* Makefile.in: Add ipa-profile.o
[official-gcc.git] / gcc / ChangeLog
bloba4954b360e6a1d654207867e7e4aebb49cbac6da
1 2013-09-01  Jan Hubicka  <jh@suse.cz>
3         * Makefile.in: Add ipa-profile.o
4         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
5         * cgraph.c (struct cgraph_propagate_frequency_data,
6         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7         ipa-profile.c; replace cgraph_ by ipa_ prefix.
8         * cgraph.h (cgraph_propagate_frequency): Remove.
9         * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
10         (inline_update_callee_summaries): Update.
11         * ipa-profile.c: New file.
12         * ipa-utils.h (ipa_propagate_frequency): Declare.
13         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
14         data-streamer.h, value-prof.h
15         (symtab_remove_unreachable_nodes): Update profile.
16         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
17         account_time_size, cmp_counts, dump_histogram,
18         ipa_profile_generate_summary, ipa_profile_write_summary,
19         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
20         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
21         Move to ipa-profile.c
23 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
25         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
27 2013-09-01  Jan Hubicka  <jh@suse.cz>
29         * common.opt (fdevirtualize-speculatively): New function.
30         * invoke.texi (fdevirtualize-speculatively): Document.
31         * ipa-devirt.c: Include ipa-inline.h
32         (likely_target_p): New function.
33         (ipa_devirt): New function.
34         (gate_ipa_devirt): New function.
35         (pass_data_ipa_devirt): New static var.
36         (pass_ipa_devirt): Likewise.
37         (make_pass_ipa_devirt): New function.
38         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
39         (common_handle_option): Disable devirtualization when
40         value range profiling is available.
41         * passes.def (pass_ipa_devirt): Add.
42         * timever.def (TV_IPA_DEVIRT): New timevar.
43         * tree-pass.h (make_pass_ipa_devirt): 
45 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
47         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
48         include sanitize(undefined).
50 2013-08-31  Jan Hubicka  <jh@suse.cz>
52         * bulitins.c (expand_builtin): Do not early exit for gcov
53         instrumented functions.
55 2013-08-31  Marek Polacek  <polacek@redhat.com>
57         * ubsan.c: Include tm_p.h.
59 2013-08-31  Jan Hubicka  <jh@suse.cz>
61         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
62         warning.
64         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
65         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
66         * tree-cfg.c (verify_gimple_label): ... here.
67         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
68         (ipa_merge_profiles): New function.
69         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
70         (lto_input_function_body): Likewise.
71         * ipa-utils.h (ipa_merge_profiles): Declare.
72         * lto-streamer.h (lto_input_function_body): Update prototype.
73         (emit_label_in_global_context_p): Remove.
74         * lto-symtab.c: Include ipa-utils.h
75         (lto_cgraph_replace_node): Use ipa_merge_profiles.
77 2013-08-31  Jan Hubicka  <jh@suse.cz>
79         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
81 2013-08-31  Jan Hubicka  <jh@suse.cz>
83         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
85 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
87         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
88         "cmp" RTX before signed_comparison_operator check to account
89         for "code" changes.
91 2013-08-30  Jan Hubicka  <jh@suse.cz>
93         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
94         (detect_type_change_1): Rename to ...
95         (detect_type_change): ... this one; early return on non-polymorphic
96         types.
97         (detect_type_change_ssa): Add comp_type parameter; update
98         use of detect_type_change.
99         (compute_complex_assign_jump_func): Add param_type parameter;
100         update use of detect_type_change_ssa.
101         (compute_complex_ancestor_jump_func): Likewise.
102         (ipa_get_callee_param_type): New function.
103         (ipa_compute_jump_functions_for_edge): Compute parameter type;
104         update calls to the jump function computation functions.
106 2013-08-30  Teresa Johnson  <tejohnson@google.com>
107             Steven Bosscher  <steven@gcc.gnu.org>
109         * cfgrtl.c (fixup_new_cold_bb): New routine.
110         (commit_edge_insertions): Invoke fixup_partitions.
111         (find_partition_fixes): New routine.
112         (fixup_partitions): Ditto.
113         (verify_hot_cold_block_grouping): Update comments.
114         (rtl_verify_edges): Invoke find_partition_fixes.
115         (rtl_verify_bb_pointers): Update comments.
116         (rtl_verify_bb_layout): Ditto.
117         * basic-block.h (probably_never_executed_edge_p): Declare.
118         (fixup_partitions): Ditto.
119         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
120         * bb-reorder.c (sanitize_hot_paths): New function.
121         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
122         sanitize_hot_paths.
123         * predict.c (probably_never_executed_edge_p): New routine.
124         * cfg.c (check_bb_profile): Add partition insanity warnings.
126 2013-08-30  Meador Inge  <meadori@codesourcery.com>
128         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
130 2013-08-30  Marek Polacek  <polacek@redhat.com>
132         * Makefile.in (ubsan.o): Add.
133         (c-family/c-ubsan.o): Add.
134         (builtins.o): Add ubsan.h dependency.
135         * ubsan.h: New file.
136         * ubsan.c: New file.
137         * common.opt: Add -fsanitize=undefined option.
138         (flag_sanitize): Add variable.
139         (fsanitize=): Add option.  Add Driver.
140         (fsanitize=thread): Remove option.
141         (fsanitize=address): Likewise.
142         (static-libubsan): New option.
143         * doc/invoke.texi: Document the new flag and -static-libubsan.
144         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
145         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
146         * builtin-attrs.def (ATTR_COLD): Define.
147         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
148         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
149         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
150         * flag-types.h (sanitize_code): New enum.
151         * opts.c (common_handle_option): Parse command line arguments
152         of -fsanitize=.  Add -fsanitize=unreachable option.
153         * varasm.c (get_variable_section): Adjust.
154         (assemble_noswitch_variable): Likewise.
155         (assemble_variable): Likewise.
156         (output_constant_def_contents): Likewise.
157         (categorize_decl_for_section): Likewise.
158         (place_block_symbol): Likewise.
159         (output_object_block): Likewise.
160         * builtins.def: Likewise.
161         * toplev.c (compile_file): Likewise.
162         (process_options): Likewise.
163         * cppbuiltin.c: Likewise.
164         * tsan.c (tsan_pass): Likewise.
165         (tsan_gate): Likewise.
166         (tsan_gate_O0): Likewise.
167         * cfgexpand.c (partition_stack_vars): Likewise.
168         (expand_stack_vars): Likewise.
169         (defer_stack_allocation): Likewise.
170         (expand_used_vars): Likewise.
171         * cfgcleanup.c (old_insns_match_p): Likewise.
172         * asan.c (asan_finish_file): Likewise.
173         (asan_instrument): Likewise.
174         (gate_asan): Likewise.
175         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
176         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
177         (asan_global_struct): Use pointer_sized_int_node instead
178         calling build_nonstandard_integer_type.
179         (initialize_sanitizer_builtins): Likewise.
180         (asan_finish_file): Likewise.
181         * gcc.c: Document %{%:function(args):X}.
182         (static_spec_functions): Add sanitize.
183         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
184         store funcval != NULL there.
185         (do_spec_1): Adjust handle_spec_function caller.
186         (handle_braces): Allow %:function(args) as condition.
187         (sanitize_spec_function): New function.
188         (ADD_STATIC_LIBUBSAN_LIBS): Define.
189         (LIBUBSAN_SPEC): Likewise.
190         (LIBUBSAN_EARLY_SPEC): Likewise.
191         (SANITIZER_SPEC): Handle libubsan.
192         (SANITIZER_EARLY_SPEC): Likewise.
193         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
194         instead of fsanitize=address.
195         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
196         instead of fsanitize=address*.
197         * builtins.c: Include ubsan.h.
198         (fold_builtin_0): Instrument __builtin_unreachable.
199         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
200         instead of flag_asan.
201         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
202         (pointer_sized_int_node): Define.
203         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
205 2013-08-30  Mike Stump  <mikestump@comcast.net>
207         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
208         with RE patterns.
210 2013-08-29  Jan Hubicka  <jh@suse.cz>
212         * cgraph.c (cgraph_function_body_availability): Handle weakref
213         correctly.
214         * passes.def: Remove pass_fixup_cfg.
215         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
216         track when we need to remove functions.
217         (gate_ipa_inline): Execute inlining always; add comment why.
218         (pass_data_ipa_inline): Remove TODO_remove_functions.
219         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
220         do not produce summaries.
221         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
222         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
223         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
224         constant pool and vtable.
226 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
228         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
229         New arm_neon.h's internal macros to specify 64-bit constants.
230         Avoid using stdint.h's macros.
232 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
234         * recog.c (verify_changes): Verify that changes[i].old is non-zero
235         before applying REG_P.
237 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
239         PR tree-optimization/58277
240         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
241         after seeing too many stmts with vdef in between dombb and current
242         bb, invalidate everything.
244 2013-08-30  Richard Biener  <rguenther@suse.de>
246         * fold-const.c (fold_single_bit_test): Fix overflow test.
248 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
250         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
251         and which can live in a register, always retrieve the value on entry.
252         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
253         passed by invisible reference specially.
254         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
255         (vt_add_function_parameter): Correctly deal with a parameter passed by
256         invisible reference.
258 2013-08-30  Jan Hubicka  <jh@suse.cz>
260         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
262 2013-08-30  Richard Biener  <rguenther@suse.de>
264         PR tree-optimization/58228
265         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
266         allow invariant loads in nested loop vectorization.
268 2013-08-30  Richard Biener  <rguenther@suse.de>
270         PR tree-optimization/58223
271         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
272         (has_anti_or_output_dependence): ... this and adjust to also
273         look for output dependences.
274         (mark_nodes_having_upstream_mem_writes): Adjust.
275         (rdg_flag_uses): Likewise.
277 2013-08-30  Richard Biener  <rguenther@suse.de>
279         PR tree-optimization/58010
280         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
281         assert that we have a loop-closed PHI.
283 2013-08-29  Jan Hubicka  <jh@suse.cz>
285         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
286         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
287         * lto-section-in.c (lto_free_function_in_decl_state): New function.
288         (lto_free_function_in_decl_state_for_node): New function.
290 2013-08-29  Xinliang David Li  <davidxl@google.com>
292         * loop-unroll.c (report_unroll_peel): Minor message change.
293         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
294         Emit alignment peeling message with default -fopt-info.
295         (vect_loop_versioning): Emit loop version info message.
296         * tree-vectorizer.c (vectorize_loops): Minor message change.
297         (execute_vect_slp): Ditto.
299 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
301         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
302         of the clone from the DECL_NAME of the original function.
304 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
306         * passes.c (register_pass): Add overload.
307         * tree-pass.h (register_pass): Forward declare it.  Add comment.
309 2013-08-29  Jan Hubicka  <jh@suse.cz>
311         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
312         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
313         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
314         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
315         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
316         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
317         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
318         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
320 2013-08-29  Teresa Johnson  <tejohnson@google.com>
322         * dumpfile.c (dump_loc): Output column number.
323         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
324         * doc/invoke.texi: Document optall -fopt-info flag.
325         * profile.c (read_profile_edge_counts): Use new dump framework.
326         (compute_branch_probabilities): Ditto.
327         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
328         when pass not in any opt group.
329         * pass_manager.h (pass_manager::get_pass_profile): New method.
330         * value-prof.c (check_counter): Use new dump framework.
331         (check_ic_target): Ditto.
332         * coverage.c (get_coverage_counts): Ditto.
333         (coverage_init): Setup new dump framework.
335 2013-08-29  Richard Biener  <rguenther@suse.de>
337         PR tree-optimization/58246
338         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
339         handle the dominance check inside a basic-block.
341 2013-08-29  Richard Biener  <rguenther@suse.de>
343         PR middle-end/57287
344         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
345         of default defs that appear in abnormal PHI nodes.
347 2013-08-29  Richard Biener  <rguenther@suse.de>
349         PR tree-optimization/57685
350         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
351         single-use operands to avoid exponential complexity.
353 2013-08-28  Dehao Chen  <dehao@google.com>
355         * ipa-inline.c (edge_badness): Fix integer underflow.
357 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
359         * gtm-builtins.def (_ITM_free): Declare leaf.
361 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
363         PR target/58067
364         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
365         (*tls_local_dynamic_base_64_largepic): Likewise.
366         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
367         Remove predicate from call operand.
368         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
369         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
371 2013-08-28  Jeff Law  <law@redhat.com>
373         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
374         checks for the number of predecessors and successors allowed.
375         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
376         which require copying a joiner block if there is a request which
377         is a subpath that requires no joiner block copying.
379 2013-08-28  Jan Hubicka  <jh@suse.cz>
381         * lto-streamer-out.c (DFS_write_tree_body): Drop
382         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
383         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
384         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
385         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
386         Do not read DECL_ERROR_ISSUED.
387         (unpack_ts_decl_with_vis_value_fields): Do not read
388         DECL_DEFER_OUTPUT.
389         (lto_input_ts_binfo_tree_pointers): Do not read
390         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
391         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
392         write DECL_ERROR_ISSUED..
393         (pack_ts_decl_with_vis_value_fields): Do not write
394         DECL_DEFER_OUTPUT.
395         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
396         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
397         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
398         * tree.h (tree_decl_common): Update comment.
399         (DECL_ERROR_ISSUED): Remove.
401 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
403         PR middle-end/58257
404         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
406 2013-08-28  Jan Hubicka  <jh@suse.cz>
408         * builtins.def (free): Declare leaf.
410 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
412         * gdbhooks.py: New.
413         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
414         * configure: Regenerate.
416 2013-08-27  Martin Jambor  <mjambor@suse.cz>
418         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
419         (ipa_ancestor_jf_data): Likewise.
420         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
421         (ipa_get_jf_pass_through_type_preserved): New function.
422         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
423         (ipa_get_jf_ancestor_type_preserved): New function.
424         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
425         (ipa_get_jf_ancestor_result): Likewise.
426         (propagate_vals_accross_pass_through): Use
427         ipa_get_jf_pass_through_result to do all the value mappings.
428         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
429         type_preserved flag.
430         (ipa_set_jf_cst_copy): New function.
431         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
432         (ipa_set_jf_arith_pass_through): Likewise.
433         (ipa_set_ancestor_jf): Likewise.
434         (compute_complex_assign_jump_func): Set type_preserved instead of
435         punting.
436         (ipa_compute_jump_functions_for_edge): Likewise.
437         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
438         (update_jump_functions_after_inlining): Update type_preserved.
439         Explicitely create jump functions when combining one with pass_through.
440         (ipa_write_jump_function): Stream the type_preserved flags.
441         (ipa_read_jump_function): Likewise.
443 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
444             Aldy Hernandez  <aldyh@redhat.com>
446         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
447         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
448         * function.h (struct function): Add has_force_vect_loops and
449         has_simduid_loops.
450         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
451         * gimple.c (gimple_build_omp_critical): Add KIND argument and
452         handle it.
453         * gimple.def: Update CLAUSES comments.
454         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
455         (gimple_build_omp_for): Add argument to prototype.
456         (gimple_omp_for_kind): New.
457         (gimple_omp_for_set_kind): New.
458         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
459         GOVD_DATA_SHARE_CLASS.
460         (enum omp_region_type): Add ORT_SIMD.
461         (gimple_add_tmp_var): Handle ORT_SIMD.
462         (gimplify_var_or_parm_decl): Same.
463         (is_gimple_stmt): Same.
464         (omp_firstprivatize_variable): Same.
465         (omp_add_variable): Only use splay_tree_insert if lookup failed.
466         (omp_notice_variable): Handle ORT_SIMD.
467         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
468         (omp_check_private): Handle ORT_SIMD.
469         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
470         OMP_CLAUSE_SAFELEN.
471         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
472         Handle OMP_CLAUSE_LASTPRIVATE.
473         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
474         OMP_CLAUSE_SAFELEN.
475         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
476         (gimplify_expr): Handle OMP_SIMD.
477         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
478         (expand_GOMP_SIMD_VF): New.
479         (expand_GOMP_SIMD_LAST_LANE): New.
480         * internal-fn.def (GOMP_SIMD_LANE): New.
481         (GOMP_SIMD_VF): New.
482         (GOMP_SIMD_LAST_LANE): New.
483         * omp-low.c: Include target.h.
484         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
485         OMP_CLAUSE_SAFELEN.
486         (check_omp_nesting_restrictions): Same.
487         (omp_max_vf): New.
488         (lower_rec_simd_input_clauses): New.
489         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
490         OMP_CLAUSE_LINEAR.
491         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
492         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
493         (expand_omp_build_assign): New.
494         (expand_omp_for_init_counts): New.
495         (expand_omp_for_init_vars): New.
496         (extract_omp_for_update_vars): New.
497         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
498         and rewrite accordingly.
499         (expand_omp_simd): New.
500         (expand_omp_for): Use expand_omp_simd.
501         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
502         (lower_omp_for): Do not lower the body.
503         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
504         in their own loops.
505         * tree-flow.h (find_omp_clause): Remove prototype.
506         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
507         forcing vectorization of the loop, or if flag_tree_vectorize.
508         (gate_tree_if_conversion): Similarly.
509         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
510         gimple_build_omp_for.
511         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
512         * tree-parloops (create_parallel_loop): Pass kind argument to
513         gimple_build_omp_for.
514         * tree-pretty-print.c (dump_omp_clause): Add cases for
515         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
516         OMP_CLAUSE__SIMDUID_.
517         (dump_generic_node): Handle OMP_SIMD.
518         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
519         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
520         unroll OMP_SIMD loops here.
521         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
522         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
523         loop->safelen.
524         (vect_analyze_data_refs): Handle simd loops.
525         * tree-vect-loop.c (vectorizable_live_operation): Handle
526         IFN_GOMP_SIMD*.
527         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
528         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
529         (vectorizable_load): Same.
530         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
531         (struct simduid_to_vf): New.
532         (simduid_to_vf::hash): New.
533         (simduid_to-vf::equal): New.
534         (struct simd_array_to_simduid): New.
535         (simd_array_to_simduid::hash): New.
536         (simd_array_to_simduid::equal): New.
537         (adjust_simduid_builtins): New.
538         (struct note_simd_array_uses_struct): New.
539         (note_simd_array_uses_cb): New.
540         (note_simd_array_uses): New.
541         (vectorize_loops): Handle simd hints and adjust simd builtins
542         accordingly.
543         * tree-vectorizer.h (struct _stmt_vec_info): Add
544         simd_lane_access_p field.
545         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
546         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
547         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
548         (omp_clause_code_name): Same.
549         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
550         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
551         * tree.def (OMP_SIMD): New entry.
552         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
553         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
554         (OMP_CLAUSE_DECL): Adjust range for new clauses.
555         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
556         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
557         (OMP_CLAUSE_LINEAR_STEP): New.
558         (OMP_CLAUSE_SAFELEN_EXPR): New.
559         (OMP_CLAUSE__SIMDUID__DECL): New.
560         (find_omp_clause): New prototype.
562 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
564         * config/i386/driver-i386.c (host_detect_local_cpu): Update
565         Haswell processor detection.
567 2013-08-27  Christian Widmer  <shadow@umbrox.de>
569         PR target/57927
570         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
571         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
572         AVX2 capable processors.
574 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
576         * config/aarch64/arm_neon.h: Replace all inline asm implementations
577         of vget_low_* with implementations in terms of other intrinsics.
579 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
581         PR middle-end/57219
582         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
583         values to -1, 0 and 1.
585 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
587         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
588         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
589         (<optab><mode>3_insn): Remove.
590         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
591         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
592         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
593         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
594         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
595         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
596         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
597         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
598         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
599         (ror<mode>3_insn): Likewise.
600         * config/aarch64/predicates.md (aarch64_simd_register): New.
602 2013-08-27  Richard Biener  <rguenther@suse.de>
604         PR tree-optimization/57521
605         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
606         one edge is non-critical.
607         (find_phi_replacement_condition): Make sure to use a non-critical
608         edge.  Cleanup and remove old bug workarounds.
609         (bb_postdominates_preds): Remove.
610         (if_convertible_loop_p_1): Do not compute post-dominators.
611         (combine_blocks): Do not free post-dominators.
612         (main_tree_if_conversion): Likewise.
613         (pass_data_if_conversion): Add TODO_verify_ssa.
615 2013-08-27  DJ Delorie  <dj@redhat.com>
617         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
619 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
621         * function.c (assign_parm_find_data_types): Set passed_mode and
622         nominal_mode to the TYPE_MODE of nominal_type for the built
623         pointer type in case of the struct-pass-by-reference.
625 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
627         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
628         (UINT16_TYPE): Change default to "unsigned int".
630         * config/avr/avr.opt (mfract-convert-truncate): New option.
631         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
632         is set, round negative fractional integers according to n1169
633         when converting to integer types.
635 2013-08-26  Jan Hubicka  <jh@suse.cz>
637         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
638         methods can not be called indirectly when their address is not taken.
640 2013-08-26  Jan Hubicka  <jh@suse.cz>
642         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
643         ctor_for_folding.
645 2013-08-26  Jan Hubicka  <jh@suse.cz>
647         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
648         can be unshared.
650 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
652         * reload.c (find_valid_class): Allow classes that do not include
653         FIRST_PSEUDO_REGISTER - 1.
655 2013-08-26  Jan Hubicka  <jh@suse.cz>
657         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
658         fix edge count/frequency when speculation failed; fix type check
659         for the direct call.
661 2013-08-26  Jan Hubicka  <jh@suse.cz>
663         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
665 2013-08-26  Jan Hubicka  <jh@suse.cz>
667         * ipa-inline-transform.c (inline_transform): Be ready for basic block
668         to be changed by edge redirection.
670 2013-08-26  Jan Hubicka  <jh@suse.cz>
672         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
673         formating; add sanity check.
674         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
675         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
676         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
677         (resolve_noninline_speculation): Update callee keys, too.
679 2013-08-26  Jan Hubicka  <jh@suse.cz>
681         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
682         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
684 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
686         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
687         into proper place.
689 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
691         * config/i386/i386.c (ix86_debug_options): Remove prototype.
692         (x86_64_elf_select_section): Ditto.
693         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
694         arguments.
695         (ix86_pass_by_reference): Ditto.
696         (output_set_got): Ditto.
697         (ix86_unary_operator_ok): Ditto.
698         (ix86_expand_builtin): Ditto.
700 2013-08-23  Jan Hubicka  <jh@suse.cz>
702         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
704 2013-08-23  Jan Hubicka  <jh@suse.cz>
706         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
707         (tree_decl_with_vis): Add FINAL field.
709 2013-08-23  Jeff Law  <law@redhat.com>
711         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
712         the debugging dump when the expression is fully redundant.
714 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
716         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
717         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
718         * pretty-print.c (pp_formatted_text_data): Likewise.
719         (pp_write_text_to_stream): Likewise.
720         (pp_write_text_as_dot_label_to_stream): Likewise.
721         (pp_append_r): Likewise.
722         (pp_format): Likewise.
723         (pp_flush): Likewise.
724         (pp_clear_output_area): Likewise.
725         (pp_append_text): Likewise.
726         (pp_formatted_text): Likewise.
727         (pp_remaining_character_count_for_line): Likewise.
728         (pp_newline): Likewise.
729         (pp_character): Likewise.
730         (output_buffer::~output_buffer): Define.
731         (pretty_printer::~pretty_printer): Destruct output buffer.
732         * pretty-print.h (output_buffer::~output_buffer): Declare.
733         (pretty_printer::~pretty_printer): Declare virtual.
735 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
737         PR other/57324
738         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
739         HOST_WIDE_INT_M1U): New macros.
740         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
741         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
742         unsigned -1 for lshift.
743         * cse.c (cse_insn): Likewise.
744         * double-int.c (rshift_double, lshift_double): Likewise.
745         * builtins.c (fold_builtin_bitop): Likewise.
746         * combine.c (force_to_mode): Likewise.
747         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
748         * simplify-rtx.c (simplify_const_unary_operation,
749         simplify_const_binary_operation): Likewise.
750         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
751         check_va_list_escapes): Likewise.
752         * rtlanal.c (nonzero_bits1): Likewise.
753         * expmed.c (expand_smod_pow2): Likewise.
754         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
756 2013-08-23  Jan Hubicka  <jh@suse.cz>
758         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
759         as having address taken.
761 2013-08-23  Jan Hubicka  <jh@suse.cz>
763         * ipa-utils.h (method_class_type): Declare.
764         * ipa-devirt.c (method_class_type): Export.
766         * cgraphunit.c (analyze_functions): Do basic devirtualization;
767         do not walk base classes of anonymous types.
769 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
771         PR rtl-optimization/58220
772         PR regression/58221
773         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
774         handle SEQUENCE insns properly.
776 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
778         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
779         definition.
780         (pp_newline_and_indent): Likewise.
781         (pp_separate_with): Likewise.
782         * pretty-print.c (pp_newline_and_flush): Define.
783         (pp_newline_and_indent): Likewise.
784         (pp_separate_with): Likewise.
786 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
788         PR target/58218
789         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
790         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
792 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
794         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
795         * gcc/config/i386/i386.md (*movti_internal): Use
796         predicate to determine if EVEX is needed.
797         (*movsi_internal): Ditto.
798         (*movdf_internal): Ditto.
799         (*movsf_internal): Ditto.
800         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
802 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
804         PR tree-optimization/58209
805         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
806         (find_tail_calls): Give up for pointer result types if m is non-NULL.
807         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
808         emit POINTER_PLUS_EXPR.
809         (create_tailcall_accumulator): For pointer result type accumulate in
810         sizetype type.
812 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
814         * configure.ac: Add backslashes missing from the last change.
815         * configure: Regenerate.
817 2013-08-22  Jan Hubicka  <jh@susue.cz>
819         * ipa.c (function_and_variable_visibility): First remember function
820         was global and then make it local.
822 2013-08-22  Julian Brown  <julian@codesourcery.com>
824         * configure.ac: Add aarch64 to list of arches which use "nop" in
825         debug_line test.
826         * configure: Regenerate.
828 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
830         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
831         gnu_libc_has_function.
832         * config/s390/tpf.h: Likewise.
834 2013-08-22  Jan Hubicka  <jh@susue.cz>
836         * timevar.c (validate_phases): Add cast.
838 2013-08-22  Jan Hubicka  <jh@susue.cz>
840         * timevar.c (validate_phases): Use size_t for memory.
841         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
843 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
845         * pretty-print.h (output_buffer::output_buffer): Declare.
846         (pretty_printer::pretty_printer): Likewise.
847         (pp_construct): Remove.
848         * pretty-print.c (output_buffer::output_buffer): Define.
849         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
850         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
851         (print_gimple_expr): Likewise.
852         (print_gimple_seq): Likewise.
853         (gimple_dump_bb): Likewise.
854         * sched-vis.c (dump_value_slim): Likewise.
855         (dump_insn_slim): Likewise.
856         (dump_rtl_slim): Likewise.
857         (str_pattern_slim): Likewise.
858         * tree-mudflap.c (mf_varname_tree): Likewise.
859         * graph.c (print_graph_cfg): Likewise.
860         (start_graph_dump): Likewise.
861         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
862         placement-new.
863         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
864         pretty printer initialization.
865         * coretypes.h (diagnostic_context): Remove superflous type alias
866         declaration.
867         (pretty_printer): Likewise.  Declare directly as a class.
868         (pretty_print_info): Remove declaration as class.
869         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
870         and pp_clear_output_area.
871         (asan_add_global): Likewise.
873 2013-08-22  Jan Hubicka  <jh@suse.cz>
875         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
876         * ipa-utils.h (update_type_inheritance_graph): Declare.
877         (possible_polymorphic_call_target_p): Declare.
878         (possible_polymorphic_call_target_p): New.
879         * ipa-devirt.c: Update toplevel comments.
880         (cached_polymorphic_call_targets): Move up.
881         (odr_type_d): Move ID down.
882         (polymorphic_type_binfo_p): Update comment.
883         (odr_hasher::remove): Likewise;
884         (get_odr_type): Set anonymous_namespace.
885         (dump_odr_type): Dump it.
886         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
887         (maybe_record_node): Record node in cached_polymorphic_call_targets.
888         (record_binfo): Add comment.
889         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
890         built.
891         (devirt_node_removal_hook): Do not iCE when cache is freed.
892         (possible_polymorphic_call_target_p): New predicate.
893         (update_type_inheritance_graph): New function.
895 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
896             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
897             Sergey Lega  <sergey.s.lega@intel.com>
898             Anna Tikhonova  <anna.tikhonova@intel.com>
899             Ilya Tocar  <ilya.tocar@intel.com>
900             Andrey Turetskiy  <andrey.turetskiy@intel.com>
901             Ilya Verbin  <ilya.verbin@intel.com>
902             Kirill Yukhin  <kirill.yukhin@intel.com>
903             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
905         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
906         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
907         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
908         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
909         (OPTION_MASK_ISA_AVX2_UNSET): Update.
910         (OPTION_MASK_ISA_AVX512F_UNSET): New.
911         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
912         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
913         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
914         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
915         OPT_mavx512pf, OPT_mavx512er cases.
916         * config/i386/constraints.md (v): New constraint.
917         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
918         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
919         (bit_AVX512CD): New.
920         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
921         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
922         * config/i386/i386-c.c (ix86_target_macros_internal):
923         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
924         __AVX512PF__.
925         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
926         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
927         * config/i386/i386.c (regclass_map, dbx_register_map)
928         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
929         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
930         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
931         -mavx512pf options.
932         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
933         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
934         -mavx512pf options.  Fix formatting.
935         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
936         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
937         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
938         -mavx512cd, -mavx512pf options.
939         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
940         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
941         (ix86_preferred_output_reload_class): Replace SSE_REGS with
942         ALL_SSE_REGS.
943         (ix86_hard_regno_mode_ok): Support 512-bit registers.
944         (ix86_set_reg_reg_cost): Ditto.
945         (x86_order_regs_for_local_alloc): Ditto.
946         (MAX_VECT_LEN): Extend to 64-byte.
947         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
948         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
949         (TARGET_AVX512ER, TARGET_AVX512CD): New.
950         (BIGGEST_ALIGNMENT): Extend to 512-bits.
951         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
952         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
953         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
954         (SSE_REG_MODE_P): Support new modes.
955         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
956         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
957         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
958         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
959         (REG_CLASS_CONTENTS): Add new registers.
960         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
961         (EXT_REX_SSE_REGNO_P): New.
962         (HI_REGISTER_NAMES): Add new registers.
963         * config/i386/i386.md: Define constants for new registers.
964         (mode): Add new 512-bit modes.
965         (prefix): Support evex prefix.
966         (isa): Support avx512f, noavx512f, fma_avx512f.
967         (ssemodesuffix): Add new 512-bit modes.
968         (movxi): New.
969         (*movxi_internal_avx512f): Ditto.
970         (*movdi_internal): Replace constraint "x" with the new constraint "v".
971         Support MODE_XI.
972         (*movsi_internal): Likewise.
973         (*movdf_internal): Likewise.
974         (*movsf_internal): Likewise.
975         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
976         (<code><mode>3): Likewise.
977         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
978         New.
979         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
980         with the new constraint "v".
981         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
982         modes.
983         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
984         with the new constraint "v".
985         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
986         (<sse2>_storedqu<avxsizesuffix>): Likewise.
987         (*<plusminus_insn><mode>3): Likewise.
988         (<sse>_vm<plusminus_insn><mode>3): Likewise.
989         (*mul<mode>3): Likewise.
990         (<sse>_vmmul<mode>3): Likewise.
991         (<sse>_div<mode>3): Likewise.
992         (<sse>_vmdiv<mode>3): Likewise.
993         (<sse>_sqrt<mode>2): Likewise.
994         (<sse>_vmsqrt<mode>2): Likewise.
995         (*<code><mode>3_finite): Likewise.
996         (*<code><mode>3) <smaxmin>: Likewise.
997         (<sse>_vm<code><mode>3): Likewise.
998         (*<code><mode>3) <any_logic>: Likewise.
999         (*fma_fmadd_<mode>): Likewise.
1000         (*fma_fmsub_<mode>): Likewise.
1001         (*fma_fnmadd_<mode>): Likewise.
1002         (*fma_fnmsub_<mode>): Likewise.
1003         (*fma_fmaddsub_<mode>): Likewise.
1004         (*fma_fmsubadd_<mode>): Likewise.
1005         (*fmai_fmadd_<mode>): Likewise.
1006         (*fmai_fmsub_<mode>): Likewise.
1007         (*fmai_fnmadd_<mode>): Likewise.
1008         (*fmai_fnmsub_<mode>): Likewise.
1009         (sse_cvtsi2ss): Likewise.
1010         (sse_cvtsi2ssq): Likewise.
1011         (sse_cvtss2si): Likewise.
1012         (sse_cvtss2si_2): Likewise.
1013         (sse_cvtss2siq): Likewise.
1014         (sse_cvtss2siq_2): Likewise.
1015         (sse_cvttss2si): Likewise.
1016         (sse_cvtss2siq_2): Likewise.
1017         (float<sseintvecmodelower><mode>2): Likewise.
1018         (sse2_cvtsd2si_2): Likewise.
1019         (sse2_cvtsd2siq_2): Likewise.
1020         (*<plusminus_insn><mode>3): Likewise.
1021         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1022         (*<sse4_1_avx2>_mul<mode>3): Likewise.
1023         (ashr<mode>3): Likewise.
1024         (<shift_insn><mode>3): Likewise.
1025         (avx2_<code><mode>3): Likewise.
1026         (*avx2_<code><mode>3): Likewise.
1027         (*andnot<mode>3): Likewise.
1028         (*<code><mode>3) <any_logic>: Likewise.
1029         (abs<mode>2): Likewise.
1030         (avx2_permvar<mode>): Likewise.
1031         (avx2_perm<mode>_1): Likewise.
1032         (*avx_vpermilp<mode>): Likewise.
1033         (avx_vpermilvar<mode>3): Likewise.
1034         (avx2_ashrv<mode>): Likewise.
1035         (avx2_<shift_insn>v<mode>): Likewise.
1036         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1037         -mavx512cd.
1038         * doc/rtl.texi: Document XImode.
1040 2013-08-21  Jeff Law  <law@redhat.com>
1042         * tree-flow.h (register_jump_thread): Pass vector of edges
1043         instead of each important edge.
1044         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1045         thread path into a vector and pass that to register_jump_thread.
1046         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1047         passed in edge vector to the current 3-edge form.
1049         Revert:
1050         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1052         * dce.c (fini_dce): Call df_analyze again just in case
1053         delete_unmarked_insns removed anything.
1055 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
1057         * reload.h (struct reg_equivs): Rename to ..
1058         (struct reg_equivs_s): .. this.
1060 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
1062         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1064 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1066         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1068 2013-08-21  Jeff Law  <law@redhat.com>
1070         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1071         simplify assignments too.  If the RHS collapses to a singleton
1072         range, then return the value for the range.
1074 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
1076         * config/i386/sse.md (V16): Rename to...
1077         (VMOVE): this.
1078         (mov<mode>): Update iterator name.
1079         (*mov<mode>_internal): Ditto.
1080         (push<mode>1): Ditto.
1081         (movmisalign<mode>): Ditto.
1083 2013-08-20  Jan Hubicka  <jh@suse.cz>
1085         PR bootstrap/58186
1086         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1087         entry for direct edges.
1088         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1090 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1092         Revert my last two changes, r201865 and r201864:
1094         Revert r201865:
1095         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1097         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1098         instances can own GC refs.
1100         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1101         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1102         (gcc::context::gt_pch_nx): Likewise.
1103         (gcc::context::gt_pch_nx):  Likewise.
1104         * ggc.h (gt_ggc_mx <T>): New.
1105         (gt_pch_nx_with_op <T>): New.
1106         (gt_pch_nx <T>): New.
1107         * passes.c (opt_pass::gt_ggc_mx): New.
1108         (opt_pass::gt_pch_nx): New.
1109         (opt_pass::gt_pch_nx_with_op): New.
1110         (pass_manager::gt_ggc_mx): New.
1111         (pass_manager::gt_pch_nx): New.
1112         (pass_manager::gt_pch_nx_with_op): New.
1113         (pass_manager::operator new): Use
1114         ggc_internal_cleared_alloc_stat rather than xcalloc.
1115         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1116         (pass_manager::gt_ggc_mx): New.
1117         (pass_manager::gt_pch_nx): New.
1118         (pass_manager::gt_pch_nx_with_op): New.
1119         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1120         (opt_pass::operator new): New.
1121         (opt_pass::gt_ggc_mx): New.
1122         (opt_pass::gt_pch_nx): New.
1123         (opt_pass::gt_pch_nx_with_op): New.
1125         Revert r201864:
1126         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1128         * Makefile.in (GTFILES): Add context.h.
1129         * context.c (gcc::context::operator new): New.
1130         (gcc::context::gt_ggc_mx): New.
1131         (gcc::context::gt_pch_nx): New.
1132         (gcc::context::gt_pch_nx): New.
1133         * context.h (gcc::context): Add GTY((user)) marking.
1134         (gcc::context::operator new): New.
1135         (gcc::context::gt_ggc_mx): New.
1136         (gcc::context::gt_pch_nx): New.
1137         (gcc::context::gt_pch_nx): New.
1138         (g): Add GTY marking.
1139         (gt_ggc_mx (gcc::context *)): New.
1140         (gt_pch_nx (gcc::context *)): New.
1141         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1142         void *cookie)): New.
1143         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1145 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1147         * dce.c (fini_dce): Call df_analyze again just in case
1148         delete_unmarked_insns removed anything.
1150 2013-08-20  Teresa Johnson  <tejohnson@google.com>
1152         PR rtl-optimizations/57451
1153         * final.c (reemit_insn_block_notes): Prevent lexical blocks
1154         from crossing split section boundaries.
1156 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1158         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1159         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1160         with MULTLIB_DEFAULTS.
1162 2013-08-20  Nick Clifton  <nickc@redhat.com>
1164         * target.def (narrow_volatile_bitfield): Note that the default
1165         value is false, not !TARGET_STRICT_ALIGN.
1166         * doc/tm.texi: Regenerate.
1168 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
1170         Fix LIB_SPEC for systems without libpthread.
1172         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1173         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1174         for Android.
1175         * config/i386/linux-common.h: Likewise.
1176         * config/mips/linux-common.h: Likewise.
1178 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
1180         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1181         checks.
1183 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1185         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1186         instances can own GC refs.
1188         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1189         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1190         (gcc::context::gt_pch_nx): Likewise.
1191         (gcc::context::gt_pch_nx):  Likewise.
1192         * ggc.h (gt_ggc_mx <T>): New.
1193         (gt_pch_nx_with_op <T>): New.
1194         (gt_pch_nx <T>): New.
1195         * passes.c (opt_pass::gt_ggc_mx): New.
1196         (opt_pass::gt_pch_nx): New.
1197         (opt_pass::gt_pch_nx_with_op): New.
1198         (pass_manager::gt_ggc_mx): New.
1199         (pass_manager::gt_pch_nx): New.
1200         (pass_manager::gt_pch_nx_with_op): New.
1201         (pass_manager::operator new): Use
1202         ggc_internal_cleared_alloc_stat rather than xcalloc.
1203         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1204         (pass_manager::gt_ggc_mx): New.
1205         (pass_manager::gt_pch_nx): New.
1206         (pass_manager::gt_pch_nx_with_op): New.
1207         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1208         (opt_pass::operator new): New.
1209         (opt_pass::gt_ggc_mx): New.
1210         (opt_pass::gt_pch_nx): New.
1211         (opt_pass::gt_pch_nx_with_op): New.
1213 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1215         * Makefile.in (GTFILES): Add context.h.
1216         * context.c (gcc::context::operator new): New.
1217         (gcc::context::gt_ggc_mx): New.
1218         (gcc::context::gt_pch_nx): New.
1219         (gcc::context::gt_pch_nx): New.
1220         * context.h (gcc::context): Add GTY((user)) marking.
1221         (gcc::context::operator new): New.
1222         (gcc::context::gt_ggc_mx): New.
1223         (gcc::context::gt_pch_nx): New.
1224         (gcc::context::gt_pch_nx): New.
1225         (g): Add GTY marking.
1226         (gt_ggc_mx (gcc::context *)): New.
1227         (gt_pch_nx (gcc::context *)): New.
1228         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1229         void *cookie)): New.
1230         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1232 2013-08-20  Alan Modra  <amodra@gmail.com>
1234         PR target/57865
1235         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1236         (rs6000_emit_epilogue): Likewise.
1238 2013-08-19  Dehao Chen  <dehao@google.com>
1240         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1242 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
1243             Jakub Jelinek  <jakub@redhat.com>
1245         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1246         (BUILT_IN_FABSD64): Likewise.
1247         (BUILT_IN_FABSD128): Likewise.
1248         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1249         (fold_builtin_1): Likewise.
1250         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1251         destination and source operands.
1252         (*abstd2_fpr): Likewise.
1253         (*nabstd2_fpr): Likewise.
1255 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
1257         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1258         JUMP_P and INSN_P.
1260 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
1262         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1263         incomplete.
1265 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1267         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1268         * builtins.c (default_libc_has_function): New.
1269         (gnu_libc_has_function): Ditto.
1270         (no_c99_libc_has_function): Ditto.
1271         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1272         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1273         (fold_builtin_sincos): Likewise.
1274         (fold_builtin_cexp): Likewise.
1275         * builtins.def (DEF_C94_BUILTIN): Likewise.
1276         (DEF_C99_BUILTIN): Likewise.
1277         (DEF_C99_C90RES_BUILTIN): Likewise.
1278         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1279         definitions to using this define.
1280         * config/darwin-protos.h (darwin_libc_has_function): New.
1281         * config/darwin.c (darwin_libc_has_function): Ditto.
1282         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1283         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1284         * config/darwin.h: Ditto.
1285         * config/elfos.h: Ditto.
1286         * config/freebsd.h: Ditto.
1287         * config/i386/cygming.h: Ditto.
1288         * config/i386/djgpp.h: Ditto.
1289         * config/i386/i386-interix.h: Ditto.
1290         * config/microblaze/microblaze.h: Ditto.
1291         * config/mmix/mmix.h: Ditto.
1292         * config/gnu-user.h: Ditto.
1293         * config/ia64/hpux.h: Ditto.
1294         * config/pa/pa-hpux.h: Ditto.
1295         * config/pdp11/pdp11.h: Ditto.
1296         * config/picochip/picochip.h: Ditto.
1297         * config/linux.h: Ditto.
1298         * config/netbsd.h: Ditto.
1299         * config/openbsd.h: Ditto.
1300         * config/rs6000/aix43.h: Ditto.
1301         * config/rs6000/aix51.h: Ditto.
1302         * config/rs6000/aix52.h: Ditto.
1303         * config/rs6000/aix53.h: Ditto.
1304         * config/rs6000/aix61.h: Ditto.
1305         * config/rs6000/darwin.h: Ditto.
1306         * config/rs6000/linux.h: Ditto.
1307         * config/rs6000/linux64.h: Ditto.
1308         * config/s390/tpf.h: Ditto.
1309         * config/sol2-10.h: Ditto.
1310         * config/sol2.h: Ditto.
1311         * config/vms/vms.h: Ditto.
1312         * config/vxworks.h: Ditto.
1313         * config/linux-android.c (linux_android_libc_has_function):
1314         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1315         * config/linux-protos.h (linux_android_libc_has_function):
1316         New declaration.
1317         * config/i386/i386.c (ix86_libc_has_function): New.
1318         * config/i386/i386-protos.h
1319         (ix86_libc_has_function): New declaration.
1320         * config/i386/i386.md
1321         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1322         ("isinf<mode>2): Likewise.
1323         * convert.c (convert_to_integer): Using new target hook
1324         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1325         TARGET_C99_FUNCTIONS.
1326         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1327         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1328         * coretypes.h (function_class): New enum for different
1329         classes of functions.
1330         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1331         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1332         (TARGET_HAS_SINCOS): Likewise.
1333         (TARGET_LIBC_HAS_FUNCTION): New.
1334         * doc/tm.texi: Regenerated.
1335         * targhooks.h (default_libc_has_function): New declaration.
1336         (no_c99_libc_has_function): Ditto.
1337         (gnu_libc_has_function): Ditto.
1338         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1339         and TARGET_HAS_SINCOS.
1341 2013-08-18  Jan Hubicka  <jh@suse.cz>
1343         * Makeifle-in (ipa-devirt.o): New.
1344         (GTFILES): Add ipa-utils.h and ipa-devirt.c
1345         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1346         (analyze_functions): Look into possible targets of polymorphic call.
1347         * dumpfile.c (dump_files): Add type-inheritance dump.
1348         * dumpfile.h (TDI_inheritance): New.
1349         * ipa-devirt.c: New file.
1350         * ipa-utils.h (odr_type_d): Forward declare.
1351         (odr_type): New type.
1352         (build_type_inheritance_graph): Declare.
1353         (possible_polymorphic_call_targets): Declare and introduce inline
1354         variant when only edge is pased.
1355         (dump_possible_polymorphic_call_targets): Likewise.
1356         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1357         * tree.c (type_in_anonymous_namespace_p): Break out from ...
1358         (types_same_for_odr): ... here.
1359         * tree.h (type_in_anonymous_namespace_p): Declare.
1361 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
1363         PR tree-optimization/58006
1364         * tree-parloops.c (take_address_of): Don't ICE if get_name
1365         returns NULL.
1366         (eliminate_local_variables_stmt): Remove clobber stmts.
1368 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
1370         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1371         error message is issued for an alias to undefined symbol.
1373 2013-08-18  Jan Hubicka  <jh@suse.cz>
1375         * cgraph.c (cgraph_create_indirect_edge): Discover
1376         polymorphic calls and record basic info into indirect_info.
1377         * gimple-fold.c (gimple_fold_call): When doing BINFO based
1378         devirtualization, ignore objc function calls.
1379         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1380         call with no parm index info.
1381         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1382         * tree.c (virtual_method_call_p): New function.
1383         * tree.h (virtual_method_call_p): Declare.
1385 2013-08-16  Jan Hubicka  <jh@suse.cz>
1387         PR middle-end/58179
1388         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1390 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
1392         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1393         attribute.
1395 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1397         * gengtype.c (type_for_name): Add special-case support for
1398         locating types within the "gcc::" namespace.
1399         (open_base_files): Emit a "using namespace gcc" directive.
1401 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1403         PR target/58160
1404         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1405         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1407         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1408         array instead of each individual operand as a separate argument.
1409         (emit_fusion_gpr_load): Likewise.
1410         (expand_fusion_gpr_load): Add new function declaration.
1412         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1413         signature to have the operands passed as an array, instead of as
1414         separate arguments.  Allow ZERO_EXTEND to be in the memory
1415         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
1416         depend on the register live/dead flags when peepholes are run.
1417         (expand_fusion_gpr_load): New function to be called from the
1418         peephole2 pass, to change the register that addis sets to be the
1419         target register.
1420         (emit_fusion_gpr_load): Change the calling signature to have the
1421         operands passed as an array, instead of as separate arguments.
1422         Allow ZERO_EXTEND to be in the memory address, and also
1423         SIGN_EXTEND if -mpower8-fusion-sign.
1425         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1426         unspec enumeration.
1427         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1428         adjust the register addis loads up in the peephole2 pass.  Do not
1429         depend on the register live/dead state when the peephole pass is done.
1431 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1433         * gengtype.c (create_user_defined_type): Ensure that the kind
1434         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1435         declaration is seen before the GTY((user)) marking.
1437 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1439         PR target/58105
1440         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1442 2013-08-16  Jan Hubicka  <jh@suse.cz>
1444         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1445         arugment expected_type.
1446         (gimple_fold_call): Use it.
1447         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1448         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1449         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1450         (try_make_edge_direct_virtual_call): Likewise.
1451         * tree.c (obj_type_ref_class): New.
1452         * tree.h (obj_type_ref_class): Use it.
1454 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1456         * sched-vis.c (rtl_slim_pp_initialized): Remove.
1457         (rtl_slim_pp): Likewise.
1458         (init_rtl_slim_pretty_print): Likewise.
1459         (dump_value_slim):  Don't call it.  Use local pretty printer.
1460         (dump_insn_slim): Likewise.
1461         (dump_rtl_slim): Likewise.
1462         (str_pattern_slim): Likewise.
1463         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1464         Simplify.
1466 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
1468         PR tree-optimization/58164
1469         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1470         walk gimple_goto_dest of GIMPLE_GOTO.
1472         PR tree-optimization/58165
1473         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1474         bi_call must be the last stmt in a bb, don't split_block, instead
1475         use fallthru edge from it and give up if there is none.
1476         Release conds vector when returning early.
1478 2013-08-14  Xinliang David Li  <davidxl@google.com>
1480         * config/i386/i386.c (ix86_option_override_internal):
1481         Remove unused variable and field.
1483 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1485         PR target/57949
1486         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1487         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1488         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1489         and Linux, correct BLKmode alignment when 128-bit alignment is
1490         required and compatibility flag is not set.
1491         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1492         for zero-size arguments when compatibility flag is not set.
1494 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1496         PR tree-optimization/58145
1497         * tree-sra.c (build_ref_for_offset): If prev_base has
1498         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1500 2013-08-14  Xinliang David Li  <davidxl@google.com>
1502         * config/i386/i386.c (ix86_option_override_internal):
1503         Fix uninitialized variable error.
1505 2013-08-14  Xinliang David Li  <davidxl@google.com>
1507         * config/i386/i386.opt: Define two new options.
1508         * config/i386/x86-tune.def: Add arch selector field in macros.
1509         * config/i386/i386.h: Adjust macro definition.
1510         * config/i386/i386.c (ix86_option_override_internal):
1511         Refactor the code.
1512         (parse_mtune_ctrl_str): New function.
1513         (set_ix86_tune_features): New function.
1514         (ix86_function_specific_restore): Call the new helper function.
1516 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
1518         PR rtl-optimization/57662
1519         * sel-sched.c (code_motion_process_successors): When the current insn
1520         is removed after the recursive traversal, break from the loop.
1521         Add comments and debug printouts.
1523 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1524             Alexandre Oliva  <aoliva@redhat.com>
1526         PR target/58067
1527         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1528         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1529         there also UNSPEC_PLTOFF.
1531 2013-08-14  Marek Polacek  <polacek@redhat.com>
1533         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1534         NUM_CONDITIONS bit positions.
1536 2013-08-13  Cary Coutant  <ccoutant@google.com>
1538         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1539         (attr_checksum): Hash vector contents instead of pointer.
1540         (attr_checksum_ordered): Likewise.
1542 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
1544         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1545         when Pmode != word_mode.  Add length_address attribute.
1546         (sse3_monitor_<mode>): Merge from sse3_monitor and
1547         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
1548         Pmode != word_mode.  Update insn length attribute.
1549         * config/i386/i386.c (ix86_option_override_internal): Update
1550         ix86_gen_monitor selection for merged sse3_monitor insn.
1552 2013-08-13  Julian Brown  <julian@codesourcery.com>
1554         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1555         perform invalid legitimization on greater-than-word-size modes for
1556         TARGET_E500_DOUBLE.
1558 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
1560         * ira.c (setup_class_translate_array): Use aclass instead of cl
1561         for classes not fully covered by allocno classes.
1563 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
1565         PR tree-optimization/57661
1566         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1567         * tree-inline.c (tree_function_versioning): Initialize it.
1568         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1569         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1570         that is not being copied.
1572         PR sanitizer/56417
1573         * asan.c (instrument_strlen_call): Fix typo in comment.
1574         Use char * type even for the lhs of POINTER_PLUS_EXPR.
1576 2013-08-13  Steve Ellcey  <sellcey@mips.com>
1578         * config/mips/mips.md (prefetch): Use lw instead of ld on
1579         loongson in 32bit mode.
1581 2013-08-13  Nick Clifton  <nickc@redhat.com>
1583         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1585 2013-08-13  Jan Hubicka  <jh@suse.cz>
1587         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1588         introduced edge; fix typo in sanity check.
1589         (cgraph_resolve_speculation): Export; improve diagnostic.
1590         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1591         speculation at type mismatch.
1592         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1593         (cgraph_resolve_speculation): Declare.
1594         (symtab_can_be_discarded): New function.
1595         * value-prof.c (gimple_ic_transform): Remove actual transform code.
1596         * ipa-inline-transform.c (speculation_removed): New global var.
1597         (clone_inlined_nodes): See if speculation can be removed.
1598         (inline_call): If speculations was removed, we growths may not match.
1599         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1600         (speculation_useful_p): New function.
1601         (resolve_noninline_speculation): New function.
1602         (inline_small_functions): Resolve useless speculations.
1603         * ipa-inline.h (speculation_useful_p): Declare
1604         * ipa.c (can_replace_by_local_alias): Simplify.
1605         (ipa_profile): Produce speculative calls in non-lto, too;
1606         add simple cost model; produce local aliases.
1608 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
1610         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1611         PASS_MANAGER_H.
1613 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
1615         * config/i386/i386.c (ix86_function_versions): Use error + inform.
1617 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
1619         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1620         iterator instead of X87MODEF.
1622 2013-08-12  Perez Read  <netfirewall@gmail.com>
1624         PR target/58132
1625         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
1626         operand 0 for intel asm alternative.
1627         (*movabs<mode>_2): Ditto for operand 1.
1629 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
1631         * config/aarch64/arm_none.h
1632         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1634 2013-08-12  Nick Clifton  <nickc@redhat.com>
1636         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1638 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1640         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1641         expand for QI/HImode operand to produce more effictive code for
1642         unsigned char(short) --> float(double) conversion.
1644 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1646         * doc/invoke.texi: Mention that -ftls-model does not force the final
1647         model.
1649 2013-08-12  Marek Polacek  <polacek@redhat.com>
1650             Marc Glisse  <marc.glisse@inria.fr>
1652         PR tree-optimization/57980
1653         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1654         when creating -1 constant.
1656 2013-08-10  Jan Hubicka  <jh@suse.cz>
1658         Workaround binutils PR14342.
1659         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1660         (gimple_init_edge_profiler): Likewise.
1661         (gimple_gen_ic_func_profiler): Likewise.
1663 2013-08-09  Jan Hubicka  <jh@suse.cz>
1665         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1667 2013-08-09  Xinliang David Li  <davidxl@google.com>
1669         * config/i386/stringop.def: New file.
1670         * config/i386/stringop.opt: New file.
1671         * config/i386/i386-opts.h: Include stringopt.def.
1672         * config/i386/i386.opt: Include stringopt.opt.
1673         * config/i386/i386.c (ix86_option_override_internal):
1674         Override default size based stringop inline strategies with options.
1675         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1676         New function.
1678 2013-08-09  Jan Hubicka  <jh@suse.cz>
1680         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1682 2013-08-09  Jan Hubicka  <jh@suse.cz>
1684         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1685         CGRAPH_FREQ_MAX.
1686         (dump_cgraph_node): Dump profile-id.
1687         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1688         and common_target_probability.
1689         * lto-cgraph.c (lto_output_edge): Stream common targets.
1690         (lto_output_node): Stream profile ids.
1691         (input_node): Stream profile ids.
1692         (input_edge): Stream common targets.
1693         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1694         * ipa.c: Include value-prof.h
1695         (ipa_profile_generate_summary): Turn indirect call statement histograms
1696         into common targets.
1697         (ipa_profile): Turn common targets into speculative edges.
1699 2013-08-09  Jan Hubicka  <jh@suse.cz>
1701         * cgraph.h (cgraph_node): Add profile_id.
1702         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1703         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1704         (del_node_map): Update.
1705         (find_func_by_funcdef_no): Replace by ...
1706         (find_func_by_profile_id): ... this one.
1707         (gimple_ic_transform): Do not remove useful histograms when
1708         speculation is not done; dump info when indirect call removal
1709         can happen at LTO.
1710         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1711         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1712         (__gcov_indirect_call_profiler_v2): .. this one.
1713         * profile.h (init_node_map): Update.
1714         * coverage.c (coverage_compute_profile_id): New function.
1715         * coverage.h (coverage_compute_profile_id): Declare.
1716         * tree-profile.c (init_ic_make_global_vars): Make
1717         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1718         (gimple_init_edge_profiler): Update prototype of
1719         __gcov_indirect_call_profiler.
1720         (gimple_gen_ic_func_profiler): Simplify.
1721         (tree_profiling): Use init_node_map
1723 2013-08-09  Jan Hubicka  <jh@suse.cz>
1725         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1726         non-speculative refs.
1727         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1728         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1729         (cgraph_set_call_stmt): Likewise.
1730         (cgraph_create_edge_1): Fix release checking compilatoin;
1731         clear lto_stmt_uid.
1732         (cgraph_free_edge): Free indirect info.
1733         (cgraph_turn_edge_to_speculative): New function.
1734         (cgraph_speculative_call_info): New function.
1735         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1736         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1737         (dump_cgraph_node): Dump speculation.
1738         (verify_edge_count_and_frequency): Accept speculative edges.
1739         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1740         (verify_cgraph_node): Handle speculation.
1741         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1742         (cgraph_set_call_stmt): Update prototype.
1743         (cgraph_make_edge_direct): Update prototype.
1744         (cgraph_speculative_call_info): Declare.
1745         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1746         to change; update call of ipa_find_references.
1747         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1748         lto_stmt_uid and speculative flags.
1749         (ipa_dump_references): Dump speculation.
1750         (ipa_clone_references): Clone speculative flag.
1751         (ipa_clone_referring): Likewise.
1752         (ipa_clone_ref): New function.
1753         (ipa_find_reference): Look into lto_stmt_uids
1754         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1755         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1756         (ipa_find_reference): Update declaration.
1757         (ipa_clone_ref): Declare.
1758         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1759         stream speculative flag.
1760         (lto_output_ref): Stream statements uids and speculation.
1761         (input_ref): Likewise.
1762         (input_edge): Stream speuclation.
1763         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1764         (cgraph_set_call_stmt_including_clones): Handle speculation.
1765         * ipa-inline.c (heap_edge_removal_hook): New function.
1766         (inline_small_functions): Register it.
1767         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1768         also initialize refs.
1769         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1770         edge to change.
1771         (try_make_edge_direct_simple_call): Likewise.
1772         (try_make_edge_direct_simple_call): Likewise.
1773         (update_indirect_edges_after_inlining): Likewise.
1774         (remove_described_reference): Look proper lto_stmt_uid.
1775         (propagate_controlled_uses): Likewise.
1776         (propagate_controlled_uses): Liekwise.
1777         * tree-inline.c (copy_bb): Copy speculative edges.
1778         (redirect_all_calls): New function.
1779         (copy_cfg_body): Do redirection after loop info is updated.
1780         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1782 2013-08-09  Jan Hubicka  <jh@suse.cz>
1784         * lto-streamer-out.c (output_function): Renumber PHIs.
1785         * lto-streamer-in.c (input_function): Likewise.
1787 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1789         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1790         (get_lane_unsigned): Likewise.
1791         (dup_lane_scalar): Likewise.
1792         (get_lane): enable for VALL.
1793         * config/aarch64/aarch64-simd.md
1794         (aarch64_dup_lane_scalar<mode>): Remove.
1795         (aarch64_get_lane_signed<mode>): Likewise.
1796         (aarch64_get_lane_unsigned<mode>): Likewise.
1797         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1798         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1799         (aarch64_get_lane<mode>): Enable for all vector modes.
1800         (aarch64_get_lanedi): Remove misleading constraints.
1801         * config/aarch64/arm_neon.h
1802         (__aarch64_vget_lane_any): Define.
1803         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1804         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1805         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1806         * config/aarch64/iterators.md (VDQQH): New.
1807         (VDQQHS): Likewise.
1808         (vwcore): Likewise.
1810 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1812         * configure.ac: Add GAS check for LEON instructions on SPARC.
1813         * configure: Regenerate.
1814         * config.in: Likewise.
1815         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1816         sparc*-*-* block.
1817         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1818         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1819         for LEON or LEON3.
1820         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1821         (AS_LEON_FLAG): New macro.
1822         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1823         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1824         Deal with LEON and LEON3 for the memory model.
1825         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1826         (atomic_compare_and_swap<mode>_1): Likewise.
1827         (*atomic_compare_and_swap<mode>_1): Likewise.
1829 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1831         * config/arm/neon.md (vcond): Fix floating-point vector
1832         comparisons against 0.
1834 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1836         * lra-constraints.c (emit_spill_move): Remove assert.
1837         (process_alt_operands): Add more debugging
1838         output.  Increase reject for spilling into memory.  Decrease
1839         reject for reloading scratch.
1840         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1842 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1844         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1845         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1846         micromips incompatible.  Add nan2008.
1847         (MULTILIB_DIRNAMES): Add nan2008.
1848         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1849         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1850         and micromips incompatible.  Add nan2008.
1851         (MULTILIB_DIRNAMES): Add nan2008.
1852         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1854 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1856         PR rtl-optimization/58079
1857         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1858         simplify_comparison has widened a comparison with an integer.
1860 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1862         * config/arm/neon.md (movmisalign<mode>): Disable when we
1863         don't allow unaligned accesses.
1864         (*movmisalign<mode>_neon_store): Likewise.
1865         (*movmisalign<mode>_neon_load): Likewise.
1866         (*movmisalign<mode>_neon_store): Likewise.
1867         (*movmisalign<mode>_neon_load): Likewise.
1869 2013-08-08  Jan Hubicka  <jh@suse.cz>
1871         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1872         (make_pass_rebuild_cgraph_edges): Also clear references.
1873         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1874         * ipa-inline-transform.c (inline_transform): Remove all references
1875         after inlining.
1876         * cgraphunit.c (expand_function): Remove all references after
1877         expansion.
1878         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1879         (ipa_find_reference): Rewrite to iterator.
1880         (remove_stmt_references): Likewise.
1881         (ipa_clear_stmts_in_references): New function.
1882         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1883         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1884         clear references.
1885         * ipa-split.c (split_function): Remove references in split function.
1887 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1889         PR target/57431
1890         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1891         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1893 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1895         PR target/56979
1896         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1897         suggested mode for the assignment isn't compatible with the
1898         registers required.
1900 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1902         PR target/58065
1903         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1905 2013-08-07  Xinliang David Li  <davidxl@google.com>
1907         * config/i386/i386.opt: New option -mtune-ctrl=.
1908         * config/i386/x86-tune.def: New file.
1909         * config/i386/i386.h: include x86-tune.def.
1910         * config/i386/i386.c (ix86_option_override_internal):
1911         Parsing -mtune-ctrl= option and set tune features.
1913 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
1915         PR other/12081
1916         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1917         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1918         to gen_2arg_fn_t.
1920 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1922         * rtl.h (update_alignments): Declare.
1923         * final.c (grow_label_align): New function extracted from...
1924         (shorten_branches): ...here.  Call it.
1925         (update_alignments): New function.
1926         * reorg.c (sibling_labels): New variable.
1927         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1928         the new label along with it onto the sibling_labels vector.
1929         (fill_simple_delay_slots): Adjust call to get_label_before.
1930         (fill_slots_from_thread): Likewise.
1931         (relax_delay_slots): Likewise.
1932         (make_return_insns): Likewise.
1933         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1935 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1937         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1938         document its semantics.
1939         (diagnostic_report_diagnostic): Adjust accordingly.
1941 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1943         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1944         (sparc_option_override): ...and port to new C++ pass API.
1945         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1947 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1949         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1951 2013-08-06  Caroline Tice  <cmtice@google.com>
1953         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1954         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1955         * tree-pass.h: Add pass_vtable_verify.
1956         * varasm.c (assemble_variable): Add code to properly set the comdat
1957         section and name for the .vtable_map_vars section.
1958         (assemble_vtyv_preinit_initializer): New function.
1959         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1960         LINK_ONCE flag.
1961         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1962         * vtable-verify.c: New file.
1963         * vtable-verify.h: New file.
1964         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1965         initialiation levels.
1966         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1967         * passes.def: Insert pass_vtable_verify.
1968         * aclocal.m4: Reorder includes.
1969         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1970         -fvtv-counts options.
1971         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1972         as appropriate, if -fvtable-verify=... is used.
1973         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1974         -fvtable-verify=... is used.
1975         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1976         (vtable-verify.o): Add new build rule.
1977         (GTFILES): Add vtable-verify.c to list.
1978         * common.opt (fvtable-verify=): New flag.
1979         (vtv_priority): Values for fvtable-verify= flag.
1980         (fvtv-counts): New flag.
1981         (fvtv-debug): New flag.
1982         * tree.h (save_vtable_map_decl): New extern function decl.
1984 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1986         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1987         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1988         (pass_data_rl78_devirt): ...new pass_data instance and...
1989         (make_pass_rl78_devirt): ...new function.
1990         (rl78_asm_file_start): Port pass registration to new C++ API.
1992 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1994         * coretypes.h (rtl_opt_pass): Add.
1995         (gcc::context): Add.
1996         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1997         (epiphany_init): Port to new C++ pass API.
1998         (epiphany_optimize_mode_switching): Likewise.
1999         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2000         (pass_manager::get_pass_mode_switching): New.
2001         (pass_manager::get_pass_peephole2): New.
2002         * mode-switching.c (pass_mode_switching): Add clone method.
2003         * recog.c (pass_peephole2): Add clone method.
2004         (pass_split_all_insns): Add clone method.
2006 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
2008         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2009         (mips_option_override): ...here, porting to new C++ API for passes.
2011 2013-08-06  Jan Hubicka  <jh@suse.cz>
2013         * cgraph.c (cgraph_get_body): New function based on lto.c
2014         implementation.
2015         * cgraph.h (cgraph_get_body): Declare.
2016         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2017         LTO paths.
2018         * cgraphunit.c (expand_function): Get body prior expanding.
2019         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2020         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2021         really need.
2022         * passes.c (do_per_function_toporder): Get body.
2023         * tree-inline.c (expand_call_inline): Get body prior inlining it.
2024         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2026 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2028         PR fortran/57987
2029         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2030         is not re-finalized.  Rename second parameter to no_collect.
2032 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2034         PR middle-end/58041
2035         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2036         MEM_REF has proper alignment information.
2038 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
2040         PR other/12081
2041         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2042         class insn_gen_fn.
2043         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2044         rtx (*) (rtx, ...) with insn_gen_fn.
2045         * genoutput.c (output_insn_data): Cast gen_? function pointers to
2046         insn_gen_fn::stored_funcptr.  Add initializer braces.
2048 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2050         Rewrite how instances of passes are cloned to remove assumptions
2051         about their sizes (thus allowing pass subclasses to have
2052         additional data fields, albeit non-GC-managed ones at this point).
2054         * passes.c (make_pass_instance): Now that passes have clone
2055         methods, rewrite this function to eliminate XNEW and memcpy
2056         calls that used hardcoded sizes.  Since this function no longer
2057         creates pass instances, rename it to...
2058         (add_pass_instance): ...this.  Document the old way that passes were
2059         numbered and flagged, and rework this function to continue using it.
2060         (next_pass_1): Add an initial_pass argument for use by
2061         add_pass_instance.
2062         (position_pass): When adding multiple instances of a pass, use
2063         the pass's clone method, rather than relying on the XNEW/memcpy
2064         within the former make_pass_instance (now add_pass_instance).
2065         (pass_manager::pass_manager): When invoking next_pass_1, also supply
2066         the initial instance of the current pass within the pass manager.
2068 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2070         This is the automated part of the conversion of passes from C
2071         structs to C++ classes.
2073         Patch autogenerated by refactor_passes.py from
2074         https://github.com/davidmalcolm/gcc-refactoring-scripts
2075         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2077         * asan.c (pass_asan): Convert from a global struct to a subclass of
2078         gimple_opt_pass along with...
2079         (pass_data_asan): ...new pass_data instance and...
2080         (make_pass_asan): ...new function.
2081         (pass_asan_O0): Convert from a global struct to a subclass of
2082         gimple_opt_pass along with...
2083         (pass_data_asan_O0): ...new pass_data instance and...
2084         (make_pass_asan_O0): ...new function.
2085         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2086         subclass of rtl_opt_pass along with...
2087         (pass_data_inc_dec): ...new pass_data instance and...
2088         (make_pass_inc_dec): ...new function.
2089         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2090         a subclass of rtl_opt_pass along with...
2091         (pass_data_reorder_blocks): ...new pass_data instance and...
2092         (make_pass_reorder_blocks): ...new function.
2093         (pass_duplicate_computed_gotos): Convert from a global struct to a
2094         subclass of rtl_opt_pass along with...
2095         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2096         (make_pass_duplicate_computed_gotos): ...new function.
2097         (pass_partition_blocks): Convert from a global struct to a subclass of
2098         rtl_opt_pass along with...
2099         (pass_data_partition_blocks): ...new pass_data instance and...
2100         (make_pass_partition_blocks): ...new function.
2101         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2102         struct to a subclass of rtl_opt_pass along with...
2103         (pass_data_branch_target_load_optimize1): ...new pass_data instance
2104         and...
2105         (make_pass_branch_target_load_optimize1): ...new function.
2106         (pass_branch_target_load_optimize2): Convert from a global struct to a
2107         subclass of rtl_opt_pass along with...
2108         (pass_data_branch_target_load_optimize2): ...new pass_data instance
2109         and...
2110         (make_pass_branch_target_load_optimize2): ...new function.
2111         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2112         of rtl_opt_pass along with...
2113         (pass_data_jump): ...new pass_data instance and...
2114         (make_pass_jump): ...new function.
2115         (pass_jump2): Convert from a global struct to a subclass of
2116         rtl_opt_pass along with...
2117         (pass_data_jump2): ...new pass_data instance and...
2118         (make_pass_jump2): ...new function.
2119         * cfgexpand.c (pass_expand): Convert from a global struct to a
2120         subclass of rtl_opt_pass along with...
2121         (pass_data_expand): ...new pass_data instance and...
2122         (make_pass_expand): ...new function.
2123         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2124         of rtl_opt_pass along with...
2125         (pass_data_free_cfg): ...new pass_data instance and...
2126         (make_pass_free_cfg): ...new function.
2127         (pass_into_cfg_layout_mode): Convert from a global struct to a
2128         subclass of rtl_opt_pass along with...
2129         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2130         (make_pass_into_cfg_layout_mode): ...new function.
2131         (pass_outof_cfg_layout_mode): Convert from a global struct to a
2132         subclass of rtl_opt_pass along with...
2133         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2134         (make_pass_outof_cfg_layout_mode): ...new function.
2135         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2136         struct to a subclass of gimple_opt_pass along with...
2137         (pass_data_build_cgraph_edges): ...new pass_data instance and...
2138         (make_pass_build_cgraph_edges): ...new function.
2139         (pass_rebuild_cgraph_edges): Convert from a global struct to a
2140         subclass of gimple_opt_pass along with...
2141         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2142         (make_pass_rebuild_cgraph_edges): ...new function.
2143         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2144         subclass of gimple_opt_pass along with...
2145         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2146         and...
2147         (make_pass_remove_cgraph_callee_edges): ...new function.
2148         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2149         struct to a subclass of rtl_opt_pass along with...
2150         (pass_data_stack_adjustments): ...new pass_data instance and...
2151         (make_pass_stack_adjustments): ...new function.
2152         * combine.c (pass_combine): Convert from a global struct to a subclass
2153         of rtl_opt_pass along with...
2154         (pass_data_combine): ...new pass_data instance and...
2155         (make_pass_combine): ...new function.
2156         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2157         global struct to a subclass of rtl_opt_pass along with...
2158         (pass_data_compare_elim_after_reload): ...new pass_data instance
2159         and...
2160         (make_pass_compare_elim_after_reload): ...new function.
2161         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2162         of rtl_opt_pass along with...
2163         (pass_data_rtl_cprop): ...new pass_data instance and...
2164         (make_pass_rtl_cprop): ...new function.
2165         * cse.c (pass_cse): Convert from a global struct to a subclass of
2166         rtl_opt_pass along with...
2167         (pass_data_cse): ...new pass_data instance and...
2168         (make_pass_cse): ...new function.
2169         (pass_cse2): Convert from a global struct to a subclass of
2170         rtl_opt_pass along with...
2171         (pass_data_cse2): ...new pass_data instance and...
2172         (make_pass_cse2): ...new function.
2173         (pass_cse_after_global_opts): Convert from a global struct to a
2174         subclass of rtl_opt_pass along with...
2175         (pass_data_cse_after_global_opts): ...new pass_data instance and...
2176         (make_pass_cse_after_global_opts): ...new function.
2177         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2178         of rtl_opt_pass along with...
2179         (pass_data_ud_rtl_dce): ...new pass_data instance and...
2180         (make_pass_ud_rtl_dce): ...new function.
2181         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2182         rtl_opt_pass along with...
2183         (pass_data_fast_rtl_dce): ...new pass_data instance and...
2184         (make_pass_fast_rtl_dce): ...new function.
2185         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2186         a subclass of rtl_opt_pass along with...
2187         (pass_data_df_initialize_opt): ...new pass_data instance and...
2188         (make_pass_df_initialize_opt): ...new function.
2189         (pass_df_initialize_no_opt): Convert from a global struct to a
2190         subclass of rtl_opt_pass along with...
2191         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2192         (make_pass_df_initialize_no_opt): ...new function.
2193         (pass_df_finish): Convert from a global struct to a subclass of
2194         rtl_opt_pass along with...
2195         (pass_data_df_finish): ...new pass_data instance and...
2196         (make_pass_df_finish): ...new function.
2197         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2198         rtl_opt_pass along with...
2199         (pass_data_rtl_dse1): ...new pass_data instance and...
2200         (make_pass_rtl_dse1): ...new function.
2201         (pass_rtl_dse2): Convert from a global struct to a subclass of
2202         rtl_opt_pass along with...
2203         (pass_data_rtl_dse2): ...new pass_data instance and...
2204         (make_pass_rtl_dse2): ...new function.
2205         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2206         subclass of rtl_opt_pass along with...
2207         (pass_data_dwarf2_frame): ...new pass_data instance and...
2208         (make_pass_dwarf2_frame): ...new function.
2209         * except.c (pass_set_nothrow_function_flags): Convert from a global
2210         struct to a subclass of rtl_opt_pass along with...
2211         (pass_data_set_nothrow_function_flags): ...new pass_data instance
2212         and...
2213         (make_pass_set_nothrow_function_flags): ...new function.
2214         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2215         subclass of rtl_opt_pass along with...
2216         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2217         and...
2218         (make_pass_convert_to_eh_region_ranges): ...new function.
2219         * final.c (pass_compute_alignments): Convert from a global struct to a
2220         subclass of rtl_opt_pass along with...
2221         (pass_data_compute_alignments): ...new pass_data instance and...
2222         (make_pass_compute_alignments): ...new function.
2223         (pass_final): Convert from a global struct to a subclass of
2224         rtl_opt_pass along with...
2225         (pass_data_final): ...new pass_data instance and...
2226         (make_pass_final): ...new function.
2227         (pass_shorten_branches): Convert from a global struct to a subclass of
2228         rtl_opt_pass along with...
2229         (pass_data_shorten_branches): ...new pass_data instance and...
2230         (make_pass_shorten_branches): ...new function.
2231         (pass_clean_state): Convert from a global struct to a subclass of
2232         rtl_opt_pass along with...
2233         (pass_data_clean_state): ...new pass_data instance and...
2234         (make_pass_clean_state): ...new function.
2235         * function.c (pass_instantiate_virtual_regs): Convert from a global
2236         struct to a subclass of rtl_opt_pass along with...
2237         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2238         (make_pass_instantiate_virtual_regs): ...new function.
2239         (pass_leaf_regs): Convert from a global struct to a subclass of
2240         rtl_opt_pass along with...
2241         (pass_data_leaf_regs): ...new pass_data instance and...
2242         (make_pass_leaf_regs): ...new function.
2243         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2244         subclass of rtl_opt_pass along with...
2245         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2246         and...
2247         (make_pass_thread_prologue_and_epilogue): ...new function.
2248         (pass_match_asm_constraints): Convert from a global struct to a
2249         subclass of rtl_opt_pass along with...
2250         (pass_data_match_asm_constraints): ...new pass_data instance and...
2251         (make_pass_match_asm_constraints): ...new function.
2252         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2253         subclass of rtl_opt_pass along with...
2254         (pass_data_rtl_fwprop): ...new pass_data instance and...
2255         (make_pass_rtl_fwprop): ...new function.
2256         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2257         rtl_opt_pass along with...
2258         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2259         (make_pass_rtl_fwprop_addr): ...new function.
2260         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2261         rtl_opt_pass along with...
2262         (pass_data_rtl_pre): ...new pass_data instance and...
2263         (make_pass_rtl_pre): ...new function.
2264         (pass_rtl_hoist): Convert from a global struct to a subclass of
2265         rtl_opt_pass along with...
2266         (pass_data_rtl_hoist): ...new pass_data instance and...
2267         (make_pass_rtl_hoist): ...new function.
2268         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2269         subclass of gimple_opt_pass along with...
2270         (pass_data_lower_cf): ...new pass_data instance and...
2271         (make_pass_lower_cf): ...new function.
2272         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2273         from a global struct to a subclass of gimple_opt_pass along with...
2274         (pass_data_strength_reduction): ...new pass_data instance and...
2275         (make_pass_strength_reduction): ...new function.
2276         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2277         of rtl_opt_pass along with...
2278         (pass_data_rtl_ifcvt): ...new pass_data instance and...
2279         (make_pass_rtl_ifcvt): ...new function.
2280         (pass_if_after_combine): Convert from a global struct to a subclass of
2281         rtl_opt_pass along with...
2282         (pass_data_if_after_combine): ...new pass_data instance and...
2283         (make_pass_if_after_combine): ...new function.
2284         (pass_if_after_reload): Convert from a global struct to a subclass of
2285         rtl_opt_pass along with...
2286         (pass_data_if_after_reload): ...new pass_data instance and...
2287         (make_pass_if_after_reload): ...new function.
2288         * init-regs.c (pass_initialize_regs): Convert from a global struct to
2289         a subclass of rtl_opt_pass along with...
2290         (pass_data_initialize_regs): ...new pass_data instance and...
2291         (make_pass_initialize_regs): ...new function.
2292         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2293         of ipa_opt_pass_d along with...
2294         (pass_data_ipa_cp): ...new pass_data instance and...
2295         (make_pass_ipa_cp): ...new function.
2296         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2297         global struct to a subclass of gimple_opt_pass along with...
2298         (pass_data_inline_parameters): ...new pass_data instance and...
2299         (make_pass_inline_parameters): ...new function.
2300         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2301         subclass of gimple_opt_pass along with...
2302         (pass_data_early_inline): ...new pass_data instance and...
2303         (make_pass_early_inline): ...new function.
2304         (pass_ipa_inline): Convert from a global struct to a subclass of
2305         ipa_opt_pass_d along with...
2306         (pass_data_ipa_inline): ...new pass_data instance and...
2307         (make_pass_ipa_inline): ...new function.
2308         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2309         struct to a subclass of gimple_opt_pass along with...
2310         (pass_data_local_pure_const): ...new pass_data instance and...
2311         (make_pass_local_pure_const): ...new function.
2312         (pass_ipa_pure_const): Convert from a global struct to a subclass of
2313         ipa_opt_pass_d along with...
2314         (pass_data_ipa_pure_const): ...new pass_data instance and...
2315         (make_pass_ipa_pure_const): ...new function.
2316         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2317         to a subclass of ipa_opt_pass_d along with...
2318         (pass_data_ipa_reference): ...new pass_data instance and...
2319         (make_pass_ipa_reference): ...new function.
2320         * ipa-split.c (pass_split_functions): Convert from a global struct to
2321         a subclass of gimple_opt_pass along with...
2322         (pass_data_split_functions): ...new pass_data instance and...
2323         (make_pass_split_functions): ...new function.
2324         (pass_feedback_split_functions): Convert from a global struct to a
2325         subclass of gimple_opt_pass along with...
2326         (pass_data_feedback_split_functions): ...new pass_data instance and...
2327         (make_pass_feedback_split_functions): ...new function.
2328         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2329         global struct to a subclass of simple_ipa_opt_pass along with...
2330         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2331         instance and...
2332         (make_pass_ipa_function_and_variable_visibility): ...new function.
2333         (pass_ipa_free_inline_summary): Convert from a global struct to a
2334         subclass of simple_ipa_opt_pass along with...
2335         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2336         (make_pass_ipa_free_inline_summary): ...new function.
2337         (pass_ipa_whole_program_visibility): Convert from a global struct to a
2338         subclass of ipa_opt_pass_d along with...
2339         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2340         and...
2341         (make_pass_ipa_whole_program_visibility): ...new function.
2342         (pass_ipa_profile): Convert from a global struct to a subclass of
2343         ipa_opt_pass_d along with...
