account for bitpos when computing alignment.
28866 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
28868         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
28869         (ix86_target_macros_internal): Likewise.
28871         * config/i386/i386.c (m_CORE2I7): Removed.
28872         (m_CORE_HASWELL): New macro.
28873         (m_CORE_ALL): Likewise.
28874         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
28875         (initial_ix86_arch_features): Likewise.
28876         (processor_target_table): Initializations for Core avx2.
28877         (cpu_names): New names "core-avx2".
28878         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
28879         PROCESSOR_CORE_HASWELL.
28880         (ix86_issue_rate): New case.
28881         (ia32_multipass_dfa_lookahead): Likewise.
28882         (ix86_sched_init_global): Likewise.
28884         * config/i386/i386.h (TARGET_HASWELL): New macro.
28885         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
28886         (processor_type): New PROCESSOR_HASWELL.
28888 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
28890         PR tree-optimization/55955
28891         * tree-vect-loop.c (vectorizable_reduction): Give up early on
28892         *SHIFT_EXPR and *ROTATE_EXPR codes.
28894         PR tree-optimization/48766
28895         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
28896         -ftrapv disable -fwrapv.
28898 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28900         PR target/55974
28901         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
28902         etc. to 1 and not to __flash.
28903         Use LL suffix for __INT24_MAX__ with -mint8.
28904         Use ULL suffix for __UINT24_MAX__ with -mint8.
28906 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28908         * config/avr/avr-arch.h
28909         (struct base_arch_s): Use typedef avr_arch_t instead.
28910         (struct arch_info_s): Use typedef avr_arch_info_t instead.
28911         (struct mcu_type_s): Use typedef avr_mcu_t instead.
28912         * config/avr/avr.c: Same.
28913         * config/avr/avr-devices.c: Same.
28914         * config/avr/driver-avr.c: Same.
28915         * config/avr/gen-avr-mmcu-texi.c: Same.
28916         * config/avr/avr-mcus.def: Adjust comment.
28918 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
28920         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
28921         * config/aarch64/iterators.md (VALLDI): New.
28923 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
28924             Andi Kleen  <ak@linux.intel.com>
28926         PR target/55948
28927         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
28928         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
28929         memmodel flag.
28931 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28933         * config/avr/avr-stdint.h: Remove trailing blanks.
28934         * config/avr/avr-log.h: Same.
28935         * config/avr/avr-arch.h: Same.
28936         * config/avr/avr-devices.c: Same.
28937         * config/avr/avr-dimode.md: Same.
28938         * config/avr/predicates.md: Same.
28939         * config/avr/avr-c.c: Same.  And fix typo.
28941         * config/avr/avr-protos.h: Same.  And:
28942         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
28943         (init_cumulative_args): Rename to avr_init_cumulative_args.
28944         (expand_prologue): Rename to avr_expand_prologue.
28945         (expand_epilogue): Rename to avr_expand_epilogue.
28946         (adjust_insn_length): Rename to avr_adjust_insn_length.
28947         (notice_update_cc): Rename to avr_notice_update_cc.
28948         (final_prescan_insn): Rename to avr_final_prescan_insn.
28949         * config/avr/avr.c: Same.
28950         * config/avr/avr.h: Same.
28951         * config/avr/avr.md: Remove trailing blanks.
28952         (prologue): Use avr_expand_prologue.
28953         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
28955 2013-01-14  Richard Biener  <rguenther@suse.de>
28957         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
28958         verify_location, collect_subblocks): New functions.
28959         (verify_gimple_in_cfg): Verify that locations only reference
28960         BLOCKs in the functions BLOCK tree.
28962 2013-01-14  Richard Biener  <rguenther@suse.de>
28964         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
28965         PHI argument.
28966         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
28967         unshare reference.
28968         (insert_out_of_ssa_copy_on_edge): Likewise.
28969         (rewrite_close_phi_out_of_ssa): Likewise.
28970         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
28971         debug expressions.
28972         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
28973         propagated constants.
28974         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
28975         can not be shared.
28977 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
28979         * config/avr/avr-modes.def: Add GPL copyright notice.
28981 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
28983         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
28984         MEMMODEL_MASK to determine memory model.
28985         (atomic_store<mode>): Ditto from operands[2].
28986         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
28988 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
28990         PR fortran/55935
28991         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
28992         (fold_gimple_assign): Don't call unshare_expr here.
