2013-09-03 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blobe107860ae1a2279c62a3b56051865322e5996d51
1 2013-09-03  Richard Biener  <rguenther@suse.de>
3         * tree-affine.c (add_elt_to_tree): Fix association issue,
4         avoid useless converts and make sure to always return a
5         properly typed result.
7 2013-09-03  Richard Biener  <rguenther@suse.de>
9         PR middle-end/57656
10         * fold-const.c (negate_expr_p): Fix division case.
11         (negate_expr): Likewise.
13 2013-09-03  Richard Biener  <rguenther@suse.de>
15         PR lto/58285
16         * tree-streamer-out.c: Include tm.h.
17         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
19 2013-09-03  Jan Hubicka  <jh@suse.cz>
21         * tree-profile.c (tree_profiling): Cleanup CFG when done.
23 2013-09-03  Alan Modra  <amodra@gmail.com>
25         * config.gcc (powerpc*-*-linux*): Add support for little-endian
26         multilibs to big-endian target and vice versa.
27         * config/rs6000/t-linux64: Use := assignment on all vars.
28         (MULTILIB_EXTRA_OPTS): Remove fPIC.
29         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
30         * config/rs6000/t-linux64le: New file.
31         * config/rs6000/t-linux64bele: New file.
32         * config/rs6000/t-linux64lebe: New file.
34 2013-09-02  Jan Hubicka  <jh@suse.cz>
36         * ipa-inline-transform.c (inline_transform): Do not
37         optimize_inline_calls when not optimizing.
39 2013-09-02  Jan Hubicka  <jh@suse.cz>
41         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
42         duplicated nodes for assembler names.
43         * symtab.c (symtab_unregister_node): Do not attempt to unlink
44         hard registers from assembler name hash.
46 2013-09-02  Jan Hubicka  <jh@suse.cz>
48         * ipa-split.c (execute_split_functions): Split externally visible
49         functions called once.
51 2013-09-02  Martin Jambor  <mjambor@suse.cz>
53         PR ipa/58106
54         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
55         linked list.  When finding the correct duplicate, also consider also
56         the caller in additon to its inlined_to node.
58 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
60         * config/aarch64/aarch64-simd-builtins.def
61         (dup_lane_scalar): Remove.
62         * config/aarch64/aarch64-simd.md
63         (aarch64_simd_dup): Add 'w->w' alternative.
64         (aarch64_dup_lane<mode>): Allow for VALL.
65         (aarch64_dup_lane_scalar<mode>): Remove.
66         (aarch64_dup_lane_<vswap_width_name><mode>): New.
67         (aarch64_get_lane_signed<mode>): Add w->w altenative.
68         (aarch64_get_lane_unsigned<mode>): Likewise.
69         (aarch64_get_lane<mode>): Likewise.
70         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
71         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
72         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
73         (VCON): Change container of V2SF.
74         (vswap_width_name): Likewise.
75         * config/aarch64/arm_neon.h
76         (__aarch64_vdup_lane_any): New.
77         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
78         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
79         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
81 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
83         PR middle-end/56382
84         * expr.c (emit_move_complex): Do not move complex FP values as parts if
85         the source or the destination is a single hard register.
87 2013-09-02  Richard Biener  <rguenther@suse.de>
89         PR middle-end/57511
90         * tree-scalar-evolution.c (instantiate_scev_name): Allow
91         non-linear SCEVs.
93 2013-09-02  Richard Biener  <rguenther@suse.de>
95         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
96         arithmetic to sizetype.
98 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
100         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
101         Find auto-increment use both before and after candidate.
103 2013-09-02  Marek Polacek  <polacek@redhat.com>
105         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
107 2013-09-01  Jan Hubicka  <jh@suse.cz>
109         * Makefile.in: Add ipa-profile.o
110         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
111         * cgraph.c (struct cgraph_propagate_frequency_data,
112         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
113         ipa-profile.c; replace cgraph_ by ipa_ prefix.
114         * cgraph.h (cgraph_propagate_frequency): Remove.
115         * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
116         (inline_update_callee_summaries): Update.
117         * ipa-profile.c: New file.
118         * ipa-utils.h (ipa_propagate_frequency): Declare.
119         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
120         data-streamer.h, value-prof.h
121         (symtab_remove_unreachable_nodes): Update profile.
122         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
123         account_time_size, cmp_counts, dump_histogram,
124         ipa_profile_generate_summary, ipa_profile_write_summary,
125         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
126         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
127         Move to ipa-profile.c
129 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
131         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
133 2013-09-01  Jan Hubicka  <jh@suse.cz>
135         * common.opt (fdevirtualize-speculatively): New function.
136         * invoke.texi (fdevirtualize-speculatively): Document.
137         * ipa-devirt.c: Include ipa-inline.h
138         (likely_target_p): New function.
139         (ipa_devirt): New function.
140         (gate_ipa_devirt): New function.
141         (pass_data_ipa_devirt): New static var.
142         (pass_ipa_devirt): Likewise.
143         (make_pass_ipa_devirt): New function.
144         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
145         (common_handle_option): Disable devirtualization when
146         value range profiling is available.
147         * passes.def (pass_ipa_devirt): Add.
148         * timever.def (TV_IPA_DEVIRT): New timevar.
149         * tree-pass.h (make_pass_ipa_devirt): 
151 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
153         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
154         include sanitize(undefined).
156 2013-08-31  Diego Novillo  <dnovillo@google.com>
158         * Makefile.in (TREE_CORE_H): Define.
159         (TREE_H): Use.
160         (GTFILES): Add tree-core.h.
161         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
162         size the array.
163         * tree-core.h: New file.
164         Move all data structures, enum, typedefs, global
165         declarations and constants from ...
166         * tree.h: ... here.
168 2013-08-31  Jan Hubicka  <jh@suse.cz>
170         * bulitins.c (expand_builtin): Do not early exit for gcov
171         instrumented functions.
173 2013-08-31  Marek Polacek  <polacek@redhat.com>
175         * ubsan.c: Include tm_p.h.
177 2013-08-31  Jan Hubicka  <jh@suse.cz>
179         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
180         warning.
182         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
183         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
184         * tree-cfg.c (verify_gimple_label): ... here.
185         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
186         (ipa_merge_profiles): New function.
187         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
188         (lto_input_function_body): Likewise.
189         * ipa-utils.h (ipa_merge_profiles): Declare.
190         * lto-streamer.h (lto_input_function_body): Update prototype.
191         (emit_label_in_global_context_p): Remove.
192         * lto-symtab.c: Include ipa-utils.h
193         (lto_cgraph_replace_node): Use ipa_merge_profiles.
195 2013-08-31  Jan Hubicka  <jh@suse.cz>
197         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
199 2013-08-31  Jan Hubicka  <jh@suse.cz>
201         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
203 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
205         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
206         "cmp" RTX before signed_comparison_operator check to account
207         for "code" changes.
209 2013-08-30  Jan Hubicka  <jh@suse.cz>
211         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
212         (detect_type_change_1): Rename to ...
213         (detect_type_change): ... this one; early return on non-polymorphic
214         types.
215         (detect_type_change_ssa): Add comp_type parameter; update
216         use of detect_type_change.
217         (compute_complex_assign_jump_func): Add param_type parameter;
218         update use of detect_type_change_ssa.
219         (compute_complex_ancestor_jump_func): Likewise.
220         (ipa_get_callee_param_type): New function.
221         (ipa_compute_jump_functions_for_edge): Compute parameter type;
222         update calls to the jump function computation functions.
224 2013-08-30  Teresa Johnson  <tejohnson@google.com>
225             Steven Bosscher  <steven@gcc.gnu.org>
227         * cfgrtl.c (fixup_new_cold_bb): New routine.
228         (commit_edge_insertions): Invoke fixup_partitions.
229         (find_partition_fixes): New routine.
230         (fixup_partitions): Ditto.
231         (verify_hot_cold_block_grouping): Update comments.
232         (rtl_verify_edges): Invoke find_partition_fixes.
233         (rtl_verify_bb_pointers): Update comments.
234         (rtl_verify_bb_layout): Ditto.
235         * basic-block.h (probably_never_executed_edge_p): Declare.
236         (fixup_partitions): Ditto.
237         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
238         * bb-reorder.c (sanitize_hot_paths): New function.
239         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
240         sanitize_hot_paths.
241         * predict.c (probably_never_executed_edge_p): New routine.
242         * cfg.c (check_bb_profile): Add partition insanity warnings.
244 2013-08-30  Meador Inge  <meadori@codesourcery.com>
246         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
248 2013-08-30  Marek Polacek  <polacek@redhat.com>
250         * Makefile.in (ubsan.o): Add.
251         (c-family/c-ubsan.o): Add.
252         (builtins.o): Add ubsan.h dependency.
253         * ubsan.h: New file.
254         * ubsan.c: New file.
255         * common.opt: Add -fsanitize=undefined option.
256         (flag_sanitize): Add variable.
257         (fsanitize=): Add option.  Add Driver.
258         (fsanitize=thread): Remove option.
259         (fsanitize=address): Likewise.
260         (static-libubsan): New option.
261         * doc/invoke.texi: Document the new flag and -static-libubsan.
262         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
263         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
264         * builtin-attrs.def (ATTR_COLD): Define.
265         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
266         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
267         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
268         * flag-types.h (sanitize_code): New enum.
269         * opts.c (common_handle_option): Parse command line arguments
270         of -fsanitize=.  Add -fsanitize=unreachable option.
271         * varasm.c (get_variable_section): Adjust.
272         (assemble_noswitch_variable): Likewise.
273         (assemble_variable): Likewise.
274         (output_constant_def_contents): Likewise.
275         (categorize_decl_for_section): Likewise.
276         (place_block_symbol): Likewise.
277         (output_object_block): Likewise.
278         * builtins.def: Likewise.
279         * toplev.c (compile_file): Likewise.
280         (process_options): Likewise.
281         * cppbuiltin.c: Likewise.
282         * tsan.c (tsan_pass): Likewise.
283         (tsan_gate): Likewise.
284         (tsan_gate_O0): Likewise.
285         * cfgexpand.c (partition_stack_vars): Likewise.
286         (expand_stack_vars): Likewise.
287         (defer_stack_allocation): Likewise.
288         (expand_used_vars): Likewise.
289         * cfgcleanup.c (old_insns_match_p): Likewise.
290         * asan.c (asan_finish_file): Likewise.
291         (asan_instrument): Likewise.
292         (gate_asan): Likewise.
293         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
294         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
295         (asan_global_struct): Use pointer_sized_int_node instead
296         calling build_nonstandard_integer_type.
297         (initialize_sanitizer_builtins): Likewise.
298         (asan_finish_file): Likewise.
299         * gcc.c: Document %{%:function(args):X}.
300         (static_spec_functions): Add sanitize.
301         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
302         store funcval != NULL there.
303         (do_spec_1): Adjust handle_spec_function caller.
304         (handle_braces): Allow %:function(args) as condition.
305         (sanitize_spec_function): New function.
306         (ADD_STATIC_LIBUBSAN_LIBS): Define.
307         (LIBUBSAN_SPEC): Likewise.
308         (LIBUBSAN_EARLY_SPEC): Likewise.
309         (SANITIZER_SPEC): Handle libubsan.
310         (SANITIZER_EARLY_SPEC): Likewise.
311         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
312         instead of fsanitize=address.
313         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
314         instead of fsanitize=address*.
315         * builtins.c: Include ubsan.h.
316         (fold_builtin_0): Instrument __builtin_unreachable.
317         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
318         instead of flag_asan.
319         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
320         (pointer_sized_int_node): Define.
321         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
323 2013-08-30  Mike Stump  <mikestump@comcast.net>
325         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
326         with RE patterns.
328 2013-08-29  Jan Hubicka  <jh@suse.cz>
330         * cgraph.c (cgraph_function_body_availability): Handle weakref
331         correctly.
332         * passes.def: Remove pass_fixup_cfg.
333         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
334         track when we need to remove functions.
335         (gate_ipa_inline): Execute inlining always; add comment why.
336         (pass_data_ipa_inline): Remove TODO_remove_functions.
337         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
338         do not produce summaries.
339         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
340         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
341         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
342         constant pool and vtable.
344 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
346         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
347         New arm_neon.h's internal macros to specify 64-bit constants.
348         Avoid using stdint.h's macros.
350 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
352         * recog.c (verify_changes): Verify that changes[i].old is non-zero
353         before applying REG_P.
355 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
357         PR tree-optimization/58277
358         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
359         after seeing too many stmts with vdef in between dombb and current
360         bb, invalidate everything.
362 2013-08-30  Richard Biener  <rguenther@suse.de>
364         * fold-const.c (fold_single_bit_test): Fix overflow test.
366 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
368         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
369         and which can live in a register, always retrieve the value on entry.
370         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
371         passed by invisible reference specially.
372         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
373         (vt_add_function_parameter): Correctly deal with a parameter passed by
374         invisible reference.
376 2013-08-30  Jan Hubicka  <jh@suse.cz>
378         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
380 2013-08-30  Richard Biener  <rguenther@suse.de>
382         PR tree-optimization/58228
383         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
384         allow invariant loads in nested loop vectorization.
386 2013-08-30  Richard Biener  <rguenther@suse.de>
388         PR tree-optimization/58223
389         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
390         (has_anti_or_output_dependence): ... this and adjust to also
391         look for output dependences.
392         (mark_nodes_having_upstream_mem_writes): Adjust.
393         (rdg_flag_uses): Likewise.
395 2013-08-30  Richard Biener  <rguenther@suse.de>
397         PR tree-optimization/58010
398         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
399         assert that we have a loop-closed PHI.
401 2013-08-29  Jan Hubicka  <jh@suse.cz>
403         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
404         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
405         * lto-section-in.c (lto_free_function_in_decl_state): New function.
406         (lto_free_function_in_decl_state_for_node): New function.
408 2013-08-29  Xinliang David Li  <davidxl@google.com>
410         * loop-unroll.c (report_unroll_peel): Minor message change.
411         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
412         Emit alignment peeling message with default -fopt-info.
413         (vect_loop_versioning): Emit loop version info message.
414         * tree-vectorizer.c (vectorize_loops): Minor message change.
415         (execute_vect_slp): Ditto.
417 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
419         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
420         of the clone from the DECL_NAME of the original function.
422 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
424         * passes.c (register_pass): Add overload.
425         * tree-pass.h (register_pass): Forward declare it.  Add comment.
427 2013-08-29  Jan Hubicka  <jh@suse.cz>
429         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
430         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
431         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
432         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
433         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
434         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
435         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
436         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
438 2013-08-29  Teresa Johnson  <tejohnson@google.com>
440         * dumpfile.c (dump_loc): Output column number.
441         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
442         * doc/invoke.texi: Document optall -fopt-info flag.
443         * profile.c (read_profile_edge_counts): Use new dump framework.
444         (compute_branch_probabilities): Ditto.
445         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
446         when pass not in any opt group.
447         * pass_manager.h (pass_manager::get_pass_profile): New method.
448         * value-prof.c (check_counter): Use new dump framework.
449         (check_ic_target): Ditto.
450         * coverage.c (get_coverage_counts): Ditto.
451         (coverage_init): Setup new dump framework.
453 2013-08-29  Richard Biener  <rguenther@suse.de>
455         PR tree-optimization/58246
456         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
457         handle the dominance check inside a basic-block.
459 2013-08-29  Richard Biener  <rguenther@suse.de>
461         PR middle-end/57287
462         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
463         of default defs that appear in abnormal PHI nodes.
465 2013-08-29  Richard Biener  <rguenther@suse.de>
467         PR tree-optimization/57685
468         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
469         single-use operands to avoid exponential complexity.
471 2013-08-28  Dehao Chen  <dehao@google.com>
473         * ipa-inline.c (edge_badness): Fix integer underflow.
475 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
477         * gtm-builtins.def (_ITM_free): Declare leaf.
479 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
481         PR target/58067
482         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
483         (*tls_local_dynamic_base_64_largepic): Likewise.
484         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
485         Remove predicate from call operand.
486         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
487         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
489 2013-08-28  Jeff Law  <law@redhat.com>
491         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
492         checks for the number of predecessors and successors allowed.
493         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
494         which require copying a joiner block if there is a request which
495         is a subpath that requires no joiner block copying.
497 2013-08-28  Jan Hubicka  <jh@suse.cz>
499         * lto-streamer-out.c (DFS_write_tree_body): Drop
500         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
501         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
502         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
503         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
504         Do not read DECL_ERROR_ISSUED.
505         (unpack_ts_decl_with_vis_value_fields): Do not read
506         DECL_DEFER_OUTPUT.
507         (lto_input_ts_binfo_tree_pointers): Do not read
508         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
509         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
510         write DECL_ERROR_ISSUED..
511         (pack_ts_decl_with_vis_value_fields): Do not write
512         DECL_DEFER_OUTPUT.
513         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
514         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
515         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
516         * tree.h (tree_decl_common): Update comment.
517         (DECL_ERROR_ISSUED): Remove.
519 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
521         PR middle-end/58257
522         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
524 2013-08-28  Jan Hubicka  <jh@suse.cz>
526         * builtins.def (free): Declare leaf.
528 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
530         * gdbhooks.py: New.
531         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
532         * configure: Regenerate.
534 2013-08-27  Martin Jambor  <mjambor@suse.cz>
536         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
537         (ipa_ancestor_jf_data): Likewise.
538         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
539         (ipa_get_jf_pass_through_type_preserved): New function.
540         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
541         (ipa_get_jf_ancestor_type_preserved): New function.
542         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
543         (ipa_get_jf_ancestor_result): Likewise.
544         (propagate_vals_accross_pass_through): Use
545         ipa_get_jf_pass_through_result to do all the value mappings.
546         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
547         type_preserved flag.
548         (ipa_set_jf_cst_copy): New function.
549         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
550         (ipa_set_jf_arith_pass_through): Likewise.
551         (ipa_set_ancestor_jf): Likewise.
552         (compute_complex_assign_jump_func): Set type_preserved instead of
553         punting.
554         (ipa_compute_jump_functions_for_edge): Likewise.
555         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
556         (update_jump_functions_after_inlining): Update type_preserved.
557         Explicitely create jump functions when combining one with pass_through.
558         (ipa_write_jump_function): Stream the type_preserved flags.
559         (ipa_read_jump_function): Likewise.
561 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
562             Aldy Hernandez  <aldyh@redhat.com>
564         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
565         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
566         * function.h (struct function): Add has_force_vect_loops and
567         has_simduid_loops.
568         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
569         * gimple.c (gimple_build_omp_critical): Add KIND argument and
570         handle it.
571         * gimple.def: Update CLAUSES comments.
572         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
573         (gimple_build_omp_for): Add argument to prototype.
574         (gimple_omp_for_kind): New.
575         (gimple_omp_for_set_kind): New.
576         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
577         GOVD_DATA_SHARE_CLASS.
578         (enum omp_region_type): Add ORT_SIMD.
579         (gimple_add_tmp_var): Handle ORT_SIMD.
580         (gimplify_var_or_parm_decl): Same.
581         (is_gimple_stmt): Same.
582         (omp_firstprivatize_variable): Same.
583         (omp_add_variable): Only use splay_tree_insert if lookup failed.
584         (omp_notice_variable): Handle ORT_SIMD.
585         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
586         (omp_check_private): Handle ORT_SIMD.
587         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
588         OMP_CLAUSE_SAFELEN.
589         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
590         Handle OMP_CLAUSE_LASTPRIVATE.
591         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
592         OMP_CLAUSE_SAFELEN.
593         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
594         (gimplify_expr): Handle OMP_SIMD.
595         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
596         (expand_GOMP_SIMD_VF): New.
597         (expand_GOMP_SIMD_LAST_LANE): New.
598         * internal-fn.def (GOMP_SIMD_LANE): New.
599         (GOMP_SIMD_VF): New.
600         (GOMP_SIMD_LAST_LANE): New.
601         * omp-low.c: Include target.h.
602         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
603         OMP_CLAUSE_SAFELEN.
604         (check_omp_nesting_restrictions): Same.
605         (omp_max_vf): New.
606         (lower_rec_simd_input_clauses): New.
607         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
608         OMP_CLAUSE_LINEAR.
609         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
610         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
611         (expand_omp_build_assign): New.
612         (expand_omp_for_init_counts): New.
613         (expand_omp_for_init_vars): New.
614         (extract_omp_for_update_vars): New.
615         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
616         and rewrite accordingly.
617         (expand_omp_simd): New.
618         (expand_omp_for): Use expand_omp_simd.
619         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
620         (lower_omp_for): Do not lower the body.
621         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
622         in their own loops.
623         * tree-flow.h (find_omp_clause): Remove prototype.
624         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
625         forcing vectorization of the loop, or if flag_tree_vectorize.
626         (gate_tree_if_conversion): Similarly.
627         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
628         gimple_build_omp_for.
629         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
630         * tree-parloops (create_parallel_loop): Pass kind argument to
631         gimple_build_omp_for.
632         * tree-pretty-print.c (dump_omp_clause): Add cases for
633         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
634         OMP_CLAUSE__SIMDUID_.
635         (dump_generic_node): Handle OMP_SIMD.
636         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
637         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
638         unroll OMP_SIMD loops here.
639         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
640         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
641         loop->safelen.
642         (vect_analyze_data_refs): Handle simd loops.
643         * tree-vect-loop.c (vectorizable_live_operation): Handle
644         IFN_GOMP_SIMD*.
645         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
646         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
647         (vectorizable_load): Same.
648         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
649         (struct simduid_to_vf): New.
650         (simduid_to_vf::hash): New.
651         (simduid_to-vf::equal): New.
652         (struct simd_array_to_simduid): New.
653         (simd_array_to_simduid::hash): New.
654         (simd_array_to_simduid::equal): New.
655         (adjust_simduid_builtins): New.
656         (struct note_simd_array_uses_struct): New.
657         (note_simd_array_uses_cb): New.
658         (note_simd_array_uses): New.
659         (vectorize_loops): Handle simd hints and adjust simd builtins
660         accordingly.
661         * tree-vectorizer.h (struct _stmt_vec_info): Add
662         simd_lane_access_p field.
663         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
664         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
665         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
666         (omp_clause_code_name): Same.
667         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
668         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
669         * tree.def (OMP_SIMD): New entry.
670         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
671         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
672         (OMP_CLAUSE_DECL): Adjust range for new clauses.
673         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
674         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
675         (OMP_CLAUSE_LINEAR_STEP): New.
676         (OMP_CLAUSE_SAFELEN_EXPR): New.
677         (OMP_CLAUSE__SIMDUID__DECL): New.
678         (find_omp_clause): New prototype.
680 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
682         * config/i386/driver-i386.c (host_detect_local_cpu): Update
683         Haswell processor detection.
685 2013-08-27  Christian Widmer  <shadow@umbrox.de>
687         PR target/57927
688         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
689         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
690         AVX2 capable processors.
692 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
694         * config/aarch64/arm_neon.h: Replace all inline asm implementations
695         of vget_low_* with implementations in terms of other intrinsics.
697 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
699         PR middle-end/57219
700         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
701         values to -1, 0 and 1.
703 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
705         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
706         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
707         (<optab><mode>3_insn): Remove.
708         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
709         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
710         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
711         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
712         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
713         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
714         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
715         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
716         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
717         (ror<mode>3_insn): Likewise.