2344         (pass_data_ipa_profile): ...new pass_data instance and...
2345         (make_pass_ipa_profile): ...new function.
2346         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2347         ipa_opt_pass_d along with...
2348         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2349         (make_pass_ipa_cdtor_merge): ...new function.
2350         * ira.c (pass_ira): Convert from a global struct to a subclass of
2351         rtl_opt_pass along with...
2352         (pass_data_ira): ...new pass_data instance and...
2353         (make_pass_ira): ...new function.
2354         (pass_reload): Convert from a global struct to a subclass of
2355         rtl_opt_pass along with...
2356         (pass_data_reload): ...new pass_data instance and...
2357         (make_pass_reload): ...new function.
2358         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2359         subclass of rtl_opt_pass along with...
2360         (pass_data_cleanup_barriers): ...new pass_data instance and...
2361         (make_pass_cleanup_barriers): ...new function.
2362         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2363         of rtl_opt_pass along with...
2364         (pass_data_loop2): ...new pass_data instance and...
2365         (make_pass_loop2): ...new function.
2366         (pass_rtl_loop_init): Convert from a global struct to a subclass of
2367         rtl_opt_pass along with...
2368         (pass_data_rtl_loop_init): ...new pass_data instance and...
2369         (make_pass_rtl_loop_init): ...new function.
2370         (pass_rtl_loop_done): Convert from a global struct to a subclass of
2371         rtl_opt_pass along with...
2372         (pass_data_rtl_loop_done): ...new pass_data instance and...
2373         (make_pass_rtl_loop_done): ...new function.
2374         (pass_rtl_move_loop_invariants): Convert from a global struct to a
2375         subclass of rtl_opt_pass along with...
2376         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2377         (make_pass_rtl_move_loop_invariants): ...new function.
2378         (pass_rtl_unswitch): Convert from a global struct to a subclass of
2379         rtl_opt_pass along with...
2380         (pass_data_rtl_unswitch): ...new pass_data instance and...
2381         (make_pass_rtl_unswitch): ...new function.
2382         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2383         subclass of rtl_opt_pass along with...
2384         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2385         and...
2386         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2387         (pass_rtl_doloop): Convert from a global struct to a subclass of
2388         rtl_opt_pass along with...
2389         (pass_data_rtl_doloop): ...new pass_data instance and...
2390         (make_pass_rtl_doloop): ...new function.
2391         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2392         a subclass of rtl_opt_pass along with...
2393         (pass_data_lower_subreg): ...new pass_data instance and...
2394         (make_pass_lower_subreg): ...new function.
2395         (pass_lower_subreg2): Convert from a global struct to a subclass of
2396         rtl_opt_pass along with...
2397         (pass_data_lower_subreg2): ...new pass_data instance and...
2398         (make_pass_lower_subreg2): ...new function.
2399         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2400         struct to a subclass of ipa_opt_pass_d along with...
2401         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2402         (make_pass_ipa_lto_gimple_out): ...new function.
2403         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2404         of ipa_opt_pass_d along with...
2405         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2406         (make_pass_ipa_lto_finish_out): ...new function.
2407         * mode-switching.c (pass_mode_switching): Convert from a global struct
2408         to a subclass of rtl_opt_pass along with...
2409         (pass_data_mode_switching): ...new pass_data instance and...
2410         (make_pass_mode_switching): ...new function.
2411         * modulo-sched.c (pass_sms): Convert from a global struct to a
2412         subclass of rtl_opt_pass along with...
2413         (pass_data_sms): ...new pass_data instance and...
2414         (make_pass_sms): ...new function.
2415         * omp-low.c (pass_expand_omp): Convert from a global struct to a
2416         subclass of gimple_opt_pass along with...
2417         (pass_data_expand_omp): ...new pass_data instance and...
2418         (make_pass_expand_omp): ...new function.
2419         (pass_lower_omp): Convert from a global struct to a subclass of
2420         gimple_opt_pass along with...
2421         (pass_data_lower_omp): ...new pass_data instance and...
2422         (make_pass_lower_omp): ...new function.
2423         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2424         of gimple_opt_pass along with...
2425         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2426         (make_pass_diagnose_omp_blocks): ...new function.
2427         * passes.c (pass_early_local_passes): Convert from a global struct to
2428         a subclass of simple_ipa_opt_pass along with...
2429         (pass_data_early_local_passes): ...new pass_data instance and...
2430         (make_pass_early_local_passes): ...new function.
2431         (pass_all_early_optimizations): Convert from a global struct to a
2432         subclass of gimple_opt_pass along with...
2433         (pass_data_all_early_optimizations): ...new pass_data instance and...
2434         (make_pass_all_early_optimizations): ...new function.
2435         (pass_all_optimizations): Convert from a global struct to a subclass
2436         of gimple_opt_pass along with...
2437         (pass_data_all_optimizations): ...new pass_data instance and...
2438         (make_pass_all_optimizations): ...new function.
2439         (pass_all_optimizations_g): Convert from a global struct to a subclass
2440         of gimple_opt_pass along with...
2441         (pass_data_all_optimizations_g): ...new pass_data instance and...
2442         (make_pass_all_optimizations_g): ...new function.
2443         (pass_rest_of_compilation): Convert from a global struct to a subclass
2444         of rtl_opt_pass along with...
2445         (pass_data_rest_of_compilation): ...new pass_data instance and...
2446         (make_pass_rest_of_compilation): ...new function.
2447         (pass_postreload): Convert from a global struct to a subclass of
2448         rtl_opt_pass along with...
2449         (pass_data_postreload): ...new pass_data instance and...
2450         (make_pass_postreload): ...new function.
2451         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2452         subclass of rtl_opt_pass along with...
2453         (pass_data_gcse2): ...new pass_data instance and...
2454         (make_pass_gcse2): ...new function.
2455         * postreload.c (pass_postreload_cse): Convert from a global struct to
2456         a subclass of rtl_opt_pass along with...
2457         (pass_data_postreload_cse): ...new pass_data instance and...
2458         (make_pass_postreload_cse): ...new function.
2459         * predict.c (pass_profile): Convert from a global struct to a subclass
2460         of gimple_opt_pass along with...
2461         (pass_data_profile): ...new pass_data instance and...
2462         (make_pass_profile): ...new function.
2463         (pass_strip_predict_hints): Convert from a global struct to a subclass
2464         of gimple_opt_pass along with...
2465         (pass_data_strip_predict_hints): ...new pass_data instance and...
2466         (make_pass_strip_predict_hints): ...new function.
2467         * recog.c (pass_peephole2): Convert from a global struct to a subclass
2468         of rtl_opt_pass along with...
2469         (pass_data_peephole2): ...new pass_data instance and...
2470         (make_pass_peephole2): ...new function.
2471         (pass_split_all_insns): Convert from a global struct to a subclass of
2472         rtl_opt_pass along with...
2473         (pass_data_split_all_insns): ...new pass_data instance and...
2474         (make_pass_split_all_insns): ...new function.
2475         (pass_split_after_reload): Convert from a global struct to a subclass
2476         of rtl_opt_pass along with...
2477         (pass_data_split_after_reload): ...new pass_data instance and...
2478         (make_pass_split_after_reload): ...new function.
2479         (pass_split_before_regstack): Convert from a global struct to a
2480         subclass of rtl_opt_pass along with...
2481         (pass_data_split_before_regstack): ...new pass_data instance and...
2482         (make_pass_split_before_regstack): ...new function.
2483         (pass_split_before_sched2): Convert from a global struct to a subclass
2484         of rtl_opt_pass along with...
2485         (pass_data_split_before_sched2): ...new pass_data instance and...
2486         (make_pass_split_before_sched2): ...new function.
2487         (pass_split_for_shorten_branches): Convert from a global struct to a
2488         subclass of rtl_opt_pass along with...
2489         (pass_data_split_for_shorten_branches): ...new pass_data instance
2490         and...
2491         (make_pass_split_for_shorten_branches): ...new function.
2492         * ree.c (pass_ree): Convert from a global struct to a subclass of
2493         rtl_opt_pass along with...
2494         (pass_data_ree): ...new pass_data instance and...
2495         (make_pass_ree): ...new function.
2496         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2497         subclass of rtl_opt_pass along with...
2498         (pass_data_stack_regs): ...new pass_data instance and...
2499         (make_pass_stack_regs): ...new function.
2500         (pass_stack_regs_run): Convert from a global struct to a subclass of
2501         rtl_opt_pass along with...
2502         (pass_data_stack_regs_run): ...new pass_data instance and...
2503         (make_pass_stack_regs_run): ...new function.
2504         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2505         subclass of rtl_opt_pass along with...
2506         (pass_data_cprop_hardreg): ...new pass_data instance and...
2507         (make_pass_cprop_hardreg): ...new function.
2508         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2509         subclass of rtl_opt_pass along with...
2510         (pass_data_reginfo_init): ...new pass_data instance and...
2511         (make_pass_reginfo_init): ...new function.
2512         * regmove.c (pass_regmove): Convert from a global struct to a subclass
2513         of rtl_opt_pass along with...
2514         (pass_data_regmove): ...new pass_data instance and...
2515         (make_pass_regmove): ...new function.
2516         * regrename.c (pass_regrename): Convert from a global struct to a
2517         subclass of rtl_opt_pass along with...
2518         (pass_data_regrename): ...new pass_data instance and...
2519         (make_pass_regrename): ...new function.
2520         * reorg.c (pass_delay_slots): Convert from a global struct to a
2521         subclass of rtl_opt_pass along with...
2522         (pass_data_delay_slots): ...new pass_data instance and...
2523         (make_pass_delay_slots): ...new function.
2524         (pass_machine_reorg): Convert from a global struct to a subclass of
2525         rtl_opt_pass along with...
2526         (pass_data_machine_reorg): ...new pass_data instance and...
2527         (make_pass_machine_reorg): ...new function.
2528         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2529         of rtl_opt_pass along with...
2530         (pass_data_sched): ...new pass_data instance and...
2531         (make_pass_sched): ...new function.
2532         (pass_sched2): Convert from a global struct to a subclass of
2533         rtl_opt_pass along with...
2534         (pass_data_sched2): ...new pass_data instance and...
2535         (make_pass_sched2): ...new function.
2536         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2537         to a subclass of rtl_opt_pass along with...
2538         (pass_data_stack_ptr_mod): ...new pass_data instance and...
2539         (make_pass_stack_ptr_mod): ...new function.
2540         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2541         to a subclass of rtl_opt_pass along with...
2542         (pass_data_rtl_store_motion): ...new pass_data instance and...
2543         (make_pass_rtl_store_motion): ...new function.
2544         * tracer.c (pass_tracer): Convert from a global struct to a subclass
2545         of gimple_opt_pass along with...
2546         (pass_data_tracer): ...new pass_data instance and...
2547         (make_pass_tracer): ...new function.
2548         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2549         to a subclass of gimple_opt_pass along with...
2550         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2551         (make_pass_diagnose_tm_blocks): ...new function.
2552         (pass_lower_tm): Convert from a global struct to a subclass of
2553         gimple_opt_pass along with...
2554         (pass_data_lower_tm): ...new pass_data instance and...
2555         (make_pass_lower_tm): ...new function.
2556         (pass_tm_init): Convert from a global struct to a subclass of
2557         gimple_opt_pass along with...
2558         (pass_data_tm_init): ...new pass_data instance and...
2559         (make_pass_tm_init): ...new function.
2560         (pass_tm_mark): Convert from a global struct to a subclass of
2561         gimple_opt_pass along with...
2562         (pass_data_tm_mark): ...new pass_data instance and...
2563         (make_pass_tm_mark): ...new function.
2564         (pass_tm_edges): Convert from a global struct to a subclass of
2565         gimple_opt_pass along with...
2566         (pass_data_tm_edges): ...new pass_data instance and...
2567         (make_pass_tm_edges): ...new function.
2568         (pass_tm_memopt): Convert from a global struct to a subclass of
2569         gimple_opt_pass along with...
2570         (pass_data_tm_memopt): ...new pass_data instance and...
2571         (make_pass_tm_memopt): ...new function.
2572         (pass_ipa_tm): Convert from a global struct to a subclass of
2573         simple_ipa_opt_pass along with...
2574         (pass_data_ipa_tm): ...new pass_data instance and...
2575         (make_pass_ipa_tm): ...new function.
2576         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2577         subclass of gimple_opt_pass along with...
2578         (pass_data_call_cdce): ...new pass_data instance and...
2579         (make_pass_call_cdce): ...new function.
2580         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2581         subclass of gimple_opt_pass along with...
2582         (pass_data_build_cfg): ...new pass_data instance and...
2583         (make_pass_build_cfg): ...new function.
2584         (pass_split_crit_edges): Convert from a global struct to a subclass of
2585         gimple_opt_pass along with...
2586         (pass_data_split_crit_edges): ...new pass_data instance and...
2587         (make_pass_split_crit_edges): ...new function.
2588         (pass_warn_function_return): Convert from a global struct to a
2589         subclass of gimple_opt_pass along with...
2590         (pass_data_warn_function_return): ...new pass_data instance and...
2591         (make_pass_warn_function_return): ...new function.
2592         (pass_warn_function_noreturn): Convert from a global struct to a
2593         subclass of gimple_opt_pass along with...
2594         (pass_data_warn_function_noreturn): ...new pass_data instance and...
2595         (make_pass_warn_function_noreturn): ...new function.
2596         (pass_warn_unused_result): Convert from a global struct to a subclass
2597         of gimple_opt_pass along with...
2598         (pass_data_warn_unused_result): ...new pass_data instance and...
2599         (make_pass_warn_unused_result): ...new function.
2600         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2601         a subclass of gimple_opt_pass along with...
2602         (pass_data_merge_phi): ...new pass_data instance and...
2603         (make_pass_merge_phi): ...new function.
2604         * tree-complex.c (pass_lower_complex): Convert from a global struct to
2605         a subclass of gimple_opt_pass along with...
2606         (pass_data_lower_complex): ...new pass_data instance and...
2607         (make_pass_lower_complex): ...new function.
2608         (pass_lower_complex_O0): Convert from a global struct to a subclass of
2609         gimple_opt_pass along with...
2610         (pass_data_lower_complex_O0): ...new pass_data instance and...
2611         (make_pass_lower_complex_O0): ...new function.
2612         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2613         subclass of gimple_opt_pass along with...
2614         (pass_data_lower_eh): ...new pass_data instance and...
2615         (make_pass_lower_eh): ...new function.
2616         (pass_refactor_eh): Convert from a global struct to a subclass of
2617         gimple_opt_pass along with...
2618         (pass_data_refactor_eh): ...new pass_data instance and...
2619         (make_pass_refactor_eh): ...new function.
2620         (pass_lower_resx): Convert from a global struct to a subclass of
2621         gimple_opt_pass along with...
2622         (pass_data_lower_resx): ...new pass_data instance and...
2623         (make_pass_lower_resx): ...new function.
2624         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2625         of gimple_opt_pass along with...
2626         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2627         (make_pass_lower_eh_dispatch): ...new function.
2628         (pass_cleanup_eh): Convert from a global struct to a subclass of
2629         gimple_opt_pass along with...
2630         (pass_data_cleanup_eh): ...new pass_data instance and...
2631         (make_pass_cleanup_eh): ...new function.
2632         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2633         to a subclass of simple_ipa_opt_pass along with...
2634         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2635         (make_pass_ipa_lower_emutls): ...new function.
2636         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2637         a subclass of gimple_opt_pass along with...
2638         (pass_data_if_conversion): ...new pass_data instance and...
2639         (make_pass_if_conversion): ...new function.
2640         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2641         subclass of gimple_opt_pass along with...
2642         (pass_data_build_ssa): ...new pass_data instance and...
2643         (make_pass_build_ssa): ...new function.
2644         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2645         global struct to a subclass of gimple_opt_pass along with...
2646         (pass_data_loop_distribution): ...new pass_data instance and...
2647         (make_pass_loop_distribution): ...new function.
2648         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2649         subclass of gimple_opt_pass along with...
2650         (pass_data_mudflap_1): ...new pass_data instance and...
2651         (make_pass_mudflap_1): ...new function.
2652         (pass_mudflap_2): Convert from a global struct to a subclass of
2653         gimple_opt_pass along with...
2654         (pass_data_mudflap_2): ...new pass_data instance and...
2655         (make_pass_mudflap_2): ...new function.
2656         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2657         subclass of gimple_opt_pass along with...
2658         (pass_data_mudflap_1): ...new pass_data instance and...
2659         (make_pass_mudflap_1): ...new function.
2660         (pass_mudflap_2): Convert from a global struct to a subclass of
2661         gimple_opt_pass along with...
2662         (pass_data_mudflap_2): ...new pass_data instance and...
2663         (make_pass_mudflap_2): ...new function.
2664         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2665         gimple_opt_pass along with...
2666         (pass_data_nrv): ...new pass_data instance and...
2667         (make_pass_nrv): ...new function.
2668         (pass_return_slot): Convert from a global struct to a subclass of
2669         gimple_opt_pass along with...
2670         (pass_data_return_slot): ...new pass_data instance and...
2671         (make_pass_return_slot): ...new function.
2672         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2673         to a subclass of gimple_opt_pass along with...
2674         (pass_data_object_sizes): ...new pass_data instance and...
2675         (make_pass_object_sizes): ...new function.
2676         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2677         global struct to a subclass of gimple_opt_pass along with...
2678         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2679         and...
2680         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2681         (pass_fixup_cfg): Convert from a global struct to a subclass of
2682         gimple_opt_pass along with...
2683         (pass_data_fixup_cfg): ...new pass_data instance and...
2684         (make_pass_fixup_cfg): ...new function.
2685         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2686         (make_pass_mudflap_1): ...new function.
2687         (pass_mudflap_2): Replace declaration with that of...
2688         (make_pass_mudflap_2): ...new function.
2689         (pass_asan): Replace declaration with that of...
2690         (make_pass_asan): ...new function.
2691         (pass_asan_O0): Replace declaration with that of...
2692         (make_pass_asan_O0): ...new function.
2693         (pass_tsan): Replace declaration with that of...
2694         (make_pass_tsan): ...new function.
2695         (pass_tsan_O0): Replace declaration with that of...
2696         (make_pass_tsan_O0): ...new function.
2697         (pass_lower_cf): Replace declaration with that of...
2698         (make_pass_lower_cf): ...new function.
2699         (pass_refactor_eh): Replace declaration with that of...
2700         (make_pass_refactor_eh): ...new function.
2701         (pass_lower_eh): Replace declaration with that of...
2702         (make_pass_lower_eh): ...new function.
2703         (pass_lower_eh_dispatch): Replace declaration with that of...
2704         (make_pass_lower_eh_dispatch): ...new function.
2705         (pass_lower_resx): Replace declaration with that of...
2706         (make_pass_lower_resx): ...new function.
2707         (pass_build_cfg): Replace declaration with that of...
2708         (make_pass_build_cfg): ...new function.
2709         (pass_early_tree_profile): Replace declaration with that of...
2710         (make_pass_early_tree_profile): ...new function.
2711         (pass_cleanup_eh): Replace declaration with that of...
2712         (make_pass_cleanup_eh): ...new function.
2713         (pass_sra): Replace declaration with that of...
2714         (make_pass_sra): ...new function.
2715         (pass_sra_early): Replace declaration with that of...
2716         (make_pass_sra_early): ...new function.
2717         (pass_early_ipa_sra): Replace declaration with that of...
2718         (make_pass_early_ipa_sra): ...new function.
2719         (pass_tail_recursion): Replace declaration with that of...
2720         (make_pass_tail_recursion): ...new function.
2721         (pass_tail_calls): Replace declaration with that of...
2722         (make_pass_tail_calls): ...new function.
2723         (pass_tree_loop): Replace declaration with that of...
2724         (make_pass_tree_loop): ...new function.
2725         (pass_tree_loop_init): Replace declaration with that of...
2726         (make_pass_tree_loop_init): ...new function.
2727         (pass_lim): Replace declaration with that of...
2728         (make_pass_lim): ...new function.
2729         (pass_tree_unswitch): Replace declaration with that of...
2730         (make_pass_tree_unswitch): ...new function.
2731         (pass_predcom): Replace declaration with that of...
2732         (make_pass_predcom): ...new function.
2733         (pass_iv_canon): Replace declaration with that of...
2734         (make_pass_iv_canon): ...new function.
2735         (pass_scev_cprop): Replace declaration with that of...
2736         (make_pass_scev_cprop): ...new function.
2737         (pass_empty_loop): Replace declaration with that of...
2738         (make_pass_empty_loop): ...new function.
2739         (pass_record_bounds): Replace declaration with that of...
2740         (make_pass_record_bounds): ...new function.
2741         (pass_graphite): Replace declaration with that of...
2742         (make_pass_graphite): ...new function.
2743         (pass_graphite_transforms): Replace declaration with that of...
2744         (make_pass_graphite_transforms): ...new function.
2745         (pass_if_conversion): Replace declaration with that of...
2746         (make_pass_if_conversion): ...new function.
2747         (pass_loop_distribution): Replace declaration with that of...
2748         (make_pass_loop_distribution): ...new function.
2749         (pass_vectorize): Replace declaration with that of...
2750         (make_pass_vectorize): ...new function.
2751         (pass_slp_vectorize): Replace declaration with that of...
2752         (make_pass_slp_vectorize): ...new function.
2753         (pass_complete_unroll): Replace declaration with that of...
2754         (make_pass_complete_unroll): ...new function.
2755         (pass_complete_unrolli): Replace declaration with that of...
2756         (make_pass_complete_unrolli): ...new function.
2757         (pass_parallelize_loops): Replace declaration with that of...
2758         (make_pass_parallelize_loops): ...new function.
2759         (pass_loop_prefetch): Replace declaration with that of...
2760         (make_pass_loop_prefetch): ...new function.
2761         (pass_iv_optimize): Replace declaration with that of...
2762         (make_pass_iv_optimize): ...new function.
2763         (pass_tree_loop_done): Replace declaration with that of...
2764         (make_pass_tree_loop_done): ...new function.
2765         (pass_ch): Replace declaration with that of...
2766         (make_pass_ch): ...new function.
2767         (pass_ccp): Replace declaration with that of...
2768         (make_pass_ccp): ...new function.
2769         (pass_phi_only_cprop): Replace declaration with that of...
2770         (make_pass_phi_only_cprop): ...new function.
2771         (pass_build_ssa): Replace declaration with that of...
2772         (make_pass_build_ssa): ...new function.
2773         (pass_build_alias): Replace declaration with that of...
2774         (make_pass_build_alias): ...new function.
2775         (pass_build_ealias): Replace declaration with that of...
2776         (make_pass_build_ealias): ...new function.
2777         (pass_dominator): Replace declaration with that of...
2778         (make_pass_dominator): ...new function.
2779         (pass_dce): Replace declaration with that of...
2780         (make_pass_dce): ...new function.
2781         (pass_dce_loop): Replace declaration with that of...
2782         (make_pass_dce_loop): ...new function.
2783         (pass_cd_dce): Replace declaration with that of...
2784         (make_pass_cd_dce): ...new function.
2785         (pass_call_cdce): Replace declaration with that of...
2786         (make_pass_call_cdce): ...new function.
2787         (pass_merge_phi): Replace declaration with that of...
2788         (make_pass_merge_phi): ...new function.
2789         (pass_split_crit_edges): Replace declaration with that of...
2790         (make_pass_split_crit_edges): ...new function.
2791         (pass_pre): Replace declaration with that of...
2792         (make_pass_pre): ...new function.
2793         (pass_profile): Replace declaration with that of...
2794         (make_pass_profile): ...new function.
2795         (pass_strip_predict_hints): Replace declaration with that of...
2796         (make_pass_strip_predict_hints): ...new function.
2797         (pass_lower_complex_O0): Replace declaration with that of...
2798         (make_pass_lower_complex_O0): ...new function.
2799         (pass_lower_complex): Replace declaration with that of...
2800         (make_pass_lower_complex): ...new function.
2801         (pass_lower_vector): Replace declaration with that of...
2802         (make_pass_lower_vector): ...new function.
2803         (pass_lower_vector_ssa): Replace declaration with that of...
2804         (make_pass_lower_vector_ssa): ...new function.
2805         (pass_lower_omp): Replace declaration with that of...
2806         (make_pass_lower_omp): ...new function.
2807         (pass_diagnose_omp_blocks): Replace declaration with that of...
2808         (make_pass_diagnose_omp_blocks): ...new function.
2809         (pass_expand_omp): Replace declaration with that of...
2810         (make_pass_expand_omp): ...new function.
2811         (pass_expand_omp_ssa): Replace declaration with that of...
2812         (make_pass_expand_omp_ssa): ...new function.
2813         (pass_object_sizes): Replace declaration with that of...
2814         (make_pass_object_sizes): ...new function.
2815         (pass_strlen): Replace declaration with that of...
2816         (make_pass_strlen): ...new function.
2817         (pass_fold_builtins): Replace declaration with that of...
2818         (make_pass_fold_builtins): ...new function.
2819         (pass_stdarg): Replace declaration with that of...
2820         (make_pass_stdarg): ...new function.
2821         (pass_early_warn_uninitialized): Replace declaration with that of...
2822         (make_pass_early_warn_uninitialized): ...new function.
2823         (pass_late_warn_uninitialized): Replace declaration with that of...
2824         (make_pass_late_warn_uninitialized): ...new function.
2825         (pass_cse_reciprocals): Replace declaration with that of...
2826         (make_pass_cse_reciprocals): ...new function.
2827         (pass_cse_sincos): Replace declaration with that of...
2828         (make_pass_cse_sincos): ...new function.
2829         (pass_optimize_bswap): Replace declaration with that of...
2830         (make_pass_optimize_bswap): ...new function.
2831         (pass_optimize_widening_mul): Replace declaration with that of...
2832         (make_pass_optimize_widening_mul): ...new function.
2833         (pass_warn_function_return): Replace declaration with that of...
2834         (make_pass_warn_function_return): ...new function.
2835         (pass_warn_function_noreturn): Replace declaration with that of...
2836         (make_pass_warn_function_noreturn): ...new function.
2837         (pass_cselim): Replace declaration with that of...
2838         (make_pass_cselim): ...new function.
2839         (pass_phiopt): Replace declaration with that of...
2840         (make_pass_phiopt): ...new function.
2841         (pass_forwprop): Replace declaration with that of...
2842         (make_pass_forwprop): ...new function.
2843         (pass_phiprop): Replace declaration with that of...
2844         (make_pass_phiprop): ...new function.
2845         (pass_tree_ifcombine): Replace declaration with that of...
2846         (make_pass_tree_ifcombine): ...new function.
2847         (pass_dse): Replace declaration with that of...
2848         (make_pass_dse): ...new function.
2849         (pass_nrv): Replace declaration with that of...
2850         (make_pass_nrv): ...new function.
2851         (pass_rename_ssa_copies): Replace declaration with that of...
2852         (make_pass_rename_ssa_copies): ...new function.
2853         (pass_sink_code): Replace declaration with that of...
2854         (make_pass_sink_code): ...new function.
2855         (pass_fre): Replace declaration with that of...
2856         (make_pass_fre): ...new function.
2857         (pass_check_data_deps): Replace declaration with that of...
2858         (make_pass_check_data_deps): ...new function.
2859         (pass_copy_prop): Replace declaration with that of...
2860         (make_pass_copy_prop): ...new function.
2861         (pass_vrp): Replace declaration with that of...
2862         (make_pass_vrp): ...new function.
2863         (pass_uncprop): Replace declaration with that of...
2864         (make_pass_uncprop): ...new function.
2865         (pass_return_slot): Replace declaration with that of...
2866         (make_pass_return_slot): ...new function.
2867         (pass_reassoc): Replace declaration with that of...
2868         (make_pass_reassoc): ...new function.
2869         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2870         (make_pass_rebuild_cgraph_edges): ...new function.
2871         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2872         (make_pass_remove_cgraph_callee_edges): ...new function.
2873         (pass_build_cgraph_edges): Replace declaration with that of...
2874         (make_pass_build_cgraph_edges): ...new function.
2875         (pass_local_pure_const): Replace declaration with that of...
2876         (make_pass_local_pure_const): ...new function.
2877         (pass_tracer): Replace declaration with that of...
2878         (make_pass_tracer): ...new function.
2879         (pass_warn_unused_result): Replace declaration with that of...
2880         (make_pass_warn_unused_result): ...new function.
2881         (pass_diagnose_tm_blocks): Replace declaration with that of...
2882         (make_pass_diagnose_tm_blocks): ...new function.
2883         (pass_lower_tm): Replace declaration with that of...
2884         (make_pass_lower_tm): ...new function.
2885         (pass_tm_init): Replace declaration with that of...
2886         (make_pass_tm_init): ...new function.
2887         (pass_tm_mark): Replace declaration with that of...
2888         (make_pass_tm_mark): ...new function.
2889         (pass_tm_memopt): Replace declaration with that of...
2890         (make_pass_tm_memopt): ...new function.
2891         (pass_tm_edges): Replace declaration with that of...
2892         (make_pass_tm_edges): ...new function.
2893         (pass_split_functions): Replace declaration with that of...
2894         (make_pass_split_functions): ...new function.
2895         (pass_feedback_split_functions): Replace declaration with that of...
2896         (make_pass_feedback_split_functions): ...new function.
2897         (pass_strength_reduction): Replace declaration with that of...
2898         (make_pass_strength_reduction): ...new function.
2899         (pass_ipa_lower_emutls): Replace declaration with that of...
2900         (make_pass_ipa_lower_emutls): ...new function.
2901         (pass_ipa_function_and_variable_visibility): Replace declaration with
2902         that of...
2903         (make_pass_ipa_function_and_variable_visibility): ...new function.
2904         (pass_ipa_tree_profile): Replace declaration with that of...
2905         (make_pass_ipa_tree_profile): ...new function.
2906         (pass_early_local_passes): Replace declaration with that of...
2907         (make_pass_early_local_passes): ...new function.
2908         (pass_ipa_whole_program_visibility): Replace declaration with that
2909         of...
2910         (make_pass_ipa_whole_program_visibility): ...new function.
2911         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2912         (make_pass_ipa_lto_gimple_out): ...new function.
2913         (pass_ipa_increase_alignment): Replace declaration with that of...
2914         (make_pass_ipa_increase_alignment): ...new function.
2915         (pass_ipa_inline): Replace declaration with that of...
2916         (make_pass_ipa_inline): ...new function.
2917         (pass_ipa_free_lang_data): Replace declaration with that of...
2918         (make_pass_ipa_free_lang_data): ...new function.
2919         (pass_ipa_free_inline_summary): Replace declaration with that of...
2920         (make_pass_ipa_free_inline_summary): ...new function.
2921         (pass_ipa_cp): Replace declaration with that of...
2922         (make_pass_ipa_cp): ...new function.
2923         (pass_ipa_reference): Replace declaration with that of...
2924         (make_pass_ipa_reference): ...new function.
2925         (pass_ipa_pure_const): Replace declaration with that of...
2926         (make_pass_ipa_pure_const): ...new function.
2927         (pass_ipa_pta): Replace declaration with that of...
2928         (make_pass_ipa_pta): ...new function.
2929         (pass_ipa_lto_finish_out): Replace declaration with that of...
2930         (make_pass_ipa_lto_finish_out): ...new function.
2931         (pass_ipa_tm): Replace declaration with that of...
2932         (make_pass_ipa_tm): ...new function.
2933         (pass_ipa_profile): Replace declaration with that of...
2934         (make_pass_ipa_profile): ...new function.
2935         (pass_ipa_cdtor_merge): Replace declaration with that of...
2936         (make_pass_ipa_cdtor_merge): ...new function.
2937         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2938         of...
2939         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2940         (pass_init_datastructures): Replace declaration with that of...
2941         (make_pass_init_datastructures): ...new function.
2942         (pass_fixup_cfg): Replace declaration with that of...
2943         (make_pass_fixup_cfg): ...new function.
2944         (pass_expand): Replace declaration with that of...
2945         (make_pass_expand): ...new function.
2946         (pass_instantiate_virtual_regs): Replace declaration with that of...
2947         (make_pass_instantiate_virtual_regs): ...new function.
2948         (pass_rtl_fwprop): Replace declaration with that of...
2949         (make_pass_rtl_fwprop): ...new function.
2950         (pass_rtl_fwprop_addr): Replace declaration with that of...
2951         (make_pass_rtl_fwprop_addr): ...new function.
2952         (pass_jump): Replace declaration with that of...
2953         (make_pass_jump): ...new function.
2954         (pass_jump2): Replace declaration with that of...
2955         (make_pass_jump2): ...new function.
2956         (pass_lower_subreg): Replace declaration with that of...
2957         (make_pass_lower_subreg): ...new function.
2958         (pass_cse): Replace declaration with that of...
2959         (make_pass_cse): ...new function.
2960         (pass_fast_rtl_dce): Replace declaration with that of...
2961         (make_pass_fast_rtl_dce): ...new function.
2962         (pass_ud_rtl_dce): Replace declaration with that of...
2963         (make_pass_ud_rtl_dce): ...new function.
2964         (pass_rtl_dce): Replace declaration with that of...
2965         (make_pass_rtl_dce): ...new function.
2966         (pass_rtl_dse1): Replace declaration with that of...
2967         (make_pass_rtl_dse1): ...new function.
2968         (pass_rtl_dse2): Replace declaration with that of...
2969         (make_pass_rtl_dse2): ...new function.
2970         (pass_rtl_dse3): Replace declaration with that of...
2971         (make_pass_rtl_dse3): ...new function.
2972         (pass_rtl_cprop): Replace declaration with that of...
2973         (make_pass_rtl_cprop): ...new function.
2974         (pass_rtl_pre): Replace declaration with that of...
2975         (make_pass_rtl_pre): ...new function.
2976         (pass_rtl_hoist): Replace declaration with that of...
2977         (make_pass_rtl_hoist): ...new function.
2978         (pass_rtl_store_motion): Replace declaration with that of...
2979         (make_pass_rtl_store_motion): ...new function.
2980         (pass_cse_after_global_opts): Replace declaration with that of...
2981         (make_pass_cse_after_global_opts): ...new function.
2982         (pass_rtl_ifcvt): Replace declaration with that of...
2983         (make_pass_rtl_ifcvt): ...new function.
2984         (pass_into_cfg_layout_mode): Replace declaration with that of...
2985         (make_pass_into_cfg_layout_mode): ...new function.
2986         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2987         (make_pass_outof_cfg_layout_mode): ...new function.
2988         (pass_loop2): Replace declaration with that of...
2989         (make_pass_loop2): ...new function.
2990         (pass_rtl_loop_init): Replace declaration with that of...
2991         (make_pass_rtl_loop_init): ...new function.
2992         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2993         (make_pass_rtl_move_loop_invariants): ...new function.
2994         (pass_rtl_unswitch): Replace declaration with that of...
2995         (make_pass_rtl_unswitch): ...new function.
2996         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2997         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2998         (pass_rtl_doloop): Replace declaration with that of...
2999         (make_pass_rtl_doloop): ...new function.
3000         (pass_rtl_loop_done): Replace declaration with that of...
3001         (make_pass_rtl_loop_done): ...new function.
3002         (pass_web): Replace declaration with that of...
3003         (make_pass_web): ...new function.
3004         (pass_cse2): Replace declaration with that of...
3005         (make_pass_cse2): ...new function.
3006         (pass_df_initialize_opt): Replace declaration with that of...
3007         (make_pass_df_initialize_opt): ...new function.
3008         (pass_df_initialize_no_opt): Replace declaration with that of...
3009         (make_pass_df_initialize_no_opt): ...new function.
3010         (pass_reginfo_init): Replace declaration with that of...
3011         (make_pass_reginfo_init): ...new function.
3012         (pass_inc_dec): Replace declaration with that of...
3013         (make_pass_inc_dec): ...new function.
3014         (pass_stack_ptr_mod): Replace declaration with that of...
3015         (make_pass_stack_ptr_mod): ...new function.
3016         (pass_initialize_regs): Replace declaration with that of...
3017         (make_pass_initialize_regs): ...new function.
3018         (pass_combine): Replace declaration with that of...
3019         (make_pass_combine): ...new function.
3020         (pass_if_after_combine): Replace declaration with that of...
3021         (make_pass_if_after_combine): ...new function.
3022         (pass_ree): Replace declaration with that of...
3023         (make_pass_ree): ...new function.
3024         (pass_partition_blocks): Replace declaration with that of...
3025         (make_pass_partition_blocks): ...new function.
3026         (pass_match_asm_constraints): Replace declaration with that of...
3027         (make_pass_match_asm_constraints): ...new function.
3028         (pass_regmove): Replace declaration with that of...
3029         (make_pass_regmove): ...new function.
3030         (pass_split_all_insns): Replace declaration with that of...
3031         (make_pass_split_all_insns): ...new function.
3032         (pass_fast_rtl_byte_dce): Replace declaration with that of...
3033         (make_pass_fast_rtl_byte_dce): ...new function.
3034         (pass_lower_subreg2): Replace declaration with that of...
3035         (make_pass_lower_subreg2): ...new function.
3036         (pass_mode_switching): Replace declaration with that of...
3037         (make_pass_mode_switching): ...new function.
3038         (pass_sms): Replace declaration with that of...
3039         (make_pass_sms): ...new function.
3040         (pass_sched): Replace declaration with that of...
3041         (make_pass_sched): ...new function.
3042         (pass_ira): Replace declaration with that of...
3043         (make_pass_ira): ...new function.
3044         (pass_reload): Replace declaration with that of...
3045         (make_pass_reload): ...new function.
3046         (pass_clean_state): Replace declaration with that of...
3047         (make_pass_clean_state): ...new function.
3048         (pass_branch_prob): Replace declaration with that of...
3049         (make_pass_branch_prob): ...new function.
3050         (pass_value_profile_transformations): Replace declaration with that
3051         of...
3052         (make_pass_value_profile_transformations): ...new function.
3053         (pass_postreload_cse): Replace declaration with that of...
3054         (make_pass_postreload_cse): ...new function.
3055         (pass_gcse2): Replace declaration with that of...
3056         (make_pass_gcse2): ...new function.
3057         (pass_split_after_reload): Replace declaration with that of...
3058         (make_pass_split_after_reload): ...new function.
3059         (pass_branch_target_load_optimize1): Replace declaration with that
3060         of...
3061         (make_pass_branch_target_load_optimize1): ...new function.
3062         (pass_thread_prologue_and_epilogue): Replace declaration with that
3063         of...
3064         (make_pass_thread_prologue_and_epilogue): ...new function.
3065         (pass_stack_adjustments): Replace declaration with that of...
3066         (make_pass_stack_adjustments): ...new function.
3067         (pass_peephole2): Replace declaration with that of...
3068         (make_pass_peephole2): ...new function.
3069         (pass_if_after_reload): Replace declaration with that of...
3070         (make_pass_if_after_reload): ...new function.
3071         (pass_regrename): Replace declaration with that of...
3072         (make_pass_regrename): ...new function.
3073         (pass_cprop_hardreg): Replace declaration with that of...
3074         (make_pass_cprop_hardreg): ...new function.
3075         (pass_reorder_blocks): Replace declaration with that of...
3076         (make_pass_reorder_blocks): ...new function.
3077         (pass_branch_target_load_optimize2): Replace declaration with that
3078         of...
3079         (make_pass_branch_target_load_optimize2): ...new function.
3080         (pass_leaf_regs): Replace declaration with that of...
3081         (make_pass_leaf_regs): ...new function.
3082         (pass_split_before_sched2): Replace declaration with that of...
3083         (make_pass_split_before_sched2): ...new function.
3084         (pass_compare_elim_after_reload): Replace declaration with that of...
3085         (make_pass_compare_elim_after_reload): ...new function.
3086         (pass_sched2): Replace declaration with that of...
3087         (make_pass_sched2): ...new function.
3088         (pass_stack_regs): Replace declaration with that of...
3089         (make_pass_stack_regs): ...new function.
3090         (pass_stack_regs_run): Replace declaration with that of...
3091         (make_pass_stack_regs_run): ...new function.
3092         (pass_df_finish): Replace declaration with that of...
3093         (make_pass_df_finish): ...new function.
3094         (pass_compute_alignments): Replace declaration with that of...
3095         (make_pass_compute_alignments): ...new function.
3096         (pass_duplicate_computed_gotos): Replace declaration with that of...
3097         (make_pass_duplicate_computed_gotos): ...new function.
3098         (pass_variable_tracking): Replace declaration with that of...
3099         (make_pass_variable_tracking): ...new function.
3100         (pass_free_cfg): Replace declaration with that of...
3101         (make_pass_free_cfg): ...new function.
3102         (pass_machine_reorg): Replace declaration with that of...
3103         (make_pass_machine_reorg): ...new function.
3104         (pass_cleanup_barriers): Replace declaration with that of...
3105         (make_pass_cleanup_barriers): ...new function.
3106         (pass_delay_slots): Replace declaration with that of...
3107         (make_pass_delay_slots): ...new function.
3108         (pass_split_for_shorten_branches): Replace declaration with that of...
3109         (make_pass_split_for_shorten_branches): ...new function.
3110         (pass_split_before_regstack): Replace declaration with that of...
3111         (make_pass_split_before_regstack): ...new function.
3112         (pass_convert_to_eh_region_ranges): Replace declaration with that
3113         of...
3114         (make_pass_convert_to_eh_region_ranges): ...new function.
3115         (pass_shorten_branches): Replace declaration with that of...
3116         (make_pass_shorten_branches): ...new function.
3117         (pass_set_nothrow_function_flags): Replace declaration with that of...
3118         (make_pass_set_nothrow_function_flags): ...new function.
3119         (pass_dwarf2_frame): Replace declaration with that of...
3120         (make_pass_dwarf2_frame): ...new function.
3121         (pass_final): Replace declaration with that of...
3122         (make_pass_final): ...new function.
3123         (pass_rtl_seqabstr): Replace declaration with that of...
3124         (make_pass_rtl_seqabstr): ...new function.
3125         (pass_release_ssa_names): Replace declaration with that of...
3126         (make_pass_release_ssa_names): ...new function.
3127         (pass_early_inline): Replace declaration with that of...
3128         (make_pass_early_inline): ...new function.
3129         (pass_inline_parameters): Replace declaration with that of...
3130         (make_pass_inline_parameters): ...new function.
3131         (pass_update_address_taken): Replace declaration with that of...
3132         (make_pass_update_address_taken): ...new function.
3133         (pass_convert_switch): Replace declaration with that of...
3134         (make_pass_convert_switch): ...new function.
3135         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3136         to a subclass of simple_ipa_opt_pass along with...
3137         (pass_data_ipa_tree_profile): ...new pass_data instance and...
3138         (make_pass_ipa_tree_profile): ...new function.
3139         * tree-sra.c (pass_sra_early): Convert from a global struct to a
3140         subclass of gimple_opt_pass along with...
3141         (pass_data_sra_early): ...new pass_data instance and...
3142         (make_pass_sra_early): ...new function.
3143         (pass_sra): Convert from a global struct to a subclass of
3144         gimple_opt_pass along with...
3145         (pass_data_sra): ...new pass_data instance and...
3146         (make_pass_sra): ...new function.
3147         (pass_early_ipa_sra): Convert from a global struct to a subclass of
3148         gimple_opt_pass along with...
3149         (pass_data_early_ipa_sra): ...new pass_data instance and...
3150         (make_pass_early_ipa_sra): ...new function.
3151         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3152         subclass of gimple_opt_pass along with...
3153         (pass_data_ccp): ...new pass_data instance and...
3154         (make_pass_ccp): ...new function.
3155         (pass_fold_builtins): Convert from a global struct to a subclass of
3156         gimple_opt_pass along with...
3157         (pass_data_fold_builtins): ...new pass_data instance and...
3158         (make_pass_fold_builtins): ...new function.
3159         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3160         subclass of gimple_opt_pass along with...
3161         (pass_data_copy_prop): ...new pass_data instance and...
3162         (make_pass_copy_prop): ...new function.
3163         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3164         global struct to a subclass of gimple_opt_pass along with...
3165         (pass_data_rename_ssa_copies): ...new pass_data instance and...
3166         (make_pass_rename_ssa_copies): ...new function.
3167         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3168         subclass of gimple_opt_pass along with...
3169         (pass_data_dce): ...new pass_data instance and...
3170         (make_pass_dce): ...new function.
3171         (pass_dce_loop): Convert from a global struct to a subclass of
3172         gimple_opt_pass along with...
3173         (pass_data_dce_loop): ...new pass_data instance and...
3174         (make_pass_dce_loop): ...new function.
3175         (pass_cd_dce): Convert from a global struct to a subclass of
3176         gimple_opt_pass along with...
3177         (pass_data_cd_dce): ...new pass_data instance and...
3178         (make_pass_cd_dce): ...new function.
3179         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3180         subclass of gimple_opt_pass along with...
3181         (pass_data_dominator): ...new pass_data instance and...
3182         (make_pass_dominator): ...new function.
3183         (pass_phi_only_cprop): Convert from a global struct to a subclass of
3184         gimple_opt_pass along with...
3185         (pass_data_phi_only_cprop): ...new pass_data instance and...
3186         (make_pass_phi_only_cprop): ...new function.
3187         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3188         subclass of gimple_opt_pass along with...
3189         (pass_data_dse): ...new pass_data instance and...
3190         (make_pass_dse): ...new function.
3191         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3192         a subclass of gimple_opt_pass along with...
3193         (pass_data_forwprop): ...new pass_data instance and...
3194         (make_pass_forwprop): ...new function.
3195         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3196         struct to a subclass of gimple_opt_pass along with...
3197         (pass_data_tree_ifcombine): ...new pass_data instance and...
3198         (make_pass_tree_ifcombine): ...new function.
3199         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3200         subclass of gimple_opt_pass along with...
3201         (pass_data_ch): ...new pass_data instance and...
3202         (make_pass_ch): ...new function.
3203         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3204         subclass of gimple_opt_pass along with...
3205         (pass_data_tree_loop): ...new pass_data instance and...
3206         (make_pass_tree_loop): ...new function.
3207         (pass_tree_loop_init): Convert from a global struct to a subclass of
3208         gimple_opt_pass along with...
3209         (pass_data_tree_loop_init): ...new pass_data instance and...
3210         (make_pass_tree_loop_init): ...new function.
3211         (pass_lim): Convert from a global struct to a subclass of
3212         gimple_opt_pass along with...
3213         (pass_data_lim): ...new pass_data instance and...
3214         (make_pass_lim): ...new function.
3215         (pass_tree_unswitch): Convert from a global struct to a subclass of
3216         gimple_opt_pass along with...
3217         (pass_data_tree_unswitch): ...new pass_data instance and...
3218         (make_pass_tree_unswitch): ...new function.
3219         (pass_predcom): Convert from a global struct to a subclass of
3220         gimple_opt_pass along with...
3221         (pass_data_predcom): ...new pass_data instance and...
3222         (make_pass_predcom): ...new function.
3223         (pass_vectorize): Convert from a global struct to a subclass of
3224         gimple_opt_pass along with...
3225         (pass_data_vectorize): ...new pass_data instance and...
3226         (make_pass_vectorize): ...new function.
3227         (pass_graphite): Convert from a global struct to a subclass of
3228         gimple_opt_pass along with...
3229         (pass_data_graphite): ...new pass_data instance and...
3230         (make_pass_graphite): ...new function.
3231         (pass_graphite_transforms): Convert from a global struct to a subclass
3232         of gimple_opt_pass along with...
3233         (pass_data_graphite_transforms): ...new pass_data instance and...
3234         (make_pass_graphite_transforms): ...new function.
3235         (pass_check_data_deps): Convert from a global struct to a subclass of
3236         gimple_opt_pass along with...
3237         (pass_data_check_data_deps): ...new pass_data instance and...
3238         (make_pass_check_data_deps): ...new function.
3239         (pass_iv_canon): Convert from a global struct to a subclass of
3240         gimple_opt_pass along with...
3241         (pass_data_iv_canon): ...new pass_data instance and...
3242         (make_pass_iv_canon): ...new function.
3243         (pass_scev_cprop): Convert from a global struct to a subclass of
3244         gimple_opt_pass along with...
3245         (pass_data_scev_cprop): ...new pass_data instance and...
3246         (make_pass_scev_cprop): ...new function.
3247         (pass_record_bounds): Convert from a global struct to a subclass of
3248         gimple_opt_pass along with...
3249         (pass_data_record_bounds): ...new pass_data instance and...
3250         (make_pass_record_bounds): ...new function.
3251         (pass_complete_unroll): Convert from a global struct to a subclass of
3252         gimple_opt_pass along with...
3253         (pass_data_complete_unroll): ...new pass_data instance and...
3254         (make_pass_complete_unroll): ...new function.
3255         (pass_complete_unrolli): Convert from a global struct to a subclass of
3256         gimple_opt_pass along with...
3257         (pass_data_complete_unrolli): ...new pass_data instance and...
3258         (make_pass_complete_unrolli): ...new function.
3259         (pass_parallelize_loops): Convert from a global struct to a subclass
3260         of gimple_opt_pass along with...
3261         (pass_data_parallelize_loops): ...new pass_data instance and...
3262         (make_pass_parallelize_loops): ...new function.
3263         (pass_loop_prefetch): Convert from a global struct to a subclass of
3264         gimple_opt_pass along with...
3265         (pass_data_loop_prefetch): ...new pass_data instance and...
3266         (make_pass_loop_prefetch): ...new function.
3267         (pass_iv_optimize): Convert from a global struct to a subclass of
3268         gimple_opt_pass along with...
3269         (pass_data_iv_optimize): ...new pass_data instance and...
3270         (make_pass_iv_optimize): ...new function.
3271         (pass_tree_loop_done): Convert from a global struct to a subclass of
3272         gimple_opt_pass along with...
3273         (pass_data_tree_loop_done): ...new pass_data instance and...
3274         (make_pass_tree_loop_done): ...new function.
3275         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3276         struct to a subclass of gimple_opt_pass along with...
3277         (pass_data_cse_reciprocals): ...new pass_data instance and...
3278         (make_pass_cse_reciprocals): ...new function.
3279         (pass_cse_sincos): Convert from a global struct to a subclass of
3280         gimple_opt_pass along with...
3281         (pass_data_cse_sincos): ...new pass_data instance and...
3282         (make_pass_cse_sincos): ...new function.
3283         (pass_optimize_bswap): Convert from a global struct to a subclass of
3284         gimple_opt_pass along with...
3285         (pass_data_optimize_bswap): ...new pass_data instance and...
3286         (make_pass_optimize_bswap): ...new function.
3287         (pass_optimize_widening_mul): Convert from a global struct to a
3288         subclass of gimple_opt_pass along with...
3289         (pass_data_optimize_widening_mul): ...new pass_data instance and...
3290         (make_pass_optimize_widening_mul): ...new function.
3291         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3292         subclass of gimple_opt_pass along with...
3293         (pass_data_phiopt): ...new pass_data instance and...
3294         (make_pass_phiopt): ...new function.
3295         (pass_cselim): Convert from a global struct to a subclass of
3296         gimple_opt_pass along with...
3297         (pass_data_cselim): ...new pass_data instance and...
3298         (make_pass_cselim): ...new function.
3299         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3300         subclass of gimple_opt_pass along with...
3301         (pass_data_phiprop): ...new pass_data instance and...
3302         (make_pass_phiprop): ...new function.
3303         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3304         subclass of gimple_opt_pass along with...
3305         (pass_data_pre): ...new pass_data instance and...
3306         (make_pass_pre): ...new function.
3307         (pass_fre): Convert from a global struct to a subclass of
3308         gimple_opt_pass along with...
3309         (pass_data_fre): ...new pass_data instance and...
3310         (make_pass_fre): ...new function.
3311         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3312         subclass of gimple_opt_pass along with...
3313         (pass_data_reassoc): ...new pass_data instance and...
3314         (make_pass_reassoc): ...new function.
3315         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3316         subclass of gimple_opt_pass along with...
3317         (pass_data_sink_code): ...new pass_data instance and...
3318         (make_pass_sink_code): ...new function.
3319         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3320         subclass of gimple_opt_pass along with...
3321         (pass_data_strlen): ...new pass_data instance and...
3322         (make_pass_strlen): ...new function.
3323         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3324         struct to a subclass of gimple_opt_pass along with...
3325         (pass_data_build_alias): ...new pass_data instance and...
3326         (make_pass_build_alias): ...new function.
3327         (pass_build_ealias): Convert from a global struct to a subclass of
3328         gimple_opt_pass along with...
3329         (pass_data_build_ealias): ...new pass_data instance and...
3330         (make_pass_build_ealias): ...new function.
3331         (pass_ipa_pta): Convert from a global struct to a subclass of
3332         simple_ipa_opt_pass along with...
3333         (pass_data_ipa_pta): ...new pass_data instance and...
3334         (make_pass_ipa_pta): ...new function.
3335         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3336         subclass of gimple_opt_pass along with...
3337         (pass_data_uncprop): ...new pass_data instance and...
3338         (make_pass_uncprop): ...new function.
3339         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3340         global struct to a subclass of gimple_opt_pass along with...
3341         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3342         (make_pass_late_warn_uninitialized): ...new function.
3343         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3344         to a subclass of gimple_opt_pass along with...
3345         (pass_data_init_datastructures): ...new pass_data instance and...
3346         (make_pass_init_datastructures): ...new function.
3347         (pass_early_warn_uninitialized): Convert from a global struct to a
3348         subclass of gimple_opt_pass along with...
3349         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3350         (make_pass_early_warn_uninitialized): ...new function.
3351         (pass_update_address_taken): Convert from a global struct to a
3352         subclass of gimple_opt_pass along with...
3353         (pass_data_update_address_taken): ...new pass_data instance and...
3354         (make_pass_update_address_taken): ...new function.
3355         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3356         struct to a subclass of gimple_opt_pass along with...
3357         (pass_data_release_ssa_names): ...new pass_data instance and...
3358         (make_pass_release_ssa_names): ...new function.
3359         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3360         subclass of gimple_opt_pass along with...
3361         (pass_data_stdarg): ...new pass_data instance and...
3362         (make_pass_stdarg): ...new function.
3363         * tree-switch-conversion.c (pass_convert_switch): Convert from a
3364         global struct to a subclass of gimple_opt_pass along with...
3365         (pass_data_convert_switch): ...new pass_data instance and...
3366         (make_pass_convert_switch): ...new function.
3367         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3368         to a subclass of gimple_opt_pass along with...
3369         (pass_data_tail_recursion): ...new pass_data instance and...
3370         (make_pass_tail_recursion): ...new function.
3371         (pass_tail_calls): Convert from a global struct to a subclass of
3372         gimple_opt_pass along with...
3373         (pass_data_tail_calls): ...new pass_data instance and...
3374         (make_pass_tail_calls): ...new function.
3375         * tree-vect-generic.c (pass_lower_vector): Convert from a global
3376         struct to a subclass of gimple_opt_pass along with...
3377         (pass_data_lower_vector): ...new pass_data instance and...
3378         (make_pass_lower_vector): ...new function.
3379         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3380         gimple_opt_pass along with...
3381         (pass_data_lower_vector_ssa): ...new pass_data instance and...
3382         (make_pass_lower_vector_ssa): ...new function.
3383         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3384         to a subclass of gimple_opt_pass along with...
3385         (pass_data_slp_vectorize): ...new pass_data instance and...
3386         (make_pass_slp_vectorize): ...new function.
3387         (pass_ipa_increase_alignment): Convert from a global struct to a
3388         subclass of simple_ipa_opt_pass along with...
3389         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3390         (make_pass_ipa_increase_alignment): ...new function.
3391         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3392         gimple_opt_pass along with...
3393         (pass_data_vrp): ...new pass_data instance and...
3394         (make_pass_vrp): ...new function.
3395         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3396         subclass of simple_ipa_opt_pass along with...
3397         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3398         (make_pass_ipa_free_lang_data): ...new function.
3399         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3400         gimple_opt_pass along with...
3401         (pass_data_tsan): ...new pass_data instance and...
3402         (make_pass_tsan): ...new function.
3403         (pass_tsan_O0): Convert from a global struct to a subclass of
3404         gimple_opt_pass along with...
3405         (pass_data_tsan_O0): ...new pass_data instance and...
3406         (make_pass_tsan_O0): ...new function.
3407         * var-tracking.c (pass_variable_tracking): Convert from a global
3408         struct to a subclass of rtl_opt_pass along with...
3409         (pass_data_variable_tracking): ...new pass_data instance and...
3410         (make_pass_variable_tracking): ...new function.
3411         * web.c (pass_web): Convert from a global struct to a subclass of
3412         rtl_opt_pass along with...
3413         (pass_data_web): ...new pass_data instance and...
3414         (make_pass_web): ...new function.
3415         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3416         declaration with that of...
3417         (make_pass_mode_switch_use): ...new function.
3418         (pass_resolve_sw_modes): Replace declaration with that of...
3419         (make_pass_resolve_sw_modes): ...new function.
3420         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3421         from a global struct to a subclass of rtl_opt_pass along with...
3422         (pass_data_mode_switch_use): ...new pass_data instance and...
3423         (make_pass_mode_switch_use): ...new function.
3424         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3425         from a global struct to a subclass of rtl_opt_pass along with...
3426         (pass_data_resolve_sw_modes): ...new pass_data instance and...
3427         (make_pass_resolve_sw_modes): ...new function.
3428         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3429         struct to a subclass of rtl_opt_pass along with...
3430         (pass_data_insert_vzeroupper): ...new pass_data instance and...
3431         (make_pass_insert_vzeroupper): ...new function.
3432         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3433         global struct to a subclass of rtl_opt_pass along with...
3434         (pass_data_work_around_errata): ...new pass_data instance and...
3435         (make_pass_work_around_errata): ...new function.
3436         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3437         struct to a subclass of rtl_opt_pass along with...
3438         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3439         (make_pass_mips_machine_reorg2): ...new function.
3441 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3443         * passes.c (pass_manager::operator new): New.
3445 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3447         Handwritten part of conversion of passes to C++ classes.
3449         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3450         (toplev.o): Add dep on PASS_MANAGER_H.
3451         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3452         of early local pases to reflect this moving from a global to a
3453         member of gcc::pass_manager.
3454         (cgraph_add_new_function): Likewise.
3455         * lto-cgraph.c (lto_output_node): Update for conversion of
3456         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3457         * passes.c (opt_pass::clone): New.
3458         (opt_pass::gate): New.
3459         (opt_pass::execute): New.
3460         (opt_pass::opt_pass): New.
3461         (pass_manager::execute_early_local_passes): New.
3462         (pass_manager::execute_pass_mode_switching): new.
3463         (finish_optimization_passes): Convert to...
3464         (pass_manager::finish_optimization_passes): ...this.
3465         (finish_optimization_passes): Update for conversion of passes to
3466         C++ classes.
3467         (register_dump_files_1): Use has_gate since we cannot portably
3468         check a vtable entry against NULL.
3469         (dump_one_pass): Likewise.
3470         (ipa_write_summaries_2): Likewise.
3471         (ipa_write_optimization_summaries_1): Likewise.
3472         (ipa_read_summaries_1): Likewise.
3473         (ipa_read_optimization_summaries_1): Likewise.
3474         (execute_ipa_stmt_fixups): Likewise.
3475         (pass_manager::pass_manager): Rewrite pass-creation, invoking
3476         pass-creation functions rather than wiring up globals, and
3477         storing the results in fields of pass_manager generated using
3478         pass-instances.def.
3479         (pass_manager::dump_profile_report): Update for conversion of
3480         passes to C++ classes.
3481         (pass_manager::execute_ipa_summary_passes): Likewise.
3482         (execute_one_ipa_transform_pass): Likewise.
3483         (execute_one_pass): Use has_gate and has_execute since we cannot
3484         portably check a vtable entry against NULL.
3485         * pass_manager.h (pass_manager::finish_optimization_passes): New.
3486         (pass_manager): Use pass-instances.def to add fields for the
3487         various pass instances.
3488         * toplev.c (finalize): Update for move of
3489         finish_optimization_passes to a method of gcc::pass_manager.
3490         * toplev.h (finish_optimization_passes): Move to method of class
3491         pass_manager.
3492         * tree-pass.h (struct pass_data): New.
3493         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3494         (opt_pass::gate): Convert to virtual function.
3495         (opt_pass::~opt_pass): New.
3496         (opt_pass::clone): New.
3497         (opt_pass::execute): Convert to virtual function.
3498         (opt_pass::opt_pass): New.
3499         (opt_pass::ctxt_): new.
3500         (gimple_opt_pass): Convert to subclass of opt_pass.
3501         (gimple_opt_pass::gimple_opt_pass): New.
3502         (rtl_opt_pass): Convert to subclass of opt_pass.
3503         (rtl_opt_pass::rtl_opt_pass): New.
3504         (ipa_opt_pass_d): Convert to subclass of opt_pass.
3505         (ipa_opt_pass_d::ipa_opt_pass_d): New.
3506         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3507         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3508         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3509         invocation of pass_mode_switching to reflect this moving from a
3510         global to a member of gcc::pass_manager.
3511         (ix86_option_override): Rework how pass_insert_vzeroupper is
3512         added to the pass_manager to reflect autogenerated changes.
3513         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3515 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
3517         PR rtl-optimization/57708
3518         * recog.c (peep2_find_free_register): Validate all regs in a
3519         multi-reg mode.
3521 2013-08-05  Jan Hubicka  <jh@suse.cz>
3523         PR lto/57602
3524         * cgraph.c (verify_cgraph_node): Accept local flags from other
3525         partitions.
3526         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3527         (function_and_variable_visibility): Likewise.
3528         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3530 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3532         * graph.c (init_graph_slim_pretty_print): Remove.
3533         (print_graph_cfg): Do not call it.  Use local pretty printer.
3534         (start_graph_dump): Likewise.
3536 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3538         * gimple-pretty-print.c (buffer): Remove.
3539         (initialized): Likewise.
3540         (maybe_init_pretty_print): Likewise.
3541         (print_gimple_stmt): Do not call it.  Use non-static local
3542         pretty_printer variable.
3543         (print_gimple_expr): Likewise.
3544         (print_gimple_seq): Likewise.
3545         (gimple_dump_bb): Likewise.
3547 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3549         * asan.c (asan_pp): Remove.
3550         (asan_pp_initialized): Likewise.
3551         (asan_pp_initialize): Likewise.
3552         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
3553         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
3554         (asan_add_global): Likewise.
3556 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3558         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
3559         * diagnostic.h (diagnostic_flush_buffer): Adjust.
3560         * pretty-print.c (pp_formatted_text_data): Likewise.
3561         (pp_indent): Rename from pp_base_indent.
3562         (pp_format): Rename from pp_base_format.
3563         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3564         (pp_format_verbatim): Rename from pp_base_format_verbatim.
3565         (pp_flush): Rename from pp_base_flush.
3566         (pp_set_line_maximum_length): Rename from
3567         pp_base_set_line_maximum_length.
3568         (pp_clear_output_area): Rename from pp_base_clear_output_area.
3569         (pp_set_prefix): Rename from pp_base_set_prefix.
3570         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3571         (pp_emit_prefix): Rename from pp_base_emit_prefix.
3572         (pp_append_text): Rename from pp_base_append_text.
3573         (pp_formatted_text): Rename from pp_base_formatted_text.
3574         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3575         (pp_remaining_character_count_for_line): Rename from
3576         pp_base_remaining_character_count_for_line.
3577         (pp_newline): Rename from pp_base_newline.
3578         (pp_character): Rename from pp_base_character.
3579         (pp_string): Rename from pp_base_string.
3580         (pp_maybe_space): Rename from pp_base_maybe_space.
3581         * asan.c (asan_pp_string): Adjust.
3582         (asan_emit_stack_protection): Likewise.
3583         (asan_add_global): Likewise.
3584         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3585         * tree-mudflap.c (mf_varname_tree): Likewise.
3586         * tree-pretty-print.c (pp_tree_identifier): Rename from
3587         pp_base_tree_identifier.
3588         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3589         Declare as function.
3591 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3593         * pretty-print.h (pp_bar_bar): New.
3594         (pp_ampersand_ampersand): Likewise.
3595         (pp_less_equal): Likewise.
3596         (pp_greater_equal): Likewise.
3597         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3598         printer functions instead of pp_string or operators and punctuators.
3599         (dump_gimple_call): Likewise.
3600         (dump_gimple_omp_for): Likewise.
3601         (dump_gimple_transaction): Likewise.
3602         (dump_gimple_phi): Likewise.
3603         (pp_gimple_stmt_1): Likewise.
3604         * sched-vis.c (print_insn): Likewise.
3605         * tree-mudflap.c (mf_varname_tree): Likewise.
3606         * tree-pretty-print.c (dump_block_node): Likewise.
3607         (dump_generic_node): Likewise.
3609 2013-08-02  Jan Hubicka  <jh@suse.cz>
3611         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3612         boundaries.
3613         * lto-streamer-out.c (tree_is_indexable): Results decls and
3614         parm decls are not indexable.
3615         (DFS_write_tree_body): Do not follow args and results.
3616         (hash_tree): Likewise.
3617         (output_functions): Rearrange so struct function is needed
3618         only when real body is output; be able to also ouptut abstract
3619         functions; output DECL_ARGUMENTS and DECL_RESULT.
3620         (lto_output): When not in WPA, ale store abstract functions.
3621         (write_symbol): Do not care about RESULT_DECL.
3622         (output_symbol_p): Handle correctly sbtract decls.
3623         * lto-streamer-in.c (input_function): Rearrange so struct
3624         function can be NULL at entry; allow streaming of
3625         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3626         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3627         sanity check during LTO.
3628         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3629         RESULT_DECl and DECL_ARGUMENTS.
3630         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3631         Likewise.
3633 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3635         * pretty-print.h (pp_underscore): New.
3636         (pp_comma): Tidy.
3637         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3638         printer functions instead of pp_character.
3639         (dump_binary_rhs): Likewise.
3640         (dump_ternary_rhs): Likewise.
3641         (dump_gimple_call_args): Likewise.
3642         (pp_points_to_solution): Likewise.
3643         (dump_gimple_call): Likewise.
3644         (dump_gimple_switch): Likewise.
3645         (dump_gimple_cond): Likewise.
3646         (dump_gimple_bind): Likewise.
3647         (dump_gimple_try): Likewise.
3648         (dump_gimple_omp_for): Likewise.
3649         (dump_gimple_omp_continue): Likewise.
3650         (dump_gimple_omp_single): Likewise.
3651         (dump_gimple_omp_sections): Likewise.
3652         (dump_gimple_omp_block): Likewise.
3653         (dump_gimple_omp_critical): Likewise.
3654         (dump_gimple_transaction): Likewise.
3655         (dump_gimple_asm): Likewise.
3656         (dump_gimple_phi): Likewise.
3657         (dump_gimple_omp_parallel): Likewise.
3658         (dump_gimple_omp_task): Likewise.
3659         (dump_gimple_omp_atomic_load): Likewise.
3660         (dump_gimple_omp_atomic_store): Likewise.
3661         (dump_gimple_mem_ops): Likewise.
3662         (pp_gimple_stmt_1): Likewise.
3663         (pp_cfg_jump): Likewise.
3664         (dump_implicit_edges): Likewise.
3665         (gimple_dump_bb_for_graph): Likewise.
3666         * graph.c (draw_cfg_node): Likewise.
3667         * langhooks.c (lhd_print_error_function): Likewise.
3668         * sched-vis.c (print_exp): Likewise.
3669         (print_value): Likewise.
3670         (print_pattern): Likewise.
3671         (print_insn): Likewise.
3672         (rtl_dump_bb_for_graph): Likewise.
3673         * tree-pretty-print.c (dump_function_declaration): Likewise.
3674         (dump_array_domain): Likewise.
3675         (dump_omp_clause): Likewise.
3676         (dump_location): Likewise.
3677         (dump_generic_node): Likewise.
3678         (print_struct_decl): Likewise.
3679         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3681 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3683         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3684         candidate table when replacing a candidate statement.
3685         (replace_rhs_if_not_dup): Likewise.
3686         (replace_one_candidate): Likewise.
3688 2013-08-02  Jan Hubicka  <jh@suse.cz>
3689             Martin Liska  <marxin.liska@gmail.com>
3691         * cgraphunit.c (add_new_function): Fix logic when adding from
3692         late IPA pass.
3693         (assemble_thunk): Rename to ...
3694         (expand_thunk); .. this one; export; get it working with
3695         general functions; make produced gimple valid.
3696         * cgraph.h (expand_thunk): Declare.
3698 2013-08-02  Jan Hubicka  <jh@suse.cz>
3700         * ipa-cp.c (gather_context_independent_values): Use
3701         ipa_get_param_move_cost.
3702         (get_replacement_map): Remove PARAM; move parameter folding
3703         into tree-inline.c
3704         (create_specialized_node): Update.
3705         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3706         assert that we have gimple body; update move_cost.
3707         (count_formal_params): Assert that we have gimple body.
3708         (ipa_dump_param): New function.
3709         (ipa_alloc_node_params): Break out from ...
3710         (ipa_initialize_node_params): ... here.
3711         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3712         (ipa_write_node_info): Stream move costs.
3713         (ipa_read_node_info): Read move costs.
3714         (ipa_update_after_lto_read): Do not recompute node params.
3715         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3716         (ipa_get_param): Check we are not in WPA.
3717         (ipa_get_param_move_cost): New.
3718         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3719         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3720         parm numbers to be present.
3722 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3724         PR rtl-optimization/58048
3725         * lra-constraints.c (process_alt_operands): Don't check asm
3726         operand on register.
3728 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3730         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3731         the implied StoreLoad barrier for atomic operations if before.
3733 2013-08-02  Jan Hubicka  <jh@suse.cz>
3734             Martin Liska  <marxin.liska@gmail.com>
3736         * cgraph.c (cgraph_function_body_availability): Do not check
3737         cgraph flags.
3738         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3739         symtab_node_availability): Declare.
3740         * ipa.c (can_replace_by_local_alias): New.
3741         (function_and_variable_visibility): Use it.
3742         * symtab.c (symtab_for_node_and_aliases,
3743         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3745 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3747         PR rtl-optimization/57963
3748         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3749         (lra_constraints): Use them.
3751 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3753         * config/arm/types.md (define_attr "type"): Add "load_acq"
3754         and "store_rel".
3755         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3756         changes.
3757         (cortex_a53_store1): Likewise.
3759 2013-08-01  Jan Hubicka  <jh@suse.cz>
3761         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3762         partitions are not needed.
3764 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3766         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3767         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3768         MAYBE_NON_Q_CLASS_P where appropriate.
3770 2013-08-01  Jan Hubicka  <jh@suse.cz>
3772         * cgraph.h (release_function_body): Declare.
3773         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3774         of unused delcarations.
3776 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3778         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3779         RTL form when subtracting a constant.
3781 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3783         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3784         Generate canonical plus rtx with negated immediate instead of minus
3785         where appropriate.
3786         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3788 2013-08-01  Jan Hubicka  <jh@suse.cz>
3790         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3791         (cgraph_release_function_body): Likewise.
3792         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3793         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3794         to used_as_abstract_origin.
3795         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3796         symbols used as abstract origins.
3797         * cgraphunit.c (analyze_functions): Update.
3798         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3799         used_as_abstract_origin.
3800         * tree-inline.c (tree_function_versioning): Update
3801         used_as_abstract_origin; be ready for DECL_RESULT and
3802         DECL_ARGUMENTS to be NULL.
3804         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3805         for abstract functions.
3806         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3807         real symbols.
3809 2013-08-01  Jan Hubicka  <jh@suse.cz>
3811         * profile.c (compute_value_histograms): Fix thinko.
3813 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3815         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3816         aarch-common-protos.h to extra_headers.
3817         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3818         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3819         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3821 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3823         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3824         Include "../arm/types.md".  Define "type" attribute for all patterns.
3825         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3826         attribute changes.
3828 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3830         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3831         to support power8 load fusion.
3832         (fusion_gpr_mem_load): Likewise.
3834         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3836         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3837         declarations for power8 load fusion.
3838         (emit_fusion_gpr_load): Likewise.
3840         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3841         tuning for power8, turn on fusion mode by default.  Turn on sign
3842         extending fusion mode if normal fusion mode is on, and we are at
3843         -O2 or -O3.
3844         (fusion_gpr_load_p): New function, return true if we can fuse an
3845         addis instruction with a dependent load to a GPR.
3846         (emit_fusion_gpr_load): Emit the instructions for power8 load
3847         fusion to GPRs.
3849         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3850         (VSX load fusion peepholes): New peepholes to fuse together an
3851         addi instruction with a VSX load instruction.
3853         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3854         peepholes to fuse an addis instruction with a load to a GPR base
3855         register.  If we are supporting sign extending fusions, convert
3856         sign extending loads to zero extending loads and add an explicit
3857         sign extension.
3859 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3861         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3862         aarch-common-protos.h to extra_headers.
3863         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3864         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3865         (arm_early_store_addr_dep): Likewise.
3866         (arm_no_early_alu_shift_dep): Likewise.
3867         (arm_no_early_alu_shift_value_dep): Likewise.
3868         (arm_no_early_mul_dep): Likewise.
3869         (arm_no_early_store_addr_dep): Likewise.
3870         (arm_mac_accumulator_is_mul_result): Likewise.
3871         (arm_mac_accumulator_is_result): Likewise.
3872         * config/arm/aarch-common.c: ... here.  New file.
3873         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3874         here to ...
3875         (arm_early_store_addr_dep): Likewise.
3876         (arm_no_early_alu_shift_dep): Likewise.
3877         (arm_no_early_alu_shift_value_dep): Likewise.
3878         (arm_no_early_mul_dep): Likewise.
3879         (arm_no_early_store_addr_dep): Likewise.
3880         (arm_mac_accumulator_is_mul_result): Likewise.
3881         (arm_mac_accumulator_is_result): Likewise.
3882         * config/arm/aarch-common-protos.h: ... here.  New file.
3883         * config/arm/t-arm (aarch-common.o): Define.
3885 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3887         * config/arm/arm.md: Include new file "types.md".
3888         (define_attr "type"): Move from here to ...
3889         (define_attr "mul32"): Likewise.
3890         (define_attr "mul64"): Likewise.
3891         * config/arm/types.md: ... here.  New file.
3893 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3895         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3896         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3898 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3900         * gen-pass-instances.awk: Fix offset of substr().
3902 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3904         * Makefile.in (pass-instances.def): New.
3905         (passes.o): Replace dependency on passes.def with one on
3906         pass-instances.def
3908         * gen-pass-instances.awk: New.
3910         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3911         rather than passes.def, updating local definition of NEXT_PASS
3912         macro to add an extra NUM parameter (currently unused).
3914 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3916         * Makefile.in (PASS_MANAGER_H): New.
3917         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3918         (passes.o): Likewise.
3919         (statistics.o): Likewise.
3920         (cgraphunit.o): Likewise.
3921         (context.o): Depend on PASS_MANAGER_H.
3923         * pass_manager.h: New.
3925         * cgraphunit.c (cgraph_add_new_function): Update for moves
3926         of globals to fields of pass_manager.
3927         (analyze_function): Likewise.
3928         (expand_function): Likewise.
3929         (ipa_passes): Likewise.
3930         (compile): Likewise.
3932         * context.c (context::context): New.
3933         * context.h  (context::context): New.
3934         (context::get_passes): New.
3935         (context::passes_): New.
3937         * lto-cgraph.c (input_node): Update for moves of globals to
3938         fields of pass_manager.
3940         * passes.c (all_passes): Remove, in favor of a field of the
3941         same name within the new class pass_manager.
3942         (all_small_ipa_passes): Likewise.
3943         (all_lowering_passes): Likewise.
3944         (all_regular_ipa_passes): Likewise.
3945         (all_late_ipa_passes): Likewise.
3946         (all_lto_gen_passes): Likewise.
3947         (passes_by_id): Likewise.
3948         (passes_by_id_size): Likewise.
3949         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3950         the new class pass_manager.
3951         (set_pass_for_id): Convert to...
3952         (pass_manager::set_pass_for_id): ...method.
3953         (get_pass_for_id): Convert to...
3954         (pass_manager::get_pass_for_id): ...method.
3955         (register_one_dump_file): Move body of implementation into...
3956         (pass_manager::register_one_dump_file): ...here.
3957         (register_dump_files_1): Convert to...
3958         (pass_manager::register_dump_files_1): ...method.
3959         (register_dump_files): Convert to...
3960         (pass_manager::register_dump_files): ...method.
3961         (create_pass_tab): Update for moves of globals to fields of
3962         pass_manager.
3963         (dump_passes): Move body of implementation into...
3964         (pass_manager::dump_passes): ...here.
3965         (register_pass): Move body of implementation into...
3966         (pass_manager::register_pass): ...here.
3967         (init_optimization_passes): Convert into...
3968         (pass_manager::pass_manager): ...constructor for new
3969         pass_manager class, and initialize the pass_lists array.
3970         (check_profile_consistency): Update for moves of globals to
3971         fields of pass_manager.
3972         (dump_profile_report): Move body of implementation into...
3973         (pass_manager::dump_profile_report): ...here.
3974         (ipa_write_summaries_1): Update for moves of pass lists from
3975         being globals to fields of pass_manager.
3976         (ipa_write_optimization_summaries): Likewise.
3977         (ipa_read_summaries):  Likewise.
3978         (ipa_read_optimization_summaries): Likewise.
3979         (execute_all_ipa_stmt_fixups): Likewise.
3981         * statistics.c (statistics_fini): Update for moves of globals to
3982         fields of pass_manager.
3984         * toplev.c (general_init): Replace call to
3985         init_optimization_passes with construction of the pass_manager
3986         instance.
3988         * tree-pass.h (all_passes): Remove, in favor of a field of the
3989         same name within the new class pass_manager.
3990         (all_small_ipa_passes): Likewise.
3991         (all_lowering_passes): Likewise.
3992         (all_regular_ipa_passes): Likewise.
3993         (all_lto_gen_passes): Likewise.
3994         (all_late_ipa_passes): Likewise.
3995         (passes_by_id): Likewise.
3996         (passes_by_id_size): Likewise.
3997         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3998         the new class pass_manager.
3999         (get_pass_for_id): Remove.
4001 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4003         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4004         configs.
4006 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4008         * arm.md (mulhi3): New expand pattern.
4010 2013-07-30  Jan Hubicka  <jh@suse.cz>
4011             Martin Liska  <marxin.liska@gmail.com>
4013         * profile.c (compute_value_histograms): Do not ICE when
4014         there is mismatch only on some counters.
4016 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4018         PR rtl-optimization/57637
4019         * function.c (move_insn_for_shrink_wrap): Also check the
4020         GEN set of the LIVE problem for the liveness analysis
4021         if it exists, otherwise give up.
4023 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4025         PR tree-optimization/57993
4026         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4027         replaced statement in the candidate table.
4028         (phi_add_costs): Return infinite cost when the hidden basis does
4029         not dominate all phis on which the candidate is dependent.
4030         (replace_one_candidate): Record replaced statement in the
4031         candidate table.
4033 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
4035         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4036         (ashlv2si3): New expander.
4037         (*ashlv2si3_i): New define_insn_and_split.
4038         * predicates.md (float_operation): Allow patterns with three
4039         basic sub-patterns.
4041         PR rtl-optimization/58021
4042         * mode-switching.c (create_pre_exit): Always split off preceding
4043         insns if we are not at the basic block head.
4045 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
4047         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4048         (UCLIBC_DYNAMIC_LINKER): New macro.
4049         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4050         `-mnan=2008'.
4051         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4052         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
4053         `-mnan=2008'.
4054         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4055         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4056         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4057         for SF and DF modes.  Use ieee_quad_format for TF mode.
4058         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4059         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4060         (mips_option_override): Handle `-mnan=legacy'.
4061         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4062         `-mabs=2008' and `-mnan=2008'.
4063         (OPTION_DEFAULT_SPECS): Add "nan" default.
4064         (ASM_SPEC): Handle `-mnan='.
4065         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4066         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4067         comment accordingly.
4068         (neg<mode>2): Likewise.
4069         * config/mips/mips.opt (mabs, mnan): New options.
4070         * doc/install.texi (Configuration): Document `--with-nan=' option.
4071         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4072         `-mnan=' options.
4073         (MIPS Options): Document them.
4074         * config.gcc <mips*-*-*>: Handle `--with-nan='.
4075         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4076         * configure: Regenerate.
4077         * config.in: Regenerate.
4079 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4081         * config/i386/i386.md (float post-reload splitters): Do not check
4082         for subregs of SSE registers.