28993         (fold_ctor_reference): Call unshare_expr.
28995 2013-01-13  Terry Guo  <terry.guo@arm.com>
28997         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
28998         * doc/fragments.texi: Document MULTILIB_REUSE.
28999         * gcc.c (multilib_reuse): New internal spec.
29000         (set_multilib_dir): Also search multilib from multilib_reuse.
29001         * genmultilib (tmpmultilib3): Refactor code.
29002         (tmpmultilib4): Ditto.
29003         (multilib_reuse): New multilib argument.
29005 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
29007         * Makefile.in: Update copyright.
29009 2013-01-12  Tom de Vries  <tom@codesourcery.com>
29011         PR middle-end/55890
29012         * calls.c (expand_call): Check if arg_nr is valid.
29014 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
29016         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
29017         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
29018         documentation.  Add missing '__' in front of
29019         __builtin_ia32_packssdw256.
29021 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29023         PR target/55719
29024         * config/s390/s390.c (s390_preferred_reload_class): Do not return
29025         NO_REGS for larl operands.
29026         (s390_reload_larl_operand): Use s390_load_address instead of
29027         emit_move_insn.
29029 2013-01-11  Richard Biener  <rguenther@suse.de>
29031         * tree-cfg.c (verify_node_sharing_1): Split out from ...
29032         (verify_node_sharing): ... here.
29033         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
29035 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
29037         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
29038         Substitute TREECHECKING.
29039         * configure: Regenerate.
29040         * Makefile.in (TREECHECKING): New.
29042 2013-01-11  Richard Guenther  <rguenther@suse.de>
29044         PR tree-optimization/44061
29045         * tree-vrp.c (extract_range_basic): Compute zero as
29046         value-range for __builtin_constant_p of function parameters.
29048 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
29050         Update copyright years.
29052 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
29054         PR rtl-optimization/55672
29055         * lra-eliminations.c (mark_not_eliminable): Permit addition with
29056         const to be eliminable.
29058 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
29060         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
29061         * configure: Regenerate.
29063 2013-01-10  Richard Biener  <rguenther@suse.de>
29065         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
29067 2013-01-10  Richard Biener  <rguenther@suse.de>
29069         PR bootstrap/55792
29070         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
29071         locations for virtual PHI arguments.
29072         (rewrite_update_phi_arguments): Likewise.
29074 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
29076         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
29077         on to assembler.
29079 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
29081         PR tree-optimization/55921
29082         * tree-complex.c (expand_complex_asm): New function.
29083         (expand_complex_operations_1): Call it for GIMPLE_ASM.
29085 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29087         PR target/55718
29088         * config/s390/s390.c (s390_symref_operand_p)
29089         (s390_loadrelative_operand_p): Merge the two functions.
29090         (s390_check_qrst_address, print_operand_address): Add parameters
29091         to s390_loadrelative_operand_p invokation.
29092         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
29093         (s390_reload_larl_operand, s390_secondary_reload): Use
29094         s390_loadrelative_operand_p instead of s390_symref_operand_p.
29095         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
29097 2013-01-09  Mike Stump  <mikestump@comcast.net>
29099         * dse.c (record_store): Remove unnecessary assert.
29101 2013-01-09  Jan Hubicka  <jh@suse.cz>
29103         PR tree-optimization/55569
29104         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
29105         * cfgloop.h (scale_loop_profile): Likewise.
29107 2013-01-09  Jan Hubicka  <jh@suse.cz>
29109         PR lto/45375
29110         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
29111         functions.
29112         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
29114 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
29116         PR middle-end/55114
29117         * expr.h (maybe_emit_group_store): Declare.
29118         * expr.c (maybe_emit_group_store): New function.
29119         * builtins.c (expand_builtin_int_roundingfn): Call it.
29120         (expand_builtin_int_roundingfn_2): Likewise.
29122 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
29124         PR rtl-optimization/55829
29125         * lra-constraints.c (match_reload): Add code for absent output.
29126         (curr_insn_transform): Add code for reloads of matched inputs
29127         without output.
29129 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
29131         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
29132         attribute of movddup insn to DF.
29133         (*vec_interleave_lowv2df): Ditto.
29134         (vec_dupv2df): Ditto.