718         * config/aarch64/predicates.md (aarch64_simd_register): New.
720 2013-08-27  Richard Biener  <rguenther@suse.de>
722         PR tree-optimization/57521
723         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
724         one edge is non-critical.
725         (find_phi_replacement_condition): Make sure to use a non-critical
726         edge.  Cleanup and remove old bug workarounds.
727         (bb_postdominates_preds): Remove.
728         (if_convertible_loop_p_1): Do not compute post-dominators.
729         (combine_blocks): Do not free post-dominators.
730         (main_tree_if_conversion): Likewise.
731         (pass_data_if_conversion): Add TODO_verify_ssa.
733 2013-08-27  DJ Delorie  <dj@redhat.com>
735         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
737 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
739         * function.c (assign_parm_find_data_types): Set passed_mode and
740         nominal_mode to the TYPE_MODE of nominal_type for the built
741         pointer type in case of the struct-pass-by-reference.
743 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
745         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
746         (UINT16_TYPE): Change default to "unsigned int".
748         * config/avr/avr.opt (mfract-convert-truncate): New option.
749         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
750         is set, round negative fractional integers according to n1169
751         when converting to integer types.
753 2013-08-26  Jan Hubicka  <jh@suse.cz>
755         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
756         methods can not be called indirectly when their address is not taken.
758 2013-08-26  Jan Hubicka  <jh@suse.cz>
760         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
761         ctor_for_folding.
763 2013-08-26  Jan Hubicka  <jh@suse.cz>
765         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
766         can be unshared.
768 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
770         * reload.c (find_valid_class): Allow classes that do not include
771         FIRST_PSEUDO_REGISTER - 1.
773 2013-08-26  Jan Hubicka  <jh@suse.cz>
775         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
776         fix edge count/frequency when speculation failed; fix type check
777         for the direct call.
779 2013-08-26  Jan Hubicka  <jh@suse.cz>
781         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
783 2013-08-26  Jan Hubicka  <jh@suse.cz>
785         * ipa-inline-transform.c (inline_transform): Be ready for basic block
786         to be changed by edge redirection.
788 2013-08-26  Jan Hubicka  <jh@suse.cz>
790         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
791         formating; add sanity check.
792         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
793         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
794         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
795         (resolve_noninline_speculation): Update callee keys, too.
797 2013-08-26  Jan Hubicka  <jh@suse.cz>
799         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
800         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
802 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
804         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
805         into proper place.
807 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
809         * config/i386/i386.c (ix86_debug_options): Remove prototype.
810         (x86_64_elf_select_section): Ditto.
811         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
812         arguments.
813         (ix86_pass_by_reference): Ditto.
814         (output_set_got): Ditto.
815         (ix86_unary_operator_ok): Ditto.
816         (ix86_expand_builtin): Ditto.
818 2013-08-23  Jan Hubicka  <jh@suse.cz>
820         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
822 2013-08-23  Jan Hubicka  <jh@suse.cz>
824         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
825         (tree_decl_with_vis): Add FINAL field.
827 2013-08-23  Jeff Law  <law@redhat.com>
829         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
830         the debugging dump when the expression is fully redundant.
832 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
834         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
835         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
836         * pretty-print.c (pp_formatted_text_data): Likewise.
837         (pp_write_text_to_stream): Likewise.
838         (pp_write_text_as_dot_label_to_stream): Likewise.
839         (pp_append_r): Likewise.
840         (pp_format): Likewise.
841         (pp_flush): Likewise.
842         (pp_clear_output_area): Likewise.
843         (pp_append_text): Likewise.
844         (pp_formatted_text): Likewise.
845         (pp_remaining_character_count_for_line): Likewise.
846         (pp_newline): Likewise.
847         (pp_character): Likewise.
848         (output_buffer::~output_buffer): Define.
849         (pretty_printer::~pretty_printer): Destruct output buffer.
850         * pretty-print.h (output_buffer::~output_buffer): Declare.
851         (pretty_printer::~pretty_printer): Declare virtual.
853 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
855         PR other/57324
856         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
857         HOST_WIDE_INT_M1U): New macros.
858         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
859         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
860         unsigned -1 for lshift.
861         * cse.c (cse_insn): Likewise.
862         * double-int.c (rshift_double, lshift_double): Likewise.
863         * builtins.c (fold_builtin_bitop): Likewise.
864         * combine.c (force_to_mode): Likewise.
865         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
866         * simplify-rtx.c (simplify_const_unary_operation,
867         simplify_const_binary_operation): Likewise.
868         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
869         check_va_list_escapes): Likewise.
870         * rtlanal.c (nonzero_bits1): Likewise.
871         * expmed.c (expand_smod_pow2): Likewise.
872         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
874 2013-08-23  Jan Hubicka  <jh@suse.cz>
876         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
877         as having address taken.
879 2013-08-23  Jan Hubicka  <jh@suse.cz>
881         * ipa-utils.h (method_class_type): Declare.
882         * ipa-devirt.c (method_class_type): Export.
884         * cgraphunit.c (analyze_functions): Do basic devirtualization;
885         do not walk base classes of anonymous types.
887 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
889         PR rtl-optimization/58220
890         PR regression/58221
891         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
892         handle SEQUENCE insns properly.
894 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
896         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
897         definition.
898         (pp_newline_and_indent): Likewise.
899         (pp_separate_with): Likewise.
900         * pretty-print.c (pp_newline_and_flush): Define.
901         (pp_newline_and_indent): Likewise.
902         (pp_separate_with): Likewise.
904 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
906         PR target/58218
907         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
908         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
910 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
912         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
913         * gcc/config/i386/i386.md (*movti_internal): Use
914         predicate to determine if EVEX is needed.
915         (*movsi_internal): Ditto.
916         (*movdf_internal): Ditto.
917         (*movsf_internal): Ditto.
918         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
920 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
922         PR tree-optimization/58209
923         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
924         (find_tail_calls): Give up for pointer result types if m is non-NULL.
925         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
926         emit POINTER_PLUS_EXPR.
927         (create_tailcall_accumulator): For pointer result type accumulate in
928         sizetype type.
930 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
932         * configure.ac: Add backslashes missing from the last change.
933         * configure: Regenerate.
935 2013-08-22  Jan Hubicka  <jh@susue.cz>
937         * ipa.c (function_and_variable_visibility): First remember function
938         was global and then make it local.
940 2013-08-22  Julian Brown  <julian@codesourcery.com>
942         * configure.ac: Add aarch64 to list of arches which use "nop" in
943         debug_line test.
944         * configure: Regenerate.
946 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
948         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
949         gnu_libc_has_function.
950         * config/s390/tpf.h: Likewise.
952 2013-08-22  Jan Hubicka  <jh@susue.cz>
954         * timevar.c (validate_phases): Add cast.
956 2013-08-22  Jan Hubicka  <jh@susue.cz>
958         * timevar.c (validate_phases): Use size_t for memory.
959         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
961 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
963         * pretty-print.h (output_buffer::output_buffer): Declare.
964         (pretty_printer::pretty_printer): Likewise.
965         (pp_construct): Remove.
966         * pretty-print.c (output_buffer::output_buffer): Define.
967         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
968         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
969         (print_gimple_expr): Likewise.
970         (print_gimple_seq): Likewise.
971         (gimple_dump_bb): Likewise.
972         * sched-vis.c (dump_value_slim): Likewise.
973         (dump_insn_slim): Likewise.
974         (dump_rtl_slim): Likewise.
975         (str_pattern_slim): Likewise.
976         * tree-mudflap.c (mf_varname_tree): Likewise.
977         * graph.c (print_graph_cfg): Likewise.
978         (start_graph_dump): Likewise.
979         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
980         placement-new.
981         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
982         pretty printer initialization.
983         * coretypes.h (diagnostic_context): Remove superflous type alias
984         declaration.
985         (pretty_printer): Likewise.  Declare directly as a class.
986         (pretty_print_info): Remove declaration as class.
987         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
988         and pp_clear_output_area.
989         (asan_add_global): Likewise.
991 2013-08-22  Jan Hubicka  <jh@suse.cz>
993         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
994         * ipa-utils.h (update_type_inheritance_graph): Declare.
995         (possible_polymorphic_call_target_p): Declare.
996         (possible_polymorphic_call_target_p): New.
997         * ipa-devirt.c: Update toplevel comments.
998         (cached_polymorphic_call_targets): Move up.
999         (odr_type_d): Move ID down.
1000         (polymorphic_type_binfo_p): Update comment.
1001         (odr_hasher::remove): Likewise;
1002         (get_odr_type): Set anonymous_namespace.
1003         (dump_odr_type): Dump it.
1004         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
1005         (maybe_record_node): Record node in cached_polymorphic_call_targets.
1006         (record_binfo): Add comment.
1007         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
1008         built.
1009         (devirt_node_removal_hook): Do not iCE when cache is freed.
1010         (possible_polymorphic_call_target_p): New predicate.
1011         (update_type_inheritance_graph): New function.
1013 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1014             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1015             Sergey Lega  <sergey.s.lega@intel.com>
1016             Anna Tikhonova  <anna.tikhonova@intel.com>
1017             Ilya Tocar  <ilya.tocar@intel.com>
1018             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1019             Ilya Verbin  <ilya.verbin@intel.com>
1020             Kirill Yukhin  <kirill.yukhin@intel.com>
1021             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1023         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1024         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1025         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1026         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1027         (OPTION_MASK_ISA_AVX2_UNSET): Update.
1028         (OPTION_MASK_ISA_AVX512F_UNSET): New.
1029         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1030         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1031         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1032         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1033         OPT_mavx512pf, OPT_mavx512er cases.
1034         * config/i386/constraints.md (v): New constraint.
1035         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1036         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1037         (bit_AVX512CD): New.
1038         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1039         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1040         * config/i386/i386-c.c (ix86_target_macros_internal):
1041         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1042         __AVX512PF__.
1043         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1044         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1045         * config/i386/i386.c (regclass_map, dbx_register_map)
1046         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1047         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1048         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1049         -mavx512pf options.
1050         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1051         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
1052         -mavx512pf options.  Fix formatting.
1053         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1054         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
1055         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1056         -mavx512cd, -mavx512pf options.
1057         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1058         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1059         (ix86_preferred_output_reload_class): Replace SSE_REGS with
1060         ALL_SSE_REGS.
1061         (ix86_hard_regno_mode_ok): Support 512-bit registers.
1062         (ix86_set_reg_reg_cost): Ditto.
1063         (x86_order_regs_for_local_alloc): Ditto.
1064         (MAX_VECT_LEN): Extend to 64-byte.
1065         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1066         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1067         (TARGET_AVX512ER, TARGET_AVX512CD): New.
1068         (BIGGEST_ALIGNMENT): Extend to 512-bits.
1069         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1070         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1071         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1072         (SSE_REG_MODE_P): Support new modes.
1073         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1074         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1075         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1076         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1077         (REG_CLASS_CONTENTS): Add new registers.
1078         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1079         (EXT_REX_SSE_REGNO_P): New.
1080         (HI_REGISTER_NAMES): Add new registers.
1081         * config/i386/i386.md: Define constants for new registers.
1082         (mode): Add new 512-bit modes.
1083         (prefix): Support evex prefix.
1084         (isa): Support avx512f, noavx512f, fma_avx512f.
1085         (ssemodesuffix): Add new 512-bit modes.
1086         (movxi): New.
1087         (*movxi_internal_avx512f): Ditto.
1088         (*movdi_internal): Replace constraint "x" with the new constraint "v".
1089         Support MODE_XI.
1090         (*movsi_internal): Likewise.
1091         (*movdf_internal): Likewise.
1092         (*movsf_internal): Likewise.
1093         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1094         (<code><mode>3): Likewise.
1095         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1096         New.
1097         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1098         with the new constraint "v".
1099         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1100         modes.
1101         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1102         with the new constraint "v".
1103         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1104         (<sse2>_storedqu<avxsizesuffix>): Likewise.
1105         (*<plusminus_insn><mode>3): Likewise.
1106         (<sse>_vm<plusminus_insn><mode>3): Likewise.
1107         (*mul<mode>3): Likewise.
1108         (<sse>_vmmul<mode>3): Likewise.
1109         (<sse>_div<mode>3): Likewise.
1110         (<sse>_vmdiv<mode>3): Likewise.
1111         (<sse>_sqrt<mode>2): Likewise.
1112         (<sse>_vmsqrt<mode>2): Likewise.
1113         (*<code><mode>3_finite): Likewise.
1114         (*<code><mode>3) <smaxmin>: Likewise.
1115         (<sse>_vm<code><mode>3): Likewise.
1116         (*<code><mode>3) <any_logic>: Likewise.
1117         (*fma_fmadd_<mode>): Likewise.
1118         (*fma_fmsub_<mode>): Likewise.
1119         (*fma_fnmadd_<mode>): Likewise.
1120         (*fma_fnmsub_<mode>): Likewise.
1121         (*fma_fmaddsub_<mode>): Likewise.
1122         (*fma_fmsubadd_<mode>): Likewise.
1123         (*fmai_fmadd_<mode>): Likewise.
1124         (*fmai_fmsub_<mode>): Likewise.
1125         (*fmai_fnmadd_<mode>): Likewise.
1126         (*fmai_fnmsub_<mode>): Likewise.
1127         (sse_cvtsi2ss): Likewise.
1128         (sse_cvtsi2ssq): Likewise.
1129         (sse_cvtss2si): Likewise.
1130         (sse_cvtss2si_2): Likewise.
1131         (sse_cvtss2siq): Likewise.
1132         (sse_cvtss2siq_2): Likewise.
1133         (sse_cvttss2si): Likewise.
1134         (sse_cvtss2siq_2): Likewise.
1135         (float<sseintvecmodelower><mode>2): Likewise.
1136         (sse2_cvtsd2si_2): Likewise.
1137         (sse2_cvtsd2siq_2): Likewise.
1138         (*<plusminus_insn><mode>3): Likewise.
1139         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1140         (*<sse4_1_avx2>_mul<mode>3): Likewise.
1141         (ashr<mode>3): Likewise.
1142         (<shift_insn><mode>3): Likewise.
1143         (avx2_<code><mode>3): Likewise.
1144         (*avx2_<code><mode>3): Likewise.
1145         (*andnot<mode>3): Likewise.
1146         (*<code><mode>3) <any_logic>: Likewise.
1147         (abs<mode>2): Likewise.
1148         (avx2_permvar<mode>): Likewise.
1149         (avx2_perm<mode>_1): Likewise.
1150         (*avx_vpermilp<mode>): Likewise.
1151         (avx_vpermilvar<mode>3): Likewise.
1152         (avx2_ashrv<mode>): Likewise.
1153         (avx2_<shift_insn>v<mode>): Likewise.
1154         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1155         -mavx512cd.
1156         * doc/rtl.texi: Document XImode.
1158 2013-08-21  Jeff Law  <law@redhat.com>
1160         * tree-flow.h (register_jump_thread): Pass vector of edges
1161         instead of each important edge.
1162         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1163         thread path into a vector and pass that to register_jump_thread.
1164         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1165         passed in edge vector to the current 3-edge form.
1167         Revert:
1168         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1170         * dce.c (fini_dce): Call df_analyze again just in case
1171         delete_unmarked_insns removed anything.
1173 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
1175         * reload.h (struct reg_equivs): Rename to ..
1176         (struct reg_equivs_s): .. this.
1178 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
1180         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1182 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1184         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1186 2013-08-21  Jeff Law  <law@redhat.com>
1188         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1189         simplify assignments too.  If the RHS collapses to a singleton
1190         range, then return the value for the range.
1192 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
1194         * config/i386/sse.md (V16): Rename to...
1195         (VMOVE): this.
1196         (mov<mode>): Update iterator name.
1197         (*mov<mode>_internal): Ditto.
1198         (push<mode>1): Ditto.
1199         (movmisalign<mode>): Ditto.
1201 2013-08-20  Jan Hubicka  <jh@suse.cz>
1203         PR bootstrap/58186
1204         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1205         entry for direct edges.
1206         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1208 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1210         Revert my last two changes, r201865 and r201864:
1212         Revert r201865:
1213         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1215         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1216         instances can own GC refs.
1218         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1219         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1220         (gcc::context::gt_pch_nx): Likewise.
1221         (gcc::context::gt_pch_nx):  Likewise.
1222         * ggc.h (gt_ggc_mx <T>): New.
1223         (gt_pch_nx_with_op <T>): New.
1224         (gt_pch_nx <T>): New.
1225         * passes.c (opt_pass::gt_ggc_mx): New.
1226         (opt_pass::gt_pch_nx): New.
1227         (opt_pass::gt_pch_nx_with_op): New.
1228         (pass_manager::gt_ggc_mx): New.
1229         (pass_manager::gt_pch_nx): New.
1230         (pass_manager::gt_pch_nx_with_op): New.
1231         (pass_manager::operator new): Use
1232         ggc_internal_cleared_alloc_stat rather than xcalloc.
1233         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1234         (pass_manager::gt_ggc_mx): New.
1235         (pass_manager::gt_pch_nx): New.
1236         (pass_manager::gt_pch_nx_with_op): New.
1237         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1238         (opt_pass::operator new): New.
1239         (opt_pass::gt_ggc_mx): New.
1240         (opt_pass::gt_pch_nx): New.
1241         (opt_pass::gt_pch_nx_with_op): New.
1243         Revert r201864:
1244         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1246         * Makefile.in (GTFILES): Add context.h.
1247         * context.c (gcc::context::operator new): New.
1248         (gcc::context::gt_ggc_mx): New.
1249         (gcc::context::gt_pch_nx): New.
1250         (gcc::context::gt_pch_nx): New.
1251         * context.h (gcc::context): Add GTY((user)) marking.
1252         (gcc::context::operator new): New.
1253         (gcc::context::gt_ggc_mx): New.
1254         (gcc::context::gt_pch_nx): New.
1255         (gcc::context::gt_pch_nx): New.
1256         (g): Add GTY marking.
1257         (gt_ggc_mx (gcc::context *)): New.
1258         (gt_pch_nx (gcc::context *)): New.
1259         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1260         void *cookie)): New.
1261         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1263 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1265         * dce.c (fini_dce): Call df_analyze again just in case
1266         delete_unmarked_insns removed anything.
1268 2013-08-20  Teresa Johnson  <tejohnson@google.com>
1270         PR rtl-optimizations/57451
1271         * final.c (reemit_insn_block_notes): Prevent lexical blocks
1272         from crossing split section boundaries.
1274 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1276         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1277         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1278         with MULTLIB_DEFAULTS.
1280 2013-08-20  Nick Clifton  <nickc@redhat.com>
1282         * target.def (narrow_volatile_bitfield): Note that the default
1283         value is false, not !TARGET_STRICT_ALIGN.
1284         * doc/tm.texi: Regenerate.
1286 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
1288         Fix LIB_SPEC for systems without libpthread.
1290         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1291         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1292         for Android.
1293         * config/i386/linux-common.h: Likewise.
1294         * config/mips/linux-common.h: Likewise.
1296 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
1298         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1299         checks.
1301 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1303         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1304         instances can own GC refs.
1306         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1307         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1308         (gcc::context::gt_pch_nx): Likewise.
1309         (gcc::context::gt_pch_nx):  Likewise.
1310         * ggc.h (gt_ggc_mx <T>): New.
1311         (gt_pch_nx_with_op <T>): New.
1312         (gt_pch_nx <T>): New.
1313         * passes.c (opt_pass::gt_ggc_mx): New.
1314         (opt_pass::gt_pch_nx): New.
1315         (opt_pass::gt_pch_nx_with_op): New.
1316         (pass_manager::gt_ggc_mx): New.
1317         (pass_manager::gt_pch_nx): New.
1318         (pass_manager::gt_pch_nx_with_op): New.
1319         (pass_manager::operator new): Use
1320         ggc_internal_cleared_alloc_stat rather than xcalloc.
1321         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1322         (pass_manager::gt_ggc_mx): New.
1323         (pass_manager::gt_pch_nx): New.
1324         (pass_manager::gt_pch_nx_with_op): New.
1325         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1326         (opt_pass::operator new): New.
1327         (opt_pass::gt_ggc_mx): New.
1328         (opt_pass::gt_pch_nx): New.
1329         (opt_pass::gt_pch_nx_with_op): New.
1331 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1333         * Makefile.in (GTFILES): Add context.h.
1334         * context.c (gcc::context::operator new): New.
1335         (gcc::context::gt_ggc_mx): New.
1336         (gcc::context::gt_pch_nx): New.
1337         (gcc::context::gt_pch_nx): New.
1338         * context.h (gcc::context): Add GTY((user)) marking.
1339         (gcc::context::operator new): New.
1340         (gcc::context::gt_ggc_mx): New.
1341         (gcc::context::gt_pch_nx): New.
1342         (gcc::context::gt_pch_nx): New.
1343         (g): Add GTY marking.
1344         (gt_ggc_mx (gcc::context *)): New.
1345         (gt_pch_nx (gcc::context *)): New.
1346         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1347         void *cookie)): New.
1348         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1350 2013-08-20  Alan Modra  <amodra@gmail.com>
1352         PR target/57865
1353         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1354         (rs6000_emit_epilogue): Likewise.
1356 2013-08-19  Dehao Chen  <dehao@google.com>
1358         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1360 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
1361             Jakub Jelinek  <jakub@redhat.com>
1363         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1364         (BUILT_IN_FABSD64): Likewise.
1365         (BUILT_IN_FABSD128): Likewise.
1366         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1367         (fold_builtin_1): Likewise.
1368         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1369         destination and source operands.
1370         (*abstd2_fpr): Likewise.
1371         (*nabstd2_fpr): Likewise.
1373 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
1375         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1376         JUMP_P and INSN_P.
1378 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
1380         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1381         incomplete.
1383 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1385         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1386         * builtins.c (default_libc_has_function): New.
1387         (gnu_libc_has_function): Ditto.
1388         (no_c99_libc_has_function): Ditto.
1389         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1390         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1391         (fold_builtin_sincos): Likewise.
1392         (fold_builtin_cexp): Likewise.
1393         * builtins.def (DEF_C94_BUILTIN): Likewise.
1394         (DEF_C99_BUILTIN): Likewise.
1395         (DEF_C99_C90RES_BUILTIN): Likewise.
1396         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1397         definitions to using this define.
1398         * config/darwin-protos.h (darwin_libc_has_function): New.
1399         * config/darwin.c (darwin_libc_has_function): Ditto.
1400         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1401         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1402         * config/darwin.h: Ditto.
1403         * config/elfos.h: Ditto.
1404         * config/freebsd.h: Ditto.
1405         * config/i386/cygming.h: Ditto.
1406         * config/i386/djgpp.h: Ditto.
1407         * config/i386/i386-interix.h: Ditto.
1408         * config/microblaze/microblaze.h: Ditto.
1409         * config/mmix/mmix.h: Ditto.
1410         * config/gnu-user.h: Ditto.
1411         * config/ia64/hpux.h: Ditto.
1412         * config/pa/pa-hpux.h: Ditto.
1413         * config/pdp11/pdp11.h: Ditto.
1414         * config/picochip/picochip.h: Ditto.
1415         * config/linux.h: Ditto.
1416         * config/netbsd.h: Ditto.
1417         * config/openbsd.h: Ditto.
1418         * config/rs6000/aix43.h: Ditto.
1419         * config/rs6000/aix51.h: Ditto.