4084 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4085             H.J. Lu  <hongjiu.lu@intel.com>
4087         PR target/57954
4088         PR target/57988
4089         * config/i386/i386.md (post-reload splitter
4090         to avoid partial SSE reg dependency stalls): New pattern.
4092 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4094         * config/s390/s390.md ("movcc"): Swap load and store instructions.
4096 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
4098         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4099         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4101 2013-07-26  Cary Coutant  <ccoutant@google.com>
4103         * dwarf2out.c (die_checksum_ordered): Don't include template
4104         instantiations in signature.
4105         (is_template_parameter): New function.
4106         (is_template_instantiation): New function.
4107         (generate_skeleton_bottom_up): Don't include template instantiations
4108         in type unit DIE.
4109         (generate_skeleton): Likewise.
4110         (break_out_comdat_types): Move recursive call to break out nested
4111         types earlier.
4112         (prune_unused_types_mark_generic_parms_dies): Call
4113         is_template_parameter.
4115 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
4117         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4118         uses vector registers.
4119         * config/aarch64/iterators.md: Add attributes rtn and vas.
4121 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4122             Richard Earnshaw  <richard.earnshaw@arm.com>
4124         * combine.c (simplify_comparison): Re-canonicalize operands
4125         where appropriate.
4126         * config/arm/arm.md (movcond_addsi): New splitter.
4128 2013-07-25  Sterling Augustine  <saugustine@google.com>
4130         * dwarf2out.c (size_of_pubnames): Move code to...
4131         (include_pubname_in_output): ...here.  New.
4132         (want_pubnames): Rearrange.
4133         (output_pubnames): Call include_pubname_in_output.  Move assertion.
4135 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4137         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4139 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4141         PR target/38836
4142         * doc/extend.texi: Remove obsolete builtins. Fix
4143         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4145 2013-07-25  Jan Hubicka  <jh@suse.cz>
4147         * cgraph.c (release_function_body): Break out from ...
4148         (cgraph_release_function_body): ... this one; also release DECL_RESULT
4149         and DECL_ARGUMENTS.
4150         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4151         old_tree in the map.
4152         (create_specialized_node): Update.
4153         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4154         into index.
4155         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4156         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4157         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4158         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4159         DECL_RESULT.
4161 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4163         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4164         addsi3_carryin_alt2_<optab>): Correct output template.
4166 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4168         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4169         Adjust for arm_restrict_it.
4170         Remove trailing whitespace.
4172 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
4174         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4175         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4177         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4179 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
4181         PR rtl-optimization/57960
4182         * lra-constraints.c (process_alt_operands): Use the right mode
4183         when checking strict_low.
4185 2013-07-25  Jan Hubicka  <jh@suse.cz>
4187         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4188         * cgraph.c (cgraph_remove_node): Do not release function body
4189         when in cgraph streaming.
4190         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4191         in other partitions are not considered reachable; fix handling of
4192         clones.
4194 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4196         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4198 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4200         PR target/19599
4201         PR target/57731
4202         PR target/57837
4203         * config/arm/arm.md ("*sibcall_insn): Replace use of
4204         Ss with US. Adjust output for v5 and v4t.
4205         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4207         * config/arm/constraints.md ("Ss"): Rename to US.
4209 2013-07-25  Terry Guo  <terry.guo@arm.com>
4211         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4212         shift_add/shift_sub0/shift_sub1 RTXs.
4214 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
4215             Anton Blanchard  <anton@au1.ibm.com>
4217         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4218         (altivec_vpks<VI_char>ss): Likewise.
4219         (altivec_vpks<VI_char>us): Likewise.
4220         (altivec_vpku<VI_char>us): Likewise.
4221         (altivec_vpku<VI_char>um): Likewise.
4223 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
4225         Introduce context class.
4227         * Makefile.in (CONTEXT_H): New.
4228         (OBJS): Add context.o.
4229         (toplev.o): Add CONTEXT_H to dependencies.
4230         (context.o): New.
4232         * toplev.c (general_init): Create the singleton gcc::context instance.
4234         * context.c: New.
4236         * context.h: New.
4238 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
4240         PR rtl-optimization/57968
4241         * mode-switching.c (create_pre_exit): Allow instructions that
4242         don't set a return register to need a non-exit mode.
4244 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4245             Anton Blanchard  <anton@au1.ibm.com>
4247         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4248         operands to vperm for little endian.
4249         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4250         of lvsl to create the control mask for a vperm for little endian.
4252 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4253             Anton Blanchard  <anton@au1.ibm.com>
4255         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4256         two operands for little-endian.
4258 2013-07-23  Steve Ellcey  <sellcey@mips.com>
4260         * config/mips/mips.c (mips_case_values_threshold): New.
4261         (TARGET_CASE_VALUES_THRESHOLD): Define.
4263 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4264             Anton Blanchard  <anton@au1.ibm.com>
4266         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4267         selection of field for vector splat in little endian mode.
4269 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4271         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4272         expanders to rs6000.md.
4273         (ior<mode>3): Likewise.
4274         (and<mode>3): Likewise.
4275         (one_cmpl<mode>2): Likewise.
4276         (nor<mode>3): Likewise.
4277         (andc<mode>3): Likewise.
4278         (eqv<mode>3): Likewise.
4279         (nand<mode>3): Likewise.
4280         (orc<mode>3): Likewise.
4282         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4283         declaration.
4285         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4286         to split multi-word logical operations.
4287         (rs6000_split_logical_di): Likewise.
4288         (rs6000_split_logical): Likewise.
4290         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4291         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4292         and allow TImode operations in 32-bit.
4293         (vsx_and<mode>3_64bit): Likewise.
4294         (vsx_ior<mode>3_32bit): Likewise.
4295         (vsx_ior<mode>3_64bit): Likewise.
4296         (vsx_xor<mode>3_32bit): Likewise.
4297         (vsx_xor<mode>3_64bit): Likewise.
4298         (vsx_one_cmpl<mode>2_32bit): Likewise.
4299         (vsx_one_cmpl<mode>2_64bit): Likewise.
4300         (vsx_nor<mode>3_32bit): Likewise.
4301         (vsx_nor<mode>3_64bit): Likewise.
4302         (vsx_andc<mode>3_32bit): Likewise.
4303         (vsx_andc<mode>3_64bit): Likewise.
4304         (vsx_eqv<mode>3_32bit): Likewise.
4305         (vsx_eqv<mode>3_64bit): Likewise.
4306         (vsx_nand<mode>3_32bit): Likewise.
4307         (vsx_nand<mode>3_64bit): Likewise.
4308         (vsx_orc<mode>3_32bit): Likewise.
4309         (vsx_orc<mode>3_64bit): Likewise.
4311         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4312         logical types in GPRs.
4314         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4315         logical insns to rs6000.md, and allow TImode operations in
4316         32-bit.
4317         (altivec_ior<mode>3): Likewise.
4318         (altivec_xor<mode>3): Likewise.
4319         (altivec_one_cmpl<mode>2): Likewise.
4320         (altivec_nor<mode>3): Likewise.
4321         (altivec_andc<mode>3): Likewise.
4323         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4324         attributes for moving the 128-bit logical operations into
4325         rs6000.md.
4326         (BOOL_REGS_OUTPUT): Likewise.
4327         (BOOL_REGS_OP1): Likewise.
4328         (BOOL_REGS_OP2): Likewise.
4329         (BOOL_REGS_UNARY): Likewise.
4330         (BOOL_REGS_AND_CR0): Likewise.
4331         (one_cmpl<mode>2): Add support for DI logical operations on
4332         32-bit, splitting the operations to 32-bit.
4333         (anddi3): Likewise.
4334         (iordi3): Likewise.
4335         (xordi3): Likewise.
4336         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4337         changes to combine the 32/64-bit code, allow logical operations on
4338         TI mode in 32-bit, and to use similar match_operator patterns like
4339         scalar mode uses.  Combine the Altivec and VSX code for logical
4340         operations, and move it here.
4341         (ior<mode>3, 128-bit types): Likewise.
4342         (xor<mode>3, 128-bit types): Likewise.
4343         (one_cmpl<mode>3, 128-bit types): Likewise.
4344         (nor<mode>3, 128-bit types): Likewise.
4345         (andc<mode>3, 128-bit types): Likewise.
4346         (eqv<mode>3, 128-bit types): Likewise.
4347         (nand<mode>3, 128-bit types): Likewise.
4348         (orc<mode>3, 128-bit types): Likewise.
4349         (and<mode>3_internal): Likewise.
4350         (bool<mode>3_internal): Likewise.
4351         (boolc<mode>3_internal1): Likewise.
4352         (boolc<mode>3_internal2): Likewise.
4353         (boolcc<mode>3_internal1): Likewise.
4354         (boolcc<mode>3_internal2): Likewise.
4355         (eqv<mode>3_internal1): Likewise.
4356         (eqv<mode>3_internal2): Likewise.
4357         (one_cmpl1<mode>3_internal): Likewise.
4359 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4361         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4362         Rename flag_stack_usage to flag_stack_usage_info.
4364 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4366         * config/microblaze/sync.md: New file.
4367         * config/microblaze/microblaze.md: Include sync.md
4368         * config/microblaze/microblaze.c: Add print_operand 'y'.
4369         * config/microblaze/constraints.md: Add memory_contraint
4370         'Q' which is a single register.
4372 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
4374         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4376 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4378         * reload.c (find_reloads): Exit loop once we find this operand
4379         cannot be reloaded somehow for this alternative.
4381         * reload.c (find_reloads): Exit loop once we find a hard register.
4383         * rtlanal.c (computed_jump_p): Exit loop once we find label
4384         reference is used.
4386         * i386.c (ix86_pad_returns): Exit loop after setting replace.
4388         * cfgloopmanip.c (remove_path): Exit loop after setting
4389         irred_invalidated.
4391         * gensupport.c (subst_dup): Avoid loop if code is not
4392         MATCH_DUP nor MATCH_OP_DUP.
4394 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
4396         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4398 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4400         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4401         true for SP_REGNUM if mode == ptr_mode.
4402         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4403         with value R0_REGNUM + 31.
4405 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4407         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4408         pad pointer-typed argument downward.
4410 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4412         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4413         and __ILP32__ when the ILP32 model is in use.
4415 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4417         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4418         (aarch64_load_symref_appropriately): In the case of
4419         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4420         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4421         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4422         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4423         if 'mode' doesn't equal to 'ptr_mode'.
4424         (aarch64_output_mi_thunk): Add an assertion on the alignment of
4425         'vcall_offset'; change to call aarch64_emit_move differently depending
4426         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4427         to calculate the upper bound of 'vcall_offset'.
4428         (aarch64_cannot_force_const_mem): Change to also return true if
4429         mode != ptr_mode.
4430         (aarch64_legitimize_reload_address): In the case of large
4431         displacements, add new local variable 'xmode' and an assertion
4432         based on it; change to use 'xmode' to generate the new rtx and
4433         reload.
4434         (aarch64_asm_trampoline_template): Change to generate the template
4435         differently depending on TARGET_ILP32 or not; change to use
4436         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4437         (aarch64_trampoline_size): Removed.
4438         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4439         and replace immediate literals with it.  Change to use 'ptr_mode'
4440         instead of 'DImode' and call convert_memory_address if the mode
4441         of 'fnaddr' doesn't equal to 'ptr_mode'.
4442         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4443         to output symbol.
4444         (aarch64_elf_asm_destructor): Likewise.
4445         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4446         on TARGET_ILP32 instead of aarch64_trampoline_size.
4447         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4448         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4449         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4450         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
4451         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4452         (storewb_pair<GPI:mode>_<P:mode>): ... this.
4453         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4454         depending on the value of 'mode'.
4455         (add_losym_<mode>): New.
4456         (ldr_got_small_<mode>): New, based on ldr_got_small.
4457         (ldr_got_small): Remove.
4458         (ldr_got_small_sidi): New.
4459         * config/aarch64/iterators.md (P): New.
4460         (PTR): Change to 'ptr_mode' in the condition.
4462 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4464         * config.gcc (aarch64*-*-*): Support --with-abi.
4465         (aarch64*-*-elf): Support --with-multilib-list.
4466         (aarch64*-*-linux*): Likewise.
4467         (supported_defaults): Add abi to aarch64*-*-*.
4468         * configure.ac: Mention AArch64 for --with-multilib-list.
4469         * configure: Re-generated.
4470         * config/aarch64/biarchilp32.h: New file.
4471         * config/aarch64/biarchlp64.h: New file.
4472         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4473         (ABI_SPEC): Ditto.
4474         (MULTILIB_DEFAULTS): Ditto.
4475         (DRIVER_SELF_SPECS): Ditto.
4476         (ASM_SPEC): Update to also substitute -mabi.
4477         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4478         file whose name depends on -mabi= and -mbig-endian.
4479         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4480         TARGET_ILP32.
4481         (POINTER_SIZE): New define.
4482         (POINTERS_EXTEND_UNSIGNED): Ditto.
4483         (enum aarch64_abi_type): New enumeration tag.
4484         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4485         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4486         (TARGET_ILP32): New define.
4487         * config/aarch64/aarch64.opt (mabi): New.
4488         (aarch64_abi): New.
4489         (ilp32, lp64): New values for -mabi.
4490         * config/aarch64/t-aarch64 (comma): New define.
4491         (MULTILIB_OPTIONS): Ditto.
4492         (MULTILIB_DIRNAMES): Ditto.
4493         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4494         * doc/invoke.texi: Document -mabi for AArch64.
4496 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
4498         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4500 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4501             Anton Blanchard  <anton@au1.ibm.com>
4503         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4504         endianness when selecting field to splat.
4506 2013-07-22  Eric Christopher  <echristo@gmail.com>
4508         * dwarf2out.c (die_odr_checksum): New function to use
4509         CHECKSUM_ macros and ULEB128 for DIE tag.
4510         (generate_type_signature): Use.
4512 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
4514         * config.gcc (sparc*-*-*): Accept leon3 processor.
4515         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4516         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4517         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4518         * config/sparc/sparc.opt (enum processor_type): Add leon3.
4519         (mfix-ut699): Adjust comment.
4520         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4521         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4522         (CPP_CPU_SPEC): Likewise.
4523         (ASM_CPU_SPEC): Likewise.
4524         * config/sparc/sparc.c (leon3_cost): New constant.
4525         (sparc_option_override): Add leon3 support.
4526         (mem_ref): New function.
4527         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4528         (sparc_do_work_around_errata): Look into the instruction in the delay
4529         slot and adjust accordingly.  Add fix for the data cache nullify issues
4530         of the UT699.  Change insertion position for the NOP.
4531         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4532         (leon3_load): New reservation.
4533         (leon_store): Bump latency to 2.
4534         (grfpu): New automaton.
4535         (grfpu_alu): New unit.
4536         (grfpu_ds): Likewise.
4537         (leon_fp_alu): Adjust.
4538         (leon_fp_mult): Delete.
4539         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4540         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4541         * config/sparc/sparc.md (cpu): Add leon3.
4542         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4543         (swapsi): Likewise.
4544         (atomic_test_and_set): Likewise.
4545         (ldstub): Likewise.
4547 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
4549         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
4550         default for R5900 targets.
4551         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4552         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4553         * config/mips/mips.c (mips_option_override): Report an error for
4554         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
4555         for -march=r5900 -mhard-float.
4557 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4559         * df-problems.c (can_move_insns_across): Exit loop once we
4560         find a non-fixed, non-global register.
4562         * ipa-pure-const.c (propagate_nothrow): Exit loop after
4563         setting can_throw.
4565         * omega.c (omega_eliminate_red): Break after setting red_found.
4566         (omega_problem_has_red_equations): Similarly after setting found.
4567         (omega_query_variable): Similarly after setting coupled.
4569 2013-07-22  Marek Polacek  <polacek@redhat.com>
4571         * gimplify.c: Don't include gimple.h twice.
4573 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4575         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4576         instead of TARGET_THUMB1.
4577         (Pz): New constraint.
4578         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4579         encodings.
4580         (compare_negsi_si): Likewise.
4581         (compare_addsi2_op0): Likewise.
4582         (compare_addsi2_op1): Likewise.
4583         (addsi3_carryin_<optab>): Likewise.
4584         (addsi3_carryin_alt2_<optab>): Likewise.
4585         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4586         for arm_restrict_it.
4587         (subsi3_carryin): Likewise.
4588         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4589         (minmax_arithsi): Disable for arm_restrict_it.
4590         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4591         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4592         (satsi_<SAT:code>_shift): Likewise.
4593         (arm_shiftsi3): Add alternative for 16-bit encoding.
4594         (arm32_movhf): Disable for arm_restrict_it.
4595         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4596         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4598 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
4600         * config/arm/arm.md (attribute "insn"): Delete.
4601         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4602         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4603         (not_shiftsi): Update for attribute change.
4604         (not_shiftsi_compare0): Likewise.
4605         (not_shiftsi_compare0_scratch): Likewise.
4606         (arm_one_cmplsi2): Likewise.
4607         (thumb1_one_cmplsi2): Likewise.
4608         (notsi_compare0): Likewise.
4609         (notsi_compare0_scratch): Likewise.
4610         (thumb1_movdi_insn): Likewise.
4611         (arm_movsi_insn): Likewise.
4612         (movhi_insn_arch4): Likewise.
4613         (movhi_bytes): Likewise.
4614         (arm_movqi_insn): Likewise.
4615         (thumb1_movqi_insn): Likewise.
4616         (arm32_movhf): Likewise.
4617         (thumb1_movhf): Likewise.
4618         (arm_movsf_soft_insn): Likewise.
4619         (thumb1_movsf_insn): Likewise.
4620         (thumb_movdf_insn): Likewise.
4621         (movsicc_insn): Likewise.
4622         (movsfcc_soft_insn): Likewise.
4623         (and_scc): Likewise.
4624         (cond_move): Likewise.
4625         (if_move_not): Likewise.
4626         (if_not_move): Likewise.
4627         (if_shift_move): Likewise.
4628         (if_move_shift): Likewise.
4629         (if_shift_shift): Likewise.
4630         (if_not_arith): Likewise.
4631         (if_arith_not): Likewise.
4632         (cond_move_not): Likewise.
4633         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4634         (neon_mov<mode>): Likewise.
4635         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4636         (thumb2_movsi_vfp): Likewise.
4637         (movsf_vfp): Likewise.
4638         (thumb2_movsf_vfp): Likewise.
4639         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4640         change.
4641         (cortexa7_older_only): Likewise.
4642         (cortexa7_younger): Likewise.
4643         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4644         (1020alu_shift_op): Likewise.
4645         (1020alu_shift_reg_op): Likewise.
4646         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4647         (alu_shift_op): Likewise.
4648         (alu_shift_reg_op): Likewise.
4649         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4650         (11_alu_shift_op): Likewise.
4651         (11_alu_shift_reg_op): Likewise.
4652         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4653         (9_alu_shift_reg_op): Likewise.
4654         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4655         change.
4656         (cortex_a15_alu_shift): Likewise.
4657         (cortex_a15_alu_shift_reg): Likewise.
4658         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4659         (cortex_a5_alu_shift): Likewise.
4660         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4661         change.
4662         (cortex_a53_alu_shift): Likewise.
4663         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4664         change.
4665         (cortex_a7_alu_reg): Likewise.
4666         (cortex_a7_alu_shift): Likewise.
4667         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4668         (cortex_a8_alu_shift): Likewise.
4669         (cortex_a8_alu_shift_reg): Likewise.
4670         (cortex_a8_mov): Likewise.
4671         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4672         (cortex_a9_dp_shift): Likewise.
4673         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4674         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4675         (cortex_r4_mov): Likewise.
4676         (cortex_r4_alu_shift): Likewise.
4677         (cortex_r4_alu_shift_reg): Likewise.
4678         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4679         (526_alu_shift_op): Likewise.
4680         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4681         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4682         (626te_alu_shift_op): Likewise.
4683         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4684         (726te_alu_op): Likewise.
4685         (726te_alu_shift_op): Likewise.
4686         (726te_alu_shift_reg_op): Likewise.
4687         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4688         (mp626_alu_shift_op): Likewise.
4689         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4690         (pj4_alu_e1_conds): Likewise.
4691         (pj4_alu): Likewise.
4692         (pj4_alu_conds): Likewise.
4693         (pj4_shift): Likewise.
4694         (pj4_shift_conds): Likewise.
4695         (pj4_alu_shift): Likewise.
4696         (pj4_alu_shift_conds): Likewise.
4698 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4700         * config/arm/predicates.md (shiftable_operator_strict_it):
4701         New predicate.
4702         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4703         Disable cond_exec version for arm_restrict_it.
4704         (thumb2_smaxsi3): Convert to generate cond_exec.
4705         (thumb2_sminsi3): Likewise.
4706         (thumb32_umaxsi3): Likewise.
4707         (thumb2_uminsi3): Likewise.
4708         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4709         (thumb2_neg_abssi2): Likewise.
4710         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4711         (thumb2_movsicc_insn): Adjust alternatives.
4712         (thumb2_mov_negscc): Disable for arm_restrict_it.
4713         (thumb2_mov_negscc_strict_it): New pattern.
4714         (thumb2_mov_notscc_strict_it): New pattern.
4715         (thumb2_mov_notscc): Disable for arm_restrict_it.
4716         (thumb2_ior_scc): Likewise.
4717         (thumb2_ior_scc_strict_it): New pattern.
4718         (thumb2_cond_move): Adjust for arm_restrict_it.
4719         (thumb2_cond_arith): Disable for arm_restrict_it.
4720         (thumb2_cond_arith_strict_it): New pattern.
4721         (thumb2_cond_sub): Adjust for arm_restrict_it.
4722         (thumb2_movcond): Likewise.
4723         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4724         (thumb2_zero_extendhisi2_v6): Likewise.
4725         (thumb2_zero_extendqisi2_v6): Likewise.
4726         (orsi_notsi_si): Likewise.
4727         (orsi_not_shiftsi_si): Likewise.
4729 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4731         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4732         instruction sequence is 1 byte shorter.
4734 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4736         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4737         it is not needed after split.
4739 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4741         PR target/51784
4742         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4743         second label for nonlocal goto receivers. Don't output pic base labels
4744         unless we're producing PIC; mark that action unreachable().
4745         (ix86_save_reg): If the function contains a nonlocal label, save the
4746         PIC base reg.
4747         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4748         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4749         (update_pic_label_number_if_needed): New.
4750         (machopic_output_function_base_name): Adjust for nonlocal receiver
4751         case.
4752         (machopic_should_output_picbase_label): New.
4753         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4754         (nonlocal_goto_receiver): New insn and split.
4756 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4758         * config/aarch64/aarch64-builtins.c
4759         (aarch64_fold_builtin): Fold abs in all modes.
4760         * config/aarch64/aarch64-simd-builtins.def
4761         (abs): Enable for all modes.
4762         * config/aarch64/arm_neon.h
4763         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4764         (vabs_f64): Add missing intrinsic.
4766 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4768         * config/aarch64/arm_neon.h (vabs_s64): New function
4770 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4772         PR target/57516
4773         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4774         * config/avr/avr.md (adjust_len): Add `round'.
4775         * config/avr/avr-protos.h (avr_out_round): New prototype.
4776         (avr_out_plus): Add `out_label' argument.
4777         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4778         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4779         Handle the case where `insn' is just a pattern.
4780         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4781         (avr_out_round): New function.
4782         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4784 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
4786         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4787         Add check for flag_stack_usage to handle -fstack-usage support
4789 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
4791         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4792         interaction for new Power8 flags and VSX.
4794 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4796         PR middle-end/57698
4797         * tree-inline.c (expand_call_inline): Emit errors during
4798         early_inlining only if optimization is not turned on.
4800 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4802         * passes.def: New.
4804         * passes.c (init_optimization_passes): Move the construction of
4805         the pass hierarchy into a new passes.def file.
4807         * Makefile.in (passes.o): Add dependency on passes.def.
4809 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4811         * passes.c (init_optimization_passes): Introduce macros for
4812         constructing the tree of passes (INSERT_PASSES_AFTER,
4813         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4814         TERMINATE_PASS_LIST).
4816 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4817             Wei Mi  <wmi@google.com>
4819         PR rtl-optimization/57878
4820         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4821         top.
4822         (reload_pseudo_compare_func): Check nregs first for reload
4823         pseudos.
4825 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4827         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4829 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4831         * read-rtl.c (validate_const_int): Once an invalid character is
4832         seen, quit the loop.
4834         * gengtype.c (write_roots): Similarly once we find the "deletable"
4835         or "if_marked" option.
4837 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4839         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4840         "xtab" and "sat".  Move value "clz" from here to ...
4841         (attriubte "type"): ... here.
4842         (satsi_<SAT:code>): Delete "insn" attribute.
4843         (satsi_<SAT:code>_shift): Likewise.
4844         (arm_zero_extendqisi2addsi): Likewise.
4845         (arm_extendqisi2addsi): Likewise.
4846         (clzsi2): Update for attribute changes.
4847         (rbitsi2): Likewise.
4848         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4849         attribute.
4850         (arm_usatsihi): Likewise.
4851         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4853 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4855         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4856         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4857         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4858         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4859         in alphabetical order.
4860         (attribute "core_cycles"): Update for attribute changes.
4861         (arm_addsi3): Likewise.
4862         (addsi3_compare0): Likewise.
4863         (addsi3_compare0_scratch): Likewise.
4864         (addsi3_compare_op1): Likewise.
4865         (addsi3_compare_op2): Likewise.
4866         (compare_addsi2_op0): Likewise.
4867         (compare_addsi2_op1): Likewise.
4868         (addsi3_carryin_shift_<optab>): Likewise.
4869         (subsi3_carryin_shift): Likewise.
4870         (rsbsi3_carryin_shift): Likewise.
4871         (arm_subsi3_insn): Likewise.
4872         (subsi3_compare0): Likewise.
4873         (subsi3_compare): Likewise.
4874         (arm_andsi3_insn): Likewise.
4875         (thumb1_andsi3_insn): Likewise.
4876         (andsi3_compare0): Likewise.
4877         (andsi3_compare0_scratch): Likewise.
4878         (zeroextractsi_compare0_scratch
4879         (andsi_not_shiftsi_si): Likewise.
4880         (iorsi3_insn): Likewise.
4881         (iorsi3_compare0): Likewise.
4882         (iorsi3_compare0_scratch): Likewise.
4883         (arm_xorsi3): Likewise.
4884         (thumb1_xorsi3_insn): Likewise.
4885         (xorsi3_compare0): Likewise.
4886         (xorsi3_compare0_scratch): Likewise.
4887         (satsi_<SAT:code>_shift): Likewise.
4888         (rrx): Likewise.
4889         (arm_shiftsi3): Likewise.
4890         (shiftsi3_compare0): Likewise.
4891         (not_shiftsi): Likewise.
4892         (not_shiftsi_compare0): Likewise.
4893         (not_shiftsi_compare0_scratch): Likewise.
4894         (arm_one_cmplsi2): Likewise.
4895         (thumb_one_complsi2): Likewise.
4896         (notsi_compare0): Likewise.
4897         (notsi_compare0_scratch): Likewise.
4898         (thumb1_zero_extendhisi2): Likewise.
4899         (arm_zero_extendhisi2): Likewise.
4900         (arm_zero_extendhisi2_v6): Likewise.
4901         (arm_zero_extendhisi2addsi): Likewise.
4902         (thumb1_zero_extendqisi2): Likewise.
4903         (thumb1_zero_extendqisi2_v6): Likewise.
4904         (arm_zero_extendqisi2): Likewise.
4905         (arm_zero_extendqisi2_v6): Likewise.
4906         (arm_zero_extendqisi2addsi): Likewise.
4907         (thumb1_extendhisi2): Likewise.
4908         (arm_extendhisi2): Likewise.
4909         (arm_extendhisi2_v6): Likewise.
4910         (arm_extendqisi): Likewise.
4911         (arm_extendqisi_v6): Likewise.
4912         (arm_extendqisi2addsi): Likewise.
4913         (thumb1_extendqisi2): Likewise.
4914         (thumb1_movdi_insn): Likewise.
4915         (arm_movsi_insn): Likewise.
4916         (movsi_compare0): Likewise.
4917         (movhi_insn_arch4): Likewise.
4918         (movhi_bytes): Likewise.
4919         (arm_movqi_insn): Likewise.
4920         (thumb1_movqi_insn): Likewise.
4921         (arm32_movhf): Likewise.
4922         (thumb1_movhf): Likewise.
4923         (arm_movsf_soft_insn): Likewise.
4924         (thumb1_movsf_insn): Likewise.
4925         (movdf_soft_insn): Likewise.
4926         (thumb_movdf_insn): Likewise.
4927         (arm_cmpsi_insn): Likewise.
4928         (cmpsi_shiftsi): Likewise.
4929         (cmpsi_shiftsi_swp): Likewise.
4930         (arm_cmpsi_negshiftsi_si): Likewise.
4931         (movsicc_insn): Likewise.
4932         (movsfcc_soft_insn): Likewise.
4933         (arith_shiftsi): Likewise.
4934         (arith_shiftsi_compare0
4935         (arith_shiftsi_compare0_scratch
4936         (sub_shiftsi): Likewise.
4937         (sub_shiftsi_compare0
4938         (sub_shiftsi_compare0_scratch
4939         (and_scc): Likewise.
4940         (cond_move): Likewise.
4941         (if_plus_move): Likewise.
4942         (if_move_plus): Likewise.
4943         (if_move_not): Likewise.
4944         (if_not_move): Likewise.
4945         (if_shift_move): Likewise.
4946         (if_move_shift): Likewise.
4947         (if_shift_shift): Likewise.
4948         (if_not_arith): Likewise.
4949         (if_arith_not): Likewise.
4950         (cond_move_not): Likewise.
4951         (thumb1_ashlsi3): Set type attribute.
4952         (thumb1_ashrsi3): Likewise.
4953         (thumb1_lshrsi3): Likewise.
4954         (thumb1_rotrsi3): Likewise.
4955         (shiftsi3_compare0_scratch): Likewise.
4956         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4957         (neon_mov<mode>): Likewise.
4958         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4959         attribute changes.
4960         (thumb2_movsi_insn): Likewise.
4961         (thumb2_cmpsi_neg_shiftsi): Likewise.
4962         (thumb2_extendqisi_v6): Likewise.
4963         (thumb2_zero_extendhisi2_v6): Likewise.
4964         (thumb2_zero_extendqisi2_v6): Likewise.
4965         (thumb2_shiftsi3_short): Likewise.
4966         (thumb2_addsi3_compare0_scratch): Likewise.
4967         (orsi_not_shiftsi_si): Likewise.
4968         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4969         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4970         changes.
4971         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4972         (1020alu_shift_op): Likewise.
4973         (1020alu_shift_reg_op): Likewise.
4974         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4975         (alu_shift_op): Likewise.
4976         (alu_shift_reg_op): Likewise.
4977         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4978         (11_alu_shift_op): Likewise.
4979         (11_alu_shift_reg_op): Likewise.
4980         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4981         (9_alu_shift_reg_op): Likewise.
4982         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4983         changes.
4984         (cortex_a15_alu_shift): Likewise.
4985         (cortex_a15_alu_shift_reg): Likewise.
4986         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4987         changes.
4988         (cortex_a5_alu_shift): Likewise.
4989         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4990         changes.
4991         (cortex_a53_alu_shift): Likewise.
4992         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4993         changes.
4994         (cortex_a7_alu_reg): Likewise.
4995         (cortex_a7_alu_shift): Likewise.
4996         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4997         changes.
4998         (cortex_a8_alu_shift): Likewise.
4999         (cortex_a8_alu_shift_reg): Likewise.
5000         (cortex_a8_mov): Likewise.
5001         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5002         (cortex_a9_dp_shift): Likewise.
5003         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5004         changes.
5005         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5006         changes.
5007         (cortex_r4_mov): Likewise.
5008         (cortex_r4_alu_shift): Likewise.
5009         (cortex_r4_alu_shift_reg): Likewise.
5010         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5011         (526_alu_shift_op): Likewise.
5012         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5013         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5014         (626te_alu_shift_op): Likewise.
5015         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5016         (726te_alu_op): Likewise.
5017         (726te_alu_shift_op): Likewise.
5018         (726te_alu_shift_reg_op): Likewise.
5019         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5020         (mp626_alu_shift_op): Likewise.
5021         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5022         (pj4_alu_e1_conds): Likewise.
5023         (pj4_alu): Likewise.
5024         (pj4_alu_conds): Likewise.
5025         (pj4_shift): Likewise.
5026         (pj4_shift_conds): Likewise.
5027         (pj4_alu_shift): Likewise.
5028         (pj4_alu_shift_conds): Likewise.
5029         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5030         changes.
5031         (cortexa7_older_only): Likewise.
5032         (cortexa7_younger): Likewise.
5034 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5036         * ipa-pure-const.c (generate_summary): Rename to...
5037         (pure_const_generate_summary): ... this.
5039 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
5041         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5043 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
5045         PR target/57909
5046         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5047         usage in HI mode.
5049 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5051         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5052         enabled without -march=zEC12.
5053         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5054         flags to be set.
5056 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5058         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5059         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5060         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5061         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5062         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5063         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5064         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5065         ISA_HAS_FP4.
5066         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5067         and ISA_HAS_NMADD3_NMSUB3.
5068         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5069         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5070         (nmsub4<mode>, nmsub3<mode>): Likewise.
5071         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5073 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5075         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5076         TARGET_MIPS5400 checking.
5078 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
5079             Peter Bergner  <bergner@vnet.ibm.com>
5081         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5082         registers in the comment.
5083         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5084         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5085         rather than FIRST_PSEUDO_REGISTERS.
5087 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
5089         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5090         enable extra ISA flags with TARGET_HTM.
5092 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5094         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5095         Fix comment typos.
5097 2013-07-15  Cong Hou  <congh@google.com>
5099         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5100         in compare function for sorting.
5102 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
5104         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5105         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5106         * config/rs6000/rs6000.opt: Add -mhtm option.
5107         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5108         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5109         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5110         __HTM__ if the HTM instructions are available.
5111         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5112         htm_spr_reg_operand): New define_predicates.
5113         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5114         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5115         Include htm.md.
5116         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5117         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5118         HTM builtin functions.
5119         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5120         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5121         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5122         (rs6000_builtin_mask_calculate): Likewise.
5123         (rs6000_option_override_internal): Likewise.
5124         (bdesc_htm): Add new HTM builtin support.
5125         (htm_spr_num): New function.
5126         (htm_spr_regno): Likewise.
5127         (rs6000_htm_spr_icode): Likewise.
5128         (htm_expand_builtin): Likewise.
5129         (htm_init_builtins): Likewise.
5130         (rs6000_expand_builtin): Add support for HTM builtin functions.
5131         (rs6000_init_builtins): Likewise.
5132         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5133         option.
5134         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5135         (TARGET_HTM, MASK_HTM): Define macros.
5136         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5137         (FIXED_REGISTERS): Likewise.
5138         (CALL_USED_REGISTERS): Likewise.
5139         (CALL_REALLY_USED_REGISTERS): Likewise.
5140         (REG_ALLOC_ORDER): Likewise.
5141         (enum reg_class): Likewise.
5142         (REG_CLASS_NAMES): Likewise.
5143         (REG_CLASS_CONTENTS): Likewise.
5144         (REGISTER_NAMES): Likewise.
5145         (ADDITIONAL_REGISTER_NAMES): Likewise.
5146         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5147         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5148         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5149         * config/rs6000/htm.md: New file.
5150         * config/rs6000/htmintrin.h: New file.
5151         * config/rs6000/htmxlintrin.h: New file.
5153 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5155         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5156         Define SYMBOL_TINY_GOT, update comment.
5157         * config/aarch64/aarch64.c
5158         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5159         (aarch64_expand_mov_immediate): Likewise.
5160         (aarch64_print_operand): Likewise.
5161         (aarch64_classify_symbol): Likewise.
5162         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5163         (ldr_got_tiny): Define.
5165 2013-07-13  Tobias Grosser  <tobias@grosser.es>
5167         PR tree-optimization/54094
5168         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5169           scheduling dimension for the parallelism check from the polyhedral
5170           information in the AST.
5171         * graphite-dependences.c (carries_deps): Do not assume the schedule is
5172           in 2D + 1 form.
5174 2013-07-13  Jason Merrill  <jason@redhat.com>
5176         * print-tree.c (debug_vec_tree): Use debug_raw.
5177         (debug_raw (vec<tree, va_gc> &)): New.
5178         (debug_raw (vec<tree, va_gc> *)): New.
5179         * tree.h: Declare them.
5181 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
5183         * ifcvt.c (ifcvt_after_combine): New static variable.
5184         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5185         for size.
5186         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
5187         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5188         rest_of_handle_if_after_reload): Pass new argument for if_convert.
5190 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
5192         * config/mips/mips.c (mips_expand_call): Remove empty statement.
5194 2013-07-12  Michael Matz  <matz@suse.de>
5196         PR middle-end/55771
5197         * convert.c (convert_to_real): Reject non-float inner types.
5199 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
5201         * config/aarch64/aarch64-protos.h
5202         (aarch64_simd_immediate_valid_for_move): Remove.
5203         * config/aarch64/aarch64.c (simd_immediate_info): New member.
5204         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5205         cases.
5206         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5208 2013-07-11  Steve Ellcey  <sellcey@mips.com>
5210         * config/mips/mips.c (mips_conditional_register_usage): Do not
5211         use t[0-7] registers in MIPS16 mode when optimizing for size.
5213 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
5215         * config/i386/i386.c (dispatch_function_versions): Fix array
5216         indexing of function_version_info to match actual_versions.
5218 2013-07-11  Teresa Johnson  <tejohnson@google.com>
5220         * vec.h (struct va_gc): Move release out-of-line.
5221         (va_gc::release): Call ggc_free on released vec.
5223 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5225         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5226         Require GOT register as additional operand in UNSPEC.
5227         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5228         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5229         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5230         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5231         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5232         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5233         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5235 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
5237         PR target/57631
5238         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5239         name seen by assembler/linker rather if available.
5241 2013-07-11  Andreas Schwab  <schwab@suse.de>
5243         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5245 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
5247         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5249 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
5251         * doc/tm.texi.in: Move hook documentation to ....
5252         * target.def: ... here.
5254         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5255         text on @hook line.
5256         * doc/tm.texi: Regenerate.
5258 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
5260         PR c++/57869
5261         * doc/invoke.texi: Document -Wconditionally-supported.
5263 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5265         PR target/57844
5266         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5267         of my_fp.
5269 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5271         PR target/57506
5272         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5273         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5274         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5275         Remove duplicate devices.
5276         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5277         * config/avr/t-multilib: Regenerate.
5278         * config/avr/avr-tables.opt: Regenerate.
5279         * doc/avr-mmcu.texi: Regenerate.
5281 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5283         PR target/56987
5284         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5286 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
5288         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5289         the cost of MULT when optimizing for size.
5291 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5293         * config/cr16/cr16-protos.h: Don't include target.h.
5295 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
5297         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5298         adjust register size for TDmode and TFmode for VSX registers.
5300 2013-07-08  Kai Tietz  <ktietz@redhat.com>
5302         PR target/56892
5303         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5304         hook_bool_const_tree_true.
5306 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5308         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5309         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5310         * config/s390/s390.md: Define FPR*_REGNUM constants.
5311         Fix FPR2_REGNUM constant (18 -> 17).
5312         ("*trunc<BFP:mode><DFP_ALL:mode>2")
5313         ("*trunc<DFP_ALL:mode><BFP:mode>2")
5314         ("trunc<BFP:mode><DFP_ALL:mode>2")
5315         ("trunc<DFP_ALL:mode><BFP:mode>2")
5316         ("*extend<BFP:mode><DFP_ALL:mode>2")
5317         ("*extend<DFP_ALL:mode><BFP:mode>2")
5318         ("extend<BFP:mode><DFP_ALL:mode>2")
5319         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5320         FPR4_REGNUM.
5322 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
5324         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5326 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5328         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5329         and cfun_fpr_bit_p to cfun_fpr_save_p.
5330         (s390_frame_area, s390_register_info, s390_frame_info)
5331         (s390_emit_prologue, s390_emit_epilogue)
5332         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5333         register numbers.
5334         * config/s390/s390.h: Define *_REGNUM macros for floating point
5335         register numbers.
5337 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
5339         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5341 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
5343         PR rtl-optimization/57786
5344         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5345         and break out of the loop when it is set to false.
5347 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
5349         PR target/57819
5350         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5351         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5352         (const_int 63)) 0)).
5353         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5354         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5355         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5357         PR rtl-optimization/57829
5358         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5359         mask bits outside of mode are just sign-extension from mode to HWI.
5361 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5363         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5364         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5365         adjust_address instead of change_address to keep info about alignment.
5366         (emit_strmov): Remove.
5367         (emit_memmov): New function.
5368         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5369         (expand_movmem_epilogue): Likewise and return updated rtx for
5370         destination.
5371         (expand_constant_movmem_prologue): Likewise and return updated rtx for
5372         destination and source.
5373         (decide_alignment): Refactor, handle vector_loop.
5374         (ix86_expand_movmem): Likewise.
5375         (ix86_expand_setmem): Likewise.
5376         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5378 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
5380         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5381         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5383 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5385         * config/i386/sse.md (sse_movlhps): Change alternative 3
5386         of operand 2 to "m".
5388 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5390         PR target/57807
5391         * config/i386/sse.md (iptr): New mode attribute.
5392         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5393         (<sse>_vm<plusminus_insn><mode>3): Ditto.
5394         (<sse>_vmmul<mode>3): Ditto.
5395         (<sse>_vmdiv<mode>3): Ditto.
5396         (sse_vmrcpv4sf2): Ditto.
5397         (<sse>_vmsqrt<mode>2): Ditto.
5398         (sse_vmrsqrtv4sf2): Ditto.
5399         (<sse>_vm<code><mode>3): Ditto.
5400         (avx_vmcmp<mode>3): Ditto.
5401         (<sse>_vmmaskcmp<mode>3): Ditto.
5402         (<sse>_comi): Ditto.
5403         (<sse>_ucomi): Ditto.
5404         (*xop_vmfrcz_<mode>): Ditto.
5405         (*fmai_fmadd_<mode>): Ditto.
5406         (*fmai_fmsub_<mode>): Ditto.
5407         (*fmai_fnmadd_<mode>): Ditto.
5408         (*fmai_fnmsub_<mode>): Ditto.
5409         (*fma4i_vmfmadd_<mode>): Ditto.
5410         (*fma4i_vmfmsub_<mode>): Ditto.
5411         (*fma4i_vmfnmadd_<mode>): Ditto.
5412         (*fma4i_vmfnmsub_<mode>): Ditto.
5413         (*xop_vmfrcz_<mode>): Ditto.
5414         (sse_cvtps2pi): Ditto.
5415         (sse_cvttps2pi): Ditto.
5416         (sse_cvtss2si): Ditto.
5417         (sse_cvtss2si_2): Ditto.
5418         (sse_cvtss2siq_2): Ditto.
5419         (sse_cvttss2si): Ditto.
5420         (sse_cvttss2siq): Ditto.
5421         (sse_cvtsd2si): Ditto.
5422         (sse_cvtsd2si_2): Ditto.
5423         (sse_cvtsd2siq_2): Ditto.
5424         (sse_cvttsd2si): Ditto.
5425         (sse_cvttsd2siq): Ditto.
5426         (sse_cvtsd2ss): Ditto.
5427         (sse_cvtss2sd): Ditto.
5428         (avx2_pbroadcast<mode>): Ditto.
5429         (avx2_pbroadcast<mode>_1): Ditto.
5430         (*avx_vperm_broadcast_v4sf): Ditto.
5432         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5433         (sse_movlhps): Ditto.
5434         (sse_storehps): Ditto.
5435         (sse_loadhps): Ditto.
5436         (sse_storelps): Ditto.
5437         (sse_loadlps): Ditto.
5438         (*vec_concatv4sf): Ditto.
5439         (*vec_interleave_highv2df): Ditto.
5440         (*vec_interleave_lowv2df): Ditto.
5441         (*vec_extractv2df_1_sse): Ditto.
5442         (*vec_extractv2df_0_sse): Ditto.
5443         (sse2_storelpd): Ditto.
5444         (sse2_loadlpd): Ditto.
5445         (sse2_movsd): Ditto.
5446         (*vec_concatv4si): Ditto.
5447         (vec_concatv2di): Ditto.
5449         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5450         for Intel asm dialect.
5451         (mmx_punpcklwd): Ditto.
5452         (mmx_punpckldq): Ditto.
5454         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5455         for intel assembler dialect.
5457 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
5459         PR target/29776
5460         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5461         for BUILT_IN_C{LZ,LRSB}*.
5462         * tree.h (CASE_INT_FN): Add FN##IMAX case.
5463         * tree-vrp.c (extract_range_basic): Handle
5464         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
5465         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5466         fall thru to code calling set_value*.
5467         * builtins.c (expand_builtin): Remove *IMAX cases.
5468         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5469         if width is bigger than 2*HWI.
5471 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
5473         PR rtl-optimization/55342
5474         * lra-int.h (lra_subreg_reload_pseudos): New.
5475         * lra.c: Add undoing optional reloads to the block diagram.
5476         (lra_subreg_reload_pseudos): New.
5477         (lra_optional_reload_pseudos): Change comments.
5478         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
5479         lra_optional_reload_pseudos after undo transformations.
5480         * lra-assigns.c (pseudo_prefix_title): New.
5481         (lra_setup_reg_renumber): Use it.
5482         (spill_for): Ditto.  Check subreg reload pseudos too.
5483         (assign_by_spills): Consider subreg reload pseudos too.
5484         * lra-constraints.c (simplify_operand_subreg): Use
5485         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5486         (curr_insn_transform): Recognize and do optional reloads.
5487         (undo_optional_reloads): New.
5488         (lra_undo_inheritance): Call undo_optional_reloads.
5490 2013-07-05  Thomas Quinot  <quinot@adacore.com>
5492         * tree-complex.c (expand_complex_operations_1): Fix typo.
5494 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
5496         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5497         (tune_params): New member 'const vec_costs'.
5498         * config/aarch64/aarch64.c (generic_vector_cost): New.
5499         (generic_tunings): New member 'generic_vector_cost'.
5500         (aarch64_builtin_vectorization_cost): New.
5501         (aarch64_add_stmt_cost): New.
5502         (TARGET_VECTORIZE_ADD_STMT_COST): New.
5503         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5505 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
5507         PR target/57777
5508         * config/i386/predicates.md (vsib_address_operand): Disallow
5509         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5511 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
5513         PR middle-end/55030
5514         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5515         expand_builtin_setjmp_receiver.
5516         (expand_label): Adjust, call expand_builtin_setjmp_receiver
5517         with NULL for the label parameter.
5518         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5519         the frame-pointer.  Adjust comments.
5520         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5521         only if LABEL is non-NULL.
5523 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
5525         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5526         (ARM_ABI_AAPCS64): Ditto.
5527         (arm_abi): Ditto.
5528         (ARM_DEFAULT_ABI): Ditto.
5530 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
5532         * config/aarch64/aarch64-builtins.c
5533         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5534         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5535         (st1): Likewise.
5536         * config/aarch64/aarch64-simd.md
5537         (aarch64_ld1<VALL:mode>): New.
5538         (aarch64_st1<VALL:mode>): Likewise.
5539         * config/aarch64/arm_neon.h
5540         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5542 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
5544         * config/i386/i386.c (gate_insert_vzeroupper): Check if
5545         target ISA is AVX.
5546         (ix86_option_override_internal):Turn on all -mavx target flags by
5547         default as they are dependent on AVX anyway.
5549 2013-07-02  Cary Coutant  <ccoutant@google.com>
5551         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5552         deterministic hash.
5553         (loc_checksum_ordered): Likewise.
5554         (hash_loc_operands): Remove inline keyword.
5556 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
5558         PR tree-optimization/57741
5559         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5560         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5561         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5562         Allow REAL_CST step_exprs if flag_associative_math.
5563         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5565 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5567         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5569 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5571         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5573 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5575         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5576         encoding.
5577         (iorsi3_insn): Likewise.
5578         (arm_xorsi3): Likewise.
5580 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
5582         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
5583         to ...
5584         (attribute "type"): ... here, and prefix with "wmmx_".
5585         (attribute "core_cycles"): Update for attribute changes.
5586         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5587         (tbcstv4hi): Likewise.
5588         (tbcstv2si): Likewise.
5589         (iwmmxt_iordi3): Likewise.
5590         (iwmmxt_xordi3): Likewise.
5591         (iwmmxt_anddi3): Likewise.
5592         (iwmmxt_nanddi3): Likewise.
5593         (iwmmxt_arm_movdi): Likewise.
5594         (iwmmxt_movsi_insn): Likewise.
5595         (mov<mode>_internal): Likewise.
5596         (and<mode>3_iwmmxt): Likewise.
5597         (ior<mode>3_iwmmxt): Likewise.
5598         (xor<mode>3_iwmmxt): Likewise.
5599         (add<mode>3_iwmmxt): Likewise.
5600         (ssaddv8qi3): Likewise.
5601         (ssaddv4hi3): Likewise.
5602         (ssaddv2si3): Likewise.
5603         (usaddv8qi3): Likewise.
5604         (usaddv4hi3): Likewise.
5605         (usaddv2si3): Likewise.
5606         (sub<mode>3_iwmmxt): Likewise.
5607         (sssubv8qi3): Likewise.
5608         (sssubv4hi3): Likewise.
5609         (sssubv2si3): Likewise.
5610         (ussubv8qi3): Likewise.
5611         (ussubv4hi3): Likewise.
5612         (ussubv2si3): Likewise.
5613         (mulv4hi3_iwmmxt): Likewise.
5614         (smulv4hi3_highpart): Likewise.
5615         (umulv4hi3_highpart): Likewise.
5616         (iwmmxt_wmacs): Likewise.
5617         (iwmmxt_wmacsz): Likewise.
5618         (iwmmxt_wmacu): Likewise.
5619         (iwmmxt_wmacuz): Likewise.
5620         (iwmmxt_clrdi): Likewise.
5621         (iwmmxt_clrv8qi): Likewise.
5622         (iwmmxt_clr4hi): Likewise.
5623         (iwmmxt_clr2si): Likewise.
5624         (iwmmxt_uavgrndv8qi3): Likewise.
5625         (iwmmxt_uavgrndv4hi3): Likewise.
5626         (iwmmxt_uavgv8qi3): Likewise.
5627         (iwmmxt_uavgv4hi3): Likewise.
5628         (iwmmxt_tinsrb): Likewise.
5629         (iwmmxt_tinsrh): Likewise.
5630         (iwmmxt_tinsrw): Likewise.
5631         (iwmmxt_textrmub): Likewise.
5632         (iwmmxt_textrmsb): Likewise.
5633         (iwmmxt_textrmuh): Likewise.
5634         (iwmmxt_textrmsh): Likewise.
5635         (iwmmxt_textrmw): Likewise.
5636         (iwmxxt_wshufh): Likewise.
5637         (eqv8qi3): Likewise.
5638         (eqv4hi3): Likewise.
5639         (eqv2si3): Likewise.
5640         (gtuv8qi3): Likewise.
5641         (gtuv4hi3): Likewise.
5642         (gtuv2si3): Likewise.
5643         (gtv8qi3): Likewise.
5644         (gtv4hi3): Likewise.
5645         (gtv2si3): Likewise.
5646         (smax<mode>3_iwmmxt): Likewise.
5647         (umax<mode>3_iwmmxt): Likewise.
5648         (smin<mode>3_iwmmxt): Likewise.
5649         (umin<mode>3_iwmmxt): Likewise.
5650         (iwmmxt_wpackhss): Likewise.
5651         (iwmmxt_wpackwss): Likewise.
5652         (iwmmxt_wpackdss): Likewise.
5653         (iwmmxt_wpackhus): Likewise.
5654         (iwmmxt_wpackwus): Likewise.
5655         (iwmmxt_wpackdus): Likewise.
5656         (iwmmxt_wunpckihb): Likewise.
5657         (iwmmxt_wunpckihh): Likewise.
5658         (iwmmxt_wunpckihw): Likewise.
5659         (iwmmxt_wunpckilb): Likewise.
5660         (iwmmxt_wunpckilh): Likewise.
5661         (iwmmxt_wunpckilw): Likewise.
5662         (iwmmxt_wunpckehub): Likewise.
5663         (iwmmxt_wunpckehuh): Likewise.
5664         (iwmmxt_wunpckehuw): Likewise.
5665         (iwmmxt_wunpckehsb): Likewise.
5666         (iwmmxt_wunpckehsh): Likewise.
5667         (iwmmxt_wunpckehsw): Likewise.
5668         (iwmmxt_wunpckelub): Likewise.
5669         (iwmmxt_wunpckeluh): Likewise.
5670         (iwmmxt_wunpckeluw): Likewise.
5671         (iwmmxt_wunpckelsb): Likewise.
5672         (iwmmxt_wunpckelsh): Likewise.
5673         (iwmmxt_wunpckelsw): Likewise.
5674         (ror<mode>3): Likewise.
5675         (ashr<mode>3_iwmmxt): Likewise.
5676         (lshr<mode>3_iwmmxt): Likewise.
5677         (ashl<mode>3_iwmmxt): Likewise.
5678         (ror<mode>3_di): Likewise.
5679         (ashr<mode>3_di): Likewise.
5680         (lshr<mode>3_di): Likewise.
5681         (ashl<mode>3_di): Likewise.
5682         (iwmmxt_wmadds): Likewise.
5683         (iwmmxt_wmaddu): Likewise.
5684         (iwmmxt_tmia): Likewise.
5685         (iwmmxt_tmiaph): Likewise.
5686         (iwmmxt_tmiabb): Likewise.
5687         (iwmmxt_tmiatb): Likewise.
5688         (iwmmxt_tmiabt): Likewise.
5689         (iwmmxt_tmiatt): Likewise.
5690         (iwmmxt_tmovmskb): Likewise.
5691         (iwmmxt_tmovmskh): Likewise.
5692         (iwmmxt_tmovmskw): Likewise.
5693         (iwmmxt_waccb): Likewise.
5694         (iwmmxt_wacch): Likewise.
5695         (iwmmxt_waccw): Likewise.
5696         (iwmmxt_waligni): Likewise.
5697         (iwmmxt_walignr): Likewise.
5698         (iwmmxt_walignr0): Likewise.
5699         (iwmmxt_walignr1): Likewise.
5700         (iwmmxt_walignr2): Likewise.
5701         (iwmmxt_walignr3): Likewise.
5702         (iwmmxt_wsadb): Likewise.
5703         (iwmmxt_wsadh): Likewise.
5704         (iwmmxt_wsadbz): Likewise.
5705         (iwmmxt_wsadhz): Likewise.
5706         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5707         (iwmmxt_wabsdiffb): Likewise.
5708         (iwmmxt_wabsdiffh): Likewise.
5709         (iwmmxt_wabsdiffw): Likewise.
5710         (iwmmxt_waddsubhx): Likewise
5711         (iwmmxt_wsubaddhx): Likewise.
5712         (addc<mode>3): Likewise.
5713         (iwmmxt_avg4): Likewise.
5714         (iwmmxt_avg4r): Likewise.
5715         (iwmmxt_wmaddsx): Likewise.
5716         (iwmmxt_wmaddux): Likewise.
5717         (iwmmxt_wmaddsn): Likewise.
5718         (iwmmxt_wmaddun): Likewise.
5719         (iwmmxt_wmulwsm): Likewise.
5720         (iwmmxt_wmulwum): Likewise.
5721         (iwmmxt_wmulsmr): Likewise.
5722         (iwmmxt_wmulumr): Likewise.
5723         (iwmmxt_wmulwsmr): Likewise.
5724         (iwmmxt_wmulwumr): Likewise.
5725         (iwmmxt_wmulwl): Likewise.
5726         (iwmmxt_wqmulm): Likewise.
5727         (iwmmxt_wqmulwm): Likewise.
5728         (iwmmxt_wqmulmr): Likewise.
5729         (iwmmxt_wqmulwmr): Likewise.
5730         (iwmmxt_waddbhusm): Likewise.
5731         (iwmmxt_waddbhusl): Likewise.
5732         (iwmmxt_wqmiabb): Likewise.
5733         (iwmmxt_wqmiabt): Likewise.
5734         (iwmmxt_wqmiatb): Likewise.
5735         (iwmmxt_wqmiatt): Likewise.
5736         (iwmmxt_wqmiabbn): Likewise.
5737         (iwmmxt_wqmiabtn): Likewise.
5738         (iwmmxt_wqmiatbn): Likewise.
5739         (iwmmxt_wqmiattn): Likewise.
5740         (iwmmxt_wmiabb): Likewise.
5741         (iwmmxt_wmiabt): Likewise.
5742         (iwmmxt_wmiatb): Likewise.
5743         (iwmmxt_wmiatt): Likewise.
5744         (iwmmxt_wmiabbn): Likewise.
5745         (iwmmxt_wmiabtn): Likewise.
5746         (iwmmxt_wmiatbn): Likewise.
5747         (iwmmxt_wmiattn): Likewise.
5748         (iwmmxt_wmiawbb): Likewise.
5749         (iwmmxt_wmiawbt): Likewise.
5750         (iwmmxt_wmiawtb): Likewise.
5751         (iwmmxt_wmiawtt): Likewise.
5752         (iwmmxt_wmiawbbn): Likewise.
5753         (iwmmxt_wmiawbtn): Likewise.
5754         (iwmmxt_wmiawtbn): Likewise.
5755         (iwmmxt_wmiawttn): Likewise.
5756         (iwmmxt_wmerge): Likewise.
5757         (iwmmxt_tandc<mode>3): Likewise.
5758         (iwmmxt_torc<mode>3): Likewise.
5759         (iwmmxt_torvsc<mode>3): Likewise.
5760         (iwmmxt_textrc<mode>3): Likewise.
5761         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5762         (wmmxt_pack): Likewise.
5763         (wmmxt_mult_c1): Likewise.
5764         (wmmxt_mult_c2): Likewise.
5765         (wmmxt_alu_c1): Likewise.
5766         (wmmxt_alu_c2): Likewise.
5767         (wmmxt_alu_c3): Likewise.
5768         (wmmxt_transfer_c1): Likewise.
5769         (wmmxt_transfer_c2): Likewise.
5770         (wmmxt_transfer_c3): Likewise.
5771         (marvell_f_iwmmxt_wstr): Likewise.
5772         (marvell_f_iwmmxt_wldr): Likewise.
5774 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5776         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5778 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5780         Revert:
5781         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5782         * lra-constraints.c (need_for_split_p): Check call used hard regs
5783         living through calls.
5785         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5786         call used regs for call insn.
5788 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5790         PR target/57736
5791         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5792         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5794 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5796         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5798 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5800         * lra-constraints.c (need_for_split_p): Check call used hard regs
5801         living through calls.
5803 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5805         PR target/57744
5806         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5807         to tie with any other modes.  Eliminate Altivec vector mode tests,
5808         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5809         code, to return 0 if testing MODE2 for a condition, if we've
5810         already tested MODE1 for the same condition.
5812 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5814         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5815         layout.
5817 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5819         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5820         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5822 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5824         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5825         Define.
5826         (aarch64_symbolic_constant_p): Remove.
5827         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5828         static.  Fix line length and white space.
5829         (aarch64_symbolic_constant_p): Remove.
5830         * config/aarch64/predicates.md (aarch64_valid_symref):
5831         Use aarch64_classify_symbol_expression.
5833 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5835         * config/arm/constraints.md (Ts): New constraint.
5836         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5837         16-bit encodings.
5838         (compare_scc): Use "Ts" constraint for operand 0.
5839         (ior_scc_scc): Likewise.
5840         (and_scc_scc): Likewise.
5841         (and_scc_scc_nodom): Likewise.
5842         (ior_scc_scc_cmp): Likewise for operand 7.
5843         (and_scc_scc_cmp): Likewise.
5844         * config/arm/thumb2.md (thumb2_movsi_insn):
5845         Add alternatives for 16-bit encodings.
5846         (thumb2_movhi_insn): Likewise.
5847         (thumb2_movsicc_insn): Likewise.
5848         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5849         (thumb2_negscc): Use "Ts" constraint.
5850         Move mvn instruction outside cond_exec block.
5851         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5852         for 16-bit encodings.
5854 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5856         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5857         encoding.
5858         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5859         (mulsi3subsi): Likewise.
5860         (mulsidi3adddi): Likewise.
5861         (mulsidi3_v6): Likewise.
5862         (umulsidi3_v6): Likewise.
5863         (umulsidi3adddi_v6): Likewise.
5864         (smulsi3_highpart_v6): Likewise.
5865         (umulsi3_highpart_v6): Likewise.
5866         (mulhisi3tb): Likewise.
5867         (mulhisi3bt): Likewise.
5868         (mulhisi3tt): Likewise.
5869         (maddhisi4): Likewise.
5870         (maddhisi4tb): Likewise.
5871         (maddhisi4tt): Likewise.
5872         (maddhidi4): Likewise.
5873         (maddhidi4tb): Likewise.
5874         (maddhidi4tt): Likewise.
5875         (zeroextractsi_compare0_scratch): Likewise.
5876         (insv_zero): Likewise.
5877         (insv_t2): Likewise.
5878         (anddi_notzesidi_di): Likewise.
5879         (anddi_notsesidi_di): Likewise.
5880         (andsi_notsi_si): Likewise.
5881         (iordi_zesidi_di): Likewise.
5882         (xordi_zesidi_di): Likewise.
5883         (andsi_iorsi3_notsi): Likewise.
5884         (smax_0): Likewise.
5885         (smax_m1): Likewise.
5886         (smin_0): Likewise.
5887         (not_shiftsi): Likewise.
5888         (unaligned_loadsi): Likewise.
5889         (unaligned_loadhis): Likewise.
5890         (unaligned_loadhiu): Likewise.
5891         (unaligned_storesi): Likewise.
5892         (unaligned_storehi): Likewise.
5893         (extv_reg): Likewise.
5894         (extzv_t2): Likewise.
5895         (divsi3): Likewise.
5896         (udivsi3): Likewise.
5897         (arm_zero_extendhisi2addsi): Likewise.
5898         (arm_zero_extendqisi2addsi): Likewise.
5899         (compareqi_eq0): Likewise.
5900         (arm_extendhisi2_v6): Likewise.
5901         (arm_extendqisi2addsi): Likewise.
5902         (arm_movt): Likewise.
5903         (thumb2_ldrd): Likewise.
5904         (thumb2_ldrd_base): Likewise.
5905         (thumb2_ldrd_base_neg): Likewise.
5906         (thumb2_strd): Likewise.
5907         (thumb2_strd_base): Likewise.
5908         (thumb2_strd_base_neg): Likewise.
5909         (arm_negsi2): Add alternative for 16-bit encoding.
5910         (arm_one_cmplsi2): Likewise.
5912 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5914         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5915         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5916         (movdfcc): Likewise.
5917         * config/arm/vfp.md (*thumb2_movsf_vfp):
5918         Disable predication for arm_restrict_it.
5919         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5920         (*thumb2_movdfcc_vfp): Likewise.
5921         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5922         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5923         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5924         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5925         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5926         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5927         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5928         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5929         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5930         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5931         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5932         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5933         Disable predication for arm_restrict_it.
5935 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5937         * config/i386/bmiintrin.h (_bextr_u32): New.
5938         (_bextr_u64): Ditto.
5940 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5942         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5943         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5944         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5945         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5946         * config/mips/n32-elf.h: ...this new file.
5948 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5950         PR target/57224
5951         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5952         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5954 2013-06-27  Catherine Moore  <clm@codesourcery.com>
5956         * config/mips/mips-tables.opt: Regenerate.
5957         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5958         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5959         * doc/invoke.texi: Add -m14kc.
5961 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5963         PR target/57623
5964         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5965         constraints of operand 1 and 2.
5967         PR target/57623
5968         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5969         to match RTL canonicalization.  Swap predicates and
5970         constraints of operand 1 and 2.
5972 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5974         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5975         Process OP_INOUT regs for splitting too.
5977 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5979         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5980         decl before the loop, initialize to NULL.
5981         (vectorizable_load): Initialize ptr_incr to NULL.
5983 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5985         PR lto/57208
5986         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5987         * ipa-ref.c (ipa_maybe_record_reference): New function.
5988         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5989         * ipa-cp.c (create_specialized_node): Record potential references from
5990         aggvals.
5991         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5993 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5995         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5996         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5997         'mode' to force_reg.
5998         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5999         (aarch64_expand_mov_immediate): Likewise.
6001 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6003         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6004         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6006 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6008         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6009         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6010         (struct machine_function): Add tbegin_p.
6011         (s390_canonicalize_comparison): Fold CC mode compares to
6012         conditional jump if possible.
6013         (s390_emit_jump): Return the emitted jump.
6014         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6015         Handle CCRAWmode compares.
6016         (s390_option_override): Default to -mhtm if available.
6017         (s390_reg_clobbered_rtx): Handle floating point regs as well.
6018         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6019         FPRs instead of df_regs_ever_live_p.
6020         (s390_optimize_nonescaping_tx): New function.
6021         (s390_init_frame_layout): Extend clobbered_regs array to cover
6022         FPRs as well.
6023         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6024         (s390_expand_tbegin): New function.
6025         (enum s390_builtin): New enum definition.
6026         (code_for_builtin): New array definition.
6027         (s390_init_builtins): New function.
6028         (s390_expand_builtin): New function.
6029         (TARGET_INIT_BUILTINS): Define.
6030         (TARGET_EXPAND_BUILTIN): Define.
6031         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6032         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6033         (s390_alc_comparison): Likewise.
6034         * config/s390/s390-modes.def: Add CCRAWmode.
6035         * config/s390/s390.h (processor_flags): Add PF_TX.
6036         (TARGET_CPU_HTM): Define macro.
6037         (TARGET_HTM): Define macro.
6038         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6039         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6040         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6041         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6042         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6043         (TBEGIN_MASK, TBEGINC_MASK): New constants.
6044         ("*cc_to_int"): Move up.
6045         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6046         constants other than 0.
6047         ("*ccraw_to_int"): New insn and splitter definition.
6048         ("tbegin", "tbegin_nofloat", "tbegin_retry")
6049         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6050         ("tx_assist"): New expander.
6051         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6052         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6053         * config/s390/s390.opt: Add -mhtm option.
6054         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6055         * config/s390/htmxlintrin.h: New file.
6056         * config/s390/htmintrin.h: New file.
6057         * config/s390/s390intrin.h: New file.
6058         * doc/extend.texi: Document htm builtins.
6059         * config.gcc: Add the new header files to extra_headers.
6061 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
6063         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6064         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6066 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
6067             Pat Haugen  <pthaugen@us.ibm.com>
6068             Peter Bergner  <bergner@vnet.ibm.com>
6070         * config/rs6000/power8.md: New.
6071         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6072         setting for power8 entry.
6073         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6074         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6075         test for Power4/Power5 only.
6076         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6077         support.
6078         (force_new_group): Adjust comment.
6079         * config/rs6000/rs6000.md: Include power8.md.
6081 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
6083         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6084         * config/arm/arm-protos.h (arm_max_conditional_execute): New
6085         declaration.
6086         (tune_params): Update comment.
6087         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6088         (arm_max_conditional_execute): New function.
6089         (thumb2_final_prescan_insn): Use max_insn_skipped and
6090         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6092 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
6094         PR tree-optimization/57705
6095         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6096         SSA_NAME step, provided that it is not defined inside the loop.
6097         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6098         (get_initial_def_for_induction): Handle SSA_NAME IV step.
6100 2013-06-25  Martin Jambor  <mjambor@suse.cz>
6102         PR middle-end/57670
6103         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6104         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6105         calls in the dump.
6106         (ipa_note_param_call): Initialize member_ptr flag.
6107         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6108         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6109         (ipa_write_indirect_edge_info): Stream member_ptr flag.
6110         (ipa_read_indirect_edge_info): Likewise.
6112 2013-06-25  Richard Biener  <rguenther@suse.de>
6114         PR middle-end/56977
6115         * passes.c (init_optimization_passes): Move pass_fold_builtins
6116         and pass_dce earlier with -Og.
6118 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
6120         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6121         <BIT_FIELD_REF>: Remove trailing TAB.
6122         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6123         remove blank line.
6125 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6127         PR tree-optimization/57358
6128         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6129         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6130         (ipa_analyze_params_uses): Generate pessimistic info when true.
6132 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6134         PR tree-optimization/57539
6135         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6136         global.inlined_to of the new node to it.  All callers changed.
6137         * ipa-inline-transform.c (clone_inlined_nodes): New variable
6138         inlining_into, pass it to cgraph_clone_node.
6139         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6140         ipa_free_edge_args_substructures.
6141         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6142         rdesc linked list.  Do not assert rdesc edges have inlined caller.
6143         Assert we have found an rdesc in the rdesc list.
6145 2013-06-24  Richard Biener  <rguenther@suse.de>
6147         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6148         (pointer_set_lookup): Declare.
6149         (class pointer_map): New template class implementing a
6150         generic pointer to T map.
6151         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6152         pointer_map<T>::contains, pointer_map<T>::insert,
6153         pointer_map<T>::traverse): New functions.
6154         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6155         (pointer_set_lookup): New function.
6156         (pointer_set_contains): Use pointer_set_lookup.
6157         (pointer_set_insert): Likewise.
6158         (insert_aux): Remove.
6159         (struct pointer_map_t): Embed a pointer_set_t.
6160         (pointer_map_create): Adjust.
6161         (pointer_map_destroy): Likewise.
6162         (pointer_map_contains): Likewise.
6163         (pointer_map_insert): Likewise.
6164         (pointer_map_traverse): Likewise.
6165         * tree-streamer.h (struct streamer_tree_cache_d): Use a
6166         pointer_map<unsigned> instead of a pointer_map_t.
6167         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6168         (streamer_tree_cache_lookup): Likewise.
6169         (streamer_tree_cache_create): Likewise.
6170         (streamer_tree_cache_delete): Likewise.
6171         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6172         pointer_map<unsigned> instead of a pointer_map_t.
6173         (lto_init_tree_ref_encoder): Adjust.
6174         (lto_destroy_tree_ref_encoder): Likewise.
6175         * lto-section-out.c (lto_output_decl_index): Likewise.
6176         (lto_record_function_out_decl_state): Likewise.
6177         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6179 2013-06-24  Richard Biener  <rguenther@suse.de>
6181         PR tree-optimization/57488
6182         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6184 2013-06-24  Alan Modra  <amodra@gmail.com>
6186         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6187         (gen_easy_altivec_constant): Likewise.
6188         * config/rs6000/predicates.md (easy_vector_constant_add_self,
6189         easy_vector_constant_msb): Likewise.
6191 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
6193         PR target/57688
6194         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6195         add missing return true.
6197 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
6199         PR target/52483
6200         * config/sh/predicates.md (general_extend_operand): Invoke
6201         general_movsrc_operand for memory operands.
6202         (general_movsrc_operand): Allow reg+reg addressing, do not use
6203         general_operand for memory operands.
6205 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
6207         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
6208         when current target options does not apply.
6209         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6210         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6211         * config/i386/bmiintrin.h: Pass appropriate target
6212         attributes to header.
6213         * config/i386/mmintrin.h: Ditto.
6214         * config/i386/nmmintrin.h: Ditto.
6215         * config/i386/avx2intrin.h: Ditto.
6216         * config/i386/fxsrintrin.h: Ditto.
6217         * config/i386/tbmintrin.h: Ditto.
6218         * config/i386/xsaveintrin.h: Ditto.
6219         * config/i386/f16cintrin.h: Ditto.
6220         * config/i386/xtestintrin.h: Ditto.
6221         * config/i386/xsaveoptintrin.h: Ditto.
6222         * config/i386/bmi2intrin.h: Ditto.
6223         * config/i386/lzcntintrin.h: Ditto.
6224         * config/i386/smmintrin.h: Ditto.
6225         * config/i386/wmmintrin.h: Ditto.
6226         * config/i386/x86intrin.h: Remove all header include guards.
6227         * config/i386/prfchwintrin.h: Ditto.
6228         * config/i386/pmmintrin.h: Ditto.
6229         * config/i386/tmmintrin.h: Ditto.
6230         * config/i386/xmmintrin.h: Ditto.
6231         * config/i386/popcntintrin.h: Ditto.
6232         * config/i386/rdseedintrin.h: Ditto.
6233         * config/i386/ammintrin.h: Ditto.
6234         * config/i386/emmintrin.h: Ditto.
6235         * config/i386/immintrin.h: Remove all header include guards.
6236         * config/i386/fma4intrin.h: Ditto.
6237         * config/i386/lwpintrin.h: Ditto.
6238         * config/i386/xopintrin.h: Ditto.
6239         * config/i386/ia32intrin.h: Ditto.
6240         * config/i386/avxintrin.h: Ditto.
6241         * config/i386/rtmintrin.h: Ditto.
6242         * config/i386/fmaintrin.h: Ditto.
6243         * config/i386/mm3dnow.h: Ditto.
6245 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
6247         * common/config/i386/i386-common.c: Handle LZCNT.
6249 2013-06-22  Andi Kleen  <ak@linux.intel.com>
6251         * doc/extend.texi: Use __atomic_store_n instead of
6252         __atomic_store in HLE example.
6254 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
6256         * config/sh/sh.c: Remove <cstdlib> workaround.
6258 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6260         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
6262 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6264         * doc/extend.texi: Document that __atomic_clear and
6265           __atomic_test_and_set should only be used with bool.
6267 2013-06-20  Jan Hubicka  <jh@suse.cz>
6269         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6270         types_same_for_odr.
6271         * tree.c (decls_same_for_odr): New function.
6272         (same_for_edr): New function.
6273         (types_same_for_odr): New function.
6274         (get_binfo_at_offset): Use it.
6275         * tree.h (types_same_for_odr): Declare.
6277 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
6278             Jason Merrill  <jason@redhat.com>
6280         * system.h: Include <cstdlib> as well as <stdlib.h>.
6282 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
6284         PR target/57655
6285         * config/i386/i386.c (construct_container): Report error if
6286         long double is used with disabled x87 float returns.
6288 2013-06-20  Jan Hubicka  <jh@suse.cz>
6290         * lto-cgraph.c (input_symtab): Do not set cgraph state.
6292 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
6294         PR rtl-optimization/57425
6295         PR rtl-optimization/57569
6296         * alias.c (write_dependence_p): Remove parameters mem_mode and
6297         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
6298         Changed all callers.
6299         (canon_anti_dependence): Get comments and semantics in sync.
6300         Add parameter mem_canonicalized.  Changed all callers.
6301         * rtl.h (canon_anti_dependence): Update prototype.
6303 2013-06-20  Richard Biener  <rguenther@suse.de>
6305         * data-streamer-in.c (streamer_read_uhwi): Optimize single
6306         byte case, inline streamer_read_uchar and defer section
6307         overrun check.
6309 2013-06-20  Richard Biener  <rguenther@suse.de>
6311         PR tree-optimization/57584
6312         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6313         SSA names into the expanded expression that take part in
6314         abnormal coalescing.
6316 2013-06-19  Sharad Singhai  <singhai@google.com>
6318         * gcov.c (print_usage): Handle new option.
6319         (process_args): Ditto.
6320         (get_gcov_intermediate_filename): New function.
6321         (output_intermediate_file): New function.
6322         (output_gcov_file): New function
6323         (generate_results): Handle new option.
6324         (release_function): Relase demangled name.
6325         (read_graph_file): Handle demangled name.
6326         (output_lines): Ditto.
6327         * doc/gcov.texi: Document gcov intermediate format.
6329 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
6331         PR bootstrap/57604
6332         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6333         (lra_emit_add): Use the functions.  Add comment about Y as an
6334         address segment.
6336 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
6338         PR driver/57652
6339         * collect2.c (collect_atexit): New.
6340         (collect_exit): Delete.
6341         (main): Register collect_atexit with atexit.
6342         (collect_wait): Change collect_exit to exit.
6343         (do_wait): Same.
6344         * collect2.h (collect_exit): Delete.
6345         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6347 2013-06-19  Wei Mi  <wmi@google.com>
6349         PR rtl-optimization/57518
6350         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6351         if regno is used in paradoxical subreg.
6352         (update_equiv_regs): Check pdx_subregs[regno] before
6353         set a reg to be equivalent with a mem.
6355 2013-06-19  Matthias Klose  <doko@ubuntu.com>
6357         PR driver/57651
6358         * file-find.h (find_a_file): Add a mode parameter.
6359         * file-find.c (find_a_file): Likewise.
6360         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6361         with X_OK for the executables.
6362         * collect2.c (main): Call find_a_file with X_OK.
6364 2013-06-19  Steve Ellcey  <sellcey@mips.com>
6366         PR target/56942
6367         * config/mips/mips.md (casesi_internal_mips16_<mode>):
6368         Use NEXT_INSN instead of next_real_insn.
6370 2013-06-19  Jan Hubicka  <jh@suse.cz>
6372         * cgraph.h (const_value_known_p): Replace by ...
6373         (ctor_for_folding): .. this one.
6374         * cgraphunit.c (process_function_and_variable_attributes): Use it.
6375         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6376         * expr.c (expand_expr_real_1): Likewise.
6377         (string_constant): Likewise.
6378         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6379         * ipa.c (process_references): Likewise.
6380         (symtab_remove_unreachable_nodes): Likewise.
6381         * ipa-inline-analysis.c (param_change_prob): Likewise.
6382         * gimple-fold.c (canonicalize_constructor_val): Likewise.
6383         (get_base_constructor): Likwise.
6384         * varpool.c (varpool_remove_node): Likewise.
6385         (varpool_remove_initializer): LIkewise.
6386         (dump_varpool_node): LIkwise.
6387         (const_value_known_p): Rewrite to ...
6388         (ctor_for_folding): ... this one.
6390 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
6392         PR driver/57651
6393         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6394         PERSONALITY in $PATH derived prefixes.
6396 2013-06-19  Jeff Law  <law@redhat.com>
6398         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6399         in comment.
6401         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6402         (simplify_bitwise_binary): Use it to simpify certain binary ops on
6403         booleans.
6405 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
6407         * config/arm/vfp.md: Move VFP instruction classification documentation
6408         to ...
6409         * config/arm/arm.md: ... here.  Update instruction classification
6410         documentation.
6412 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
6414         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6415         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6416         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
6418 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
6420         PR target/57609
6421         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6422         with NEXT_INSN.  Use tablejump_p to check for jump table data
6423         insns.
6425 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
6427         PR c++/56544
6428         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6429         that now in C++ the value is correct per the C++ standards.
6431 2013-06-19  Richard Biener  <rguenther@suse.de>
6433         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6434         for global context.
6436 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6438         Revert:
6439         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6441         PR target/57609
6442         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6443         with next_active_insn.
6445 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
6447         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6448         functions are inlined during failures to flag an error.
6449         * tree-inline.c (expand_call_inline): Allow the error to be flagged
6450         in early inline pass.
6452 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
6454         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6455         in comments.
6457 2013-06-18  Julian Brown  <julian@codesourcery.com>
6459         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6460         Permit virtual register pre-reload if !strict.
6461         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6462         change.
6463         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6464         prototype.
6465         * config/arm/neon.md (movmisalign<mode>): Use
6466         neon_perm_struct_or_reg_operand instead of
6467         neon_struct_or_register_operand.
6468         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6469         neon_permissive_struct_operand instead of neon_struct_operand.
6470         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6471         neon_vector_mem_operand.
6472         * config/arm/predicates.md (neon_struct_operand): Adjust call to
6473         neon_vector_mem_operand.
6474         (neon_permissive_struct_operand): New.
6475         (neon_struct_or_register_operand): Rename to...
6476         (neon_perm_struct_or_reg_operand): This. Adjust call to
6477         neon_vector_mem_operand.
6479 2013-06-18  Richard Biener  <rguenther@suse.de>
6481         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6482         * lto-streamer.h: Include pointer-set.h.
6483         (struct lto_decl_slot): Remove.
6484         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6485         Remove next_index entry.
6486         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6487         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6488         (lto_init_tree_ref_encoder): Adjust.
6489         (lto_destroy_tree_ref_encoder): Likewise.
6490         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6491         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6492         (lto_output_decl_index): Adjust.
6493         (lto_new_out_decl_state): Likewise.
6494         (lto_record_function_out_decl_state): Likewise.
6495         * lto-streamer-out.c (copy_function): Likewise.
6497 2013-06-18  Richard Biener  <rguenther@suse.de>
6499         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6500         * cgraphunit.c: Include cfgloop.h.