29136 2013-01-09  Jan Hubicka  <jh@suse.cz>
29138         PR tree-optimiation/55875
29139         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
29140         EVERY_ITERATION parameter.
29141         (number_of_iterations_exit): Check if exit is executed every iteration.
29142         (idx_infer_loop_bounds): Similarly here.
29143         (n_of_executions_at_most): Simplify
29144         to only test for cases where statement is dominated by the
29145         particular bound; handle correctly the "postdominance" test.
29146         (scev_probably_wraps_p): Use max loop iterations info
29147         as a global bound first.
29149 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
29150             Nick Clifton  <nickc@redhat.com>
29152         * config/v850/v850.md (cbranchsf4): New pattern.
29153         (cstoresf4): New pattern.
29154         (cbranchdf4): New pattern.
29155         (cstoredf4): New pattern.
29156         (movsicc): Disallow floating point comparisons.
29157         (cmpsf_le_insn): Fix order of operators.
29158         (cmpsf_lt_insn): Likewise.
29159         (cmpsf_eq_insn): Likewise.
29160         (cmpdf_le_insn): Likewise.
29161         (cmpdf_lt_insn): Likewise.
29162         (cmpdf_eq_insn): Likewise.
29163         (cmpsf_ge_insn): Use LE comparison.
29164         (cmpdf_ge_insn): Likewise.
29165         (cmpsf_gt_insn): Use LT comparison.
29166         (cmpdf_gt_insn): Likewise.
29167         (cmpsf_ne_insn): Delete pattern.
29168         (cmpdf_ne_insn): Delete pattern.
29169         * config/v850/v850.c (v850_gen_float_compare): Use
29170         gen_cmpdf_eq_insn for NE comparison.
29171         (v850_float_z_comparison_operator)
29172         (v850_float_nz_comparison_operator): Move from here ...
29173         * config/v850/predicates.md: ... to here.  Move GT and GE
29174         comparisons into v850_float_z_comparison_operator.
29175         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
29176         Delete prototype.
29177         (v850_float_nz_comparison_operator): Likewise.
29179 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29181         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
29182         with calls to gen_insvsi/gen_insvdi.
29184 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
29186         * config/i386/i386.c (initial_ix86_tune_features): Set up
29187         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
29189 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
29190             Jakub Jelinek  <jakub@redhat.com>
29192         PR tree-optimization/48189
29193         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
29194         If nitercst is 0, don't predict the exit edge.
29196 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
29198         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
29199         in asm_fprintf with reg_names.
29200         (aarch64_print_operand_address): Likewise.
29201         (aarch64_return_addr): Likewise.
29202         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
29204 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29206         * config/pa/pa.h (VAL_U6_BITS_P): Define.
29207         (INT_U6_BITS): Likewise.
29208         * config/pa/predicates.md (uint6_operand): New predicate.
29209         (shift5_operand, shift6_operand): Likewise.
29210         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
29211         arith32_operand.
29212         (lshrdi3): Use shift6_operand.
29213         (shrpsi4, shrpdi4): New insn patterns.
29214         (extzv): Delete expander.
29215         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
29216         predicates in unamed zero extract patterns.  Tighten common constraint.
29217         (extv): Delete expander.
29218         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
29219         predicates in unamed sign extract patterns.  Tighten common constraint.
29220         (insv): Delete expander.
29221         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
29222         predicates in unamed insert patterns.  Tighten common constraint.
29223         Change uint32_operand predicate to uint6_operand predicate in unamed
29224         DImode pattern to insert constant values of type 1...1xxxx.
29226 2013-01-04  Jan Hubicka  <jh@suse.cz>
29228         PR tree-optimization/55823
29229         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
29230         issue.
29232 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29233             Uros Bizjak  <ubizjak@gmail.com>
29235         PR rtl-optimization/55845
29236         * df-problems.c (can_move_insns_across): Stop scanning at
29237         volatile_insn_p source instruction or give up if
29238         across_from .. across_to range contains any volatile_insn_p
29239         instructions.
29241 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29243         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
29244         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
29245         Declare.
29246         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
29247         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
29249 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29251         PR fortran/55341
29252         * asan.c (asan_clear_shadow): New function.
29253         (asan_emit_stack_protection): Use it.
29255 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29257         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
29258         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
29259         with tab instead of space.
29261 2013-01-08  Nick Clifton  <nickc@redhat.com>
29263         * config/rl78/rl78.c (rl78_expand_prologue): Always select
29264         register bank 0 at the start of an interrupt handler.