1420         * config/rs6000/aix52.h: Ditto.
1421         * config/rs6000/aix53.h: Ditto.
1422         * config/rs6000/aix61.h: Ditto.
1423         * config/rs6000/darwin.h: Ditto.
1424         * config/rs6000/linux.h: Ditto.
1425         * config/rs6000/linux64.h: Ditto.
1426         * config/s390/tpf.h: Ditto.
1427         * config/sol2-10.h: Ditto.
1428         * config/sol2.h: Ditto.
1429         * config/vms/vms.h: Ditto.
1430         * config/vxworks.h: Ditto.
1431         * config/linux-android.c (linux_android_libc_has_function):
1432         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1433         * config/linux-protos.h (linux_android_libc_has_function):
1434         New declaration.
1435         * config/i386/i386.c (ix86_libc_has_function): New.
1436         * config/i386/i386-protos.h
1437         (ix86_libc_has_function): New declaration.
1438         * config/i386/i386.md
1439         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1440         ("isinf<mode>2): Likewise.
1441         * convert.c (convert_to_integer): Using new target hook
1442         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1443         TARGET_C99_FUNCTIONS.
1444         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1445         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1446         * coretypes.h (function_class): New enum for different
1447         classes of functions.
1448         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1449         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1450         (TARGET_HAS_SINCOS): Likewise.
1451         (TARGET_LIBC_HAS_FUNCTION): New.
1452         * doc/tm.texi: Regenerated.
1453         * targhooks.h (default_libc_has_function): New declaration.
1454         (no_c99_libc_has_function): Ditto.
1455         (gnu_libc_has_function): Ditto.
1456         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1457         and TARGET_HAS_SINCOS.
1459 2013-08-18  Jan Hubicka  <jh@suse.cz>
1461         * Makeifle-in (ipa-devirt.o): New.
1462         (GTFILES): Add ipa-utils.h and ipa-devirt.c
1463         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1464         (analyze_functions): Look into possible targets of polymorphic call.
1465         * dumpfile.c (dump_files): Add type-inheritance dump.
1466         * dumpfile.h (TDI_inheritance): New.
1467         * ipa-devirt.c: New file.
1468         * ipa-utils.h (odr_type_d): Forward declare.
1469         (odr_type): New type.
1470         (build_type_inheritance_graph): Declare.
1471         (possible_polymorphic_call_targets): Declare and introduce inline
1472         variant when only edge is pased.
1473         (dump_possible_polymorphic_call_targets): Likewise.
1474         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1475         * tree.c (type_in_anonymous_namespace_p): Break out from ...
1476         (types_same_for_odr): ... here.
1477         * tree.h (type_in_anonymous_namespace_p): Declare.
1479 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
1481         PR tree-optimization/58006
1482         * tree-parloops.c (take_address_of): Don't ICE if get_name
1483         returns NULL.
1484         (eliminate_local_variables_stmt): Remove clobber stmts.
1486 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
1488         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1489         error message is issued for an alias to undefined symbol.
1491 2013-08-18  Jan Hubicka  <jh@suse.cz>
1493         * cgraph.c (cgraph_create_indirect_edge): Discover
1494         polymorphic calls and record basic info into indirect_info.
1495         * gimple-fold.c (gimple_fold_call): When doing BINFO based
1496         devirtualization, ignore objc function calls.
1497         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1498         call with no parm index info.
1499         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1500         * tree.c (virtual_method_call_p): New function.
1501         * tree.h (virtual_method_call_p): Declare.
1503 2013-08-16  Jan Hubicka  <jh@suse.cz>
1505         PR middle-end/58179
1506         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1508 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
1510         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1511         attribute.
1513 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1515         * gengtype.c (type_for_name): Add special-case support for
1516         locating types within the "gcc::" namespace.
1517         (open_base_files): Emit a "using namespace gcc" directive.
1519 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1521         PR target/58160
1522         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1523         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1525         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1526         array instead of each individual operand as a separate argument.
1527         (emit_fusion_gpr_load): Likewise.
1528         (expand_fusion_gpr_load): Add new function declaration.
1530         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1531         signature to have the operands passed as an array, instead of as
1532         separate arguments.  Allow ZERO_EXTEND to be in the memory
1533         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
1534         depend on the register live/dead flags when peepholes are run.
1535         (expand_fusion_gpr_load): New function to be called from the
1536         peephole2 pass, to change the register that addis sets to be the
1537         target register.
1538         (emit_fusion_gpr_load): Change the calling signature to have the
1539         operands passed as an array, instead of as separate arguments.
1540         Allow ZERO_EXTEND to be in the memory address, and also
1541         SIGN_EXTEND if -mpower8-fusion-sign.
1543         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1544         unspec enumeration.
1545         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1546         adjust the register addis loads up in the peephole2 pass.  Do not
1547         depend on the register live/dead state when the peephole pass is done.
1549 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1551         * gengtype.c (create_user_defined_type): Ensure that the kind
1552         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1553         declaration is seen before the GTY((user)) marking.
1555 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1557         PR target/58105
1558         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1560 2013-08-16  Jan Hubicka  <jh@suse.cz>
1562         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1563         arugment expected_type.
1564         (gimple_fold_call): Use it.
1565         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1566         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1567         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1568         (try_make_edge_direct_virtual_call): Likewise.
1569         * tree.c (obj_type_ref_class): New.
1570         * tree.h (obj_type_ref_class): Use it.
1572 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1574         * sched-vis.c (rtl_slim_pp_initialized): Remove.
1575         (rtl_slim_pp): Likewise.
1576         (init_rtl_slim_pretty_print): Likewise.
1577         (dump_value_slim):  Don't call it.  Use local pretty printer.
1578         (dump_insn_slim): Likewise.
1579         (dump_rtl_slim): Likewise.
1580         (str_pattern_slim): Likewise.
1581         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1582         Simplify.
1584 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
1586         PR tree-optimization/58164
1587         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1588         walk gimple_goto_dest of GIMPLE_GOTO.
1590         PR tree-optimization/58165
1591         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1592         bi_call must be the last stmt in a bb, don't split_block, instead
1593         use fallthru edge from it and give up if there is none.
1594         Release conds vector when returning early.
1596 2013-08-14  Xinliang David Li  <davidxl@google.com>
1598         * config/i386/i386.c (ix86_option_override_internal):
1599         Remove unused variable and field.
1601 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1603         PR target/57949
1604         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1605         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1606         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1607         and Linux, correct BLKmode alignment when 128-bit alignment is
1608         required and compatibility flag is not set.
1609         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1610         for zero-size arguments when compatibility flag is not set.
1612 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1614         PR tree-optimization/58145
1615         * tree-sra.c (build_ref_for_offset): If prev_base has
1616         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1618 2013-08-14  Xinliang David Li  <davidxl@google.com>
1620         * config/i386/i386.c (ix86_option_override_internal):
1621         Fix uninitialized variable error.
1623 2013-08-14  Xinliang David Li  <davidxl@google.com>
1625         * config/i386/i386.opt: Define two new options.
1626         * config/i386/x86-tune.def: Add arch selector field in macros.
1627         * config/i386/i386.h: Adjust macro definition.
1628         * config/i386/i386.c (ix86_option_override_internal):
1629         Refactor the code.
1630         (parse_mtune_ctrl_str): New function.
1631         (set_ix86_tune_features): New function.
1632         (ix86_function_specific_restore): Call the new helper function.
1634 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
1636         PR rtl-optimization/57662
1637         * sel-sched.c (code_motion_process_successors): When the current insn
1638         is removed after the recursive traversal, break from the loop.
1639         Add comments and debug printouts.
1641 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1642             Alexandre Oliva  <aoliva@redhat.com>
1644         PR target/58067
1645         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1646         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1647         there also UNSPEC_PLTOFF.
1649 2013-08-14  Marek Polacek  <polacek@redhat.com>
1651         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1652         NUM_CONDITIONS bit positions.
1654 2013-08-13  Cary Coutant  <ccoutant@google.com>
1656         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1657         (attr_checksum): Hash vector contents instead of pointer.
1658         (attr_checksum_ordered): Likewise.
1660 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
1662         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1663         when Pmode != word_mode.  Add length_address attribute.
1664         (sse3_monitor_<mode>): Merge from sse3_monitor and
1665         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
1666         Pmode != word_mode.  Update insn length attribute.
1667         * config/i386/i386.c (ix86_option_override_internal): Update
1668         ix86_gen_monitor selection for merged sse3_monitor insn.
1670 2013-08-13  Julian Brown  <julian@codesourcery.com>
1672         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1673         perform invalid legitimization on greater-than-word-size modes for
1674         TARGET_E500_DOUBLE.
1676 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
1678         * ira.c (setup_class_translate_array): Use aclass instead of cl
1679         for classes not fully covered by allocno classes.
1681 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
1683         PR tree-optimization/57661
1684         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1685         * tree-inline.c (tree_function_versioning): Initialize it.
1686         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1687         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1688         that is not being copied.
1690         PR sanitizer/56417
1691         * asan.c (instrument_strlen_call): Fix typo in comment.
1692         Use char * type even for the lhs of POINTER_PLUS_EXPR.
1694 2013-08-13  Steve Ellcey  <sellcey@mips.com>
1696         * config/mips/mips.md (prefetch): Use lw instead of ld on
1697         loongson in 32bit mode.
1699 2013-08-13  Nick Clifton  <nickc@redhat.com>
1701         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1703 2013-08-13  Jan Hubicka  <jh@suse.cz>
1705         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1706         introduced edge; fix typo in sanity check.
1707         (cgraph_resolve_speculation): Export; improve diagnostic.
1708         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1709         speculation at type mismatch.
1710         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1711         (cgraph_resolve_speculation): Declare.
1712         (symtab_can_be_discarded): New function.
1713         * value-prof.c (gimple_ic_transform): Remove actual transform code.
1714         * ipa-inline-transform.c (speculation_removed): New global var.
1715         (clone_inlined_nodes): See if speculation can be removed.
1716         (inline_call): If speculations was removed, we growths may not match.
1717         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1718         (speculation_useful_p): New function.
1719         (resolve_noninline_speculation): New function.
1720         (inline_small_functions): Resolve useless speculations.
1721         * ipa-inline.h (speculation_useful_p): Declare
1722         * ipa.c (can_replace_by_local_alias): Simplify.
1723         (ipa_profile): Produce speculative calls in non-lto, too;
1724         add simple cost model; produce local aliases.
1726 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
1728         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1729         PASS_MANAGER_H.
1731 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
1733         * config/i386/i386.c (ix86_function_versions): Use error + inform.
1735 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
1737         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1738         iterator instead of X87MODEF.
1740 2013-08-12  Perez Read  <netfirewall@gmail.com>
1742         PR target/58132
1743         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
1744         operand 0 for intel asm alternative.
1745         (*movabs<mode>_2): Ditto for operand 1.
1747 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
1749         * config/aarch64/arm_none.h
1750         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1752 2013-08-12  Nick Clifton  <nickc@redhat.com>
1754         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1756 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1758         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1759         expand for QI/HImode operand to produce more effictive code for
1760         unsigned char(short) --> float(double) conversion.
1762 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1764         * doc/invoke.texi: Mention that -ftls-model does not force the final
1765         model.
1767 2013-08-12  Marek Polacek  <polacek@redhat.com>
1768             Marc Glisse  <marc.glisse@inria.fr>
1770         PR tree-optimization/57980
1771         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1772         when creating -1 constant.
1774 2013-08-10  Jan Hubicka  <jh@suse.cz>
1776         Workaround binutils PR14342.
1777         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1778         (gimple_init_edge_profiler): Likewise.
1779         (gimple_gen_ic_func_profiler): Likewise.
1781 2013-08-09  Jan Hubicka  <jh@suse.cz>
1783         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1785 2013-08-09  Xinliang David Li  <davidxl@google.com>
1787         * config/i386/stringop.def: New file.
1788         * config/i386/stringop.opt: New file.
1789         * config/i386/i386-opts.h: Include stringopt.def.
1790         * config/i386/i386.opt: Include stringopt.opt.
1791         * config/i386/i386.c (ix86_option_override_internal):
1792         Override default size based stringop inline strategies with options.
1793         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1794         New function.
1796 2013-08-09  Jan Hubicka  <jh@suse.cz>
1798         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1800 2013-08-09  Jan Hubicka  <jh@suse.cz>
1802         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1803         CGRAPH_FREQ_MAX.
1804         (dump_cgraph_node): Dump profile-id.
1805         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1806         and common_target_probability.
1807         * lto-cgraph.c (lto_output_edge): Stream common targets.
1808         (lto_output_node): Stream profile ids.
1809         (input_node): Stream profile ids.
1810         (input_edge): Stream common targets.
1811         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1812         * ipa.c: Include value-prof.h
1813         (ipa_profile_generate_summary): Turn indirect call statement histograms
1814         into common targets.
1815         (ipa_profile): Turn common targets into speculative edges.
1817 2013-08-09  Jan Hubicka  <jh@suse.cz>
1819         * cgraph.h (cgraph_node): Add profile_id.
1820         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1821         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1822         (del_node_map): Update.
1823         (find_func_by_funcdef_no): Replace by ...
1824         (find_func_by_profile_id): ... this one.
1825         (gimple_ic_transform): Do not remove useful histograms when
1826         speculation is not done; dump info when indirect call removal
1827         can happen at LTO.
1828         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1829         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1830         (__gcov_indirect_call_profiler_v2): .. this one.
1831         * profile.h (init_node_map): Update.
1832         * coverage.c (coverage_compute_profile_id): New function.
1833         * coverage.h (coverage_compute_profile_id): Declare.
1834         * tree-profile.c (init_ic_make_global_vars): Make
1835         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1836         (gimple_init_edge_profiler): Update prototype of
1837         __gcov_indirect_call_profiler.
1838         (gimple_gen_ic_func_profiler): Simplify.
1839         (tree_profiling): Use init_node_map
1841 2013-08-09  Jan Hubicka  <jh@suse.cz>
1843         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1844         non-speculative refs.
1845         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1846         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1847         (cgraph_set_call_stmt): Likewise.
1848         (cgraph_create_edge_1): Fix release checking compilatoin;
1849         clear lto_stmt_uid.
1850         (cgraph_free_edge): Free indirect info.
1851         (cgraph_turn_edge_to_speculative): New function.
1852         (cgraph_speculative_call_info): New function.
1853         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1854         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1855         (dump_cgraph_node): Dump speculation.
1856         (verify_edge_count_and_frequency): Accept speculative edges.
1857         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1858         (verify_cgraph_node): Handle speculation.
1859         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1860         (cgraph_set_call_stmt): Update prototype.
1861         (cgraph_make_edge_direct): Update prototype.
1862         (cgraph_speculative_call_info): Declare.
1863         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1864         to change; update call of ipa_find_references.
1865         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1866         lto_stmt_uid and speculative flags.
1867         (ipa_dump_references): Dump speculation.
1868         (ipa_clone_references): Clone speculative flag.
1869         (ipa_clone_referring): Likewise.
1870         (ipa_clone_ref): New function.
1871         (ipa_find_reference): Look into lto_stmt_uids
1872         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1873         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1874         (ipa_find_reference): Update declaration.
1875         (ipa_clone_ref): Declare.
1876         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1877         stream speculative flag.
1878         (lto_output_ref): Stream statements uids and speculation.
1879         (input_ref): Likewise.
1880         (input_edge): Stream speuclation.
1881         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1882         (cgraph_set_call_stmt_including_clones): Handle speculation.
1883         * ipa-inline.c (heap_edge_removal_hook): New function.
1884         (inline_small_functions): Register it.
1885         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1886         also initialize refs.
1887         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1888         edge to change.
1889         (try_make_edge_direct_simple_call): Likewise.
1890         (try_make_edge_direct_simple_call): Likewise.
1891         (update_indirect_edges_after_inlining): Likewise.
1892         (remove_described_reference): Look proper lto_stmt_uid.
1893         (propagate_controlled_uses): Likewise.
1894         (propagate_controlled_uses): Liekwise.
1895         * tree-inline.c (copy_bb): Copy speculative edges.
1896         (redirect_all_calls): New function.
1897         (copy_cfg_body): Do redirection after loop info is updated.
1898         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1900 2013-08-09  Jan Hubicka  <jh@suse.cz>
1902         * lto-streamer-out.c (output_function): Renumber PHIs.
1903         * lto-streamer-in.c (input_function): Likewise.
1905 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1907         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1908         (get_lane_unsigned): Likewise.
1909         (dup_lane_scalar): Likewise.
1910         (get_lane): enable for VALL.
1911         * config/aarch64/aarch64-simd.md
1912         (aarch64_dup_lane_scalar<mode>): Remove.
1913         (aarch64_get_lane_signed<mode>): Likewise.
1914         (aarch64_get_lane_unsigned<mode>): Likewise.
1915         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1916         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1917         (aarch64_get_lane<mode>): Enable for all vector modes.
1918         (aarch64_get_lanedi): Remove misleading constraints.
1919         * config/aarch64/arm_neon.h
1920         (__aarch64_vget_lane_any): Define.
1921         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1922         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1923         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1924         * config/aarch64/iterators.md (VDQQH): New.
1925         (VDQQHS): Likewise.
1926         (vwcore): Likewise.
1928 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1930         * configure.ac: Add GAS check for LEON instructions on SPARC.
1931         * configure: Regenerate.
1932         * config.in: Likewise.
1933         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1934         sparc*-*-* block.
1935         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1936         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1937         for LEON or LEON3.
1938         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1939         (AS_LEON_FLAG): New macro.
1940         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1941         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1942         Deal with LEON and LEON3 for the memory model.
1943         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1944         (atomic_compare_and_swap<mode>_1): Likewise.
1945         (*atomic_compare_and_swap<mode>_1): Likewise.
1947 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1949         * config/arm/neon.md (vcond): Fix floating-point vector
1950         comparisons against 0.
1952 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1954         * lra-constraints.c (emit_spill_move): Remove assert.
1955         (process_alt_operands): Add more debugging
1956         output.  Increase reject for spilling into memory.  Decrease
1957         reject for reloading scratch.
1958         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1960 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1962         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1963         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1964         micromips incompatible.  Add nan2008.
1965         (MULTILIB_DIRNAMES): Add nan2008.
1966         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1967         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1968         and micromips incompatible.  Add nan2008.
1969         (MULTILIB_DIRNAMES): Add nan2008.
1970         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1972 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1974         PR rtl-optimization/58079
1975         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1976         simplify_comparison has widened a comparison with an integer.
1978 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1980         * config/arm/neon.md (movmisalign<mode>): Disable when we
1981         don't allow unaligned accesses.
1982         (*movmisalign<mode>_neon_store): Likewise.
1983         (*movmisalign<mode>_neon_load): Likewise.
1984         (*movmisalign<mode>_neon_store): Likewise.
1985         (*movmisalign<mode>_neon_load): Likewise.
1987 2013-08-08  Jan Hubicka  <jh@suse.cz>
1989         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1990         (make_pass_rebuild_cgraph_edges): Also clear references.
1991         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1992         * ipa-inline-transform.c (inline_transform): Remove all references
1993         after inlining.
1994         * cgraphunit.c (expand_function): Remove all references after
1995         expansion.
1996         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1997         (ipa_find_reference): Rewrite to iterator.
1998         (remove_stmt_references): Likewise.
1999         (ipa_clear_stmts_in_references): New function.
2000         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
2001         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
2002         clear references.
2003         * ipa-split.c (split_function): Remove references in split function.
2005 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2007         PR target/57431
2008         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
2009         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
2011 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2013         PR target/56979
2014         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2015         suggested mode for the assignment isn't compatible with the
2016         registers required.
2018 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2020         PR target/58065
2021         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2023 2013-08-07  Xinliang David Li  <davidxl@google.com>
2025         * config/i386/i386.opt: New option -mtune-ctrl=.
2026         * config/i386/x86-tune.def: New file.
2027         * config/i386/i386.h: include x86-tune.def.
2028         * config/i386/i386.c (ix86_option_override_internal):
2029         Parsing -mtune-ctrl= option and set tune features.
2031 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
2033         PR other/12081
2034         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2035         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2036         to gen_2arg_fn_t.
2038 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2040         * rtl.h (update_alignments): Declare.
2041         * final.c (grow_label_align): New function extracted from...
2042         (shorten_branches): ...here.  Call it.
2043         (update_alignments): New function.
2044         * reorg.c (sibling_labels): New variable.
2045         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
2046         the new label along with it onto the sibling_labels vector.
2047         (fill_simple_delay_slots): Adjust call to get_label_before.
2048         (fill_slots_from_thread): Likewise.
2049         (relax_delay_slots): Likewise.
2050         (make_return_insns): Likewise.
2051         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2053 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2055         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2056         document its semantics.
2057         (diagnostic_report_diagnostic): Adjust accordingly.
2059 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2061         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2062         (sparc_option_override): ...and port to new C++ pass API.
2063         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2065 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
2067         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2069 2013-08-06  Caroline Tice  <cmtice@google.com>
2071         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2072         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2073         * tree-pass.h: Add pass_vtable_verify.
2074         * varasm.c (assemble_variable): Add code to properly set the comdat
2075         section and name for the .vtable_map_vars section.
2076         (assemble_vtyv_preinit_initializer): New function.
2077         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
2078         LINK_ONCE flag.
2079         * output.h: Add function decl for assemble_vtv_preinit_initializer.
2080         * vtable-verify.c: New file.
2081         * vtable-verify.h: New file.
2082         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2083         initialiation levels.
2084         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2085         * passes.def: Insert pass_vtable_verify.
2086         * aclocal.m4: Reorder includes.
2087         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
2088         -fvtv-counts options.
2089         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2090         as appropriate, if -fvtable-verify=... is used.
2091         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2092         -fvtable-verify=... is used.
2093         * Makefile.in (OBJS):  Add vtable-verify.o to list.
2094         (vtable-verify.o): Add new build rule.
2095         (GTFILES): Add vtable-verify.c to list.
2096         * common.opt (fvtable-verify=): New flag.
2097         (vtv_priority): Values for fvtable-verify= flag.
2098         (fvtv-counts): New flag.
2099         (fvtv-debug): New flag.
2100         * tree.h (save_vtable_map_decl): New extern function decl.
2102 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2104         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2105         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2106         (pass_data_rl78_devirt): ...new pass_data instance and...
2107         (make_pass_rl78_devirt): ...new function.
2108         (rl78_asm_file_start): Port pass registration to new C++ API.
2110 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2112         * coretypes.h (rtl_opt_pass): Add.
2113         (gcc::context): Add.
2114         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2115         (epiphany_init): Port to new C++ pass API.
2116         (epiphany_optimize_mode_switching): Likewise.
2117         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2118         (pass_manager::get_pass_mode_switching): New.
2119         (pass_manager::get_pass_peephole2): New.
2120         * mode-switching.c (pass_mode_switching): Add clone method.
2121         * recog.c (pass_peephole2): Add clone method.
2122         (pass_split_all_insns): Add clone method.
2124 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
2126         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2127         (mips_option_override): ...here, porting to new C++ API for passes.
2129 2013-08-06  Jan Hubicka  <jh@suse.cz>
2131         * cgraph.c (cgraph_get_body): New function based on lto.c
2132         implementation.
2133         * cgraph.h (cgraph_get_body): Declare.
2134         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2135         LTO paths.