6501         (init_lowered_empty_function): Initialize the loop tree.
6502         (assemble_thunk): Insert new BBs into loops.
6504 2013-06-18  Richard Biener  <rguenther@suse.de>
6506         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6507         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6508         the map from cache entry to cache index optional.
6509         (streamer_tree_cache_replace_tree): Adjust accordingly.
6510         (streamer_tree_cache_append): Likewise.
6511         (streamer_tree_cache_delete): Likewise.
6512         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6513         streamer cache map from cache entry to cache index.
6514         * lto-streamer-out.c (create_output_block): Adjust.
6516 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
6518         * config/arm/arm.md (attribute "insn"): Move multiplication and
6519         division attributes to...
6520         (attribute "type"): ... here.  Remove mult.
6521         (attribute "mul32"): New attribute.
6522         (attribute "mul64"): Add umaal.
6523         (*arm_mulsi3): Update attributes.
6524         (*arm_mulsi3_v6): Likewise.
6525         (*thumb_mulsi3): Likewise.
6526         (*thumb_mulsi3_v6): Likewise.
6527         (*mulsi3_compare0): Likewise.
6528         (*mulsi3_compare0_v6): Likewise.
6529         (*mulsi_compare0_scratch): Likewise.
6530         (*mulsi_compare0_scratch_v6): Likewise.
6531         (*mulsi3addsi): Likewise.
6532         (*mulsi3addsi_v6): Likewise.
6533         (*mulsi3addsi_compare0): Likewise.
6534         (*mulsi3addsi_compare0_v6): Likewise.
6535         (*mulsi3addsi_compare0_scratch): Likewise.
6536         (*mulsi3addsi_compare0_scratch_v6): Likewise.
6537         (*mulsi3subsi): Likewise.
6538         (*mulsidi3adddi): Likewise.
6539         (*mulsi3addsi_v6): Likewise.
6540         (*mulsidi3adddi_v6): Likewise.
6541         (*mulsidi3_nov6): Likewise.
6542         (*mulsidi3_v6): Likewise.
6543         (*umulsidi3_nov6): Likewise.
6544         (*umulsidi3_v6): Likewise.
6545         (*umulsidi3adddi): Likewise.
6546         (*umulsidi3adddi_v6): Likewise.
6547         (*smulsi3_highpart_nov6): Likewise.
6548         (*smulsi3_highpart_v6): Likewise.
6549         (*umulsi3_highpart_nov6): Likewise.
6550         (*umulsi3_highpart_v6): Likewise.
6551         (mulhisi3): Likewise.
6552         (*mulhisi3tb): Likewise.
6553         (*mulhisi3bt): Likewise.
6554         (*mulhisi3tt): Likewise.
6555         (maddhisi4): Likewise.
6556         (*maddhisi4tb): Likewise.
6557         (*maddhisi4tt): Likewise.
6558         (maddhidi4): Likewise.
6559         (*maddhidi4tb): Likewise.
6560         (*maddhidi4tt): Likewise.
6561         (divsi3): Likewise.
6562         (udivsi3): Likewise.
6563         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6564         (thumb2_mulsi_short_compare0): Likewise.
6565         (thumb2_mulsi_short_compare0_scratch): Likewise.
6566         * config/arm/arm1020e.md (1020mult1): Update attribute change.
6567         (1020mult2): Likewise.
6568         (1020mult3): Likewise.
6569         (1020mult4): Likewise.
6570         (1020mult5): Likewise.
6571         (1020mult6): Likewise.
6572         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6573         change.
6574         (cortex_a15_mult64): Likewise.
6575         (cortex_a15_sdiv): Likewise.
6576         (cortex_a15_udiv): Likewise.
6577         * config/arm/arm1026ejs.md (mult1): Update attribute change.
6578         (mult2): Likewise.
6579         (mult3): Likewise.
6580         (mult4): Likewise.
6581         (mult5): Likewise.
6582         (mult6): Likewise.
6583         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6584         (pj4_ir_div): Likewise.
6585         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6586         (11_mult2): Likewise.
6587         (11_mult3): Likewise.
6588         (11_mult4): Likewise.
6589         (11_mult5): Likewise.
6590         (11_mult6): Likewise.
6591         (11_mult7): Likewise.
6592         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6593         (cortex_a8_mla): Likewise.
6594         (cortex_a8_mull): Likewise.
6595         (cortex_a8_smulwy): Likewise.
6596         (cortex_a8_smlald): Likewise.
6597         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6598         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6599         (cortex_r4_mul_3): Likewise.
6600         (cortex_r4_mla_4): Likewise.
6601         (cortex_r4_mla_3): Likewise.
6602         (cortex_r4_smlald): Likewise.
6603         (cortex_r4_mull): Likewise.
6604         (cortex_r4_sdiv): Likewise.
6605         (cortex_r4_udiv): Likewise.
6606         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6607         (cortex_a7_idiv): Likewise.
6608         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6609         (9_mult2): Likewise.
6610         (9_mult3): Likewise.
6611         (9_mult4): Likewise.
6612         (9_mult5): Likewise.
6613         (9_mult6): Likewise.
6614         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6615         (cortex_a53_sdiv): Likewise.
6616         (cortex_a53_udiv): Likewise.
6617         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6618         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6619         (mp626_mult2): Likewise.
6620         (mp626_mult3): Likewise.
6621         (mp626_mult4): Likewise.
6622         * config/arm/fa526.md (526_mult1): Update attribute change.
6623         (526_mult2): Likewise.
6624         * config/arm/arm-generic.md (mult): Update attribute change.
6625         (mult_ldsched_strongarm): Likewise.
6626         (mult_ldsched): Likewise.
6627         (multi_cycle): Likewise.
6628         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6629         * config/arm/fa606te.md (606te_mult1): Update attribute change.
6630         (606te_mult2): Likewise.
6631         (606te_mult3): Likewise.
6632         (606te_mult4): Likewise.
6633         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6634         (cortex_a9_mac16): Likewise.
6635         (cortex_a9_multiply): Likewise.
6636         (cortex_a9_mac): Likewise.
6637         (cortex_a9_multiply_long): Likewise.
6638         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6639         (626te_mult2): Likewise.
6640         (626te_mult3): Likewise.
6641         (626te_mult4): Likewise.
6643 2013-06-18  Richard Biener  <rguenther@suse.de>
6645         PR lto/57334
6646         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6648 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6650         PR target/57609
6651         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6652         with next_active_insn.
6654 2013-06-18  Alan Modra  <amodra@gmail.com>
6656         * config/rs6000/rs6000.h (enum data_align): New.
6657         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6658         (DATA_ABI_ALIGNMENT): Define.
6659         (CONSTANT_ALIGNMENT): Correct comment.
6660         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6661         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6663 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6665         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6666         ATTRIBUTE_UNUSED marking.
6668 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6670         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6671         alternative and update.
6672         (aarch64_dup_lanedi): Delete.
6673         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6674         * config/aarch64/aarch64-simd-builtins.def: Update.
6676 2013-06-17  Richard Biener  <rguenther@suse.de>
6678         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6679         (lto_input_scc): Declare.
6680         (lto_input_tree_1): Likewise.
6681         (struct lto_stats_d): Add num_tree_bodies_output and
6682         num_pickle_refs_output.
6683         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6684         (lto_read_tree_1): Split out from ...
6685         (lto_read_tree): ... this.
6686         (lto_input_scc): New function.
6687         (lto_input_tree_1): Split out from ...
6688         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6689         (lto_data_in_create): Create the streamer cache without hashes.
6690         * lto-streamer-out.c (create_output_block): Create the streamer
6691         cache with hashes when not doing WPA.
6692         (lto_write_tree_1): Split out from ...
6693         (lto_write_tree): ... this.
6694         (get_symbol_initial_value): New function.
6695         (lto_output_tree_1): Split out from ...
6696         (lto_output_tree): ... this.  Write trees as series of SCCs
6697         using a DFS walk via DFS_write_tree.
6698         (struct sccs, struct scc_entry): New types.
6699         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6700         (DFS_write_tree_body): New function.
6701         (DFS_write_tree): Likewise.
6702         (hash_tree): Likewise.
6703         (scc_entry_compare): Likewise.
6704         (hash_scc): Likewise.
6705         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6706         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6707         TREE_CHAIN as regular reference.
6708         (streamer_read_integer_cst): Remove.
6709         (streamer_get_pickled_tree): Adjust.
6710         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6711         of DECL_EXTERNALs in BLOCK_VARS for now.
6712         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6713         reference.
6714         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6715         Add hash value argument and record that if hashes are recorded
6716         in the cache.
6717         (streamer_tree_cache_insert_1): Adjust.
6718         (streamer_tree_cache_insert): Likewise.
6719         (streamer_tree_cache_insert_at): Rename to ...
6720         (streamer_tree_cache_replace_tree): ... this and adjust.
6721         (streamer_tree_cache_append): Adjust.
6722         (record_common_node): Likewise.
6723         (streamer_tree_cache_create): Add argument whether to
6724         record hash values together with trees.
6725         (streamer_tree_cache_delete): Adjust.
6726         * tree-streamer.h (struct streamer_tree_cache_d): Add
6727         vector of hashes.
6728         (streamer_read_integer_cst): Remove.
6729         (streamer_tree_cache_insert): Adjust.
6730         (streamer_tree_cache_append): Likewise.
6731         (streamer_tree_cache_insert_at): Rename to ...
6732         (streamer_tree_cache_replace_tree): ... this and adjust.
6733         (streamer_tree_cache_create): Add argument whether to record hashes.
6734         (streamer_tree_cache_get): Rename to ...
6735         (streamer_tree_cache_get_tree): ... this.
6736         (streamer_tree_cache_get_hash): New function.
6737         * tree.c (cache_integer_cst): New function.
6738         * tree.h (cache_integer_cst): Declare.
6739         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6740         * lto-symtab.c (lto_varpool_replace_node): Only release
6741         DECL_INITIAL of non-prevailing decls.
6742         * varpool.c (varpool_remove_initializer): Do not release
6743         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6745 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6747         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6748         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6749         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6750         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6751         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6752         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6753         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6754         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6755         instead of TARGET_64BIT.
6756         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6757         Require ISA_HAS_<D>DIV.
6759 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6761         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6762         (mips*-*-linux*): Move default with_llsc setting to where other
6763         defaults are set.
6764         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6765         with_arch block.
6766         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6767         Likewise.  Remove default with_tune setting.  Move default float
6768         setting to its own block.  Handle with_llsc in the same block as above.
6770 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
6772         PR rtl-optimization/57425
6773         PR rtl-optimization/57569
6774         * alias.c (write_dependence_p): Add new parameters mem_mode,
6775         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6776         Changed all callers.
6777         (canon_anti_dependence): New function.
6778         * cse.c (check_dependence): Use canon_anti_dependence.
6779         * cselib.c (cselib_invalidate_mem): Likewise.
6780         * rtl.h (canon_anti_dependence): Declare.
6782 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6784         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6785         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6786         ".set mips3" for 64-bit targets.
6788 2013-06-15  Dehao Chen  <dehao@google.com>
6790         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6791         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6792         (gimple_check_call_args): Likewise.
6793         * value-prof.c (check_ic_target): Likewise.
6794         * ipa-inline.c (early_inliner): Likewise.
6795         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6796         * cgraph.c (cgraph_create_edge_1): Likewise.
6797         (cgraph_make_edge_direct): Likewise.
6799 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6801         PR target/57615
6802         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6803         rs6000_output_move_128bit to handle emitting quad memory
6804         operations.  Set attribute length to 8 bytes.
6806 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
6808         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6809         New pattern.
6810         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6811         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6812         (aarch64_<su>mlsl<mode>): Likewise.
6814 2013-06-14  Mike Stump  <mikestump@comcast.net>
6816         * Makefile.in (TARGET_H): Add insn-codes.h.
6818 2013-06-14  Alan Modra  <amodra@gmail.com>
6820         PR middle-end/57134
6821         PR middle-end/57586
6822         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6823         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6824         bitfield expansion when EXPAND_MEMORY.
6825         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6827 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6829         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6830         test for clearing quad memory on 32-bit later.
6832 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6834         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6835         (fold_negate_expr): Likewise.
6836         (fold_real_zero_addition_p): Handle vectors.
6837         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6839 2013-06-14  Alan Modra  <amodra@gmail.com>
6841         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6843 2013-06-13  Jan Hubicka  <jh@suse.cz>
6845         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6846         Local comdats are not externally visible.
6847         * symtab.c (dump_symtab_base): Dump externally visible.
6848         (verify_symtab_base): Verify back links in the symtab hash.
6850 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6852         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6853         CONVERT_EXPR as equal nodes.
6855 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6857         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6859 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6861         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6862         Generalize to complex and vector.
6863         * tree.c (build_all_ones_cst): New function.
6864         * tree.h (build_all_ones_cst): Declare it.
6866 2013-06-13  Alan Modra  <amodra@gmail.com>
6868         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6869         * config/rs6000/rs6000.md (signbittf2): New insn.
6870         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6871         (abstf2_internal, cmptf_internal2): Likewise.
6872         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6874 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6875             Pat Haugen  <pthaugen@us.ibm.com>
6876             Peter Bergner  <bergner@vnet.ibm.com>
6878         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6879         power8 byte, half-word, and quad-word atomic instructions.
6880         (emit_store_conditional): Likewise.
6881         (rs6000_expand_atomic_compare_and_swap): Likewise.
6882         (rs6000_expand_atomic_op): Likewise.
6884         * config/rs6000/sync.md (larx): Add new modes for power8.
6885         (stcx): Likewise.
6886         (AINT): New mode iterator to include TImode as well as normal
6887         integer modes on power8.
6888         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6889         that VSX registers are not considered.  Use AINT mode iterator
6890         instead of INT1 to allow inclusion of quad word atomic operations
6891         on power8.
6892         (load_locked<mode>): Likewise.
6893         (store_conditional<mode>): Likewise.
6894         (atomic_compare_and_swap<mode>): Likewise.
6895         (atomic_exchange<mode>): Likewise.
6896         (atomic_nand<mode>): Likewise.
6897         (atomic_fetch_<fetchop_name><mode>): Likewise.
6898         (atomic_nand_fetch<mode>): Likewise.
6899         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6900         each type.
6901         (ATOMIC): On power8, add QImode, HImode modes.
6902         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6903         modes that promote to SImode.
6904         (load_lockedti): Convert TImode arguments to PTImode, so that we
6905         get a guaranteed even/odd register pair.
6906         (load_lockedpti): Likewise.
6907         (store_conditionalti): Likewise.
6908         (store_conditionalpti): Likewise.
6910         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6911         atomic load/store instructions.
6912         (HSI): Likewise.
6914 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6916         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6917         loads.
6918         (insn_count): New attribute, with most cases extracted from...
6919         (length): ...here.  Redefine most cases in terms of insn_count.
6920         (single_insn): Delete.
6921         (can_delay): Use insn_count to check for single instructions.
6922         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6923         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6924         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6925         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6926         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6927         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6928         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6929         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6930         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6931         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6932         rather than "length".
6933         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6934         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6935         Use "insn_count" rather than "length".
6936         * config/mips/mips-dsp.md
6937         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6938         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6939         length attributes.
6941 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6943         PR tree-optimization/57361
6944         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6946 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6948         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6949         to split.
6950         (aarch64_simd_combine<mode>): New instruction expansion.
6951         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6952         function prototype.
6953         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6954         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6956 2013-06-12  Jan Hubicka  <jh@suse.cz>
6958         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6959         decl has when in streaming stage.
6960         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6961         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6963 2013-06-12  Roland Stigge  <stigge@antcom.de>
6965         PR target/57578
6966         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6968 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6970         PR tree-optimization/57537
6971         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6972         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6974 2013-06-12  Richard Biener  <rguenther@suse.de>
6976         * data-streamer.h (streamer_write_char_stream): CSE
6977         obs->current_pointer.
6978         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6979         streamer_write_char_stream manually and optimize the resulting loop.
6980         (streamer_write_hwi_stream): Likewise.
6982 2013-06-12  Jan Hubicka  <jh@suse.cz>
6984         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6985         * cgraph.h (varpool_create_empty_node): Declare.
6986         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6987         duplicated nodes.
6988         * symtab.c (symtab_unregister_node): Be lax about missin entries
6989         in node hash.
6990         (symtab_get_node): Update comment.
6991         * varpool.c (varpool_create_empty_node): Break out from ...
6992         (varpool_node_for_decl): ... here.
6993         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6995 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6997         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6998         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6999         part.  Use straight-line flow at the end.
7000         <COMPONENT_REF>: Remove superfluous else.
7001         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7003 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7005         PR target/56564
7006         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7007         target hook even for !TREE_PUBLIC decls.  If no resolution info
7008         is available, return false for common and external decls.
7010 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7012         * config/rl78/constraints.md (U): New constraint.
7013         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7014         valloc attribute.
7016 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7018         PR target/57589
7019         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7020         to allow returning address to AT_PLATFORM name.
7022 2013-06-11  Jan Hubicka  <jh@suse.cz>
7024         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7025         * cgraph.h (symtab_node_base): Add weakref flag.
7026         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7027         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7028         (output_weakrefs): Use weakref flag.
7029         * fold-const.c (simple_operand_p): Handle WEAK.
7030         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7031         * ipa.c (varpool_externally_visible_p): Drop weakref.
7032         (function_and_variable_visibility): Update comment; fix weakref
7033         sanity checks; do not clear DECL_WEAK on them.
7034         * lto-cgraph.c (lto_output_node): update.
7035         (lto_output_varpool_node): Update.
7036         (input_overwrite_node): Update.
7037         (input_node): Update.
7038         (input_varpool_node): Update.
7039         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7040         (lto_symtab_merge_symbols): Add sanity check.
7041         (lto_symtab_prevailing_decl): Do not special case weakrefs.
7042         * passes.c (rest_of_decl_compilation): Set static flag, too.
7043         * symtab.c (dump_symtab_base): Dump weakref.
7044         (verify_symtab_base): Sanity check weakrefs.
7045         (symtab_make_decl_local): Remove duplicated code.
7046         (symtab_alias_ultimate_target): Simplify.
7047         * varpool.c (varpool_create_variable_alias): Set weakref flag.
7049 2013-06-11  DJ Delorie  <dj@redhat.com>
7051         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7052         (rl78_unwind_word_mode): New.
7054 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7056         * final.c (debug_prefix_maps): Make static.
7058 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7060         * function.c (initial_trampoline): Remove stray copy.
7062 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
7064         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7066 2013-06-11  Martin Jambor  <mjambor@suse.cz>
7068         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7069         within bounds at the beginning of the function.
7071 2013-06-11  Alan Modra  <amodra@gmail.com>
7073         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
7074         reporting.
7075         (get_named_section): Don't NULL !DECL_P decl.
7077 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
7079         * doc/invoke.texi (core-avx2): Document.
7080         (slm): Likewise.
7081         (atom): Updated with MOVBE.
7083 2013-06-11  Richard Biener  <rguenther@suse.de>
7085         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7087 2013-06-11  Anton Blanchard  <anton@samba.org>
7089         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7090         correct shift value in little-endian mode.
7092 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
7094         PR target/56564
7095         * varasm.c (get_variable_align): Move #endif to the right place.
7097 2013-06-10  Cary Coutant  <ccoutant@google.com>
7099         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7100         for hash so that hash table traversal order is deterministic.
7102 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
7103             Pat Haugen  <pthaugen@us.ibm.com>
7104             Peter Bergner  <bergner@vnet.ibm.com>
7106         * config/rs6000/vector.md (GPR move splitter): Do not split moves
7107         of vectors in GPRS if they are direct moves or quad word load or
7108         store moves.
7110         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7111         declaration.
7112         (direct_move_p): Likewise.
7113         (quad_load_store_p): Likewise.
7115         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7116         classes into bins based on the physical register type.
7117         (reg_class_to_reg_type): Likewise.
7118         (IS_STD_REG_TYPE): Likewise.
7119         (IS_FP_VECT_REG_TYPE): Likewise.
7120         (reload_fpr_gpr): Arrays to determine what insn to use if we can
7121         use direct move instructions.
7122         (reload_gpr_vsx): Likewise.
7123         (reload_vsx_gpr): Likewise.
7124         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7125         information that is a simplification of register classes.  Also
7126         precalculate direct move reload helpers.
7127         (direct_move_p): New function to return true if the operation can
7128         be done as a direct move instruciton.
7129         (quad_load_store_p): New function to return true if the operation
7130         is a quad memory operation.
7131         (rs6000_legitimize_address): If quad memory, only allow register
7132         indirect for TImode addresses.
7133         (rs6000_legitimate_address_p): Likewise.
7134         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7135         (rs6000_reload_register_type): Likewise.
7136         (register_to_reg_type): Return register type.
7137         (rs6000_secondary_reload_simple_move): New helper function for
7138         secondary reload and secondary memory needed to identify anything
7139         that is a simple move, and does not need reloading.
7140         (rs6000_secondary_reload_direct_move): New helper function for
7141         secondary reload to identify cases that can be done with several
7142         instructions via the direct move instructions.
7143         (rs6000_secondary_reload_move): New helper function for secondary
7144         reload to identify moves between register types that can be done.
7145         (rs6000_secondary_reload): Add support for quad memory operations
7146         and for direct move.
7147         (rs6000_secondary_memory_needed): Likewise.
7148         (rs6000_debug_secondary_memory_needed): Change argument names.
7149         (rs6000_output_move_128bit): New function to return the move to
7150         use for 128-bit moves, including knowing about the various
7151         limitations of quad memory operations.
7153         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7154         memory operations.  call rs6000_output_move_128bit for the actual
7155         instruciton(s) to generate.
7156         (vsx_movti_64bit): Likewise.
7158         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7159         (UNSPEC_P8V_MTVSRWZ): Likewise.
7160         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7161         (UNSPEC_P8V_MTVSRD): Likewise.
7162         (UNSPEC_P8V_XXPERMDI): Likewise.
7163         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7164         (UNSPEC_FUSION_GPR): Likewise.
7165         (FMOVE128_GPR): New iterator for direct move.
7166         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7167         (f32_sv): Likewise.
7168         (f32_dm): Likewise.
7169         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7170         loads and direct move instructions.
7171         (zero_extendsidi2_lfiwzx): Likewise.
7172         (extendsidi2_lfiwax): Likewise.
7173         (extendsidi2_nocell): Likewise.
7174         (floatsi<mode>2_lfiwax): Likewise.
7175         (lfiwax): Likewise.
7176         (floatunssi<mode>2_lfiwzx): Likewise.
7177         (lfiwzx): Likewise.
7178         (fix_trunc<mode>_stfiwx): Likewise.
7179         (fixuns_trunc<mode>_stfiwx): Likewise.
7180         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7181         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7182         (parity<mode>2_cmpb): Set length/type attr.
7183         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7184         for 'mr.' to fast_compare.
7185         (bpermd_<mode>): Change type attr to popcnt.
7186         (p8_fmrgow_<mode>): New insns for power8 direct move support.
7187         (p8_mtvsrwz_1): Likewise.
7188         (p8_mtvsrwz_2): Likewise.
7189         (reload_fpr_from_gpr<mode>): Likewise.
7190         (p8_mtvsrd_1): Likewise.
7191         (p8_mtvsrd_2): Likewise.
7192         (p8_xxpermdi_<mode>): Likewise.
7193         (reload_vsx_from_gpr<mode>): Likewise.
7194         (reload_vsx_from_gprsf): Likewise.
7195         (p8_mfvsrd_3_<mode>): LIkewise.
7196         (reload_gpr_from_vsx<mode>): Likewise.
7197         (reload_gpr_from_vsxsf): Likewise.
7198         (p8_mfvsrd_4_disf): Likewise.
7199         (multi-word GPR splits): Do not split direct moves or quad memory
7200         operations.
7202 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
7204         * tree-into-ssa.c (interesting_blocks): Make static.
7206 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
7208         PR target/56564
7209         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7210         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7211         Use DATA_ABI_ALIGNMENT for that case instead if defined.
7212         (get_variable_align): New function.
7213         (get_variable_section, emit_bss, emit_common,
7214         assemble_variable_contents, place_block_symbol): Use
7215         get_variable_align instead of DECL_ALIGN.
7216         (assemble_noswitch_variable): Add align argument, use it
7217         instead of DECL_ALIGN.
7218         (assemble_variable): Adjust caller.  Use get_variable_align
7219         instead of DECL_ALIGN.
7220         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7221         caller.
7222         (DATA_ABI_ALIGNMENT): Define.
7223         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7224         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
7225         opt is false, only return the psABI mandated alignment increase.
7226         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7227         (DATA_ABI_ALIGNMENT): ... this.
7228         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7229         (DATA_ABI_ALIGNMENT): ... this.
7230         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7231         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7232         (DATA_ABI_ALIGNMENT): ... this.
7233         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7234         * doc/tm.texi: Regenerated.
7236 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
7238         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7239         cmp_code to construct REG_EQUAL note.
7241 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
7243         PR target/57568
7244         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7245         that operands[2] doesn't overlap with operands[0].
7247 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
7248             Jan Hubicka  <jh@suse.cz>
7250         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7251         hack to mark symbols as used.
7253 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
7255         PR rtl-optimization/57559
7256         * lra-constraints.c (process_alt_operands): Don't discourage
7257         memory with known offset for offsetable memory constraint.
7258         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7260 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
7262         * varasm.c (struct oc_local_state): Reorder fields.
7263         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7264         and adjust accordingly.
7265         (output_constructor): Reorder initialization code and adjust call to
7266         output_constructor_bitfield.
7268 2013-06-07  Jan Hubicka  <jh@suse.cz>
7270         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7272 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
7274         * tree-object-size.c (unknown): Make const.
7276 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7278         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7279         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7280         for last alternative in the cpu_facility attribute.
7282 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7284         PR target/56315
7285         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7286         (xordi3): Change operand 2 constraint to arm_xordi_operand.
7287         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7288         * config/arm/constraints.md (Dg): New constraint.
7289         * config/arm/neon.md (xordi3_neon): Remove.
7290         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7291         * config/arm/predicates.md (arm_xordi_operand): New predicate.
7293 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7295         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7296         Clean up alternatives.
7298 2013-06-07  Alan Modra  <amodra@gmail.com>
7300         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7301         va_list_gpr_size.
7303 2013-06-07  Alan Modra  <amodra@gmail.com>
7305         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7307 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7309         * config/arm/constraints.md (Df): New constraint.
7310         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7311         Correct length attribute for last two alternatives.
7313 2013-06-07  Alan Modra  <amodra@gmail.com>
7315         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7316         override user -mfp-in-toc.
7317         (offsettable_ok_by_alignment): Consider just the current access
7318         rather than the whole object, unless BLKmode.  Handle
7319         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7320         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7321         for -mcmodel=medium.
7322         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7323         override user -mfp-in-toc or -msum-in-toc.  Default to
7324         -mno-fp-in-toc for -mcmodel=medium.
7326 2013-06-06  DJ Delorie  <dj@redhat.com>
7328         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7329         TARGET_VALID_POINTER_MODE.
7331 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
7332             Pat Haugen  <pthaugen@us.ibm.com>
7333             Peter Bergner  <bergner@vnet.ibm.com>
7335         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7336         Document new power8 builtins.
7338         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7339         condition code register, to allow 128-bit logical operations to be
7340         done in the VSX or GPR registers.
7341         (nor<mode>3): Use the canonical form for nor.
7342         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7343         vclz*, and vpopcnt* vector instructions.
7344         (nand<mode>3): Likewise.
7345         (orc<mode>3): Likewise.
7346         (clz<mode>2): LIkewise.
7347         (popcount<mode>2): Likewise.
7349         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7350         that only the GPRs are recognized.
7352         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7353         support for new power8 builtins.
7355         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7356         builtin functions.
7357         (xscvdpspn): Likewise.
7358         (vclz): Likewise.
7359         (vclzb): Likewise.
7360         (vclzh): Likewise.
7361         (vclzw): Likewise.
7362         (vclzd): Likewise.
7363         (vpopcnt): Likewise.
7364         (vpopcntb): Likewise.
7365         (vpopcnth): Likewise.
7366         (vpopcntw): Likewise.
7367         (vpopcntd): Likewise.
7368         (vgbbd): Likewise.
7369         (vmrgew): Likewise.
7370         (vmrgow): Likewise.
7371         (eqv): Likewise.
7372         (eqv_v16qi3): Likewise.
7373         (eqv_v8hi3): Likewise.
7374         (eqv_v4si3): Likewise.
7375         (eqv_v2di3): Likewise.
7376         (eqv_v4sf3): Likewise.
7377         (eqv_v2df3): Likewise.
7378         (nand): Likewise.
7379         (nand_v16qi3): Likewise.
7380         (nand_v8hi3): Likewise.
7381         (nand_v4si3): Likewise.
7382         (nand_v2di3): Likewise.
7383         (nand_v4sf3): Likewise.
7384         (nand_v2df3): Likewise.
7385         (orc): Likewise.
7386         (orc_v16qi3): Likewise.
7387         (orc_v8hi3): Likewise.
7388         (orc_v4si3): Likewise.
7389         (orc_v2di3): Likewise.
7390         (orc_v4sf3): Likewise.
7391         (orc_v2df3): Likewise.
7393         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7394         allow power8 quad mode in 64-bit.
7395         (rs6000_builtin_vectorized_function): Add support to vectorize
7396         ISA 2.07 count leading zeros, population count builtins.
7397         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7398         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7399         (builtin_function_type): Add vgbbd builtin function which takes an
7400         unsigned argument.
7401         (altivec_expand_vec_perm_const): Add support for new power8 merge
7402         instructions.
7404         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7405         that does not include TImdoe for use with 32-bit.
7406         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7407         instructions.
7408         (UNSPEC_VSX_CVDPSPN): Likewise.
7409         (vsx_xscvdpspn): Likewise.
7410         (vsx_xscvspdpn): Likewise.
7411         (vsx_xscvdpspn_scalar): Likewise.
7412         (vsx_xscvspdpn_directmove): Likewise.
7413         (vsx_and<mode>3): Split logical operations into 32-bit and
7414         64-bit. Add support to do logical operations on TImode as well as
7415         VSX vector types.  Allow logical operations to be done in either
7416         VSX registers or in general purpose registers in 64-bit mode.  Add
7417         splitters if GPRs were used. For AND, add clobber of CCmode to
7418         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
7419         encoding.
7420         (vsx_and<mode>3_32bit): Likewise.
7421         (vsx_and<mode>3_64bit): Likewise.
7422         (vsx_ior<mode>3): Likewise.
7423         (vsx_ior<mode>3_32bit): Likewise.
7424         (vsx_ior<mode>3_64bit): Likewise.
7425         (vsx_xor<mode>3): Likewise.
7426         (vsx_xor<mode>3_32bit): Likewise.
7427         (vsx_xor<mode>3_64bit): Likewise.
7428         (vsx_one_cmpl<mode>2): Likewise.
7429         (vsx_one_cmpl<mode>2_32bit): Likewise.
7430         (vsx_one_cmpl<mode>2_64bit): Likewise.
7431         (vsx_nor<mode>3): Likewise.
7432         (vsx_nor<mode>3_32bit): Likewise.
7433         (vsx_nor<mode>3_64bit): Likewise.
7434         (vsx_andc<mode>3): Likewise.
7435         (vsx_andc<mode>3_32bit): Likewise.
7436         (vsx_andc<mode>3_64bit): Likewise.
7437         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7438         and xxlorc instructions.
7439         (vsx_eqv<mode>3_64bit): Likewise.
7440         (vsx_nand<mode>3_32bit): Likewise.
7441         (vsx_nand<mode>3_64bit): Likewise.
7442         (vsx_orc<mode>3_32bit): Likewise.
7443         (vsx_orc<mode>3_64bit): Likewise.
7445         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7447         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7448         instruction.
7449         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7450         (p8_vmrgow): Likewise.
7451         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7452         GPRs to be split under VSX.
7453         (p8v_clz<mode>2): Add power8 count leading zero support.
7454         (p8v_popcount<mode>2): Add power8 population count support.
7455         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7456         support.
7458         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7459         instruction.
7461         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7462         builtin functions.
7463         (vec_nand): Likewise.
7464         (vec_vclz): Likewise.
7465         (vec_vclzb): Likewise.
7466         (vec_vclzd): Likewise.
7467         (vec_vclzh): Likewise.
7468         (vec_vclzw): Likewise.
7469         (vec_vgbbd): Likewise.
7470         (vec_vmrgew): Likewise.
7471         (vec_vmrgow): Likewise.
7472         (vec_vpopcnt): Likewise.
7473         (vec_vpopcntb): Likewise.
7474         (vec_vpopcntd): Likewise.
7475         (vec_vpopcnth): Likewise.
7476         (vec_vpopcntw): Likewise.
7478 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7480         PR rtl-optimization/57468
7481         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7482         spilled pseudos.
7484 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7486         PR rtl-optimization/57459
7487         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7488         type when setting live regs.
7490 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7492         * config/s390/s390.opt (mlra): New option.
7493         * config/s390/s390.c (s390_decompose_address): Check displacement
7494         for all registers for LRA.
7495         (s390_secondary_reload): Don't used secondary reloads for LRA.
7496         (s390_lra_p): New function.
7497         (TARGET_LRA_P): Define.
7498         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7499         of attribute cpu_facility to zarch for the last alternative.
7500         (*cmpmem_short): Ditto.
7502 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
7504         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7505         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
7506         (arm_expand_prologue): Likewise.
7508 2013-06-06  Teresa Johnson  <tejohnson@google.com>
7510         PR c++/53743
7511         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7512         as this is now done by redirect_edge_and_branch_force.
7513         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7514         barriers, and fix interaction with splitting.
7515         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7516         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7517         reflect changes made in the routine.
7518         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7519         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7520         since this is called in cfglayout mode, and replace partition fixup
7521         with assert as that is now done by force_nonfallthru_and_redirect.
7522         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7523         already be marked with region crossing note.
7524         (insert_section_boundary_note): Make non-static, gate on flag
7525         has_bb_partition, rewrite to also check for multiple partitions.
7526         (rest_of_handle_reorder_blocks): Remove call to
7527         insert_section_boundary_note, now done later during free_cfg.
7528         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7529         * bb-reorder.h (insert_section_boundary_note): Declare.
7530         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7531         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7532         invoke insert_section_boundary_note.
7533         (try_redirect_by_replacing_jump): Remove unnecessary
7534         check for region crossing note.
7535         (fixup_partition_crossing): New function.
7536         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7537         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7538         in non-cfglayout mode.
7539         (force_nonfallthru_and_redirect): Fixup partition boundaries,
7540         remove old code that tried to do this. Emit barrier correctly
7541         when we are in cfglayout mode.
7542         (last_bb_in_partition): New function.
7543         (rtl_split_edge): Correctly fixup partition boundaries.
7544         (commit_one_edge_insertion): Remove old code that tried to
7545         fixup region crossing edge since this is now handled in
7546         split_block, and set up insertion point correctly since
7547         block may now end in a jump.
7548         (verify_hot_cold_block_grouping): Guard against checking when not in
7549         linearized RTL mode.
7550         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7551         notes.
7552         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7553         rtl_verify_flow_info, so not called in cfglayout mode.
7554         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7555         (fixup_reorder_chain): Remove old code that attempted to fixup region
7556         crossing note as this is now handled in force_nonfallthru_and_redirect.
7557         (duplicate_insn_chain): Don't duplicate switch section notes.
7558         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7559         note.
7560         * basic-block.h (emit_barrier_after_bb): Declare.
7562 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7564         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7565         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7566         arm_usatsihi): Adjust alternatives for arm_restrict_it.
7568 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7570         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7571         where appropriate.
7572         * config/arm/ldmstm.md: Regenerate.
7574 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7576         * config/arm/sync.md (atomic_loaddi_1):
7577         Disable predication for arm_restrict_it.
7578         (arm_load_exclusive<mode>): Likewise.
7579         (arm_load_exclusivesi): Likewise.
7580         (arm_load_exclusivedi): Likewise.
7581         (arm_load_acquire_exclusive<mode>): Likewise.
7582         (arm_load_acquire_exclusivesi): Likewise.
7583         (arm_load_acquire_exclusivedi): Likewise.
7584         (arm_store_exclusive<mode>): Likewise.
7585         (arm_store_exclusive<mode>): Likewise.
7586         (arm_store_release_exclusivedi): Likewise.
7587         (arm_store_release_exclusive<mode>): Likewise.
7589 2013-06-06  Richard Biener  <rguenther@suse.de>
7591         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7592         after LTO_null.
7593         (lto_tag_is_tree_code_p): Adjust.
7594         (lto_tag_is_gimple_code_p): Likewise.
7595         (lto_gimple_code_to_tag): Likewise.
7596         (lto_tag_to_gimple_code): Likewise.
7597         (lto_tree_code_to_tag): Likewise.
7598         (lto_tag_to_tree_code): Likewise.
7599         * data-streamer.h (streamer_write_hwi_in_range): Use
7600         uhwi streaming to stream the normalized range.
7601         (streamer_read_hwi_in_range): Likewise.
7603 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7605         * config/arm/arm.md (enabled_for_depr_it): New attribute.
7606         (predicable_short_it): Likewise.
7607         (predicated): Likewise.
7608         (enabled): Handle above.
7609         (define_cond_exec): Set predicated attribute to yes.
7611 2013-06-05  Mike Stump  <mikestump@comcast.net>
7613         * gdbinit.in (__FUNCTION__): Add.
7615 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
7617         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7618         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7620 2013-06-05  Jan Hubicka  <jh@suse.cz>
7622         * varasm.c (mark_decl_referenced): Revert the removal until targets
7623         are fixed.
7625 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
7627         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7628         instead of mark_decl_referenced.
7630 2013-06-05  Jan Hubicka  <jh@suse.cz>
7632         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7633         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7634         and symtab_used_from_object_file_p.
7635         (cgraph_make_node_local_1): Clear forced_by_abi.
7636         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7637         * cgraph.h (symtab_node_base): Add forced_by_abi.
7638         (decide_is_variable_needed): Remove.
7639         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7640         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7641         (decide_is_symbol_needed): ... this one; handle symbols in general;
7642         always analyze virtuals; honnor forced_by_abi.
7643         (cgraph_finalize_function): Update.
7644         (varpool_finalize_decl): Update.
7645         (symbol_defined_and_needed): Remove.
7646         (analyze_functions): Update.
7647         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7648         output_refs, input_overwrite_node): Handle forced_by_abi.
7649         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7650         (address_taken_from_non_vtable_p): ... this one.
7651         (comdat_can_be_unshared_p_1): New function.
7652         (cgraph_comdat_can_be_unshared_p): Rename to ...
7653         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7654         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7655         (function_and_variable_visibility): Clear forced_by_abi as needed.
7656         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7657         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7658         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7659         * varpool.c (decide_is_variable_needed): Remove.
7661 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7663         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7664         (arm_option_override): Override arm_restrict_it where appropriate.
7665         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7666         * config/arm/arm.opt (mrestrict-it): New command-line option.
7667         * doc/invoke.texi: Document -mrestrict-it.
7669 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7671         * tsan.c (tsan_atomic_table): Make const.
7673 2013-06-05  Richard Biener  <rguenther@suse.de>
7675         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7676         index associated with the tree we are supposed to replace.
7677         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7678         TREE_ASM_WRITTEN as zero for everything but SSA names.
7680 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7682         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7684 2013-06-04  Jan Hubicka  <jh@suse.cz>
7686         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7687         (input_node, input_varpool_node): Handle correctly external same
7688         body aliases.
7689         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7690         nodes at ltrans stage.
7692 2013-06-04  Jan Hubicka  <jh@suse.cz>
7694         * ipa-inline.c (update_caller_keys): Fix availability test.
7695         (update_callee_keys): Likewise.
7696         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7697         to follow ELF standard.
7699 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7701         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7702         (mips64r5900el-*-elf*): New configurations.
7703         * config/mips/mips-cpus.def (r5900): New processor.
7704         * config/mips/mips-tables.opt: Regenerate.
7705         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7706         (mips_issue_rate): Handle PROCESSOR_R5900.
7707         (mips_reorg_process_insns): Force reorder mode for the R5900.
7708         * config/mips/mips.h (TARGET_MIPS5900): Define.
7709         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7710         TARGET_MIPS5900.
7711         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7712         TARGET_MIPS5900.
7713         * config/mips/mips.md (processor): Add r5900.
7714         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7716 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7718         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7719         into function to generate MOVI instruction.
7720         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7721         (aarch64_preferred_simd_mode): Turn into wrapper.
7722         (aarch64_output_scalar_simd_mov_immediate): New function.
7723         * config/aarch64/aarch64-protos.h: Add prototype for above.
7725 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7727         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7728         element_char member.
7729         (sizetochar): Return signed char.
7730         (aarch64_simd_valid_immediate): Remove elchar and other
7731         unnecessary variables.
7732         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7733         Calculate element_char as required.
7734         * config/aarch64/aarch64-protos.h: Update and move prototype
7735         for aarch64_output_simd_mov_immediate.
7736         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7737         Update arguments.
7739 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7741         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7742         information completed by aarch64_simd_valid_immediate.
7743         (aarch64_legitimate_constant_p): Update arguments.
7744         (aarch64_simd_valid_immediate): Work with struct rather than many
7745         pointers.
7746         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7747         (aarch64_simd_make_constant): Update arguments.
7748         (aarch64_output_simd_mov_immediate): Work with struct rather than
7749         many pointers.  Output immediate directly rather than as operand.
7750         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7751         Update prototype.
7752         * config/aarch64/constraints.md (Dn): Update arguments.
7754 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7756         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7757         longer static.
7758         (aarch64_simd_immediate_valid_for_move): Remove.
7759         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7760         (aarch64_simd_make_constant): Update call.
7761         (aarch64_output_simd_mov_immediate): Update call.
7762         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7763         Add prototype.
7764         * config/aarch64/constraints.md (Dn): Update call.
7766 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7768         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7769         return type to bool for prototype.
7770         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7771         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7772         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7774 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7776         * config/mips/mips.opt (meva): New.
7777         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7778         (ASM_SPEC): Handle -meva.
7779         * doc/invoke.texi (meva):  Document.
7781 2013-06-04  Alan Modra  <amodra@gmail.com>
7783         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7784         constant output.
7786 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7788         * rtl.def: Add extra fourth optional field to define_cond_exec.
7789         * gensupport.c (process_one_cond_exec): Process attributes from
7790         define_cond_exec.
7791         * doc/md.texi: Document fourth field in define_cond_exec.
7793 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7795         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7796         out the processing order as in store_bit_field_1.
7798 2013-06-04  Jan Hubicka  <jh@suse.cz>
7800         PR middle-end/57500
7801         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7802         non-VAR_DECL node if it does not exist yet.
7804 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7806         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7807         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7808         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7809         target_cpu_default setting.
7811 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7813         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7814         default to -fopt-info=optimized instead of all.
7815         * doc/invoke.texi: Ditto.
7816         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7817         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7818         (execute_vect_slp): Emit BB vectorization success under
7819         MSG_OPTIMIZED_LOCATIONS.
7820         * tree-vect-slp.c (vect_slp_transform_bb): Change
7821         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7822         * tree-vect-loop.c (vect_transform_loop): Ditto.
7824 2013-06-03  Jason Merrill  <jason@redhat.com>
7826         PR c++/57415
7827         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7828         Use TARGET_EXPR for C++.
7830 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7832         PR rtl-optimization/57268
7833         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7834         if DEBUG_INSN_P (insn).
7836         Reapply
7837         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7839         PR rtl-optimization/57268
7840         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7841         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7843 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7845         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7846         (ix86_avoid_lea_for_addr): Likewise.
7847         (exact_dependency_1): Likewise.
7848         (ix86_adjust_cost): Likewise.
7849         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7850         removed.
7851         (do_reorder_for_imul): Fix typo, formatting and
7852         !reload_completed check removed.
7853         (ix86_sched_reorder): Fix typo and formatting.
7854         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7855         list.
7857 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7859         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7861 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7863         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7864         <STRING_CST>: Likewise.
7865         <VECTOR_CST>: Likewise.
7867 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7868             Mikael Morin  <mikael@gcc.gnu.org>
7870         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7871         * config.in: Regenerated.
7872         * configure: Regenerated.
7874 2013-06-01  Jan Hubicka  <jh@suse.cz>
7876         PR middle-end/57366
7877         * cgraphunit.c (compile): When weakref is not supported,
7878         set up transparent aliases before final output pass.
7879         * varasm.c (assemble_alias): Do not try to do it here.
7881 2013-06-01  Jan Hubicka  <jh@suse.cz>
7883         PR middle-end/57467
7884         * passes.c (for_per_function): Skip unanalyzed functions.
7886 2013-06-01  Jan Hubicka  <jh@suse.cz>
7888         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7889         (lto_symtab_merge_symbols_1): ... this one.
7890         (lto_symtab_merge_cgraph_nodes): Rename to ...
7891         (lto_symtab_merge_symbols): ... this one; simplify.
7892         * cgraph.c (same_body_aliases_done): Rename to ...
7893         (cpp_implicit_aliases_done): ... this one.
7894         (cgraph_create_function_alias): Update.
7895         (cgraph_same_body_alias): Update.
7896         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7897         (verify_edge_corresponds_to_fndecl): Simplify.
7898         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7899         (cgraph_node): Remove same_body_alias.
7900         (varpool_node): Remove alias_of and extra_name_alias.
7901         (same_body_aliases_done): Rename to ..
7902         (cpp_implicit_aliases_done): ... this one.
7903         (symtab_alias_ultimate_target): Add default parameter.
7904         (symtab_resolve_alias): New function.
7905         (fixup_same_cpp_alias_visibility): Declare.
7906         (cgraph_function_node): Add default parameter.
7907         (cgraph_node_asm_name): Likewise.
7908         (cgraph_function_or_thunk_node): Add default parameter; do
7909         not ICE when it is NULL.
7910         (varpool_variable_node): Likewise.
7911         * tree-emutls.c (create_emultls_var): Update.
7912         (ipa_lower_emutls): Update.
7913         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7914         (cgraph_reset_node): Reset alias info.
7915         (cgraph_finalize_function): Update.
7916         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7917         (analyze_function): Simplify.
7918         (cgraph_process_same_body_aliases): Simplify.
7919         (analyze_functions): Fixup same body aliases.
7920         (handle_alias_pairs): Simplify.
7921         (assemble_thunk): Update.
7922         (assemble_thunks_and_aliases): Update.
7923         (output_weakrefs): Rewrite.
7924         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7925         (lto_output_varpool_node): Likewise.
7926         (compute_ltrans_boundary): Remve assert.
7927         (get_alias_symbol): New functoin.
7928         (input_node): Rewrite alias handling.
7929         (input_varpool_node): Likewise.
7930         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7931         * ipa.c (process_references): Handle weakrefs correctly.
7932         (symtab_remove_unreachable_nodes): Likewise.
7933         * trans-mem.c (get_cg_data): Update.
7934         (ipa_tm_create_version_alias): Update.
7935         (ipa_tm_execute): Update.
7936         * symtab.c (dump_symtab_base): Dump aliases.
7937         (verify_symtab_base): Verify aliases.
7938         (symtab_node_availability): New function.
7939         (symtab_alias_ultimate_target): Simplify.
7940         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7941         handle all the fixup cases.
7942         (symtab_resolve_alias): New function.
7943         * passes.c (ipa_write_summaries): Handle weakrefs.
7944         * varpool.c (varpool_analyze_node): Simplify.
7945         (assemble_aliases): Update.
7946         (varpool_create_variable_alias): Simplify.
7947         (varpool_extra_name_alias): Simplify.
7948         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7949         (lto_symtab_merge_symbols): ... this one.
7951 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7953         Revert
7954         PR rtl-optimization/57268
7955         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7956         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7958 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7960         Partially reverted:
7961         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7963         PR middle-end/57073
7964         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7965         further up.
7967 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7969         PR rtl-optimization/57268
7970         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7971         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7973 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7975         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7976         unordered comparison operators when -fno-trapping-math is in effect
7977         on the e500.
7978         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7979         and implement unordered comparison operators properly on the e500.
7981 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7983         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7984         for constant scalar integers.
7985         (simplify_relational_operation_1): Likewise.
7987 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7989         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7990         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7991         Fix comment.
7993 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7994             Igor Zamyatin  <igor.zamyatin@intel.com>
7996         Silvermont (SLM) architecture performance tuning.
7997         * config/i386/i386.h (enum ix86_tune_indices): Add
7998         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7999         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8001         * config/i386/i386.c (initial_ix86_tune_features)
8002         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8003         (ix86_lea_outperforms): Handle Silvermont tuning.
8004         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8005         call.
8006         (ix86_use_lea_for_mov): Likewise.
8007         (ix86_avoid_lea_for_addr): Likewise.
8008         (ix86_lea_for_add_ok): Likewise.
8009         (exact_dependency_1): New function.
8010         (exact_store_load_dependency): Likewise.
8011         (ix86_adjust_cost): Handle Silvermont tuning.
8012         (do_reoder_for_imul): Likewise.
8013         (swap_top_of_ready_list): New function.
8014         (ix86_sched_reorder): Changed to handle Silvermont tuning.
8016         * config/i386/i386.md (peepholes that split memory operand in fp
8017         converts): New.
8019 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8021         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8022         Remove un-necessary braces.
8024 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8026         * config/aarch64/aarch64.c (aarch64_classify_symbol):
8027         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8029 2013-05-31  Tobias Burnus  <burnus@net-b.de>
8031         PR middle-end/57073
8032         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8034 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8036         PR target/56315
8037         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8038         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8039         * config/arm/neon.md (iordi3_neon): Remove.
8040         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8041         * config/arm/predicates.md (imm_for_neon_logic_operand):
8042         Move to earlier in the file.
8043         (neon_logic_op2): Likewise.
8044         (arm_iordi_operand_neon): New predicate.
8046 2013-05-31  Richard Biener  <rguenther@suse.de>
8048         PR tree-optimization/57478
8049         PR tree-optimization/57453
8050         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8051         are life as well.
8053 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8055         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8056         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8058 2013-05-30  Tobias Burnus  <burnus@net-b.de>
8059             Thomas Koenig  <tkoenig@gcc.gnu.org>
8061         PR middle-end/57073
8062         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8063         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8065 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
8067         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8069 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
8071         * target.def (register_usage_leveling_p): New hook.
8072         * targhooks.c (default_register_usage_leveling_p): New.
8073         * targhooks.h (default_register_usage_leveling_p): New prototype.
8074         * lra-assigns.c (register_usage_leveling_p): Use the hook.
8075         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8076         * doc/tm.texi: Update.
8077         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8079 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
8081         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8082         (*insv_reg<mode>): New define_insn.
8084 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8086         PR rtl-optimization/57439
8087         * postreload.c (move2add_valid_value_p): Check that we have
8088         a zero subreg_regno_offset when accessing the register in
8089         the requested mode.
8091 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8092             Igor Zamyatin  <igor.zamyatin@intel.com>
8094         Silvermont (SLM) architecture pipeline model, tuning and
8095         insn selection.
8096         * config.gcc: Add slm config options and target.
8098         * config/i386/slm.md: New.
8100         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8102         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
8103         PROCESSOR_SLM.
8104         (ix86_target_macros_internal): Likewise.
8106         * gcc/config/i386/i386.c (slm_cost): New cost.
8107         (m_SLM): New macro flag.
8108         (initial_ix86_tune_features): Set m_SLM.
8109         (x86_accumulate_outgoing_args): Likewise.
8110         (x86_arch_always_fancy_math_387): Likewise.
8111         (processor_target_table): Add slm cost.
8112         (cpu_names): Add slm cpu name.
8113         (x86_option_override_internal): Set SLM ISA.
8114         (ix86_issue_rate): New case PROCESSOR_SLM.
8115         (ia32_multipass_dfa_lookahead): Likewise.
8116         (fold_builtin_cpu): Add slm.
8118         * config/i386/i386.h (TARGET_SLM): New target macro.
8119         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8120         (processor_type): Add PROCESSOR_SLM.
8122         * config/i386/i386.md (cpu): Add new value "slm".
8123         (slm.md): Include slm.md.
8125 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
8126             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8128         * config/arm/arm-protos.h: Add and update function protos.
8129         * config/arm/arm.c (use_simple_return_p): New added.
8130         (thumb2_expand_return): Check simple_return flag.
8131         * config/arm/arm.md: Add simple_return and conditional simple_return.
8132         * config/arm/iterators.md: Add iterator for return and simple_return.
8134 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8136         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8137         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8138         (arm_emit_vfp_multi_reg_pop): Likewise.
8139         (thumb2_emit_ldrd_pop): Likewise.
8140         (arm_expand_epilogue): Add misc REG_CFA notes.
8141         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8143 2013-05-29  Lawrence Crowl  <crowl@google.com>
8145         * config/arm/t-arm: Update for below.
8147         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8148         Change type to hash_table.  Update dependent calls and types.
8150         * config/i386/t-cygming: Update for below.
8152         * config/i386/t-interix: Update for below.
8154         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8155         Change type to hash_table.  Update dependent calls and types.
8156         (i386_find_on_wrapper_list::wrappers): Likewise.
8158         * config/ia64/t-ia64: Update for below.
8160         * config/ia64/ia64.c (bundle_state_table):
8161         Change type to hash_table.  Update dependent calls and types.
8163         * config/mips/mips.c (mips_reorg_process_insns::htab):
8164         Change type to hash_table.  Update dependent calls and types.
8166         * config/sol2.c (solaris_comdat_htab):
8167         Change type to hash_table.  Update dependent calls and types.
8169         * config/t-sol2: Update for above.
8171 2013-05-29  Teresa Johnson  <tejohnson@google.com>
8173         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8174         functions are not yet marked as defined.
8176 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
8177             Pat Haugen  <pthaugen@us.ibm.com>
8178             Peter Bergner  <bergner@vnet.ibm.com>
8180         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8181         instructions.
8182         (VEC_A): Likewise.
8183         (VEC_C): Likewise.
8184         (vrotl<mode>3): Likewise.
8185         (vashl<mode>3): Likewise.
8186         (vlshr<mode>3): Likewise.
8187         (vashr<mode>3): Likewise.
8189         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8190         support for power8 V2DI builtins.
8192         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8193         power8 V2DI builtins.
8194         (vupkhsw): Likewise.
8195         (vupklsw): Likewise.
8196         (vaddudm): Likewise.
8197         (vminsd): Likewise.
8198         (vmaxsd): Likewise.
8199         (vminud): Likewise.
8200         (vmaxud): Likewise.
8201         (vpkudum): Likewise.
8202         (vpksdss): Likewise.
8203         (vpkudus): Likewise.
8204         (vpksdus): Likewise.
8205         (vrld): Likewise.
8206         (vsld): Likewise.
8207         (vsrd): Likewise.
8208         (vsrad): Likewise.
8209         (vsubudm): Likewise.
8210         (vcmpequd): Likewise.
8211         (vcmpgtsd): Likewise.
8212         (vcmpgtud): Likewise.
8213         (vcmpequd_p): Likewise.
8214         (vcmpgtsd_p): Likewise.
8215         (vcmpgtud_p): Likewise.
8216         (vupkhsw): Likewise.
8217         (vupklsw): Likewise.
8218         (vaddudm): Likewise.
8219         (vmaxsd): Likewise.
8220         (vmaxud): Likewise.
8221         (vminsd): Likewise.
8222         (vminud): Likewise.
8223         (vpksdss): Likewise.
8224         (vpksdus): Likewise.
8225         (vpkudum): Likewise.
8226         (vpkudus): Likewise.
8227         (vrld): Likewise.
8228         (vsld): Likewise.
8229         (vsrad): Likewise.
8230         (vsrd): Likewise.
8231         (vsubudm): Likewise.
8233         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8234         support for power8 V2DI instructions.
8236         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8237         power8 V2DI instructions.  Combine pack and unpack insns to use an
8238         iterator for each mode.  Check whether a particular mode supports
8239         Altivec instructions instead of just checking TARGET_ALTIVEC.
8240         (UNSPEC_VPKUWUM): Likewise.
8241         (UNSPEC_VPKSHSS): Likewise.
8242         (UNSPEC_VPKSWSS): Likewise.
8243         (UNSPEC_VPKUHUS): Likewise.
8244         (UNSPEC_VPKSHUS): Likewise.
8245         (UNSPEC_VPKUWUS): Likewise.
8246         (UNSPEC_VPKSWUS): Likewise.
8247         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8248         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8249         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8250         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8251         (UNSPEC_VUPKHSB): Likewise.
8252         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8253         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8254         (UNSPEC_VUPKHSH): Likewise.
8255         (UNSPEC_VUPKLSB): Likewise.
8256         (UNSPEC_VUPKLSH): Likewise.
8257         (VI2): Likewise.
8258         (VI_char): Likewise.
8259         (VI_scalar): Likewise.
8260         (VI_unit): Likewise.
8261         (VP): Likewise.
8262         (VP_small): Likewise.
8263         (VP_small_lc): Likewise.
8264         (VU_char): Likewise.
8265         (add<mode>3): Likewise.
8266         (altivec_vaddcuw): Likewise.
8267         (altivec_vaddu<VI_char>s): Likewise.
8268         (altivec_vadds<VI_char>s): Likewise.
8269         (sub<mode>3): Likewise.
8270         (altivec_vsubcuw): Likewise.
8271         (altivec_vsubu<VI_char>s): Likewise.
8272         (altivec_vsubs<VI_char>s): Likewise.
8273         (altivec_vavgs<VI_char>): Likewise.
8274         (altivec_vcmpbfp): Likewise.
8275         (altivec_eq<mode>): Likewise.
8276         (altivec_gt<mode>): Likewise.
8277         (altivec_gtu<mode>): Likewise.
8278         (umax<mode>3): Likewise.
8279         (smax<mode>3): Likewise.
8280         (umin<mode>3): Likewise.
8281         (smin<mode>3): Likewise.
8282         (altivec_vpkuhum): Likewise.
8283         (altivec_vpkuwum): Likewise.
8284         (altivec_vpkshss): Likewise.
8285         (altivec_vpkswss): Likewise.
8286         (altivec_vpkuhus): Likewise.
8287         (altivec_vpkshus): Likewise.
8288         (altivec_vpkuwus): Likewise.
8289         (altivec_vpkswus): Likewise.
8290         (altivec_vpks<VI_char>ss): Likewise.
8291         (altivec_vpks<VI_char>us): Likewise.
8292         (altivec_vpku<VI_char>us): Likewise.
8293         (altivec_vpku<VI_char>um): Likewise.
8294         (altivec_vrl<VI_char>): Likewise.
8295         (altivec_vsl<VI_char>): Likewise.
8296         (altivec_vsr<VI_char>): Likewise.
8297         (altivec_vsra<VI_char>): Likewise.
8298         (altivec_vsldoi_<mode>): Likewise.
8299         (altivec_vupkhsb): Likewise.
8300         (altivec_vupkhs<VU_char>): Likewise.
8301         (altivec_vupkls<VU_char>): Likewise.
8302         (altivec_vupkhsh): Likewise.
8303         (altivec_vupklsb): Likewise.
8304         (altivec_vupklsh): Likewise.
8305         (altivec_vcmpequ<VI_char>_p): Likewise.
8306         (altivec_vcmpgts<VI_char>_p): Likewise.
8307         (altivec_vcmpgtu<VI_char>_p): Likewise.
8308         (abs<mode>2): Likewise.
8309         (vec_unpacks_hi_v16qi): Likewise.
8310         (vec_unpacks_hi_v8hi): Likewise.
8311         (vec_unpacks_lo_v16qi): Likewise.
8312         (vec_unpacks_hi_<VP_small_lc>): Likewise.
8313         (vec_unpacks_lo_v8hi): Likewise.
8314         (vec_unpacks_lo_<VP_small_lc>): Likewise.
8315         (vec_pack_trunc_v8h): Likewise.
8316         (vec_pack_trunc_v4si): Likewise.
8317         (vec_pack_trunc_<mode>): Likewise.
8319         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8320         V2DI builtins.
8321         (vec_vmaxsd): Likewise.
8322         (vec_vmaxud): Likewise.
8323         (vec_vminsd): Likewise.
8324         (vec_vminud): Likewise.
8325         (vec_vpksdss): Likewise.
8326         (vec_vpksdus): Likewise.
8327         (vec_vpkudum): Likewise.
8328         (vec_vpkudus): Likewise.
8329         (vec_vrld): Likewise.
8330         (vec_vsld): Likewise.
8331         (vec_vsrad): Likewise.
8332         (vec_vsrd): Likewise.
8333         (vec_vsubudm): Likewise.
8334         (vec_vupkhsw): Likewise.
8335         (vec_vupklsw): Likewise.
8337 2013-05-29  Jan Hubicka  <jh@suse.cz>
8339         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8340         flags; reorder rest of fields in more consistent way.
8341         (varpool_node): Remove analyzed, finalized and alias.
8342         (cgraph_ndoe): Likewise.
8343         (symtab_alias_ultimate_target): New function.
8344         (cgraph_function_node): Move offline.
8345         (cgraph_reset_node): Declare.
8346         (cgraph_comdat_can_be_unshared_p): Remove.
8347         (varpool_remove_initializer): Declare.
8348         (varpool_first_defined_variable, varpool_next_defined_variable
8349         cgraph_first_defined_function, cgraph_next_defined_function): Update.
8350         (cgraph_function_with_gimple_body_p): Update.
8351         (varpool_all_refs_explicit_p): Update.
8352         (symtab_alias_target): New function.
8353         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8354         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8355         (cgraph_function_or_thunk_node): Simplify using
8356         symtab_alias_ultimate_target.
8357         (varpool_variable_node): Likewise.
8358         * cgraph.c (cgraph_create_function_alias): Update.
8359         (cgraph_add_thunk): Update.
8360         (cgraph_remove_node): Update.
8361         (dump_cgraph_node): Do not dump removed flags.
8362         (cgraph_function_body_availability): Update.
8363         (cgraph_propagate_frequency): Update.
8364         (verify_cgraph_node): Check sanity of local flag.
8365         (cgraph_function_node): Move here from cgraph.h; revamp for
8366         cgraph_function_or_thunk_node.
8367         * lto-symtab.c (lto_varpool_replace_node): Update.
8368         (lto_symtab_resolve_can_prevail_p): Update.
8369         (lto_symtab_merge_cgraph_nodes): Update.
8370         * ipa-cp.c (determine_versionability, initialize_node_lattices,
8371         propagate_constants_accross_call, devirtualization_time_bonus,
8372         ipcp_propagate_stage): Update.
8373         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8374         * ipa-inline-transform.c (clone_inlined_nodes,
8375         preserve_function_body_p): Update.
8376         * ipa-reference.c (propagate): Update.
8377         (write_node_summary_p): Update.
8378         * toplev.c (wrapup_global_declaration_2): Update.
8379         * cgraphunit.c (cgraph_analyze_function): Rename to ...
8380         (analyze_function) ... this one.
8381         (cgraph_process_new_functions): Update.
8382         (cgraph_reset_node): Export.
8383         (cgraph_finalize_function): Update.
8384         (cgraph_add_new_function): Update.
8385         (process_function_and_variable_attributes): Update.
8386         (varpool_finalize_decl): Update.
8387         (symbol_finalized): Remove.
8388         (symbol_finalized_and_needed): Rename to ...
8389         (symbol_defined_and_needed): ... update.
8390         (cgraph_analyze_functions): Update.
8391         (handle_alias_pairs): Update.
8392         (mark_functions_to_output): Update.
8393         (assemble_thunk): Update.
8394         (output_in_order): Update.
8395         (output_weakrefs): Update.
8396         (finalize_compilation_unit): Update.
8397         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8398         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8399         input_node, input_varpool_node): Update.
8400         * dbxout.c (dbxout_expand_expr): Update.
8401         * cgraphclones.c (cgraph_clone_node): Update.
8402         (cgraph_copy_node_for_versioning): Update.
8403         (cgraph_materialize_clone): Update.
8404         (cgraph_materialize_all_clones): Update.
8405         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8406         propagate_pure_const, propagate_nothrow): Update.
8407         * lto-streamer-out.c (lto_output, write_symbol): Update.
8408         * ipa-utils.c (ipa_reverse_postorder): Update.
8409         * ipa-inline.c (can_inline_edge_p): Update.
8410         (update_caller_keys, ipa_inline): Update.
8411         * dwarf2out.c (reference_to_unused,
8412         premark_types_used_by_global_vars_helper): Update.
8413         * tree-eh.c (tree_could_trap_p): Update.
8414         * ipa-split.c (consider_split, execute_split_functions): Update.
8415         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8416         has_addr_references_p): Update; move ahead in file for better
8417         readability.
8418         (process_references): Simplify.
8419         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8420         bodies are removed.
8421         (cgraph_comdat_can_be_unshared_p): Make static.
8422         (cgraph_externally_visible_p): Update.
8423         (varpool_externally_visible_p): Update.
8424         (function_and_variable_visibility): Update.
8425         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8426         ipa_tm_mark_force_output_node): Update.
8427         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8428         estimate_edge_devirt_benefit, inline_generate_summary,
8429         inline_write_summary): Update.
8430         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8431         * ipa-prop.c (ipa_compute_jump_functions): Update.
8432         (ipa_print_node_params, ipa_prop_read_section,
8433         ipa_update_after_lto_read, read_replacements_section): Update.
8434         * varasm.c (mark_decl_referenced): Update.
8435         (assemble_alias, dump_tm_clone_pairs): Update.
8436         * tree-inline.c (copy_bb): Update.
8437         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8438         Update.
8439         * symtab.c (dump_symtab_base): Print new flags.
8440         (verify_symtab_base): Verify new flags.
8441         (symtab_alias_ultimate_target): New function.
8442         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8443         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8444         Update.
8445         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8446         Update.
8447         * i386.c (ix86_get_function_versions_dispatcher,
8448         ix86_generate_version_dispatcher_body): Update.
8449         (fold_builtin_cpu): Use varpool_add_new_variable.
8450         * varpool.c (varpool_remove_initializer): Break out from ...
8451         (varpool_remove_node): ... this one.
8452         (dump_varpool_node, varpool_node_for_asm,
8453         cgraph_variable_initializer_availability, varpool_analyze_node,
8454         varpool_assemble_decl, varpool_remove_unreferenced_decls,
8455         varpool_finalize_named_section_flags, varpool_create_variable_alias):
8456         Update.
8458 2013-05-29  Jan Hubicka  <jh@suse.cz>
8460         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8462 2013-05-29  Easwaran Raman  <eraman@google.com>
8464         PR tree-optimization/57442
8465         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8466         when control exits the main loop.
8468 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
8470         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8471         and RX600.
8472         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8473         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8474         * rx/t-rx: Add rx100 under multi library matches option for nofpu
8475         option.
8477 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8479         PR tree-optimization/57441
8480         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8481         Don't limit size of incr_vec to number of candidates.
8483 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
8485         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8486         and mips16 directories.
8487         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8488         (MULTILIB_DIRNAMES): Ditto.
8489         (MULTILIB_EXCEPTIONS): Add new exceptions.
8490         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8491         (MULTILIB_DIRNAMES): Ditto.
8492         (MULTILIB_EXCEPTIONS): Add new exceptions.
8494 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8495             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8497         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8498         SYMBOL_TINY_ABSOLUTE.
8499         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8500         SYMBOL_TINY_ABSOLUTE.
8501         (aarch64_expand_mov_immediate): Likewise.
8502         (aarch64_classify_symbol): Likewise.
8503         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8504         Permit SYMBOL_TINY_ABSOLUTE.
8505         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8507 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8508             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8510         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8511         Refactor if/switch.  Replace gcc_assert with if.
8513 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8515         * config/i386/i386.c (initial_ix86_tune_features): Enable
8516         FP Reassociation for AMD bdver1 and bdver2.
8518 2013-05-29  Martin Jambor  <mjambor@suse.cz>
8520         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8521         and IMAGPART_EXPR do not occur within other handled_components.
8523 2013-05-29  Richard Biener  <rguenther@suse.de>
8525         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8526         access on whether the use is in the BB we currently try to
8527         vectorize.
8528         (vect_bb_vectorization_profitable_p): Pass the BB we currently
8529         vectorize to vect_bb_slp_scalar_cost.
8531 2013-05-29  Richard Biener  <rguenther@suse.de>
8533         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8534         computing scalar cost offsetted by stmts that are kept live
8535         by scalar uses.
8536         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8537         for computation of scalar cost.
8539 2013-05-28  Steve Ellcey  <sellcey@mips.com>
8541         * config/mips/mips-cpus.def (mips32r2): Change processor type.
8543 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8545         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8546         array notation built-in reduction functions.
8547         * doc/passes.texi (Passes): Added documentation about changes done
8548         for Cilk Plus.
8549         * doc/invoke.texi (C Dialect Options): Added documentation about
8550         the -fcilkplus flag.
8551         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8552         (BUILTINS_DEF): Depend on cilkplus.def.
8553         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
8554         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8555         * cilkplus.def: New file.
8557 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
8559         PR rtl-optimization/57439
8560         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8562 2013-05-28  Easwaran Raman  <eraman@google.com>
8564         PR tree-optimization/57337
8565         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8566         (find_insert_point): Correctly identify the insertion point
8567         when two statements with the same UID is compared.
8569 2013-05-28  Richard Biener  <rguenther@suse.de>
8571         PR tree-optimization/56787
8572         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8573         from the list of data references.
8574         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8575         clobbers.
8576         (vect_analyze_loop_operations): Likewise.
8577         (vect_transform_loop): Remove clobbers.
8579 2013-05-28  Martin Jambor  <mjambor@suse.cz>
8581         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8582         and REALPART_EXPRs have scalar type.
8584 2013-05-28  Richard Biener  <rguenther@suse.de>
8586         PR tree-optimization/57411
8587         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8588         virtual operands.
8589         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8590         virtual operand propagation.
8592 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8594         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8595         destination register for bmasksi_vis.
8596         (vector_init_bshuffle): Likewise.
8597         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8599 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8601         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8602         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8603         mode if the instruction isn't available in the original mode.
8604         * config/sparc/sparc.opt (mfix-ut699): New option.
8605         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8606         (divdf3): Turn into expander.
8607         (divdf3_nofix): New insn.
8608         (divdf3_fix): Likewise.
8609         (divsf3): Disable if -mfix-ut699.
8610         (sqrtdf2): Turn into expander.
8611         (sqrtdf2_nofix): New insn.
8612         (sqrtdf2_fix): Likewise.
8613         (sqrtsf2): Disable if -mfix-ut699.
8615 2013-05-27  Richard Biener  <rguenther@suse.de>
8617         PR middle-end/57412
8618         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8619         block for the new loop.
8621 2013-05-27  Richard Biener  <rguenther@suse.de>
8623         PR tree-optimization/57343
8624         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8625         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8626         (number_of_iterations_cond): Do not build the folded tree.
8628 2013-05-27  Richard Biener  <rguenther@suse.de>
8630         Revert
8631         PR middle-end/57381
8632         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8633         OEP_CONSTANT_ADDRESS_OF retained.
8635         PR tree-optimization/57417
8636         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8637         for unchanged base.
8638         (set_ssa_val_to): Compare addresses using
8639         get_addr_base_and_unit_offset.
8641 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
8643         PR rtl-optimization/56833
8644         * postreload.c (move2add_record_mode): New function.
8645         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8646         (move2add_use_add2_insn): Use move2add_record_sym_value.
8647         (move2add_use_add3_insn): Likewise.
8648         (reload_cse_move2add): Use move2add_valid_value_p and
8649         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8650         affected regs by setting reg_mode to VOIDmode.
8651         (move2add_note_store): Don't pretend the inside of a SUBREG is
8652         the actual destination.  Invalidate single/leading registers by
8653         setting reg_mode to VOIDmode.
8654         Use move2add_record_sym_value, move2add_valid_value_p and
8655         move2add_record_mode.
8657 2013-05-27  Richard Biener  <rguenther@suse.de>
8659         PR tree-optimization/57396
8660         * tree-affine.c (double_int_constant_multiple_p): Properly
8661         return false for val == 0 and div != 0.
8663 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8665         * config/mips/mips.h: Use #elif in preprocessor conditions.
8667 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8669         PR target/53916
8670         * config/mips/constraints.md (kl): New constraint.
8671         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8672         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8673         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8674         from LO for MIPS16.
8675         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8677 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8679         PR target/55777
8680         * config/mips/mips.c (mips_can_inline_p): New function.
8681         (TARGET_CAN_INLINE_P): Define.
8683 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8685         * sched-int.h (ds_t, dw_t): Make unsigned int.
8686         Fix documentation that describes how all the ds_t bits are used.
8687         Reserve the last bit for delayed-branch scheduling.
8688         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8689         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8690         (gen_dep_weak_1): Remove prototype.
8691         * sched-deps.c (get_dep_weak_1): Make static.
8692         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8693         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8694         * doc/tm.texi: Regenerate.
8695         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8697 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8699         PR debug/56950
8700         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8702 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8703             Sandra Loosemore  <sandra@codesourcery.com>
8705         * config.gcc (powerpc-*): Allow native for with-cpu.
8707 2013-05-24  Jeff Law  <law@redhat.com>
8709         PR tree-optimization/57124
8710         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8711         conversion feeding a condition if the range has an overflow
8712         if -fstrict-overflow.  Add warnings for when we do make the
8713         transformation.
8715 2013-05-24  Dehao Chen  <dehao@google.com>
8717         * tree-cfg.c (locus_discrim_map): Fix the typo.
8718         (locus_discrim_hasher): Likewise.
8719         (locus_discrim_hasher::hash): Likewise.
8720         (locus_discrim_hasher::equal): Likewise.
8722 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8724         PR tree-optimization/57294
8725         * cgraph.h (ipa_record_stmt_references): Declare.
8726         * cgraphbuild.c (ipa_record_stmt_references): New function.
8727         (build_cgraph_edges): Use ipa_record_stmt_references.
8728         (rebuild_cgraph_edges): Likewise.
8729         (cgraph_rebuild_references): Likewise.
8730         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8731         associated with the old statement and build references from the
8732         newly built statements.
8733         * ipa-ref.c (ipa_remove_stmt_references): New function.
8734         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8736 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8738         * lra-constraints.c (emit_spill_move): Use smaller mode for
8739         mem-mem moves.
8740         (check_and_process_move): Consider mem-reg moves for secondary
8741         too.
8742         (curr_insn_transform): Don't lose insns emitted before for
8743         secondary memory moves.
8744         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8745         reg set up in the current insn.
8747 2013-05-24  Dehao Chen  <dehao@google.com>
8749         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8750         hash function.
8751         (locus_descrim_hasher::equal): Likewise.
8752         (build_gimple_cfg): New discriminator assignment algorithm.
8753         (make_edges): Likewise.
8754         (next_discriminator_for_locus): Likewise.
8755         (same_line_p): Likewise.
8756         (assign_discriminators): Likewise.
8757         (make_cond_expr_edges): Likewise.
8758         (make_gimple_switch_edges): Likewise.
8759         (make_goto_expr_edges): Likewise.
8760         (make_gimple_asm_edges): Likewise.
8762 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8764         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8765         X format specifier to only display bottom 16 bits.
8766         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8767         immediate to match for operand 2, since it will be masked.
8769 2013-05-24  Richard Biener  <rguenther@suse.de>
8771         PR tree-optimization/57287
8772         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8773         all SSA names that occur in abnormal PHIs.
8775 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8777         PR tree-ssa/57385
8778         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8779         that index is not negative.
8781 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8783         PR rtl-optimization/55177
8784         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8785         (simplify_byte_swapping_operation): New.
8786         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8787         (simplify_relational_operation_1): Deal with BSWAP.
8789 2013-05-23  Richard Henderson  <rth@redhat.com>
8791         PR target/56742
8792         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8793         (ix86_reorg): Call it.
8795 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8797         PR target/57379
8798         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8799         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8800         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8802 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8804         PR debug/57351
8805         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8807 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8808             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8810         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8811         * config/aarch64/constraints.md (Usa): Remove.
8812         * doc/md.texi (AArch64 Usa): Remove.
8814 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8815             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8817         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8818         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8819         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8820         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8822 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
8824         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8825         instruction (AdvSIMD).
8826         * config/aarch64/aarch64-builtins.c
8827         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8828         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8830 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8832         PR middle-end/57347
8833         * tree.h (contains_bitfld_component_ref_p): Declare.
8834         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8835         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8836         caller.
8837         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8838         not access a bit-field.  Assert all final offsets are byte-aligned.
8840 2013-05-23  Richard Biener  <rguenther@suse.de>
8842         PR tree-optimization/57380
8843         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8844         least one invariant or re-used load.
8845         * passes.c (init_optimization_passes): Move pass_phiprop before
8846         pass_forwprop.
8848 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8850         * config/aarch64/aarch64-simd.md
8851         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8853 2013-05-23  Richard Biener  <rguenther@suse.de>
8855         PR middle-end/57381
8856         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8857         OEP_CONSTANT_ADDRESS_OF retained.
8859 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8861         PR middle-end/57344
8862         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8863         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8865 2013-05-23  Richard Biener  <rguenther@suse.de>
8867         PR rtl-optimization/57341
8868         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8869         instead of true_dependence.
8871 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8873         * bb-reorder.c (branch_threshold): Make const.
8874         (exec_threshold): Ditto.
8876 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8877             Pat Haugen  <pthaugen@us.ibm.com>
8878             Peter Bergner  <bergner@vnet.ibm.com>
8880         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8881         documentation for the power8 crypto builtins.
8883         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8885         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8886         macros for defining power8 builtin functions.
8887         (BU_P8V_AV_2): Likewise.
8888         (BU_P8V_AV_P): Likewise.
8889         (BU_P8V_VSX_1): Likewise.
8890         (BU_P8V_OVERLOAD_1): Likewise.
8891         (BU_P8V_OVERLOAD_2): Likewise.
8892         (BU_CRYPTO_1): Likewise.
8893         (BU_CRYPTO_2): Likewise.
8894         (BU_CRYPTO_3): Likewise.
8895         (BU_CRYPTO_OVERLOAD_1): Likewise.
8896         (BU_CRYPTO_OVERLOAD_2): Likewise.
8897         (XSCVSPDP): Fix typo, point to the correct instruction.
8898         (VCIPHER): Add power8 crypto builtins.
8899         (VCIPHERLAST): Likewise.
8900         (VNCIPHER): Likewise.
8901         (VNCIPHERLAST): Likewise.
8902         (VPMSUMB): Likewise.
8903         (VPMSUMH): Likewise.
8904         (VPMSUMW): Likewise.
8905         (VPERMXOR_V2DI): Likewise.
8906         (VPERMXOR_V4SI: Likewise.
8907         (VPERMXOR_V8HI: Likewise.
8908         (VPERMXOR_V16QI: Likewise.
8909         (VSHASIGMAW): Likewise.
8910         (VSHASIGMAD): Likewise.
8911         (VPMSUM): Likewise.
8912         (VPERMXOR): Likewise.
8913         (VSHASIGMA): Likewise.
8915         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8916         __CRYPTO__ if the crypto instructions are available.
8917         (altivec_overloaded_builtins): Add support for overloaded power8
8918         builtins.
8920         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8921         support for power8 crypto builtins.
8922         (builtin_function_type): Likewise.
8923         (altivec_init_builtins): Add support for builtins that take vector
8924         long long (V2DI) arguments.
8926         * config/rs6000/crypto.md: New file, define power8 crypto
8927         instructions.
8929 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8930             Pat Haugen  <pthaugen@us.ibm.com>
8931             Peter Bergner  <bergner@vnet.ibm.com>
8933         * doc/invoke.texi (Option Summary): Add power8 options.
8934         (RS/6000 and PowerPC Options): Likewise.
8936         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8937         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8938         wm, wn, wr documentation.
8940         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8941         registers if direct move instructions are enabled.
8942         (wn): New constraint for no registers.
8943         (wq): New constraint for quad word even GPR registers.
8944         (wr): New constraint if 64-bit instructions are enabled.
8945         (wv): New constraint if power8 vector instructions are enabled.
8946         (wQ): New constraint for quad word memory locations.
8948         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8949         constraint for 0..15 for crypto instructions.
8950         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8951         as GPR and floating point registers.
8952         (int_reg_operand): New predicate to match only GPR registers.
8953         (base_reg_operand): New predicate to match base registers.
8954         (quad_int_reg_operand): New predicate to match even GPR registers
8955         for quad memory operations.
8956         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8957         operations in both GPR and VSX registers.