29265         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
29266         MDBH registers.
29268 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29270         * config/aarch64/aarch64-simd.md
29271         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
29272         (aarch64_simd_bsl): Likewise.
29273         (aarch64_vcond_internal<mode>): Likewise.
29274         (vcond<mode><mode>): Likewise.
29275         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
29276         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
29278 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29280         * config/aarch64/aarch64-builtins.c
29281         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
29283 2013-01-08  Martin Jambor  <mjambor@suse.cz>
29285         PR debug/55579
29286         * tree-sra.c (analyze_access_subtree): Return true also after
29287         potentially creating a debug-only replacement.
29289 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29291         PR middle-end/55890
29292         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
29294         PR tree-optimization/54120
29295         * tree-vrp.c (range_fits_type_p): Don't allow
29296         src_precision < precision from signed vr to unsigned_p
29297         if vr->min or vr->max is negative.
29298         (simplify_float_conversion_using_ranges): Test can_float_p
29299         against CODE_FOR_nothing.
29301 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29302             Richard Biener  <rguenther@suse.de>
29304         PR middle-end/55851
29305         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
29306         types instead of just INTEGER_TYPE types.
29308 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
29310         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
29311         TF_SIZE): Define.
29313 2013-01-07  Steve Ellcey  <sellcey@mips.com>
29315         PR target/42661
29316         * config/mips/mips.opt: Change mad to mmad to match documentation.
29318 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29320         PR target/55897
29321         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
29322         .progmemx.data now.
29324 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29326         PR target/55897
29327         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
29328         (avr_addrspace_t): Add .section_name field.
29329         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
29330         array size.
29331         (avr_addrspace): Same.  Initialize .section_name.  Remove last
29332         NULL entry.  Put __memx into .progmemx.data.
29333         (progmem_section_prefix): Remove.
29334         (avr_asm_init_sections): No need to initialize progmem_section.
29335         (avr_asm_named_section): Use avr_addrspace[].section_name to get
29336         section name prefix.
29337         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
29338         retrieve the progmem section.
29339         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
29340         boundary to run over avr_addrspace[].
29341         (avr_register_target_pragmas): Ditto.
29343 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
29345         * varasm.c (output_constant_def_contents): For asan_protect_global
29346         protected strings, adjust DECL_ALIGN if needed, before testing for
29347         anchored symbols.
29348         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
29349         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
29350         normal decls.
29351         (output_object_block): For asan protected decls, emit asan padding
29352         after their contents.
29353         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
29354         (asan_finish_file): Test it here instead.
29356 2013-01-07  Nick Clifton  <nickc@redhat.com>
29357             Matthias Klose  <doko@debian.org>
29358             Doug Kwan  <dougkwan@google.com>
29359             H.J. Lu  <hongjiu.lu@intel.com>
29361         PR driver/55470
29362         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
29364         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
29366         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
29368         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
29370         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
29372 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29374         PR target/54461
29375         * doc/install.texi (Cross-Compiler-Specific Options): Document
29376         --with-avrlibc.
29378 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
29380         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
29381         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
29382         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
29383         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
29384         vqmovun_high_s64): Fix source operand number and update copyright.
29386 2013-01-07  Richard Biener  <rguenther@suse.de>
29388         PR middle-end/55890
29389         * gimple.h (gimple_call_builtin_p): New overload.
29390         * gimple.c (validate_call): New function.
29391         (gimple_call_builtin_p): Likewise.
29392         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29393         Use gimple_call_builtin_p.
29394         (find_func_clobbers): Likewise.
29395         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
29396         (strlen_optimize_stmt): Likewise.
29398 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29400         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
29401         (vld1q_dup_*): Likewise.
29402         (vld1_*): Likewise.
29403         (vld1q_*): Likewise.
29404         (vld1_lane_*): Likewise.
29405         (vld1q_lane_*): Likewise.
29407 2013-01-07  Richard Biener  <rguenther@suse.de>
29409         * lto-streamer.h (LTO_minor_version): Bump to 2.
29411 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29413         * config/aarch64/aarch64-protos.h
29414         (aarch64_const_double_zero_rtx_p): Rename to...
29415         (aarch64_float_const_zero_rtx_p): ...this.
29416         (aarch64_float_const_representable_p): New.