2136         * cgraphunit.c (expand_function): Get body prior expanding.
2137         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2138         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2139         really need.
2140         * passes.c (do_per_function_toporder): Get body.
2141         * tree-inline.c (expand_call_inline): Get body prior inlining it.
2142         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2144 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2146         PR fortran/57987
2147         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2148         is not re-finalized.  Rename second parameter to no_collect.
2150 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2152         PR middle-end/58041
2153         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2154         MEM_REF has proper alignment information.
2156 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
2158         PR other/12081
2159         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2160         class insn_gen_fn.
2161         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2162         rtx (*) (rtx, ...) with insn_gen_fn.
2163         * genoutput.c (output_insn_data): Cast gen_? function pointers to
2164         insn_gen_fn::stored_funcptr.  Add initializer braces.
2166 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2168         Rewrite how instances of passes are cloned to remove assumptions
2169         about their sizes (thus allowing pass subclasses to have
2170         additional data fields, albeit non-GC-managed ones at this point).
2172         * passes.c (make_pass_instance): Now that passes have clone
2173         methods, rewrite this function to eliminate XNEW and memcpy
2174         calls that used hardcoded sizes.  Since this function no longer
2175         creates pass instances, rename it to...
2176         (add_pass_instance): ...this.  Document the old way that passes were
2177         numbered and flagged, and rework this function to continue using it.
2178         (next_pass_1): Add an initial_pass argument for use by
2179         add_pass_instance.
2180         (position_pass): When adding multiple instances of a pass, use
2181         the pass's clone method, rather than relying on the XNEW/memcpy
2182         within the former make_pass_instance (now add_pass_instance).
2183         (pass_manager::pass_manager): When invoking next_pass_1, also supply
2184         the initial instance of the current pass within the pass manager.
2186 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2188         This is the automated part of the conversion of passes from C
2189         structs to C++ classes.
2191         Patch autogenerated by refactor_passes.py from
2192         https://github.com/davidmalcolm/gcc-refactoring-scripts
2193         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2195         * asan.c (pass_asan): Convert from a global struct to a subclass of
2196         gimple_opt_pass along with...
2197         (pass_data_asan): ...new pass_data instance and...
2198         (make_pass_asan): ...new function.
2199         (pass_asan_O0): Convert from a global struct to a subclass of
2200         gimple_opt_pass along with...
2201         (pass_data_asan_O0): ...new pass_data instance and...
2202         (make_pass_asan_O0): ...new function.
2203         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2204         subclass of rtl_opt_pass along with...
2205         (pass_data_inc_dec): ...new pass_data instance and...
2206         (make_pass_inc_dec): ...new function.
2207         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2208         a subclass of rtl_opt_pass along with...
2209         (pass_data_reorder_blocks): ...new pass_data instance and...
2210         (make_pass_reorder_blocks): ...new function.
2211         (pass_duplicate_computed_gotos): Convert from a global struct to a
2212         subclass of rtl_opt_pass along with...
2213         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2214         (make_pass_duplicate_computed_gotos): ...new function.
2215         (pass_partition_blocks): Convert from a global struct to a subclass of
2216         rtl_opt_pass along with...
2217         (pass_data_partition_blocks): ...new pass_data instance and...
2218         (make_pass_partition_blocks): ...new function.
2219         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2220         struct to a subclass of rtl_opt_pass along with...
2221         (pass_data_branch_target_load_optimize1): ...new pass_data instance
2222         and...
2223         (make_pass_branch_target_load_optimize1): ...new function.
2224         (pass_branch_target_load_optimize2): Convert from a global struct to a
2225         subclass of rtl_opt_pass along with...
2226         (pass_data_branch_target_load_optimize2): ...new pass_data instance
2227         and...
2228         (make_pass_branch_target_load_optimize2): ...new function.
2229         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2230         of rtl_opt_pass along with...
2231         (pass_data_jump): ...new pass_data instance and...
2232         (make_pass_jump): ...new function.
2233         (pass_jump2): Convert from a global struct to a subclass of
2234         rtl_opt_pass along with...
2235         (pass_data_jump2): ...new pass_data instance and...
2236         (make_pass_jump2): ...new function.
2237         * cfgexpand.c (pass_expand): Convert from a global struct to a
2238         subclass of rtl_opt_pass along with...
2239         (pass_data_expand): ...new pass_data instance and...
2240         (make_pass_expand): ...new function.
2241         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2242         of rtl_opt_pass along with...
2243         (pass_data_free_cfg): ...new pass_data instance and...
2244         (make_pass_free_cfg): ...new function.
2245         (pass_into_cfg_layout_mode): Convert from a global struct to a
2246         subclass of rtl_opt_pass along with...
2247         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2248         (make_pass_into_cfg_layout_mode): ...new function.
2249         (pass_outof_cfg_layout_mode): Convert from a global struct to a
2250         subclass of rtl_opt_pass along with...
2251         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2252         (make_pass_outof_cfg_layout_mode): ...new function.
2253         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2254         struct to a subclass of gimple_opt_pass along with...
2255         (pass_data_build_cgraph_edges): ...new pass_data instance and...
2256         (make_pass_build_cgraph_edges): ...new function.
2257         (pass_rebuild_cgraph_edges): Convert from a global struct to a
2258         subclass of gimple_opt_pass along with...
2259         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2260         (make_pass_rebuild_cgraph_edges): ...new function.
2261         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2262         subclass of gimple_opt_pass along with...
2263         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2264         and...
2265         (make_pass_remove_cgraph_callee_edges): ...new function.
2266         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2267         struct to a subclass of rtl_opt_pass along with...
2268         (pass_data_stack_adjustments): ...new pass_data instance and...
2269         (make_pass_stack_adjustments): ...new function.
2270         * combine.c (pass_combine): Convert from a global struct to a subclass
2271         of rtl_opt_pass along with...
2272         (pass_data_combine): ...new pass_data instance and...
2273         (make_pass_combine): ...new function.
2274         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2275         global struct to a subclass of rtl_opt_pass along with...
2276         (pass_data_compare_elim_after_reload): ...new pass_data instance
2277         and...
2278         (make_pass_compare_elim_after_reload): ...new function.
2279         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2280         of rtl_opt_pass along with...
2281         (pass_data_rtl_cprop): ...new pass_data instance and...
2282         (make_pass_rtl_cprop): ...new function.
2283         * cse.c (pass_cse): Convert from a global struct to a subclass of
2284         rtl_opt_pass along with...
2285         (pass_data_cse): ...new pass_data instance and...
2286         (make_pass_cse): ...new function.
2287         (pass_cse2): Convert from a global struct to a subclass of
2288         rtl_opt_pass along with...
2289         (pass_data_cse2): ...new pass_data instance and...
2290         (make_pass_cse2): ...new function.
2291         (pass_cse_after_global_opts): Convert from a global struct to a
2292         subclass of rtl_opt_pass along with...
2293         (pass_data_cse_after_global_opts): ...new pass_data instance and...
2294         (make_pass_cse_after_global_opts): ...new function.
2295         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2296         of rtl_opt_pass along with...
2297         (pass_data_ud_rtl_dce): ...new pass_data instance and...
2298         (make_pass_ud_rtl_dce): ...new function.
2299         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2300         rtl_opt_pass along with...
2301         (pass_data_fast_rtl_dce): ...new pass_data instance and...
2302         (make_pass_fast_rtl_dce): ...new function.
2303         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2304         a subclass of rtl_opt_pass along with...
2305         (pass_data_df_initialize_opt): ...new pass_data instance and...
2306         (make_pass_df_initialize_opt): ...new function.
2307         (pass_df_initialize_no_opt): Convert from a global struct to a
2308         subclass of rtl_opt_pass along with...
2309         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2310         (make_pass_df_initialize_no_opt): ...new function.
2311         (pass_df_finish): Convert from a global struct to a subclass of
2312         rtl_opt_pass along with...
2313         (pass_data_df_finish): ...new pass_data instance and...
2314         (make_pass_df_finish): ...new function.
2315         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2316         rtl_opt_pass along with...
2317         (pass_data_rtl_dse1): ...new pass_data instance and...
2318         (make_pass_rtl_dse1): ...new function.
2319         (pass_rtl_dse2): Convert from a global struct to a subclass of
2320         rtl_opt_pass along with...
2321         (pass_data_rtl_dse2): ...new pass_data instance and...
2322         (make_pass_rtl_dse2): ...new function.
2323         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2324         subclass of rtl_opt_pass along with...
2325         (pass_data_dwarf2_frame): ...new pass_data instance and...
2326         (make_pass_dwarf2_frame): ...new function.
2327         * except.c (pass_set_nothrow_function_flags): Convert from a global
2328         struct to a subclass of rtl_opt_pass along with...
2329         (pass_data_set_nothrow_function_flags): ...new pass_data instance
2330         and...
2331         (make_pass_set_nothrow_function_flags): ...new function.
2332         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2333         subclass of rtl_opt_pass along with...
2334         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2335         and...
2336         (make_pass_convert_to_eh_region_ranges): ...new function.
2337         * final.c (pass_compute_alignments): Convert from a global struct to a
2338         subclass of rtl_opt_pass along with...
2339         (pass_data_compute_alignments): ...new pass_data instance and...
2340         (make_pass_compute_alignments): ...new function.
2341         (pass_final): Convert from a global struct to a subclass of
2342         rtl_opt_pass along with...
2343         (pass_data_final): ...new pass_data instance and...
2344         (make_pass_final): ...new function.
2345         (pass_shorten_branches): Convert from a global struct to a subclass of
2346         rtl_opt_pass along with...
2347         (pass_data_shorten_branches): ...new pass_data instance and...
2348         (make_pass_shorten_branches): ...new function.
2349         (pass_clean_state): Convert from a global struct to a subclass of
2350         rtl_opt_pass along with...
2351         (pass_data_clean_state): ...new pass_data instance and...
2352         (make_pass_clean_state): ...new function.
2353         * function.c (pass_instantiate_virtual_regs): Convert from a global
2354         struct to a subclass of rtl_opt_pass along with...
2355         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2356         (make_pass_instantiate_virtual_regs): ...new function.
2357         (pass_leaf_regs): Convert from a global struct to a subclass of
2358         rtl_opt_pass along with...
2359         (pass_data_leaf_regs): ...new pass_data instance and...
2360         (make_pass_leaf_regs): ...new function.
2361         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2362         subclass of rtl_opt_pass along with...
2363         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2364         and...
2365         (make_pass_thread_prologue_and_epilogue): ...new function.
2366         (pass_match_asm_constraints): Convert from a global struct to a
2367         subclass of rtl_opt_pass along with...
2368         (pass_data_match_asm_constraints): ...new pass_data instance and...
2369         (make_pass_match_asm_constraints): ...new function.
2370         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2371         subclass of rtl_opt_pass along with...
2372         (pass_data_rtl_fwprop): ...new pass_data instance and...
2373         (make_pass_rtl_fwprop): ...new function.
2374         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2375         rtl_opt_pass along with...
2376         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2377         (make_pass_rtl_fwprop_addr): ...new function.
2378         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2379         rtl_opt_pass along with...
2380         (pass_data_rtl_pre): ...new pass_data instance and...
2381         (make_pass_rtl_pre): ...new function.
2382         (pass_rtl_hoist): Convert from a global struct to a subclass of
2383         rtl_opt_pass along with...
2384         (pass_data_rtl_hoist): ...new pass_data instance and...
2385         (make_pass_rtl_hoist): ...new function.
2386         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2387         subclass of gimple_opt_pass along with...
2388         (pass_data_lower_cf): ...new pass_data instance and...
2389         (make_pass_lower_cf): ...new function.
2390         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2391         from a global struct to a subclass of gimple_opt_pass along with...
2392         (pass_data_strength_reduction): ...new pass_data instance and...
2393         (make_pass_strength_reduction): ...new function.
2394         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2395         of rtl_opt_pass along with...
2396         (pass_data_rtl_ifcvt): ...new pass_data instance and...
2397         (make_pass_rtl_ifcvt): ...new function.
2398         (pass_if_after_combine): Convert from a global struct to a subclass of
2399         rtl_opt_pass along with...
2400         (pass_data_if_after_combine): ...new pass_data instance and...
2401         (make_pass_if_after_combine): ...new function.
2402         (pass_if_after_reload): Convert from a global struct to a subclass of
2403         rtl_opt_pass along with...
2404         (pass_data_if_after_reload): ...new pass_data instance and...
2405         (make_pass_if_after_reload): ...new function.
2406         * init-regs.c (pass_initialize_regs): Convert from a global struct to
2407         a subclass of rtl_opt_pass along with...
2408         (pass_data_initialize_regs): ...new pass_data instance and...
2409         (make_pass_initialize_regs): ...new function.
2410         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2411         of ipa_opt_pass_d along with...
2412         (pass_data_ipa_cp): ...new pass_data instance and...
2413         (make_pass_ipa_cp): ...new function.
2414         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2415         global struct to a subclass of gimple_opt_pass along with...
2416         (pass_data_inline_parameters): ...new pass_data instance and...
2417         (make_pass_inline_parameters): ...new function.
2418         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2419         subclass of gimple_opt_pass along with...
2420         (pass_data_early_inline): ...new pass_data instance and...
2421         (make_pass_early_inline): ...new function.
2422         (pass_ipa_inline): Convert from a global struct to a subclass of
2423         ipa_opt_pass_d along with...
2424         (pass_data_ipa_inline): ...new pass_data instance and...
2425         (make_pass_ipa_inline): ...new function.
2426         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2427         struct to a subclass of gimple_opt_pass along with...
2428         (pass_data_local_pure_const): ...new pass_data instance and...
2429         (make_pass_local_pure_const): ...new function.
2430         (pass_ipa_pure_const): Convert from a global struct to a subclass of
2431         ipa_opt_pass_d along with...
2432         (pass_data_ipa_pure_const): ...new pass_data instance and...
2433         (make_pass_ipa_pure_const): ...new function.
2434         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2435         to a subclass of ipa_opt_pass_d along with...
2436         (pass_data_ipa_reference): ...new pass_data instance and...
2437         (make_pass_ipa_reference): ...new function.
2438         * ipa-split.c (pass_split_functions): Convert from a global struct to
2439         a subclass of gimple_opt_pass along with...
2440         (pass_data_split_functions): ...new pass_data instance and...
2441         (make_pass_split_functions): ...new function.
2442         (pass_feedback_split_functions): Convert from a global struct to a
2443         subclass of gimple_opt_pass along with...
2444         (pass_data_feedback_split_functions): ...new pass_data instance and...
2445         (make_pass_feedback_split_functions): ...new function.
2446         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2447         global struct to a subclass of simple_ipa_opt_pass along with...
2448         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2449         instance and...
2450         (make_pass_ipa_function_and_variable_visibility): ...new function.
2451         (pass_ipa_free_inline_summary): Convert from a global struct to a
2452         subclass of simple_ipa_opt_pass along with...
2453         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2454         (make_pass_ipa_free_inline_summary): ...new function.
2455         (pass_ipa_whole_program_visibility): Convert from a global struct to a
2456         subclass of ipa_opt_pass_d along with...
2457         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2458         and...
2459         (make_pass_ipa_whole_program_visibility): ...new function.
2460         (pass_ipa_profile): Convert from a global struct to a subclass of
2461         ipa_opt_pass_d along with...
2462         (pass_data_ipa_profile): ...new pass_data instance and...
2463         (make_pass_ipa_profile): ...new function.
2464         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2465         ipa_opt_pass_d along with...
2466         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2467         (make_pass_ipa_cdtor_merge): ...new function.
2468         * ira.c (pass_ira): Convert from a global struct to a subclass of
2469         rtl_opt_pass along with...
2470         (pass_data_ira): ...new pass_data instance and...
2471         (make_pass_ira): ...new function.
2472         (pass_reload): Convert from a global struct to a subclass of
2473         rtl_opt_pass along with...
2474         (pass_data_reload): ...new pass_data instance and...
2475         (make_pass_reload): ...new function.
2476         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2477         subclass of rtl_opt_pass along with...
2478         (pass_data_cleanup_barriers): ...new pass_data instance and...
2479         (make_pass_cleanup_barriers): ...new function.
2480         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2481         of rtl_opt_pass along with...
2482         (pass_data_loop2): ...new pass_data instance and...
2483         (make_pass_loop2): ...new function.
2484         (pass_rtl_loop_init): Convert from a global struct to a subclass of
2485         rtl_opt_pass along with...
2486         (pass_data_rtl_loop_init): ...new pass_data instance and...
2487         (make_pass_rtl_loop_init): ...new function.
2488         (pass_rtl_loop_done): Convert from a global struct to a subclass of
2489         rtl_opt_pass along with...
2490         (pass_data_rtl_loop_done): ...new pass_data instance and...
2491         (make_pass_rtl_loop_done): ...new function.
2492         (pass_rtl_move_loop_invariants): Convert from a global struct to a
2493         subclass of rtl_opt_pass along with...
2494         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2495         (make_pass_rtl_move_loop_invariants): ...new function.
2496         (pass_rtl_unswitch): Convert from a global struct to a subclass of
2497         rtl_opt_pass along with...
2498         (pass_data_rtl_unswitch): ...new pass_data instance and...
2499         (make_pass_rtl_unswitch): ...new function.
2500         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2501         subclass of rtl_opt_pass along with...
2502         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2503         and...
2504         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2505         (pass_rtl_doloop): Convert from a global struct to a subclass of
2506         rtl_opt_pass along with...
2507         (pass_data_rtl_doloop): ...new pass_data instance and...
2508         (make_pass_rtl_doloop): ...new function.
2509         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2510         a subclass of rtl_opt_pass along with...
2511         (pass_data_lower_subreg): ...new pass_data instance and...
2512         (make_pass_lower_subreg): ...new function.
2513         (pass_lower_subreg2): Convert from a global struct to a subclass of
2514         rtl_opt_pass along with...
2515         (pass_data_lower_subreg2): ...new pass_data instance and...
2516         (make_pass_lower_subreg2): ...new function.
2517         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2518         struct to a subclass of ipa_opt_pass_d along with...
2519         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2520         (make_pass_ipa_lto_gimple_out): ...new function.
2521         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2522         of ipa_opt_pass_d along with...
2523         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2524         (make_pass_ipa_lto_finish_out): ...new function.
2525         * mode-switching.c (pass_mode_switching): Convert from a global struct
2526         to a subclass of rtl_opt_pass along with...
2527         (pass_data_mode_switching): ...new pass_data instance and...
2528         (make_pass_mode_switching): ...new function.
2529         * modulo-sched.c (pass_sms): Convert from a global struct to a
2530         subclass of rtl_opt_pass along with...
2531         (pass_data_sms): ...new pass_data instance and...
2532         (make_pass_sms): ...new function.
2533         * omp-low.c (pass_expand_omp): Convert from a global struct to a
2534         subclass of gimple_opt_pass along with...
2535         (pass_data_expand_omp): ...new pass_data instance and...
2536         (make_pass_expand_omp): ...new function.
2537         (pass_lower_omp): Convert from a global struct to a subclass of
2538         gimple_opt_pass along with...
2539         (pass_data_lower_omp): ...new pass_data instance and...
2540         (make_pass_lower_omp): ...new function.
2541         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2542         of gimple_opt_pass along with...
2543         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2544         (make_pass_diagnose_omp_blocks): ...new function.
2545         * passes.c (pass_early_local_passes): Convert from a global struct to
2546         a subclass of simple_ipa_opt_pass along with...
2547         (pass_data_early_local_passes): ...new pass_data instance and...
2548         (make_pass_early_local_passes): ...new function.
2549         (pass_all_early_optimizations): Convert from a global struct to a
2550         subclass of gimple_opt_pass along with...
2551         (pass_data_all_early_optimizations): ...new pass_data instance and...
2552         (make_pass_all_early_optimizations): ...new function.
2553         (pass_all_optimizations): Convert from a global struct to a subclass
2554         of gimple_opt_pass along with...
2555         (pass_data_all_optimizations): ...new pass_data instance and...
2556         (make_pass_all_optimizations): ...new function.
2557         (pass_all_optimizations_g): Convert from a global struct to a subclass
2558         of gimple_opt_pass along with...
2559         (pass_data_all_optimizations_g): ...new pass_data instance and...
2560         (make_pass_all_optimizations_g): ...new function.
2561         (pass_rest_of_compilation): Convert from a global struct to a subclass
2562         of rtl_opt_pass along with...
2563         (pass_data_rest_of_compilation): ...new pass_data instance and...
2564         (make_pass_rest_of_compilation): ...new function.
2565         (pass_postreload): Convert from a global struct to a subclass of
2566         rtl_opt_pass along with...
2567         (pass_data_postreload): ...new pass_data instance and...
2568         (make_pass_postreload): ...new function.
2569         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2570         subclass of rtl_opt_pass along with...
2571         (pass_data_gcse2): ...new pass_data instance and...
2572         (make_pass_gcse2): ...new function.
2573         * postreload.c (pass_postreload_cse): Convert from a global struct to
2574         a subclass of rtl_opt_pass along with...
2575         (pass_data_postreload_cse): ...new pass_data instance and...
2576         (make_pass_postreload_cse): ...new function.
2577         * predict.c (pass_profile): Convert from a global struct to a subclass
2578         of gimple_opt_pass along with...
2579         (pass_data_profile): ...new pass_data instance and...
2580         (make_pass_profile): ...new function.
2581         (pass_strip_predict_hints): Convert from a global struct to a subclass
2582         of gimple_opt_pass along with...
2583         (pass_data_strip_predict_hints): ...new pass_data instance and...
2584         (make_pass_strip_predict_hints): ...new function.
2585         * recog.c (pass_peephole2): Convert from a global struct to a subclass
2586         of rtl_opt_pass along with...
2587         (pass_data_peephole2): ...new pass_data instance and...
2588         (make_pass_peephole2): ...new function.
2589         (pass_split_all_insns): Convert from a global struct to a subclass of
2590         rtl_opt_pass along with...
2591         (pass_data_split_all_insns): ...new pass_data instance and...
2592         (make_pass_split_all_insns): ...new function.
2593         (pass_split_after_reload): Convert from a global struct to a subclass
2594         of rtl_opt_pass along with...
2595         (pass_data_split_after_reload): ...new pass_data instance and...
2596         (make_pass_split_after_reload): ...new function.
2597         (pass_split_before_regstack): Convert from a global struct to a
2598         subclass of rtl_opt_pass along with...
2599         (pass_data_split_before_regstack): ...new pass_data instance and...
2600         (make_pass_split_before_regstack): ...new function.
2601         (pass_split_before_sched2): Convert from a global struct to a subclass
2602         of rtl_opt_pass along with...
2603         (pass_data_split_before_sched2): ...new pass_data instance and...
2604         (make_pass_split_before_sched2): ...new function.
2605         (pass_split_for_shorten_branches): Convert from a global struct to a
2606         subclass of rtl_opt_pass along with...
2607         (pass_data_split_for_shorten_branches): ...new pass_data instance
2608         and...
2609         (make_pass_split_for_shorten_branches): ...new function.
2610         * ree.c (pass_ree): Convert from a global struct to a subclass of
2611         rtl_opt_pass along with...
2612         (pass_data_ree): ...new pass_data instance and...
2613         (make_pass_ree): ...new function.
2614         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2615         subclass of rtl_opt_pass along with...