8958         (quad_memory_operand): New predicate for quad memory operations.
8959         (reg_or_indexed_operand): New predicate for direct move support.
8961         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8962         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8963         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8964         (POWERPC_MASKS): Add power8 options.
8965         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8966         various options.
8968         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8969         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8971         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8972         (-mpower8-fusion): New power8 options.
8973         (-mpower8-fusion-sign): Likewise.
8974         (-mpower8-vector): Likewise.
8975         (-mcrypto): Likewise.
8976         (-mdirect-move): Likewise.
8977         (-mquad-memory): Likewise.
8979         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8980         power8.
8981         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8982         registers.
8983         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8984         (rs6000_debug_vector_unit): Add p8_vector.
8985         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8986         definitions.  Also print fusion state.
8987         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8988         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8989         (rs6000_option_override_internal): Add support for power8.
8990         (rs6000_common_init_builtins): Add debugging for skipped builtins
8991         if -mdebug=builtin.
8992         (rs6000_adjust_cost): Add power8 support.
8993         (rs6000_issue_rate): Likewise.
8994         (insn_must_be_first_in_group): Likewise.
8995         (insn_must_be_last_in_group): Likewise.
8996         (force_new_group): Likewise.
8997         (rs6000_register_move_cost): Likewise.
8998         (rs6000_opt_masks): Likewise.
9000         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9001         power8 capable assembler, default to power7 options.
9002         (TARGET_DIRECT_MOVE): Likewise.
9003         (TARGET_CRYPTO): Likewise.
9004         (TARGET_P8_VECTOR): Likewise.
9005         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9006         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9007         (VECTOR_MEM_P8_VECTOR_P): Likewise.
9008         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9009         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9010         (TARGET_XSCVDPSPN): Likewise.
9011         (TARGET_XSCVSPDPN): Likewsie.
9012         (TARGET_SYNC_HI_QI): Likewise.
9013         (TARGET_SYNC_TI): Likewise.
9014         (MASK_CRYPTO): Likewise.
9015         (MASK_DIRECT_MOVE): Likewise.
9016         (MASK_P8_FUSION): Likewise.
9017         (MASK_P8_VECTOR): Likewise.
9018         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9019         temporary used by some of the direct move instructions to get two FP
9020         temporary registers does not force creation of a stack frame.
9021         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9022         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9023         that any VSX registers are tieable, even if they are also an
9024         Altivec vector mode.
9025         (r6000_reg_class_enum): Add wm, wr, wv constraints.
9026         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9027         (RS6000_BTM_CRYPTO): Likewise.
9028         (RS6000_BTM_COMMON): Likewise.
9030         * config/rs6000/rs6000.md (cpu attribute): Add power8.
9031         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9032         (enum rs6000_vector): Add power8 vector support.
9034 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9036         PR target/19599
9037         PR target/57340
9038         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9039         (any_sibcall_could_use_r3): this and handle indirect calls.
9040         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9042 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9044         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9046 2013-05-22  Richard Biener  <rguenther@suse.de>
9048         PR middle-end/57349
9049         * profile.c (branch_prob): Do not split blocks that are
9050         abnormally receiving from ECF_RETURNS_TWICE functions.
9052 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9054         * recog.c (offsettable_address_addr_space_p): Fix calculation of
9055         address mode.  Move pointer mode initialization to the same place.
9057 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9059         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9060         while it has any effect.
9062 2013-05-21  Easwaran Raman  <eraman@google.com>
9064         PR tree-optimization/57322
9065         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9066         UID of the statement added to the BB to be 1.
9068 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
9070         PR tree-optimization/57331
9071         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9072         of conversion from pointer type to integral type with integer.
9074 2013-05-21  Martin Jambor  <mjambor@suse.cz>
9076         PR lto/57289
9077         * ipa-prop.c (ipa_read_node_info): Process param_used and
9078         controlled_uses in the same order as when writing.
9080 2013-05-21  Magnus Granberg  <baldrick@free.fr>
9082         PR plugins/56754
9083         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9085 2013-05-21  Richard Biener  <rguenther@suse.de>
9087         PR tree-optimization/57318
9088         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9089         estimate stmts with side-effects as likely eliminated.
9091 2013-05-21  Richard Biener  <rguenther@suse.de>
9093         PR tree-optimization/57330
9094         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9095         preserve the call stmts fntype.
9097 2013-05-21  Richard Biener  <rguenther@suse.de>
9099         PR tree-optimization/57303
9100         * tree-ssa-sink.c (statement_sink_location): Improve killing
9101         stmt detection and properly handle self-assignments.
9103 2013-05-21  Christian Bruel  <christian.bruel@st.com>
9105         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9106         spanning registers. LEAF_REG_REMAP is supported only for contiguous
9107         registers. Set register size out of the PARALLEL loop.
9109 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
9111         PR target/56547
9112         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9113         (*fmasf4, *fmasf4_media): New insns.
9115 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9117         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9118         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9119         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9120         (mips_idiv_insns): Update the comments to say that the returned
9121         instruction counts are in units of BASE_INSN_LENGTH.
9122         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9123         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
9124         using 2 rather than 4 as the length of indirect MIPS16 and
9125         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
9126         length of a NOP.  Don't divide MIPS16 lengths by 2.
9127         (mips16_split_long_branches): Assume a branch is long if the
9128         length is greater than 4 rather than 8.
9129         * config/mips/mips.md (length): Give MIPS16 lengths directly,
9130         rather than multiplying them by 2.  Multiply instruction counts
9131         by BASE_INSN_LENGTH rather than 4.
9132         (*jump_mips16, tls_get_tp_mips16_<mode>)
9133         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9135 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9137         * config/mips/mips.md (extended_mips16): Remove branch case.
9138         (length): Remove duplicated extended_mips16 test.
9140 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9142         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9144 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
9146         * recog.h (Recog_data): Rename to...
9147         (recog_data_d): ...this.
9148         (recog_data): Update accordingly.
9149         * recog.c (recog_data): Likewise.
9150         * reload.c (save_recog_data): Likewise.
9151         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9152         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9154 2013-05-17  Julian Brown  <julian@codesourcery.com>
9156         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9157         found in a REG_EQUAL note, invalidate it.
9159 2013-05-17  Easwaran Raman  <eraman@google.com>
9161         * tree-ssa-reassoc.c (find_insert_point): New function.
9162         (insert_stmt_after): Likewise.
9163         (get_def_stmt): Likewise.
9164         (ensure_ops_are_available): Likewise.
9165         (not_dominated_by): Likewise.
9166         (rewrite_expr_tree): Do not move statements beyond what is
9167         necessary. Remove call to swap_ops_for_binary_stmt...
9168         (reassociate_bb): ... and move it here.
9169         (build_and_add_sum): Assign UIDs for new statements.
9170         (linearize_expr): Likewise.
9171         (do_reassoc): Renumber gimple statement UIDs.
9173 2013-05-17  Jan Hubicka  <jh@suse.cz>
9175         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9176         weakrefs.
9177         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9178         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9179         weakrefs.
9180         (output_weakrefs): Update.
9182 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
9183             Martin Jambor  <mjambor@suse.cz>
9185         PR middle-end/57276
9186         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9187         value that corresponds to the given aggval is found in values vector.
9189 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
9191         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9192         sse, sse2, sse3, ssse3 and sse4a flags to options.
9194 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
9196         * gengtype-state.c: (s_expr_writer): New class, to handle
9197         prettifying of output layout of s-expressions.
9198         (state_writer): New class, to write out gtype.state.
9199         (state_written_type_count): Move this variable into member data of
9200         state_writer.
9201         (s_expr_writer::s_expr_writer): New code: constructor for new class
9202         (state_writer::state_writer(): ditto
9203         (s_expr_writer::write_new_line): New function
9204         (s_expr_writer::write_any_indent): ditto
9205         (s_expr_writer::begin_s_expr): ditto
9206         (s_expr_writer::end_s_expr): ditto
9207         (write_state_fileloc): convert to method of state_writer...
9208         (state_writer:: write_state_fileloc): ...and use methods of
9209         s_expr_writer to write indentation into the gtype.state output file
9210         to visually represent the hierarchical structure of the list
9211         structures
9212         (write_state_fields): ditto, renaming to...
9213         (state_writer::write_state_fields)
9214         (write_state_a_string): ditto, renaming to...
9215         (state_writer::write_state_a_string)
9216         (write_state_string_option): ditto, renaming to...
9217         (state_writer::write_state_string_option)
9218         (write_state_type_option): ditto, renaming to...
9219         (state_writer::write_state_type_option)
9220         (write_state_nested_option): ditto, renaming to...
9221         (state_writer::write_state_nested_option)
9222         (write_state_option): ditto, renaming to...
9223         (state_writer::write_state_option)
9224         (write_state_options): ditto, renaming to...
9225         (state_writer::write_state_options)
9226         (write_state_lang_bitmap): ditto, renaming to...
9227         (state_writer::write_state_lang_bitmap)
9228         (write_state_version): ditto, renaming to...
9229         (state_writer::write_state_version)
9230         (write_state_scalar_type): ditto, renaming to...
9231         (state_writer::write_state_scalar_type)
9232         (write_state_string_type): ditto, renaming to...
9233         (state_writer::write_state_string_type)
9234         (write_state_undefined_type): ditto, renaming to...
9235         (state_writer::write_state_undefined_type)
9236         (write_state_struct_union_type): ditto, renaming to...
9237         (state_writer::write_state_struct_union_type)
9238         (write_state_struct_type): ditto, renaming to...
9239         (state_writer::write_state_struct_type)
9240         (write_state_user_struct_type): ditto, renaming to...
9241         (state_writer::write_state_user_struct_type)
9242         (write_state_lang_struct_type): ditto, renaming to...
9243         (state_writer::write_state_lang_struct_type)
9244         (write_state_param_struct_type): ditto, renaming to...
9245         (state_writer::write_state_param_struct_type)
9246         (write_state_pointer_type): ditto, renaming to...
9247         (state_writer::write_state_pointer_type)
9248         (write_state_array_type): ditto, renaming to...
9249         (state_writer::write_state_array_type)
9250         (write_state_gc_used): ditto, renaming to...
9251         (state_writer::write_state_gc_used)
9252         (write_state_common_type_content): ditto, renaming to...
9253         (state_writer::write_state_common_type_content)
9254         (write_state_type): ditto, renaming to...
9255         (state_writer::write_state_type)
9256         (write_state_pair_list): ditto, renaming to...
9257         (state_writer::write_state_pair_list)
9258         (write_state_pair): ditto, renaming to...
9259         (state_writer::write_state_pair)
9260         (write_state_typedefs): ditto, renaming to...
9261         (state_writer::write_state_typedefs)
9262         (write_state_structures): ditto, renaming to...
9263         (state_writer::write_state_structures)
9264         (write_state_param_structs): ditto, renaming to...
9265         (state_writer::write_state_param_structs)
9266         (write_state_variables): ditto, renaming to...
9267         (state_writer::write_state_variables)
9268         (write_state_srcdir): ditto, renaming to...
9269         (state_writer::write_state_srcdir)
9270         (write_state_files_list): ditto, renaming to...
9271         (state_writer::write_state_files_list)
9272         (write_state_languages): ditto, renaming to...
9273         (state_writer::write_state_languages)
9274         (write_state): create a state_writer instance and use it when
9275         writing out the state file
9277 2013-05-17  Mike Stump  <mikestump@comcast.net>
9279         PR rtl-optimization/57304
9280         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9281         accessing DF_REF_REAL_LOC.
9283 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9285         PR rtl-optimization/57281
9286         PR rtl-optimization/57300
9287         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9288         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9289         what the other splitter did if the registers are dead.
9291 2013-05-17  Richard Biener  <rguenther@suse.de>
9293         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9294         MEM_REF offsets.
9296 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9298         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9299         linking.
9301 2013-05-17  Marek Polacek  <polacek@redhat.com>
9303         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9304         length when doing non-zero store of storing '\0' to '\0'.
9306 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9308         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9309         vect_external_def oprnd1 with loop_vinfo, try to emit
9310         optional cast, negation and and stmts on the loop preheader
9311         edge instead of into the pattern def seq.
9313         PR tree-optimization/57051
9314         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9315         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9317 2013-05-16  Nick Clifton  <nickc@redhat.com>
9319         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9320         (rl78_is_naked_func): New function.
9321         (rl78_expand_prologue): Skip prologue generation for naked functions.
9322         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9323         * doc/extend.texi (naked): Add RL78 to the list of processors
9324         that supports this attribute.
9326 2013-05-16  Jeff Law  <law@redhat.com>
9328         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9330 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9332         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9333         cache parameters using detect_caches_amd also for CYRIX,
9334         NSC and TM2 signatures.
9336 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9337             Dzianis Kahanovich  <mahatma@eu.by>
9339         PR target/45359
9340         PR target/46396
9341         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9342         VIA/Centaur processors and determine their cache parameters
9343         using detect_caches_amd.
9345 2013-05-16  Teresa Johnson  <tejohnson@google.com>
9347         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9348         (rtl_verify_edges): New function.
9349         (rtl_verify_bb_insns): Ditto.
9350         (rtl_verify_bb_pointers): Ditto.
9351         (rtl_verify_bb_insn_chain): Ditto.
9352         (rtl_verify_fallthru): Ditto.
9353         (rtl_verify_bb_layout): Ditto.
9354         (rtl_verify_flow_info_1): Outline checks into new functions.
9355         (rtl_verify_flow_info): Ditto.
9357 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
9359         * cfghooks.c (copy_bbs): Add update_dominance argument.
9360         * cfghooks.h (copy_bbs): Update prototype.
9361         * tree-cfg.c (gimple_duplicate_sese_region):
9362         Add update_dominance argument.
9363         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9364         * tree-ssa-loop-ch.c (copy_loop_headers): Update
9365         gimple_duplicate_sese_region call.
9366         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9367         Update copy_bbs call.
9368         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9369         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9371 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9373         * tree-vectorizer.h (NUM_PATTERNS): Increment.
9374         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9375         vect_recog_rotate_pattern.
9376         (vect_recog_rotate_pattern): New function.
9378 2013-05-16  Jason Merrill  <jason@redhat.com>
9380         * Makefile.in (LLINKER): New variable.
9381         (mostlyclean): Remove link mutex.
9382         * configure.ac: Handle --enable-link-mutex.
9383         * lock-and-run.sh: New script.
9385 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9387         PR target/19599
9388         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9389         for NULL decl.
9391 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9393         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9395 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
9397         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9398         * config/arm/arm.c (next_consecutive_mem): New function.
9399         (gen_movmem_ldrd_strd): Likewise.
9400         * config/arm/arm.md (movmemqi): Update condition and code.
9401         (unaligned_loaddi, unaligned_storedi): New patterns.
9403 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9405         * config.gcc: Obsolete *-*-solaris2.9*.
9406         * doc/install.texi (Specific, *-*-solaris2*): Document it.
9408 2013-05-16  Richard Biener  <rguenther@suse.de>
9410         * passes.c (init_optimization_passes): Move pass_parallelize_loops
9411         earlier, after GRAPHITE transforms and IV canonicalization.
9413 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9415         * omp-low.c (extract_omp_for_data): For collapsed loops,
9416         if at least one of the loops is known at compile time to
9417         iterate zero times, set count to 0.
9418         (expand_omp_regimplify_p): New function.
9419         (expand_omp_for_generic): For collapsed loops, if at least
9420         one of the loops isn't known to iterate at least once,
9421         add runtime check with setting count to 0.
9422         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9423         For unsigned types if it isn't known at compile time that
9424         the loop will iterate at least once, add runtime check to bypass
9425         the whole loop if initial condition isn't true.
9427 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
9429         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9431 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
9433         PR middle-end/57286
9434         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9435         transformations to avoid an infinite loop.
9437 2013-05-16  Marek Polacek  <polacek@redhat.com>
9439         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9441 2013-05-15  Leif Ekblad  <leif@rdos.net>
9443         * config/i386/i386.c (ix86_decompose_address): Use
9444         DEFAULT_TLS_SEG_REG to access TLS segment register.
9445         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9446         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9447         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9449 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
9451         PR target/57260
9452         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9453         sibling calls to functions that would normally be lazily bound,
9454         unless $gp is call-clobbered.
9456 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
9458         * config/i386/i386.c (ix86_option_override_internal): Update
9459         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
9460         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9461         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9462         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9463         of TARGET_3DNOW.
9464         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9466 2013-05-15  Andreas Schwab  <schwab@suse.de>
9468         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9469         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
9470         third operand.
9472 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9474         * loop-unroll.c (report_unroll_peel): Check decision before
9475         emitting unroll/peel message.
9477 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9479         * function.h (has_bb_partition): New rtl_data flag.
9480         (bb_reorder_complete): Ditto.
9481         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9482         instead of flag_reorder_blocks_and_partition.
9483         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9484         with some enhancements.
9485         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9486         * bb-reorder.c (connect_traces): Check for has_bb_partition
9487         instead of flag_reorder_blocks_and_partition.
9488         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9489         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9490         verify_hot_cold_block_grouping.
9491         (partition_hot_cold_basic_blocks): Set has_bb_partition.
9493 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9495         PR target/19599
9496         * config/arm/predicates.md (call_insn_operand): New predicate.
9497         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
9498         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9499         if insn is not a tail call.
9500         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9501         registers.
9502         * config/arm/arm.h (enum reg_class): New caller save register class.
9503         (REG_CLASS_NAMES): Likewise.
9504         (REG_CLASS_CONTENTS): Likewise.
9505         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9506         without decls.
9508 2013-05-15  Richard Biener  <rguenther@suse.de>
9510         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9511         of MSG_OPTIMIZED_LOCATIONS.
9512         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9513         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9514         message.
9515         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9516         of MSG_OPTIMIZED_LOCATIONS.
9517         (execute_vect_slp): Likewise.
9518         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9519         (vect_create_cond_for_alias_checks): Likewise.
9520         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9521         (vect_recog_widen_mult_pattern): Likewise.
9522         (vect_recog_widen_sum_pattern): Likewise.
9523         (vect_recog_over_widening_pattern): Likewise.
9524         (vect_recog_widen_shift_pattern): Likewise.
9525         (vect_recog_vector_vector_shift_pattern): Likewise.
9526         (vect_recog_divmod_pattern): Likewise.
9527         (vect_recog_mixed_size_cond_pattern): Likewise.
9528         (vect_recog_bool_pattern): Likewise.
9529         (vect_pattern_recog_1): Likewise.
9531 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9533         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9534         non-functions to builtin_unreachable.
9535         * ipa-inline-transform.c (inline_call): Do not assert estimates were
9536         correct when new direct edges were discovered.
9538 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9540         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9541         header, print symbol order instead of node uid, print more information
9542         about indirect edge targets.
9543         (ipa_make_edge_direct_to_target): Print symbol order instead of node
9544         uids.
9545         (ipa_make_edge_direct_to_target): Likewise.
9546         (remove_described_reference): Likewise.
9547         (propagate_controlled_uses): Likewise.
9548         (ipa_print_node_params): Also print symbol order.
9549         (ipcp_transform_function): Print symbol order instead of node uids.
9550         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9551         (cgraph_get_create_real_symbol_node): Likewise.
9552         * ipa-cp.c (print_lattice): Likewise.
9553         (print_all_lattices): Likewise.
9554         (determine_versionability): Likewise.
9555         (initialize_node_lattices): Likewise.
9556         (estimate_local_effects): Likewise.
9557         (update_profiling_info): Likewise.
9558         (create_specialized_node): Likewise.
9559         (perhaps_add_new_callers): Likewise.
9560         (decide_about_value): Likewise.
9561         (decide_whether_version_node): Likewise.
9562         (identify_dead_nodes): Likewise.
9563         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9564         (dump_inline_summary): Likewise.
9565         (estimate_node_size_and_time): Likewise.
9566         (inline_analyze_function): Likewise.
9567         * ipa-inline.c (report_inline_failed_reason): Likewise.
9568         (want_early_inline_function_p): Likewise.
9569         (edge_badness): Likewise.
9570         (update_edge_key): Likewise.
9571         (inline_small_functions): Likewise.  Add dumping of order to two other
9572         dumps.
9573         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9574         instead of node uids.
9575         (propagate_pure_const): Likewise.
9576         (propagate_pure_const): Likewise.
9577         * ipa-utils.c (dump_cgraph_node_set): Likewise.
9578         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9579         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9580         of node uids.
9581         * tree-pretty-print.c (dump_function_header): Likewise.
9582         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9583         Print symbol order instead of node uids.
9585 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9587         * config/s390/s390.c (s390_register_move_cost): Don't impose the
9588         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9590 2013-05-15  Richard Biener  <rguenther@suse.de>
9592         PR tree-optimization/57275
9593         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9594         return value for fail to do runtime alias checks for gather loads.
9596 2013-05-15  Jan Hubicka  <jh@suse.cz>
9598         PR lto/57038
9599         PR lto/47375
9600         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9601         weakrefs are not external.
9602         (lto_symtab_merge_decls): Fix thinko when dealing with
9603         non-lto_symtab decls.
9604         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9605         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9606         * varpool.c (dump_varpool_node): Dump more flags.
9608 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9610         * config/i386/i386.c (processor_alias_table): Add instruction
9611         FSGSBASE for AMD bdver3 architecture.
9613 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9615         * tree.c (warn_deprecated_use): Print file:line using locus color.
9616         * diagnostic.c (diagnostic_report_current_module): Print file:line
9617         and file:line:column using locus color.
9619 2013-05-14  Mike Stump  <mikestump@comcast.net>
9621         * gdbinit.in: Add __null.
9623 2013-05-14  Mike Stump  <mikestump@comcast.net>
9625         * recog.h: Rename struct recog_data to Recog_data.
9626         * recog.c: Likewise.
9627         * reload.c (can_reload_into): Likewise.
9628         * config/picochip/picochip.c: Likewise.
9630 2013-05-14  Mike Stump  <mikestump@comcast.net>
9632         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9634 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9636         * resource.h (struct resources): Remove unch_memory member.
9637         (CLEAR_RESOURCE): Don't clear unch_memory.
9638         * resource.c (mark_referenced_resources): Don't set it.
9639         (mark_set_resources): Likewise.
9640         (mark_target_live_regs): Don't clear it.
9641         (init_resource_info): Likewise.
9642         * reorg.c (resource_conflicts_p): Don't compare it.
9643         (redundant_insn): Don't set it.
9645         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9646         Remove prototypes.
9647         * emit-rtl.c (next_label): Remove unused function.
9648         (skip_consecutive_labels, link_cc0_insns): Move to ...
9649         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9650         only place where these functions are used, and make them static.
9652 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9654         * fold-const.c (fold_negate_expr): Handle vectors.
9655         (fold_truth_not_expr): Make it static.
9656         (fold_invert_truthvalue): New static function.
9657         (invert_truthvalue_loc): Handle vectors. Do not call
9658         fold_truth_not_expr directly.
9659         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9660         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9661         (fold_comparison): Handle vector constants.
9662         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9663         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9664         * tree.h (fold_truth_not_expr): Remove declaration.
9666 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9668         * config/aarch64/aarch64-simd.md
9669         (aarch64_vcond_internal<mode>): Rename to...
9670         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9671         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9672         float modes. Clarify all iterator modes.
9673         (vcond<mode><mode>): Use new name for vcond expanders.
9674         (vcond<v_cmp_result><mode>): Likewise.
9675         (vcondu<mode><mode>: Likewise.
9676         * config/aarch64/iterators.md (VDQF_COND): New.
9678 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9680         PR bootstrap/57266
9681         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9682         variable for the shift amount. Check that we shift by non-negative
9683         amounts.
9685 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9687         PR target/42017
9688         * config/arm/arm.h (EPILOGUE_USES): Only return true
9689         for LR_REGNUM after epilogue_completed.
9691 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
9693         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9694         is error_mark_node.
9696 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9698         PR target/57261
9699         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9700         and Solaris 11+/x86 with gld.
9701         * configure: Regenerate.
9703 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9705         * expmed.c (expand_shift_1): Canonicalize rotates by
9706         constant bitsize / 2 to bitsize - 1.
9707         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9708         case ROTATERT>: Likewise.
9710         Revert:
9711         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9713         * config/i386/i386.md (rotateinv): New code attr.
9714         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9715         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9716         roll $31, %eax, etc.
9718 2013-05-14  Richard Biener  <rguenther@suse.de>
9720         PR middle-end/57235
9721         * tree-eh.c (sink_clobbers): Give up for successors with
9722         multiple predecessors and no virtual uses.
9724 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9726         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9727         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9729 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9731         PR middle-end/57251
9732         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9733         the case when both op0 and op1 have VOIDmode.
9735 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9737         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9738         in multiply-accumulate mode.
9740 2013-05-13  Guozhi Wei  <carrot@google.com>
9742         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9744 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9746         PR target/56975
9747         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9748         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9749         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9750         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9751         (ix86_expand_prologue): Likewise.
9752         (ix86_expand_split_stack_prologue): Likewise.
9753         (legitimate_pic_address_disp_p): Likewise.
9754         (legitimize_pic_address): Likewise.
9755         (legitimize_tls_address): Likewise.
9756         (legitimize_pe_coff_symbol): Likewise.
9757         (output_pic_addr_const): Likewise.
9758         (construct_plt_address): Likewise.
9759         (ix86_expand_call): Likewise.
9760         (x86_output_mi_thunk): Likewise.
9761         (x86_function_profiler): Likewise.
9763 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9765         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9766         similar switch cases.
9767         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9768         (aarch64_simd_mov_to_<mode>low): Delete.
9769         (aarch64_simd_mov_to_<mode>high): Delete.
9770         (move_lo_quad_<mode>): Add w<-r alternative.
9771         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9772         (aarch64_simd_mov_from_*): Update type attribute.
9773         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9774         statement.
9776 2013-05-13  Jan Hubicka  <jh@suse.cz>
9778         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9779         * config/i386/i386.c (ix86_compute_frame_layout,
9780         ix86_expand_epilogue, emit_i387_cw_initialization,
9781         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9782         ix86_local_alignment): Fix use of size/speed predicates.
9784 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9786         PR tree-optimization/45216
9787         PR tree-optimization/57157
9788         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9789         the (-Y) & (B - 1) variant if OP is |.
9790         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9791         return shifted.  Use (-op1) & (prec - 1) as other_amount
9792         instead of prec - op1.
9794 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9796         PR middle-end/42371
9797         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9798         (ipa_constant_data): New type.
9799         (ipa_jump_func): Use ipa_constant_data to hold information about
9800         constant jump functions.
9801         (ipa_get_jf_constant): Adjust to jump function type changes.
9802         (ipa_get_jf_constant_rdesc): New function.
9803         (ipa_param_descriptor): New field controlled_uses.
9804         (ipa_get_controlled_uses): New function.
9805         (ipa_set_controlled_uses): Likewise.
9806         * ipa-ref.h (ipa_find_reference): Declare.
9807         * ipa-prop.c (ipa_cst_ref_desc): New type.
9808         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9809         changes.
9810         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9811         New parameter cs.  Adjust all callers.
9812         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9813         (remove_described_reference): New function.
9814         (jfunc_rdesc_usable): Likewise.
9815         (try_make_edge_direct_simple_call): Decrement controlled use count,
9816         attempt to remove reference if it hits zero.
9817         (combine_controlled_uses_counters): New function.
9818         (propagate_controlled_uses): Likewise.
9819         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9820         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9821         (ipa_print_node_params): Print described use counter.
9822         (ipa_write_jump_function): Adjust to jump function type changes.
9823         (ipa_read_jump_function): New parameter CS, pass it to
9824         ipa_set_jf_constant.  Adjust caller.
9825         (ipa_write_node_info): Stream controlled use count
9826         (ipa_read_node_info): Likewise.
9827         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9828         asserting.
9829         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9830         count.  Remove cloning-added reference if it reaches zero.
9831         * ipa-ref.c (ipa_find_reference): New function.
9833 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9835         * config/i386/i386.c (processor_target_table): Modified default
9836         alignment values for AMD BD and BT architectures.
9838 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9840         * tree-vect-generic.c (uniform_vector_p): Move ...
9841         * tree.c (uniform_vector_p): ... here.
9842         * tree.h (uniform_vector_p): Declare it.
9843         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9844         into a scalar.
9846 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9848         PR tree-optimization/57230
9849         * tree-ssa-strlen.c (handle_char_store): Record length for
9850         array store from STRING_CST.
9852         PR tree-optimization/57230
9853         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9854         check.
9856 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9858         * config/epiphany/epiphany.c (epiphany_init): Check size of
9859         NUM_MODES_FOR_MODE_SWITCHING.
9860         (epiphany_expand_prologue):
9861         Remove CONFIG_REGNUM initial value handling code.
9862         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9863         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9864         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9865         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9866         Don't return 1 for FP_MODE_NONE.
9867         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9868         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9869         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9870         * config/epiphany/epiphany.md (save_config): New pattern.
9872 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9874         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9876 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9878         * config/i386/i386.md (memory): Handle sseishft1.
9879         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9880         (*vec_extractv2di_1): Ditto.
9882 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9884         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9885         saved registers.
9887 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9889         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9890         Add mthumb/march=armv7-a multilib.
9891         Add mthumb/march=armv7-r multilib.
9892         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9894 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9896         * config/v850/t-rtems: Add more multilibs.
9898 2013-05-10  Richard Biener  <rguenther@suse.de>
9900         PR tree-optimization/57214
9901         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9902         not propagate from SSA names that occur in abnormal PHI nodes.
9904 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9906         * stor-layout.c (element_precision): New function.
9907         * machmode.h (element_precision): Declare it.
9908         * tree.c (build_minus_one_cst): New function.
9909         (element_precision): Likewise.
9910         * tree.h (build_minus_one_cst): Declare new function.
9911         (element_precision): Likewise.
9912         * fold-const.c (operand_equal_p): Use element_precision.
9913         (fold_binary_loc): Handle vector types.
9914         * convert.c (convert_to_integer): Use element_precision.
9915         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9916         separately.
9918 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9920         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9921         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9922         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9923         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9924         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9925         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9926         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9927         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9928         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9929         (Uuw8): New constraints.
9930         (Usb4): Move into alphabetical order.
9931         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9932         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9933         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9934         previously unnamed patterns.
9935         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9936         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9937         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9938         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9939         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9940         of specific lengths.
9942 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9944         * config/i386/i386.md (rotateinv): New code attr.
9945         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9946         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9947         roll $31, %eax, etc.
9949         PR tree-optimization/45216
9950         PR tree-optimization/57157
9951         * tree-ssa-forwprop.c (simplify_rotate): New function.
9952         (ssa_forward_propagate_and_combine): Call it.
9954 2013-05-10  Richard Biener  <rguenther@suse.de>
9956         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9957         disable peeling when we version for aliasing.
9958         (vector_alignment_reachable_p): Honor explicit user alignment.
9959         (vect_supportable_dr_alignment): Likewise.
9960         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9961         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9962         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9963         then peeling to arrange for the cost-model check to come first.
9965 2013-05-10  Alan Modra  <amodra@gmail.com>
9967         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9968         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9969         * configure: Regenerate.
9971 2013-05-10  Alan Modra  <amodra@gmail.com>
9973         PR target/55033
9974         * varasm.c (default_elf_select_section): Move !DECL_P check..
9975         (get_named_section): ..to here before calling get_section_name.
9976         Adjust assertion.
9977         (default_section_type_flags): Add DECL_P check.
9978         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9979         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9981 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9983         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9984         When using gen_stack_adjust_str with a register offset, add a
9985         REG_FRAME_RELATED_EXPR note.
9987 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9989         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9990         (*vec_extractv4si_zext_mem): Ditto.
9991         (*vec_extractv2di): Add 0->x and x->x alternatives.
9992         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9993         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9995 2013-05-09  Jason Merrill  <jason@redhat.com>
9997         N3639 C++1y VLA support
9998         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9999         DECL_VALUE_EXPR.
10001         * tree.c (build_constructor_va): New.
10002         * tree.h: Declare it.
10004 2013-05-09  Martin Jambor  <mjambor@suse.cz>
10006         PR lto/57084
10007         * gimple-fold.c (canonicalize_constructor_val): Call
10008         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10010 2013-05-09  Jan Hubicka  <jh@suse.cz>
10011             Richard Biener  <rguenther@suse.de>
10013         PR lto/54095
10014         * symtab.c (symtab_make_decl_local): Do not add private names.
10016 2013-05-09  Jan Hubicka  <jh@suse.cz>
10018         PR lto/54095
10019         * symtab.c (insert_to_assembler_name_hash): Handle clones.
10020         (unlink_from_assembler_name_hash): Likewise.
10021         (symtab_prevail_in_asm_name_hash, symtab_register_node,
10022         symtab_unregister_node, symtab_initialize_asm_name_hash,
10023         change_decl_assembler_name): Update.
10025 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
10027         * config/aarch64/aarch64.md: New movtf split.
10028         (*movtf_aarch64): Update.
10029         (aarch64_movdi_tilow): Handle TF modes and rename to
10030         aarch64_movdi_<mode>low.
10031         (aarch64_movdi_tihigh): Handle TF modes and rename to
10032         aarch64_movdi_<mode>high
10033         (aarch64_movtihigh_di): Handle TF modes and rename to
10034         aarch64_mov<mode>high_di
10035         (aarch64_movtilow_di): Handle TF modes and rename to
10036         aarch64_mov<mode>low_di
10037         (aarch64_movtilow_tilow): Remove spurious whitespace.
10038         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10039         splits.
10040         (aarch64_print_operand): Update.
10042 2013-05-09  Alan Modra  <amodra@gmail.com>
10044         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10045         powerpc64le.
10046         * configure: Regenerate.
10048 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10050         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10051         splitter preparation statements.
10052         * config/i386/sse.md (*vec_extract* splitters): Ditto.
10053         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10054         adjust_address_nv.
10056 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10058         * gimple-ssa-strength-reduction.c (count_candidates): Change
10059         return value to int.
10060         (analyze_candidates_and_replace): Change type of length to int.
10062 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10064         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10065         (*vec_extract<mode>): Use VI12_128 mode iterator.
10066         (*vec_extract<mode>_mem): Ditto.
10067         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10068         attribute.
10070 2013-05-08  Diego Novillo  <dnovillo@google.com>
10072         PR bootstrap/54659
10074         Revert:
10075         2012-08-17  Diego Novillo  <dnovillo@google.com>
10077         PR bootstrap/54281
10078         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10079         * config.in: Regenerate.
10080         * configure: Regenerate.
10081         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10083 2013-05-08  Jan Hubicka  <jh@suse.cz>
10085         PR lto/54095
10086         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10087         * cgraph.h (symtab_node_base): Add unique_name.
10088         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10089         input_overwrite_node, input_varpool_node): Stream unique_name.
10090         * cgraphclones.c (cgraph_create_virtual_clone,
10091         cgraph_function_versioning): Set unique_name.
10092         * ipa.c (function_and_variable_visibility): Set unique_name.
10094 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10096         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10097         (alloc_cand_and_find_basis): Restrict conditional candidate
10098         processing to CAND_MULTs.
10100 2013-05-08  Jan Hubicka  <jh@suse.cz>
10102         PR lto/54095
10103         lto-symtab.c (lto_symtab_symbol_p): New function.
10104         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10105         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10106         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10107         Skip static symbols.
10109 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
10111         PR tree-optimization/57200
10112         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10113         Only call inform if the preceding warning_at returns true.
10115 2013-05-07  Han Shen  <shenhan@google.com>
10117         * cfgexpand.c (record_or_union_type_has_array_p): New function.
10118         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10119         * common.opt (fstack-protector-strong): New option.
10120         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10121         * doc/invoke.texi (Optimization Options): Document
10122         "-fstack-protector-strong".
10123         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10125 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10127         * config/mips/mips.c (mips_machine_reorg2): Return 0.
10129 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
10131         * ira.c (update_equiv_regs): Add insn having equiv memory even if
10132         it is not lhs of the insn.
10133         (setup_reg_equiv): Remove insn having equiv memory which it is not
10134         lhs of the insn.
10135         * lra-constraints.c (process_address): Try to improve generation
10136         code for address base + disp.
10137         (lra_constraints): Make correct the code for checking insn setting
10138         up backward equivalence.  Remove insn only if it is in the init
10139         insn list.
10140         * lra-eliminations.c (update_reg_eliminate): Change return value.
10141         (lra_eliminate): Use the result.
10143 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
10145         * config/i386/sse.md (ssescalarnummask): New mode attribute.
10146         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10147         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10148         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10149         register target operands.
10150         (*vec_extractv8hi_sse2): New pattern.
10151         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10152         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10153         (*vec_extract<mode>_mem): New insn and split pattern.
10155 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
10157         * config/arm/arm.c (arm_asan_shadow_offset): New function.
10158         (TARGET_ASAN_SHADOW_OFFSET): Define.
10159         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10160         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10162 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10164         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10165         (incr_vec_index): Return -1 if increment not found.
10166         (create_add_on_incoming_edge): Assert if increment not found.
10167         (record_increment): Limit number of increments recorded.
10168         (all_phi_incrs_profitable): Return false if an increment not found.
10169         (replace_profitable_candidates): Don't process increments that were
10170         not recorded.
10171         (analyze_candidates_and_replace): Limit size of incr_vec.
10173 2013-05-07  Richard Biener  <rguenther@suse.de>
10175         * calls.c (special_function_p): setjmp-like functions are leaf.
10176         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10177         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10179 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
10181         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10182         (aarch64_simd_mov<mode>): New expander.
10183         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10184         (aarch64_simd_mov_to_<mode>high): Likewise.
10185         (aarch64_simd_mov_from_<mode>low): Likewise.
10186         (aarch64_simd_mov_from_<mode>high): Likewise.
10187         (aarch64_dup_lane<mode>): Update.
10188         (aarch64_dup_lanedi): New instruction pattern.
10189         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10190         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10192 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10194         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10195         (replace_mult_candidate): Remove unnecessary argument; remove
10196         unnecessary parameter from call to introduce_cast_before_cand.
10197         (replace_unconditional_candidate): Remove unnecessary parameter
10198         from call to replace_mult_candidate.
10199         (replace_conditional_candidate): Likewise.
10200         (insert_initializers): Use make_temp_ssa_name.
10201         (introduce_cast_before_cand): Remove unnecessary argument; use
10202         make_temp_ssa_name.
10203         (replace_one_candidate): Remove unnecessary argument; remove
10204         unnecessary parameter from calls to introduce_cast_before_cand.
10205         (replace_profitable_candidates): Remove unnecessary parameters
10206         from calls to replace_one_candidate.
10208 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10210         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10211         phi def as possibly hiding a basis for a CAND_ADD whose operands
10212         have been commuted in the analysis.
10213         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10215 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10217         * config/aarch64/aarch64.md
10218         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10219         shift value between 0-4.
10221 2013-05-07  Richard Biener  <rguenther@suse.de>
10223         * double-int.h (rshift): New overload.
10224         * double-int.c (rshift): New function.
10225         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10226         (create_reference_ops_from_ref): Remove.
10227         (vn_reference_insert): Use shared ops for constructing the
10228         reference and copy it.
10230 2013-05-07  Richard Biener  <rguenther@suse.de>
10232         PR middle-end/57190
10233         * tree-eh.c (sink_clobbers): Properly propagate
10234         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10236 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
10238         PR tree-optimization/57149
10239         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10240         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10241         collect_phi_def_edges, execute_late_warn_uninitialized): Use
10242         uninit_undefined_value_p instead of ssa_undefined_value_p.
10244         PR debug/57184
10245         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10246         for modifier == EXPAND_INITIALIZER.
10248 2013-05-07  Anton Blanchard  <anton@samba.org>
10250         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10251         for powerpc64 little endian.
10252         * configure: Regenerate.
10254 2013-05-06  Graham Stott  <grahams@btinternet.com>
10256         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10257         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10258         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10259         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10261 2013-05-06  Graham Stott  <grahams@btinternet.com>
10263         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10264         codes which allow non-lvalues.
10266 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
10268         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10269         components are all 1s.
10270         (integer_minus_onep): New function.
10271         * tree.h (integer_minus_onep): Declare it.
10272         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10273         integer_minus_onep instead of integer_all_onesp.
10275 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10277         PR target/52933
10278         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10279         variations of these patterns.
10281 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10283         * config/i386/i386.md (isa): Add x64_sse4 member.
10284         (enabled): Handle x64_sse4.
10285         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10286         instruction for 64bit SSE4_1 targets.  Update insn attributes.
10287         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10288         instruction for SSE4_1 targets.  Update insn attributes.
10289         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10290         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10291         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10292         const_1 selector.
10293         (*vec_extractv4si): Rename from *sse4_1_pextrd.
10294         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10295         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10297 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10299         PR target/57108
10300         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10302 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
10304         * final.c (do_assembler_dialects): Don't handle curly braces and
10305         vertical bar escaped by % as dialect delimiters.
10306         (output_asm_insn): Print curly braces and vertical bar if escaped
10307         by % and ASSEMBLER_DIALECT defined.
10308         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10309         * doc/tm.texi: Regenerated.
10311 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10313         * config/mips/mips.c: Include tree-pass.h.
10314         (mips_reorg): Split in pre- and post-dbr_schedule parts.
10315         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10316         (pass_mips_machine_reorg2): New machine specific pass.
10317         (insert_pass_mips_machine_reorg2): New pass plugin definition.
10318         (mips_option_override): Register the new pass.
10319         * rtl.h (cleanup_barriers): Remove prototype.
10320         (dbr_schedule): Likewise.
10321         * jump.c (cleanup_barriers): Make static.
10322         * reorg.c (dbr_schedule): Likewise.
10324 2013-05-06  Richard Biener  <rguenther@suse.de>
10326         PR tree-optimization/57185
10327         * tree-parloops.c (add_field_for_reduction): Handle anonymous
10328         SSA names properly.
10330 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10332         PR target/57106
10333         * config/i386/i386.c (add_parameter_dependencies): Add dependence
10334         between "first_arg" and "insn", not "last" and "insn".
10336 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10338         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10339         (find_candidates_in_block): Re-enable slsr_process_phi.
10340         (create_phi_basis): Fix double counting of candidate adjustment.
10342 2013-05-06  Richard Biener  <rguenther@suse.de>
10344         PR middle-end/57147
10345         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10346         the edge is also fallthru, preserve it and just clear the
10347         abnormal flag.
10348         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10349         also complex, preserve that and just clear the fallthru flag.
10350         * tree-inline.c (update_ssa_across_abnormal_edges): Also
10351         update virtual operands.
10353 2013-05-06  Alan Modra  <amodra@gmail.com>
10355         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10356         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10357         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10358         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10359         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10360         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10362 2013-05-06  Alan Modra  <amodra@gmail.com>
10364         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10365         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
10366         (DEFAULT_ASM_ENDIAN): Define.
10367         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10368         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10369         Update -K PIC clause from sysv4.h.
10370         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10371         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10373 2013-05-06  Alan Modra  <amodra@gmail.com>
10375         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10376         twice for little-endian.
10377         (ashrdi3_no_power, ashrdi3): Support little-endian.
10379 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10381         PR target/55303
10382         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10383         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10384         related expanders.
10385         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10386         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10387         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10388         New predicates.
10390 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
10391             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
10393         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10394         * config/pa/pa.opt: Make mbig-switch a no-op.
10395         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10396         (CASE_VECTOR_MODE): Always return SImode.
10397         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10398         for the !TARGET_BIG_SWITCH case.
10399         * config/pa/pa-linux.h: Likewise.
10400         * config/pa/pa-openbsd.h: Likewise.
10401         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10402         * config/pa/pa.md (short_jump): Remove define_insn.
10403         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10404         (casesi0): Remove define_insn.
10405         (type): Remove btable_branch.
10406         (pa_combine_type): Likewise.
10407         (in_nullified_branch_delay): Likewise.
10408         (in_call_delay): Likewise.
10409         (define_delay): Likewise.
10410         (define_insn_reservation "Z3"): Likewise.
10411         (define_insn_reservation "Z4"): Likewise.
10412         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10413         (pa_adjust_insn_length): Remove adjustment for btable branches.
10414         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10415         and mno-big-switch
10417 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
10419         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10420         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10421         Add m->r,x alternatives.
10422         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10423         splitters using SWI48x mode iterator.
10424         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
10425         TARGET_64BIT.  Add m->x alternative.
10426         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10427         Add o->x alternative.  Enable for TARGET_SSE.
10428         (sse_storeq): Remove expander.
10429         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10430         with memory input operand.
10431         (*vec_extractv2di_1 splitter): New.
10432         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10433         * config/i386/i386.md (ssevecmodelower): New mode attribute.
10435 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10437         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
10438         (INT_LOWPART): Delete.
10439         (extract_MB): Adjust.
10440         (extract_ME): Adjust.
10441         (print_operand): Adjust.
10443 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10445         * config/rs6000/predicates.md (reg_or_add_cint_operand,
10446         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10447         (reg_or_logical_cint_operand, easy_fp_constant,
10448         logical_const_operand): Delete "CONST_DOUBLE" case.
10449         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10450         "HOST_BITS_PER_WIDE_INT == 64" test.
10451         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
10452         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10453         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10454         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10455         test.
10456         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10457         CONST_DOUBLE DImode/VOIDmode case.
10458         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10459         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10460         CONST_DOUBLE VOIDmode case.
10461         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10462         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10463         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10464         Delete CONST_DOUBLE case.
10465         (splitters for mov FMOVE64 const_double): Delete
10466         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10467         "HOST_BITS_PER_WIDE_INT >= 64" test.
10468         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10469         case.
10470         (mov DI const_double): Delete.
10472 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
10474         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10475         on op shows all bits zero in mode of a lowpart subreg, return zero.
10477 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
10479         PR target/57150
10480         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10481         to save TFmode registers and DImode to save TImode registers for
10482         caller save operations.
10483         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10484         mark being partially clobbered since they only use the first
10485         double word.
10487         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10488         and TDmode only use the upper 64-bits of each VSX register.
10490 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10492         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10493         (find_candidates_in_block): Disable slsr_process_phi.
10495 2013-05-03  Guozhi Wei  <carrot@google.com>
10497         * coverage.c (coverage_obj_init): Move the construction of gcov
10498         constructor to ...
10499         (build_init_ctor): ... here.
10501 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10503         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10504         (slsr_cand_d): Redefine def_phi.
10505         (stride_status, phi_adjust_status, count_phis_status): New enums.
10506         (find_phi_def): New.
10507         (find_basis_for_base_expr): New.
10508         (find_basis_for_candidate): Handle hidden bases.
10509         (alloc_cand_and_find_basis): Handle phi candidates.
10510         (slsr_process_phi): New.
10511         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10512         (create_mul_imm_cand): Likewise.
10513         (create_add_ssa_cand): Exclude phi base candidates.
10514         (create_add_imm_cand): Likewise.
10515         (slsr_process_cast): Likewise.
10516         (slsr_process_copy): Likewise.
10517         (find_candidates_in_block): Handle phi candidates.
10518         (dump_candidate): Likewise.
10519         (unconditional_cands): Delete.
10520         (unconditional_cands_with_known_stride_p): Delete.
10521         (phi_dependent_cand_p): New.
10522         (cand_increment): Handle phi-dependent candidates.
10523         (replace_dependent): Delete.
10524         (replace_mult_candidate): New.
10525         (replace_unconditional_candidate): New.
10526         (incr_vec_index): Move to avoid forward reference.
10527         (create_add_on_incoming_edge): New.
10528         (create_phi_basis): New.
10529         (replace_dependents): Delete.
10530         (replace_conditional_candidate): New.
10531         (phi_add_costs): New.
10532         (replace_uncond_cands_and_profitable_phis): New.
10533         (record_increment): Handle phi adjustments.
10534         (record_phi_increments): New.
10535         (record_increments): Handle phi adjustments.
10536         (phi_incr_cost): New.
10537         (lowest_cost_path): Handle phis.
10538         (total_savings): Likewise.
10539         (analyze_increments): Likewise.
10540         (ncd_with_phi): New.
10541         (ncd_of_cand_and_phis): New.
10542         (nearest_common_dominator_for_cands): Handle phi increments.
10543         (all_phi_incrs_profitable): New.
10544         (replace_profitable_candidates): Handle phi-dependent candidates.
10545         (analyze_candidates_and_replace): Likewise.
10547 2013-05-03  Teresa Johnson  <tejohnson@google.com>
10549         PR bootstrap/57154
10550         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10551         do not exceed REG_BR_PROB_BASE.
10553 2013-05-03  Jeff Law  <law@redhat.com>
10555         PR tree-optimization/57144
10556         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10557         operand of the condition will bit into the new type when eliminating
10558         a cast feeding a condition.
10560 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
10562         PR rtl-optimization/57130
10563         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10564         of COMPARE as in_code to the recursive call if needed.
10566 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
10568         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10569         (enabled): Handle new members.
10570         * config/i386/sse.md (*vec_concatv2si): Merge from
10571         *vec_concatv2si_sse2 and vec_concatv2si_sse.
10572         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10574 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
10576         PR tree-optimization/57027
10577         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10578         for fnms opportunity, check we got the prerequisite kind
10579         of tree / gimple before using accessor functions.
10581 2013-05-03  Richard Biener  <rguenther@suse.de>
10583         * double-int.h (lshift): New overload without precision
10584         and arith argument.
10585         (operator *=, operator +=, operator -=): Move ...
10586         * double-int.c (operator *=, operator +=, operator -=): ... here
10587         and implement more efficiently.
10588         (mul_double_with_sign): Remove.
10589         (lshift_double): Adjust to take unsinged shift argument, push
10590         dispatching code to callers.
10591         (mul_double_wide_with_sign): Add early out for callers that
10592         are not interested in high parts or overflow.
10593         (lshift): New function.
10594         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10595         dispatch code here.
10596         (lrotate, rrotate): Use logical shifts.
10597         * expr.c (get_inner_reference): Use lshift.
10598         * fixed-value.c (do_fixed_divide): Likewise.
10599         * tree-dfa.c (get_ref_base_and_extent): Likewise.
10600         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10601         (indirect_refs_may_alias_p): Likewise.
10602         (stmt_kills_ref_p_1): Likewise.
10604 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10606         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10608 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10610         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10611         scalar form of FABD instruction.
10613 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
10615         * lra-constraints.c (process_alt_operands): Add checking alt
10616         number to choose the best alternative.
10618 2013-05-02  Richard Biener  <rguenther@suse.de>
10620         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10621         bitmap and its handling.
10622         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10624 2013-05-02  Richard Biener  <rguenther@suse.de>
10626         PR middle-end/57140
10627         * tree-inline.c (copy_loops): Properly handle removed loops.
10628         (copy_cfg_body): Mark destination loops for fixup if source
10629         loops needed fixup.
10631 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
10633         PR target/56732
10634         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10635         generating simple_return for naked functions.
10637 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10639         PR middle-end/56988
10640         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10641         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10642         flags match.
10643         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10644         ipa_agg_replacement_value structures.
10645         (known_aggs_to_agg_replacement_list): Likewise.
10646         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10647         (read_agg_replacement_chain): Likewise.
10648         (ipcp_transform_function): Also check that by_ref flags match.
10650 2013-05-02  Richard Biener  <rguenther@suse.de>
10652         * graphds.h (struct graph): Add obstack member.
10653         * graphds.c (new_graph): Initialize obstack and allocate
10654         vertices from it.
10655         (add_edge): Allocate edge from the obstack.
10656         (free_graph): Free the obstack instead of all edges and vertices.
10658 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10660         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10661         divides.
10662         * cfg.c (update_bb_profile_for_threading): Ditto.
10663         * tree-inline.c (copy_bb): Ditto.
10664         (copy_edges_for_bb): Ditto.
10665         (initialize_cfun): Ditto.
10666         (copy_cfg_body): Ditto.
10667         (expand_call_inline): Ditto.
10668         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10669         (estimate_node_size_and_time): Ditto.
10670         (inline_merge_summary): Ditto.
10671         * cgraphclones.c (cgraph_clone_edge): Ditto.
10672         (cgraph_clone_node): Ditto.
10673         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10674         (compute_trg_info): Ditto.
10676 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10678         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10679         S reg when fp attribute set.
10680         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10682 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10684         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10685         New pattern.
10686         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10687         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10688         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10690 2013-05-02  Richard Biener  <rguenther@suse.de>
10692         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10693         (struct instantiate_cache_entry): New type.
10694         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10695         (struct instantiate_cache_type): New type.
10696         (set_instantiated_value, get_instantiated_value): Remove.
10697         (get_instantiated_value_entry): New function.
10698         (instantiate_scev_name): Use the new cache and adjust.
10699         (instantiate_scev_poly): Adjust.
10700         (instantiate_scev_binary): Likewise.
10701         (instantiate_array_ref): Likewise.
10702         (instantiate_scev_convert): Likewise.
10703         (instantiate_scev_not): Likewise.
10704         (instantiate_scev_3): Likewise.
10705         (instantiate_scev_2): Likewise.
10706         (instantiate_scev_r): Likewise.
10707         (instantiate_scev): Likewise.
10708         (resolve_mixers): Likewise.
10710 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10712         PR target/57091
10713         * lra-constraints.c (best_small_class_operands_num): Remove.
10714         (process_alt_operands): Remove small_class_operands_num.  Take
10715         small classes operands into losers and only if the operand is not
10716         matched.  Modify debugging output.
10717         (curr_insn_transform): Remove best_small_class_operands_num.
10718         Print insn name.
10720 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10722         * config/aarch64/aarch64-builtins.c
10723         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10724         * config/aarch64/aarch64-simd-builtins.def
10725         (reduc_splus_): Add new modes.
10726         (reduc_uplus_): New.
10727         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10728         (reduc_uplus_v4sf): Likewise.
10729         (reduc_splus_v4sf): Likewise.
10730         (aarch64_addv<mode>): Likewise.
10731         (reduc_uplus_<mode>): Likewise.
10732         (reduc_splus_<mode>): Likewise.
10733         (aarch64_addvv2di): Likewise.
10734         (reduc_uplus_v2di): Likewise.
10735         (reduc_splus_v2di): Likewise.
10736         (aarch64_addvv2si): Likewise.
10737         (reduc_uplus_v2si): Likewise.
10738         (reduc_splus_v2si): Likewise.
10739         (reduc_<sur>plus_<mode>): New.
10740         (reduc_<sur>plus_v2di): Likewise.
10741         (reduc_<sur>plus_v2si): Likewise.
10742         (reduc_<sur>plus_v4sf): Likewise.
10743         (aarch64_addpv4sf): Likewise.
10744         * config/aarch64/arm_neon.h
10745         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10746         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10747         add UNSPEC_SADDV, UNSPEC_UADDV.
10748         (SUADDV): New.
10749         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10751 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10753         * config/aarch64/arm_neon.h
10754         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10756 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10758         * config/aarch64/aarch64-builtins
10759         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10761 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10763         * config/aarch64/aarch64-simd-builtins.def
10764         (reduc_smax_): New.
10765         (reduc_smin_): Likewise.
10766         (reduc_umax_): Likewise.
10767         (reduc_umin_): Likewise.
10768         (reduc_smax_nan_): Likewise.
10769         (reduc_smin_nan_): Likewise.
10770         (fmax): Remove.
10771         (fmin): Likewise.
10772         (smax): Update for V2SF, V4SF and V2DF modes.
10773         (smin): Likewise.
10774         (smax_nan): New.
10775         (smin_nan): Likewise.
10776         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10777         (<su><maxmin><mode>3): ...This, refactor.
10778         (s<maxmin><mode>3): New.
10779         (<maxmin_uns><mode>3): Likewise.
10780         (reduc_<maxmin_uns>_<mode>): Refactor.
10781         (reduc_<maxmin_uns>_v4sf): Likewise.
10782         (reduc_<maxmin_uns>_v2si): Likewise.
10783         (aarch64_<fmaxmin><mode>: Remove.
10784         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10785         new builtin names.
10786         (vmin<q>_f<32,64>): Likewise.
10787         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10788         (FMAXMIN): New.
10789         (su): Add mappings for smax, smin, umax, umin.
10790         (maxmin): New.
10791         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10792         (FMAXMIN): Rename as...
10793         (FMAXMIN_UNS): ...This.
10794         (maxminv): Remove.
10795         (fmaxminv): Likewise.
10796         (fmaxmin): Likewise.
10797         (maxmin_uns): New.
10798         (maxmin_uns_op): Likewise.
10800 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10802         * config/aarch64/arm_neon.h
10803         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10804         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10805         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10807 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10809         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10810         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10812 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10814         * config/aarch64/aarch64-simd.md
10815         (vcond<mode>_internal): Handle special cases for constant masks.
10816         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10817         (vcondu<mode><mode>): Likewise.
10818         (vcond<v_cmp_result><mode>): New.
10820 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10822         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10823         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10824         * config/aarch64/aarch64-simd-builtins.def
10825         (cmeq): Update to BUILTIN_VALLDI.
10826         (cmgt): Likewise.
10827         (cmge): Likewise.
10828         (cmle): Likewise.
10829         (cmlt): Likewise.
10830         * config/aarch64/arm_neon.h
10831         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10832         to builtins or C as appropriate.
10834 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10836         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10837         (cmgeu): ...This.
10838         (cmhi): Rename to...
10839         (cmgtu): ...This.
10840         * config/aarch64/aarch64-simd.md
10841         (simd_mode): Add SF.
10842         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10843         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10844         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10845         (cstore<mode>_neg): ...This.
10846         * config/aarch64/iterators.md
10847         (VALLF): new.
10848         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10849         (COMPARISONS): New.
10850         (UCOMPARISONS): Likewise.
10851         (optab): Add missing comparisons.
10852         (n_optab): New.
10853         (cmp_1): Likewise.
10854         (cmp_2): Likewise.
10855         (CMP): Likewise.
10856         (cmp): Remove.
10857         (VCMP_S): Likewise.
10858         (VCMP_U): Likewise.
10859         (V_cmp_result): Add DF, SF modes.
10860         (v_cmp_result): Likewise.
10861         (v): Likewise.
10862         (vmtype): Likewise.
10863         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10865 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10867         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10868         define_insn to define_insn_and_split.
10869         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10870         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10871         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10872         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10873         (thumb2_negscc): Likewise.
10875 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10877         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10879 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10881         * config/arm/thumb2.md: Remove trailing whitespaces.
10883 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10885         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10886         Use gen_int_mode rather than GEN_INT.
10888 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10890         * value-prof.c (stream_in_histogram_value): Remove the strayed
10891         debug_gimple_stmt.
10893 2013-04-30  Richard Biener  <rguenther@suse.de>
10895         PR middle-end/57122
10896         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10898 2013-04-30  Richard Biener  <rguenther@suse.de>
10900         PR middle-end/57107
10901         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10903 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10905         PR rtl-optimization/56957
10906         PR rtl-optimization/57105
10907         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10908         variable.  Use just INSN_UID for determining whether an insn
10909         should be only disconnected from the insn stream.
10910         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10912 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10914         PR tree-optimization/57104
10915         * tsan.c (instrument_expr): Don't instrument accesses to
10916         DECL_HARD_REGISTER VAR_DECLs.
10918 2013-04-30  Richard Biener  <rguenther@suse.de>
10920         * function.h (loops_for_fn): New inline function.
10921         (set_loops_for_fn): Likewise.
10922         * cfgloop.h (place_new_loop): Add struct function parameter.
10923         (get_loop): Likewise.
10924         (get_loops): Likewise.
10925         (number_of_loops): Likewise.
10926         (fel_next): Adjust.
10927         (fel_init): Likewise.
10928         * cfg.c (get_loop_copy): Adjust.
10929         * cfgloop.c (flow_loops_dump): Likewise.
10930         (record_loop_exits): Likewise.
10931         (verify_loop_structure): Likewise.
10932         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10933         (estimate_reg_pressure_cost): Likewise.
10934         (mark_loop_exit_edges): Likewise.
10935         * cfgloopmanip.c (place_new_loop): Likewise.
10936         (add_loop): Likewise.
10937         (duplicate_loop): Likewise.
10938         * graph.c (draw_cfg_nodes): Likewise.
10939         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10940         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10941         (extract_affine_chrec): Likewise.
10942         (build_scop_iteration_domain): Likewise.
10943         * graphite.c (graphite_initialize): Likewise.
10944         * ira-build.c (create_loop_tree_nodes): Likewise.
10945         (more_one_region_p): Likewise.
10946         (rebuild_regno_allocno_maps): Likewise.
10947         (mark_loops_for_removal): Likewise.
10948         (mark_all_loops_for_removal): Likewise.
10949         (remove_unnecessary_regions): Likewise.
10950         (ira_build): Likewise.
10951         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10952         * loop-init.c (fix_loop_structure): Likewise.
10953         (gate_rtl_move_loop_invariants): Likewise.
10954         (gate_rtl_unswitch): Likewise.
10955         (gate_rtl_unroll_and_peel_loops): Likewise.
10956         (rtl_doloop): Likewise.
10957         * lto-streamer-in.c (input_cfg): Likewise.
10958         * lto-streamer-out.c (output_cfg): Likewise.
10959         * modulo-sched.c (sms_schedule): Likewise.
10960         * predict.c (tree_estimate_probability): Likewise.
10961         (tree_estimate_probability_driver): Likewise.
10962         (estimate_loops): Likewise.
10963         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10964         (move_sese_region_to_fn): Likewise.
10965         (debug_loop_num): Likewise.
10966         * tree-chrec.c (chrec_evaluate): Likewise.
10967         (hide_evolution_in_other_loops_than_loop): Likewise.
10968         (chrec_component_in_loop_num): Likewise.
10969         (reset_evolution_in_loop): Likewise.
10970         (evolution_function_is_invariant_rec_p): Likewise.
10971         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10972         * tree-inline.c (copy_loops): Likewise.
10973         (copy_cfg_body): Likewise.
10974         (tree_function_versioning): Likewise.
10975         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10976         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10977         Likewise.
10978         (add_to_evolution_1): Likewise.
10979         (scev_const_prop): Likewise.
10980         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10981         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10982         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10983         (tree_ssa_lim_initialize): Likewise.
10984         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10985         (verify_loop_closed_ssa): Likewise.
10986         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10987         (tree_ssa_loop_im): Likewise.
10988         (tree_ssa_loop_unswitch): Likewise.
10989         (tree_vectorize): Likewise.
10990         (check_data_deps): Likewise.
10991         (tree_ssa_loop_ivcanon): Likewise.
10992         (tree_ssa_loop_bounds): Likewise.
10993         (tree_complete_unroll): Likewise.
10994         (tree_complete_unroll_inner): Likewise.
10995         (tree_parallelize_loops): Likewise.
10996         (tree_ssa_loop_prefetch): Likewise.
10997         (tree_ssa_loop_ivopts): Likewise.
10998         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10999         * tree-vectorizer.c (vectorize_loops): Likewise.
11001 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
11003         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11004         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11005         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11006         with EABI_LINK_SPEC.
11008 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11010         PR target/44578
11011         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11012         alternative.
11014 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
11016         PR target/57097
11017         * lra-constraints.c (process_alt_operands): Discourage a bit more
11018         using memory for pseudos.  Print cost dump for alternatives.
11019         Modify cost values for conflicts with early clobbers.
11020         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11022 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11024         PR target/57098
11025         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11027 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11029         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11030         from/to S register.
11031         (movdi_aarch64): Support LDR/STR from/to D register.
11033 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11035         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11036         or higher by default.
11038 2013-04-29  Richard Biener  <rguenther@suse.de>
11040         PR middle-end/57075
11041         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11042         even if not adding abnormal edges for calls that can make
11043         abnormal gotos.
11045 2013-04-29  Richard Biener  <rguenther@suse.de>
11047         PR middle-end/57103
11048         * tree-cfg.c (move_stmt_op): Fix condition under which to update
11049         TREE_BLOCK.
11050         (move_stmt_r): Remove redundant checking.
11052 2013-04-29  Teresa Johnson  <tejohnson@google.com>
11054         PR bootstrap/57077
11055         * basic-block.h (apply_scale): New function.
11056         (apply_probability): Use apply_scale.
11057         * gimple-streamer-in.c (input_bb): Ditto.
11058         * lto-streamer-in.c (input_cfg): Ditto.
11059         * lto-cgraph.c (merge_profile_summaries): Ditto.
11060         * tree-optimize.c (execute_fixup_cfg): Ditto.
11061         * tree-inline.c (copy_bb): Update comment to use apply_scale.
11062         (copy_edges_for_bb): Ditto.
11063         (copy_cfg_body): Ditto.
11065 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11067         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11068         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11069         (tail_merge_optimize): Handle current_loops == NULL.
11071 2013-04-26  Jeff Law  <law@redhat.com>
11073         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11074         (simplify_cond_using_ranges): Generalize code to simplify
11075         COND_EXPRs where one argument is a constant and the other
11076         is an SSA_NAME created by an integral type conversion.
11078 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11080         * config/arm/arm.md (store_minmaxsi): Use only when
11081         optimize_insn_for_size_p.
11083 2013-04-29  Christian Bruel  <christian.bruel@st.com>
11085         PR target/57108
11086         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11088 2013-04-29  Richard Biener  <rguenther@suse.de>
11090         PR middle-end/57089
11091         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11092         loop tree make sure to schedule a fixup for the child as well.
11093         (expand_omp_for_generic): Properly add loops.
11094         (expand_omp_for_static_nochunk): Likewise.
11095         (expand_omp_for_static_chunk): Likewise.
11096         (expand_omp_for): For the degenerate case fixup loops.
11097         (expand_omp_sections): Fix default bb placement in loops.
11098         (expand_omp_atomic_pipeline): Properly add loops.
11100 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11102         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11104 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11106         * tree-ssa-tail-merge.c: Update header comment.
11108 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11110         * config/aarch64/arm_neon.h
11111         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11112         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11113         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11114         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11115         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11116         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11117         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11118         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11120 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11122         * config/aarch64/aarch64-simd.md
11123         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11124         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11125         fix_trunc, fixuns_trunc.
11126         (ftrunc<VDQF:mode>2): New.
11127         * config/aarch64/iterators.md (optab): Add fix, fixuns.
11128         (fix_trunc_optab): New.
11130 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11132         * config/aarch64/aarch64-builtins.c
11133         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11134         iceilf, lround, iroundf.
11136 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11138         PR target/54349
11139         * config/i386/i386.h (enum ix86_tune_indices)
11140         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11141         New, split from X86_TUNE_INTER_UNIT_MOVES.
11142         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11143         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11144         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11145         (TARGET_INTER_UNIT_MOVES): Remove.
11146         * config/i386/i386.c (initial_ix86_tune_features): Update.
11147         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11148         (ix86_expand_convert_uns_didf_sse): Use
11149         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11150         (ix86_expand_vector_init_one_nonzero): Ditto.
11151         (ix86_expand_vector_init_interleave): Ditto.
11152         (inline_secondary_memory_needed): Return true for moves from SSE class
11153         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11154         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11155         * config/i386/constraints.md (Yi, Ym): Depend on
11156         TARGET_INTER_UNIT_MOVES_TO_VEC.
11157         (Yj, Yn): New constraints.
11158         * config/i386/i386.md (*movdi_internal): Change constraints of
11159         operand 1 from Yi to Yj and from Ym to Yn.
11160         (*movsi_internal): Ditto.
11161         (*movdf_internal): Ditto.
11162         (*movsf_internal): Ditto.
11163         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11164         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11165         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11166         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11167         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11168         * config/i386/sse.md (movdi_to_sse): Ditto.
11169         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11170         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11171         TARGET_INTER_UNIT_MOVES.
11172         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11173         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11174         instead of TARGET_INTER_UNIT_MOVES.
11175         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11176         operand 1 from Yi to Yj and from Ym to Yn.
11178 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11180         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11181         (float_truncate_hi_): Likewise.
11182         (float_extend_lo_): Likewise.
11183         (float_truncate_lo_): Likewise.
11184         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11185         (aarch64_float_extend_lo_v2df): Likewise.
11186         (vec_unpacks_hi_v4sf): Likewise.
11187         (aarch64_float_truncate_lo_v2sf): Likewise.
11188         (aarch64_float_truncate_hi_v4sf): Likewise.
11189         (vec_pack_trunc_v2df): Likewise.
11190         (vec_pack_trunc_df): Likewise.
11192 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11194         * config/aarch64/aarch64-builtins.c
11195         (aarch64_fold_builtin): Fold float conversions.
11196         * config/aarch64/aarch64-simd-builtins.def
11197         (floatv2si, floatv4si, floatv2di): New.
11198         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11199         * config/aarch64/aarch64-simd.md
11200         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11201         * config/aarch64/iterators.md (FLOATUORS): New.
11202         (optab): Add float, floatuns.
11203         (su_optab): Likewise.
11205 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11207         * config/aarch64/aarch64-builtins.c
11208         (aarch64_builtin_vectorized_function): Use new names for
11209         fcvt builtins.
11210         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11211         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11212         (fcvtzu): Split as...
11213         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11214         (fcvtas): Split as...
11215         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11216         (fcvtau): Split as...
11217         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11218         (fcvtps): Split as...
11219         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11220         (fcvtpu): Split as...
11221         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11222         (fcvtms): Split as...
11223         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11224         (fcvtmu): Split as...
11225         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11226         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11227         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11228         (lfrintnusf, lfrintnudf): Likewise.
11229         * config/aarch64/aarch64-simd.md
11230         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11231         define_insn.
11232         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11233         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11234         (fcvt_pattern): Likewise.
11236 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11238         * config/aarch64/aarch64-simd.md
11239         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11240         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11242 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11244         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11245         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11246         (vrnd<a,m,n,p>_f32): Implement using builtins.
11247         (vrnd<i,x><q>_f<32, 64>): New.
11249 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11251         * config/aarch64/aarch64-builtins.c
11252         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11253         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11254         (frintz): Rename to...
11255         (btrunc): ...this.
11256         (frintp): Rename to...
11257         (ceil): ...this.
11258         (frintm): Rename to...
11259         (floor): ...this.
11260         (frinti): Rename to...
11261         (nearbyint): ...this.
11262         (frintx): Rename to...
11263         (rint): ...this.
11264         (frinta): Rename to...
11265         (round): ...this.
11266         * config/aarch64/aarch64-simd.md
11267         (aarch64_frint<frint_suffix><mode>): Delete.
11268         (<frint_pattern><mode>2): Convert to insn.
11269         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11270         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11271         (frint_pattern): Likewise.
11272         (frint_suffix): Likewise.
11274 2013-04-29  Richard Biener  <rguenther@suse.de>
11276         PR tree-optimization/57081
11277         * loop-init.c: Include tree-flow.h.
11278         (loop_optimizer_finalize): Free number of iteration estimates.
11279         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11281 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
11283         PR tree-optimization/57083
11284         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11285         non-singleton shift count range, zero extend low_bound for uns case.
11287         * config/i386/predicates.md (general_vector_operand): New predicate.
11288         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11289         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11290         if they aren't nonimmediate operands.  If their original values
11291         satisfy const_vector_equal_evenodd_p, don't shift them.
11292         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11293         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
11294         if not nonimmediate_operand.
11295         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11296         instead of register_operand.
11297         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11299 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
11301         * stor-layout.c (finalize_size_functions): Allocate a structure and
11302         reset cfun before dumping the functions.
11304 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
11306         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11308         PR target/56866
11309         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11310         use xop_pmacsdqh if uns_p.
11311         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11312         the immediate rotate count.
11314 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
11316         * rtl.h (struct rtx_def): Add comment for field jump.
11317         (LRA_SUBREG_P): New macro.
11318         * recog.c (register_operand): Check LRA_SUBREG_P.
11319         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11320         stack frame.
11321         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11322         (lra_final_code_change): Skip subreg change for operators.
11323         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11324         if there are no operand changes.
11325         * lra-constraints.c (curr_insn_set): New.
11326         (match_reload): Set LRA_SUBREG_P.
11327         (emit_spill_move): Ditto.
11328         (check_and_process_move): Use curr_insn_set. Process only single
11329         set insns.  Don't initialize sec_mem_p and change_p.
11330         (simplify_operand_subreg): Use LRA_SUBREG_P.
11331         (reg_in_class_p): New function.
11332         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11333         of #ifdef.  Add code to remove cycling.
11334         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11335         non-null disp.  Reload inner instead of disp when base and index
11336         are null.  Try to put lo_sum into register.
11337         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11338         (check_and_process_move): Move code for move cost check to
11339         simple_move_p.  Remove equiv_substitution.
11340         (simple_move_p): New function.
11341         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11342         curr_insn_set.  Call check_and_process_move only for single set
11343         insns.  Use the new function.  Move call of check_and_process_move
11344         after operand equiv substitution and address process.
11346 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11348         PR go/57045
11349         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11350         with nonlocal goto receivers or returns twice calls, ignore
11351         unininitialized values from abnormal edges to nl goto receiver
11352         or returns twice call.
11354 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11356         PR tree-optimization/57051
11357         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11358         and VEC_RSHIFT_EXPR if shift count is a multiple of element
11359         bitsize.
11361 2013-04-26  Richard Biener  <rguenther@suse.de>
11363         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11364         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
11365         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11366         (fixup_loop_arrays_after_move): New function.
11367         (move_sese_region_to_fn): Properly outline the loop tree parts
11368         of the SESE region.
11370 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
11372         * config/i386/i386.md (type, unit): Fix long lines.
11374 2013-04-26  Richard Biener  <rguenther@suse.de>
11376         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11377         (lto-streamer-out.o): Likewise.
11378         * cfgloop.c (init_loops_structure): Export, add struct function
11379         argument and adjust.
11380         (flow_loops_find): Adjust.
11381         * cfgloop.h (enum loop_estimation): Add EST_LAST.
11382         (init_loops_structure): Declare.
11383         * lto-streamer-in.c: Include cfgloop.h.
11384         (input_cfg): Input the loop tree.
11385         * lto-streamer-out.c: Include cfgloop.h.
11386         (output_cfg): Output the loop tree.
11387         (output_struct_function_base): Do not drop PROP_loops.
11389 2013-03-26  Richard Biener  <rguenther@suse.de>
11391         * tree-cfg.c (execute_build_cfg): Build the loop tree.
11392         (pass_build_cfg): Provide PROP_loops.
11393         (move_sese_region_to_fn): Remove loops that are outlined into fn
11394         for now.
11395         * tree-inline.c: Include cfgloop.h.
11396         (initialize_cfun): Do not drop PROP_loops.
11397         (copy_loops): New function.
11398         (copy_cfg_body): Copy loop structure.
11399         (tree_function_versioning): Initialize destination loop tree.
11400         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11401         (pass_parallelize_loops): Do IL verification.
11402         * loop-init.c (loop_optimizer_init): Fixup loops if required.
11403         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11404         the CFG make sure we fixup loops as well.
11405         * tree-ssa-tail-merge.c: Include cfgloop.h.
11406         (replace_block_by): When merging loop latches mark loops for fixup.
11407         * lto-streamer-out.c (output_struct_function_base): Drop
11408         PROP_loops for now.
11409         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11410         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11411         * ipa-split.c: Include cfgloop.h.
11412         (split_function): Add the new return block to the loop tree root.
11413         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11414         whether we have removed the forwarder block.
11415         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11416         * cfgloop.h (place_new_loop): Declare.
11417         * cfgloopmanip.c (place_new_loop): Export.
11418         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11419         (tree-switch-conversion.o): Likewise.
11420         (tree-complex.o): Likewise.
11421         (tree-inline.o): Likewise.
11422         (tree-ssa-tailmerge.o): Likewise.
11423         (ipa-split.o): Likewise.
11424         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11425         (tree-ssa-copy.o): Likewise.
11426         * tree-switch-conversion.c: Include cfgloop.h
11427         (process_switch): If we emit a bit-test cascade, schedule loops
11428         for fixup.
11429         * tree-complex.c: Include cfgloop.h.
11430         (expand_complex_div_wide): Properly add new basic-blocks to loops.
11431         * asan.c: Include cfgloop.h.
11432         (create_cond_insert_point): Properly add new basic-blocks to
11433         loops, schedule loop fixup.
11434         * cfgloop.c (verify_loop_structure): Check that looks are not
11435         marked for fixup.
11436         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11437         to loops.
11438         (expand_omp_for_generic): Likewise.
11439         (expand_omp_sections): Likewise.
11440         (expand_omp_atomic_pipeline): Schedule loops for fixup.
11441         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11442         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11443         is initialized, not when loops are present.
11444         * tree-parloops.c (parallelize_loops): Remove checking here.
11445         * passes.c (init_optimization_passes): Schedule a copy-propagation
11446         pass before complete unrolling of inner loops.
11448 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11450         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11451         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11452         (colorize_init): Add argument to _WIN32 version.
11453         * toplev.c: Include diagnostic-color.h.
11454         (process_options): Default to -fdiagnostics-color=auto if
11455         GCC_COLORS env var is in the environment.
11456         * common.opt (fdiagnostics-color=): Add Var and Init.
11457         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11458         env var is in the environment, the default is auto rather than never.
11460         * diagnostic.h (file_name_as_prefix): Add context argument.
11461         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
11462         the string as locus.
11463         * langhooks.c (lhd_print_error_function): Adjust caller.
11465 2013-04-25  Lawrence Crowl  <crowl@google.com>
11467         * var-tracking.c (shared_hash_def::htab):
11468         Change type to hash_table.  Update dependent calls and types.
11470 2013-04-25  Lawrence Crowl  <crowl@google.com>
11472         * Makefile.in: Update as needed below.
11474         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11475         Move declaration to after the type's method definitons.
11477         * attribs.c (htab_t scoped_attributes::attribute_hash):
11478         Change type to hash_table.  Update dependent calls and types.
11480         * bitmap.c (htab_t bitmap_desc_hash):
11481         Change type to hash_table.  Update dependent calls and types.
11483         * cselib.c (htab_t cselib_hash_table):
11484         Change type to hash_table.  Update dependent calls and types.
11486         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11487         (hash_string_slot_node): Move implementation into lto-streamer.h
11488         struct string_slot_hasher.
11489         (eq_string_slot_node): Likewise.
11491         * data-streamer-out.c: Update output_block::string_hash_table
11492         dependent calls and types.
11494         * dwarf2cfi.c (htab_t trace_index):
11495         Change type to hash_table.  Update dependent calls and types.
11497         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11498         Change type to hash_table.  Update dependent calls and types.
11499         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11500         (htab_t optimize_external_refs::map): Likewise.
11501         (htab_t output_comp_unit::extern_map): Likewise.
11502         (htab_t output_comdat_type_unit::extern_map): Likewise.
11503         (htab_t output_macinfo::macinfo_htab): Likewise.
11504         (htab_t optimize_location_lists::htab): Likewise.
11505         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11507         * except.c (htab_t ehspec_hash_type):
11508         Change type to hash_table.  Update dependent calls and types.
11509         (assign_filter_values::ttypes): Likewise.
11510         (assign_filter_values::ehspec): Likewise.
11511         (sjlj_assign_call_site_values::ar_hash): Likewise.
11512         (convert_to_eh_region_ranges::ar_hash): Likewise.
11514         * gcse.c (htab_t pre_ldst_table):
11515         Change type to hash_table.  Update dependent calls and types.
11517         * ggc-common.c (htab_t saving_htab):
11518         Change type to hash_table.  Update dependent calls and types.
11519         (htab_t loc_hash): Likewise.
11520         (htab_t ptr_hash): Likewise.
11521         (call_count): Rename ggc_call_count.
11522         (call_alloc): Rename ggc_call_alloc.
11523         (loc_descriptor): Rename make_loc_descriptor.
11524         (add_statistics): Rename ggc_add_statistics.
11526         * ggc-common.c (saving_htab):
11527         Change type to hash_table.  Update dependent calls and types.
11529         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11530         (push_gimplify_context): Likewise.
11531         (pop_gimplify_context): Likewise.
11532         (struct gimple_temp_hash_elt): Added.
11533         (struct gimplify_hasher): Likewise.
11534         (struct gimplify_ctx.temp_htab):
11535         Change type to hash_table.  Update dependent calls and types.
11537         * gimple-fold.c: Include gimplify-ctx.h.
11539         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11540         Change type to hash_table.  Update dependent calls and types.
11541         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11542         avoid potential global name collision.
11544         * gimplify.c: Include gimplify-ctx.h.
11545         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11546         (htab_t gimplify_ctx::temp_htab):
11547         Update dependent calls and types for new type hash_table.
11548         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11549         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11551         * gimplify-ctx.h: New.
11552         (struct gimple_temp_hash_elt): Move from gimplify.c.
11553         (class gimplify_hasher): New.
11554         (struct gimplify_ctx): Move from gimple.h.