29417         (aarch64_output_simd_mov_immediate): Likewise.
29418         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
29419         move immediate case.
29420         * config/aarch64/aarch64.c
29421         (aarch64_const_double_zero_rtx_p): Rename to...
29422         (aarch64_float_const_zero_rtx_p): ...this.
29423         (aarch64_print_operand): Allow printing of new constants.
29424         (aarch64_valid_floating_const): New.
29425         (aarch64_legitimate_constant_p): Check for valid floating-point
29426         constants.
29427         (aarch64_simd_valid_immediate): Likewise.
29428         (aarch64_vect_float_const_representable_p): New.
29429         (aarch64_float_const_representable_p): Likewise.
29430         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
29431         (aarch64_output_simd_mov_immediate): New.
29432         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
29433         (*movdf_aarch64): Likewise.
29434         * config/aarch64/constraints.md (Ufc): New.
29435         (Y): call aarch64_float_const_zero_rtx.
29436         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
29438 2013-01-07  Richard Biener  <rguenther@suse.de>
29440         PR tree-optimization/55888
29441         PR tree-optimization/55862
29442         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
29443         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
29444         not if it is contained therein.
29446 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29448         * config/avr/t-avr: Typo.
29450 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29452         PR55243
29453         * config/avr/t-avr: Don't automatically rebuild
29454         $(srcdir)/config/avr/t-multilib
29455         $(srcdir)/config/avr/avr-tables.opt
29456         $(srcdir)/doc/avr-mmcu.texi
29457         (avr-mcus): New phony target to build them on request.
29458         (s-avr-mlib, s-avr-mmcu-texi): Remove.
29459         * avr/avr-mcus.def: Adjust comments.
29461 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
29463         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
29465 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29467         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
29469 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29471         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
29473 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
29475         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
29476         to generate profiling.
29477         * config/rs6000/aix64.h (LIB_SPEC): Same.
29479 2013-01-04  Andrew Pinski  <apinski@cavium.com>
29481         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
29482         New function.
29483         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
29485 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
29487         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
29488         unconditionally.
29489         (ix86_expand_move): Ditto.
29490         (ix86_zero_extend_to_Pmode): Ditto.
29491         (ix86_expand_call): Ditto.
29492         (ix86_expand_special_args_builtin): Ditto.
29493         (ix86_expand_builtin): Ditto.
29495 2013-01-04  Richard Biener  <rguenther@suse.de>
29497         PR tree-optimization/55862
29498         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
29499         translating them through PHI nodes.
29501 2013-01-04  Martin Jambor  <mjambor@suse.cz>
29503         PR tree-optimization/55755
29504         * tree-sra.c (sra_modify_assign): Do not check that an access has no
29505         children when trying to avoid producing a VIEW_CONVERT_EXPR.
29507 2013-01-04  Marek Polacek  <polacek@redhat.com>
29509         PR middle-end/55859
29510         * opts.c (default_options_optimization): Clarify error message.
29512 2013-01-04  Richard Biener  <rguenther@suse.de>
29514         PR middle-end/55863
29515         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
29516         reassociation.
29518 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29520         PR target/53789
29521         * config/pa/pa.md (movsi): Revert previous change.
29522         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
29523         references.
29525 2013-01-03  Richard Henderson  <rth@redhat.com>
29527         * config/i386/i386.c (ix86_expand_move): Always assign to op1
29528         after eliminating TLS symbols.
29530 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
29532         PR bootstrap/50167
29533         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
29534         * graphite-poly.c (debug_gmp_value): Likewise.
29536 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
29538         PR target/55712
29539         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
29540         selected code model, define __code_mode_small__, __code_model_medium__,
29541         __code_model_large__, __code_model_32__ or __code_model_kernel__.
29542         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
29543         xchg temporary register with %k.  Declare temporary register as
29544         early clobbered.
29545         [__x86_64__]: For medium and large code models, preserve %rbx register.
29547 2013-01-03  Richard Biener  <rguenther@suse.de>
29549         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
29550         (dump_subscript): Adjust.
29551         (finalize_ddr_dependent): Do not dump redundant info.
29552         (analyze_siv_subscript): Adjust.
29553         (subscript_dependence_tester): Likewise.
29554         (compute_affine_dependence): Likewise.