2616         (pass_data_stack_regs): ...new pass_data instance and...
2617         (make_pass_stack_regs): ...new function.
2618         (pass_stack_regs_run): Convert from a global struct to a subclass of
2619         rtl_opt_pass along with...
2620         (pass_data_stack_regs_run): ...new pass_data instance and...
2621         (make_pass_stack_regs_run): ...new function.
2622         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2623         subclass of rtl_opt_pass along with...
2624         (pass_data_cprop_hardreg): ...new pass_data instance and...
2625         (make_pass_cprop_hardreg): ...new function.
2626         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2627         subclass of rtl_opt_pass along with...
2628         (pass_data_reginfo_init): ...new pass_data instance and...
2629         (make_pass_reginfo_init): ...new function.
2630         * regmove.c (pass_regmove): Convert from a global struct to a subclass
2631         of rtl_opt_pass along with...
2632         (pass_data_regmove): ...new pass_data instance and...
2633         (make_pass_regmove): ...new function.
2634         * regrename.c (pass_regrename): Convert from a global struct to a
2635         subclass of rtl_opt_pass along with...
2636         (pass_data_regrename): ...new pass_data instance and...
2637         (make_pass_regrename): ...new function.
2638         * reorg.c (pass_delay_slots): Convert from a global struct to a
2639         subclass of rtl_opt_pass along with...
2640         (pass_data_delay_slots): ...new pass_data instance and...
2641         (make_pass_delay_slots): ...new function.
2642         (pass_machine_reorg): Convert from a global struct to a subclass of
2643         rtl_opt_pass along with...
2644         (pass_data_machine_reorg): ...new pass_data instance and...
2645         (make_pass_machine_reorg): ...new function.
2646         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2647         of rtl_opt_pass along with...
2648         (pass_data_sched): ...new pass_data instance and...
2649         (make_pass_sched): ...new function.
2650         (pass_sched2): Convert from a global struct to a subclass of
2651         rtl_opt_pass along with...
2652         (pass_data_sched2): ...new pass_data instance and...
2653         (make_pass_sched2): ...new function.
2654         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2655         to a subclass of rtl_opt_pass along with...
2656         (pass_data_stack_ptr_mod): ...new pass_data instance and...
2657         (make_pass_stack_ptr_mod): ...new function.
2658         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2659         to a subclass of rtl_opt_pass along with...
2660         (pass_data_rtl_store_motion): ...new pass_data instance and...
2661         (make_pass_rtl_store_motion): ...new function.
2662         * tracer.c (pass_tracer): Convert from a global struct to a subclass
2663         of gimple_opt_pass along with...
2664         (pass_data_tracer): ...new pass_data instance and...
2665         (make_pass_tracer): ...new function.
2666         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2667         to a subclass of gimple_opt_pass along with...
2668         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2669         (make_pass_diagnose_tm_blocks): ...new function.
2670         (pass_lower_tm): Convert from a global struct to a subclass of
2671         gimple_opt_pass along with...
2672         (pass_data_lower_tm): ...new pass_data instance and...
2673         (make_pass_lower_tm): ...new function.
2674         (pass_tm_init): Convert from a global struct to a subclass of
2675         gimple_opt_pass along with...
2676         (pass_data_tm_init): ...new pass_data instance and...
2677         (make_pass_tm_init): ...new function.
2678         (pass_tm_mark): Convert from a global struct to a subclass of
2679         gimple_opt_pass along with...
2680         (pass_data_tm_mark): ...new pass_data instance and...
2681         (make_pass_tm_mark): ...new function.
2682         (pass_tm_edges): Convert from a global struct to a subclass of
2683         gimple_opt_pass along with...
2684         (pass_data_tm_edges): ...new pass_data instance and...
2685         (make_pass_tm_edges): ...new function.
2686         (pass_tm_memopt): Convert from a global struct to a subclass of
2687         gimple_opt_pass along with...
2688         (pass_data_tm_memopt): ...new pass_data instance and...
2689         (make_pass_tm_memopt): ...new function.
2690         (pass_ipa_tm): Convert from a global struct to a subclass of
2691         simple_ipa_opt_pass along with...
2692         (pass_data_ipa_tm): ...new pass_data instance and...
2693         (make_pass_ipa_tm): ...new function.
2694         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2695         subclass of gimple_opt_pass along with...
2696         (pass_data_call_cdce): ...new pass_data instance and...
2697         (make_pass_call_cdce): ...new function.
2698         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2699         subclass of gimple_opt_pass along with...
2700         (pass_data_build_cfg): ...new pass_data instance and...
2701         (make_pass_build_cfg): ...new function.
2702         (pass_split_crit_edges): Convert from a global struct to a subclass of
2703         gimple_opt_pass along with...
2704         (pass_data_split_crit_edges): ...new pass_data instance and...
2705         (make_pass_split_crit_edges): ...new function.
2706         (pass_warn_function_return): Convert from a global struct to a
2707         subclass of gimple_opt_pass along with...
2708         (pass_data_warn_function_return): ...new pass_data instance and...
2709         (make_pass_warn_function_return): ...new function.
2710         (pass_warn_function_noreturn): Convert from a global struct to a
2711         subclass of gimple_opt_pass along with...
2712         (pass_data_warn_function_noreturn): ...new pass_data instance and...
2713         (make_pass_warn_function_noreturn): ...new function.
2714         (pass_warn_unused_result): Convert from a global struct to a subclass
2715         of gimple_opt_pass along with...
2716         (pass_data_warn_unused_result): ...new pass_data instance and...
2717         (make_pass_warn_unused_result): ...new function.
2718         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2719         a subclass of gimple_opt_pass along with...
2720         (pass_data_merge_phi): ...new pass_data instance and...
2721         (make_pass_merge_phi): ...new function.
2722         * tree-complex.c (pass_lower_complex): Convert from a global struct to
2723         a subclass of gimple_opt_pass along with...
2724         (pass_data_lower_complex): ...new pass_data instance and...
2725         (make_pass_lower_complex): ...new function.
2726         (pass_lower_complex_O0): Convert from a global struct to a subclass of
2727         gimple_opt_pass along with...
2728         (pass_data_lower_complex_O0): ...new pass_data instance and...
2729         (make_pass_lower_complex_O0): ...new function.
2730         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2731         subclass of gimple_opt_pass along with...
2732         (pass_data_lower_eh): ...new pass_data instance and...
2733         (make_pass_lower_eh): ...new function.
2734         (pass_refactor_eh): Convert from a global struct to a subclass of
2735         gimple_opt_pass along with...
2736         (pass_data_refactor_eh): ...new pass_data instance and...
2737         (make_pass_refactor_eh): ...new function.
2738         (pass_lower_resx): Convert from a global struct to a subclass of
2739         gimple_opt_pass along with...
2740         (pass_data_lower_resx): ...new pass_data instance and...
2741         (make_pass_lower_resx): ...new function.
2742         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2743         of gimple_opt_pass along with...
2744         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2745         (make_pass_lower_eh_dispatch): ...new function.
2746         (pass_cleanup_eh): Convert from a global struct to a subclass of
2747         gimple_opt_pass along with...
2748         (pass_data_cleanup_eh): ...new pass_data instance and...
2749         (make_pass_cleanup_eh): ...new function.
2750         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2751         to a subclass of simple_ipa_opt_pass along with...
2752         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2753         (make_pass_ipa_lower_emutls): ...new function.
2754         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2755         a subclass of gimple_opt_pass along with...
2756         (pass_data_if_conversion): ...new pass_data instance and...
2757         (make_pass_if_conversion): ...new function.
2758         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2759         subclass of gimple_opt_pass along with...
2760         (pass_data_build_ssa): ...new pass_data instance and...
2761         (make_pass_build_ssa): ...new function.
2762         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2763         global struct to a subclass of gimple_opt_pass along with...
2764         (pass_data_loop_distribution): ...new pass_data instance and...
2765         (make_pass_loop_distribution): ...new function.
2766         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2767         subclass of gimple_opt_pass along with...
2768         (pass_data_mudflap_1): ...new pass_data instance and...
2769         (make_pass_mudflap_1): ...new function.
2770         (pass_mudflap_2): Convert from a global struct to a subclass of
2771         gimple_opt_pass along with...
2772         (pass_data_mudflap_2): ...new pass_data instance and...
2773         (make_pass_mudflap_2): ...new function.
2774         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2775         subclass of gimple_opt_pass along with...
2776         (pass_data_mudflap_1): ...new pass_data instance and...
2777         (make_pass_mudflap_1): ...new function.
2778         (pass_mudflap_2): Convert from a global struct to a subclass of
2779         gimple_opt_pass along with...
2780         (pass_data_mudflap_2): ...new pass_data instance and...
2781         (make_pass_mudflap_2): ...new function.
2782         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2783         gimple_opt_pass along with...
2784         (pass_data_nrv): ...new pass_data instance and...
2785         (make_pass_nrv): ...new function.
2786         (pass_return_slot): Convert from a global struct to a subclass of
2787         gimple_opt_pass along with...
2788         (pass_data_return_slot): ...new pass_data instance and...
2789         (make_pass_return_slot): ...new function.
2790         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2791         to a subclass of gimple_opt_pass along with...
2792         (pass_data_object_sizes): ...new pass_data instance and...
2793         (make_pass_object_sizes): ...new function.
2794         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2795         global struct to a subclass of gimple_opt_pass along with...
2796         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2797         and...
2798         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2799         (pass_fixup_cfg): Convert from a global struct to a subclass of
2800         gimple_opt_pass along with...
2801         (pass_data_fixup_cfg): ...new pass_data instance and...
2802         (make_pass_fixup_cfg): ...new function.
2803         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2804         (make_pass_mudflap_1): ...new function.
2805         (pass_mudflap_2): Replace declaration with that of...
2806         (make_pass_mudflap_2): ...new function.
2807         (pass_asan): Replace declaration with that of...
2808         (make_pass_asan): ...new function.
2809         (pass_asan_O0): Replace declaration with that of...
2810         (make_pass_asan_O0): ...new function.
2811         (pass_tsan): Replace declaration with that of...
2812         (make_pass_tsan): ...new function.
2813         (pass_tsan_O0): Replace declaration with that of...
2814         (make_pass_tsan_O0): ...new function.
2815         (pass_lower_cf): Replace declaration with that of...
2816         (make_pass_lower_cf): ...new function.
2817         (pass_refactor_eh): Replace declaration with that of...
2818         (make_pass_refactor_eh): ...new function.
2819         (pass_lower_eh): Replace declaration with that of...
2820         (make_pass_lower_eh): ...new function.
2821         (pass_lower_eh_dispatch): Replace declaration with that of...
2822         (make_pass_lower_eh_dispatch): ...new function.
2823         (pass_lower_resx): Replace declaration with that of...
2824         (make_pass_lower_resx): ...new function.
2825         (pass_build_cfg): Replace declaration with that of...
2826         (make_pass_build_cfg): ...new function.
2827         (pass_early_tree_profile): Replace declaration with that of...
2828         (make_pass_early_tree_profile): ...new function.
2829         (pass_cleanup_eh): Replace declaration with that of...
2830         (make_pass_cleanup_eh): ...new function.
2831         (pass_sra): Replace declaration with that of...
2832         (make_pass_sra): ...new function.
2833         (pass_sra_early): Replace declaration with that of...
2834         (make_pass_sra_early): ...new function.
2835         (pass_early_ipa_sra): Replace declaration with that of...
2836         (make_pass_early_ipa_sra): ...new function.
2837         (pass_tail_recursion): Replace declaration with that of...
2838         (make_pass_tail_recursion): ...new function.
2839         (pass_tail_calls): Replace declaration with that of...
2840         (make_pass_tail_calls): ...new function.
2841         (pass_tree_loop): Replace declaration with that of...
2842         (make_pass_tree_loop): ...new function.
2843         (pass_tree_loop_init): Replace declaration with that of...
2844         (make_pass_tree_loop_init): ...new function.
2845         (pass_lim): Replace declaration with that of...
2846         (make_pass_lim): ...new function.
2847         (pass_tree_unswitch): Replace declaration with that of...
2848         (make_pass_tree_unswitch): ...new function.
2849         (pass_predcom): Replace declaration with that of...
2850         (make_pass_predcom): ...new function.
2851         (pass_iv_canon): Replace declaration with that of...
2852         (make_pass_iv_canon): ...new function.
2853         (pass_scev_cprop): Replace declaration with that of...
2854         (make_pass_scev_cprop): ...new function.
2855         (pass_empty_loop): Replace declaration with that of...
2856         (make_pass_empty_loop): ...new function.
2857         (pass_record_bounds): Replace declaration with that of...
2858         (make_pass_record_bounds): ...new function.
2859         (pass_graphite): Replace declaration with that of...
2860         (make_pass_graphite): ...new function.
2861         (pass_graphite_transforms): Replace declaration with that of...
2862         (make_pass_graphite_transforms): ...new function.
2863         (pass_if_conversion): Replace declaration with that of...
2864         (make_pass_if_conversion): ...new function.
2865         (pass_loop_distribution): Replace declaration with that of...
2866         (make_pass_loop_distribution): ...new function.
2867         (pass_vectorize): Replace declaration with that of...
2868         (make_pass_vectorize): ...new function.
2869         (pass_slp_vectorize): Replace declaration with that of...
2870         (make_pass_slp_vectorize): ...new function.
2871         (pass_complete_unroll): Replace declaration with that of...
2872         (make_pass_complete_unroll): ...new function.
2873         (pass_complete_unrolli): Replace declaration with that of...
2874         (make_pass_complete_unrolli): ...new function.
2875         (pass_parallelize_loops): Replace declaration with that of...
2876         (make_pass_parallelize_loops): ...new function.
2877         (pass_loop_prefetch): Replace declaration with that of...
2878         (make_pass_loop_prefetch): ...new function.
2879         (pass_iv_optimize): Replace declaration with that of...
2880         (make_pass_iv_optimize): ...new function.
2881         (pass_tree_loop_done): Replace declaration with that of...
2882         (make_pass_tree_loop_done): ...new function.
2883         (pass_ch): Replace declaration with that of...
2884         (make_pass_ch): ...new function.
2885         (pass_ccp): Replace declaration with that of...
2886         (make_pass_ccp): ...new function.
2887         (pass_phi_only_cprop): Replace declaration with that of...
2888         (make_pass_phi_only_cprop): ...new function.
2889         (pass_build_ssa): Replace declaration with that of...
2890         (make_pass_build_ssa): ...new function.
2891         (pass_build_alias): Replace declaration with that of...
2892         (make_pass_build_alias): ...new function.
2893         (pass_build_ealias): Replace declaration with that of...
2894         (make_pass_build_ealias): ...new function.
2895         (pass_dominator): Replace declaration with that of...
2896         (make_pass_dominator): ...new function.
2897         (pass_dce): Replace declaration with that of...
2898         (make_pass_dce): ...new function.
2899         (pass_dce_loop): Replace declaration with that of...
2900         (make_pass_dce_loop): ...new function.
2901         (pass_cd_dce): Replace declaration with that of...
2902         (make_pass_cd_dce): ...new function.
2903         (pass_call_cdce): Replace declaration with that of...
2904         (make_pass_call_cdce): ...new function.
2905         (pass_merge_phi): Replace declaration with that of...
2906         (make_pass_merge_phi): ...new function.
2907         (pass_split_crit_edges): Replace declaration with that of...
2908         (make_pass_split_crit_edges): ...new function.
2909         (pass_pre): Replace declaration with that of...
2910         (make_pass_pre): ...new function.
2911         (pass_profile): Replace declaration with that of...
2912         (make_pass_profile): ...new function.
2913         (pass_strip_predict_hints): Replace declaration with that of...
2914         (make_pass_strip_predict_hints): ...new function.
2915         (pass_lower_complex_O0): Replace declaration with that of...
2916         (make_pass_lower_complex_O0): ...new function.
2917         (pass_lower_complex): Replace declaration with that of...
2918         (make_pass_lower_complex): ...new function.
2919         (pass_lower_vector): Replace declaration with that of...
2920         (make_pass_lower_vector): ...new function.
2921         (pass_lower_vector_ssa): Replace declaration with that of...
2922         (make_pass_lower_vector_ssa): ...new function.
2923         (pass_lower_omp): Replace declaration with that of...
2924         (make_pass_lower_omp): ...new function.
2925         (pass_diagnose_omp_blocks): Replace declaration with that of...
2926         (make_pass_diagnose_omp_blocks): ...new function.
2927         (pass_expand_omp): Replace declaration with that of...
2928         (make_pass_expand_omp): ...new function.
2929         (pass_expand_omp_ssa): Replace declaration with that of...
2930         (make_pass_expand_omp_ssa): ...new function.
2931         (pass_object_sizes): Replace declaration with that of...
2932         (make_pass_object_sizes): ...new function.
2933         (pass_strlen): Replace declaration with that of...
2934         (make_pass_strlen): ...new function.
2935         (pass_fold_builtins): Replace declaration with that of...
2936         (make_pass_fold_builtins): ...new function.
2937         (pass_stdarg): Replace declaration with that of...
2938         (make_pass_stdarg): ...new function.
2939         (pass_early_warn_uninitialized): Replace declaration with that of...
2940         (make_pass_early_warn_uninitialized): ...new function.
2941         (pass_late_warn_uninitialized): Replace declaration with that of...
2942         (make_pass_late_warn_uninitialized): ...new function.
2943         (pass_cse_reciprocals): Replace declaration with that of...
2944         (make_pass_cse_reciprocals): ...new function.
2945         (pass_cse_sincos): Replace declaration with that of...
2946         (make_pass_cse_sincos): ...new function.
2947         (pass_optimize_bswap): Replace declaration with that of...
2948         (make_pass_optimize_bswap): ...new function.
2949         (pass_optimize_widening_mul): Replace declaration with that of...
2950         (make_pass_optimize_widening_mul): ...new function.
2951         (pass_warn_function_return): Replace declaration with that of...
2952         (make_pass_warn_function_return): ...new function.
2953         (pass_warn_function_noreturn): Replace declaration with that of...
2954         (make_pass_warn_function_noreturn): ...new function.
2955         (pass_cselim): Replace declaration with that of...
2956         (make_pass_cselim): ...new function.
2957         (pass_phiopt): Replace declaration with that of...
2958         (make_pass_phiopt): ...new function.
2959         (pass_forwprop): Replace declaration with that of...
2960         (make_pass_forwprop): ...new function.
2961         (pass_phiprop): Replace declaration with that of...
2962         (make_pass_phiprop): ...new function.
2963         (pass_tree_ifcombine): Replace declaration with that of...
2964         (make_pass_tree_ifcombine): ...new function.
2965         (pass_dse): Replace declaration with that of...
2966         (make_pass_dse): ...new function.
2967         (pass_nrv): Replace declaration with that of...
2968         (make_pass_nrv): ...new function.
2969         (pass_rename_ssa_copies): Replace declaration with that of...
2970         (make_pass_rename_ssa_copies): ...new function.
2971         (pass_sink_code): Replace declaration with that of...
2972         (make_pass_sink_code): ...new function.
2973         (pass_fre): Replace declaration with that of...
2974         (make_pass_fre): ...new function.
2975         (pass_check_data_deps): Replace declaration with that of...
2976         (make_pass_check_data_deps): ...new function.
2977         (pass_copy_prop): Replace declaration with that of...
2978         (make_pass_copy_prop): ...new function.
2979         (pass_vrp): Replace declaration with that of...
2980         (make_pass_vrp): ...new function.
2981         (pass_uncprop): Replace declaration with that of...
2982         (make_pass_uncprop): ...new function.
2983         (pass_return_slot): Replace declaration with that of...
2984         (make_pass_return_slot): ...new function.
2985         (pass_reassoc): Replace declaration with that of...
2986         (make_pass_reassoc): ...new function.
2987         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2988         (make_pass_rebuild_cgraph_edges): ...new function.
2989         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2990         (make_pass_remove_cgraph_callee_edges): ...new function.
2991         (pass_build_cgraph_edges): Replace declaration with that of...
2992         (make_pass_build_cgraph_edges): ...new function.
2993         (pass_local_pure_const): Replace declaration with that of...
2994         (make_pass_local_pure_const): ...new function.
2995         (pass_tracer): Replace declaration with that of...
2996         (make_pass_tracer): ...new function.
2997         (pass_warn_unused_result): Replace declaration with that of...
2998         (make_pass_warn_unused_result): ...new function.
2999         (pass_diagnose_tm_blocks): Replace declaration with that of...
3000         (make_pass_diagnose_tm_blocks): ...new function.
3001         (pass_lower_tm): Replace declaration with that of...
3002         (make_pass_lower_tm): ...new function.
3003         (pass_tm_init): Replace declaration with that of...
3004         (make_pass_tm_init): ...new function.
3005         (pass_tm_mark): Replace declaration with that of...
3006         (make_pass_tm_mark): ...new function.
3007         (pass_tm_memopt): Replace declaration with that of...
3008         (make_pass_tm_memopt): ...new function.
3009         (pass_tm_edges): Replace declaration with that of...
3010         (make_pass_tm_edges): ...new function.
3011         (pass_split_functions): Replace declaration with that of...
3012         (make_pass_split_functions): ...new function.
3013         (pass_feedback_split_functions): Replace declaration with that of...
3014         (make_pass_feedback_split_functions): ...new function.
3015         (pass_strength_reduction): Replace declaration with that of...
3016         (make_pass_strength_reduction): ...new function.
3017         (pass_ipa_lower_emutls): Replace declaration with that of...
3018         (make_pass_ipa_lower_emutls): ...new function.
3019         (pass_ipa_function_and_variable_visibility): Replace declaration with
3020         that of...
3021         (make_pass_ipa_function_and_variable_visibility): ...new function.
3022         (pass_ipa_tree_profile): Replace declaration with that of...
3023         (make_pass_ipa_tree_profile): ...new function.
3024         (pass_early_local_passes): Replace declaration with that of...
3025         (make_pass_early_local_passes): ...new function.
3026         (pass_ipa_whole_program_visibility): Replace declaration with that
3027         of...
3028         (make_pass_ipa_whole_program_visibility): ...new function.
3029         (pass_ipa_lto_gimple_out): Replace declaration with that of...
3030         (make_pass_ipa_lto_gimple_out): ...new function.
3031         (pass_ipa_increase_alignment): Replace declaration with that of...
3032         (make_pass_ipa_increase_alignment): ...new function.
3033         (pass_ipa_inline): Replace declaration with that of...
3034         (make_pass_ipa_inline): ...new function.
3035         (pass_ipa_free_lang_data): Replace declaration with that of...
3036         (make_pass_ipa_free_lang_data): ...new function.
3037         (pass_ipa_free_inline_summary): Replace declaration with that of...
3038         (make_pass_ipa_free_inline_summary): ...new function.
3039         (pass_ipa_cp): Replace declaration with that of...
3040         (make_pass_ipa_cp): ...new function.
3041         (pass_ipa_reference): Replace declaration with that of...
3042         (make_pass_ipa_reference): ...new function.
3043         (pass_ipa_pure_const): Replace declaration with that of...
3044         (make_pass_ipa_pure_const): ...new function.
3045         (pass_ipa_pta): Replace declaration with that of...
3046         (make_pass_ipa_pta): ...new function.
3047         (pass_ipa_lto_finish_out): Replace declaration with that of...
3048         (make_pass_ipa_lto_finish_out): ...new function.