11555         (htab_t gimplify_ctx::temp_htab):
11556         Change type to hash_table.  Update dependent calls and types.
11558         * graphite-clast-to-gimple.c: Include graphite-htab.h.
11559         (htab_t ivs_params::newivs_index):
11560         Change type to hash_table.  Update dependent calls and types.
11561         (htab_t ivs_params::params_index): Likewise.
11562         (htab_t print_generated_program::params_index): Likewise.
11563         (htab_t gloog::newivs_index): Likewise.
11564         (htab_t gloog::params_index): Likewise.
11566         * graphite.c: Include graphite-htab.h.
11567         4htab_t graphite_transform_loops::bb_pbb_mapping):
11568         Change type to hash_table.  Update dependent calls and types.
11570         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11571         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11572         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11574         * graphite-dependences.c: Include graphite-htab.h.
11575         (loop_is_parallel_p): Change hash table type of parameter.
11577         * graphite-htab.h: New.
11578         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11579         (extern find_pbb_via_hash): Move from graphite-poly.h.
11580         (extern loop_is_parallel_p): Move from graphite-poly.h.
11581         (extern get_loop_body_pbbs): Move from graphite-poly.h.
11583         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11584         (extern loop_is_parallel_p): Move to graphite-htab.h.
11585         (extern get_loop_body_pbbs): Move to graphite-htab.h.
11587         * haifa-sched.c (htab_t delay_htab):
11588         Change type to hash_table.  Update dependent calls and types.
11589         (htab_t delay_htab_i2): Likewise.
11591         * ira-color.c (htab_t allocno_hard_regs_htab):
11592         Change type to hash_table.  Update dependent calls and types.
11594         * ira-costs.c (htab_t cost_classes_htab):
11595         Change type to hash_table.  Update dependent calls and types.
11597         * loop-invariant.c (htab_t merge_identical_invariants::eq):
11598         Change type to hash_table.  Update dependent calls and types.
11600         * loop-iv.c (htab_t bivs):
11601         Change type to hash_table.  Update dependent calls and types.
11603         * loop-unroll.c (htab_t opt_info::insns_to_split):
11604         Change type to hash_table.  Update dependent calls and types.
11605         (htab_t opt_info::insns_with_var_to_expand): Likewise.
11607         * lto-streamer.h (struct string_slot): Move from data-streamer.h
11608         (struct string_slot_hasher): New.
11609         (htab_t output_block::string_hash_table):
11610         Change type to hash_table.  Update dependent calls and types.
11612         * lto-streamer-in.c (freeing_string_slot_hasher): New.
11613         (htab_t file_name_hash_table):
11614         Change type to hash_table.  Update dependent calls and types.
11616         * lto-streamer-out.c: Update output_block::string_hash_table dependent
11617         calls and types.
11619         * lto-streamer.c (htab_t tree_htab):
11620         Change type to hash_table.  Update dependent calls and types.
11622         * omp-low.c: Include gimplify-ctx.h.
11624         * passes.c (htab_t name_to_pass_map):
11625         Change type to hash_table.  Update dependent calls and types.
11626         (pass_traverse): Rename to passes_pass_traverse.
11628         * plugin.c (htab_t event_tab):
11629         Change type to hash_table.  Update dependent calls and types.
11631         * postreload-gcse.c (htab_t expr_table):
11632         Change type to hash_table.  Update dependent calls and types.
11633         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11635         * sese.c (debug_rename_map_1): Make extern.
11636         (htab_t copy_bb_and_scalar_dependences::rename_map):
11637         Change type to hash_table.  Update dependent calls and types.
11639         * sese.h (extern debug_rename_map): Move to .c file.
11641         * store-motion.c (htab_t store_motion_mems_table):
11642         Change type to hash_table.  Update dependent calls and types.
11644         * trans-mem.c (htab_t tm_new_mem_hash):
11645         Change type to hash_table.  Update dependent calls and types.
11647         * tree-browser.c (htab_t TB_up_ht):
11648         Change type to hash_table.  Update dependent calls and types.
11650         * tree-cfg.c (htab_t discriminator_per_locus):
11651         Change type to hash_table.  Update dependent calls and types.
11653         * tree-complex.c: Include tree-hasher.h
11654         (htab_t complex_variable_components):
11655         Change type to hash_table.  Update dependent calls and types.
11657         * tree-eh.c (htab_t finally_tree):
11658         Change type to hash_table.  Update dependent calls and types.
11660         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11661         struct int_tree_hasher.
11662         (extern int_tree_map_eq): Likewise.
11663         (uid_decl_map_hash): Removed.
11664         (extern decl_tree_map_eq): Likewise.
11666         * tree-hasher.h: New.
11667         (struct int_tree_hasher): New.
11668         (typedef int_tree_htab_type): New.
11670         * tree-inline.c: Include gimplify-ctx.h.
11672         * tree-mudflap.c: Include gimplify-ctx.h.
11674         * tree-parloops.c: Include tree-hasher.h.
11675         (htab_t eliminate_local_variables_stmt::decl_address):
11676         Change type to hash_table.  Update dependent calls and types.
11677         (htab_t separate_decls_in_region::decl_copies): Likewise.
11679         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11680         Change type to hash_table.  Update dependent calls and types.
11682         * tree-sra.c (candidates):
11683         Change type to hash_table.  Update dependent calls and types.
11685         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11686         in tree-flow.h.
11687         (int_tree_map_hash): Likewise.
11689         * tree-ssa-dom.c (htab_t avail_exprs):
11690         Change type to hash_table.  Update dependent calls and types.
11692         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11693         Change type to hash_table.  Update dependent calls and types.
11695         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11696         Change type to hash_table.  Update dependent calls and types.
11698         * tree-ssa-phiopt.c (seen_ssa_names):
11699         Change type to hash_table.  Update dependent calls and types.
11701         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11702         Change type to hash_table.  Update dependent calls and types.
11704         * tree-ssa-uncprop.c (equiv):
11705         Change type to hash_table.  Update dependent calls and types.
11707 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11709         PR rtl-optimization/57003
11710         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11711         call note_stores with kill_clobbered_value callback again after
11712         killing regs_invalidated_by_call.
11714 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11716         * config/aarch64/aarch64-simd.md
11717         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11718         (aarch64_simd_bsl<mode>): Likewise.
11719         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11721 2013-04-25  Marek Polacek  <polacek@redhat.com>
11723         PR tree-optimization/57066
11724         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11726 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11728         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11730 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11732         * config/aarch64/aarch64-builtins.c
11733         (aarch64_fold_builtin): New.
11734         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11735         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11736         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11737         * config/aarch64/arm_neon.h
11738         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11740 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11741             Tejas Belagod  <tejas.belagod@arm.com>
11743         * config/aarch64/aarch64-builtins.c
11744         (aarch64_gimple_fold_builtin): New.
11745         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11746         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11747         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11748         (addvv4sf): Update.
11749         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11751 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11753         * config/aarch64/aarch64.md
11754         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11756 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11758         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11759         (*ngcsi_uxtw): New pattern.
11761 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11762             Julian Brown  <julian@codesourcery.com>
11764         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11765         (TB_DREG): Add T_V4HF.
11766         (v4hf_UP): New macro.
11767         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11768         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11769         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11770         built-ins.
11771         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11772         (arm_vector_mode_supported_p): Handle V4HF.
11773         (arm_mangle_map): Handle V4HFmode.
11774         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11775         * config/arm/arm_neon_builtins.def: Add entries for
11776         vcvtv4hfv4sf, vcvtv4sfv4hf.
11777         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11778         (neon_vcvtv4hfv4sf): Likewise.
11779         * config/arm/neon-gen.ml: Handle half-precision floating point
11780         features.
11781         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11782         * config/arm/arm_neon.h: Regenerate.
11783         * config/arm/neon.ml (type elts): Add F16.
11784         (type vectype): Add T_float16x4, T_floatHF.
11785         (type vecmode): Add V4HF.
11786         (type features): Add Requires_FP_bit feature.
11787         (elt_width): Handle F16.
11788         (elt_class): Likewise.
11789         (elt_of_class_width): Likewise.
11790         (mode_of_elt): Refactor.
11791         (type_for_elt): Handle F16, fix error messages.
11792         (vectype_size): Handle T_float16x4.
11793         (vcvt_sh): New function.
11794         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11795         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11796         (string_of_mode): Handle V4HF.
11797         * doc/arm-neon-intrinsics.texi: Regenerate.
11799 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11801         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11802         format specifier in 'X' case.
11804 2013-04-25  Alan Modra  <amodra@gmail.com>
11806         PR target/57052
11807         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11808         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11809         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11810         Repeat for many other rotate/shift and mask patterns using subregs.
11811         Name lshiftrt insns.
11812         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11813         on WORDS_BIG_ENDIAN.
11815 2013-04-25  Alan Modra  <amodra@gmail.com>
11817         * config.gcc: Support little-endian powerpc-linux targets.
11818         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11819         (LINK_OS_LINUX_SPEC): Define.
11820         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11821         Preserve MASK_LITTLE_ENDIAN.
11822         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11823         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11824         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11825         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11826         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11827         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11828         Correct fp word order for little-endian.  Don't shift toc entries
11829         smaller than a word for little-endian.
11830         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11831         (bswapdi2 splits): Correct low-part subreg for little-endian.
11832         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11833         low/high where such is correct only for be.
11834         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11835         little-endian for -mcall-aixdesc.
11837 2013-04-25  Alan Modra  <amodra@gmail.com>
11839         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11840         replace_equiv_address_nv.
11842 2013-04-25  Alan Modra  <amodra@gmail.com>
11844         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11846 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11848         Revert:
11849         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11850         * rtl.h (struct rtx_def): ...
11852 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11854         PR rtl-optimizations/57046
11855         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11856         for multi-reg splits.
11858 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11860         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11862 2013-04-24  Sterling Augustine  <saugustine@google.com>
11864         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11865         (comp_dir_string, debug_str_dwo_section): New.
11866         (DEBUG_STR_DWO_SECTION): Rename to ...
11867         (DEBUG_DWO_STR_SECTION): ... this.
11868         (DEBUG_NORM_STR_SECTION): Delete.
11869         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11870         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11871         (find_AT_string): Move most logic to ...
11872         (find_AT_string_in_table): ... here.  New.
11873         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11874         add_skeleton_AT_string.  Delete logic.
11875         (output_skeleton_debug_sections): Remove call to
11876         add_top_level_skeleton_die_attrs.
11877         (add_comp_dir_attribute): Move logic to comp_dir_string.
11878         (dwarf2out_init): Initialize debug_str_dwo_section.
11879         (output_indirect_string): Call find_string_form.
11880         (output_indirect_strings): Rewrite.
11881         (prune_unused_types): Empty skeleton_debug_str_hash.
11882         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11883         (dwarf2out_finish):  Call output_indirect_strings.
11885 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11887         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11889 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11891         * rtl.h (struct rtx_def): Add comment for field jump.
11892         (LRA_SUBREG_P): New macro.
11893         * recog.c (register_operand): Check LRA_SUBREG_P.
11894         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11895         stack frame.
11896         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11897         (lra_final_code_change): Skip subreg change for operators.
11898         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11899         if there are no operand changes.
11900         * lra-constraints.c (curr_insn_set): New.
11901         (match_reload): Set LRA_SUBREG_P.
11902         (emit_spill_move): Ditto.
11903         (check_and_process_move): Use curr_insn_set. Process only single
11904         set insns.  Don't initialize sec_mem_p and change_p.
11905         (simplify_operand_subreg): Use LRA_SUBREG_P.
11906         (reg_in_class_p): New function.
11907         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11908         of #ifdef.  Add code to remove cycling.
11909         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11910         non-null disp.  Reload inner instead of disp when base and index
11911         are null.  Try to put lo_sum into register.
11912         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11913         (check_and_process_move): Move code for move cost check to
11914         simple_move_p.  Remove equiv_substitution.
11915         (simple_move_p): New function.
11916         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11917         curr_insn_set.  Call check_and_process_move only for single set
11918         insns.  Use the new function.  Move call of check_and_process_move
11919         after operand equiv substitution and address process.
11921 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11923         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11924         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11925         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11927 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11929         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11931 2013-04-24  Marek Polacek  <polacek@redhat.com>
11933         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11934         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11935         (select_loops_exit_conditions): Likewise.
11936         (number_of_iterations_for_all_loops): Likewise.
11937         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11938         (scev_analysis): Likewise.
11940 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11941             Chao-ying Fu  <fu@mips.com>
11943         * config/mips/micromips.md (jraddiusp): New pattern.
11944         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11945         instruction if possible.
11947 2013-04-24  Alan Modra  <amodra@gmail.com>
11949         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11951 2013-04-24  Julian Brown  <julian@codesourcery.com>
11952             Chung-Lin Tang  <cltang@codesourcery.com>
11954         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11955         dependency behavior in enumeration type DIE generation. Add TODO note
11956         to comments about future DW_FORM_sdata/udata re-work of related code.
11958 2013-04-23  Lawrence Crowl  <crowl@google.com>
11960         * Makefile.in: Update as needed below.
11962         * hash-table.h (class hash_table):
11963         Correct many methods with parameter types compare_type to the correct
11964         value_type.  (Correct code was unlikely to notice the change.)
11965         (hash_table::elements_with_deleted) New.
11966         (class hashtable::iterator): New.
11967         (hashtable::begin()): New.
11968         (hashtable::end()): New.
11969         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11971         * statistics.c (statistics_hashes):
11972         Change type to hash_table.  Update dependent calls and types.
11974         * tree-into-ssa.c (var_infos):
11975         Change type to hash_table.  Update dependent calls and types.
11977         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11978         Change type to hash_table.  Update dependent calls and types.
11980         * tree-ssa-loop-im.c (struct mem_ref.refs):
11981         Change type to hash_table.  Update dependent calls and types.
11983         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11984         Change type to hash_table.  Update dependent calls and types.
11986         * tree-ssa-sccvn.c (vn_tables_s::nary):
11987         Change type to hash_table.  Update dependent calls and types.
11988         (vn_tables_s::phis): Likewise.
11989         (vn_tables_s::references): Likewise.
11991         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11992         (vn_reference_eq): Update parameter and return types.
11994         * tree-ssa-structalias.c (pointer_equiv_class_table):
11995         Change type to hash_table.  Update dependent calls and types.
11996         (location_equiv_class_table): Likewise.
11998         * tree-vect-data-refs.c: Consequential changes for making
11999         peeling a hash_table.
12001         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12002         (destroy_loop_vec_info): Dependent hash_table update.
12004         * tree-vectorizer.h (peeling_htab):
12005         Change type to hash_table.  Update dependent calls and types.
12007 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
12009         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12010         to check the register content is equal or not.
12011         * lra-constraints.c (match_reload): Use lra_assign_reg_val
12012         to assign register content record.
12013         * lra-eliminations.c (update_reg_eliminate): Use
12014         lra_update_reg_val_offset to update register content offset.
12015         * lra-int.h (struct lra_reg): Add offset member.
12016         (lra_reg_val_equal_p): New static inline function.
12017         (lra_update_reg_val_offset): New static inline function.
12018         (lra_assign_reg_val): New static inline function.
12019         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12020         to assign register content record.
12021         (initialize_lra_reg_info_element): Initial offset to zero.
12023 2013-04-23  Catherine Moore  <clm@codesourcery.com>
12025         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12026         operands.  Record compression.
12028 2013-04-23  Xinliang David Li  <davidxl@google.com>
12030         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12032 2013-04-23  Richard Biener  <rguenther@suse.de>
12034         PR middle-end/57036
12035         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12036         parameter, only add abnormal goto edges from the copied body
12037         if the call could perform abnormal gotos.
12038         (copy_cfg_body): Adjust.
12040 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
12042         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12044 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
12046         * coretypes.h (gimple_stmt_iterator): Add struct to make
12047         compatible with C.
12049 2013-04-23  Richard Biener  <rguenther@suse.de>
12051         PR tree-optimization/57026
12052         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12053         from SSA names occuring in abnormal PHI nodes.
12055 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12057         * lto/lto.c (print_lto_report_1): Fix LTO report names.
12059 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12061         * lto/lto.c (print_lto_report_1): Declare early.
12062         (read_cgraph_and_symbols): Call print_lto_report_1 early.
12064 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12066         * common.opt (-flto-report-wpa): Add.
12067         * doc/invoke.texi (-flto-report-wpa): Add.
12068         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12069         (lto_main): dito.
12071 2013-04-22  Xinliang David Li  <davidxl@google.com>
12073         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12074         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12075         * Makefile.in: New dependency
12077             David Daney <ddaney.cavm@gmail.com>
12079         * configure.ac (gcc_cv_as_micromips_support):  Use the
12080         --fatal-warnings option.
12081         * configure: Regenerate.
12083 2013-04-22  Marek Polacek  <polacek@redhat.com>
12085         PR sanitizer/56990
12086         * tsan.c (instrument_expr): Don't instrument expression
12087         in case its size is zero.
12089 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
12091         PR target/57032
12092         Revert:
12093         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
12095         * config/alpha/alpha.c (TARGET_LRA_P): New define.
12097 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12099         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12100         (gimple_stmt_iterator): New typedef.
12101         * gimple.h (gimple_stmt_iterator): Rename to...
12102         (gimple_stmt_iterator_d): ... This.
12103         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12104         trees be valid for GIMPLE and GENERIC.
12105         (TARGET_GIMPLE_FOLD_BUILTIN): New.
12106         * gimple-fold.c (gimple_fold_call): Call target hook
12107         gimple_fold_builtin.
12108         * hooks.c (hook_bool_gsiptr_false): New.
12109         * hooks.h (hook_bool_gsiptr_false): New.
12110         * target.def (fold_stmt): New.
12111         * doc/tm.texi: Regenerate.
12113 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
12115         PR target/57018
12116         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12117         a set sp if no stack realignment.
12119 2013-04-22  Nick Clifton  <nickc@redhat.com>
12121         * config.gcc (tilegx-linux): Extend extra_objs rather than
12122         overwriting it.
12123         (tilepro-linux): Likewise.
12125 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12127         * config/aarch64/aarch64-builtins.c
12128         (CF): Remove.
12129         (CF0, CF1, CF2, CF3, CF4, CF10): New.
12130         (VAR<1-12>): Add MAP parameter.
12131         (BUILTIN_*): Likewise.
12132         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12133         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12134         (aarch64_ushl_n<mode>): Likewise.
12135         (aarch64_sshr_n<mode>): Likewise.
12136         (aarch64_ushr_n<mode>): Likewise.
12137         (aarch64_<maxmin><mode>): Likewise.
12138         (aarch64_sqrt<mode>): Likewise.
12139         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12140         (vshr<q>_n_*): Likewise.
12142 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12144         * config/aarch64/aarch64-builtins.c
12145         (aarch64_simd_builtin_type_mode): Handle SF types.
12146         (sf_UP): Define.
12147         (BUILTIN_GPF): Define.
12148         (aarch64_init_simd_builtins): Handle SF types.
12149         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12150         (frecps): Likewise.
12151         (frecpx): Likewise.
12152         * config/aarch64/aarch64-simd.md
12153         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12154         (aarch64_frecpe<mode>): New.
12155         (aarch64_frecps<mode>): Likewise.
12156         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12157         (v8type): Add frecp<esx>.
12158         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12159         (aarch64_frecps<mode>): Likewise.
12160         * config/aarch64/iterators.md (FRECP): New.
12161         (frecp_suffix): Likewise.
12162         * config/aarch64/arm_neon.h
12163         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12165 2013-04-22  Christian Bruel  <christian.bruel@st.com>
12167         PR target/56995
12168         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12169         (REG_CLASS_NAMES): Idem.
12170         (REG_CLASS_CONTENTS): Idem.
12171         (REGCLASS_HAS_FP_REG): Idem.
12172         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12173         (sh_conditional_register_usage): Idem.
12175 2013-04-21  Jeff Law  <law@redhat.com>
12177         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12178         (ssa_forward_propagate_and_combine): Use it.
12180 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12182         * lra.c: Update the flow chart diagram.
12184 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12186         PR rtl-optimization/56847
12187         * lra-constraints.c (process_alt_operands): Discourage alternative
12188         with non-matche doffsettable memory constraint fro memory with
12189         known offset.
12191 2013-04-19  Richard Biener  <rguenther@suse.de>
12193         PR tree-optimization/56982
12194         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12195         function.
12196         * gimplify.c (gimplify_call_expr): Notice special calls.
12197         (gimplify_modify_expr): Likewise.
12198         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12199         abnormal control flow receivers.
12200         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12201         in the same way as cfun->has_nonlocal_labels.
12202         (gimple_purge_dead_abnormal_call_edges): Likewise.
12203         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12204         receivers start a basic-block.
12206 2013-04-19  Richard Biener  <rguenther@suse.de>
12208         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12209         member ...
12210         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
12211         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12212         (SLP_TREE_LOAD_PERMUTATION): Add.
12213         (vect_transform_slp_perm_load): Adjust prototype.
12214         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12215         (vect_free_slp_instance): Likewise.
12216         (vect_create_new_slp_node): Likewise.
12217         (vect_supported_slp_permutation_p): Remove.
12218         (vect_slp_rearrange_stmts): Adjust.
12219         (vect_supported_load_permutation_p): Likewise.  Inline
12220         vect_supported_slp_permutation_p here.
12221         (vect_analyze_slp_instance): Compute load permutations per
12222         slp node instead of per instance.
12223         (vect_get_slp_defs): Adjust.
12224         (vect_transform_slp_perm_load): Likewise.
12225         (vect_schedule_slp_instance): Remove redundant code.
12226         (vect_schedule_slp): Remove hack for PR56270, add it ...
12227         * tree-vect-stmts.c (vectorizable_load): ... here, do not
12228         CSE loads for SLP.  Adjust.
12230 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12232         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12233         spelling in two comments.
12235 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12237         PR target/56797
12238         * config/arm/arm.c (load_multiple_sequence): Require SP
12239         as base register for loads if SP is in the register list.
12241 2013-04-19  Martin Jambor  <mjambor@suse.cz>
12243         PR tree-optimization/56718
12244         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12245         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12246         and made public.  Adjusted all callers.
12247         (ipa_intraprocedural_devirtualization): New function.
12248         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12249         (ipa_intraprocedural_devirtualization): Likewise.
12250         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12252 2013-04-19  Richard Biener  <rguenther@suse.de>
12254         PR tree-optimization/57000
12255         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12257 2013-04-19  Terry Guo  <terry.guo@arm.com>
12259         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12260         Replace with ...
12261         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
12262         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12263         (cortex_m4_fmacs): Use new reservations.
12264         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12266 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
12268         PR rtl-optimization/56999
12269         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12270         related code.
12271         (lra_coalesce): Remove split_origin_bitmap and related code.
12272         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
12273         ranges if necessary.
12275 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
12277         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12278         New array.
12279         (ix86_expand_call): Remove clobbered_registers array and use
12280         x86_64_ms_sysv_extra_clobbered_registers instead.
12281         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12282         Declare here.
12283         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12284         predicate.
12285         * config/i386/i386.md (*call_rex64_ms_sysv): Use
12286         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
12287         (*call_value_rex64_ms_sysv): Ditto.
12289 2013-04-18  Cary Coutant  <ccoutant@google.com>
12291         * dwarf2out.c (output_pubnames): Check die_perennial_p of
12292         parent instead of die_mark.
12294 2013-04-18  Diego Novillo  <dnovillo@google.com>
12296         * gimple.c (create_gimple_tmp): New.
12297         (get_expr_type): New.
12298         (build_assign): New.
12299         (build_type_cast): New.
12300         * gimple.h (enum ssa_mode): Define.
12301         (gimple_seq_set_location): New.
12302         * asan.c (build_check_stmt): Change some gimple_build_* calls
12303         to use build_assign and build_type_cast.
12305 2013-04-18  Richard Biener  <rguenther@suse.de>
12307         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12308         handle negative step.  Remove redundant checks.
12309         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12310         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12311         for negative step and grouped loads fail to vectorize.
12313 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12315         * emit-rtl.c (reset_insn_used_flags): New function.
12316         (reset_all_used_flags): Use it.
12317         (verify_insn_sharing): New function.
12318         (verify_rtl_sharing): Fix verification for SEQUENCEs.
12320 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
12322         PR tree-optimization/56984
12323         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12324         and (x >> M) >= N don't register any assertion if N << M is the
12325         minimum value.
12327 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12329         * lower-subreg.c (resolve_simple_move): If called self-recursive,
12330         do not delete_insn insns that have not yet been emitted, only
12331         unlink them with remove_insn.
12332         * df-scan.c (df_insn_delete): Revert r197492.
12334 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
12336         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12337         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12339 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12341         * config/arm/arm.md (movsicc_insn): Convert define_insn into
12342         define_insn_and_split.
12343         (and_scc,ior_scc,negscc): Likewise.
12344         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12346 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12348         * config/arm/arm.c (use_return_insn): Return 0 for targets that
12349         can benefit from using a sequence of LDRD instructions in epilogue
12350         instead of a single LDM instruction.
12352 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12354         PR 45688
12355         * doc/extend.texi: Fix typo.
12357 2013-04-17  Richard Biener  <rguenther@suse.de>
12359         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12360         (vect_build_slp_tree): ... here.
12361         (vect_build_slp_tree_1): Compute which stmts of the SLP group
12362         match.  Remove special-casing of mismatched complex loads.
12363         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12364         re-try the match with swapped commutative operands.
12365         (vect_supported_load_permutation_p): Remove special-casing of
12366         mismatched complex loads.
12367         (vect_analyze_slp_instance): Adjust.
12369 2013-04-17  Richard Biener  <rguenther@suse.de>
12371         PR rtl-optimization/56921
12372         * cfgloop.h (struct loop): Add simple_loop_desc member.
12373         (struct niter_desc): Mark with GTY(()).
12374         (simple_loop_desc): Do not use aux field but simple_loop_desc.
12375         * loop-iv.c (get_simple_loop_desc): Likewise.
12376         (free_simple_loop_desc): Likewise.
12378         Revert
12379         2013-04-16  Richard Biener  <rguenther@suse.de>
12381         PR rtl-optimization/56921
12382         * loop-init.c (pass_rtl_move_loop_invariants): Add
12383         TODO_do_not_ggc_collect to todo_flags_finish.
12384         (pass_rtl_unswitch): Same.
12385         (pass_rtl_unroll_and_peel_loops): Same.
12386         (pass_rtl_doloop): Same.
12388 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
12390         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12391         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12392         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12393         references.
12394         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12395         * tree-streamer.c (record_common_node): Adjust reference in comment.
12397 2013-04-17  Terry Guo  <terry.guo@arm.com>
12399         * config/arm/cortex-m4.md: Add a new bypass.
12401 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12403         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12404         New pattern.
12405         (*subs_<optab><mode>_multp2): New pattern.
12406         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12407         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12409 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12411         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12412         (*subs_mul_imm_<mode>): New pattern.
12414 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
12416         PR target/56948
12417         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12418         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
12419         (vsx_movti_32bit): Change j->wa to O->wa.
12421 2013-04-16  Richard Biener  <rguenther@suse.de>
12423         PR rtl-optimization/56921
12424         * loop-init.c (pass_rtl_move_loop_invariants): Add
12425         TODO_do_not_ggc_collect to todo_flags_finish.
12426         (pass_rtl_unswitch): Same.
12427         (pass_rtl_unroll_and_peel_loops): Same.
12428         (pass_rtl_doloop): Same.
12430 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12432         * config/arm/arm.c (emit_multi_reg_push): New declaration
12433         for an existing function.
12434         (arm_emit_strd_push): New function.
12435         (arm_expand_prologue): Used here.
12436         (arm_emit_ldrd_pop): New function.
12437         (arm_expand_epilogue): Used here.
12438         (arm_get_frame_offsets): Update condition.
12439         (arm_emit_multi_reg_pop): Add a special case for load of a single
12440         register with writeback.
12442 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
12444         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12445         description.
12447 2013-04-16  Richard Biener  <rguenther@suse.de>
12449         PR tree-optimization/56756
12450         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12451         (first_mem_ref_loc): New.
12452         (execute_sm): Place the load temporarily before a previous
12453         access instead of in the latch edge to ensure its SSA dependencies
12454         are defined at points dominating the load.
12456 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
12458         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12459         correct fix by moving header and footer insn to the footer of
12460         the merged basic block.  Clear BB_END of the merged-away block.
12462         PR middle-end/43631
12463         * emit-rtl.c (make_note_raw): New function.
12464         (link_insn_into_chain): New static inline function.
12465         (add_insn): Use it.
12466         (add_insn_before, add_insn_after): Factor insn chain linking code...
12467         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12468         using link_insn_into_chain.
12469         (note_outside_basic_block_p): New helper function for emit_note_after
12470         and emit_note_before.
12471         (emit_note_after): Use nobb variant of add_insn_after if the note
12472         should not be contained in a basic block.
12473         (emit_note_before): Use nobb variant of add_insn_before if the note
12474         should not be contained in a basic block.
12475         (emit_note_copy): Use make_note_raw.
12476         (emit_note): Likewise.
12477         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12478         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12479         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12480         the moved barrier the tail of the basic block it follows.
12481         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12483 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
12485         PR tree-optimization/56962
12486         * gimple-ssa-strength-reduction.c (record_increment): Only set
12487         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12488         either rhs1 or rhs2 is equal to c->base_expr.
12490 2013-04-15  Richard Biener  <rguenther@suse.de>
12492         PR tree-optimization/56933
12493         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12494         member.
12495         (GROUP_READ_WRITE_DEPENDENCE): Remove.
12496         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12497         * tree-vect-data-refs.c (vect_analyze_group_access): Move
12498         dependence check ...
12499         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12500         ... here.
12501         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12502         GROUP_READ_WRITE_DEPENDENCE.
12504 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12506         * emit-rtl.c (reset_all_used_flags): New function.
12507         (verify_rtl_sharing): Call reset_all_used_flags before and after
12508         performing the checks.
12510 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12512         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12513         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12514         * config/arm/constraints.md (De): New constraint.
12515         * config/arm/neon.md (anddi3_neon): Delete.
12516         (neon_vand<mode>): Expand to standard anddi3 pattern.
12517         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12518         Move earlier in the file.
12519         (neon_inv_logic_op2): Likewise.
12520         (arm_anddi_operand_neon): New predicate.
12522 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12524         * configure.ac (gcc_cv_ld_as_needed): Set
12525         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12526         Use -z ignore, -z record on *-*-solaris2*.
12527         (HAVE_LD_AS_NEEDED): Update comment.
12528         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12529         * configure: Regenerate.
12530         * config.in: Regenerate.
12531         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12532         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12533         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12534         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12535         equivalents.  Fix markup.
12536         * doc/tm.texi: Regenerate.
12538 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
12540         * config/i386/i386.opt: New option mstack-protector-guard=.
12541         * config/i386/i386-opts.h: Add enum stack_protector_guard.
12542         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12543         TARGET_SSP_TLS_GUARD.
12544         * config/i386/i386.c (ix86_option_override_internal): Set
12545         ix86_stack_protector_guard.
12546         * config/i386/i386.md (stack_protect_set): Enable for
12547         TARGET_SSP_TLS_GUARD only.
12548         (stack_protect_set_<mode>): Ditto.
12549         (stack_protect_test): Ditto.
12550         (stack_protect_test_<mode>): Ditto.
12551         * doc/invoke.texi (i386 Option): Document.
12553 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
12555         PR target/56890
12556         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12557         (S_MODES): Set H_MODE bit.
12558         (SF_MODES): Set only S_MODE and SF_MODE bits.
12559         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12560         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12561         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12562         <MODE_FLOAT>: Likewise.
12564 2013-04-15  Joey Ye  <joey.ye@arm.com>
12566         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12568 2013-04-15  Joey Ye  <joey.ye@arm.com>
12570         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12571         for real far jump.
12572         (thumb_far_jump_used_p): Count instruction size and set
12573         far_jump_used.
12575 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
12577         * reorg.c (fill_simple_delay_slots): Reindent block of code.
12578         * resource.c (mark_target_live_regs): Reformat conditional block.
12580 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12582         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12583         notes, they are emitted only just before final.
12584         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12586 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12588         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12589         * cfgrtl.c (delete_insn): Call it here instead.
12590         * lra-spills.c (lra_final_code_change): Use delete_insn.
12591         * haifa-sched.c (sched_remove_insn): Likewise.
12592         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12593         returning to the nop pool.
12594         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12595         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
12597 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12599         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12600         * doc/tm.texi: Regenerated.
12602 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
12604         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12605         QImode checks.
12607 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12609         * df-core.c (df_find_def): Compare register numbers.
12610         (df_find_use): Likewise.
12612 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
12614         PR target/56903
12615         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12616         lra_in_progress for return.
12618 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12620         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12621         define_insn into define_insn_and_split and emit movsicc patterns.
12623 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12625         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12627 2013-04-12  Richard Biener  <rguenther@suse.de>
12629         * tree-pass.h (TODO_do_not_ggc_collect): New.
12630         * passes.c (execute_one_ipa_transform_pass): Honor
12631         TODO_do_not_ggc_collect.
12632         (execute_one_pass): Likewise.
12634         Revert
12635         2013-04-10  Richard Biener  <rguenther@suse.de>
12637         * passes.c (init_optimization_passes): Remove reload pass.
12638         * ira.c (do_reload): Merge into ...
12639         (ira): ... this.
12640         (rest_of_handle_reload): Remove.
12641         (pass_reload): Likewise.
12642         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12643         of reload for vzeroupper pass placement.
12645 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12647         PR tree-optimization/56918
12648         PR tree-optimization/56920
12649         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12650         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12651         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12652         use wide_mul_with_sign method.
12654 2013-04-12  Richard Biener  <rguenther@suse.de>
12656         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12657         not be considered a gimple constant.
12659 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12661         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12662         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12663         operations.
12665 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12666             Jakub Jelinek  <jakub@redhat.com>
12668         * opts.c: Include diagnostic-color.h.
12669         (common_handle_option): Handle OPT_fdiagnostics_color_.
12670         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12671         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12672         (diagnostic-color.o): New.
12673         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12674         (diagnostic_color_rule): New enum.
12675         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12676         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12677         the location string.
12678         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12679         either NULL, or color kind.
12680         * diagnostic-color.c: New file.
12681         * diagnostic-color.h: New file.
12682         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12683         arguments.
12684         * doc/invoke.texi (-fdiagnostics-color): Document.
12685         * pretty-print.h (pp_show_color): Define.
12686         (struct pretty_print_info): Add show_color field.
12687         * diagnostic.c: Include diagnostic-color.h.
12688         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12689         macros.  Colorize error:, warning: etc. strings and also the location
12690         string.
12691         (diagnostic_show_locus): Colorize the caret line.
12692         * pretty-print.c: Include diagnostic-color.h.
12693         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12694         inside of %< %> quotes or quoted through q format modifier.
12696 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12698         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12700 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12702         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12703         code in CC_NZ mode.
12704         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12705         pattern.
12707 2013-04-11  Marek Polacek  <polacek@redhat.com>
12709         PR tree-optimization/48184
12710         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12712 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12714         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12715         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12716         (skip_simple_arithmetic): Tidy up.
12717         * tree.h (skip_simple_constant_arithmetic): Declare.
12719 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12721         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12723 2013-04-11  Richard Biener  <rguenther@suse.de>
12725         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12726         generate vector constants.
12728 2013-04-11  Richard Biener  <rguenther@suse.de>
12730         PR tree-optimization/56878
12731         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12732         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12733         New function.
12734         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12735         Prefer to align the DR with the most invariant base address.
12737 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12739         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12740         comment.
12742 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12744         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12745         floating-point vector comparisons against 0.
12747 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12749         PR tree-optimization/56899
12750         * fold-const.c (extract_muldiv_1): Apply distributive law
12751         only if TYPE_OVERFLOW_WRAPS (ctype).
12753 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12755         PR target/56124
12756         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12757         loading has side effect.
12759 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12761         * config/sparc/sparc.c: Include tree-pass.h.
12762         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12763         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12764         head of file.  Change return type.  Split off gate function.
12765         (sparc_gate_work_around_errata): New function.
12766         (pass_work_around_errata): New pass definition.
12767         (insert_pass_work_around_errata) New pass insert definition to
12768         insert pass_work_around_errata just after delayed-branch scheduling.
12769         (sparc_option_override): Insert the pass.
12770         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12772 2013-04-10  David S. Miller  <davem@davemloft.net>
12774         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12775         or -mcpu=hypersparc.
12777         * target.def (cstore_mode): New hook.
12778         * target.h: Include insn-codes.h
12779         * targhooks.c: Likewise.
12780         (default_cstore_mode): New function.
12781         * targhooks.h: Declare it.
12782         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12783         * doc/tm.texi: Rebuild.
12784         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12785         target hook, rather than inspecting the insn_data.
12786         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12787         (TARGET_CSTORE_MODE): Redefine.
12788         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12789         result patterns.
12790         * config/sparc/predicates.md (cstore_result_operand): New special
12791         predicate.
12792         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12793         Use it for operand 0.
12794         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12795         (*snesi_special): Likewise.
12796         (*snesi_zero): Likewise.
12797         (*seqsi_zero): Likewise.
12798         (*sltu_insn): Likewise.
12799         (*sgeu_insn): Likewise.
12800         (*seqdi_special): Make operand 0 and comparison operation be of
12801         DImode.
12802         (*snedi_special): Likewise.
12803         (*snedi_special_vis3): Likewise.
12804         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12805         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12806         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12807         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12808         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12809         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12810         (*sltu_extend_sp64): Likewise.
12811         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12812         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12813         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12814         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12815         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12817 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12819         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12820         (aarch64_start_file): Use the new function.
12822 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12823             Jason Merrill  <jason@redhat.com>
12825         * common.opt: Add -gdwarf.
12826         * opts.c (common_handle_option): Handle it.
12827         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12829 2013-04-10  Richard Biener  <rguenther@suse.de>
12831         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12832         (execute_one_ipa_transform_pass): But unconditionally here.
12833         (execute_one_pass): And here.
12834         (init_optimization_passes): Remove reload pass.
12835         * tree-pass.h (TODO_ggc_collect): Remove.
12836         (pass_reload): Likewise.
12837         * ira.c (do_reload): Merge into ...
12838         (ira): ... this.
12839         (rest_of_handle_reload): Remove.
12840         (pass_reload): Likewise.
12841         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12842         of reload for vzeroupper pass placement.
12843         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12844         and todo_flags_finish of all passes.
12846 2013-04-10  Richard Biener  <rguenther@suse.de>
12848         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12849         first_const_oprnd field, rename first_def_type to first_op_type.
12850         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12851         (vect_get_and_check_slp_defs): Always use the type of the
12852         operand.  Allow mixed vect_external_def, vect_constant_def types.
12853         (vect_get_constant_vectors): Handle mixed vect_external_def,
12854         vect_constant_def types.
12856 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
12858         PR tree-optimization/55524
12859         * tree-ssa-math-opts.c
12860         (convert_mult_to_fma): Don't use an fms construct
12861         when we don't have an fms operation, but fnma, and it looks
12862         likely that we'll be able to use the latter.
12864 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
12866         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12867         function.
12868         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12869         inline fail caused by overwritable functions.
12871 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12873         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12874         unnecessary bits in the constant power of two case.
12876 2013-04-10  Richard Biener  <rguenther@suse.de>
12878         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12879         broken code swapping operands.
12880         (vect_build_slp_tree): Do not compute load permutations here.
12881         (vect_analyze_slp_instance): Compute load permutations here,
12882         after building the SLP tree.
12884 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12886         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12887         of next/prev_real_insn.
12889 2013-04-09  Jan Hubicka  <jh@suse.cz>
12891         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12892         Drop aliased parameter.
12893         (function_and_variable_visibility): Do not handle alias pairs.
12894         * cgraph.c (varpool_externally_visible_p): Update prototype.
12895         * varpool.c (varpool_add_new_variable): Update.
12897 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12899         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12901 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12903         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12905         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12907 2013-04-09  Marek Polacek  <polacek@redhat.com>
12909         PR tree-optimization/48762
12910         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12912 2013-04-09  Richard Biener  <rguenther@suse.de>
12914         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12915         dealing with cost.
12916         (vect_build_slp_tree): Likewise.
12917         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12918         calculating the cost of a SLP instance.
12919         (vect_analyze_slp_instance): Use it from here, after building
12920         the SLP tree.
12922 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12924         PR middle-end/56883
12925         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12926         expand_omp_for_static_chunk): Use simple_p = true in
12927         force_gimple_operand_gsi calls when assigning to addressable decls.
12929 2013-04-09  Jeff Law  <law@redhat.com>
12931         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12932         when the boolean was created by converting a wider object which
12933         had a boolean range.
12935 2013-04-09  Richard Biener  <rguenther@suse.de>
12937         * tree-vectorizer.h (slp_void_p): Remove.
12938         (slp_tree): Typedef before _slp_tree declaration.
12939         (struct _slp_tree): Use a vector of slp_tree as children.
12940         (vect_get_place_in_interleaving_chain): Remove.
12941         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12942         Move ...
12943         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12944         and make static.
12945         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12946         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12947         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12948         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12949         Use slp_node instead of slp_void_p and adjust.
12951 2013-04-09  Richard Biener  <rguenther@suse.de>
12953         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12954         work that is not necessary.
12956 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12958         PR tree-optimization/56854
12959         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12960         forward into clobber stmts if it would change MEM_REF lhs into
12961         non-MEM_REF.
12963 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12965         * tree.c (type_hash_lookup, type_hash_add): Make static.
12966         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12968 2013-04-09  Richard Biener  <rguenther@suse.de>
12970         * tree.h (unsave_expr_now): Remove.
12971         * tree-inline.c (mark_local_for_remap_r): Remove.
12972         (unsave_expr_1): Likewise.
12973         (unsave_r): Likewise.
12974         (unsave_expr_now): Likewise.
12975         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12976         (propagate_tree_value): Likewise.
12978 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12980         * doc/rtl.texi (sequence): Rewrite documentation to match the
12981         current use of SEQUENCE rtl objects.
12982         * rtl.def (SEQUENCE): Likewise.
12984         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12985         Update documentation.
12986         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12987         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12989         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12991 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12993         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12994         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12995         methods.
12996         (estimate_edge_size_and_time): Add comment to suggest using rounding
12997         methods.
12998         (estimate_node_size_and_time): Ditto.
12999         (remap_edge_change_prob): Use helper rounding divide methods.
13000         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13001         (gimple_mod_pow2_value_transform): Ditto.
13002         (gimple_mod_subtract_transform): Ditto.
13003         (gimple_ic_transform): Ditto.
13004         (gimple_stringops_transform): Ditto.
13005         * stmt.c (conditional_probability): Ditto.
13006         (emit_case_dispatch_table): Ditto.
13007         * lto-cgraph.c (merge_profile_summaries): Ditto.
13008         * tree-optimize.c (execute_fixup_cfg): Ditto.
13009         * cfgcleanup.c (try_forward_edges): Ditto.
13010         * cfgloopmanip.c (scale_loop_profile): Ditto.
13011         (loopify): Ditto.
13012         (duplicate_loop_to_header_edge): Ditto.
13013         (lv_adjust_loop_entry_edge): Ditto.
13014         * tree-vect-loop.c (vect_transform_loop): Ditto.
13015         * profile.c (compute_branch_probabilities): Ditto.
13016         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13017         * lto-streamer-in.c (input_cfg): Ditto.
13018         * gimple-streamer-in.c (input_bb): Ditto.
13019         * ipa-cp.c (update_profiling_info): Ditto.
13020         (update_specialized_profile): Ditto.
13021         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13022         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13023         rounding methods.
13024         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13025         (compute_trg_info): Ditto.
13026         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13027         (purge_dead_edges): Ditto.
13028         * loop-unswitch.c (unswitch_loop): Ditto.
13029         * cgraphclones.c (cgraph_clone_edge): Ditto.
13030         (cgraph_clone_node): Ditto.
13031         * tree-inline.c (copy_bb): Ditto.
13032         (copy_edges_for_bb): Ditto.
13033         (initialize_cfun): Ditto.
13034         (copy_cfg_body): Ditto.
13035         (expand_call_inline): Ditto.
13037 2013-04-08  Kai Tietz  <ktietz@redhat.com>
13039         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13040         TARGET_CYGWIN64 by TARGET_64BIT.
13042 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13044         * config/epiphany/epiphany.md (GPR_1): New constant.
13045         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13046         * config/epiphany/epiphany.c (gen_compare_reg):
13047         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13048         is already in place.
13049         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13050         Don't require being called during rtl expansion; If y operlaps r0,
13051         return 0.
13052         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13053         (epiphany_expand_epilogue): Likewise.
13055         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13056         Don't use CC_FPmode for ORDERED / UNORDERED.
13057         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13059         * config/epiphany/constraints.md (CnL): New constraint.
13060         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13061         * config/epiphany/predicates.md (add_operand): Allow 1024.
13063         * config/epiphany/epiphany.md (logical_op): New code iterator.
13064         (op_mnc): New code attribute.
13065         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13066         (mov_f+1, mov_f+2): New peephole2 patterns.
13068         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13069         (cstoresi4): Also allow re-use of zero result when doing a NE
13070         comparison to a non-zero operand.
13071         Use (clobber (scratch)) for first insn if the gpr output is not needed.
13073         * config/epiphany/epiphany.md (<insn_opname>v2si3):
13074         Use gen_addsi3_i / gen_subsi3_i.
13076 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13078         PR c++/34949
13079         PR c++/50243
13080         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13081         contain anything but clobbers, at most one __builtin_stack_restore,
13082         optionally debug stmts and final resx, and if it has at least one
13083         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
13084         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13085         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13086         which isn't defaut definition, remove them.
13087         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13088         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13089         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13090         with MEM_REF LHS with SSA_NAME address.
13092 2013-04-08  Jeff Law  <law@redhat.com>
13094         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13096 2013-04-08  Richard Biener  <rguenther@suse.de>
13098         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13099         extra newline.
13100         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13101         determined vector type.
13102         (vect_analyze_data_refs): Likewise.
13103         (vect_get_new_vect_var): Adjust.
13104         (vect_create_destination_var): Preserve SSA name versions.
13105         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13106         not dump anything here.
13108 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13110         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13111         Add member lr_slot_known.
13112         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13113         if necessary.
13114         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13115         Remove code that sets lr_slot_offset according to what a previous
13116         version of epiphany_emit_save_restore used to do.
13117         (epiphany_emit_save_restore): When doing an lr save or restore,
13118         set/verify lr_slot_known and lr_slot_offset.
13120 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
13122         PR target/54338
13123         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13124         in ALL_REGS.
13126 2013-04-08  Richard Biener  <rguenther@suse.de>
13128         * alias.c (find_base_term): Fix thinko in previous change.
13130 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13132         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13133         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13134         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
13135         if possible to compute val.
13136         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13137         For QImode integers don't require anything about precision.  Use
13138         const_with_all_bytes_same to find out if the constant doesn't have
13139         repeated bytes in it.
13141 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13143         * config/s390/s390.c (s390_expand_insv): Only accept insertions
13144         within mode size.
13146 2013-04-08  Marek Polacek  <polacek@redhat.com>
13148         PR rtl-optimization/48182
13149         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13150         value to 1.
13152 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13154         PR target/55487
13155         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13156         nuses, make sure we have a label.
13158 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13160         PR target/56843
13161         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13162         (rs6000_emit_swdiv_low_precision): Remove.
13163         (rs6000_emit_swdiv): Rewrite to handle between one and four
13164         iterations of Newton-Raphson generally; modify required number of
13165         iterations for some cases.
13166         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13168 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
13170         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13171         set-but-unused variable.
13173         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13174         basic blocks of released function bodies garbage-collectable.
13176         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13177         (struct rtl_opt_pass): Add TODO_df_finish.
13179         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13181 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13183         * config/arm/constraints.md (q): New constraint.
13184         * config/arm/ldrdstrd.md: New file.
13185         * config/arm/arm.md (ldrdstrd.md) New include.
13186         (arm_movdi): Use "q" instead of "r" constraint
13187         for double-word memory access.
13188         (movdf_soft_insn): Likewise.
13189         * config/arm/vfp.md (movdi_vfp): Likewise.
13190         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13191         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13192         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13193         (mem_ok_for_ldrd_strd): Likewise.
13194         (output_move_double): Update assertion.
13196 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13198         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13200 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13202         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13203         define_insn_and_split.
13204         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13206 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13208         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13209         define_insn_and_split.
13210         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13211         (shiftsi3_compare): New pattern.
13212         (rrx): New pattern.
13213         * config/arm/unspecs.md (UNSPEC_RRX): New.
13215 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13217         * config/arm/arm.md (negdi_extendsidi): New pattern.
13218         (negdi_zero_extendsidi): Likewise.
13220 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13222         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13223         define_insn_and_split.
13224         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13225         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13227 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13229         * config/arm/arm.md (arm_subdi3): Convert define_insn into
13230         define_insn_and_split.
13231         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13232         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13234 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13236         * config/arm/arm.md (subsi3_carryin): New pattern.
13237         (subsi3_carryin_const): Likewise.
13238         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13239         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13241 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13243         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13245 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13247         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13248         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13250 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13252         * config/arm/arm.c (arm_expand_builtin): Change fcode
13253         type to unsigned int.
13255 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13257         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13259 2013-04-04  Ian Lance Taylor  <iant@google.com>
13261         * doc/standards.texi (Standards): The Go frontend supports the Go 1
13262         language standard.
13264 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
13266         PR middle-end/56729
13267         * df-scan.c (df_insn_delete): Disable failing assert.
13269 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13271         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13272         New function prototype.
13273         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13274         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13275         (arm_builtin_vectorized_function): New function.
13277 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13279         * config/arm/arm_neon_builtins.def: New file.
13280         * config/arm/arm.c (neon_builtin_data): Move contents to
13281         arm_neon_builtins.def.
13282         (enum arm_builtins): Include neon builtin definitions.
13283         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13284         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13286 2013-04-04  Marek Polacek  <polacek@redhat.com>
13288         PR tree-optimization/48186
13289         * predict.c (maybe_hot_frequency_p): Return false if
13290         HOT_BB_FREQUENCY_FRACTION is 0.
13291         (cgraph_maybe_hot_edge_p): Likewise.
13293 2013-04-04  Richard Biener  <rguenther@suse.de>
13295         PR tree-optimization/56826
13296         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13297         more accurately.
13299 2013-04-04  Richard Biener  <rguenther@suse.de>
13301         PR tree-optimization/56213
13302         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13303         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13305 2013-04-04  Richard Biener  <rguenther@suse.de>
13307         PR tree-optimization/56837
13308         * tree-loop-distribution.c (classify_partition): For non-zero
13309         values require that the value has the same precision as its
13310         mode to be useful as memset value.
13312 2013-04-03  Nick Clifton  <nickc@redhat.com>
13314         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13315         (fmssf4): Use fmsf.s on E3V5 architectures.
13316         (fnmasf4): Use fnmaf.s on E3V5 architectures.
13317         (fnmssf4): Use fnmsf.s on E3V5 architectures.
13319 2013-04-03  Jeff Law  <law@redhat.com>
13321         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13322         (lra-eliminations.o): Likewise.
13324 2013-04-03  Teresa Johnson  <tejohnson@google.com>
13326         * gcov-io.c (compute_working_sets): Moved most of body of old
13327         compute_working_sets here from profile.c.
13328         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13329         (gcov_working_set_t): Moved typedef here from basic-block.h
13330         (compute_working_set): Declare.
13331         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13332         (get_working_sets): Renamed from compute_working_set,
13333         replace most of body with call to new compute_working_sets.
13334         (get_exec_counts): Replace call to compute_working_sets
13335         to get_working_sets.
13336         * profile.h (get_working_sets): Renamed from compute_working_set.
13337         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13338         to get_working_sets.
13339         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13340         * gcov-dump.c (dump_working_sets): New function.
13342 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
13344         * hwint.c (sext_hwi, zext_hwi): New functions.
13345         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13346         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13347         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13348         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13349         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13350         (sext_hwi, zext_hwi): New functions.
13352 2013-04-03  Jeff Law  <law@redhat.com>
13354         PR tree-optimization/56799
13355         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13356         back test for widening conversion erroneously dropped in prior change.
13358 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13360         PR target/56809
13361         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13362         instead of next_real_insn.
13364 2013-04-03  Marek Polacek  <polacek@redhat.com>
13366         PR sanitizer/55702
13367         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13369 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13371         PR target/56809
13372         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13373         next_real_insn.
13374         (thumb1_output_casesi): Likewise.
13375         (thumb2_output_casesi): Likewise.
13377 2013-04-03  Richard Biener  <rguenther@suse.de>
13379         PR tree-optimization/56817
13380         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13381         Split out ...
13382         (tree_unroll_loops_completely_1): ... new function to manually
13383         walk the loop tree, properly defering outer loops of unrolled
13384         loops to later iterations.
13386 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13388         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13389         (vectorizable_load): Likewise.
13390         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13391         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13393 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13395         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13396         BIT_FIELD_REF.
13398 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13400         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13402 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
13404         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13406 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13408         PR tree-optimization/56790
13409         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13410         folding.
13412 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13414         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13415         Handle VEC_MERGE.
13416         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13417         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13418         equal arguments.
13420 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
13422         PR c/19449
13423         * tree.h (force_folding_builtin_constant_p): New decl.
13424         * builtins.c (force_folding_builtin_constant_p): New variable.
13425         (fold_builtin_constant_p): Fold immediately also if
13426         force_folding_builtin_constant_p.
13428 2013-04-03  Richard Biener  <rguenther@suse.de>
13430         PR tree-optimization/56812
13431         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13432         DRs of the same interleaving chain are independent.
13434 2013-04-02  Jason Merrill  <jason@redhat.com>
13436         * gdbinit.in (pbb): Use debug fn.
13438 2013-04-02  Lawrence Crowl  <crowl@google.com>
13440         * sese.h (struct ivtype_map_elt_s): Remove unused.
13441         (extern debug_ivtype_map): Remove unused.
13442         (extern eq_ivtype_map_elts): Remove unused.
13443         * sese.c (debug_ivtype_map): Removed unused.
13444         (debug_ivtype_map_1): Removed unused.
13445         (debug_ivtype_elt): Remove unused.
13446         (eq_ivtype_map_elts): Remove unused.
13449 2013-04-02  Kai Tietz  <ktietz@redhat.com>
13451         PR target/52790
13452         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13453         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13454         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13455         function.
13456         (legitimize_pe_coff_symbol): Likewise.
13457         (is_imported_p): New helper-function.
13458         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13459         for Windows x64 targets.
13460         (ix86_expand_prologue): Optimize for pe-coff targets.
13461         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13462         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13463         medium/large code-model.
13464         (legitimize_pic_address): Likewise.
13465         (legitimize_tls_address): Likewise.
13466         (ix86_expand_call): Likewise.
13467         (x86_output_mi_thunk): Likewise.
13468         (get_dllimport_decl): Add new beimport argument.
13469         (construct_plt_address): Don't assert for x64 pe-coff targets.
13470         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13471         targets.
13472         (SYMBOL_FLAG_STUBVAR): New macro.
13473         (SYMBOL_REF_STUBVAR_P): Likewise.
13474         * config/i386/winnt.c (stub_list): New structure.
13475         (stub_head): New local variable.
13476         (i386_pe_record_stub): New function.
13477         (i386_pe_file_end): Emit refptr-stubs.
13479 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
13481         PR rtl-optimization/56745
13482         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13483         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13485         PR c++/34949
13486         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13487         and both of them are MEM_REFs, just compare first argument for
13488         equality and attempt to deal even with differing offsets.
13490         PR c++/34949
13491         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13492         of gimple_clobber_p to be MEM_REF.
13493         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13494         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
13495         after gimplification.
13496         * asan.c (get_mem_ref_of_assignment): Don't instrument
13497         gimple_clobber_p stmts.
13498         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13499         gimple_clobber_p stmt if they have MEM_REF lhs and
13500         are dead because of another gimple_clobber_p stmt.
13501         * tree-ssa-live.c (clear_unused_block_pointer): Treat
13502         gimple_clobber_p stmts like debug stmts.
13503         (remove_unused_locals): Remove clobbers with MEM_REF lhs
13504         that refer to unused VAR_DECLs or uninitialized values.
13505         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13506         gimple_clobber_p stmts if they refer to removed parameters.
13507         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13508         formatting.
13510 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
13512         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13513         using SWI48 mode attribute.
13515 2013-04-02  Wei Mi  <wmi@google.com>
13517         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13518         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13519         *<rotate_insn><mode>3_mask in i386.md.
13521 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13523         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13525 2013-04-02  Richard Biener  <rguenther@suse.de>
13527         PR tree-optimization/56778
13528         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13529         Runtime alias tests are not supported for gather loads.
13530         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13531         stmts referenced from SSA operands before updating SSA form.
13533 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
13534             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13536         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13537         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13538         * config/arm/cortex-a53.md: New file.
13539         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13540         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13541         * config/arm/arm.c (arm_issue_rate): Likewise.
13542         * config/arm/arm-tune.md: Regenerate
13543         * config/arm/arm-tables.opt: Regenerate.
13544         * config/arm/arm-cores.def: Add cortex-a53.
13546 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
13548         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13549         non-static link.
13551 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13553         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13554         scalar load/store operations using B/H registers.
13555         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13557 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13559         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13560         scalar move.
13561         * config/aarch64/aarch64.c
13562         (aarch64_simd_scalar_immediate_valid_for_move): New.
13563         * config/aarch64/aarch64-protos.h
13564         (aarch64_simd_scalar_immediate_valid_for_move): New.
13565         * config/aarch64/constraints.md (Dh, Dq): New.
13566         * config/aarch64/iterators.md (hq): New.
13568 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
13570         * reorg.c (get_branch_condition): Deal with conditional returns.
13571         (fill_simple_delay_slots): Remove dead code dealing with jumps.
13573 2013-04-01  Wei Mi  <wmi@google.com>
13575         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13576         Truncate operand 2 using %b asm operand modifier.
13577         (*<shift_insn><mode>3_mask): Ditto.
13578         (*<rotate_insn><mode>3_mask): Ditto.
13580 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
13582         PR middle-end/56798
13583         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13585 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
13587         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13588         of next_real_insn.
13589         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13591 2013-03-30  Lawrence Crowl  <crowl@google.com>
13593         * dse.c (clear_alias_sets): Remove never set.
13594         (disqualified_clear_alias_sets): Remove never set.
13595         (clear_alias_mode_pool): Remove never set.
13596         (dse_step0): Remove condition that is never true.
13597         (canon_address): Remove condition that is never true.
13598         (dse_step7): Remove condition that is never true.
13599         (rest_of_handle_dse): Remove condition that is never true.
13600         (rest_of_handle_dse::did_global): Remove never read from above.
13601         (dse_step2_spill): Remove never called from above.
13602         (dse_step5_spill): Remove never called from above.
13604 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
13606         * doc/md.texi (Standard Names) <casesi>: Update documentation for
13607         JUMP_TABLE_DATA changes.
13608         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13609         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13610         (Insns) <jump_table_data>: New entry.
13611         * doc/tm.texi: Regenerate.
13613         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13615         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13616         for table jump at the end of a basic block using tablejump_p.
13617         * targhooks.c (default_invalid_within_doloop): Likewise.
13618         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13619         target hook implementation that is identical to the default hook.
13620         (rs6000_invalid_within_doloop): Remove.
13622         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13623         unused variable from tablejump_p call.
13625         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13626         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13627         (INSN_DELETED_P): Likewise.
13628         (emit_jump_table_data): New prototype.
13629         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13630         after 4th as unused.
13631         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13632         * sched-vis.c (print_insn): Likewise.
13633         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13634         insn for compatibility with back ends that use next_active_insn to
13635         identify jump table data.
13636         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13637         (remove_insn): Likewise.
13638         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13639         to be emitted.
13640         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13641         (emit_jump_table_data): New function.
13643         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13644         basic block, a JUMP_TABLE_DATA never is.
13645         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13646         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13647         off from code handling real insns.
13648         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13649         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13650         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13651         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13652         is not a NONDEBUG_INSN_P.
13653         * ira-costs.c (scan_one_insn): Likewise.
13654         * jump.c (mark_all_labels): Likewise.
13655         (mark_jump_label_1): Likewise.
13656         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13657         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13658         a basic block.
13659         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13660         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13661         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13662         JUMP_TABLE_DATA_P insns.
13663         (calculate_elim_costs_all_insns): Likewise.
13664         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13665         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13666         (delete_output_reload): Code style fixups.
13667         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13668         insn flags on this non-insn.
13669         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13670         as scheduling barriers, for pre-change compatibility.
13671         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13672         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13674         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13675         redundant JUMP_TABLE_DATA_P test.
13676         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13677         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13678         (frv_for_each_packet): Likewise.
13679         * config/i386/i386.c (min_insn_size): Likewise.
13680         (ix86_avoid_jump_mispredicts): Likewise.
13681         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13682         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13683         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13684         (mips16_insn_length): Robustify.
13685         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13686         (mips16_split_long_branches): Likewise.
13687         * config/pa/pa.c (pa_combine_instructions): Likewise.
13688         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13689         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13690         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13691         as contributing to pool range lengths.
13692         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13693         Remove redundant JUMP_TABLE_DATA_P test.
13694         (sh_loop_align): Likewise.
13695         (split_branches): Likewise.
13696         (sh_insn_length_adjustment): Likewise.
13697         * config/spu/spu.c (get_branch_target): Likewise.
13699 2013-03-29  Jan Hubicka  <jh@suse.cz>
13701         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13702         gcov streaming; stream hot bb threshold to ltrans.
13703         * predict.c (get_hot_bb_threshold): Break out from ....
13704         (maybe_hot_count_p): ... here.
13705         (set_hot_bb_threshold): New function.
13706         * lto-section-in.c (lto_section_name): Add profile.
13707         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13708         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13709         and data-streamer.h
13710         (histogram_entry): New structure.
13711         (histogram, histogram_pool): New global vars.
13712         (histogram_hash): New structure.
13713         (histogram_hash::hash): New method.
13714         (histogram_hash::equal): Likewise.
13715         (account_time_size): New function.
13716         (cmp_counts): New function.
13717         (dump_histogram): New function.
13718         (ipa_profile_generate_summary): New function.
13719         (ipa_profile_write_summary): New function.
13720         (ipa_profile_read_summary): New function.
13721         (ipa_profile): Decide on threshold.
13722         (pass_ipa_profile): Add ipa_profile_write_summary and
13723         ipa_profile_read_summary.
13724         * Makefile.in (ipa.o): Update dependencies.
13725         * lto-streamer.h (LTO_section_ipa_profile): New section.
13727 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13729         * tree.h (VAR_P): New.
13731 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13733         PR lto/56777
13734         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13736 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13738         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13739         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13740         (control_flow_insn_p): Likewise.
13741         * cfgrtl.c (duplicate_insn_chain): Likewise.
13742         * final.c (get_attr_length_1): Likewise.
13743         (shorten_branches): Likewise.
13744         (final_scan_insn): Likewise.
13745         * function.c (instantiate_virtual_regs): Likewise.
13746         * gcse.c (insert_insn_end_basic_block): Likewise.
13747         * ira-costs.c (scan_one_insn): Likewise.
13748         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13749         * lra.c (check_rtl): Likewise.
13750         * reload1.c (elimination_costs_in_insn): Likewise.
13751         * reorg.c (follow_jumps): Likewise.
13753         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13754         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13755         (thumb_far_jump_used_p): Likewise.
13756         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13757         (workaround_speculation): Likewise.
13758         (add_sched_insns_for_speculation): Likewise.
13759         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13760         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13761         (frv_for_each_packet): Likewise.
13762         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13763         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13764         (final_emit_insn_group_barriers): Likewise.
13765         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13766         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13767         (mips16_insn_length): Likewise.
13768         * config/pa/pa.c (pa_reorg): Likewise.
13769         (pa_combine_instructions): Likewise.
13770         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13771         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13772         (sh_reorg): Likewise.
13773         (split_branches): Likewise.
13774         * config/spu/spu.c (get_branch_target): Likewise.
13776         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13777         JUMP_TABLE_DATA_P.
13779 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13781         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13782         Fix declaration name.
13784 2013-03-28  Lawrence Crowl  <crowl@google.com>
13786         * graphds.h (struct graph.indicies): Remove unused.
13787         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13788         (SCOP_ORIGINAL_PDDRS): Remove unused.
13789         * sese.h (extern insert_loop_close_phis): Removed unused.
13790         (extern insert_guard_phis): Removed unused.
13791         (extern ivtype_map_elt_info): Removed unused.
13792         (new_ivtype_map_elt): Removed unused.
13793         * sese.c (ivtype_map_elt_info): Removed unused.
13795 2013-03-28  Lawrence Crowl  <crowl@google.com>
13797         * Makefile.in: Add several missing include dependences.
13798         (DUMPFILE_H): New.
13799         (test-dump.o): New.  This object is not added to any executable,
13800         but is present for ad-hoc testing.
13801         * bitmap.c
13802         (debug (const bitmap_head_def &)): New.
13803         (debug (const bitmap_head_def *)): New.
13804         * bitmap.h
13805         (extern debug (const bitmap_head_def &)): New.
13806         (extern debug (const bitmap_head_def *)): New.
13807         * cfg.c
13808         (debug (edge_def &)): New.
13809         (debug (edge_def *)): New.
13810         * cfghooks.c
13811         (debug (basic_block_def &)): New.
13812         (debug (basic_block_def *)): New.
13813         * dumpfile.h
13814         (dump_node (const_tree, int, FILE *)): Correct source file.
13815         * dwarf2out.c
13816         (debug (die_struct &)): New.
13817         (debug (die_struct *)): New.
13818         * dwarf2out.h
13819         (extern debug (die_struct &)): New.
13820         (extern debug (die_struct *)): New.
13821         * gimple-pretty-print.c
13822         (debug (gimple_statement_d &)): New.
13823         (debug (gimple_statement_d *)): New.
13824         * gimple-pretty-print.h
13825         (extern debug (gimple_statement_d &)): New.
13826         (extern debug (gimple_statement_d *)): New.
13827         * ira-build.c
13828         (debug (ira_allocno_copy &)): New.
13829         (debug (ira_allocno_copy *)): New.
13830         (debug (ira_allocno &)): New.
13831         (debug (ira_allocno *)): New.
13832         * ira-int.h
13833         (extern debug (ira_allocno_copy &)): New.
13834         (extern debug (ira_allocno_copy *)): New.
13835         (extern debug (ira_allocno &)): New.
13836         (extern debug (ira_allocno *)): New.
13837         * ira-lives.c
13838         (debug (live_range &)): New.
13839         (debug (live_range *)): New.
13840         * lra-int.h
13841         (debug (lra_live_range &)): New.
13842         (debug (lra_live_range *)): New.
13843         * lra-lives.c
13844         (debug (lra_live_range &)): New.
13845         (debug (lra_live_range *)): New.
13846         * omega.c
13847         (debug (omega_pb_d &)): New.
13848         (debug (omega_pb_d *)): New.
13849         * omega.h
13850         (extern debug (omega_pb_d &)): New.
13851         (extern debug (omega_pb_d *)): New.
13852         * print-rtl.c
13853         (debug (const rtx_def &)): New.
13854         (debug (const rtx_def *)): New.
13855         * print-tree.c
13856         (debug_tree (tree): Move within file.
13857         (debug_raw (const tree_node &)): New.
13858         (debug_raw (const tree_node *)): New.
13859         (dump_tree_via_hooks (const tree_node *, int)): New.
13860         (debug (const tree_node &)): New.
13861         (debug (const tree_node *)): New.
13862         (debug_verbose (const tree_node &)): New.
13863         (debug_verbose (const tree_node *)): New.
13864         (debug_head (const tree_node &)): New.
13865         (debug_head (const tree_node *)): New.
13866         (debug_body (const tree_node &)): New.
13867         (debug_body (const tree_node *)): New.
13868         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13869         (debug (vec<tree, va_gc> &)): New.
13870         (debug (vec<tree, va_gc> *)): New.
13871         * rtl.h
13872         (extern debug (const rtx_def &)): New.
13873         (extern debug (const rtx_def *)): New.
13874         * sbitmap.c
13875         (debug_raw (simple_bitmap_def &)): New.
13876         (debug_raw (simple_bitmap_def *)): New.
13877         (debug (simple_bitmap_def &)): New.
13878         (debug (simple_bitmap_def *)): New.
13879         * sbitmap.h
13880         (extern debug (simple_bitmap_def &)): New.
13881         (extern debug (simple_bitmap_def *)): New.
13882         (extern debug_raw (simple_bitmap_def &)): New.
13883         (extern debug_raw (simple_bitmap_def *)): New.
13884         * sel-sched-dump.c
13885         (debug (vinsn_def &)): New.
13886         (debug (vinsn_def *)): New.
13887         (debug_verbose (vinsn_def &)): New.
13888         (debug_verbose (vinsn_def *)): New.
13889         (debug (expr_def &)): New.
13890         (debug (expr_def *)): New.
13891         (debug_verbose (expr_def &)): New.
13892         (debug_verbose (expr_def *)): New.
13893         (debug (vec<rtx> &)): New.
13894         (debug (vec<rtx> *)): New.
13895         * sel-sched-dump.h
13896         (extern debug (vinsn_def &)): New.
13897         (extern debug (vinsn_def *)): New.
13898         (extern debug_verbose (vinsn_def &)): New.
13899         (extern debug_verbose (vinsn_def *)): New.
13900         (extern debug (expr_def &)): New.
13901         (extern debug (expr_def *)): New.
13902         (extern debug_verbose (expr_def &)): New.
13903         (extern debug_verbose (expr_def *)): New.
13904         (extern debug (vec<rtx> &)): New.
13905         (extern debug (vec<rtx> *)): New.
13906         * sel-sched-ir.h
13907         (_list_iter_cond_expr): Make inline instead of static.
13908         * sreal.c
13909         (debug (sreal &)): New.
13910         (debug (sreal *)): New.
13911         * sreal.h
13912         (extern debug (sreal &)): New.
13913         (extern debug (sreal *)): New.
13914         * tree.h
13915         (extern debug_raw (const tree_node &)): New.
13916         (extern debug_raw (const tree_node *)): New.
13917         (extern debug (const tree_node &)): New.
13918         (extern debug (const tree_node *)): New.
13919         (extern debug_verbose (const tree_node &)): New.
13920         (extern debug_verbose (const tree_node *)): New.
13921         (extern debug_head (const tree_node &)): New.
13922         (extern debug_head (const tree_node *)): New.
13923         (extern debug_body (const tree_node &)): New.
13924         (extern debug_body (const tree_node *)): New.
13925         (extern debug (vec<tree, va_gc> &)): New.
13926         (extern debug (vec<tree, va_gc> *)): New.
13927         * tree-cfg.c
13928         (debug (struct loop &)): New.
13929         (debug (struct loop *)): New.
13930         (debug_verbose (struct loop &)): New.
13931         (debug_verbose (struct loop *)): New.
13932         * tree-dump.c: Add header dependence.
13933         * tree-flow.h
13934         (extern debug (struct loop &)): New.
13935         (extern debug (struct loop *)): New.
13936         (extern debug_verbose (struct loop &)): New.
13937         (extern debug_verbose (struct loop *)): New.
13938         * tree-data-ref.c
13939         (debug (data_reference &)): New.
13940         (debug (data_reference *)): New.
13941         (debug (vec<data_reference_p> &)): New.
13942         (debug (vec<data_reference_p> *)): New.
13943         (debug (vec<ddr_p> &)): New.
13944         (debug (vec<ddr_p> *)): New.
13945         * tree-data-ref.h
13946         (extern debug (data_reference &)): New.
13947         (extern debug (data_reference *)): New.
13948         (extern debug (vec<data_reference_p> &)): New.
13949         (extern debug (vec<data_reference_p> *)): New.
13950         (extern debug (vec<ddr_p> &)): New.
13951         (extern debug (vec<ddr_p> *)): New.
13952         * tree-ssa-alias.c
13953         (debug (pt_solution &)): New.
13954         (debug (pt_solution *)): New.
13955         * tree-ssa-alias.h
13956         (extern debug (pt_solution &)): New.
13957         (extern debug (pt_solution *)): New.
13958         * tree-ssa-alias.c
13959         (debug (_var_map &)): New.
13960         (debug (_var_map *)): New.
13961         (debug (tree_live_info_d &)): New.
13962         (debug (tree_live_info_d *)): New.
13963         * tree-ssa-alias.h
13964         (extern debug (_var_map &)): New.
13965         (extern debug (_var_map *)): New.
13966         (extern debug (tree_live_info_d &)): New.
13967         (extern debug (tree_live_info_d *)): New.
13969 2013-03-28  Jan Hubicka  <jh@suse.cz>
13971         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13973 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13975         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13976         record only when desired or required.
13978 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13980         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13981         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13983 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13985         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13986         (*andsi3_compare0_uxtw): New pattern.
13987         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13988         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13990 2013-03-28  Jan Hubicka  <jh@suse.cz>
13992         * data-streamer-in.c (streamer_read_gcov_count): New function.
13993         * gimple-streamer-out.c: Include value-prof.h.
13994         (output_gimple_stmt): Output histogram.
13995         (output_bb): Use streamer_write_gcov_count.
13996         * value-prof.c: Include data-streamer.h
13997         (dump_histogram_value): Add HIST_TYPE_MAX.
13998         (stream_out_histogram_value): New function.
13999         (stream_in_histogram_value): New function.
14000         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14001         (stream_out_histogram_value, stream_in_histogram_value): Declare.
14002         * data-streamer-out.c (streamer_write_gcov_count): New function.
14003         (streamer_write_gcov_count_stream): New function.
14004         * lto-cgraph.c (lto_output_edge): Update counter streaming.
14005         (lto_output_node): Likewise.
14006         (input_node, input_edge): Likewise.
14007         * lto-streamer-out.c (output_cfg): Update streaming.
14008         * lto-streamer-in.c (input_cfg): Likewise.
14009         * data-streamer.h (streamer_write_gcov_count,
14010         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14011         * gimple-streamer-in.c: Include value-prof.h
14012         (input_gimple_stmt): Input histograms.
14013         (input_bb): Update profile streaming.
14015 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14017         * genmodes.c (emit_max_int): New function.
14018         (emit_insn_modes_h): Added call to emit_max_function.
14019         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14020         Added doc.
14021         * machmode.def: Fixed comment.
14023 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14025         * combine.c (try_combine): Removed useless assert.
14026         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14028 2013-03-28  Marek Polacek  <polacek@redhat.com>
14029             Richard Biener  <rguenther@suse.de>
14031         PR tree-optimization/56695
14032         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14033         build signed result of a vector comparison.
14034         * tree-cfg.c (verify_gimple_comparison): Check that a result
14035         of a vector comparison has signed type.
14037 2013-03-28  Richard Biener  <rguenther@suse.de>
14039         PR tree-optimization/37021
14040         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14041         do not restrict gaps between groups.
14042         * tree-vect-stmts.c (vectorizable_load): Properly account for
14043         a gap between groups.
14045 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
14047         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14048         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14049         is not enabled.
14051 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
14053         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14054         * doc/extend.texi (Named Address Spaces): Ditto.
14055         (Variable Attributes): Ditto.
14057 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14059         * config.build: Add support for cygwin x64 target.
14060         * config.gcc: Likewise.
14061         * config.host: Likewise.
14062         * configure.ac: Likewise
14063         * configure: Regenerated.
14065 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14067         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14068         * config/i386/t-cygwin-w64: New file.
14069         * config/i386/cygwin-w64.h: New file.
14070         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14071         and add support for x64-cygwin target.
14072         (CPP_SPEC): Likewise.
14073         (CXX_WRAP_SPEC_LIST): Undefine before define.
14074         (LIBGCJ_SONAME): Use 15 as version.
14076 2013-03-27  Richard Biener  <rguenther@suse.de>
14078         PR tree-optimization/56716
14079         * tree-ssa-structalias.c (perform_var_substitution): Adjust
14080         dumping for ref nodes.
14082 2013-03-27  Martin Jambor  <mjambor@suse.cz>
14084         PR tree-optimization/55334
14085         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14086         restricted pointers to arrays.
14088 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14090         * Makefile.in (.SUFFIXES): Add .cc.
14091         (.c.o): Apply same recipe for implicit rule .cc.o.
14093 2013-03-27  Richard Biener  <rguenther@suse.de>
14095         PR tree-optimization/37021
14096         * tree-vect-data-refs.c (vect_check_strided_load): Allow
14097         REALPART/IMAGPART_EXPRs around the supported refs.
14098         * tree-ssa-structalias.c (find_func_aliases): Assume that
14099         floating-point values are not used to transfer pointers.
14101 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
14103         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14104         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14105         * gcc/doc/tm.texi: Regenerate.
14106         * gcc/targhooks.h (default_has_ifunc_p): New.
14107         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14108         * gcc/config/linux-protos.h: New file.
14109         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14110         this hook for linux which disables support of indirect functions in
14111         android.
14112         * gcc/config/linux-android.c: New file.
14113         * gcc/config/t-linux-android.c: Ditto.
14114         * gcc/config.gcc: Added new object file linux-android.o.
14115         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14116         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14117         * gcc/varasm.c (do_assemble_alias): Likewise.
14118         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14119         doesn't support indirect functions.
14120         * configure: Regenerate.
14122 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
14124         PR target/56102
14125         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14126         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14127         mult-word mode.
14129 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14131         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14133 2013-03-27  Terry Guo  <terry.guo@arm.com>
14135         * config/arm/arm-cores.def: Added core cortex-r7.
14136         * config/arm/arm-tune.md: Regenerated.
14137         * config/arm/arm-tables.opt: Regenerated.
14138         * doc/invoke.texi: Added entry for core cortex-r7.
14140 2013-03-27  Walter Lee  <walt@tilera.com>
14142         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14143         double-decrement of next_scratch_regno.
14145 2013-03-27  Walter Lee  <walt@tilera.com>
14147         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14148         input operands.
14149         (insn_v1mulus): Ditto.
14150         (insn_v2muls): Ditto.
14152 2013-03-27  Walter Lee  <walt@tilera.com>
14154         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14155         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14157 2013-03-27  Walter Lee  <walt@tilera.com>
14159         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14160         (*sibcall_value): Ditto.
14162 2013-03-27  Walter Lee  <walt@tilera.com>
14164         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14165         (insn_mnz_v8qi): ... this ...
14166         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
14167         vector equivalent.
14168         (insn_v<n>mnz): Replaced by ...
14169         (insn_v1mnz): ... this ...
14170         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
14171         equivalent.
14172         (insn_mz_<mode>): Replaced by ...
14173         (insn_mz_v8qi): ... this ...
14174         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
14175         vector equivalent.
14176         (insn_v<n>mz): Replaced by ...
14177         (insn_v1mz): ... this ...
14178         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
14179         equivalent.
14181 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
14183         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14185 2013-03-26  Roland McGrath  <mcgrathr@google.com>
14187         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14188         than fprintf with a non-constant, non-format string.
14190 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14192         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14193         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
14194         operand 0 predicate.
14195         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14196         attribute.  Use general_x64nomem_operand as operand 1 predicate.
14197         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14198         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14199         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14200         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14201         (mov<mode>_insv_1): Remove expander.  Merge insn with
14202         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14203         Use general_x64nomem_operand as operand 1 predicate.
14204         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14205         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14206         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14207         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14208         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
14209         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14210         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14211         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14212         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
14213         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14214         (general_x64nomem_operand): Ditto.
14216 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14218         * config/rtems.opt: Add -pthread option.
14220 2013-03-26  Richard Biener  <rguenther@suse.de>
14222         * alias.c (find_base_term): Avoid redundant and not used recursion.
14223         (base_alias_check): Get the initial base term from the caller.
14224         (true_dependence_1): Compute and pass base terms to base_alias_check.
14225         (write_dependence_p): Likewise.
14226         (may_alias_p): Likewise.
14228 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
14230         * config/aarch64/aarch64.c (aarch64_classify_address): Support
14231         PC-relative load in SI modes and above only.
14233 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
14235         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14236         * config/arm/iwmmxt.md (WCGR0): Update.
14237         (WCGR1, WCGR2, WCGR3): Likewise.
14239 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14241         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14242         Use x64 and nox64 isa attributes.
14244 2013-03-26  Richard Biener  <rguenther@suse.de>
14246         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14247         alignment computations and rely on get_object_alignment_1
14248         for the !TYPE_P case.