29556 2013-01-03  Richard Biener  <rguenther@suse.de>
29558         Revert
29559         2013-01-03  Richard Biener  <rguenther@suse.de>
29561         PR tree-optimization/55857
29562         * tree-vect-stmts.c (vectorizable_load): Do not setup
29563         re-alignment for invariant loads.
29565         2013-01-02  Richard Biener  <rguenther@suse.de>
29567         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29568         invariant load do not generate a vector load from the scalar location.
29570 2013-01-03  Richard Biener  <rguenther@suse.de>
29572         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
29573         for not vectorizing.
29574         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
29575         not build INDIRECT_REFs, call get_name once only.
29576         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
29577         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
29579 2013-01-03  Richard Biener  <rguenther@suse.de>
29581         PR tree-optimization/55857
29582         * tree-vect-stmts.c (vectorizable_load): Do not setup
29583         re-alignment for invariant loads.
29585 2013-01-03  Richard Biener  <rguenther@suse.de>
29587         PR lto/55848
29588         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
29589         prefer a built-in decl.
29591 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29593         * gcc.c (process_command): Update copyright notice dates.
29594         * gcov.c (print_version): Likewise.
29595         * gcov-dump.c (print_version): Likewise.
29597         PR rtl-optimization/55838
29598         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
29599         iv0.step, iv1.step and step.
29601 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29602             Marc Glisse  <marc.glisse@inria.fr>
29604         PR tree-optimization/55832
29605         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
29606         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
29607         integer_{one,zero}_node.
29609 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29611         PR debug/54402
29612         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
29613         * var-tracking.c (reverse_op): Don't add reverse ops to
29614         VALUEs that have already
29615         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
29617 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
29619         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
29621 2013-01-02  Teresa Johnson  <tejohnson@google.com>
29623         * dumpfile.c (dump_loc): Print filename with location.
29624         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
29625         new location_t parameter to emit complete unroll message with
29626         new dump framework.
29627         (canonicalize_loop_induction_variables): Compute loops location
29628         and pass to try_unroll_loop_completely.
29629         * loop-unroll.c (report_unroll_peel): New function.
29630         (peel_loops_completely): Use new dump format with location
29631         for main dumpfile message, and invoke report_unroll_peel on success.
29632         (decide_unrolling_and_peeling): Ditto.
29633         (decide_peel_once_rolling): Remove old dumpfile message subsumed
29634         by report_unroll_peel.
29635         (decide_peel_completely): Ditto.
29636         (decide_unroll_constant_iterations): Ditto.
29637         (decide_unroll_runtime_iterations): Ditto.
29638         (decide_peel_simple): Ditto.
29639         (decide_unroll_stupid): Ditto.
29640         * cfgloop.c (get_loop_location): New function.
29641         * cfgloop.h (get_loop_location): Declare.
29643 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29645         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
29646         NULL.
29648 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29650         PR middle-end/55198
29651         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
29652         BLKmode objects when EXPAND_MEMORY is specified.
29654 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29656         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
29657         in loop predicate.
29658         (fold_builtin_cpu): Do not share cpu model decls across statements.
29660 2013-01-02  Jason Merrill  <jason@redhat.com>
29662         PR c++/55804
29663         * tree.c (build_array_type_1): Revert earlier change.
29665 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
29667         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
29668         "cortex-a57".
29669         * config/aarch64/aarch64-tune.md: Re-generate.
29671 2013-01-02  Richard Biener  <rguenther@suse.de>
29673         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29674         invariant load do not generate a vector load from the scalar location.
29676 2013-01-02  Richard Biener  <rguenther@suse.de>
29678         PR bootstrap/55784
29679         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
29680         * configure: Regenerate.
29682 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29684         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
29685         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
29686         (expand_builtin_int_roundingfn_2): Keep the original target around
29687         for the fallback case.
29689 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29691         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
29692         to be clear for sign changes.
29694 2013-01-01  Jan Hubicka  <jh@suse.cz>
29696         * ipa-inline-analysis.c: Fix formatting.
29698 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
29700         PR tree-optimization/55831
29701         * tree-vect-loop.c (get_initial_def_for_induction): Use
29702         gsi_after_labels instead of gsi_start_bb.
29704 Copyright (C) 2013 Free Software Foundation, Inc.
29706 Copying and distribution of this file, with or without modification,
29707 are permitted in any medium without royalty provided the copyright
29708 notice and this notice are preserved.