3049         (pass_ipa_tm): Replace declaration with that of...
3050         (make_pass_ipa_tm): ...new function.
3051         (pass_ipa_profile): Replace declaration with that of...
3052         (make_pass_ipa_profile): ...new function.
3053         (pass_ipa_cdtor_merge): Replace declaration with that of...
3054         (make_pass_ipa_cdtor_merge): ...new function.
3055         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3056         of...
3057         (make_pass_cleanup_cfg_post_optimizing): ...new function.
3058         (pass_init_datastructures): Replace declaration with that of...
3059         (make_pass_init_datastructures): ...new function.
3060         (pass_fixup_cfg): Replace declaration with that of...
3061         (make_pass_fixup_cfg): ...new function.
3062         (pass_expand): Replace declaration with that of...
3063         (make_pass_expand): ...new function.
3064         (pass_instantiate_virtual_regs): Replace declaration with that of...
3065         (make_pass_instantiate_virtual_regs): ...new function.
3066         (pass_rtl_fwprop): Replace declaration with that of...
3067         (make_pass_rtl_fwprop): ...new function.
3068         (pass_rtl_fwprop_addr): Replace declaration with that of...
3069         (make_pass_rtl_fwprop_addr): ...new function.
3070         (pass_jump): Replace declaration with that of...
3071         (make_pass_jump): ...new function.
3072         (pass_jump2): Replace declaration with that of...
3073         (make_pass_jump2): ...new function.
3074         (pass_lower_subreg): Replace declaration with that of...
3075         (make_pass_lower_subreg): ...new function.
3076         (pass_cse): Replace declaration with that of...
3077         (make_pass_cse): ...new function.
3078         (pass_fast_rtl_dce): Replace declaration with that of...
3079         (make_pass_fast_rtl_dce): ...new function.
3080         (pass_ud_rtl_dce): Replace declaration with that of...
3081         (make_pass_ud_rtl_dce): ...new function.
3082         (pass_rtl_dce): Replace declaration with that of...
3083         (make_pass_rtl_dce): ...new function.
3084         (pass_rtl_dse1): Replace declaration with that of...
3085         (make_pass_rtl_dse1): ...new function.
3086         (pass_rtl_dse2): Replace declaration with that of...
3087         (make_pass_rtl_dse2): ...new function.
3088         (pass_rtl_dse3): Replace declaration with that of...
3089         (make_pass_rtl_dse3): ...new function.
3090         (pass_rtl_cprop): Replace declaration with that of...
3091         (make_pass_rtl_cprop): ...new function.
3092         (pass_rtl_pre): Replace declaration with that of...
3093         (make_pass_rtl_pre): ...new function.
3094         (pass_rtl_hoist): Replace declaration with that of...
3095         (make_pass_rtl_hoist): ...new function.
3096         (pass_rtl_store_motion): Replace declaration with that of...
3097         (make_pass_rtl_store_motion): ...new function.
3098         (pass_cse_after_global_opts): Replace declaration with that of...
3099         (make_pass_cse_after_global_opts): ...new function.
3100         (pass_rtl_ifcvt): Replace declaration with that of...
3101         (make_pass_rtl_ifcvt): ...new function.
3102         (pass_into_cfg_layout_mode): Replace declaration with that of...
3103         (make_pass_into_cfg_layout_mode): ...new function.
3104         (pass_outof_cfg_layout_mode): Replace declaration with that of...
3105         (make_pass_outof_cfg_layout_mode): ...new function.
3106         (pass_loop2): Replace declaration with that of...
3107         (make_pass_loop2): ...new function.
3108         (pass_rtl_loop_init): Replace declaration with that of...
3109         (make_pass_rtl_loop_init): ...new function.
3110         (pass_rtl_move_loop_invariants): Replace declaration with that of...
3111         (make_pass_rtl_move_loop_invariants): ...new function.
3112         (pass_rtl_unswitch): Replace declaration with that of...
3113         (make_pass_rtl_unswitch): ...new function.
3114         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3115         (make_pass_rtl_unroll_and_peel_loops): ...new function.
3116         (pass_rtl_doloop): Replace declaration with that of...
3117         (make_pass_rtl_doloop): ...new function.
3118         (pass_rtl_loop_done): Replace declaration with that of...
3119         (make_pass_rtl_loop_done): ...new function.
3120         (pass_web): Replace declaration with that of...
3121         (make_pass_web): ...new function.
3122         (pass_cse2): Replace declaration with that of...
3123         (make_pass_cse2): ...new function.
3124         (pass_df_initialize_opt): Replace declaration with that of...
3125         (make_pass_df_initialize_opt): ...new function.
3126         (pass_df_initialize_no_opt): Replace declaration with that of...
3127         (make_pass_df_initialize_no_opt): ...new function.
3128         (pass_reginfo_init): Replace declaration with that of...
3129         (make_pass_reginfo_init): ...new function.
3130         (pass_inc_dec): Replace declaration with that of...
3131         (make_pass_inc_dec): ...new function.
3132         (pass_stack_ptr_mod): Replace declaration with that of...
3133         (make_pass_stack_ptr_mod): ...new function.
3134         (pass_initialize_regs): Replace declaration with that of...
3135         (make_pass_initialize_regs): ...new function.
3136         (pass_combine): Replace declaration with that of...
3137         (make_pass_combine): ...new function.
3138         (pass_if_after_combine): Replace declaration with that of...
3139         (make_pass_if_after_combine): ...new function.
3140         (pass_ree): Replace declaration with that of...
3141         (make_pass_ree): ...new function.
3142         (pass_partition_blocks): Replace declaration with that of...
3143         (make_pass_partition_blocks): ...new function.
3144         (pass_match_asm_constraints): Replace declaration with that of...
3145         (make_pass_match_asm_constraints): ...new function.
3146         (pass_regmove): Replace declaration with that of...
3147         (make_pass_regmove): ...new function.
3148         (pass_split_all_insns): Replace declaration with that of...
3149         (make_pass_split_all_insns): ...new function.
3150         (pass_fast_rtl_byte_dce): Replace declaration with that of...
3151         (make_pass_fast_rtl_byte_dce): ...new function.
3152         (pass_lower_subreg2): Replace declaration with that of...
3153         (make_pass_lower_subreg2): ...new function.
3154         (pass_mode_switching): Replace declaration with that of...
3155         (make_pass_mode_switching): ...new function.
3156         (pass_sms): Replace declaration with that of...
3157         (make_pass_sms): ...new function.
3158         (pass_sched): Replace declaration with that of...
3159         (make_pass_sched): ...new function.
3160         (pass_ira): Replace declaration with that of...
3161         (make_pass_ira): ...new function.
3162         (pass_reload): Replace declaration with that of...
3163         (make_pass_reload): ...new function.
3164         (pass_clean_state): Replace declaration with that of...
3165         (make_pass_clean_state): ...new function.
3166         (pass_branch_prob): Replace declaration with that of...
3167         (make_pass_branch_prob): ...new function.
3168         (pass_value_profile_transformations): Replace declaration with that
3169         of...
3170         (make_pass_value_profile_transformations): ...new function.
3171         (pass_postreload_cse): Replace declaration with that of...
3172         (make_pass_postreload_cse): ...new function.
3173         (pass_gcse2): Replace declaration with that of...
3174         (make_pass_gcse2): ...new function.
3175         (pass_split_after_reload): Replace declaration with that of...
3176         (make_pass_split_after_reload): ...new function.
3177         (pass_branch_target_load_optimize1): Replace declaration with that
3178         of...
3179         (make_pass_branch_target_load_optimize1): ...new function.
3180         (pass_thread_prologue_and_epilogue): Replace declaration with that
3181         of...
3182         (make_pass_thread_prologue_and_epilogue): ...new function.
3183         (pass_stack_adjustments): Replace declaration with that of...
3184         (make_pass_stack_adjustments): ...new function.
3185         (pass_peephole2): Replace declaration with that of...
3186         (make_pass_peephole2): ...new function.
3187         (pass_if_after_reload): Replace declaration with that of...
3188         (make_pass_if_after_reload): ...new function.
3189         (pass_regrename): Replace declaration with that of...
3190         (make_pass_regrename): ...new function.
3191         (pass_cprop_hardreg): Replace declaration with that of...
3192         (make_pass_cprop_hardreg): ...new function.
3193         (pass_reorder_blocks): Replace declaration with that of...
3194         (make_pass_reorder_blocks): ...new function.
3195         (pass_branch_target_load_optimize2): Replace declaration with that
3196         of...
3197         (make_pass_branch_target_load_optimize2): ...new function.
3198         (pass_leaf_regs): Replace declaration with that of...
3199         (make_pass_leaf_regs): ...new function.
3200         (pass_split_before_sched2): Replace declaration with that of...
3201         (make_pass_split_before_sched2): ...new function.
3202         (pass_compare_elim_after_reload): Replace declaration with that of...
3203         (make_pass_compare_elim_after_reload): ...new function.
3204         (pass_sched2): Replace declaration with that of...
3205         (make_pass_sched2): ...new function.
3206         (pass_stack_regs): Replace declaration with that of...
3207         (make_pass_stack_regs): ...new function.
3208         (pass_stack_regs_run): Replace declaration with that of...
3209         (make_pass_stack_regs_run): ...new function.
3210         (pass_df_finish): Replace declaration with that of...
3211         (make_pass_df_finish): ...new function.
3212         (pass_compute_alignments): Replace declaration with that of...
3213         (make_pass_compute_alignments): ...new function.
3214         (pass_duplicate_computed_gotos): Replace declaration with that of...
3215         (make_pass_duplicate_computed_gotos): ...new function.
3216         (pass_variable_tracking): Replace declaration with that of...
3217         (make_pass_variable_tracking): ...new function.
3218         (pass_free_cfg): Replace declaration with that of...
3219         (make_pass_free_cfg): ...new function.
3220         (pass_machine_reorg): Replace declaration with that of...
3221         (make_pass_machine_reorg): ...new function.
3222         (pass_cleanup_barriers): Replace declaration with that of...
3223         (make_pass_cleanup_barriers): ...new function.
3224         (pass_delay_slots): Replace declaration with that of...
3225         (make_pass_delay_slots): ...new function.
3226         (pass_split_for_shorten_branches): Replace declaration with that of...
3227         (make_pass_split_for_shorten_branches): ...new function.
3228         (pass_split_before_regstack): Replace declaration with that of...
3229         (make_pass_split_before_regstack): ...new function.
3230         (pass_convert_to_eh_region_ranges): Replace declaration with that
3231         of...
3232         (make_pass_convert_to_eh_region_ranges): ...new function.
3233         (pass_shorten_branches): Replace declaration with that of...
3234         (make_pass_shorten_branches): ...new function.
3235         (pass_set_nothrow_function_flags): Replace declaration with that of...
3236         (make_pass_set_nothrow_function_flags): ...new function.
3237         (pass_dwarf2_frame): Replace declaration with that of...
3238         (make_pass_dwarf2_frame): ...new function.
3239         (pass_final): Replace declaration with that of...
3240         (make_pass_final): ...new function.
3241         (pass_rtl_seqabstr): Replace declaration with that of...
3242         (make_pass_rtl_seqabstr): ...new function.
3243         (pass_release_ssa_names): Replace declaration with that of...
3244         (make_pass_release_ssa_names): ...new function.
3245         (pass_early_inline): Replace declaration with that of...
3246         (make_pass_early_inline): ...new function.
3247         (pass_inline_parameters): Replace declaration with that of...
3248         (make_pass_inline_parameters): ...new function.
3249         (pass_update_address_taken): Replace declaration with that of...
3250         (make_pass_update_address_taken): ...new function.
3251         (pass_convert_switch): Replace declaration with that of...
3252         (make_pass_convert_switch): ...new function.
3253         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3254         to a subclass of simple_ipa_opt_pass along with...
3255         (pass_data_ipa_tree_profile): ...new pass_data instance and...
3256         (make_pass_ipa_tree_profile): ...new function.
3257         * tree-sra.c (pass_sra_early): Convert from a global struct to a
3258         subclass of gimple_opt_pass along with...
3259         (pass_data_sra_early): ...new pass_data instance and...
3260         (make_pass_sra_early): ...new function.
3261         (pass_sra): Convert from a global struct to a subclass of
3262         gimple_opt_pass along with...
3263         (pass_data_sra): ...new pass_data instance and...
3264         (make_pass_sra): ...new function.
3265         (pass_early_ipa_sra): Convert from a global struct to a subclass of
3266         gimple_opt_pass along with...
3267         (pass_data_early_ipa_sra): ...new pass_data instance and...
3268         (make_pass_early_ipa_sra): ...new function.
3269         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3270         subclass of gimple_opt_pass along with...
3271         (pass_data_ccp): ...new pass_data instance and...
3272         (make_pass_ccp): ...new function.
3273         (pass_fold_builtins): Convert from a global struct to a subclass of
3274         gimple_opt_pass along with...
3275         (pass_data_fold_builtins): ...new pass_data instance and...
3276         (make_pass_fold_builtins): ...new function.
3277         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3278         subclass of gimple_opt_pass along with...
3279         (pass_data_copy_prop): ...new pass_data instance and...
3280         (make_pass_copy_prop): ...new function.
3281         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3282         global struct to a subclass of gimple_opt_pass along with...
3283         (pass_data_rename_ssa_copies): ...new pass_data instance and...
3284         (make_pass_rename_ssa_copies): ...new function.
3285         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3286         subclass of gimple_opt_pass along with...
3287         (pass_data_dce): ...new pass_data instance and...
3288         (make_pass_dce): ...new function.
3289         (pass_dce_loop): Convert from a global struct to a subclass of
3290         gimple_opt_pass along with...
3291         (pass_data_dce_loop): ...new pass_data instance and...
3292         (make_pass_dce_loop): ...new function.
3293         (pass_cd_dce): Convert from a global struct to a subclass of
3294         gimple_opt_pass along with...
3295         (pass_data_cd_dce): ...new pass_data instance and...
3296         (make_pass_cd_dce): ...new function.
3297         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3298         subclass of gimple_opt_pass along with...
3299         (pass_data_dominator): ...new pass_data instance and...
3300         (make_pass_dominator): ...new function.
3301         (pass_phi_only_cprop): Convert from a global struct to a subclass of
3302         gimple_opt_pass along with...
3303         (pass_data_phi_only_cprop): ...new pass_data instance and...
3304         (make_pass_phi_only_cprop): ...new function.
3305         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3306         subclass of gimple_opt_pass along with...
3307         (pass_data_dse): ...new pass_data instance and...
3308         (make_pass_dse): ...new function.
3309         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3310         a subclass of gimple_opt_pass along with...
3311         (pass_data_forwprop): ...new pass_data instance and...
3312         (make_pass_forwprop): ...new function.
3313         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3314         struct to a subclass of gimple_opt_pass along with...
3315         (pass_data_tree_ifcombine): ...new pass_data instance and...
3316         (make_pass_tree_ifcombine): ...new function.
3317         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3318         subclass of gimple_opt_pass along with...
3319         (pass_data_ch): ...new pass_data instance and...
3320         (make_pass_ch): ...new function.
3321         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3322         subclass of gimple_opt_pass along with...
3323         (pass_data_tree_loop): ...new pass_data instance and...
3324         (make_pass_tree_loop): ...new function.
3325         (pass_tree_loop_init): Convert from a global struct to a subclass of
3326         gimple_opt_pass along with...
3327         (pass_data_tree_loop_init): ...new pass_data instance and...
3328         (make_pass_tree_loop_init): ...new function.
3329         (pass_lim): Convert from a global struct to a subclass of
3330         gimple_opt_pass along with...
3331         (pass_data_lim): ...new pass_data instance and...
3332         (make_pass_lim): ...new function.
3333         (pass_tree_unswitch): Convert from a global struct to a subclass of
3334         gimple_opt_pass along with...
3335         (pass_data_tree_unswitch): ...new pass_data instance and...
3336         (make_pass_tree_unswitch): ...new function.
3337         (pass_predcom): Convert from a global struct to a subclass of
3338         gimple_opt_pass along with...
3339         (pass_data_predcom): ...new pass_data instance and...
3340         (make_pass_predcom): ...new function.
3341         (pass_vectorize): Convert from a global struct to a subclass of
3342         gimple_opt_pass along with...
3343         (pass_data_vectorize): ...new pass_data instance and...
3344         (make_pass_vectorize): ...new function.
3345         (pass_graphite): Convert from a global struct to a subclass of
3346         gimple_opt_pass along with...
3347         (pass_data_graphite): ...new pass_data instance and...
3348         (make_pass_graphite): ...new function.
3349         (pass_graphite_transforms): Convert from a global struct to a subclass
3350         of gimple_opt_pass along with...
3351         (pass_data_graphite_transforms): ...new pass_data instance and...
3352         (make_pass_graphite_transforms): ...new function.
3353         (pass_check_data_deps): Convert from a global struct to a subclass of
3354         gimple_opt_pass along with...
3355         (pass_data_check_data_deps): ...new pass_data instance and...
3356         (make_pass_check_data_deps): ...new function.
3357         (pass_iv_canon): Convert from a global struct to a subclass of
3358         gimple_opt_pass along with...
3359         (pass_data_iv_canon): ...new pass_data instance and...
3360         (make_pass_iv_canon): ...new function.
3361         (pass_scev_cprop): Convert from a global struct to a subclass of
3362         gimple_opt_pass along with...
3363         (pass_data_scev_cprop): ...new pass_data instance and...
3364         (make_pass_scev_cprop): ...new function.
3365         (pass_record_bounds): Convert from a global struct to a subclass of
3366         gimple_opt_pass along with...
3367         (pass_data_record_bounds): ...new pass_data instance and...
3368         (make_pass_record_bounds): ...new function.
3369         (pass_complete_unroll): Convert from a global struct to a subclass of
3370         gimple_opt_pass along with...
3371         (pass_data_complete_unroll): ...new pass_data instance and...
3372         (make_pass_complete_unroll): ...new function.
3373         (pass_complete_unrolli): Convert from a global struct to a subclass of
3374         gimple_opt_pass along with...
3375         (pass_data_complete_unrolli): ...new pass_data instance and...
3376         (make_pass_complete_unrolli): ...new function.
3377         (pass_parallelize_loops): Convert from a global struct to a subclass
3378         of gimple_opt_pass along with...
3379         (pass_data_parallelize_loops): ...new pass_data instance and...
3380         (make_pass_parallelize_loops): ...new function.
3381         (pass_loop_prefetch): Convert from a global struct to a subclass of
3382         gimple_opt_pass along with...
3383         (pass_data_loop_prefetch): ...new pass_data instance and...
3384         (make_pass_loop_prefetch): ...new function.
3385         (pass_iv_optimize): Convert from a global struct to a subclass of
3386         gimple_opt_pass along with...
3387         (pass_data_iv_optimize): ...new pass_data instance and...
3388         (make_pass_iv_optimize): ...new function.
3389         (pass_tree_loop_done): Convert from a global struct to a subclass of
3390         gimple_opt_pass along with...
3391         (pass_data_tree_loop_done): ...new pass_data instance and...
3392         (make_pass_tree_loop_done): ...new function.
3393         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3394         struct to a subclass of gimple_opt_pass along with...
3395         (pass_data_cse_reciprocals): ...new pass_data instance and...
3396         (make_pass_cse_reciprocals): ...new function.
3397         (pass_cse_sincos): Convert from a global struct to a subclass of
3398         gimple_opt_pass along with...
3399         (pass_data_cse_sincos): ...new pass_data instance and...
3400         (make_pass_cse_sincos): ...new function.
3401         (pass_optimize_bswap): Convert from a global struct to a subclass of
3402         gimple_opt_pass along with...
3403         (pass_data_optimize_bswap): ...new pass_data instance and...
3404         (make_pass_optimize_bswap): ...new function.
3405         (pass_optimize_widening_mul): Convert from a global struct to a
3406         subclass of gimple_opt_pass along with...
3407         (pass_data_optimize_widening_mul): ...new pass_data instance and...
3408         (make_pass_optimize_widening_mul): ...new function.
3409         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3410         subclass of gimple_opt_pass along with...
3411         (pass_data_phiopt): ...new pass_data instance and...
3412         (make_pass_phiopt): ...new function.
3413         (pass_cselim): Convert from a global struct to a subclass of
3414         gimple_opt_pass along with...
3415         (pass_data_cselim): ...new pass_data instance and...
3416         (make_pass_cselim): ...new function.
3417         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3418         subclass of gimple_opt_pass along with...
3419         (pass_data_phiprop): ...new pass_data instance and...
3420         (make_pass_phiprop): ...new function.
3421         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3422         subclass of gimple_opt_pass along with...
3423         (pass_data_pre): ...new pass_data instance and...
3424         (make_pass_pre): ...new function.
3425         (pass_fre): Convert from a global struct to a subclass of
3426         gimple_opt_pass along with...
3427         (pass_data_fre): ...new pass_data instance and...
3428         (make_pass_fre): ...new function.
3429         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3430         subclass of gimple_opt_pass along with...
3431         (pass_data_reassoc): ...new pass_data instance and...
3432         (make_pass_reassoc): ...new function.
3433         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3434         subclass of gimple_opt_pass along with...
3435         (pass_data_sink_code): ...new pass_data instance and...
3436         (make_pass_sink_code): ...new function.
3437         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3438         subclass of gimple_opt_pass along with...
3439         (pass_data_strlen): ...new pass_data instance and...
3440         (make_pass_strlen): ...new function.
3441         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3442         struct to a subclass of gimple_opt_pass along with...
3443         (pass_data_build_alias): ...new pass_data instance and...
3444         (make_pass_build_alias): ...new function.
3445         (pass_build_ealias): Convert from a global struct to a subclass of
3446         gimple_opt_pass along with...
3447         (pass_data_build_ealias): ...new pass_data instance and...
3448         (make_pass_build_ealias): ...new function.
3449         (pass_ipa_pta): Convert from a global struct to a subclass of
3450         simple_ipa_opt_pass along with...
3451         (pass_data_ipa_pta): ...new pass_data instance and...
3452         (make_pass_ipa_pta): ...new function.
3453         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3454         subclass of gimple_opt_pass along with...
3455         (pass_data_uncprop): ...new pass_data instance and...
3456         (make_pass_uncprop): ...new function.
3457         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3458         global struct to a subclass of gimple_opt_pass along with...
3459         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3460         (make_pass_late_warn_uninitialized): ...new function.
3461         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3462         to a subclass of gimple_opt_pass along with...
3463         (pass_data_init_datastructures): ...new pass_data instance and...
3464         (make_pass_init_datastructures): ...new function.
3465         (pass_early_warn_uninitialized): Convert from a global struct to a
3466         subclass of gimple_opt_pass along with...
3467         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3468         (make_pass_early_warn_uninitialized): ...new function.
3469         (pass_update_address_taken): Convert from a global struct to a
3470         subclass of gimple_opt_pass along with...
3471         (pass_data_update_address_taken): ...new pass_data instance and...
3472         (make_pass_update_address_taken): ...new function.
3473         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3474         struct to a subclass of gimple_opt_pass along with...
3475         (pass_data_release_ssa_names): ...new pass_data instance and...
3476         (make_pass_release_ssa_names): ...new function.
3477         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3478         subclass of gimple_opt_pass along with...
3479         (pass_data_stdarg): ...new pass_data instance and...