14249         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14251 2013-03-26  Walter Lee  <walt@tilera.com>
14253         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14254         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14256 2013-03-25  Jeff Law  <law@redhat.com>
14258         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14259         check for INTEGRAL_TYPE_P that was missing due to checking in
14260         wrong version of prior patch.
14262 2013-03-25  Walter Lee  <walt@tilera.com>
14264         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14265         TILEGX_INSN_SHUFFLEBYTES1.
14266         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14267         shufflebytes1.
14268         (tilegx_builtins): Ditto.
14269         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14271 2013-03-25  Walter Lee  <walt@tilera.com>
14273         * config/tilegx/tilegx.md (floatsisf2): New pattern.
14274         (floatunssisf2): New pattern.
14275         (floatsidf2): New pattern.
14276         (floatunssidf2): New pattern.
14278 2013-03-25  Walter Lee  <walt@tilera.com>
14280         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14281         tests for constraint J, K, N, P.
14283 2013-03-25  Walter Lee  <walt@tilera.com>
14285         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14286         Use indirect/pcrel encoding.
14287         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14288         Ditto.
14290 2013-03-25  Steve Ellcey  <sellcey@mips.com>
14292         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14293         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14294         * config/mips/mips.c (mips_option_override): Set IMADD default.
14295         * config/mips/mips.h (PTF_AVOID_IMADD): New.
14296         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14297         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14298         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14299         * doc/invoke.texi (-mimadd/-mno-imadd): New.
14301 2013-03-25  Jeff Law  <law@redhat.com>
14303         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14304         slightly to avoid creating and folding useless trees.  Simplify
14305         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14307 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
14309         * config/i386/i386.md (*zero_extendsidi2): Merge with
14310         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
14311         * config/i386/predicates.md (x86_64_zext_operand): Rename from
14312         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
14313         targets.  Clarify comment.
14315 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14317         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14318         pass-through jump functions differently.
14319         (ipa_read_jump_function): Likewise.  Also use setter functions to set
14320         up jump functions.
14322 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14324         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14325         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14326         process it.
14327         (ipa_get_indirect_edge_target): New function.
14328         (devirtualization_time_bonus): New parameter known_aggs, pass it to
14329         ipa_get_indirect_edge_target.  Update all callers.
14330         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
14331         ipa_get_indirect_edge_target_1 instead of calling
14332         ipa_get_indirect_edge_target.
14333         (create_specialized_node): Pass aggvlas to
14334         ipcp_discover_new_direct_edges.
14336 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14338         * config/arm/arm.md (f_sels, f_seld): New types.
14339         (*cmov<mode>): New pattern.
14340         * config/arm/predicates.md (arm_vsel_comparison_operator): New
14341         predicate.
14343 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14345         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14346         POSIX-printf for mingw-hosted  builds.
14348 2013-03-25  Richard Biener  <rguenther@suse.de>
14350         PR middle-end/56694
14351         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14352         must-not-throw stmt location.
14354 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14356         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14357         Emit load-acquire versions when acq is true.
14358         (arm_emit_store_exclusive): Add rel parameter.
14359         Emit store-release versions when rel is true.
14360         (arm_split_compare_and_swap): Use acquire-release instructions
14361         instead.
14362         of barriers when appropriate.
14363         (arm_split_atomic_op): Likewise.
14364         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14365         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14366         (VUNSPEC_SLX): Likewise.
14367         (VUNSPEC_LDA): Likewise.
14368         (VUNSPEC_STL): Likewise.
14369         * config/arm/sync.md (atomic_load<mode>): New pattern.
14370         (atomic_store<mode>): Likewise.
14371         (arm_load_acquire_exclusive<mode>): Likewise.
14372         (arm_load_acquire_exclusivesi): Likewise.
14373         (arm_load_acquire_exclusivedi): Likewise.
14374         (arm_store_release_exclusive<mode>): Likewise.
14376 2013-03-25  Catherine Moore  <clm@codesourcery.com>
14378         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14379         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14380         * config/mip/predicates.md (lwsp_swsp_operand,
14381         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14382         sb16_operand, db4_operand, db7_operand, ib3_operand,
14383         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14384         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14385         andi16_operand): New predicates.
14386         * config/mips/mips.md (compression): New attribute.
14387         (enabled): New attribute.
14388         (length): Consider compression in computing length.
14389         (shift_compression): New code attribute.
14390         (*add<mode>3): New operands. Record compression.
14391         (sub<mode>3): Likewise.
14392         (one_cmpl<mode>2): Likewise.
14393         (*and<mode>3): Likewise.
14394         (*ior<mode>3): Likewise.
14395         (unnamed pattern for xor): Likewise.
14396         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14397         (*<optab><mode>3): Likewise.
14398         (*mov<mode>_internal: Likewise.
14399         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14400         (mips_unsigned_immediate_p): New.
14401         (umips_lwsp_swsp_address_p): New.
14402         (m16_based_address_p): New.
14403         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14404         (mips_unsigned_immediate_p): New prototype.
14405         (lwsp_swsp_address_p): New prototype.
14406         (m16_based_address_p): New prototype.
14407         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14408         (mips_signed_immediate_p): New function.
14409         (m16_based_address_p): New function.
14410         (lwsp_swsp_address_p): New function.
14411         (mips_print_operand_punctuation): Recognize short delay slot insns
14412         for microMIPS.add<mode>3"
14414 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14416         PR target/56720
14417         * config/arm/iterators.md (v_cmp_result): New mode attribute.
14418         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14420 2013-03-25  Richard Biener  <rguenther@suse.de>
14422         PR tree-optimization/56689
14423         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14424         any edge.
14426 2013-03-25  Richard Biener  <rguenther@suse.de>
14428         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14429         of bitmap.
14430         (memory_references): Likewise.
14431         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14432         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14433         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14434         (gather_mem_refs_in_loops): Fold into ...
14435         (analyze_memory_references): ... this.  Move initialization
14436         to tree_ssa_lim_initialize.
14437         (fill_always_executed_in): Rename to ...
14438         (fill_always_executed_in_1): ... this.
14439         (fill_always_executed_in): Move contains_call computation to
14440         this new function from ...
14441         (tree_ssa_lim_initialize): ... here.
14442         (tree_ssa_lim): Call fill_always_executed_in.
14444 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14446         * postreload.c (reload_combine): Fix code detecting returns.
14448 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14450         * function.c (emit_use_return_register_into_block): On cc0 targets,
14451         do not emit the sequence between cc0 setter and user.
14453 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14455         * config/i386/predicates.md (local_symbolic_operand): Interpret
14456         dll-imported symbols as none-local.
14458 2013-03-25  Richard Biener  <rguenther@suse.de>
14460         * tree-ssa-loop-im.c (struct depend): Remove.
14461         (struct lim_aux_data): Make depends a vec of gimples.
14462         (free_lim_aux_data): Adjust.
14463         (add_dependency): Likewise.
14464         (set_level): Likewise.
14466 2013-03-25  Richard Biener  <rguenther@suse.de>
14468         PR middle-end/56434
14469         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14470         the pointer returned by calls with ECF_MALLOC set.
14472 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14474         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14476 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14478         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14479         using MMXMODE mode iterator.
14480         (*move<mode>_internal): Merge with *movv2sf_internal and
14481         *movv2sf_internal_rex64 using MMXMODE mode iterator.
14483 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14485         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14486         (record_last_mem_set_info): Likewise.
14488         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14489         of XNEWVEC followed by memset.
14490         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14492 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14494         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14495         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14496         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14497         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14498         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14499         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14500         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14501         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14502         BARRIER_P instead of GET_CODE.
14504 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14506         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14507         inaccuracy in the probing code.
14509         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14510         (ctrapdi4): Likewise.
14512 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14514         * calls.c (expand_call): Add missing guard to code handling return
14515         of non-BLKmode structures in MSB.
14516         * function.c (expand_function_end): Likewise.
14518 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14520         * combine.c (try_combine): Adjust comment.  Do not add the set of
14521         insn #0 if the destination indirectly is set or dies in insn #2.
14522         Tidy up code to distribute a new note.
14524 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14526         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14527         also for alternatives 16 and 17.
14529 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14531         * config/i386/sse.md (*mov<mode>_internal): Merge with
14532         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
14533         Emit insn template depending on type attribute.  Use
14534         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14535         movd instead of movq mnemonic for interunit moves.  Rewrite mode
14536         attribute calculation.  Remove unit attribute calculation.
14537         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14538         Set prefix_data16 attribute for DImode ssemov types.
14539         Use Ym instead of y for SSE-MMX conversion alternatives.
14540         Reorder operand constraints.
14542 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
14544         * df.h (df_insn_delete): Adjust prototype.
14545         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14546         and let it decide whether mark the basic block dirty.
14547         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14548         * df-scan.c (df_insn_info_delete): New helper function, split
14549         off from df_insn_delete.
14550         (df_scan_free_bb_info): Use it.
14551         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14552         Likewise.
14553         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
14554         that the insn is actually an insn and it has a non-NULL basic block.
14555         Do not mark basic block dirty if only deleting a DEBUG_INSN.
14557 2013-03-22  Richard Biener  <rguenther@suse.de>
14559         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14560         dep_ref members.
14561         (mem_ref_alloc): Do not allocate them.
14562         (refs_independent_p): Do not query or maintain a cache.
14564 2013-03-22  Richard Biener  <rguenther@suse.de>
14566         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14567         (gather_mem_refs_in_loops): Do not compute it.
14568         (analyze_memory_references): Do not allocate it.
14569         (tree_ssa_lim_finalize): Do not free it.
14570         (for_all_locs_in_loop): Do not query all_refs_in_loop.
14572 2013-03-22  Richard Biener  <rguenther@suse.de>
14574         * is-a.h (as_a): Use gcc_checking_assert.
14576 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
14578         * config/aarch64/aarch64.c (aarch64_print_operand): New
14579         format specifier for printing a constant in hex.
14580         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14581         format specifier for printing second operand.
14583 2013-03-22  Richard Biener  <rguenther@suse.de>
14585         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14586         bitmaps.
14587         (gather_mem_refs_in_loops): Perform store accumulation here.
14588         (create_vop_ref_mapping_loop): Remove.
14589         (create_vop_ref_mapping): Likewise.
14590         (analyze_memory_references): Initialize refs_stored_in_loop.
14591         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14592         (record_indep_loop): Remove.
14593         (record_dep_loop): New function.
14594         (ref_indep_loop_p_1): Adjust to only walk over references
14595         in the loop, not its subloops.
14596         (ref_indep_loop_p): Rename to ...
14597         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14598         maintaining a more fine-grained cache.
14599         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14600         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14602 2013-03-22  Richard Biener  <rguenther@suse.de>
14604         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14605         (struct mem_ref): Make accesses_in_loop a vec of a vec of
14606         aggregate mem_ref_loc.
14607         (free_mem_ref_locs): Inline into ...
14608         (memref_free): ... this and adjust.
14609         (mem_ref_alloc): Adjust.
14610         (mem_ref_locs_alloc): Remove.
14611         (record_mem_ref_loc): Adjust.
14612         (get_all_locs_in_loop): Rewrite into ...
14613         (for_all_locs_in_loop): ... this iterator.
14614         (rewrite_mem_ref_loc): New functor.
14615         (rewrite_mem_refs): Use for_all_locs_in_loop.
14616         (sm_set_flag_if_changed): New functor.
14617         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14618         (ref_always_accessed): New functor.
14619         (ref_always_accessed_p): Use for_all_locs_in_loop.
14621 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14623         * tree-pass.h (PROP_gimple_lvec): New.
14624         * passes.c (dump_properties): Handle PROP_gimple_lvec.
14625         (init_optimization_passes): Move pass_lower_vector.
14626         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14627         PROP_gimple_lvec.
14628         (pass_lower_vector): Provide PROP_gimple_lvec.
14629         (pass_lower_vector_ssa): Likewise.
14630         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14632 2013-03-21  Mark Wielaard  <mjw@redhat.com>
14634         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14636 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14638         * config/i386/i386.md (*movdi_internal): Disparage slightly
14639         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14640         conversion alternatives.
14642 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14644         PR middle-end/48087
14645         * diagnostic.def (DK_WERROR): New kind.
14646         * diagnostic.h (werrorcount): Define.
14647         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14648         promoted to DK_ERROR, increment DK_WERROR counter instead of
14649         DK_ERROR counter.
14650         * toplev.c (toplev_main): Call print_ignored_options even if
14651         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14652         even if just werrorcount is non-zero.
14654         PR debug/55608
14655         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14656         on failure.
14657         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14658         (string_cst_pool_decl): New function.
14659         (optimize_one_addr_into_implicit_ptr): New function.
14660         (resolve_addr_in_expr): Optimize DWARF location expression
14661         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14662         which doesn't live in memory, but has DW_AT_location or
14663         DW_AT_const_value, or refers to a string literal, into
14664         DW_OP_GNU_implicit_pointer.
14665         (optimize_location_into_implicit_ptr): New function.
14666         (resolve_addr): If removing DW_AT_location of a variable because
14667         it was DW_OP_addr of address of the variable, but the variable doesn't
14668         live in memory, try to emit const value attribute for the initializer.
14670 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14672         * tree.h (VECTOR_TYPE_P): New macro.
14673         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14674         TYPE_MODE): Use it.
14675         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14676         VEC_COND_EXPR cannot be lvalues.
14677         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14679 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14681         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14682         Restrict the transformation to equal modes.
14684 2013-03-21  Richard Biener  <rguenther@suse.de>
14686         PR tree-optimization/39326
14687         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14688         (MEM_ANALYZABLE): Adjust.
14689         (record_mem_ref_loc): Move bitmap ops ...
14690         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14691         unanalyzable refs, do not record locations for it.
14692         (analyze_memory_references): Allocate ref zero as shared
14693         unanalyzable ref.
14694         (refs_independent_p): Do not test for unanalyzed mems here.
14695         (ref_indep_loop_p_1): Special-case disambiguation against
14696         the unanalyzed ref.
14697         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14699 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14701         * config/arm/arm-protos.h (tune_params): Add
14702         prefer_neon_for_64bits field.
14703         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14704         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14705         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14706         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14707         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14708         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14709         (arm_option_override): Handle -mneon-for-64bits new option.
14710         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14711         (prefer_neon_for_64bits): Declare new variable.
14712         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14713         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14714         (arch_enabled): Handle new arch types. Remove support for onlya8
14715         and nota8.
14716         (one_cmpldi2): Use new arch names.
14717         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14718         * config/arm/arm.opt (mneon-for-64bits): Add option.
14719         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14720         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14721         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14722         of onlya8.
14723         * doc/invoke.texi (-mneon-for-64bits): Document.
14725 2013-03-21  Richard Biener  <rguenther@suse.de>
14727         PR tree-optimization/39326
14728         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14729         (sort_bbs_in_loop_postorder_cmp): New function.
14730         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14732 2013-03-21  Richard Biener  <rguenther@suse.de>
14734         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14735         (vect_insert_into_interleaving_chain): Likewise.
14736         (vect_drs_dependent_in_basic_block): Inline ...
14737         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14738         split out from ...
14739         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14740         (vect_check_interleaving): Simplify.
14741         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14742         (vect_slp_analyze_data_ref_dependences): ... this new function.
14743         (dr_group_sort_cmp): New function.
14744         (vect_analyze_data_ref_accesses): Compute data-reference groups
14745         here instead of in vect_analyze_data_ref_dependence.  Use
14746         a more efficient algorithm.
14747         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14748         vect_slp_analyze_data_ref_dependences.  Call
14749         vect_analyze_data_ref_accesses earlier.
14750         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14751         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14752         (vect_slp_analyze_data_ref_dependences): New prototype.
14754 2013-03-21  Richard Biener  <rguenther@suse.de>
14756         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14757         ref is stored in the loop.
14758         (find_refs_for_sm): Walk only over all stores.
14759         (store_motion_loop): Allocate from lim_bitmap_obstack.
14760         (store_motion): Likewise.
14762 2013-03-21  Richard Biener  <rguenther@suse.de>
14764         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14765         Update virtual SSA form.
14767 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14769         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14770         * configure: Regenerate.
14771         * config.in: Regenerate.
14772         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14773         if !HAVE_LD_EH_FRAME_CIEV3.
14775 2013-03-21  Richard Biener  <rguenther@suse.de>
14777         * tree-cfg.c (verify_expr_no_block): New function.
14778         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14779         nor DECL_VALUE_EXPR have locations with associated blocks.
14780         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14781         (clear_unused_block_pointer): Remove code dealing with
14782         blocks in DECL_DEBUG_EXPR locations.
14784 2013-03-21  Richard Biener  <rguenther@suse.de>
14786         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14787         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14788         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14789         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14790         instead of DECL_DEBUG_EXPR_IS_FROM.
14791         * gimplify.c (gimplify_modify_expr): Likewise.
14792         * tree-cfg.c (verify_expr_location_1): Likewise.
14793         * tree-complex.c (create_one_component_var): Likewise.
14794         * tree-sra.c (create_access_replacement): Likewise.
14795         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14796         (clear_unused_block_pointer): Likewise.
14797         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14798         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14799         * var-tracking.c (var_debug_decl): Likewise.
14800         (track_expr_p): Likewise.
14801         * tree-inline.c (add_local_variables): Likewise.  Set
14802         DECL_HAS_DEBUG_EXPR_P after copying it.
14803         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14804         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14806 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14808         PR bootstrap/56656
14809         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14810         * configure: Regenerate.
14811         * config.in: Regenerate.
14812         * config/i386/i386.md (*movdf_internal): Use
14813         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14814         movd instead of movq mnemonic for interunit moves.
14815         (*movdi_internal): Ditto.
14817 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14819         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14820         (abd<mode>_3): New pattern.
14821         (aba<mode>_3): New pattern.
14822         (fabd<mode>_3): New pattern.
14824 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14826         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14827         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14828         occurrence of REGISTER_PREFIX as its empty string.
14830 2013-03-20  Jeff Law  <law@redhat.com>
14832         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14833         addititional equivalences for equality comparisons between an SSA_NAME
14834         and a constant where the SSA_NAME was set from a widening conversion.
14836 2013-03-20  Walter Lee  <walt@tilera.com>
14838         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14840 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14842         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14843         depending on type attribute.
14844         (*movti_internal): Ditto.
14845         (*movtf_internal): Ditto.
14846         (*movxf_internal): Ditto.
14847         (*movdf_internal): Ditto.
14848         (*movsf_internal): Ditto.
14850 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14852         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14853         maybe_vex for sselog1 and ssemov types.
14854         (*movdi_internal): Reorder operand constraints.
14855         (*movsi_internal): Ditto.  Set prefix attribute to
14856         maybe_vex for sselog1 and ssemov types.
14857         (*movtf_internal): Set prefix attribute to maybe_vex
14858         for sselog1 and ssemov types.
14859         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14860         DImode ssemov types.  Reorder operand constraints.
14861         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14862         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14863         attribute for SImode ssemov types.  Reorder operand constraints.
14865 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14867         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14868         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14870 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
14872         * config/rs6000/predicates.md (indexed_address, update_address_mem
14873         update_indexed_address_mem): New predicates.
14874         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14875         attribute for load/store instructions.
14876         * config/rs6000/dfp.md (movsd_store): Likewise.
14877         (movsd_load): Likewise.
14878         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14879         (unnamed HI->DI extend define_insn): Likewise.
14880         (unnamed SI->DI extend define_insn): Likewise.
14881         (unnamed QI->SI extend define_insn): Likewise.
14882         (unnamed QI->HI extend define_insn): Likewise.
14883         (unnamed HI->SI extend define_insn): Likewise.
14884         (unnamed HI->SI extend define_insn): Likewise.
14885         (extendsfdf2_fpr): Likewise.
14886         (movsi_internal1): Likewise.
14887         (movsi_internal1_single): Likewise.
14888         (movhi_internal): Likewise.
14889         (movqi_internal): Likewise.
14890         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14891         attribute for load/store instructions.
14892         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14893         instructions.
14894         (mov<mode>_softfloat): Likewise.
14895         (mov<mode>_hardfloat32): Likewise.
14896         (mov<mode>_hardfloat64): Likewise.
14897         (mov<mode>_softfloat64): Likewise.
14898         (movdi_internal32): Likewise.
14899         (movdi_internal64): Likewise.
14900         (probe_stack_<mode>): Likewise.
14902 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14904         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14905         floating point, and decimal floating point to reload iterator.
14907         * config/rs6000/constraints.md (wl constraint): New constraints to
14908         return FLOAT_REGS if certain options are used to reduce the number
14909         of separate patterns that exist in the file.
14910         (wx constraint): Likewise.
14911         (wz constraint): Likewise.
14913         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14914         -mdebug=reg, print wg, wl, wx, and wz constraints.
14915         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14916         Initialize the reload functions for 64-bit binary/decimal floating
14917         point types.
14918         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14919         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14920         create the buffer on the stack to overcome not having a 32-bit
14921         load and store.
14922         (rs6000_emit_move): Likewise.
14923         (rs6000_secondary_memory_needed_rtx): Likewise.
14924         (rs6000_alloc_sdmode_stack_slot): Likewise.
14925         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14926         via xxlxor, just like DFmode 0.0.
14928         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14929         define as 1 if we are running on a power7 or newer.
14930         (enum r6000_reg_class_enum): Add new constraints.
14932         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14933         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14934         with other moves by using conditional constraits (wg).  Use LFIWZX
14935         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14936         (movsd splitter): Likewise.
14937         (movsd_hardfloat): Likewise.
14938         (movsd_softfloat): Likewise.
14940         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14941         binary and decimal floating point moves.
14942         (fmove_ok): New attributes to combine binary and decimal floating
14943         point moves, and to combine power6x (mfpgpr) moves along normal
14944         floating moves.
14945         (real_value_to_target): Likewise.
14946         (f32_lr): Likewise.
14947         (f32_lm): Likewise.
14948         (f32_li): Likewise.
14949         (f32_sr): Likewise.
14950         (f32_sm): Likewise.
14951         (f32_si): Likewise.
14952         (movsf): Combine binary and decimal floating point moves.  Combine
14953         power6x (mfpgpr) moves with other moves by using conditional
14954         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14955         (mov<mode> for SFmode/SDmode); Likewise.
14956         (SFmode/SDmode splitters): Likewise.
14957         (movsf_hardfloat): Likewise.
14958         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14959         (movsf_softfloat): Likewise.
14960         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14962         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14963         wx and wz constraints.
14965         * config/rs6000/constraints.md (wg constraint): New constraint to
14966         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14968         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14969         constraint.
14971         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14972         -mdebug=reg, print wg, wl, wx, and wz constraints.
14973         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14974         Initialize the reload functions for 64-bit binary/decimal floating
14975         point types.
14976         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14977         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14978         create the buffer on the stack to overcome not having a 32-bit
14979         load and store.
14980         (rs6000_emit_move): Likewise.
14981         (rs6000_secondary_memory_needed_rtx): Likewise.
14982         (rs6000_alloc_sdmode_stack_slot): Likewise.
14983         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14984         via xxlxor, just like DFmode 0.0.
14986         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14987         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14988         with other moves by using conditional constraits (wg).  Use LFIWZX
14989         and STFIWX for loading SDmode on power7.
14990         (movdd splitters): Likewise.
14991         (movdd_hardfloat32): Likewise.
14992         (movdd_softfloat32): Likewise.
14993         (movdd_hardfloat64_mfpgpr): Likewise.
14994         (movdd_hardfloat64): Likewise.
14995         (movdd_softfloat64): Likewise.
14997         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14998         64-bit binary and decimal floating point moves.
14999         (FMOVE64X): Likewise.
15000         (movdf): Combine 64-bit binary and decimal floating point moves.
15001         Combine power6x (mfpgpr) moves with other moves by using
15002         conditional constraits (wg).
15003         (mov<mode> for DFmode/DDmode): Likewise.
15004         (DFmode/DDmode splitters): Likewise.
15005         (movdf_hardfloat32): Likewise.
15006         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15007         (movdf_softfloat32): Likewise.
15008         (movdf_hardfloat64_mfpgpr): Likewise.
15009         (movdf_hardfloat64): Likewise.
15010         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15011         (movdf_softfloat64): Likewise.
15012         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15013         (reload_<mode>_load): Move to later in the file so they aren't in
15014         the middle of the floating point move insns.
15015         (reload_<mode>_store): Likewise.
15017         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15018         constraint.
15020         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15021         constraint if -mdebug=reg.
15022         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15023         Enable using dd reload support if needed.
15025         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15026         binary and decimal floating point moves in rs6000.md.
15027         (movtd_internal): Likewise.
15029         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15030         decimal floating point moves.
15031         (movtf): Likewise.
15032         (movtf_internal): Likewise.
15033         (mov<mode>_internal, TDmode/TFmode): Likewise.
15034         (movtf_softfloat): Likewise.
15035         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15037         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15038         movdi_internal64, using wg constraint for move direct operations.
15039         (movdi_internal64): Likewise.
15041         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15042         MODES_TIEABLE_P for selected modes.  Print the numerical value of
15043         the various virtual registers. Use GPR/FPR first/last values,
15044         instead of hard coding the register numbers.  Print which modes
15045         have reload functions registered.
15046         (rs6000_option_override_internal): If -mdebug=reg, trace the options
15047         settings before/after setting cpu, target and subtarget settings.
15048         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15049         and for secondary reload failures in rs6000_secondary_reload_inner.
15050         (rs6000_secondary_reload_fail): Likewise.
15051         (rs6000_secondary_reload_inner): Likewise.
15053         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15054         macros for first/last GPR and FPR registers.
15055         (LAST_GPR_REGNO): Likewise.
15056         (FIRST_FPR_REGNO): Likewise.
15057         (LAST_FPR_REGNO): Likewise.
15059         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15060         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15061         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15062         (vcond<mode><mode>): Likewise.
15063         (vcondu<mode><mode>): Likewise.
15064         (vector_gtu<mode>): Likewise.
15065         (vector_gte<mode>): Likewise.
15066         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15067         to prevent the compiler from converting DImode operations to TImode.
15068         (ior<mode>3): Likewise.
15069         (and<mode>3): Likewise.
15070         (one_cmpl<mode>2): Likewise.
15071         (nor<mode>3): Likewise.
15072         (andc<mode>3): Likewise.
15074         * config/rs6000/constraints.md (wt constraint): New constraint
15075         that returns VSX_REGS if TImode is allowed in VSX registers.
15077         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15078         constant under VSX.
15080         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15081         similar to TImode, but it is restricted to being in the GPRs.
15083         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15084         TImode to occupy a single VSX register.
15086         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15087         -mvsx-timode for power7/power8.
15088         (power7 cpu): Likewise.
15089         (power8 cpu): Likewise.
15091         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15092         sure that TFmode/TDmode take up two registers if they are ever
15093         allowed in the upper VSX registers.
15094         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15095         registers.
15096         (rs6000_init_hard_regno_mode_ok): Likewise.
15097         (rs6000_debug_reg_global): Add debugging for PTImode and wt
15098         constraint.  Print if LRA is turned on.
15099         (rs6000_option_override_internal): Give an error if -mvsx-timode
15100         and VSX is not enabled.
15101         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
15102         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15103         to reg+offset addressing.  Use PTImode when checking offset
15104         addresses for validity.
15105         (reg_offset_addressing_ok_p): Likewise.
15106         (rs6000_legitimate_offset_address_p): Likewise.
15107         (rs6000_legitimize_address): Likewise.
15108         (rs6000_legitimize_reload_address): Likewise.
15109         (rs6000_legitimate_address_p): Likewise.
15110         (rs6000_eliminate_indexed_memrefs): Likewise.
15111         (rs6000_emit_move): Likewise.
15112         (rs6000_secondary_reload): Likewise.
15113         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
15114         reloads to fpr registers to continue to use reg+offset addressing,
15115         but 64-bit reloads to altivec registers need reg+reg addressing.
15116         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15117         it.  Treat LO_SUM like a PLUS operation.
15118         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15119         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15120         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15121         registers to share a register with a smaller sized type, since VSX
15122         puts scalars in the upper 64-bits.
15123         (print_operand): Add support for PTImode.
15124         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15125         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15126         registers, but don't have arithmetic support.
15127         (rs6000_memory_move_cost): Add test for VSX.
15128         (rs6000_opt_masks): Add -mvsx-timode.
15130         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15131         for TImode.
15132         (VSs): Likewise.
15133         (VSr): Use wt constraint for TImode.
15134         (VSv): Drop TImode support.
15135         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15136         (vsx_movti_64bit): Likewise.
15137         (vsx_movti_32bit): Likewise.
15138         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15139         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15140         one '?' on the appropriate output constraint.  Do not allow TImode
15141         logical operations on 32-bit systems.
15142         (vsx_ior<mode>3): Likewise.
15143         (vsx_xor<mode>3): Likewise.
15144         (vsx_one_cmpl<mode>2): Likewise.
15145         (vsx_nor<mode>3): Likewise.
15146         (vsx_andc<mode>3): Likewise.
15147         (vsx_concat_<mode>): Likewise.
15148         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15150         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15151         OPTION_MASK_VSX_TIMODE.
15152         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15153         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15155         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15156         (TI2 iterator): New iterator for TImode, PTImode.
15157         (wd mode attribute): Add values for vector types.
15158         (movti_string): Replace TI move operations with operations for TImode
15159         and PTImode.  Add support for TImode being allowed in VSX registers.
15160         (mov<mode>_string, TImode/PTImode): Likewise.
15161         (movti_ppc64): Likewise.
15162         (mov<mode>_ppc64, TImode/PTImode): Likewise.
15163         (TI mode splitters): Likewise.
15165         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15166         constraint.
15168 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
15170         PR tree-optimization/56355
15171         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15172         Also handle integers with undefined overflow.
15174 2013-03-20  Catherine Moore  <clm@codesourcery.com>
15175             Maciej W. Rozycki  <macro@codesourcery.com>
15176             Tom de Vries  <tom@codesourcery.com>
15177             Nathan Sidwell  <nathan@codesourcery.com>
15178             Iain Sandoe  <iain@codesourcery.com>
15179             Nathan Froyd  <froydnj@codesourcery.com>
15180             Chao-ying Fu  <fu@mips.com>
15182         * doc/extend.texi: (micromips, nomicromips, nocompression):
15183         Document new function attributes.
15184         * doc/invoke.texi (minterlink-compressed, mmicromips,
15185         m14k, m14ke, m14kec): Document new options.
15186         (minterlink-mips16): Update documentation.
15187         * doc/md.texi (ZC, ZD): Document new constraints.
15188         * configure.ac (gcc_cv_as_micromips): Check if linker
15189         supports the .set micromips directive.
15190         * configure: Regenerate.
15191         * config.in: Regenerate.
15192         * config/mips/mips-tables.opt: Regenerate.
15193         * config/mips/micromips.md: New file.
15194         * constraints.md (ZC, ZD): New constraints.
15195         * config/mips/predicates.md (movep_src_register): New predicate.
15196         (movep_src_operand): New predicate.
15197         (non_volatile_mem_operand): New predicate.
15198         * config/mips/mips.md (multimem): New type.
15199         (length): Differentiate between 17-bit and 18-bit branch offsets.
15200         (MOVEP1, MOVEP2): New mode iterator.
15201         (mov_<load>l): Use ZC constraint.
15202         (mov_<load>r): Likewise.
15203         (mov_<store>l): Likewise.
15204         (mov_<store>r): Likewise.
15205         (*branch_equality<mode>_inverted): Add microMIPS support.
15206         (*branch_equality<mode>): Likewise.
15207         (*jump_absolute): Likewise.
15208         (indirect_jump_<mode>): Likewise.
15209         (tablejump_<mode>): Likewise.
15210         (<optab>_internal): Likewise.
15211         (sibcall_internal): Likewise.
15212         (sibcall_value_internal): Likewise.
15213         (prefetch): Use constraint ZD.
15214         * config/mips/mips.opt (minterlink-compressed): New option.
15215         (minterlink-mips16): Now an alias for minterlink-compressed.
15216         (mmicromips): New option.
15217         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15218         (compare_and_swap_12): Likewise.
15219         (sync_add<mode>): Likewise.
15220         (sync_<optab>_12): Likewise.
15221         (sync_old_<optab>_12): Likewise.
15222         (sync_new_<optab>_12): Likewise.
15223         (sync_nand_12): Likewise.
15224         (sync_old_nand_12): Likewise.
15225         (sync_new_nand_12): Likewise.
15226         (sync_sub<mode>): Likewise.
15227         (sync_old_add<mode>): Likewise.
15228         (sync_old_sub<mode>): Likewise.
15229         (sync_new_add<mode>): Likewise.
15230         (sync_new_sub<mode>): Likewise.
15231         (sync_<optab><mode>): Likewise.
15232         (sync_old_<optab><mode>): Likewise.
15233         (sync_new_<optab><mode>): Likewise.
15234         (sync_nand<mode>): Likewise.
15235         (sync_old_nand<mode>): Likewise.
15236         (sync_new_nand<mode>): Likewise.
15237         (sync_lock_test_and_set<mode>): Likewise.
15238         (test_and_set_12): Likewise.
15239         (atomic_compare_and_swap<mode>): Likewise.
15240         (atomic_exchange<mode>_llsc): Likewise.
15241         (atomic_fetch_add<mode>_llsc): Likewise.
15242         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15243         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15244         (umips_save_restore_pattern_p): Likewise.
15245         (umips_load_store_pair_p): Likewise.
15246         (umips_output_load_store_pair): Likewise.
15247         (umips_movep_target_p): Likewise.
15248         (umips_12bit_offset_address_p): Likewise.
15249         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15250         (mips_base_mips16): Rename this...
15251         (mips_base_compression_flags): ...to this. Update all uses.
15252         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15253         (mips_mips16_decl_p): Delete.
15254         (mips_nomips16_decl_p): Delete.
15255         (mips_get_compress_on_flags): New function.
15256         (mips_get_compress_off_flags): New function.
15257         (mips_get_compress_mode): New function.
15258         (mips_get_compress_on_name): New function.
15259         (mips_get_compress_off_name): New function.
15260         (mips_insert_attributes): Support multiple compression types.
15261         (mips_merge_decl_attributes): Likewise.
15262         (umips_12bit_offset_address_p): New function.
15263         (mips_start_function_definition): Emit .set micromips directive.
15264         (mips_call_may_need_jalx_p): New function.
15265         (mips_function_ok_for_sibcall): Add microMIPS support.
15266         (mips_print_operand_punctuation): Support short delay slots and
15267         compact jumps.
15268         (umips_swm_mask, umips_swm_encoding): New.
15269         (umips_build_save_restore): New function.
15270         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15271         (was_mips16_p): Remove.
15272         (old_compression_mode): New.
15273         (mips_set_compression_mode): New function.
15274         (mips_set_current_function): Add microMIPS support.
15275         (mips_option_override): Likewise.
15276         (umips_save_restore_pattern_p): New function.
15277         (umips_output_save_restore): New function.
15278         (umips_load_store_pair_p_1): New function.
15279         (umips_load_store_pair_p): New function.
15280         (umips_output_load_store_pair_1): New function.
15281         (umips_output_load_store_pair): New function.
15282         (umips_movep_target_p) New function.
15283         (mips_prepare_pch_save): Add microMIPS support.
15284         * config/mips/mips.h (TARGET_COMPRESSION): New.
15285         (TARGET_CPU_CPP_BUILTINS): Update macro
15286         to use new compression flags and to support microMIPS.
15287         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15288         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15289         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15290         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15291         (ASM_SPEC): Support mmicromips and mno-micromips.
15292         (M16STORE_REG_P): New macro.
15293         (MIPS_CALL): Support TARGET_MICROMIPS.
15294         (MICROMIPS_J): New macro.
15295         (mips_base_mips16): Rename this...
15296         (mips_base_compression_flags): ...to this.
15297         (UMIPS_12BIT_OFFSET_P): New macro.
15298         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15299         (MULTILIB_DIRNAMES): Likewise.
15300 2013-03-20  Richard Biener  <rguenther@suse.de>
15302         PR tree-optimization/56661
15303         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15304         the result does not have to be distinct.
15306 2013-03-20  Richard Biener  <rguenther@suse.de>
15308         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15309         remap_gimple_op_r.
15311 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15312             Steven Bosscher  <steven@gcc.gnu.org>
15314         PR rtl-optimization/56605
15315         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15317 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15319         PR bootstrap/56656
15320         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15321         that require movd instead of movq.
15323 2013-03-20  Richard Biener  <rguenther@suse.de>
15325         * tree-ssa-structalias.c (struct variable_info): Add pointer
15326         to the first field of an aggregate with sub-vars.  Make
15327         this and the pointer to the next subfield its ID.
15328         (vi_next): New function.
15329         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15330         storedanything_id, integer_id): Increment by one.
15331         (new_var_info, get_call_vi, lookup_call_clobber_vi,
15332         get_call_clobber_vi): Adjust.
15333         (solution_set_expand): Simplify and speedup.
15334         (solution_set_add): Inline into ...
15335         (set_union_with_increment): ... this.  Adjust accordingly.
15336         (do_sd_constraint): Likewise.
15337         (do_ds_constraint): Likewise.
15338         (do_complex_constraint): Simplify.
15339         (build_pred_graph): Adjust.
15340         (solve_graph): Likewise.  Simplify and speedup.
15341         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15342         get_constraint_for_component_ref, get_constraint_for_1,
15343         first_vi_for_offset, first_or_preceding_vi_for_offset,
15344         create_function_info_for, create_variable_info_for_1,
15345         create_variable_info_for, intra_create_variable_infos): Adjust.
15346         (init_base_vars): Push NULL for ID zero.
15347         (compute_points_to_sets): Adjust.
15349 2013-03-20  Richard Biener  <rguenther@suse.de>
15351         * cfgloop.c (verify_loop_structure): Streamline and avoid
15352         ICEing on corrupt loop tree.
15353         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15354         loop tree.
15356 2013-03-20  Richard Biener  <rguenther@suse.de>
15358         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15359         check whether an SSA update is needed.
15361 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
15363         * config/mips/constraints.md (T): Rename to...
15364         (Yf): ...this.
15365         (U): Rename to...
15366         (Yd): ...this.
15367         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15368         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15370 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15372         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15373         (*subsi3_carryin_uxtw): Likewise.
15375 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15377         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15378         (*rorsi3_insn_uxtw): Likewise.
15380 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15382         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15383         (*extrsi5_insn_uxtw): Likewise.
15385 2013-03-19  Richard Biener  <rguenther@suse.de>
15387         PR tree-optimization/56273
15388         * passes.c (init_optimization_passes): Move second VRP after DOM.
15390 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
15392         * config/i386/i386.md (*movti_internal): Merge from
15393         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
15394         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
15395         nox64 isa attributes.
15397 2013-03-18  Richard Biener  <rguenther@suse.de>
15399         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15400         (unite): Likewise.
15401         (merge_node_constraints): Likewise.
15402         (build_succ_graph): Likewise.
15403         (valid_graph_edge): Inline into single caller.
15404         (unify_nodes): Likewise.  Use bitmap_set_bit return value
15405         and cache varinfo.
15406         (scc_visit): Fix formatting and variable use.
15407         (do_sd_constraint): Use gcc_checking_assert.
15408         (do_ds_constraint): Likewise.
15409         (do_complex_constraint): Likewise.
15410         (condense_visit): Likewise.  Cleanup.
15411         (dump_pred_graph): New function.
15412         (perform_var_substitution): Dump the pred-graph before
15413         variable substitution.
15414         (find_equivalent_node): Use gcc_checking_assert.
15415         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15417 2013-03-18  Richard Biener  <rguenther@suse.de>
15419         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15420         Remove cond_expr_stmt_list argument and do not gimplify the
15421         built expression.
15422         (vect_loop_versioning): Adjust.
15423         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15424         Cleanup to use less temporaries.
15425         (vect_create_data_ref_ptr): Cleanup.
15427 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15429         PR tree-optimization/56635
15430         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15431         require types_compatible_p types.
15433 2013-03-18  Nick Clifton  <nickc@redhat.com>
15435         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15436         spurious backslash.
15438         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15439         Add missing line to comment describing function.
15441 2013-03-18  Richard Biener  <rguenther@suse.de>
15443         PR tree-optimization/56210
15444         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15445         Handle string / character search functions.
15446         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15448 2013-03-18  Richard Biener  <rguenther@suse.de>
15450         PR middle-end/56483
15451         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15452         and implement properly.
15453         * gimple.h (gimple_cond_single_var_p): Remove.
15455 2013-03-18  Richard Biener  <rguenther@suse.de>
15457         * tree-data-ref.h (find_data_references_in_loop): Declare.
15458         * tree-data-ref.c (get_references_in_stmt): Use a stack
15459         vector pre-allocated in the callers.
15460         (find_data_references_in_stmt): Adjust.
15461         (graphite_find_data_references_in_stmt): Likewise.
15462         (create_rdg_vertices): Likewise.
15463         (find_data_references_in_loop): Export.
15464         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15465         Compute dependences here...
15466         (vect_analyze_data_refs): ...not here.  When we encounter
15467         a non-vectorizable data reference in basic-block vectorization
15468         truncate the data reference vector.  Do not bother to
15469         fixup data-dependence information for gather loads.
15470         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15471         of data references, as reported.
15473 2013-03-18  Richard Biener  <rguenther@suse.de>
15475         PR tree-optimization/3713
15476         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
15477         has_constants and expr.
15478         (stmt_has_constants): Properly valueize SSA names when deciding
15479         whether the stmt has constants.
15481 2013-03-18  Richard Biener  <rguenther@suse.de>
15483         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15484         whole function when there is nothing to do.
15485         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15486         * tree-vectorizer.c (vectorize_loops): Update virtual and
15487         loop-closed SSA once.
15488         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15490 2013-03-18  Richard Biener  <rguenther@suse.de>
15492         PR middle-end/56113
15493         * domwalk.c (bb_postorder): New global static.
15494         (cmp_bb_postorder): New function.
15495         (walk_dominator_tree): Replace scheme imposing an order for
15496         visiting dominator sons by one sorting them at the time they
15497         are pushed on the stack.
15499 2013-03-18  Richard Biener  <rguenther@suse.de>
15501         PR tree-optimization/39326
15502         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15503         (struct mem_ref): Replace mem member with ao_ref typed member.
15504         (MEM_ANALYZABLE): Adjust.
15505         (memref_eq): Likewise.
15506         (mem_ref_alloc): Likewise.
15507         (gather_mem_refs_stmt): Likewise.
15508         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15509         (execute_sm_if_changed_flag_set): Adjust.
15510         (execute_sm): Likewise.
15511         (ref_always_accessed_p): Likewise.
15512         (refs_independent_p): Likewise.
15513         (can_sm_ref_p): Likewise.
15515 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15517         PR c/56566
15518         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15519         return 1 even for !unsignedp.
15521 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15523         * config/i386/i386.md (isa): Add x64 and nox64.
15524         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15525         (*pushtf): Enable *roF alternative for x64 isa only.
15526         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
15527         mode attribute of integer alternatives to DImode for TARGET_64BIT.
15528         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
15529         (*movtf_internal): Merge from *movtf_internal_rex64 and
15530         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
15531         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
15532         nox64 isa attributes.
15533         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
15534         nox64 isa attributes.
15535         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15537 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15539         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15541 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
15543         PR target/56640
15544         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15545         class names.  Remove trailing comma after "ALL_REGS".
15547 2013-03-16  Jan Hubicka  <jh@suse.cz>
15549         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15550         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15551         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15552         of cgraph_get_create_node.
15553         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15555 2013-03-16  Jason Merrill  <jason@redhat.com>
15557         PR debug/49090
15558         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15559         with DW_AT_default_value.
15561 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
15563         * BASE-VER: Set to 4.9.0.
15565 2013-03-14  Andi Kleen  <ak@linux.intel.com>
15567         PR target/56619
15568         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15569         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15570         Document _x* TSX intrinsics.
15572 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
15573             David Holsgrove  <david.holsgrove@xilinx.com>
15575         * configure.ac: Add MicroBlaze TLS support detection.
15576         * configure: Regenerate.
15577         * config/microblaze/microblaze-protos.h
15578         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15579         symbol_mentioned_p, label_mentioned_p): Add prototypes.
15580         * config/microblaze/microblaze.c (microblaze_address_type): Add
15581         ADDRESS_TLS and tls_reloc address types.
15582         (microblaze_address_info): Add tls_reloc.
15583         (TARGET_HAVE_TLS): Define.
15584         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15585          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15586          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15587         load_tls_operand,  microblaze_call_tls_get_addr,
15588         microblaze_legitimize_tls_address): New functions.
15589         (microblaze_classify_unspec): Handle UNSPEC_TLS.
15590         (get_base_reg): Use microblaze_tls_symbol_p.
15591         (microblaze_classify_address): Handle TLS.
15592         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15593         label_mentioned_p and microblaze_tls_referenced_p.
15594         (microblaze_legitimize_address): Handle TLS.
15595         (microblaze_address_insns): Handle ADDRESS_TLS.
15596         (pic_address_needs_scratch): Handle TLS.
15597         (print_operand_address): Handle TLS.
15598         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15599         (microblaze_expand_move): Handle TLS.
15600         (microblaze_legitimate_constant_p): Check
15601         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15602         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15603         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15604         (PIC_OFFSET_TABLE_REGNUM): Set.
15605         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15606         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15607         (addsi3, movsi_internal2, movdf_internal): Update constraints
15608         * config/microblaze/predicates.md (arith_plus_operand): Define
15609         (move_operand): Redefine as move_src_operand,
15610         check microblaze_tls_referenced_p.
15612 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15614         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15615         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15617 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15619         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15620         CC mode for AND.
15622 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
15624         PR tree-optimization/53265
15625         * common.opt (Waggressive-loop-optimizations): New option.
15626         * tree-ssa-loop-niter.c: Include tree-pass.h.
15627         (do_warn_aggressive_loop_optimizations): New function.
15628         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
15629         if number_of_latch_executions returned constant.
15630         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15631         early.  If number_of_latch_executions returned constant, set
15632         nb_iterations_upper_bound back to it.
15633         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15634         field.
15635         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15636         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15638         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15639         (MULTILIB_OSDIRNAMES): Set.
15640         * genmultilib: If defaultosdirname doesn't start with :: , set
15641         defaultosdirname2 instead, clear it and emit two . multilib_raw
15642         entries instead of just one.
15644 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15646         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15647         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15648         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15649         (SUBTARGET_OVERRIDE_OPTIONS): New.
15651 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15653         PR target/49880
15654         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15655         (musermode): Convert to Var(TARGET_USERMODE).
15656         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15657         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15658         * config/sh/sh.c (sh_option_override): Use
15659         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15660         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15661         condition.
15662         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15663         TARGET_SH4.
15664         (udivsi3_i4_single, divsi3_i4_single): Use
15665         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15667 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15669         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15670         default setting.
15672 2013-03-13  Richard Biener  <rguenther@suse.de>
15674         PR tree-optimization/56608
15675         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15676         calls when vectorizing basic-blocks.
15678 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15680         PR plugins/45078
15681         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15682         tm_file.
15684 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15686         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15688 2013-03-11  Jan Hubicka  <jh@suse.cz>
15690         PR lto/56557
15691         * lto-streamer-out.c (output_symbol_p): Skip references from
15692         constructors of external variables.
15694 2013-03-11  Jan Hubicka  <jh@suse.cz>
15696         PR middle-end/56571
15697         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15698         from pseudos.
15699         * emit-rtl.c (verify_rtx_sharing): Likewise.
15700         (copy_insn_1): Likewise.
15701         * rtl.c (copy_rtx): Likewise.
15703 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15705         PR target/56591
15706         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15707         output_operand_lossage message.
15709 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15711         PR target/56470
15712         * arm.c (shift_op): Validate RTL pattern on the fly.
15713         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15714         the RTL.
15716 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15718         PR target/56347
15719         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15720         new call patterns that clobber %fr12.
15721         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15722         split and postreload patterns.
15723         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15724         registers %fr12 and %fr12R as call used.
15726 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15728         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15729         (canon_address, record_store, replace_read, check_mem_read_rtx,
15730         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15731         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15732         rest_of_handle_dse): Likewise.
15734 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15736         PR middle-end/56524
15737         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15738         Add base_optabs.
15739         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15740         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15741         (save_optabs_if_changed): Replace with...
15742         (init_tree_optimization_optabs): ...this.
15743         * optabs.c (save_optabs_if_changed): Rename to...
15744         (init_tree_optimization_optabs): ...this.  Take the optimization node
15745         as argument.  Do nothing if the base optabs are already correct.
15746         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15747         to recompute optabs.
15748         * function.h (function): Remove optabs field.
15749         * function.c (invoke_set_current_function_hook): Call
15750         init_tree_optimization_optabs.  Use the result to initialize
15751         this_fn_optabs.
15753 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15755         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15756         if GTMA_HAS_NO_INSTRUMENTATION.
15757         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15758         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15759         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15760         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15761         GTMA_HAS_NO_INSTRUMENTATION.
15763 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15765         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15766         libasan_preinit.o.
15768 2013-03-08  Marek Polacek  <polacek@redhat.com>
15769             Jakub Jelinek  <jakub@redhat.com>
15771         PR tree-optimization/56478
15772         * predict.c (is_comparison_with_loop_invariant_p): Change the
15773         type of loop_step to tree.
15774         (predict_loops): Adjust.
15775         (predict_iv_comparison): Perform the computations on double_ints.
15777 2013-03-08  Richard Biener  <rguenther@suse.de>
15779         PR tree-optimization/56570
15780         * tree-cfg.c (verify_expr_location_1): Verify locations for
15781         DECL_DEBUG_EXPR.
15782         * tree-sra.c (create_access_replacement): Strip locations
15783         from DECL_DEBUG_EXPRs.
15785 2013-03-08  Richard Biener  <rguenther@suse.de>
15787         * tree-inline.c (expand_call_inline): Do not associate
15788         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15789         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15791 2013-03-08  Richard Biener  <rguenther@suse.de>
15793         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15794         or block changes with -Og.  Fix for location / block encoding
15795         changes and PHI arguments with locations.
15797 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15799         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15800         for all counters.
15801         (struct output_info): Likewise.
15802         (register_overhead): Remove bad gcc_assert.
15803         (bitmap_find_bit): If there is only a single bitmap element, do not
15804         count a miss as a search.
15805         (print_statistics): Update for counter type changes.
15806         (dump_bitmap_statistics): Likewise.  Print headers such that they
15807         are properly lined up with the printed counters.
15809 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15811         PR tree-optimization/56559
15812         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15813         check that it has only a single use.
15815 2013-03-07  Richard Biener  <rguenther@suse.de>
15817         * doc/invoke.texi (fwhole-program): Discourage use in combination
15818         with -flto.
15820 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15822         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15824         PR tree-optimization/56539
15825         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15826         instead of GSI_CONTINUE_LINKING as last argument to
15827         force_gimple_operand_gsi.  Adjust function comment.
15829         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15830         aarch64-cores.def.
15832         PR middle-end/56548
15833         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15834         promoted mode, convert the result back to the original mode.
15836 2013-03-06  Richard Biener  <rguenther@suse.de>
15838         PR middle-end/56294
15839         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15840         (insert_updated_phi_nodes_compare_uids): New function.
15841         (update_ssa): Sort symbols_to_rename after UID before
15842         traversing it to insert PHI nodes.
15844 2013-03-06  Richard Biener  <rguenther@suse.de>
15846         PR middle-end/50494
15847         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15848         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15850         Revert
15851         2013-02-13  Richard Biener  <rguenther@suse.de>
15853         PR lto/50494
15854         * varasm.c (output_constant_def_1): Get the decl representing
15855         the constant as argument.
15856         (output_constant_def): Wrap output_constant_def_1.
15857         (make_decl_rtl): Use output_constant_def_1 with the decl
15858         representing the constant.
15859         (build_constant_desc): Optionally re-use a decl already
15860         representing the constant.
15861         (tree_output_constant_def): Adjust.
15863 2013-03-06  Joey Ye  <joey.ye@arm.com>
15865         PR lto/50293
15866         * gcc.c (convert_white_space): New function.
15867         (main): Handles white space in function name.
15869 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15871         PR target/56529
15872         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15873         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15874         to SH_DIV_CALL_TABLE for TARGET_SH2.
15875         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15876         list.
15877         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15878         call-table options.
15880 2013-03-05  Sterling Augustine  <saugustine@google.com>
15881             Cary Coutant  <ccoutant@google.com>
15883         PR debug/55364
15884         * dwarf2out.c (resolve_addr): Don't call
15885         remove_loc_list_addr_table_entries a second time for the same
15886         expression.
15888 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15890         PR debug/56510
15891         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15892         (avoid_complex_debug_insns): New function.
15893         (expand_debug_locations): Call it.
15895         PR rtl-optimization/56484
15896         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15897         lifetimes of hard registers on small register class machines.
15899 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15901         * config/microblaze/microblaze-protos.h: Rename
15902         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15903         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15904         fast_interrupt.
15905         (microblaze_fast_interrupt_function_p): New function.
15906         (microblaze_is_interrupt_handler): Rename to
15907         microblaze_is_interrupt_variant and add fast_interrupt check.
15908         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15909         (save_restore_insns): Likewise.
15910         (compute_frame_size): Likewise.
15911         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15912         (microblaze_globalize_label): Likewise.
15913         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15914         * config/microblaze/microblaze.md: Use wrapper
15915         microblaze_is_interrupt_variant.
15917 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15919         * sdbout.c (sdbout_one_type): Switch to current function's section
15920         supporting cold/hot.
15922 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15924         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15925         -mxl-reorder.
15927 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15929         PR middle-end/56461
15930         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15931         if VALGRIND_GET_VBITS is defined, temporarily make object
15932         memory all defined, and restore previous valgrind addressability
15933         and definability afterwards.  Free this_object at the end.
15935         PR middle-end/56461
15936         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15937         right before calling lra_create_live_ranges, also call it
15938         when clearing live_p.  Only call lra_clear_live_ranges
15939         at the end if live_p.
15941         PR middle-end/56461
15942         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15944 2013-03-05  Richard Biener  <rguenther@suse.de>
15946         PR tree-optimization/56521
15947         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15948         value-id.
15950 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15952         PR c++/55135
15953         * except.h (remove_unreachable_eh_regions): New prototype.
15954         * except.c (remove_eh_handler_splicer): New function, split out
15955         of remove_eh_handler.
15956         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15957         warning about running it on many EH regions one at a time.
15958         (remove_unreachable_eh_regions_worker): New function, walk the
15959         EH tree in depth-first order and remove non-marked regions.
15960         (remove_unreachable_eh_regions): New function.
15961         * tree-eh.c (mark_reachable_handlers): New function, split out
15962         from remove_unreachable_handlers.
15963         (remove_unreachable_handlers): Use mark_reachable_handlers and
15964         remove_unreachable_eh_regions.
15965         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15966         and remove_unreachable_eh_regions.
15968 2013-03-05  Richard Biener  <rguenther@suse.de>
15970         PR middle-end/56525
15971         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15972         not freeing them until the end.
15974 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15976         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15978 2013-03-05  Richard Biener  <rguenther@suse.de>
15980         PR tree-optimization/56270
15981         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15982         of loads after scheduling an SLP instance.
15984 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15986         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15987         tic6x.exp.
15988         (check_gcc_parallelize): Run guality.exp as a separate job from
15989         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15990         struct-layout-1.exp with stackalign.exp.
15992         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15994         PR middle-end/56461
15995         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15996         load_index sbitmap even if some bit in it isn't set.
15998         PR middle-end/56461
15999         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16000         (discover_iteration_bound_by_body_walk): Change queues to
16001         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
16002         spelling in comment.  Call safe_push on queues[bound_index] directly.
16003         Release queues[queue_index] in every iteration unconditionally.
16004         Release bounds vector.
16006         PR middle-end/56461
16007         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16008         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16009         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16010         inner_phis vector.
16012 2013-03-05  Richard Biener  <rguenther@suse.de>
16014         PR lto/56515
16015         * tree-inline.c (remap_blocks_to_null): New function.
16016         (expand_call_inline): When expanding a call stmt without
16017         an associated block inline remap all callee blocks to NULL.
16019 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16021         PR rtl-optimization/56494
16022         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16023         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16024         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16026         PR middle-end/56461
16027         * sel-sched-ir.c (free_sched_pools): Release
16028         succs_info_pool.stack[succs_info_pool.max_top] vectors too
16029         if succs_info_pool.max_top isn't -1.
16031         PR bootstrap/56509
16032         * opts.c (opts_obstack, opts_concat): Moved to...
16033         * opts-common.c (opts_obstack, opts_concat): ... here.
16035 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16037         PR middle-end/56461
16038         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16040 2013-03-04  Martin Jambor  <mjambor@suse.cz>
16042         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16043         all appropriate places.
16045 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
16047         PR tree-optimization/56424
16048         * ipa-split.c (split_function): Do not set the RSO flag if result is
16049         not by reference and its type is a register type.
16051 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
16053         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16054         (microblaze_legitimate_pic_operand): Likewise
16055         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16056         new function microblaze_legitimate_pic_operand
16057         * config/microblaze/microblaze-protos.h
16058         (microblaze_legitimate_pic_operand): Declare.
16060 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16062         * config/microblaze/predicates.md (call_insn_simple_operand):
16063         New predicate for supported rtx code types.
16064         * config/microblaze/microblaze.md (call_internal1): Use
16065         call_insn_simple_operand predicate.
16067 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16069         PR middle-end/56461
16070         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16071         partitions.ordered_remove.
16073         PR middle-end/56461
16074         * tree-vect-stmts.c (vectorizable_conversion): Don't call
16075         vec_oprnds0.create (1) for modifier == NONE.
16077         PR middle-end/56461
16078         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16079         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16080         vec_oprnds1 right before pushing anything to it for
16081         scalar_shift_arg.
16083         PR middle-end/56461
16084         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16085         set nbbs to 0 instead of having separate code path.
16086         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16087         instead of false as last argument if returning NULL.
16089 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
16091         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16092         the attribute is now called "target" instead of "option".
16093         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16094         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
16095         attribute/pragma name for TARGET_OPTION_VALID_P and
16096         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
16097         * doc/tm.texi: Regenerated.
16099 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
16101         * config/microblaze/microblaze.c:
16102         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16103         * config/microblaze/microblaze.h: Add -mxl-reorder to
16104         DRIVER_SELF_SPECS.
16105         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16106         instructions emitted if TARGET_REORDER.
16107         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16108         or 0 for -m/-mno case, but initialises as 2 to detect default use case
16109         separately.
16111 2013-03-01  Xinliang David Li  <davidxl@google.com>
16113         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16114         walk length.
16116 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
16118         PR middle-end/56461
16119         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16120         vector even when returning true.  Fix up function comment formatting.
16122         PR middle-end/56461
16123         * ira-build.c (ira_loop_nodes_count): New variable.
16124         (create_loop_tree_nodes): Initialize it.
16125         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16127         PR middle-end/56461
16128         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16129         method on dr_chain and result_chain.
16130         * tree-vect-stmts.c (vectorizable_store): Only call
16131         result_chain.create if j == 0.
16133         PR middle-end/56461
16134         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16135         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16136         before overwriting it.
16138 2013-03-01  Tobias Burnus  <burnus@net-b.de>
16140         * doc/extended.texi (C Extensions): Change order in @menu
16141         to match @node.
16142         (Other MIPS Built-in Functions): Move last MIPS entry before
16143         "picoChip Built-in Functions".
16144         (SH Built-in Functions): Move after RX Built-in Functions.
16145         * doc/gcc.texi (Introduction): Change order in @menu
16146         to match @node.
16147         * doc/md.texi (Constraints): Ditto.
16148         * gty.texi (Type Information): Ditto.
16149         (User-provided marking routines for template types): Make
16150         subsection.
16151         * doc/invoke.texi (AArch64 Options): Move before
16152         "Adapteva Epiphany Options".
16154 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
16155             Jakub Jelinek  <jakub@redhat.com>
16157         PR sanitizer/56454
16158         * asan.c (gate_asan): Lookup no_sanitize_address instead of
16159         no_address_safety_analysis attribute.
16160         * doc/extend.texi (no_address_safety_attribute): Rename to
16161         no_sanitize_address attribute, mention no_address_safety_analysis
16162         attribute as deprecated alias.
16164 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
16166         PR middle-end/56461
16167         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16168         type to vec<vec<tree> > *.
16169         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
16170         to be vec<tree> instead of vec<tree> *, set vec_defs
16171         to vNULL and call vec_defs.create (number_of_vects), adjust other
16172         uses of vec_defs.
16173         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16174         vectorizable_condition): Adjust vect_get_slp_defs callers.
16176 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16178         * config/aarch64/aarch64.c
16179         (aarch64_float_const_representable): Remove unused variable.
16181 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16183         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16185 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16187         * config/aarch64/aarch64-builtins.c
16188         (aarch64_init_simd_builtins): Make static.
16190 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16192         * config/aarch64/aarch64.c
16193         (aarch64_simd_make_constant): Make static.
16195 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16197         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16198         with no initialization to the RHS of debug statements.
16200 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16202         PR tree-optimization/56294
16203         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16204         Adjust dumping.
16205         (get_access_replacement): Do not call create_access_replacement.
16206         Assert a replacement exists.
16207         (get_repl_default_def_ssa_name): Create the replacement declaration
16208         itself.
16210 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16212         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16213         final_end_function.
16215 2013-02-28  Marek Polacek  <polacek@redhat.com>
16217         PR rtl-optimization/56466
16218         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16219         if we're changing a loop.
16220         (peel_loops_completely): Likewise.
16222 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
16224         PR c++/55813
16225         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16227 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16229         PR target/56445
16230         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16231         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16232         INTX_FTYPE_FX, FX_FTYPE_INTX.
16233         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16235 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16237         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16238         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16239         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16240         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16241         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16242         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16243         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16244         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16245         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16246         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16247         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16248         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16249         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16250         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16251         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16252         (avrxmega6): Increase max flash segments from 5 to 6.
16253         * config/avr/t-multilib: Regenerate.
16254         * config/avr/avr-tables.opt: Regenerate.
16255         * doc/avr-mmcu.texi: Regenerate.
16257 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16259         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16260         (avr_device_to_arch): Rename to avr_device_to_ld.
16261         (avr_device_to_as): New prototype.
16262         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16263         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16264         * config/avr/driver-avr.c (avr_device_to_as): New.
16265         (avr_device_to_arch): Rename to avr_device_to_ld.
16267 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16269         PR middle-end/56461
16270         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16271         method on dr_chain and result_chain.
16273         PR middle-end/56461
16274         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16275         pointer_set_destroy on not_executed_last_iteration.
16277         PR middle-end/56461
16278         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16280         PR middle-end/56461
16281         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16282         FOR_EACH_DEFINED_FUNCTION when freeing state.
16284         PR middle-end/56461
16285         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16286         pool_free.
16287         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16288         overwriting it.
16290         PR middle-end/56461
16291         * ipa-cp.c (decide_whether_version_node): Call vec_free on
16292         known_aggs[i].items and release known_aggs vector.
16294         PR middle-end/56461
16295         * ipa-reference.c (propagate): Free node_info even for alias nodes.
16297 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16299         * config/microblaze/microblaze.c (microblaze_emit_compare):
16300         Use xor for EQ/NE comparisions.
16301         * config/microblaze/microblaze.md (cstoresf4): Add constraints
16302         (cbranchsf4): Adjust operator to comparison_operator.
16304 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16306         PR middle-end/56461
16307         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16308         vector.
16309         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16310         vec_safe_push, always update *slot.
16311         (redirect_edge_var_map_clear): Use vec_free.
16312         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16313         (free_var_map_entry): Use vec_free.
16314         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16315         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16317 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
16319         PR middle-end/45472
16320         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16321         when the may_trap_p bit of the exprs being merged differs.
16322         Reorder tests for speculativeness in the logical and operator.
16324 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16326         * incpath.c (add_standard_paths): Use reconcat instead of concat
16327         where appropriate and avoid leaking memory.
16329         * opts.h: Include obstack.h.
16330         (opts_concat): New prototype.
16331         (opts_obstack): New declaration.
16332         * opts.c (opts_concat): New function.
16333         (opts_obstack): New variable.
16334         (init_options_struct): Call gcc_init_obstack on opts_obstack.
16335         (finish_options): Use opts_concat instead of concat
16336         and XOBNEWVEC instead of XNEWVEC.
16337         * opts-common.c (generate_canonical_option, decode_cmdline_option,
16338         generate_option): Likewise.
16339         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16340         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16342         PR target/56455
16343         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16344         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16346 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16348         PR middle-end/56461
16349         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16351 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16353         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16354         (arm_block_move_unaligned_straight): Likewise.
16355         (arm_adjust_block_mem): Likewise.
16357 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16359         PR target/48901
16360         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16361         temp, cond and label.
16362         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16364         PR target/52500
16365         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16366         * config/c6x/c6x.h (dbx_register_map): Update declaration.
16368         PR target/52501
16369         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16370         of prologue/epilogue functions.
16372         PR target/52550
16373         * config/tilegx/tilegx.c (tilegx_expand_prologue):
16374         Remove unused variable cfa_offset.
16375         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16377         PR target/54639
16378         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16379         type promotion to unsigned.
16381         PR target/54640
16382         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16383         for HOST_WIDE_INT of 32 bit / same size as int.
16384         (arm_block_move_unaligned_straight): Likewise.
16385         (arm_adjust_block_mem): Likewise.
16387         PR target/54662
16388         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16389         ALL_CFLAGS.
16391 2013-02-26  Marek Polacek  <polacek@redhat.com>
16393         PR tree-optimization/56426
16394         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16396 2013-02-26  Richard Biener  <rguenther@suse.de>
16398         PR target/56444
16399         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16400         unused variable loops.
16402 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16404         PR tree-optimization/56448
16405         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16406         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16407         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16408         later operands of the references, or even first operand for
16409         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16411         PR tree-optimization/56443
16412         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16413         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16414         to type_for_mode langhook.
16416 2013-02-25  Matt Turner  <mattst88@gmail.com>
16418         * doc/invoke.texi: Document r4700.
16420 2013-02-25  Richard Biener  <rguenther@suse.de>
16422         PR tree-optimization/56175
16423         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16424         split out from ...
16425         (simplify_bitwise_binary): ... here.  Also guard the conversion
16426         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16428 2013-02-25  Catherine Moore  <clm@codesourcery.com>
16430         Revert:
16431         2013-02-24  Catherine Moore  <clm@codesourcery.com>
16432             Maciej W. Rozycki  <macro@codesourcery.com>
16433             Tom de Vries  <tom@codesourcery.com>
16434             Nathan Sidwell  <nathan@codesourcery.com>
16435             Iain Sandoe  <iain@codesourcery.com>
16436             Nathan Froyd  <froydnj@codesourcery.com>
16437             Chao-ying Fu  <fu@mips.com>
16439         * doc/extend.texi: (micromips, nomicromips, nocompression):
16440         Document new function attributes.
16441         * doc/invoke.texi (minterlink-compressed, mmicromips,
16442         m14k, m14ke, m14kec): Document new options.
16443         (minterlink-mips16): Update documentation.
16444         * doc/md.texi (ZC, ZD): Document new constraints.
16445         * configure.ac (gcc_cv_as_micromips): Check if linker
16446         supports the .set micromips directive.
16447         * configure: Regenerate.
16448         * config.in: Regenerate.
16449         * config/mips/mips-tables.opt: Regenerate.
16450         * config/mips/micromips.md: New file.
16451         * constraints.md (ZC, AD): New constraints.
16452         * config/mips/predicates.md (movep_src_register): New predicate.
16453         (movep_src_operand): New predicate.
16454         (non_volatile_mem_operand): New predicate.
16455         * config/mips/mips.md (multimem): New type.
16456         (length): Differentiate between 17-bit and 18-bit branch offsets.
16457         (MOVEP1, MOVEP2): New mode iterator.
16458         (mov_<load>l): Use ZC constraint.
16459         (mov_<load>r): Likewise.
16460         (mov_<store>l): Likewise.
16461         (mov_<store>r): Likewise.
16462         (*branch_equality<mode>_inverted): Add microMIPS support.
16463         (*branch_equality<mode>): Likewise.
16464         (*jump_absolute): Likewise.
16465         (indirect_jump_<mode>): Likewise.
16466         (tablejump_<mode>): Likewise.
16467         (<optab>_internal): Likewise.
16468         (sibcall_internal): Likewise.
16469         (sibcall_value_internal): Likewise.
16470         (prefetch): Use constraint ZD.
16471         * config/mips/mips.opt (minterlink-compressed): New option.
16472         (minterlink-mips16): Now an alias for minterlink-compressed.
16473         (mmicromips): New option.
16474         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16475         (compare_and_swap_12): Likewise.
16476         (sync_add<mode>): Likewise.
16477         (sync_<optab>_12): Likewise.
16478         (sync_old_<optab>_12): Likewise.
16479         (sync_new_<optab>_12): Likewise.
16480         (sync_nand_12): Likewise.
16481         (sync_old_nand_12): Likewise.
16482         (sync_new_nand_12): Likewise.
16483         (sync_sub<mode>): Likewise.
16484         (sync_old_add<mode>): Likewise.
16485         (sync_old_sub<mode>): Likewise.
16486         (sync_new_add<mode>): Likewise.
16487         (sync_new_sub<mode>): Likewise.
16488         (sync_<optab><mode>): Likewise.
16489         (sync_old_<optab><mode>): Likewise.
16490         (sync_new_<optab><mode>): Likewise.
16491         (sync_nand<mode>): Likewise.
16492         (sync_old_nand<mode>): Likewise.
16493         (sync_new_nand<mode>): Likewise.
16494         (sync_lock_test_and_set<mode>): Likewise.
16495         (test_and_set_12): Likewise.
16496         (atomic_compare_and_swap<mode>): Likewise.
16497         (atomic_exchange<mode>_llsc): Likewise.
16498         (atomic_fetch_add<mode>_llsc): Likewise.
16499         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16500         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16501         (umips_save_restore_pattern_p): Likewise.
16502         (umips_load_store_pair_p): Likewise.
16503         (umips_output_load_store_pair): Likewise.
16504         (umips_movep_target_p): Likewise.
16505         (umips_12bit_offset_address_p): Likewise.
16506         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16507         (mips_base_mips16): Rename this...
16508         (mips_base_compression_flags): ...to this. Update all uses.
16509         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16510         (mips_mips16_decl_p): Delete.
16511         (mips_nomips16_decl_p): Delete.
16512         (mips_get_compress_on_flags): New function.
16513         (mips_get_compress_off_flags): New function.
16514         (mips_get_compress_mode): New function.
16515         (mips_get_compress_on_name): New function.
16516         (mips_get_compress_off_name): New function.
16517         (mips_insert_attributes): Support multiple compression types.
16518         (mips_merge_decl_attributes): Likewise.
16519         (umips_12bit_offset_address_p): New function.
16520         (mips_start_function_definition): Emit .set micromips directive.
16521         (mips_call_may_need_jalx_p): New function.
16522         (mips_function_ok_for_sibcall): Add microMIPS support.
16523         (mips_print_operand_punctuation): Support short delay slots and
16524         compact jumps.
16525         (umips_swm_mask, umips_swm_encoding): New.
16526         (umips_build_save_restore): New function.
16527         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16528         (was_mips16_p): Remove.
16529         (old_compression_mode): New.
16530         (mips_set_compression_mode): New function.
16531         (mips_set_current_function): Add microMIPS support.
16532         (mips_option_override): Likewise.
16533         (umips_save_restore_pattern_p): New function.
16534         (umips_output_save_restore): New function.
16535         (umips_load_store_pair_p_1): New function.
16536         (umips_load_store_pair_p): New function.
16537         (umips_output_load_store_pair_1): New function.
16538         (umips_output_load_store_pair): New function.
16539         (umips_movep_target_p) New function.
16540         (mips_prepare_pch_save): Add microMIPS support.
16541         * config/mips/mips.h (TARGET_COMPRESSION): New.
16542         (TARGET_CPU_CPP_BUILTINS): Update macro
16543         to use new compression flags and to support microMIPS.
16544         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16545         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16546         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16547         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16548         (ASM_SPEC): Support mmicromips and mno-micromips.
16549         (M16STORE_REG_P): New macro.
16550         (MIPS_CALL): Support TARGET_MICROMIPS.
16551         (MICROMIPS_J): New macro.
16552         (mips_base_mips16): Rename this...
16553         (mips_base_compression_flags): ...to this.
16554         (UMIPS_12BIT_OFFSET_P): New macro.
16555         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16556         (MULTILIB_DIRNAMES): Likewise.
16558 2013-02-25  Tom de Vries  <tom@codesourcery.com>
16560         PR rtl-optimization/56131
16561         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16562         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16563         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16565 2013-02-25  Tobias Burnus  <burnus@net-b.de>
16567         * doc/invoke.texi (-fsanitize=): Move from optimization
16568         to debugging options.
16570 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16572         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16574 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16575             Alexander Monakov  <amonakov@ispras.ru>
16577         PR middle-end/56077
16578         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16579         flush pending lists also on non-jumps.  Adjust comment.
16581 2013-02-24  Catherine Moore  <clm@codesourcery.com>
16582             Maciej W. Rozycki  <macro@codesourcery.com>
16583             Tom de Vries  <tom@codesourcery.com>
16584             Nathan Sidwell  <nathan@codesourcery.com>
16585             Iain Sandoe  <iain@codesourcery.com>
16586             Nathan Froyd  <froydnj@codesourcery.com>
16587             Chao-ying Fu  <fu@mips.com>
16589         * doc/extend.texi: (micromips, nomicromips, nocompression):
16590         Document new function attributes.
16591         * doc/invoke.texi (minterlink-compressed, mmicromips,
16592         m14k, m14ke, m14kec): Document new options.
16593         (minterlink-mips16): Update documentation.
16594         * doc/md.texi (ZC, ZD): Document new constraints.
16595         * configure.ac (gcc_cv_as_micromips): Check if linker
16596         supports the .set micromips directive.
16597         * configure: Regenerate.
16598         * config.in: Regenerate.
16599         * config/mips/mips-tables.opt: Regenerate.
16600         * config/mips/micromips.md: New file.
16601         * constraints.md (ZC, AD): New constraints.
16602         * config/mips/predicates.md (movep_src_register): New predicate.
16603         (movep_src_operand): New predicate.
16604         (non_volatile_mem_operand): New predicate.
16605         * config/mips/mips.md (multimem): New type.
16606         (length): Differentiate between 17-bit and 18-bit branch offsets.
16607         (MOVEP1, MOVEP2): New mode iterator.
16608         (mov_<load>l): Use ZC constraint.
16609         (mov_<load>r): Likewise.
16610         (mov_<store>l): Likewise.
16611         (mov_<store>r): Likewise.
16612         (*branch_equality<mode>_inverted): Add microMIPS support.
16613         (*branch_equality<mode>): Likewise.
16614         (*jump_absolute): Likewise.
16615         (indirect_jump_<mode>): Likewise.
16616         (tablejump_<mode>): Likewise.
16617         (<optab>_internal): Likewise.
16618         (sibcall_internal): Likewise.
16619         (sibcall_value_internal): Likewise.
16620         (prefetch): Use constraint ZD.
16621         * config/mips/mips.opt (minterlink-compressed): New option.
16622         (minterlink-mips16): Now an alias for minterlink-compressed.
16623         (mmicromips): New option.
16624         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16625         (compare_and_swap_12): Likewise.
16626         (sync_add<mode>): Likewise.
16627         (sync_<optab>_12): Likewise.
16628         (sync_old_<optab>_12): Likewise.
16629         (sync_new_<optab>_12): Likewise.
16630         (sync_nand_12): Likewise.
16631         (sync_old_nand_12): Likewise.
16632         (sync_new_nand_12): Likewise.
16633         (sync_sub<mode>): Likewise.
16634         (sync_old_add<mode>): Likewise.
16635         (sync_old_sub<mode>): Likewise.
16636         (sync_new_add<mode>): Likewise.
16637         (sync_new_sub<mode>): Likewise.
16638         (sync_<optab><mode>): Likewise.
16639         (sync_old_<optab><mode>): Likewise.
16640         (sync_new_<optab><mode>): Likewise.
16641         (sync_nand<mode>): Likewise.
16642         (sync_old_nand<mode>): Likewise.
16643         (sync_new_nand<mode>): Likewise.
16644         (sync_lock_test_and_set<mode>): Likewise.
16645         (test_and_set_12): Likewise.
16646         (atomic_compare_and_swap<mode>): Likewise.
16647         (atomic_exchange<mode>_llsc): Likewise.
16648         (atomic_fetch_add<mode>_llsc): Likewise.
16649         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16650         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16651         (umips_save_restore_pattern_p): Likewise.
16652         (umips_load_store_pair_p): Likewise.
16653         (umips_output_load_store_pair): Likewise.
16654         (umips_movep_target_p): Likewise.
16655         (umips_12bit_offset_address_p): Likewise.
16656         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16657         (mips_base_mips16): Rename this...
16658         (mips_base_compression_flags): ...to this. Update all uses.
16659         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16660         (mips_mips16_decl_p): Delete.
16661         (mips_nomips16_decl_p): Delete.
16662         (mips_get_compress_on_flags): New function.
16663         (mips_get_compress_off_flags): New function.
16664         (mips_get_compress_mode): New function.
16665         (mips_get_compress_on_name): New function.
16666         (mips_get_compress_off_name): New function.
16667         (mips_insert_attributes): Support multiple compression types.
16668         (mips_merge_decl_attributes): Likewise.
16669         (umips_12bit_offset_address_p): New function.
16670         (mips_start_function_definition): Emit .set micromips directive.
16671         (mips_call_may_need_jalx_p): New function.
16672         (mips_function_ok_for_sibcall): Add microMIPS support.
16673         (mips_print_operand_punctuation): Support short delay slots and
16674         compact jumps.
16675         (umips_swm_mask, umips_swm_encoding): New.
16676         (umips_build_save_restore): New function.
16677         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16678         (was_mips16_p): Remove.
16679         (old_compression_mode): New.
16680         (mips_set_compression_mode): New function.
16681         (mips_set_current_function): Add microMIPS support.
16682         (mips_option_override): Likewise.
16683         (umips_save_restore_pattern_p): New function.
16684         (umips_output_save_restore): New function.
16685         (umips_load_store_pair_p_1): New function.
16686         (umips_load_store_pair_p): New function.
16687         (umips_output_load_store_pair_1): New function.
16688         (umips_output_load_store_pair): New function.
16689         (umips_movep_target_p) New function.
16690         (mips_prepare_pch_save): Add microMIPS support.
16691         * config/mips/mips.h (TARGET_COMPRESSION): New.
16692         (TARGET_CPU_CPP_BUILTINS): Update macro
16693         to use new compression flags and to support microMIPS.
16694         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16695         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16696         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16697         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16698         (ASM_SPEC): Support mmicromips and mno-micromips.
16699         (M16STORE_REG_P): New macro.
16700         (MIPS_CALL): Support TARGET_MICROMIPS.
16701         (MICROMIPS_J): New macro.
16702         (mips_base_mips16): Rename this...
16703         (mips_base_compression_flags): ...to this.
16704         (UMIPS_12BIT_OFFSET_P): New macro.
16705         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16706         (MULTILIB_DIRNAMES): Likewise.
16708 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16710         PR target/52555
16711         * target-globals.c (save_target_globals): For init_reg_sets and
16712         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16714 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16716         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16717         * config/aarch64/t-aarch64
16718         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16720 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16722         PR inline-asm/56148
16723         * lra-constraints.c (process_alt_operands): Reload operand
16724         conflicting with earlier clobber only if no more other conflicting
16725         operands.
16727 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16729         PR sanitizer/56393
16730         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16731         if not linking a shared library.
16733 2013-02-22  Seth LaForge  <sethml@google.com>
16735         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16737 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16739         * config/arm/arm.md (split for extendsidi): Update condition.
16740         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16741         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16742         (qhs_zextenddi_cstr): Likewise.
16744 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16746         PR middle-end/56420
16747         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16748         avoid signed wrapping.
16749         (expand_mult): Handle properly multiplication by
16750         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16751         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16752         in the compiler if coeff is HOST_WIDE_INT_MIN.
16753         (expand_divmod): Don't make ext_op1 static, change it's type to
16754         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16756         PR rtl-optimization/50339
16757         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16758         field.
16759         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16760         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16761         into splitting_ashiftrt field.
16762         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16763         ASHIFTRT.
16764         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16765         choices.
16767 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16769         PR middle-end/56108
16770         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16771         are sure to go irrevocable.
16773 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16775         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16776         scalars are valid operands.