3480         (make_pass_stdarg): ...new function.
3481         * tree-switch-conversion.c (pass_convert_switch): Convert from a
3482         global struct to a subclass of gimple_opt_pass along with...
3483         (pass_data_convert_switch): ...new pass_data instance and...
3484         (make_pass_convert_switch): ...new function.
3485         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3486         to a subclass of gimple_opt_pass along with...
3487         (pass_data_tail_recursion): ...new pass_data instance and...
3488         (make_pass_tail_recursion): ...new function.
3489         (pass_tail_calls): Convert from a global struct to a subclass of
3490         gimple_opt_pass along with...
3491         (pass_data_tail_calls): ...new pass_data instance and...
3492         (make_pass_tail_calls): ...new function.
3493         * tree-vect-generic.c (pass_lower_vector): Convert from a global
3494         struct to a subclass of gimple_opt_pass along with...
3495         (pass_data_lower_vector): ...new pass_data instance and...
3496         (make_pass_lower_vector): ...new function.
3497         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3498         gimple_opt_pass along with...
3499         (pass_data_lower_vector_ssa): ...new pass_data instance and...
3500         (make_pass_lower_vector_ssa): ...new function.
3501         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3502         to a subclass of gimple_opt_pass along with...
3503         (pass_data_slp_vectorize): ...new pass_data instance and...
3504         (make_pass_slp_vectorize): ...new function.
3505         (pass_ipa_increase_alignment): Convert from a global struct to a
3506         subclass of simple_ipa_opt_pass along with...
3507         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3508         (make_pass_ipa_increase_alignment): ...new function.
3509         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3510         gimple_opt_pass along with...
3511         (pass_data_vrp): ...new pass_data instance and...
3512         (make_pass_vrp): ...new function.
3513         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3514         subclass of simple_ipa_opt_pass along with...
3515         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3516         (make_pass_ipa_free_lang_data): ...new function.
3517         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3518         gimple_opt_pass along with...
3519         (pass_data_tsan): ...new pass_data instance and...
3520         (make_pass_tsan): ...new function.
3521         (pass_tsan_O0): Convert from a global struct to a subclass of
3522         gimple_opt_pass along with...
3523         (pass_data_tsan_O0): ...new pass_data instance and...
3524         (make_pass_tsan_O0): ...new function.
3525         * var-tracking.c (pass_variable_tracking): Convert from a global
3526         struct to a subclass of rtl_opt_pass along with...
3527         (pass_data_variable_tracking): ...new pass_data instance and...
3528         (make_pass_variable_tracking): ...new function.
3529         * web.c (pass_web): Convert from a global struct to a subclass of
3530         rtl_opt_pass along with...
3531         (pass_data_web): ...new pass_data instance and...
3532         (make_pass_web): ...new function.
3533         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3534         declaration with that of...
3535         (make_pass_mode_switch_use): ...new function.
3536         (pass_resolve_sw_modes): Replace declaration with that of...
3537         (make_pass_resolve_sw_modes): ...new function.
3538         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3539         from a global struct to a subclass of rtl_opt_pass along with...
3540         (pass_data_mode_switch_use): ...new pass_data instance and...
3541         (make_pass_mode_switch_use): ...new function.
3542         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3543         from a global struct to a subclass of rtl_opt_pass along with...
3544         (pass_data_resolve_sw_modes): ...new pass_data instance and...
3545         (make_pass_resolve_sw_modes): ...new function.
3546         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3547         struct to a subclass of rtl_opt_pass along with...
3548         (pass_data_insert_vzeroupper): ...new pass_data instance and...
3549         (make_pass_insert_vzeroupper): ...new function.
3550         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3551         global struct to a subclass of rtl_opt_pass along with...
3552         (pass_data_work_around_errata): ...new pass_data instance and...
3553         (make_pass_work_around_errata): ...new function.
3554         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3555         struct to a subclass of rtl_opt_pass along with...
3556         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3557         (make_pass_mips_machine_reorg2): ...new function.
3559 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3561         * passes.c (pass_manager::operator new): New.
3563 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3565         Handwritten part of conversion of passes to C++ classes.
3567         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3568         (toplev.o): Add dep on PASS_MANAGER_H.
3569         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3570         of early local pases to reflect this moving from a global to a
3571         member of gcc::pass_manager.
3572         (cgraph_add_new_function): Likewise.
3573         * lto-cgraph.c (lto_output_node): Update for conversion of
3574         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3575         * passes.c (opt_pass::clone): New.
3576         (opt_pass::gate): New.
3577         (opt_pass::execute): New.
3578         (opt_pass::opt_pass): New.
3579         (pass_manager::execute_early_local_passes): New.
3580         (pass_manager::execute_pass_mode_switching): new.
3581         (finish_optimization_passes): Convert to...
3582         (pass_manager::finish_optimization_passes): ...this.
3583         (finish_optimization_passes): Update for conversion of passes to
3584         C++ classes.
3585         (register_dump_files_1): Use has_gate since we cannot portably
3586         check a vtable entry against NULL.
3587         (dump_one_pass): Likewise.
3588         (ipa_write_summaries_2): Likewise.
3589         (ipa_write_optimization_summaries_1): Likewise.
3590         (ipa_read_summaries_1): Likewise.
3591         (ipa_read_optimization_summaries_1): Likewise.
3592         (execute_ipa_stmt_fixups): Likewise.
3593         (pass_manager::pass_manager): Rewrite pass-creation, invoking
3594         pass-creation functions rather than wiring up globals, and
3595         storing the results in fields of pass_manager generated using
3596         pass-instances.def.
3597         (pass_manager::dump_profile_report): Update for conversion of
3598         passes to C++ classes.
3599         (pass_manager::execute_ipa_summary_passes): Likewise.
3600         (execute_one_ipa_transform_pass): Likewise.
3601         (execute_one_pass): Use has_gate and has_execute since we cannot
3602         portably check a vtable entry against NULL.
3603         * pass_manager.h (pass_manager::finish_optimization_passes): New.
3604         (pass_manager): Use pass-instances.def to add fields for the
3605         various pass instances.
3606         * toplev.c (finalize): Update for move of
3607         finish_optimization_passes to a method of gcc::pass_manager.
3608         * toplev.h (finish_optimization_passes): Move to method of class
3609         pass_manager.
3610         * tree-pass.h (struct pass_data): New.
3611         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3612         (opt_pass::gate): Convert to virtual function.
3613         (opt_pass::~opt_pass): New.
3614         (opt_pass::clone): New.
3615         (opt_pass::execute): Convert to virtual function.
3616         (opt_pass::opt_pass): New.
3617         (opt_pass::ctxt_): new.
3618         (gimple_opt_pass): Convert to subclass of opt_pass.
3619         (gimple_opt_pass::gimple_opt_pass): New.
3620         (rtl_opt_pass): Convert to subclass of opt_pass.
3621         (rtl_opt_pass::rtl_opt_pass): New.
3622         (ipa_opt_pass_d): Convert to subclass of opt_pass.
3623         (ipa_opt_pass_d::ipa_opt_pass_d): New.
3624         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3625         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3626         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3627         invocation of pass_mode_switching to reflect this moving from a
3628         global to a member of gcc::pass_manager.
3629         (ix86_option_override): Rework how pass_insert_vzeroupper is
3630         added to the pass_manager to reflect autogenerated changes.
3631         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3633 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
3635         PR rtl-optimization/57708
3636         * recog.c (peep2_find_free_register): Validate all regs in a
3637         multi-reg mode.
3639 2013-08-05  Jan Hubicka  <jh@suse.cz>
3641         PR lto/57602
3642         * cgraph.c (verify_cgraph_node): Accept local flags from other
3643         partitions.
3644         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3645         (function_and_variable_visibility): Likewise.
3646         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3648 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3650         * graph.c (init_graph_slim_pretty_print): Remove.
3651         (print_graph_cfg): Do not call it.  Use local pretty printer.
3652         (start_graph_dump): Likewise.
3654 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3656         * gimple-pretty-print.c (buffer): Remove.
3657         (initialized): Likewise.
3658         (maybe_init_pretty_print): Likewise.
3659         (print_gimple_stmt): Do not call it.  Use non-static local
3660         pretty_printer variable.
3661         (print_gimple_expr): Likewise.
3662         (print_gimple_seq): Likewise.
3663         (gimple_dump_bb): Likewise.
3665 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3667         * asan.c (asan_pp): Remove.
3668         (asan_pp_initialized): Likewise.
3669         (asan_pp_initialize): Likewise.
3670         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
3671         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
3672         (asan_add_global): Likewise.
3674 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3676         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
3677         * diagnostic.h (diagnostic_flush_buffer): Adjust.
3678         * pretty-print.c (pp_formatted_text_data): Likewise.
3679         (pp_indent): Rename from pp_base_indent.
3680         (pp_format): Rename from pp_base_format.
3681         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3682         (pp_format_verbatim): Rename from pp_base_format_verbatim.
3683         (pp_flush): Rename from pp_base_flush.
3684         (pp_set_line_maximum_length): Rename from
3685         pp_base_set_line_maximum_length.
3686         (pp_clear_output_area): Rename from pp_base_clear_output_area.
3687         (pp_set_prefix): Rename from pp_base_set_prefix.
3688         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3689         (pp_emit_prefix): Rename from pp_base_emit_prefix.
3690         (pp_append_text): Rename from pp_base_append_text.
3691         (pp_formatted_text): Rename from pp_base_formatted_text.
3692         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3693         (pp_remaining_character_count_for_line): Rename from
3694         pp_base_remaining_character_count_for_line.
3695         (pp_newline): Rename from pp_base_newline.
3696         (pp_character): Rename from pp_base_character.
3697         (pp_string): Rename from pp_base_string.
3698         (pp_maybe_space): Rename from pp_base_maybe_space.
3699         * asan.c (asan_pp_string): Adjust.
3700         (asan_emit_stack_protection): Likewise.
3701         (asan_add_global): Likewise.
3702         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3703         * tree-mudflap.c (mf_varname_tree): Likewise.
3704         * tree-pretty-print.c (pp_tree_identifier): Rename from
3705         pp_base_tree_identifier.
3706         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3707         Declare as function.
3709 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3711         * pretty-print.h (pp_bar_bar): New.
3712         (pp_ampersand_ampersand): Likewise.
3713         (pp_less_equal): Likewise.
3714         (pp_greater_equal): Likewise.
3715         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3716         printer functions instead of pp_string or operators and punctuators.
3717         (dump_gimple_call): Likewise.
3718         (dump_gimple_omp_for): Likewise.
3719         (dump_gimple_transaction): Likewise.
3720         (dump_gimple_phi): Likewise.
3721         (pp_gimple_stmt_1): Likewise.
3722         * sched-vis.c (print_insn): Likewise.
3723         * tree-mudflap.c (mf_varname_tree): Likewise.
3724         * tree-pretty-print.c (dump_block_node): Likewise.
3725         (dump_generic_node): Likewise.
3727 2013-08-02  Jan Hubicka  <jh@suse.cz>
3729         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3730         boundaries.
3731         * lto-streamer-out.c (tree_is_indexable): Results decls and
3732         parm decls are not indexable.
3733         (DFS_write_tree_body): Do not follow args and results.
3734         (hash_tree): Likewise.
3735         (output_functions): Rearrange so struct function is needed
3736         only when real body is output; be able to also ouptut abstract
3737         functions; output DECL_ARGUMENTS and DECL_RESULT.
3738         (lto_output): When not in WPA, ale store abstract functions.
3739         (write_symbol): Do not care about RESULT_DECL.
3740         (output_symbol_p): Handle correctly sbtract decls.
3741         * lto-streamer-in.c (input_function): Rearrange so struct
3742         function can be NULL at entry; allow streaming of
3743         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3744         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3745         sanity check during LTO.
3746         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3747         RESULT_DECl and DECL_ARGUMENTS.
3748         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3749         Likewise.
3751 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3753         * pretty-print.h (pp_underscore): New.
3754         (pp_comma): Tidy.
3755         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3756         printer functions instead of pp_character.
3757         (dump_binary_rhs): Likewise.
3758         (dump_ternary_rhs): Likewise.
3759         (dump_gimple_call_args): Likewise.
3760         (pp_points_to_solution): Likewise.
3761         (dump_gimple_call): Likewise.
3762         (dump_gimple_switch): Likewise.
3763         (dump_gimple_cond): Likewise.
3764         (dump_gimple_bind): Likewise.
3765         (dump_gimple_try): Likewise.
3766         (dump_gimple_omp_for): Likewise.
3767         (dump_gimple_omp_continue): Likewise.
3768         (dump_gimple_omp_single): Likewise.
3769         (dump_gimple_omp_sections): Likewise.
3770         (dump_gimple_omp_block): Likewise.
3771         (dump_gimple_omp_critical): Likewise.
3772         (dump_gimple_transaction): Likewise.
3773         (dump_gimple_asm): Likewise.
3774         (dump_gimple_phi): Likewise.
3775         (dump_gimple_omp_parallel): Likewise.
3776         (dump_gimple_omp_task): Likewise.
3777         (dump_gimple_omp_atomic_load): Likewise.
3778         (dump_gimple_omp_atomic_store): Likewise.
3779         (dump_gimple_mem_ops): Likewise.
3780         (pp_gimple_stmt_1): Likewise.
3781         (pp_cfg_jump): Likewise.
3782         (dump_implicit_edges): Likewise.
3783         (gimple_dump_bb_for_graph): Likewise.
3784         * graph.c (draw_cfg_node): Likewise.
3785         * langhooks.c (lhd_print_error_function): Likewise.
3786         * sched-vis.c (print_exp): Likewise.
3787         (print_value): Likewise.
3788         (print_pattern): Likewise.
3789         (print_insn): Likewise.
3790         (rtl_dump_bb_for_graph): Likewise.
3791         * tree-pretty-print.c (dump_function_declaration): Likewise.
3792         (dump_array_domain): Likewise.
3793         (dump_omp_clause): Likewise.
3794         (dump_location): Likewise.
3795         (dump_generic_node): Likewise.
3796         (print_struct_decl): Likewise.
3797         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3799 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3801         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3802         candidate table when replacing a candidate statement.
3803         (replace_rhs_if_not_dup): Likewise.
3804         (replace_one_candidate): Likewise.
3806 2013-08-02  Jan Hubicka  <jh@suse.cz>
3807             Martin Liska  <marxin.liska@gmail.com>
3809         * cgraphunit.c (add_new_function): Fix logic when adding from
3810         late IPA pass.
3811         (assemble_thunk): Rename to ...
3812         (expand_thunk); .. this one; export; get it working with
3813         general functions; make produced gimple valid.
3814         * cgraph.h (expand_thunk): Declare.
3816 2013-08-02  Jan Hubicka  <jh@suse.cz>
3818         * ipa-cp.c (gather_context_independent_values): Use
3819         ipa_get_param_move_cost.
3820         (get_replacement_map): Remove PARAM; move parameter folding
3821         into tree-inline.c
3822         (create_specialized_node): Update.
3823         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3824         assert that we have gimple body; update move_cost.
3825         (count_formal_params): Assert that we have gimple body.
3826         (ipa_dump_param): New function.
3827         (ipa_alloc_node_params): Break out from ...
3828         (ipa_initialize_node_params): ... here.
3829         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3830         (ipa_write_node_info): Stream move costs.
3831         (ipa_read_node_info): Read move costs.
3832         (ipa_update_after_lto_read): Do not recompute node params.
3833         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3834         (ipa_get_param): Check we are not in WPA.
3835         (ipa_get_param_move_cost): New.
3836         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3837         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3838         parm numbers to be present.
3840 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3842         PR rtl-optimization/58048
3843         * lra-constraints.c (process_alt_operands): Don't check asm
3844         operand on register.
3846 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3848         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3849         the implied StoreLoad barrier for atomic operations if before.
3851 2013-08-02  Jan Hubicka  <jh@suse.cz>
3852             Martin Liska  <marxin.liska@gmail.com>
3854         * cgraph.c (cgraph_function_body_availability): Do not check
3855         cgraph flags.
3856         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3857         symtab_node_availability): Declare.
3858         * ipa.c (can_replace_by_local_alias): New.
3859         (function_and_variable_visibility): Use it.
3860         * symtab.c (symtab_for_node_and_aliases,
3861         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3863 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3865         PR rtl-optimization/57963
3866         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3867         (lra_constraints): Use them.
3869 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3871         * config/arm/types.md (define_attr "type"): Add "load_acq"
3872         and "store_rel".
3873         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3874         changes.
3875         (cortex_a53_store1): Likewise.
3877 2013-08-01  Jan Hubicka  <jh@suse.cz>
3879         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3880         partitions are not needed.
3882 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3884         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3885         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3886         MAYBE_NON_Q_CLASS_P where appropriate.
3888 2013-08-01  Jan Hubicka  <jh@suse.cz>
3890         * cgraph.h (release_function_body): Declare.
3891         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3892         of unused delcarations.
3894 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3896         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3897         RTL form when subtracting a constant.
3899 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3901         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3902         Generate canonical plus rtx with negated immediate instead of minus
3903         where appropriate.
3904         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3906 2013-08-01  Jan Hubicka  <jh@suse.cz>
3908         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3909         (cgraph_release_function_body): Likewise.
3910         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3911         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3912         to used_as_abstract_origin.
3913         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3914         symbols used as abstract origins.
3915         * cgraphunit.c (analyze_functions): Update.
3916         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3917         used_as_abstract_origin.
3918         * tree-inline.c (tree_function_versioning): Update
3919         used_as_abstract_origin; be ready for DECL_RESULT and
3920         DECL_ARGUMENTS to be NULL.
3922         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3923         for abstract functions.
3924         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3925         real symbols.
3927 2013-08-01  Jan Hubicka  <jh@suse.cz>
3929         * profile.c (compute_value_histograms): Fix thinko.
3931 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3933         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3934         aarch-common-protos.h to extra_headers.
3935         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3936         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3937         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3939 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3941         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3942         Include "../arm/types.md".  Define "type" attribute for all patterns.
3943         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3944         attribute changes.
3946 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3948         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3949         to support power8 load fusion.
3950         (fusion_gpr_mem_load): Likewise.
3952         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3954         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3955         declarations for power8 load fusion.
3956         (emit_fusion_gpr_load): Likewise.
3958         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3959         tuning for power8, turn on fusion mode by default.  Turn on sign
3960         extending fusion mode if normal fusion mode is on, and we are at
3961         -O2 or -O3.
3962         (fusion_gpr_load_p): New function, return true if we can fuse an
3963         addis instruction with a dependent load to a GPR.
3964         (emit_fusion_gpr_load): Emit the instructions for power8 load
3965         fusion to GPRs.
3967         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3968         (VSX load fusion peepholes): New peepholes to fuse together an
3969         addi instruction with a VSX load instruction.
3971         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3972         peepholes to fuse an addis instruction with a load to a GPR base
3973         register.  If we are supporting sign extending fusions, convert
3974         sign extending loads to zero extending loads and add an explicit
3975         sign extension.
3977 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3979         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3980         aarch-common-protos.h to extra_headers.
3981         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3982         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3983         (arm_early_store_addr_dep): Likewise.
3984         (arm_no_early_alu_shift_dep): Likewise.
3985         (arm_no_early_alu_shift_value_dep): Likewise.
3986         (arm_no_early_mul_dep): Likewise.
3987         (arm_no_early_store_addr_dep): Likewise.
3988         (arm_mac_accumulator_is_mul_result): Likewise.
3989         (arm_mac_accumulator_is_result): Likewise.
3990         * config/arm/aarch-common.c: ... here.  New file.
3991         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3992         here to ...
3993         (arm_early_store_addr_dep): Likewise.
3994         (arm_no_early_alu_shift_dep): Likewise.
3995         (arm_no_early_alu_shift_value_dep): Likewise.
3996         (arm_no_early_mul_dep): Likewise.
3997         (arm_no_early_store_addr_dep): Likewise.
3998         (arm_mac_accumulator_is_mul_result): Likewise.
3999         (arm_mac_accumulator_is_result): Likewise.
4000         * config/arm/aarch-common-protos.h: ... here.  New file.
4001         * config/arm/t-arm (aarch-common.o): Define.
4003 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
4005         * config/arm/arm.md: Include new file "types.md".
4006         (define_attr "type"): Move from here to ...
4007         (define_attr "mul32"): Likewise.
4008         (define_attr "mul64"): Likewise.
4009         * config/arm/types.md: ... here.  New file.
4011 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4013         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4014         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4016 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4018         * gen-pass-instances.awk: Fix offset of substr().
4020 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
4022         * Makefile.in (pass-instances.def): New.
4023         (passes.o): Replace dependency on passes.def with one on
4024         pass-instances.def
4026         * gen-pass-instances.awk: New.
4028         * passes.c (pass_manager::pass_manager): Use pass-instances.def
4029         rather than passes.def, updating local definition of NEXT_PASS
4030         macro to add an extra NUM parameter (currently unused).
4032 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
4034         * Makefile.in (PASS_MANAGER_H): New.
4035         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4036         (passes.o): Likewise.
4037         (statistics.o): Likewise.
4038         (cgraphunit.o): Likewise.
4039         (context.o): Depend on PASS_MANAGER_H.
4041         * pass_manager.h: New.
4043         * cgraphunit.c (cgraph_add_new_function): Update for moves
4044         of globals to fields of pass_manager.
4045         (analyze_function): Likewise.
4046         (expand_function): Likewise.
4047         (ipa_passes): Likewise.
4048         (compile): Likewise.
4050         * context.c (context::context): New.
4051         * context.h  (context::context): New.
4052         (context::get_passes): New.
4053         (context::passes_): New.
4055         * lto-cgraph.c (input_node): Update for moves of globals to
4056         fields of pass_manager.
4058         * passes.c (all_passes): Remove, in favor of a field of the
4059         same name within the new class pass_manager.
4060         (all_small_ipa_passes): Likewise.
4061         (all_lowering_passes): Likewise.
4062         (all_regular_ipa_passes): Likewise.
4063         (all_late_ipa_passes): Likewise.
4064         (all_lto_gen_passes): Likewise.
4065         (passes_by_id): Likewise.
4066         (passes_by_id_size): Likewise.
4067         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4068         the new class pass_manager.
4069         (set_pass_for_id): Convert to...
4070         (pass_manager::set_pass_for_id): ...method.
4071         (get_pass_for_id): Convert to...
4072         (pass_manager::get_pass_for_id): ...method.
4073         (register_one_dump_file): Move body of implementation into...
4074         (pass_manager::register_one_dump_file): ...here.
4075         (register_dump_files_1): Convert to...
4076         (pass_manager::register_dump_files_1): ...method.
4077         (register_dump_files): Convert to...
4078         (pass_manager::register_dump_files): ...method.
4079         (create_pass_tab): Update for moves of globals to fields of
4080         pass_manager.
4081         (dump_passes): Move body of implementation into...
4082         (pass_manager::dump_passes): ...here.
4083         (register_pass): Move body of implementation into...
4084         (pass_manager::register_pass): ...here.
4085         (init_optimization_passes): Convert into...
4086         (pass_manager::pass_manager): ...constructor for new
4087         pass_manager class, and initialize the pass_lists array.
4088         (check_profile_consistency): Update for moves of globals to
4089         fields of pass_manager.
4090         (dump_profile_report): Move body of implementation into...
4091         (pass_manager::dump_profile_report): ...here.