16778 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16780         PR tree-optimization/56310
16781         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16782         only matching indices and non-negative final offsets.
16783         (intersect_aggregates_with_edge): Pass src_idx to
16784         agg_replacements_to_vector.  Pass src_idx insstead of index to
16785         intersect_with_agg_replacements.
16787 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16789         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16790         instead of hard-wired defaults.
16792 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16794         * doc/invoke.texi (MIPS Options): Update documentation of the
16795         floating-point multiply-accumulate instruction restrictions.
16797 2013-02-21  Kostya Serebryany  <kcc@google.com>
16799         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16800         asan_shadow_offset on x86_64 linux.
16802 2013-02-21  Richard Biener  <rguenther@suse.de>
16804         PR tree-optimization/56415
16805         Revert
16806         2013-02-11  Richard Biener  <rguenther@suse.de>
16808         PR tree-optimization/56273
16809         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16810         first VRP run.
16812 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16814         PR bootstrap/56258
16815         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16816         instead of @itemx.
16818         PR inline-asm/56405
16819         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16820         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16822 2013-02-20  Jan Hubicka  <jh@suse.cz>
16824         PR tree-optimization/56265
16825         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16826         when target is referenced for first time.
16828 2013-02-20  Richard Biener  <rguenther@suse.de>
16830         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16831         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16832         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16833         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16834         not return anything.
16835         (rename_ssa_copies): Do not remove unused locals.
16836         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16837         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16838         * passes.c (execute_function_todo): Do not schedule unused locals
16839         removal if cleanup_tree_cfg did something.
16840         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16841         about the number of removed locals.
16843 2013-02-20  Richard Biener  <rguenther@suse.de>
16845         PR tree-optimization/56398
16846         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16848 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16850         PR tree-optimization/55334
16851         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16852         restricted pointers to arrays.
16854 2013-02-20  Richard Biener  <rguenther@suse.de>
16855             Jakub Jelinek  <jakub@redhat.com>
16857         PR tree-optimization/56396
16858         * tree-ssa-ccp.c (n_const_val): New static variable.
16859         (get_value): Return NULL for SSA names we don't have a lattice
16860         entry for.
16861         (ccp_initialize): Initialize n_const_val.
16862         * tree-ssa-copy.c (n_copy_of): New static variable.
16863         (init_copy_prop): Initialize n_copy_of.
16864         (get_value): Return NULL_TREE for SSA names we don't have a
16865         lattice entry for.
16867 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16869         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16871 2013-02-20  Richard Biener  <rguenther@suse.de>
16873         * genpreds.c (write_lookup_constraint): Do not compare first
16874         letter of the constraint again.
16876 2013-02-20  Richard Biener  <rguenther@suse.de>
16878         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16879         and ceil_log2.
16880         (get_use_iv_cost): Terminate hashtable walk when coming across
16881         an empty entry.
16883 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16885         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16886         reassociation for avx2 targets.
16888 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16890         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16891         Add version check for v8.10.a to enable microblaze_has_clz
16892         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16893         version and TARGET_PATTERN_COMPARE check
16894         * config/microblaze/microblaze.md: New clzsi2 instruction
16896 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16898         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16899         function before branching.
16901 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16903         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16904         DUMP_INSN_RTX_UID.
16905         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16907 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16909         PR middle-end/55889
16910         * sel-sched.c: Include ira.h.
16911         (implicit_clobber_conflict_p): New function.
16912         (moveup_expr): Use it.
16913         * Makefile.in (sel-sched.o): Depend on ira.h.
16915 2013-02-19  Richard Biener  <rguenther@suse.de>
16917         PR tree-optimization/56384
16918         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16919         (vn_hash_type): Split out from ...
16920         (vn_hash_constant_with_type): ... here.
16921         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16922         (vn_phi_eq): Compare types from vn_phi_s structure.
16923         (vn_phi_lookup): Populate vn_phi_s type.
16924         (vn_phi_insert): Likewise.
16926 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16928         PR tree-optimization/56350
16929         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16930         if haven't found reduction or nested cycle operand, rather than
16931         asserting we must find it.
16933         PR tree-optimization/56381
16934         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16935         to fold_build3.
16937 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16938             Jakub Jelinek  <jakub@redhat.com>
16940         PR target/52555
16941         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16942         (swap_optab_enable): Same.
16943         (init_all_optabs): Use argument instead of global.
16944         * tree.h (struct tree_optimization_option): New field target_optabs.
16945         * expr.h (init_all_optabs): Add argument to prototype.
16946         (TREE_OPTIMIZATION_OPTABS): New.
16947         (save_optabs_if_changed): Protoize.
16948         * optabs.h: Declare this_fn_optabs.
16949         * optabs.c (save_optabs_if_changed): New.
16950         Declare this_fn_optabs.
16951         (init_optabs): Add argument to init_all_optabs() call.
16952         * function.c (invoke_set_current_function_hook): Handle per
16953         function optabs.
16954         * function.h (struct function): New field optabs.
16955         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16956         optimization_current_node has changed.
16957         * target-globals.h (save_target_globals_default_opts): Protoize.
16958         * target-globals.c (save_target_globals_default_opts): New.
16960 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16962         PR target/56347
16963         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16964         registers %fr12 and %fr12R as call used.
16966         PR target/56214
16967         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16968         and HImode, require all displacements to be an integer multiple of
16969         their mode size.
16970         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16971         only allow QImode and HImode when reload is in progress and strict is
16972         true.  Likewise for symbolic addresses.  Use base14_operand to check
16973         displacements in REG+BASE addresses.
16975 2013-02-18  Richard Biener  <rguenther@suse.de>
16977         PR tree-optimization/56366
16978         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16979         handle sign-conversion of outer-loop initial induction value.
16981 2013-02-18  Richard Biener  <rguenther@suse.de>
16983         PR middle-end/56349
16984         * cfghooks.c (merge_blocks): If we merge a latch into another
16985         block adjust references to it.
16986         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16987         (verify_loop_structure): Verify that a recorded latch is in fact
16988         a latch.
16990 2013-02-18  Richard Biener  <rguenther@suse.de>
16992         PR tree-optimization/56321
16993         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16994         order SSA name release and virtual operand unlinking.
16996 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16998         * config/microblaze/microblaze.md (save_stack_block): Define.
16999         (restore_stack_block): Likewise.
17001 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17003         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17004         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17005         * config/microblaze/microblaze.c (microblaze_option_override):
17006         Bail out early for PIC modes when target does not support PIC.
17008 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17010         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17011         Replace with a microblaze version.
17012         (microblaze_trampoline_init): Adapt for microblaze.
17013         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17014         microblaze.
17016 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
17017             Dodji Seketeli  <dodji@redhat.com>
17019         PR asan/56330
17020         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17021         (instrument_mem_region_access): Do not forget to always put
17022         instrumentation of the of 'base' and 'base + len' in a "if (len !=
17023         0) statement, even for cases where either 'base' or 'base + len'
17024         are not instrumented -- because they have been previously
17025         instrumented.  Simplify the logic by putting all the statements
17026         instrument 'base + len' inside a sequence, and then insert that
17027         sequence right before the current insertion point.  Then, to
17028         instrument 'base + len', just get an iterator on that statement.
17029         And do not forget to update the pointer to iterator the function
17030         received as argument.
17032 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
17034         PR rtl-optimization/56348
17035         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17037 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
17039         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17040         (clean_graph_dump_file): Pass base to start_graph_dump.
17042 2013-02-14  Richard Henderson  <rth@redhat.com>
17044         PR target/55941
17045         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17047 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
17049         * collect2-aix.h: Define F_LOADONLY.
17051 2013-02-14  Richard Biener  <rguenther@suse.de>
17053         PR lto/50494
17054         * varasm.c (output_constant_def_1): Get the decl representing
17055         the constant as argument.
17056         (output_constant_def): Wrap output_constant_def_1.
17057         (make_decl_rtl): Use output_constant_def_1 with the decl
17058         representing the constant.
17059         (build_constant_desc): Optionally re-use a decl already
17060         representing the constant.
17061         (tree_output_constant_def): Adjust.
17063 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
17065         Fix an asan crash
17066         * asan.c (instrument_builtin_call):  Really put the length of the
17067         second source argument into src1_len.
17069 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17071         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17072         argument.  If it is false, don't create edge from then_bb to
17073         fallthru_bb.
17074         (insert_if_then_before_iter): Pass true to it.
17075         (build_check_stmt): Pass false to it.
17076         (transform_statements): Flush hash table only on extended basic
17077         block boundaries, rather than at the beginning of every bb.
17078         Don't flush hash table on nonfreeing_call_p calls.
17079         * tree-flow.h (nonfreeing_call_p): New prototype.
17080         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17082 2013-02-13  David S. Miller  <davem@davemloft.net>
17084         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17086 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
17088         PR target/56184
17089         * ira.c (max_regno_before_ira): Move from ...
17090         (ira): ... here.
17091         (fix_reg_equiv_init): Use max_regno_before_ira instead of
17092         vec_safe_length.
17094 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17096         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17098 2013-02-13  Richard Biener  <rguenther@suse.de>
17100         PR lto/56295
17101         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17102         globals in MEM_REFs.
17104 2013-02-13  Richard Biener  <rguenther@suse.de>
17106         * loop-init.c (loop_optimizer_init): Clear loop state when
17107         re-initializing preserved loops.
17108         * loop-unswitch.c (unswitch_single_loop): Return whether
17109         we unswitched the loop.  Do not verify loop state here.
17110         (unswitch_loops): When we unswitched a loop discover new loops.
17112 2013-02-13  Kostya Serebryany  <kcc@google.com>
17114         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17115         on x86_64 linux.
17116         * sanitizer.def: Rename __asan_init to __asan_init_v1.
17118 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
17120         Avoid instrumenting duplicated memory access in the same basic block
17121         * Makefile.in (asan.o): Add new dependency on hash-table.h
17122         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17123         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17124         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17125         (free_mem_ref_resources, has_mem_ref_been_instrumented)
17126         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17127         (get_mem_ref_of_assignment): New functions.
17128         (get_mem_refs_of_builtin_call): Extract from
17129         instrument_builtin_call and tweak a little bit to make it fit with
17130         the new signature.
17131         (instrument_builtin_call): Use the new
17132         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
17133         of is_gimple_builtin_call.
17134         (instrument_derefs, instrument_mem_region_access): Insert the
17135         instrumented memory reference into the hash table.
17136         (maybe_instrument_assignment): Renamed instrument_assignment into
17137         this, and change it to advance the iterator when instrumentation
17138         actually happened and return true in that case.  This makes it
17139         homogeneous with maybe_instrument_assignment, and thus give a
17140         chance to callers to be more 'regular'.
17141         (transform_statements): Clear the memory reference hash table
17142         whenever we enter a new BB, when we cross a function call, or when
17143         we are done transforming statements.  Use
17144         maybe_instrument_assignment instead of instrumentation.  No more
17145         need to special case maybe_instrument_assignment and advance the
17146         iterator after calling it; it's now handled just like
17147         maybe_instrument_call.  Update comment.
17149 2013-02-13  Richard Biener  <rguenther@suse.de>
17151         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17152         Fix loop discovery code.
17154 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
17156         PR inline-asm/56148
17157         * lra-constraints.c (process_alt_operands): Match early clobber
17158         operand with itself.  Check conflicts with earlyclobber only if
17159         the operand is not reloaded.  Prefer to reload conflicting operand
17160         if earlyclobber and matching operands are the same.
17162 2013-02-12  Richard Biener  <rguenther@suse.de>
17164         PR lto/56297
17165         * lto-streamer-out.c (write_symbol): Do not output symbols
17166         for hard register variables.
17168 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
17170         PR target/54222
17171         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17172         (umulsidi3_insn, mulsidi3_insn): New insns.
17174 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
17176         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17177         (struct tune_params): Add vec_costs field.
17178         * config/arm/arm.c (arm_builtin_vectorization_cost)
17179         (arm_add_stmt_cost): New functions.
17180         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17181         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17182         (arm_default_vec_cost): New struct of type cpu_vec_costs.
17183         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17184         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17185         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17186         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17188 2013-02-12  Richard Biener  <rguenther@suse.de>
17190         PR lto/56295
17191         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17192         decls again if possible.
17194 2013-02-12  Richard Biener  <rguenther@suse.de>
17196         PR middle-end/56288
17197         * tree-ssa.c (verify_ssa_name): Fix check, move
17198         SSA_NAME_IN_FREE_LIST check up.
17200 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
17201             Steven Bosscher   <steven@gcc.gnu.org>
17203         PR rtl-optimization/56151
17204         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17205         equal to op0 or op1, and last_insn pattern is CODE operation
17206         with MEM dest and one of the operands matches that MEM.
17208 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
17210         * doc/extend.texi: Document Function Multiversioning and "default"
17211         parameter string to target attribute.
17212         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17213         target attribute parameter is "default".
17214         (ix86_compare_version_priority): Remove checks for target attribute.
17215         (ix86_mangle_function_version_assembler_name): Change error to sorry.
17216         Remove check for target attribute equal to NULL. Add assert.
17217         (ix86_generate_version_dispatcher_body): Change error to sorry.
17219 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
17220             Jack Howarth  <howarth@bromo.med.uc.edu>
17221             Patrick Marlier  <patrick.marlier@gmail.com>
17223         PR libitm/55693
17224         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17225         define ENDFILE_SPEC as TM_DESTRUCTOR.
17226         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17228 2013-02-11  Alexander Potapenko  <glider@google.com>
17229             Jack Howarth  <howarth@bromo.med.uc.edu>
17230             Jakub Jelinek  <jakub@redhat.com>
17232         PR sanitizer/55617
17233         * config/darwin.c (cdtor_record): Rename ctor_record.
17234         (sort_cdtor_records): Rename sort_ctor_records.
17235         (finalize_dtors): New routine to sort destructors by
17236         priority before use in assemble_integer.
17237         (machopic_asm_out_destructor): Use finalize_dtors if needed.
17239 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
17241         PR rtl-optimization/56275
17242         * simplify-rtx.c (avoid_constant_pool_reference): Check that
17243         offset is non-negative and less than cmode size before
17244         calling simplify_subreg.
17246 2013-02-11  Richard Biener  <rguenther@suse.de>
17248         PR tree-optimization/56264
17249         * cfgloop.h (fix_loop_structure): Adjust prototype.
17250         * loop-init.c (fix_loop_structure): Return the number of
17251         newly discovered loops.
17252         * tree-cfgcleanup.c (repair_loop_structures): When new loops
17253         are discovered, do a full loop-closed SSA rewrite.
17255 2013-02-11  Richard Biener  <rguenther@suse.de>
17257         PR tree-optimization/56273
17258         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17259         first VRP run.
17260         (check_array_ref): Fix missing newline in dumps.
17261         (search_for_addr_array): Likewise.
17263 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
17265         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17267 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
17269         PR target/56256
17270         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17272 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
17274         PR rtl-optimization/56246
17275         * lra-constraints.c (simplify_operand_subreg): Try to reuse
17276         reload pseudo.
17277         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17278         constraints are satisfied.
17280 2013-02-08  Jeff Law  <law@redhat.com>
17282         PR debug/53948
17283         * emit-rtl.c (reg_is_parm_p): New function.
17284         * regs.h (reg_is_parm_p): New prototype.
17285         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17286         callee-clobbered registers.
17288 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
17290         PR target/56043
17291         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17292         If there is no implicit builtin declaration, just return NULL.
17294 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17296         * config/i386/sse.md (FMAMODEM): New mode iterator.
17297         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17298         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17300 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17302         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17303         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17304         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17306 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17308         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17309         (microblaze*-*-elf): Likewise.
17310         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17311         LINK_SPEC.
17312         * config/microblaze/microblaze-c.c: Add builtin defines for
17313         _LITTLE_ENDIAN and _BIG_ENDIAN.
17314         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17315         add to TARGET_DEFAULT flags.
17316         Expand ASM_SPEC and LINK_SPEC.
17317         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17318         * config/microblaze/microblaze.md: Update extendsidi2 and
17319         movdi_internal instructions to use low-order / high-order reg
17320         print_operands.
17321         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17322         options and inversemask / mask of LITTLE_ENDIAN.
17323         * config/microblaze/t-microblaze: Expand multilib options to
17324         include mlittle-endian (le) and update exceptions patterns.
17326 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
17328         PR rtl-optimization/56195
17329         * lra-constraints.c (get_reload_reg): Don't reuse regs
17330         if they have smaller mode than requested, if they have
17331         wider mode than requested, try to return a SUBREG.
17333         PR tree-optimization/56250
17334         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17335         if type is unsigned and code isn't MULT_EXPR.
17337 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17339         PR tree-optimization/56064
17340         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17341         bits according to mode.
17342         * fixed-value.h (fixed_from_double_int)
17343         (const_fixed_from_double_int): Adjust comments.
17345 2013-02-08  Richard Biener  <rguenther@suse.de>
17347         PR lto/56231
17348         * lto-streamer.h (struct data_in): Remove current_file, current_line
17349         and current_col members.
17350         * lto-streamer-out.c (lto_output_location): Stream changed bits
17351         en-block for efficiency.
17352         * lto-streamer-in.c (clear_line_info): Remove.
17353         (lto_input_location): Cache current file, line and column
17354         globally via local statics.  Read changed bits en-block.
17355         (input_function): Do not call clear_line_info.
17356         (lto_read_body): Likewise.
17357         (lto_input_toplevel_asms): Likewise.
17359 2013-02-08  Michael Matz  <matz@suse.de>
17361         PR tree-optimization/52448
17362         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17363         (nt_call_phase): New static.
17364         (add_or_mark_expr): Only mark accesses with newer phase than any
17365         call seen.
17366         (nonfreeing_call_p): New.
17367         (nt_init_block): Update nt_call_phase, mark blocks as visited.
17368         (nt_fini_block): Keep blocks marked as visited.
17369         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17371 2013-02-08  Richard Biener  <rguenther@suse.de>
17373         * ira.c (ira): Free broken dominator information.
17375 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17377         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17379 2013-02-08  Marek Polacek  <polacek@redhat.com>
17381         * cfgloop.c (verify_loop_structure): Add more checking of headers.
17383 2013-02-08  Richard Biener  <rguenther@suse.de>
17385         PR middle-end/56181
17386         * cfgloop.h (flow_loops_find): Adjust.
17387         (bb_loop_header_p): Declare.
17388         * cfgloop.c (bb_loop_header_p): New function split out from ...
17389         (flow_loops_find): ... here.  Adjust function signature,
17390         support incremental loop structure update.
17391         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
17392         * cfgloopmanip.c (fix_loop_structure): Move ...
17393         * loop-init.c (fix_loop_structure): ... here.
17394         (apply_loop_flags): Split out from ...
17395         (loop_optimizer_init): ... here.
17396         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
17397         in incremental mode, only remove dead loops here.
17399 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17401         PR target/54222
17402         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17403         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17404         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17405         (*round<mode>3.libgcc): New insns for fixed-modes.
17406         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17407         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17408         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17409         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17410         implementations.  Define to __builtin_avr_absFX,
17411         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17412         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17413         __builtin_avr_countlsFX, respectively.
17414         * config/avr/avr-c.c (target.h): Include it.
17415         (enum avr_builtin_id): New enum.
17416         (avr_resolve_overloaded_builtin): New static function.
17417         (avr_register_target_pragmas): Use it to set
17418         targetm.resolve_overloaded_builtin.
17419         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17420         tree nodes used by DEF_BUILTIN.
17421         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17422         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17423         <AVR_BUILTIN_xxBITS>: Same.
17425 2013-02-08  Richard Biener  <rguenther@suse.de>
17427         * cfgloop.c (verify_loop_structure): Properly handle
17428         a loop exiting to another loop header.
17429         * ira-int.h (ira_loops): Remove.
17430         * ira.c (ira_loops): Remove.
17431         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17432         (do_reload): Use loop_optimizer_finalize.
17433         * ira-build.c (create_loop_tree_nodes): Use get_loops and
17434         number_of_loops to access the loop tree.
17435         (more_one_region_p): Likewise.
17436         (finish_loop_tree_nodes): Likewise.
17437         (rebuild_regno_allocno_maps): Likewise.
17438         (mark_loops_for_removal): Likewise.
17439         (mark_all_loops_for_removal): Likewise.
17440         (remove_unnecessary_regions): Likewise.
17441         (ira_build): Likewise.
17442         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17444 2013-02-08  Richard Biener  <rguenther@suse.de>
17446         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17447         * ipa-pure-const.c (analyze_function): Avoid calling
17448         mark_irreducible_loops twice.
17449         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17451 2013-02-07  David S. Miller  <davem@davemloft.net>
17453         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17454         on 'reg'.
17455         * var-tracking.c (vt_add_function_parameter): Test the presence of
17456         HAVE_window_save properly and do not remap argument registers when
17457         we have a leaf function.
17459 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
17461         PR bootstrap/56227
17462         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17463         instead of "ll".
17464         * config/i386/i386.c (ix86_print_operand): Ditto.
17466 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17468         * lra-constraints.c (process_alt_operands): Fix recently added comment.
17470 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17472         PR rtl-optimization/56225
17473         * lra-constraints.c (process_alt_operands): Check that reload hard
17474         reg can hold value for strict_low_part.
17476 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17478         PR debug/56154
17479         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17480         dwarf2out_end_function.
17481         (in_first_function_p, maybe_at_text_label_p,
17482         first_loclabel_num_not_at_text_label): New variables.
17483         (dwarf2out_var_location): In the first function find out
17484         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17485         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17486         functions.
17488 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
17490         PR rtl-optimization/56178
17491         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17492         SUBREG of a register.  Tidy up related block of code.
17493         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17494         note if the source is a register or a SUBREG of a register.
17496 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17498         PR target/56228
17499         * config/rs6000/rs6000.md (ptrm): New mode attr.
17500         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17501         call_value_indirect_aix<pttrsize>,
17502         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17503         m in constraints.
17505 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17507         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17508         if -bnortl. Convert to strcmp and strncmp.
17510 2013-02-07  Alan Modra  <amodra@gmail.com>
17512         PR target/54009
17513         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17514         addresses won't wrap when offsetting.
17515         (rs6000_secondary_reload): Provide secondary reloads needed for
17516         wrapping LO_SUM addresses.
17518 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
17520         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17521         MACH, just __MACH__.
17523 2013-02-06  Richard Biener  <rguenther@suse.de>
17525         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17526         instead of calling fix_loop_structure.
17528 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
17530         PR middle-end/56217
17531         * omp-low.c (use_pointer_for_field): Return false if
17532         lower_send_shared_vars doesn't generate any copy-out code.
17534 2013-02-06  Tom de Vries  <tom@codesourcery.com>
17536         PR rtl-optimization/56131
17537         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17538         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
17539         of the label is NULL.  Add comment.
17541 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
17543         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17545         PR sanitizer/55374
17546         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17547         (STATIC_LIBTSAN_LIBS): Likewise.
17548         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17549         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17550         is defined, don't add anything else beyond that.
17551         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17552         (LINK_COMMAND_SPEC): Use them.
17554         PR tree-optimization/56205
17555         * tree-stdarg.c (check_all_va_list_escapes): Return true if
17556         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17557         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17559 2013-02-05  Richard Biener  <rguenther@suse.de>
17561         PR tree-optimization/53342
17562         PR tree-optimization/53185
17563         * tree-vectorizer.h (vect_check_strided_load): Remove.
17564         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17565         not disallow peeling for vectorized strided loads.
17566         (vect_check_strided_load): Make static and simplify.
17567         (vect_analyze_data_refs): Adjust.
17568         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17569         correctly when vectorizing strided loads.
17571 2013-02-05  Richard Biener  <rguenther@suse.de>
17573         * doc/install.texi: Refer to ISL, not PPL.
17575 2013-02-05  Jan Hubicka  <jh@suse.cz>
17577         PR tree-optimization/55789
17578         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17580 2013-02-05  Jan Hubicka  <jh@suse.cz>
17582         PR tree-optimization/55789
17583         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17584         the dead call anyway.
17586 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
17588         PR sanitizer/55374
17589         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17591 2013-02-04  Alexander Potapenko  <glider@google.com>
17592             Jack Howarth  <howarth@bromo.med.uc.edu>
17593             Jakub Jelinek  <jakub@redhat.com>
17595         PR sanitizer/55617
17596         * config/darwin.c (sort_ctor_records): Stabilized qsort
17597         on constructor priority by using original position.
17598         (finalize_ctors): New routine to sort constructors by
17599         priority before use in assemble_integer.
17600         (machopic_asm_out_constructor): Use finalize_ctors if needed.
17602 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
17604         PR libstdc++/54314
17605         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17606         about visibility on artificial decls.
17607         * config/sol2.c (solaris_assemble_visibility): Likewise.
17609 2013-02-04  Kai Tietz  <ktietz@redhat.com>
17611         PR target/56186
17612         * config/i386/i386.c (function_value_ms_64): Add additional valtype
17613         argument and improve checking of return-argument types for 16-byte
17614         modes.
17615         (ix86_function_value_1): Add additional valtype argument on call
17616         of function_value_64.
17617         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17618         handling infunction_value_64 function.
17620 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
17622         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17624 2013-02-04  Richard Biener  <rguenther@suse.de>
17626         PR tree-optimization/56188
17627         * tree-ssa-structalias.c (label_visit): Consider case with
17628         initially non-empty points-to set.
17629         (perform_var_substitution): Dump node mapping and clean up.
17631 2013-02-04  Richard Guenther  <rguenther@suse.de>
17633         PR lto/56168
17634         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17635         node prevail as last resort.
17636         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17637         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17639 2013-02-04  Richard Biener  <rguenther@suse.de>
17641         PR tree-optimization/56113
17642         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17643         Merge into ...
17644         (equiv_class_lookup_or_add): ... this.
17645         (label_visit): Adjust and fix error in previous patch.
17646         (perform_var_substitution): Adjust.
17648 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17650         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17651         * config/sh/sh4-300.md: Likewise.
17652         * config/sh/sh4a.md: Likewise.
17653         * config/sh/constraints.md: Likewise.
17654         * config/sh/sh.md: Likewise.
17655         * config/sh/netbsd-elf.h: Likewise.
17656         * config/sh/predicates.md: Likewise.
17657         * config/sh/sh-protos.h: Likewise.
17658         * config/sh/ushmedia.h: Likewise.
17659         * config/sh/linux.h: Likewise.
17660         * config/sh/sh.c: Likewise.
17661         * config/sh/superh.h: Likewise.
17662         * config/sh/elf.h: Likewise.
17663         * config/sh/sh4.md: Likewise.
17664         * config/sh/sh.h: Likewise.
17666 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17668         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17669         constraint to match_test floating_point_store_memory_operand().
17670         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17671         (base14_operand): New.
17672         (floating_point_store_memory_operand): New.
17673         (integer_store_memory_operand): Revise to use base14_operand and
17674         reg_plus_base_memory_operand.
17675         (move_dest_operand): Allow symbolic_memory_operands.
17676         (symbolic_memory_operand): Check for LO_SOM.
17677         (symbolic_operand): Change default case to break.
17678         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17679         CONST_DOUBLE values to be reloaded by putting them into memory when
17680         the destination is a floating point register.
17681         (movdf): Remove code to handle CONST_DOUBLE.
17682         (movsf): Likewise.
17683         (reload_indf_r1): New.
17684         (reload_insf_r1): New.
17685         Consistently use "Q" and "T" constraints with integer and floating
17686         point move instructions, respectively.
17687         (movdi): Remove FAIL.
17688         Change predicate for source operand unamed DImode move from
17689         general_operand to move_src_operand.
17690         (umulsidi3): Change predicate for destination operand to
17691         register_operand.
17692         Likewise for similar unamed patterns.
17693         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17694         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17695         (hppa_legitimize_address): Simplify mask calculation.
17696         (pa_emit_move_sequence): Revised handling of secondary reloads from
17697         REG+D addresses for floating point loads and stores.  Directly handle
17698         loading CONST0_RTX (mode) to a floating point register.
17699         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17700         to floating point registers.  Don't restrict secondary reloads to
17701         floating point registers to integer modes.  Revise some comments and
17702         cleanup some code.
17703         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17704         (pa_legitimate_address_p): New.
17705         (pa_legitimize_reload_address): New.
17706         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17707         (STRICT_REG_OK_FOR_BASE_P): New.
17708         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17709         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17711 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17712             Andrew Dixie  <andrewd@gentrack.com>
17714         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17715         flag set.
17717 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17719         * expmed.c (extract_bit_field_1): Pass the full width of the
17720         structure to get_best_reg_extraction_insn.
17722 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17724         PR target/54601
17725         * configure.ac (use_cxa_atexit): Add AIX.
17726         * configure: Regenerate.
17728         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17730 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17732         PR debug/54793
17733         * final.c (need_profile_function): New variable.
17734         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17735         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17736         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17737         notes, targetm.asm_out.function_prologue doesn't emit anything,
17738         HAVE_prologue and profiler should be emitted before prologue,
17739         set need_profile_function instead of emitting it.
17740         (final_scan_insn): If need_profile_function, emit
17741         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17742         NOTE_INSN_FUNCTION_BEG note.
17744 2013-02-01  Richard Henderson  <rth@redhat.com>
17746         * config/rs6000/rs6000.md (smulditi3): New.
17747         (umulditi3): New.
17749         * config/alpha/alpha.md (umulditi3): New.
17751 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17753         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17754         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17756 2013-02-01  Richard Biener  <rguenther@suse.de>
17758         PR tree-optimization/56113
17759         * tree-ssa-structalias.c (label_visit): Reduce work for
17760         single-predecessor nodes.
17762 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17764         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17765         range isn't testing for zero.
17767 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17769         PR middle-end/56113
17770         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17772 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17773             Nick Clifton  <nickc@redhat.com>
17775         * config/v850/constraints.md (Q): Define as a memory constraint.
17776         * config/v850/predicates.md (label_ref_operand): New predicate.
17777         (e3v5_shift_operand): New predicate.
17778         (ior_operator): New predicate.
17779         * config/v850/t-v850: Add e3v5 multilib.
17780         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17781         (v850_gen_movdi): Prototype.
17782         * config/v850/v850.c: Add support for e3v5 architecture.
17783         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17784         TARGET_V850E_UP.
17785         (construct_save_jarl): Add e3v5 long JARL support.
17786         (v850_adjust_insn_length): New function.  Adjust length of call
17787         insns when using e3v5 instructions.
17788         (v850_gen_movdi): New function: Generate instructions to move a
17789         DImode value.
17790         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17791         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17792         (TARGET_USE_FPU): Enable for e3v5.
17793         (CONST_OK_FOR_W): New macro.
17794         (ADJUST_INSN_LENGTH): Define.
17795         * config/v850/v850.md (UNSPEC_LOOP): Define.
17796         (attr cpu): Add v850e3v5.
17797         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17798         (movdi): New pattern.
17799         (movdi_internal): New pattern.
17800         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17801         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17802         (cstoresf4): Likewise.
17803         (cstoredf4): Likewise.
17804         (insv): New pattern.
17805         (rotlso3_a): New pattern.
17806         (rotlsi3_b): New pattern
17807         (rotlsi3_v850e3v5): New pattern.
17808         (doloop_begin): New pattern.
17809         (fix_loop_counter): New pattern.
17810         (doloop_end): New pattern.
17811         (branch_normal): Add e3v5 long branch support.
17812         (branch_invert): Likewise.
17813         (branch_z_normal): Likewise.
17814         (branch_z_invert): Likewise.
17815         (branch_nz_normal): Likewise.
17816         (branch_nz_invert): Likewise.
17817         (call_internal_short): Add e3v5 register-indirect JARL support.
17818         (call_internal_long): Likewise.
17819         (call_value_internal_short): Likewise.
17820         (call_value_internal_long): Likewise.
17821         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17822         (mloop): New option.
17823         * config.gcc: Add support for configuring v840e3v5 target.
17824         * doc/invoke.texi: Document new v850 specific command line options.
17826 2013-01-31  Paul Koning  <ni1d@arrl.net>
17828         PR debug/55059
17829         PR debug/54508
17830         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17831         children if parent is a class.
17832         (prune_unused_types_prune): Don't add DW_AT_declaration.
17834 2013-01-31  Richard Biener  <rguenther@suse.de>
17836         PR tree-optimization/56157
17837         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17838         match up operand with SLP child.
17840 2013-01-31  Jason Merrill  <jason@redhat.com>
17842         PR debug/54410
17843         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17844         parameters the first time.
17845         (gen_scheduled_generic_parms_dies): Check completeness here.
17847 2013-01-31  Richard Biener  <rguenther@suse.de>
17849         PR middle-end/53073
17850         * common.opt (faggressive-loop-optimizations): New flag,
17851         enabled by default.
17852         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17853         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17854         infer_loop_bounds_from_undefined by it.
17856 2013-01-31  Richard Biener  <rguenther@suse.de>
17858         PR tree-optimization/56150
17859         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17860         visit virtual operands.
17861         (find_uses_to_rename_bb): Likewise.
17863 2013-01-31  Richard Biener  <rguenther@suse.de>
17865         PR tree-optimization/56150
17866         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17867         mixed store non-store stmts.
17869 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17871         PR sanitizer/55374
17872         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17873         LIBASAN_EARLY_SPEC is defined.
17874         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17875         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17876         before %o.
17877         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17879         PR c++/55742
17880         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17881         invalid args instead of ICEing on it.
17882         (ix86_valid_target_attribute_tree): Return error_mark_node if
17883         ix86_valid_target_attribute_inner_p failed.
17884         (ix86_valid_target_attribute_p): Return false only if
17885         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17886         target("default") attribute.
17887         (sorted_attr_string): Change argument from const char * to tree,
17888         merge in all target attribute arguments rather than just one.
17889         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17890         instead of free.  Avoid using strcat.
17891         (ix86_mangle_function_version_assembler_name): Mangle
17892         target("default") as if no target attribute is present.  Adjust
17893         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17894         instead of xmalloc and XDELETEVEC instead of free.
17895         (ix86_function_versions): Don't return true if one of the decls
17896         doesn't have target attribute.  If they don't and one of the decls
17897         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17898         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17899         (ix86_supports_function_versions): Remove.
17900         (make_name): Fix up formatting.
17901         (make_dispatcher_decl): Remove resolver_name and its initialization.
17902         Avoid leaking memory.
17903         (is_function_default_version): Return true if there is
17904         target("default") attribute rather than no target attribute at all.
17905         (make_resolver_func): Avoid leaking memory.
17906         (ix86_generate_version_dispatcher_body): Likewise.
17907         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17908         * target.def (supports_function_versions): Remove.
17909         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17910         * doc/tm.texi: Regenerated.
17912 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17914         PR rtl-optimization/56144
17915         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17916         for values with side effects.
17918 2013-01-30  Richard Biener  <rguenther@suse.de>
17920         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17921         (sparseset_pop): Likewise.
17922         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17923         to be able to use quick_push in the worker loop.
17925 2013-01-30  Marek Polacek  <polacek@redhat.com>
17927         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17929 2013-01-30  Richard Biener  <rguenther@suse.de>
17931         PR lto/56147
17932         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17934 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17936         PR tree-optimization/56064
17937         * fixed-value.c (fixed_from_double_int): New function.
17938         * fixed-value.h (fixed_from_double_int): New prototype.
17939         (const_fixed_from_double_int): New static inline function.
17940         * fold-const.c (native_interpret_fixed): New static function.
17941         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17942         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17943         (native_encode_fixed): New static function.
17944         (native_encode_expr) <FIXED_CST>: Use it.
17945         (native_interpret_int): Move double_int worker code to...
17946         * double-int.c (double_int::from_buffer): ...this new static method.
17947         * double-int.h (double_int::from_buffer): Prototype it.
17949 2013-01-30  Richard Biener  <rguenther@suse.de>
17951         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17952         New pointer-map and obstack.
17953         (init_alias_vars): Allocate pointer-map and obstack.
17954         (delete_points_to_sets): Free them.
17955         (find_what_var_points_to): Cache result.
17956         (find_what_p_points_to): Adjust for changed interface of
17957         find_what_var_points_to.
17958         (compute_points_to_sets): Likewise.
17959         (ipa_pta_execute): Likewise.
17961 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17963         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17964         * configure: Regenerate.
17965         * config.in: Regenerate.
17966         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17967         #nobits/#progbits if supported.
17969 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17971         PR target/56121
17972         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17973         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17974         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17976 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17978         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17979         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17981 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17983         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17984         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17986 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17988         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17989         declaration.
17990         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17991         * config/arm/cortex-a7.md: New bypasses using
17992         arm_mac_accumulator_is_result.
17994 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17996         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17997         (cortex_a7_neon_mla): Likewise.
17998         (cortex_a7_fpfmad): New reservation.
17999         (cortex_a7_fpmacs): Use ffmas and update required units.
18000         (cortex_a7_fpmuld): Update required units and latency.
18001         (cortex_a7_fpmacd): Likewise.
18002         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18003         (cortex_a7_neon). Likewise.
18004         (bypass) Update participating units.
18006 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18008         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18009         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18010         from fmac to ffma.
18011         * config/arm/vfp11.md (vfp_farith): Use ffmas.
18012         (vfp_fmul): Use ffmad.
18013         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18014         (cortex_r4_fmacd): Use ffmad.
18015         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18016         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
18017         (cortex_a9_fmacd): Use ffmad.
18018         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18019         (cortex_a8_vfp_macd): Use ffmad.
18020         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18021         (cortex_a5_fpmacd): Use ffmad.
18022         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18023         (cortex_a15_vfp_macd): Use ffmad.
18024         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18026 2013-01-29  Jason Merrill  <jason@redhat.com>
18028         PR libstdc++/54314
18029         * varasm.c (default_assemble_visibility): Don't warn about
18030         visibility on artificial decls.
18032 2013-01-29  Richard Biener  <rguenther@suse.de>
18034         PR tree-optimization/56113
18035         * tree-ssa-structalias.c (equiv_class_lookup): Also return
18036         the bitmap leader.
18037         (label_visit): Free duplicate bitmaps and record the leader instead.
18038         (perform_var_substitution): Adjust.
18040 2013-01-29  Richard Biener  <rguenther@suse.de>
18042         PR tree-optimization/55270
18043         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18044         the CFG, schedule loops for fixup.
18046 2013-01-29  Nick Clifton  <nickc@redhat.com>
18048         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18049         SP_REG.
18051 2013-01-28  Leif Ekblad  <leif@rdos.net>
18053         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18054         * config/i386/i386.h (TARGET_RDOS): New macro.
18055         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18056         * config/i386/i386.c (ix86_option_override_internal): For 64bit
18057         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18058         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18059         DEFAULT_LARGE_SECTION_THRESHOLD.
18060         * config/i386/i386.md (R14_REG, R15_REG): New constants.
18061         * config/i386/rdos.h: New file.
18062         * config/i386/rdos64.h: New file.
18064 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
18066         PR other/54814
18067         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18068         TEST_HARD_REG_BIT.
18070 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18072         PR rtl-optimization/56117
18073         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18074         call cselib_lookup_from_insn on the MEM before calling
18075         add_insn_mem_dependence.
18077 2013-01-28  Richard Biener  <rguenther@suse.de>
18079         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18080         to a stmt that didn't have one.
18081         (copy_phis_for_bb): Likewise for PHI arguments.
18082         (copy_debug_stmt): Likewise for debug stmts.
18084 2013-01-28  Richard Biener  <rguenther@suse.de>
18086         PR tree-optimization/56034
18087         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18088         (partition_builtin_p): Adjust.
18089         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
18090         it is the last partition.
18091         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18092         up the vertex for the definition.
18093         (classify_partition): Classify whether a partition is a
18094         PKIND_REDUCTION, thus has uses outside of the loop.
18095         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18096         Merge all PKIND_REDUCTION partitions into the last partition.
18097         (tree_loop_distribution): Seed partitions from reductions as well.
18099 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18101         PR tree-optimization/56125
18102         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18103         pow(x,c) into sqrt(x) * powi(x, n/2) or
18104         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18105         optimizing for size.
18106         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18107         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18108         integer.
18110         PR tree-optimization/56094
18111         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18112         to UNKNOWN_LOCATION while gimplifying expr.
18114 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
18116         PR target/56114
18117         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18118         operand 0 in movabs insn template for -masm=intel asm alternative.
18119         (*movabs<mode>_2): Ditto for operand 1.
18121 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
18123         PR target/54663
18124         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18125         of microblaze-c.o
18127 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18129         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18130         tm_file.
18132 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18134         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18135         Undef to avoid warning.
18137 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
18139         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18140         * configure: Regenerate.
18142 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
18144         PR tree-optimization/56098
18145         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18146         for stmts with volatile ops.
18147         (cond_store_replacement): Don't optimize if assign has volatile ops.
18148         (cond_if_else_store_replacement_1): Don't optimize if either
18149         then_assign or else_assign have volatile ops.
18150         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18151         volatile ops.
18153 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18155         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18157 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18159         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18160         missing ':' in asm example.
18162 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
18164         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18165         entries into lane and laneq entries.
18166         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18167         Remove AdvSIMD scalar modes.
18168         (aarch64_sq<r>dmulh_laneq<mode>): New.
18169         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18170         modes.
18171         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18172         builtin implementations to relfect changes in RTL in aarch64-simd.md.
18173         * config/aarch64/iterators.md (VCOND): New.
18174         (VCONQ): New.
18176 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18178         PR target/54222
18179         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18180         Add NULL LIBNAME argument to existing definitions.
18181         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18182         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18183         * config/avr/avr.c (DEF_BUILTIN): Same.
18184         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18185         (avr_expand_builtin): Expand to a vanilla call if a libgcc
18186         implementation is available (DECL_ASSEMBLER_NAME is set).
18187         (avr_fold_absfx): New static function.
18188         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18189         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18190         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18191         AVR_BUILTIN_ABSLLK.
18192         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18193         (abshk, absk, abslk, absllk): Provide as static inline functions.
18195 2013-01-25  Marek Polacek  <polacek@redhat.com>
18197         PR tree-optimization/56035
18198         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18200 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
18202         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18203         (*movtf_internal_rex64): Add (!o,C) alternative
18204         (*movxf_internal_rex64): Ditto.
18205         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18207 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
18209         * doc/invoke.texi: fix typo.
18210         * doc/objc.texi: fix typo.
18212 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
18214         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18215         for the first two alternatives.
18217 2013-01-24  Diego Novillo  <dnovillo@google.com>
18219         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
18220         (ggc-zone.o): Remove.
18221         * configure.ac: Remove option --with-gc.
18222         * configure: Re-generate.
18223         * doc/install.texi: Remove documentation for --with-gc.
18224         * gengtype.c (write_enum_defn): Remove.  Update all users.
18225         (write_Types_process_field): Remove generation of gt_e_* argument.
18226         (output_type_enum): Remove.  Update all users.
18227         (write_enum_defn): Remove.  Update all users.
18228         (enum alloc_zone): Remove.  Update all users.
18229         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18230         * ggc-common.c (ggc_splay_alloc): Remove first argument.
18231         Update all callers.
18232         (struct ptr_data): Remove field TYPE.  Update all users.
18233         (gt_pch_note_object): Remove argument TYPE.  Update all users.
18234         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18235         Update all users.
18236         * ggc-none.c (ggc_alloc_typed_stat): Remove.
18237         (struct alloc_zone): Remove.
18238         (ggc_internal_alloc_zone_stat): Remove.
18239         (ggc_internal_cleared_alloc_zone_stat): Remove.
18240         * ggc-page.c (ggc_alloc_typed_stat): Remove.
18241         (ggc_pch_count_object): Remove last argument.  Update all users.
18242         (ggc_pch_alloc_object): Remove last argument.  Update all users.
18243         (struct alloc_zone): Remove.
18244         * ggc-zone.c: Remove.
18245         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
18246         (struct alloc_zone): Remove.
18247         (ggc_alloc_typed_stat): Remove.
18248         (ggc_alloc_typed): Remove.
18249         (ggc_splay_alloc): Remove first argument.
18250         (rtl_zone): Remove.  Update all users.
18251         (tree_zone): Remove.  Update all users.
18252         (tree_id_zone): Remove.  Update all users.
18253         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
18254         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
18255         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
18256         * tree-ssanames.c: Remove references to zone allocator in comments.
18258 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
18260         * config/avr/avr.c (avr_out_fract): Make register numbers that
18261         might be outside of source operand signed.
18263 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
18265         * config/i386/constraints.md (Yf): New constraint.
18266         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18267         of f constraint to conditionaly disable x87 register preferences.
18268         (*movdf_internal): Ditto.
18269         (*movsf_internal): Ditto.
18271 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
18273         PR inline-asm/55934
18274         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18275         that have operands with impossible constraints.
18276         Add a FIXME for a speed-up opportunity.
18277         * lra-constraints.c (process_alt_operands): Verify that a class
18278         selected from constraints on asms is valid for the operand mode.
18279         (curr_insn_transform): Remove incorrect comment.
18281 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
18283         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18284         TOC operand is a valid symbol ref in the constant pool.
18286 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18288         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18290 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
18292         PR target/54222
18293         * config/avr/stdfix.h: New file.
18294         * t-avr (stdfix-gcc.h): New rule to build it.
18295         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18297 2013-01-23  Kostya Serebryany  <kcc@google.com>
18299         * config/darwin.h: remove dependency on
18300         CoreFoundation (asan on Mac OS).
18302 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
18304         PR target/49069
18305         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18306         instead of cmpdi_operand for first comparison operand.
18307         Don't assert that comparison operands aren't both constants.
18309 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18311         * doc/install.texi (Downloading the Source): Update references to
18312         downloading separate components.
18314 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18316         * doc/extend.texi (__int128): Improve grammar.
18318 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
18320         PR target/56028
18321         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18322         alternative to (o,r).
18323         (*movdi_internal_rex64): Remove (!o,n) alternative.
18324         (DImode immediate->memory splitter): Remove.
18325         (DImode immediate->memory peephole2): Remove.
18326         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18327         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18328         alternative to (!o,*r).
18329         (*movtf_internal_sse): New pattern.
18330         (*movxf_internal_rex64): New pattern.
18331         (*movxf_internal): Disable for TARGET_64BIT.
18332         (*movdf_internal_rex64): Remove (!o,F) alternative.
18334 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
18336         PR middle-end/56074
18337         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18338         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18339         * tree-vect-loop-manip.c (find_loop_location): Also ignore
18340         stmt locations where LOCATION_LOCUS of the stmt location is
18341         UNKNOWN_LOCATION or BUILTINS_LOCATION.
18343         PR target/55686
18344         * config/i386/i386.md (UNSPEC_STOS): New.
18345         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18346         *strsetqi_1): Add UNSPEC_STOS.
18348 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
18350         PR c++/56067
18351         * doc/invoke.texi: Remove left over -Wsynth example.
18353 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18355         PR tree-optimization/56051
18356         * fold-const.c (fold_binary_loc): Don't fold
18357         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18358         a narrowing conversion, or widening conversion from signed
18359         to unsigned.
18361 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18363         PR rtl-optimization/56023
18364         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18365         dependent on debug instruction.
18367 2013-01-21  Martin Jambor  <mjambor@suse.cz>
18369         PR middle-end/56022
18370         * function.c (allocate_struct_function): Call
18371         invoke_set_current_function_hook earlier.
18373 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18375         * reload1.c (init_reload): Only initialize reload_obstack
18376         during the first call.
18378 2013-01-21  Marek Polacek  <polacek@redhat.com>
18380         * cfgloop.c (verify_loop_structure): Fix up grammar.
18382 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
18384         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18385         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18387 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18389         PR target/56058
18390         * config/arm/marvell-pj4.md: Update copyright year.
18391         Fix up use of alu to alu_reg and simple_alu_imm.
18393 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18395         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18397 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
18399         PR target/55433
18400         * lra-constraints.c (curr_insn_transform): Don't reuse original
18401         insn for secondary memory move when memory mode should be different.
18403 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18405         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18406         atomic_storedi_1): New patterns.
18408 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18410         btver2 pipeline descriptions.
18411         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18412         descriptions.
18413         * config/i386/i386.md (btver2_decode): New type attributes.
18414         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18415         type attributes.
18416         * config/i386/btver2.md: New file describing btver2 pipelines.
18418 2013-01-19  Andrew Pinski  <apinski@cavium.com>
18420         PR tree-optimization/52631
18421         * tree-ssa-sccvn (visit_use): Before looking up the original
18422         statement, try looking up the simplified expression.
18424 2013-01-19  Anthony Green  <green@moxielogic.com>
18426         * config/moxie/moxie.c (moxie_expand_prologue): Set
18427         current_function_static_stack_size.
18429 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18431         PR tree-optimization/56029
18432         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18433         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18435 2013-01-18  Sharad Singhai  <singhai@google.com>
18437         PR tree-optimization/55995
18438         * dumpfile.c (dump_loc): Print location only if available.
18439         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18441 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
18443         PR target/55433
18444         * lra-constraints.c (curr_insn_transform): Reuse original insn for
18445         secondary memory move.
18446         (inherit_reload_reg): Use rclass instead of cl for
18447         check_secondary_memory_needed_p.
18449 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18451         PR middle-end/56015
18452         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18453         the case where writing real complex part of target modifies op1.
18455 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
18457         * config/aarch64/aarch64-simd.md
18458         (aarch64_vcond_internal<mode>): Handle unordered cases.
18459         * config/aarch64/iterators.md (v_cmp_result): New.
18461 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
18462             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18464         * config/arm/marvell-pj4.md: New file.
18465         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18466         * config/arm/arm.md (generic_sched): Add marvell_pj4.
18467         (generic_vfp): Likewise.
18468         * config/arm/arm-cores.def: Add marvell-pj4.
18469         * config/arm/arm-tune.md: Regenerate.
18470         * config/arm/arm-tables.opt: Regenerate.
18471         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18472         * doc/invoke.texi: Document marvell-pj4.
18474 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
18476         * config/aarch64/arm_neon.h: Map scalar types to standard types.
18478 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18480         PR debug/54114
18481         PR debug/54402
18482         PR debug/49888
18483         * var-tracking.c (negative_power_of_two_p): New.
18484         (global_get_addr_cache, local_get_addr_cache): New.
18485         (get_addr_from_global_cache, get_addr_from_local_cache): New.
18486         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
18487         heading comment.
18488         (vt_stack_offset_p): Remove.
18489         (vt_canon_true_dep): Always canonicalize loc's address.
18490         (clobber_overlapping_mems): Make sure we have a MEM.
18491         (local_get_addr_clear_given_value): New.
18492         (val_reset): Clear local cached entries.
18493         (compute_bb_dataflow): Create and release the local cache.
18494         Disable duplicate MEMs clobbering.
18495         (emit_notes_in_bb): Clobber MEMs likewise.
18496         (vt_emit_notes): Create and release the local cache.
18497         (vt_initialize, vt_finalize): Create and release the global
18498         cache, respectively.
18499         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18501 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18503         PR libmudflap/53359
18504         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18505         not found in the symtab.
18507 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18509         PR debug/56006
18510         PR rtl-optimization/55547
18511         PR rtl-optimization/53827
18512         PR debug/53671
18513         PR debug/49888
18514         * alias.c (offset_overlap_p): New, factored out of...
18515         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
18516         the conservative special case for symbolic constants.  Don't
18517         adjust zero sizes on alignment.
18519 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
18521         PR rtl-optimization/52573
18522         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18523         REG_UNUSED for the same register.
18525 2013-01-17  Richard Biener  <rguenther@suse.de>
18526             Marek Polacek  <polacek@redhat.com>
18528         PR rtl-optimization/55833
18529         * loop-unswitch.c (unswitch_loops): Move loop verification...
18530         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
18531         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18532         Set it to true when we're removing a loop from hierarchy tree in
18533         an irreducible region.
18534         (fix_bb_placements): Adjust caller.
18535         (fix_loop_placements): Likewise.
18537 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
18539         * config/avr/builtins.def (DEF_BUILTIN): Factor out
18540         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18541         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18542         Remove ID.  Adjust comments.
18543         * config/avr/avr-c.c (avr_builtin_name): Remove.
18544         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18545         * config/avr/avr.c (avr_tolower): New static function.
18546         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
18547         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18548         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18549         default expansion.
18551 2013-01-17  Jan Hubicka  <jh@suse.cz>
18553         PR tree-optimization/55273
18554         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18556 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
18558         PR target/55981
18559         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18560         store through atomic_store<mode>_1.
18561         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18563 2013-01-17  Martin Jambor  <mjambor@suse.cz>
18565         PR tree-optimizations/55264
18566         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18567         for virtual methods.
18568         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18569         virtual methods before inlining is over.
18570         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18571         virtual functions.
18572         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18573         non-virtual.
18575 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
18577         PR rtl-optimization/56005
18578         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18579         pending reads for prefetch.
18581 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
18583         * config/aarch64/aarch64.md
18584         (*cstoresi_neg_uxtw): New pattern.
18585         (*cmovsi_insn_uxtw): New pattern.
18586         (*<optab>si3_uxtw): New pattern.
18587         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18588         (*<optab>si3_insn_uxtw): New pattern.
18589         (*bswapsi2_uxtw): New pattern.
18591 2013-01-16  Richard Biener  <rguenther@suse.de>
18593         * tree-inline.c (tree_function_versioning): Remove set but
18594         never used variable.
18596 2013-01-16  Richard Biener  <rguenther@suse.de>
18598         PR tree-optimization/55964
18599         * tree-flow.h (rename_variables_in_loop): Remove.
18600         (rename_variables_in_bb): Likewise.
18601         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18602         (copy_loop_before): Adjust and delete update-ssa status.
18603         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18604         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
18605         (rename_variables_in_loop): Remove.
18606         (slpeel_update_phis_for_duplicate_loop): Likewise.
18607         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18608         use available cfg machinery instead of duplicating it.
18609         Update PHI nodes and perform poor-mans SSA update here.
18610         (slpeel_tree_peel_loop_to_edge): Adjust.
18612 2013-01-16  Richard Biener  <rguenther@suse.de>
18614         PR tree-optimization/54767
18615         PR tree-optimization/53465
18616         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18617         (vrp_visit_phi_node): For PHI arguments coming via backedges
18618         drop all symbolical range information.
18619         (execute_vrp): Compute backedges.
18621 2013-01-16  Richard Biener  <rguenther@suse.de>
18623         * doc/install.texi: Update CLooG and ISL requirements to
18624         0.18.0 and 0.11.1.
18626 2013-01-16  Christian Bruel  <christian.bruel@st.com>
18628         PR target/55301
18629         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18630         (broken_move): Handle UNSPECV_SP_SWITCH_B.
18631         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18633 2013-01-16  DJ Delorie  <dj@redhat.com>
18635         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18636         (UNSPECV_SP_SWITCH_E): New.
18637         (sp_switch_1): Change to an unspec.
18638         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18639         replace $r15.
18641 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18643         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18644         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18645         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18646         (expand_mem_thread_fence): Ditto.
18647         (expand_mem_signal_fence): Ditto.
18648         (expand_atomic_load): Ditto.
18649         (expand_atomic_store): Ditto.
18651 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18653         PR rtl-optimization/55547
18654         PR rtl-optimization/53827
18655         PR debug/53671
18656         PR debug/49888
18657         * alias.c (memrefs_conflict_p): Set sizes to negative after
18658         AND adjustments.
18660 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18662         PR target/55940
18663         * function.c (thread_prologue_and_epilogue_insns): Always
18664         add crtl->drap_reg to set_up_by_prologue.set, even if
18665         stack_realign_drap is false.
18667 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18669         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18670         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18671         *call): Fix indention.
18673 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18675         PR target/55876
18676         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18677         Update comment.
18679 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18681         PR rtl-optimization/55153
18682         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18684 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18686         PR tree-optimization/55920
18687         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18688         accesses as grp_to_be_debug_replaced.
18690 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18692         PR tree-optimization/55920
18693         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18694         there is non-useless type conversion needed from debug rhs to lhs,
18695         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18697 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18698             Mikael Pettersson  <mikpe@it.uu.se>
18700         PR target/43961
18701         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18702         Thumb.
18703         (ASM_OUTPUT_CASE_LABEL): Remove.
18704         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18705         * final.c (shorten_branches): Update alignment of labels before
18706         jump tables if CASE_VECTOR_SHORTEN_MODE.
18708 2013-01-15  Richard Biener  <rguenther@suse.de>
18710         PR bootstrap/55961
18711         * system.h: Do not include gmp.h for building host tools.
18713 2013-01-15  Richard Biener  <rguenther@suse.de>
18715         PR middle-end/55882
18716         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18717         account for bitpos when computing alignment.
18719 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18721         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18722         (ix86_target_macros_internal): Likewise.
18724         * config/i386/i386.c (m_CORE2I7): Removed.
18725         (m_CORE_HASWELL): New macro.
18726         (m_CORE_ALL): Likewise.
18727         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18728         (initial_ix86_arch_features): Likewise.
18729         (processor_target_table): Initializations for Core avx2.
18730         (cpu_names): New names "core-avx2".
18731         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18732         PROCESSOR_CORE_HASWELL.
18733         (ix86_issue_rate): New case.
18734         (ia32_multipass_dfa_lookahead): Likewise.
18735         (ix86_sched_init_global): Likewise.
18737         * config/i386/i386.h (TARGET_HASWELL): New macro.
18738         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18739         (processor_type): New PROCESSOR_HASWELL.
18741 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18743         PR tree-optimization/55955
18744         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18745         *SHIFT_EXPR and *ROTATE_EXPR codes.
18747         PR tree-optimization/48766
18748         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18749         -ftrapv disable -fwrapv.
18751 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18753         PR target/55974
18754         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18755         etc. to 1 and not to __flash.
18756         Use LL suffix for __INT24_MAX__ with -mint8.
18757         Use ULL suffix for __UINT24_MAX__ with -mint8.
18759 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18761         * config/avr/avr-arch.h
18762         (struct base_arch_s): Use typedef avr_arch_t instead.
18763         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18764         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18765         * config/avr/avr.c: Same.
18766         * config/avr/avr-devices.c: Same.
18767         * config/avr/driver-avr.c: Same.
18768         * config/avr/gen-avr-mmcu-texi.c: Same.
18769         * config/avr/avr-mcus.def: Adjust comment.
18771 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18773         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18774         * config/aarch64/iterators.md (VALLDI): New.
18776 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18777             Andi Kleen  <ak@linux.intel.com>
18779         PR target/55948
18780         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18781         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18782         memmodel flag.
18784 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18786         * config/avr/avr-stdint.h: Remove trailing blanks.
18787         * config/avr/avr-log.h: Same.
18788         * config/avr/avr-arch.h: Same.
18789         * config/avr/avr-devices.c: Same.
18790         * config/avr/avr-dimode.md: Same.
18791         * config/avr/predicates.md: Same.
18792         * config/avr/avr-c.c: Same.  And fix typo.
18794         * config/avr/avr-protos.h: Same.  And:
18795         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18796         (init_cumulative_args): Rename to avr_init_cumulative_args.
18797         (expand_prologue): Rename to avr_expand_prologue.
18798         (expand_epilogue): Rename to avr_expand_epilogue.
18799         (adjust_insn_length): Rename to avr_adjust_insn_length.
18800         (notice_update_cc): Rename to avr_notice_update_cc.
18801         (final_prescan_insn): Rename to avr_final_prescan_insn.
18802         * config/avr/avr.c: Same.
18803         * config/avr/avr.h: Same.
18804         * config/avr/avr.md: Remove trailing blanks.
18805         (prologue): Use avr_expand_prologue.
18806         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18808 2013-01-14  Richard Biener  <rguenther@suse.de>
18810         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18811         verify_location, collect_subblocks): New functions.
18812         (verify_gimple_in_cfg): Verify that locations only reference
18813         BLOCKs in the functions BLOCK tree.
18815 2013-01-14  Richard Biener  <rguenther@suse.de>
18817         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18818         PHI argument.
18819         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18820         unshare reference.
18821         (insert_out_of_ssa_copy_on_edge): Likewise.
18822         (rewrite_close_phi_out_of_ssa): Likewise.
18823         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18824         debug expressions.
18825         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18826         propagated constants.
18827         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18828         can not be shared.
18830 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18832         * config/avr/avr-modes.def: Add GPL copyright notice.
18834 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18836         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18837         MEMMODEL_MASK to determine memory model.
18838         (atomic_store<mode>): Ditto from operands[2].
18839         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18841 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18843         PR fortran/55935
18844         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18845         (fold_gimple_assign): Don't call unshare_expr here.
18846         (fold_ctor_reference): Call unshare_expr.
18848 2013-01-13  Terry Guo  <terry.guo@arm.com>
18850         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18851         * doc/fragments.texi: Document MULTILIB_REUSE.
18852         * gcc.c (multilib_reuse): New internal spec.
18853         (set_multilib_dir): Also search multilib from multilib_reuse.
18854         * genmultilib (tmpmultilib3): Refactor code.
18855         (tmpmultilib4): Ditto.
18856         (multilib_reuse): New multilib argument.
18858 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18860         * Makefile.in: Update copyright.
18862 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18864         PR middle-end/55890
18865         * calls.c (expand_call): Check if arg_nr is valid.
18867 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18869         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18870         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18871         documentation.  Add missing '__' in front of
18872         __builtin_ia32_packssdw256.
18874 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18876         PR target/55719
18877         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18878         NO_REGS for larl operands.
18879         (s390_reload_larl_operand): Use s390_load_address instead of
18880         emit_move_insn.
18882 2013-01-11  Richard Biener  <rguenther@suse.de>
18884         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18885         (verify_node_sharing): ... here.
18886         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18888 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18890         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18891         Substitute TREECHECKING.
18892         * configure: Regenerate.
18893         * Makefile.in (TREECHECKING): New.
18895 2013-01-11  Richard Guenther  <rguenther@suse.de>
18897         PR tree-optimization/44061
18898         * tree-vrp.c (extract_range_basic): Compute zero as
18899         value-range for __builtin_constant_p of function parameters.
18901 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18903         Update copyright years.
18905 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18907         PR rtl-optimization/55672
18908         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18909         const to be eliminable.
18911 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18913         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18914         * configure: Regenerate.
18916 2013-01-10  Richard Biener  <rguenther@suse.de>
18918         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18920 2013-01-10  Richard Biener  <rguenther@suse.de>
18922         PR bootstrap/55792
18923         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18924         locations for virtual PHI arguments.
18925         (rewrite_update_phi_arguments): Likewise.
18927 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18929         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18930         on to assembler.
18932 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18934         PR tree-optimization/55921
18935         * tree-complex.c (expand_complex_asm): New function.
18936         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18938 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18940         PR target/55718
18941         * config/s390/s390.c (s390_symref_operand_p)
18942         (s390_loadrelative_operand_p): Merge the two functions.
18943         (s390_check_qrst_address, print_operand_address): Add parameters
18944         to s390_loadrelative_operand_p invokation.
18945         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18946         (s390_reload_larl_operand, s390_secondary_reload): Use
18947         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18948         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18950 2013-01-09  Mike Stump  <mikestump@comcast.net>
18952         * dse.c (record_store): Remove unnecessary assert.
18954 2013-01-09  Jan Hubicka  <jh@suse.cz>
18956         PR tree-optimization/55569
18957         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18958         * cfgloop.h (scale_loop_profile): Likewise.
18960 2013-01-09  Jan Hubicka  <jh@suse.cz>
18962         PR lto/45375
18963         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18964         functions.
18965         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18967 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18969         PR middle-end/55114
18970         * expr.h (maybe_emit_group_store): Declare.
18971         * expr.c (maybe_emit_group_store): New function.
18972         * builtins.c (expand_builtin_int_roundingfn): Call it.
18973         (expand_builtin_int_roundingfn_2): Likewise.
18975 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18977         PR rtl-optimization/55829
18978         * lra-constraints.c (match_reload): Add code for absent output.
18979         (curr_insn_transform): Add code for reloads of matched inputs
18980         without output.
18982 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18984         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18985         attribute of movddup insn to DF.
18986         (*vec_interleave_lowv2df): Ditto.
18987         (vec_dupv2df): Ditto.
18989 2013-01-09  Jan Hubicka  <jh@suse.cz>
18991         PR tree-optimiation/55875
18992         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18993         EVERY_ITERATION parameter.
18994         (number_of_iterations_exit): Check if exit is executed every iteration.
18995         (idx_infer_loop_bounds): Similarly here.
18996         (n_of_executions_at_most): Simplify
18997         to only test for cases where statement is dominated by the
18998         particular bound; handle correctly the "postdominance" test.
18999         (scev_probably_wraps_p): Use max loop iterations info
19000         as a global bound first.
19002 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
19003             Nick Clifton  <nickc@redhat.com>
19005         * config/v850/v850.md (cbranchsf4): New pattern.
19006         (cstoresf4): New pattern.
19007         (cbranchdf4): New pattern.
19008         (cstoredf4): New pattern.
19009         (movsicc): Disallow floating point comparisons.
19010         (cmpsf_le_insn): Fix order of operators.
19011         (cmpsf_lt_insn): Likewise.
19012         (cmpsf_eq_insn): Likewise.
19013         (cmpdf_le_insn): Likewise.
19014         (cmpdf_lt_insn): Likewise.
19015         (cmpdf_eq_insn): Likewise.
19016         (cmpsf_ge_insn): Use LE comparison.
19017         (cmpdf_ge_insn): Likewise.
19018         (cmpsf_gt_insn): Use LT comparison.
19019         (cmpdf_gt_insn): Likewise.
19020         (cmpsf_ne_insn): Delete pattern.
19021         (cmpdf_ne_insn): Delete pattern.
19022         * config/v850/v850.c (v850_gen_float_compare): Use
19023         gen_cmpdf_eq_insn for NE comparison.
19024         (v850_float_z_comparison_operator)
19025         (v850_float_nz_comparison_operator): Move from here ...
19026         * config/v850/predicates.md: ... to here.  Move GT and GE
19027         comparisons into v850_float_z_comparison_operator.
19028         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19029         Delete prototype.
19030         (v850_float_nz_comparison_operator): Likewise.
19032 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19034         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19035         with calls to gen_insvsi/gen_insvdi.
19037 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19039         * config/i386/i386.c (initial_ix86_tune_features): Set up
19040         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19042 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
19043             Jakub Jelinek  <jakub@redhat.com>
19045         PR tree-optimization/48189
19046         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19047         If nitercst is 0, don't predict the exit edge.
19049 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19051         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19052         in asm_fprintf with reg_names.
19053         (aarch64_print_operand_address): Likewise.
19054         (aarch64_return_addr): Likewise.
19055         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19057 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19059         * config/pa/pa.h (VAL_U6_BITS_P): Define.
19060         (INT_U6_BITS): Likewise.
19061         * config/pa/predicates.md (uint6_operand): New predicate.
19062         (shift5_operand, shift6_operand): Likewise.
19063         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19064         arith32_operand.
19065         (lshrdi3): Use shift6_operand.
19066         (shrpsi4, shrpdi4): New insn patterns.
19067         (extzv): Delete expander.
19068         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
19069         predicates in unamed zero extract patterns.  Tighten common constraint.
19070         (extv): Delete expander.
19071         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
19072         predicates in unamed sign extract patterns.  Tighten common constraint.
19073         (insv): Delete expander.
19074         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
19075         predicates in unamed insert patterns.  Tighten common constraint.
19076         Change uint32_operand predicate to uint6_operand predicate in unamed
19077         DImode pattern to insert constant values of type 1...1xxxx.
19079 2013-01-04  Jan Hubicka  <jh@suse.cz>
19081         PR tree-optimization/55823
19082         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19083         issue.
19085 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19086             Uros Bizjak  <ubizjak@gmail.com>
19088         PR rtl-optimization/55845
19089         * df-problems.c (can_move_insns_across): Stop scanning at
19090         volatile_insn_p source instruction or give up if
19091         across_from .. across_to range contains any volatile_insn_p
19092         instructions.
19094 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19096         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19097         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19098         Declare.
19099         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19100         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19102 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19104         PR fortran/55341
19105         * asan.c (asan_clear_shadow): New function.
19106         (asan_emit_stack_protection): Use it.
19108 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19110         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19111         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19112         with tab instead of space.
19114 2013-01-08  Nick Clifton  <nickc@redhat.com>
19116         * config/rl78/rl78.c (rl78_expand_prologue): Always select
19117         register bank 0 at the start of an interrupt handler.
19118         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19119         MDBH registers.
19121 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19123         * config/aarch64/aarch64-simd.md
19124         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19125         (aarch64_simd_bsl): Likewise.
19126         (aarch64_vcond_internal<mode>): Likewise.
19127         (vcond<mode><mode>): Likewise.
19128         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19129         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19131 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19133         * config/aarch64/aarch64-builtins.c
19134         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19136 2013-01-08  Martin Jambor  <mjambor@suse.cz>
19138         PR debug/55579
19139         * tree-sra.c (analyze_access_subtree): Return true also after
19140         potentially creating a debug-only replacement.
19142 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19144         PR middle-end/55890
19145         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19147         PR tree-optimization/54120
19148         * tree-vrp.c (range_fits_type_p): Don't allow
19149         src_precision < precision from signed vr to unsigned_p
19150         if vr->min or vr->max is negative.
19151         (simplify_float_conversion_using_ranges): Test can_float_p
19152         against CODE_FOR_nothing.
19154 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19155             Richard Biener  <rguenther@suse.de>
19157         PR middle-end/55851
19158         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19159         types instead of just INTEGER_TYPE types.
19161 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
19163         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19164         TF_SIZE): Define.
19166 2013-01-07  Steve Ellcey  <sellcey@mips.com>
19168         PR target/42661
19169         * config/mips/mips.opt: Change mad to mmad to match documentation.
19171 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19173         PR target/55897
19174         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19175         .progmemx.data now.
19177 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19179         PR target/55897
19180         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19181         (avr_addrspace_t): Add .section_name field.
19182         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19183         array size.
19184         (avr_addrspace): Same.  Initialize .section_name.  Remove last
19185         NULL entry.  Put __memx into .progmemx.data.
19186         (progmem_section_prefix): Remove.
19187         (avr_asm_init_sections): No need to initialize progmem_section.
19188         (avr_asm_named_section): Use avr_addrspace[].section_name to get
19189         section name prefix.
19190         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
19191         retrieve the progmem section.
19192         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19193         boundary to run over avr_addrspace[].
19194         (avr_register_target_pragmas): Ditto.
19196 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
19198         * varasm.c (output_constant_def_contents): For asan_protect_global
19199         protected strings, adjust DECL_ALIGN if needed, before testing for
19200         anchored symbols.
19201         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19202         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
19203         normal decls.
19204         (output_object_block): For asan protected decls, emit asan padding
19205         after their contents.
19206         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19207         (asan_finish_file): Test it here instead.
19209 2013-01-07  Nick Clifton  <nickc@redhat.com>
19210             Matthias Klose  <doko@debian.org>
19211             Doug Kwan  <dougkwan@google.com>
19212             H.J. Lu  <hongjiu.lu@intel.com>
19214         PR driver/55470
19215         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19217         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19219         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19221         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19223         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19225 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19227         PR target/54461
19228         * doc/install.texi (Cross-Compiler-Specific Options): Document
19229         --with-avrlibc.
19231 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
19233         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19234         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19235         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19236         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19237         vqmovun_high_s64): Fix source operand number and update copyright.
19239 2013-01-07  Richard Biener  <rguenther@suse.de>
19241         PR middle-end/55890
19242         * gimple.h (gimple_call_builtin_p): New overload.
19243         * gimple.c (validate_call): New function.
19244         (gimple_call_builtin_p): Likewise.
19245         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19246         Use gimple_call_builtin_p.
19247         (find_func_clobbers): Likewise.
19248         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19249         (strlen_optimize_stmt): Likewise.
19251 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19253         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19254         (vld1q_dup_*): Likewise.
19255         (vld1_*): Likewise.
19256         (vld1q_*): Likewise.
19257         (vld1_lane_*): Likewise.
19258         (vld1q_lane_*): Likewise.
19260 2013-01-07  Richard Biener  <rguenther@suse.de>
19262         * lto-streamer.h (LTO_minor_version): Bump to 2.
19264 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19266         * config/aarch64/aarch64-protos.h
19267         (aarch64_const_double_zero_rtx_p): Rename to...
19268         (aarch64_float_const_zero_rtx_p): ...this.
19269         (aarch64_float_const_representable_p): New.
19270         (aarch64_output_simd_mov_immediate): Likewise.
19271         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19272         move immediate case.
19273         * config/aarch64/aarch64.c
19274         (aarch64_const_double_zero_rtx_p): Rename to...
19275         (aarch64_float_const_zero_rtx_p): ...this.
19276         (aarch64_print_operand): Allow printing of new constants.
19277         (aarch64_valid_floating_const): New.
19278         (aarch64_legitimate_constant_p): Check for valid floating-point
19279         constants.
19280         (aarch64_simd_valid_immediate): Likewise.
19281         (aarch64_vect_float_const_representable_p): New.
19282         (aarch64_float_const_representable_p): Likewise.
19283         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19284         (aarch64_output_simd_mov_immediate): New.
19285         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19286         (*movdf_aarch64): Likewise.
19287         * config/aarch64/constraints.md (Ufc): New.
19288         (Y): call aarch64_float_const_zero_rtx.
19289         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19291 2013-01-07  Richard Biener  <rguenther@suse.de>
19293         PR tree-optimization/55888
19294         PR tree-optimization/55862
19295         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19296         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19297         not if it is contained therein.
19299 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19301         * config/avr/t-avr: Typo.
19303 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19305         PR55243
19306         * config/avr/t-avr: Don't automatically rebuild
19307         $(srcdir)/config/avr/t-multilib
19308         $(srcdir)/config/avr/avr-tables.opt
19309         $(srcdir)/doc/avr-mmcu.texi
19310         (avr-mcus): New phony target to build them on request.
19311         (s-avr-mlib, s-avr-mmcu-texi): Remove.
19312         * avr/avr-mcus.def: Adjust comments.
19314 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
19316         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19318 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19320         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19322 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19324         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19326 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
19328         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19329         to generate profiling.
19330         * config/rs6000/aix64.h (LIB_SPEC): Same.
19332 2013-01-04  Andrew Pinski  <apinski@cavium.com>
19334         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19335         New function.
19336         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19338 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
19340         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19341         unconditionally.
19342         (ix86_expand_move): Ditto.
19343         (ix86_zero_extend_to_Pmode): Ditto.
19344         (ix86_expand_call): Ditto.
19345         (ix86_expand_special_args_builtin): Ditto.
19346         (ix86_expand_builtin): Ditto.
19348 2013-01-04  Richard Biener  <rguenther@suse.de>
19350         PR tree-optimization/55862
19351         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19352         translating them through PHI nodes.
19354 2013-01-04  Martin Jambor  <mjambor@suse.cz>
19356         PR tree-optimization/55755
19357         * tree-sra.c (sra_modify_assign): Do not check that an access has no
19358         children when trying to avoid producing a VIEW_CONVERT_EXPR.
19360 2013-01-04  Marek Polacek  <polacek@redhat.com>
19362         PR middle-end/55859
19363         * opts.c (default_options_optimization): Clarify error message.
19365 2013-01-04  Richard Biener  <rguenther@suse.de>
19367         PR middle-end/55863
19368         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19369         reassociation.
19371 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19373         PR target/53789
19374         * config/pa/pa.md (movsi): Revert previous change.
19375         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19376         references.
19378 2013-01-03  Richard Henderson  <rth@redhat.com>
19380         * config/i386/i386.c (ix86_expand_move): Always assign to op1
19381         after eliminating TLS symbols.
19383 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
19385         PR bootstrap/50167
19386         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19387         * graphite-poly.c (debug_gmp_value): Likewise.
19389 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
19391         PR target/55712
19392         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19393         selected code model, define __code_mode_small__, __code_model_medium__,
19394         __code_model_large__, __code_model_32__ or __code_model_kernel__.
19395         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19396         xchg temporary register with %k.  Declare temporary register as
19397         early clobbered.
19398         [__x86_64__]: For medium and large code models, preserve %rbx register.
19400 2013-01-03  Richard Biener  <rguenther@suse.de>
19402         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19403         (dump_subscript): Adjust.
19404         (finalize_ddr_dependent): Do not dump redundant info.
19405         (analyze_siv_subscript): Adjust.
19406         (subscript_dependence_tester): Likewise.
19407         (compute_affine_dependence): Likewise.
19409 2013-01-03  Richard Biener  <rguenther@suse.de>
19411         Revert
19412         2013-01-03  Richard Biener  <rguenther@suse.de>
19414         PR tree-optimization/55857
19415         * tree-vect-stmts.c (vectorizable_load): Do not setup
19416         re-alignment for invariant loads.
19418         2013-01-02  Richard Biener  <rguenther@suse.de>
19420         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19421         invariant load do not generate a vector load from the scalar location.
19423 2013-01-03  Richard Biener  <rguenther@suse.de>
19425         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19426         for not vectorizing.
19427         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19428         not build INDIRECT_REFs, call get_name once only.
19429         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
19430         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19432 2013-01-03  Richard Biener  <rguenther@suse.de>
19434         PR tree-optimization/55857
19435         * tree-vect-stmts.c (vectorizable_load): Do not setup
19436         re-alignment for invariant loads.
19438 2013-01-03  Richard Biener  <rguenther@suse.de>
19440         PR lto/55848
19441         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19442         prefer a built-in decl.
19444 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19446         * gcc.c (process_command): Update copyright notice dates.
19447         * gcov.c (print_version): Likewise.
19448         * gcov-dump.c (print_version): Likewise.
19450         PR rtl-optimization/55838
19451         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19452         iv0.step, iv1.step and step.
19454 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19455             Marc Glisse  <marc.glisse@inria.fr>
19457         PR tree-optimization/55832
19458         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19459         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19460         integer_{one,zero}_node.
19462 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19464         PR debug/54402
19465         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19466         * var-tracking.c (reverse_op): Don't add reverse ops to
19467         VALUEs that have already
19468         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19470 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
19472         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19474 2013-01-02  Teresa Johnson  <tejohnson@google.com>
19476         * dumpfile.c (dump_loc): Print filename with location.
19477         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19478         new location_t parameter to emit complete unroll message with
19479         new dump framework.
19480         (canonicalize_loop_induction_variables): Compute loops location
19481         and pass to try_unroll_loop_completely.
19482         * loop-unroll.c (report_unroll_peel): New function.
19483         (peel_loops_completely): Use new dump format with location
19484         for main dumpfile message, and invoke report_unroll_peel on success.
19485         (decide_unrolling_and_peeling): Ditto.
19486         (decide_peel_once_rolling): Remove old dumpfile message subsumed
19487         by report_unroll_peel.
19488         (decide_peel_completely): Ditto.
19489         (decide_unroll_constant_iterations): Ditto.
19490         (decide_unroll_runtime_iterations): Ditto.
19491         (decide_peel_simple): Ditto.
19492         (decide_unroll_stupid): Ditto.
19493         * cfgloop.c (get_loop_location): New function.
19494         * cfgloop.h (get_loop_location): Declare.
19496 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19498         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19499         NULL.
19501 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19503         PR middle-end/55198
19504         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19505         BLKmode objects when EXPAND_MEMORY is specified.
19507 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19509         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19510         in loop predicate.
19511         (fold_builtin_cpu): Do not share cpu model decls across statements.
19513 2013-01-02  Jason Merrill  <jason@redhat.com>
19515         PR c++/55804
19516         * tree.c (build_array_type_1): Revert earlier change.
19518 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
19520         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19521         "cortex-a57".
19522         * config/aarch64/aarch64-tune.md: Re-generate.
19524 2013-01-02  Richard Biener  <rguenther@suse.de>
19526         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19527         invariant load do not generate a vector load from the scalar location.
19529 2013-01-02  Richard Biener  <rguenther@suse.de>
19531         PR bootstrap/55784
19532         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19533         * configure: Regenerate.
19535 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19537         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19538         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19539         (expand_builtin_int_roundingfn_2): Keep the original target around
19540         for the fallback case.
19542 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19544         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19545         to be clear for sign changes.
19547 2013-01-01  Jan Hubicka  <jh@suse.cz>
19549         * ipa-inline-analysis.c: Fix formatting.
19551 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
19553         PR tree-optimization/55831
19554         * tree-vect-loop.c (get_initial_def_for_induction): Use
19555         gsi_after_labels instead of gsi_start_bb.
19557 Copyright (C) 2013 Free Software Foundation, Inc.
19559 Copying and distribution of this file, with or without modification,
19560 are permitted in any medium without royalty provided the copyright
19561 notice and this notice are preserved.