4092         (ipa_write_summaries_1): Update for moves of pass lists from
4093         being globals to fields of pass_manager.
4094         (ipa_write_optimization_summaries): Likewise.
4095         (ipa_read_summaries):  Likewise.
4096         (ipa_read_optimization_summaries): Likewise.
4097         (execute_all_ipa_stmt_fixups): Likewise.
4099         * statistics.c (statistics_fini): Update for moves of globals to
4100         fields of pass_manager.
4102         * toplev.c (general_init): Replace call to
4103         init_optimization_passes with construction of the pass_manager
4104         instance.
4106         * tree-pass.h (all_passes): Remove, in favor of a field of the
4107         same name within the new class pass_manager.
4108         (all_small_ipa_passes): Likewise.
4109         (all_lowering_passes): Likewise.
4110         (all_regular_ipa_passes): Likewise.
4111         (all_lto_gen_passes): Likewise.
4112         (all_late_ipa_passes): Likewise.
4113         (passes_by_id): Likewise.
4114         (passes_by_id_size): Likewise.
4115         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4116         the new class pass_manager.
4117         (get_pass_for_id): Remove.
4119 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4121         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4122         configs.
4124 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4126         * arm.md (mulhi3): New expand pattern.
4128 2013-07-30  Jan Hubicka  <jh@suse.cz>
4129             Martin Liska  <marxin.liska@gmail.com>
4131         * profile.c (compute_value_histograms): Do not ICE when
4132         there is mismatch only on some counters.
4134 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4136         PR rtl-optimization/57637
4137         * function.c (move_insn_for_shrink_wrap): Also check the
4138         GEN set of the LIVE problem for the liveness analysis
4139         if it exists, otherwise give up.
4141 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4143         PR tree-optimization/57993
4144         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4145         replaced statement in the candidate table.
4146         (phi_add_costs): Return infinite cost when the hidden basis does
4147         not dominate all phis on which the candidate is dependent.
4148         (replace_one_candidate): Record replaced statement in the
4149         candidate table.
4151 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
4153         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4154         (ashlv2si3): New expander.
4155         (*ashlv2si3_i): New define_insn_and_split.
4156         * predicates.md (float_operation): Allow patterns with three
4157         basic sub-patterns.
4159         PR rtl-optimization/58021
4160         * mode-switching.c (create_pre_exit): Always split off preceding
4161         insns if we are not at the basic block head.
4163 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
4165         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4166         (UCLIBC_DYNAMIC_LINKER): New macro.
4167         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4168         `-mnan=2008'.
4169         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4170         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
4171         `-mnan=2008'.
4172         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4173         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4174         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4175         for SF and DF modes.  Use ieee_quad_format for TF mode.
4176         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4177         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4178         (mips_option_override): Handle `-mnan=legacy'.
4179         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4180         `-mabs=2008' and `-mnan=2008'.
4181         (OPTION_DEFAULT_SPECS): Add "nan" default.
4182         (ASM_SPEC): Handle `-mnan='.
4183         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4184         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4185         comment accordingly.
4186         (neg<mode>2): Likewise.
4187         * config/mips/mips.opt (mabs, mnan): New options.
4188         * doc/install.texi (Configuration): Document `--with-nan=' option.
4189         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4190         `-mnan=' options.
4191         (MIPS Options): Document them.
4192         * config.gcc <mips*-*-*>: Handle `--with-nan='.
4193         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4194         * configure: Regenerate.
4195         * config.in: Regenerate.
4197 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4199         * config/i386/i386.md (float post-reload splitters): Do not check
4200         for subregs of SSE registers.
4202 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4203             H.J. Lu  <hongjiu.lu@intel.com>
4205         PR target/57954
4206         PR target/57988
4207         * config/i386/i386.md (post-reload splitter
4208         to avoid partial SSE reg dependency stalls): New pattern.
4210 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4212         * config/s390/s390.md ("movcc"): Swap load and store instructions.
4214 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
4216         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4217         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4219 2013-07-26  Cary Coutant  <ccoutant@google.com>
4221         * dwarf2out.c (die_checksum_ordered): Don't include template
4222         instantiations in signature.
4223         (is_template_parameter): New function.
4224         (is_template_instantiation): New function.
4225         (generate_skeleton_bottom_up): Don't include template instantiations
4226         in type unit DIE.
4227         (generate_skeleton): Likewise.
4228         (break_out_comdat_types): Move recursive call to break out nested
4229         types earlier.
4230         (prune_unused_types_mark_generic_parms_dies): Call
4231         is_template_parameter.
4233 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
4235         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4236         uses vector registers.
4237         * config/aarch64/iterators.md: Add attributes rtn and vas.
4239 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4240             Richard Earnshaw  <richard.earnshaw@arm.com>
4242         * combine.c (simplify_comparison): Re-canonicalize operands
4243         where appropriate.
4244         * config/arm/arm.md (movcond_addsi): New splitter.
4246 2013-07-25  Sterling Augustine  <saugustine@google.com>
4248         * dwarf2out.c (size_of_pubnames): Move code to...
4249         (include_pubname_in_output): ...here.  New.
4250         (want_pubnames): Rearrange.
4251         (output_pubnames): Call include_pubname_in_output.  Move assertion.
4253 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4255         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4257 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4259         PR target/38836
4260         * doc/extend.texi: Remove obsolete builtins. Fix
4261         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4263 2013-07-25  Jan Hubicka  <jh@suse.cz>
4265         * cgraph.c (release_function_body): Break out from ...
4266         (cgraph_release_function_body): ... this one; also release DECL_RESULT
4267         and DECL_ARGUMENTS.
4268         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4269         old_tree in the map.
4270         (create_specialized_node): Update.
4271         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4272         into index.
4273         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4274         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4275         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4276         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4277         DECL_RESULT.
4279 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4281         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4282         addsi3_carryin_alt2_<optab>): Correct output template.
4284 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4286         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4287         Adjust for arm_restrict_it.
4288         Remove trailing whitespace.
4290 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
4292         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4293         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4295         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4297 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
4299         PR rtl-optimization/57960
4300         * lra-constraints.c (process_alt_operands): Use the right mode
4301         when checking strict_low.
4303 2013-07-25  Jan Hubicka  <jh@suse.cz>
4305         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4306         * cgraph.c (cgraph_remove_node): Do not release function body
4307         when in cgraph streaming.
4308         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4309         in other partitions are not considered reachable; fix handling of
4310         clones.
4312 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4314         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4316 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4318         PR target/19599
4319         PR target/57731
4320         PR target/57837
4321         * config/arm/arm.md ("*sibcall_insn): Replace use of
4322         Ss with US. Adjust output for v5 and v4t.
4323         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4325         * config/arm/constraints.md ("Ss"): Rename to US.
4327 2013-07-25  Terry Guo  <terry.guo@arm.com>
4329         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4330         shift_add/shift_sub0/shift_sub1 RTXs.
4332 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
4333             Anton Blanchard  <anton@au1.ibm.com>
4335         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4336         (altivec_vpks<VI_char>ss): Likewise.
4337         (altivec_vpks<VI_char>us): Likewise.
4338         (altivec_vpku<VI_char>us): Likewise.
4339         (altivec_vpku<VI_char>um): Likewise.
4341 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
4343         Introduce context class.
4345         * Makefile.in (CONTEXT_H): New.
4346         (OBJS): Add context.o.
4347         (toplev.o): Add CONTEXT_H to dependencies.
4348         (context.o): New.
4350         * toplev.c (general_init): Create the singleton gcc::context instance.
4352         * context.c: New.
4354         * context.h: New.
4356 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
4358         PR rtl-optimization/57968
4359         * mode-switching.c (create_pre_exit): Allow instructions that
4360         don't set a return register to need a non-exit mode.
4362 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4363             Anton Blanchard  <anton@au1.ibm.com>
4365         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4366         operands to vperm for little endian.
4367         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4368         of lvsl to create the control mask for a vperm for little endian.
4370 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4371             Anton Blanchard  <anton@au1.ibm.com>
4373         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4374         two operands for little-endian.
4376 2013-07-23  Steve Ellcey  <sellcey@mips.com>
4378         * config/mips/mips.c (mips_case_values_threshold): New.
4379         (TARGET_CASE_VALUES_THRESHOLD): Define.
4381 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4382             Anton Blanchard  <anton@au1.ibm.com>
4384         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4385         selection of field for vector splat in little endian mode.
4387 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4389         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4390         expanders to rs6000.md.
4391         (ior<mode>3): Likewise.
4392         (and<mode>3): Likewise.
4393         (one_cmpl<mode>2): Likewise.
4394         (nor<mode>3): Likewise.
4395         (andc<mode>3): Likewise.
4396         (eqv<mode>3): Likewise.
4397         (nand<mode>3): Likewise.
4398         (orc<mode>3): Likewise.
4400         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4401         declaration.
4403         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4404         to split multi-word logical operations.
4405         (rs6000_split_logical_di): Likewise.
4406         (rs6000_split_logical): Likewise.
4408         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4409         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4410         and allow TImode operations in 32-bit.
4411         (vsx_and<mode>3_64bit): Likewise.
4412         (vsx_ior<mode>3_32bit): Likewise.
4413         (vsx_ior<mode>3_64bit): Likewise.
4414         (vsx_xor<mode>3_32bit): Likewise.
4415         (vsx_xor<mode>3_64bit): Likewise.
4416         (vsx_one_cmpl<mode>2_32bit): Likewise.
4417         (vsx_one_cmpl<mode>2_64bit): Likewise.
4418         (vsx_nor<mode>3_32bit): Likewise.
4419         (vsx_nor<mode>3_64bit): Likewise.
4420         (vsx_andc<mode>3_32bit): Likewise.
4421         (vsx_andc<mode>3_64bit): Likewise.
4422         (vsx_eqv<mode>3_32bit): Likewise.
4423         (vsx_eqv<mode>3_64bit): Likewise.
4424         (vsx_nand<mode>3_32bit): Likewise.
4425         (vsx_nand<mode>3_64bit): Likewise.
4426         (vsx_orc<mode>3_32bit): Likewise.
4427         (vsx_orc<mode>3_64bit): Likewise.
4429         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4430         logical types in GPRs.
4432         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4433         logical insns to rs6000.md, and allow TImode operations in
4434         32-bit.
4435         (altivec_ior<mode>3): Likewise.
4436         (altivec_xor<mode>3): Likewise.
4437         (altivec_one_cmpl<mode>2): Likewise.
4438         (altivec_nor<mode>3): Likewise.
4439         (altivec_andc<mode>3): Likewise.
4441         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4442         attributes for moving the 128-bit logical operations into
4443         rs6000.md.
4444         (BOOL_REGS_OUTPUT): Likewise.
4445         (BOOL_REGS_OP1): Likewise.
4446         (BOOL_REGS_OP2): Likewise.
4447         (BOOL_REGS_UNARY): Likewise.
4448         (BOOL_REGS_AND_CR0): Likewise.
4449         (one_cmpl<mode>2): Add support for DI logical operations on
4450         32-bit, splitting the operations to 32-bit.
4451         (anddi3): Likewise.
4452         (iordi3): Likewise.
4453         (xordi3): Likewise.
4454         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4455         changes to combine the 32/64-bit code, allow logical operations on
4456         TI mode in 32-bit, and to use similar match_operator patterns like
4457         scalar mode uses.  Combine the Altivec and VSX code for logical
4458         operations, and move it here.
4459         (ior<mode>3, 128-bit types): Likewise.
4460         (xor<mode>3, 128-bit types): Likewise.
4461         (one_cmpl<mode>3, 128-bit types): Likewise.
4462         (nor<mode>3, 128-bit types): Likewise.
4463         (andc<mode>3, 128-bit types): Likewise.
4464         (eqv<mode>3, 128-bit types): Likewise.
4465         (nand<mode>3, 128-bit types): Likewise.
4466         (orc<mode>3, 128-bit types): Likewise.
4467         (and<mode>3_internal): Likewise.
4468         (bool<mode>3_internal): Likewise.
4469         (boolc<mode>3_internal1): Likewise.
4470         (boolc<mode>3_internal2): Likewise.
4471         (boolcc<mode>3_internal1): Likewise.
4472         (boolcc<mode>3_internal2): Likewise.
4473         (eqv<mode>3_internal1): Likewise.
4474         (eqv<mode>3_internal2): Likewise.
4475         (one_cmpl1<mode>3_internal): Likewise.
4477 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4479         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4480         Rename flag_stack_usage to flag_stack_usage_info.
4482 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4484         * config/microblaze/sync.md: New file.
4485         * config/microblaze/microblaze.md: Include sync.md
4486         * config/microblaze/microblaze.c: Add print_operand 'y'.
4487         * config/microblaze/constraints.md: Add memory_contraint
4488         'Q' which is a single register.
4490 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
4492         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4494 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4496         * reload.c (find_reloads): Exit loop once we find this operand
4497         cannot be reloaded somehow for this alternative.
4499         * reload.c (find_reloads): Exit loop once we find a hard register.
4501         * rtlanal.c (computed_jump_p): Exit loop once we find label
4502         reference is used.
4504         * i386.c (ix86_pad_returns): Exit loop after setting replace.
4506         * cfgloopmanip.c (remove_path): Exit loop after setting
4507         irred_invalidated.
4509         * gensupport.c (subst_dup): Avoid loop if code is not
4510         MATCH_DUP nor MATCH_OP_DUP.
4512 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
4514         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4516 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4518         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4519         true for SP_REGNUM if mode == ptr_mode.
4520         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4521         with value R0_REGNUM + 31.
4523 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4525         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4526         pad pointer-typed argument downward.
4528 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4530         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4531         and __ILP32__ when the ILP32 model is in use.
4533 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4535         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4536         (aarch64_load_symref_appropriately): In the case of
4537         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4538         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4539         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4540         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4541         if 'mode' doesn't equal to 'ptr_mode'.
4542         (aarch64_output_mi_thunk): Add an assertion on the alignment of
4543         'vcall_offset'; change to call aarch64_emit_move differently depending
4544         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4545         to calculate the upper bound of 'vcall_offset'.
4546         (aarch64_cannot_force_const_mem): Change to also return true if
4547         mode != ptr_mode.
4548         (aarch64_legitimize_reload_address): In the case of large
4549         displacements, add new local variable 'xmode' and an assertion
4550         based on it; change to use 'xmode' to generate the new rtx and
4551         reload.
4552         (aarch64_asm_trampoline_template): Change to generate the template
4553         differently depending on TARGET_ILP32 or not; change to use
4554         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4555         (aarch64_trampoline_size): Removed.
4556         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4557         and replace immediate literals with it.  Change to use 'ptr_mode'
4558         instead of 'DImode' and call convert_memory_address if the mode
4559         of 'fnaddr' doesn't equal to 'ptr_mode'.
4560         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4561         to output symbol.
4562         (aarch64_elf_asm_destructor): Likewise.
4563         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4564         on TARGET_ILP32 instead of aarch64_trampoline_size.
4565         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4566         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4567         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4568         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
4569         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4570         (storewb_pair<GPI:mode>_<P:mode>): ... this.
4571         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4572         depending on the value of 'mode'.
4573         (add_losym_<mode>): New.
4574         (ldr_got_small_<mode>): New, based on ldr_got_small.
4575         (ldr_got_small): Remove.
4576         (ldr_got_small_sidi): New.
4577         * config/aarch64/iterators.md (P): New.
4578         (PTR): Change to 'ptr_mode' in the condition.
4580 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4582         * config.gcc (aarch64*-*-*): Support --with-abi.
4583         (aarch64*-*-elf): Support --with-multilib-list.
4584         (aarch64*-*-linux*): Likewise.
4585         (supported_defaults): Add abi to aarch64*-*-*.
4586         * configure.ac: Mention AArch64 for --with-multilib-list.
4587         * configure: Re-generated.
4588         * config/aarch64/biarchilp32.h: New file.
4589         * config/aarch64/biarchlp64.h: New file.
4590         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4591         (ABI_SPEC): Ditto.
4592         (MULTILIB_DEFAULTS): Ditto.
4593         (DRIVER_SELF_SPECS): Ditto.
4594         (ASM_SPEC): Update to also substitute -mabi.
4595         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4596         file whose name depends on -mabi= and -mbig-endian.
4597         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4598         TARGET_ILP32.
4599         (POINTER_SIZE): New define.
4600         (POINTERS_EXTEND_UNSIGNED): Ditto.
4601         (enum aarch64_abi_type): New enumeration tag.
4602         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4603         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4604         (TARGET_ILP32): New define.
4605         * config/aarch64/aarch64.opt (mabi): New.
4606         (aarch64_abi): New.
4607         (ilp32, lp64): New values for -mabi.
4608         * config/aarch64/t-aarch64 (comma): New define.
4609         (MULTILIB_OPTIONS): Ditto.
4610         (MULTILIB_DIRNAMES): Ditto.
4611         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4612         * doc/invoke.texi: Document -mabi for AArch64.
4614 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
4616         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4618 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4619             Anton Blanchard  <anton@au1.ibm.com>
4621         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4622         endianness when selecting field to splat.
4624 2013-07-22  Eric Christopher  <echristo@gmail.com>
4626         * dwarf2out.c (die_odr_checksum): New function to use
4627         CHECKSUM_ macros and ULEB128 for DIE tag.
4628         (generate_type_signature): Use.
4630 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
4632         * config.gcc (sparc*-*-*): Accept leon3 processor.
4633         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4634         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4635         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4636         * config/sparc/sparc.opt (enum processor_type): Add leon3.
4637         (mfix-ut699): Adjust comment.
4638         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4639         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4640         (CPP_CPU_SPEC): Likewise.
4641         (ASM_CPU_SPEC): Likewise.
4642         * config/sparc/sparc.c (leon3_cost): New constant.
4643         (sparc_option_override): Add leon3 support.
4644         (mem_ref): New function.
4645         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4646         (sparc_do_work_around_errata): Look into the instruction in the delay
4647         slot and adjust accordingly.  Add fix for the data cache nullify issues
4648         of the UT699.  Change insertion position for the NOP.
4649         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4650         (leon3_load): New reservation.
4651         (leon_store): Bump latency to 2.
4652         (grfpu): New automaton.
4653         (grfpu_alu): New unit.
4654         (grfpu_ds): Likewise.
4655         (leon_fp_alu): Adjust.
4656         (leon_fp_mult): Delete.
4657         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4658         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4659         * config/sparc/sparc.md (cpu): Add leon3.
4660         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4661         (swapsi): Likewise.
4662         (atomic_test_and_set): Likewise.
4663         (ldstub): Likewise.
4665 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
4667         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
4668         default for R5900 targets.
4669         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4670         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4671         * config/mips/mips.c (mips_option_override): Report an error for
4672         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
4673         for -march=r5900 -mhard-float.
4675 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4677         * df-problems.c (can_move_insns_across): Exit loop once we
4678         find a non-fixed, non-global register.
4680         * ipa-pure-const.c (propagate_nothrow): Exit loop after
4681         setting can_throw.
4683         * omega.c (omega_eliminate_red): Break after setting red_found.
4684         (omega_problem_has_red_equations): Similarly after setting found.
4685         (omega_query_variable): Similarly after setting coupled.
4687 2013-07-22  Marek Polacek  <polacek@redhat.com>
4689         * gimplify.c: Don't include gimple.h twice.
4691 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4693         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4694         instead of TARGET_THUMB1.
4695         (Pz): New constraint.
4696         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4697         encodings.
4698         (compare_negsi_si): Likewise.
4699         (compare_addsi2_op0): Likewise.
4700         (compare_addsi2_op1): Likewise.
4701         (addsi3_carryin_<optab>): Likewise.
4702         (addsi3_carryin_alt2_<optab>): Likewise.
4703         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4704         for arm_restrict_it.
4705         (subsi3_carryin): Likewise.
4706         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4707         (minmax_arithsi): Disable for arm_restrict_it.
4708         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4709         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4710         (satsi_<SAT:code>_shift): Likewise.
4711         (arm_shiftsi3): Add alternative for 16-bit encoding.
4712         (arm32_movhf): Disable for arm_restrict_it.
4713         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4714         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4716 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
4718         * config/arm/arm.md (attribute "insn"): Delete.
4719         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4720         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4721         (not_shiftsi): Update for attribute change.
4722         (not_shiftsi_compare0): Likewise.
4723         (not_shiftsi_compare0_scratch): Likewise.
4724         (arm_one_cmplsi2): Likewise.
4725         (thumb1_one_cmplsi2): Likewise.
4726         (notsi_compare0): Likewise.
4727         (notsi_compare0_scratch): Likewise.
4728         (thumb1_movdi_insn): Likewise.
4729         (arm_movsi_insn): Likewise.
4730         (movhi_insn_arch4): Likewise.
4731         (movhi_bytes): Likewise.
4732         (arm_movqi_insn): Likewise.
4733         (thumb1_movqi_insn): Likewise.
4734         (arm32_movhf): Likewise.
4735         (thumb1_movhf): Likewise.
4736         (arm_movsf_soft_insn): Likewise.
4737         (thumb1_movsf_insn): Likewise.
4738         (thumb_movdf_insn): Likewise.
4739         (movsicc_insn): Likewise.
4740         (movsfcc_soft_insn): Likewise.
4741         (and_scc): Likewise.
4742         (cond_move): Likewise.
4743         (if_move_not): Likewise.
4744         (if_not_move): Likewise.
4745         (if_shift_move): Likewise.
4746         (if_move_shift): Likewise.
4747         (if_shift_shift): Likewise.
4748         (if_not_arith): Likewise.
4749         (if_arith_not): Likewise.
4750         (cond_move_not): Likewise.
4751         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4752         (neon_mov<mode>): Likewise.
4753         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4754         (thumb2_movsi_vfp): Likewise.
4755         (movsf_vfp): Likewise.
4756         (thumb2_movsf_vfp): Likewise.
4757         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4758         change.
4759         (cortexa7_older_only): Likewise.
4760         (cortexa7_younger): Likewise.
4761         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4762         (1020alu_shift_op): Likewise.
4763         (1020alu_shift_reg_op): Likewise.
4764         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4765         (alu_shift_op): Likewise.
4766         (alu_shift_reg_op): Likewise.
4767         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4768         (11_alu_shift_op): Likewise.
4769         (11_alu_shift_reg_op): Likewise.
4770         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4771         (9_alu_shift_reg_op): Likewise.
4772         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4773         change.
4774         (cortex_a15_alu_shift): Likewise.
4775         (cortex_a15_alu_shift_reg): Likewise.
4776         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4777         (cortex_a5_alu_shift): Likewise.
4778         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4779         change.
4780         (cortex_a53_alu_shift): Likewise.
4781         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4782         change.
4783         (cortex_a7_alu_reg): Likewise.
4784         (cortex_a7_alu_shift): Likewise.
4785         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4786         (cortex_a8_alu_shift): Likewise.
4787         (cortex_a8_alu_shift_reg): Likewise.
4788         (cortex_a8_mov): Likewise.
4789         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4790         (cortex_a9_dp_shift): Likewise.
4791         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4792         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4793         (cortex_r4_mov): Likewise.
4794         (cortex_r4_alu_shift): Likewise.
4795         (cortex_r4_alu_shift_reg): Likewise.
4796         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4797         (526_alu_shift_op): Likewise.
4798         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4799         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4800         (626te_alu_shift_op): Likewise.
4801         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4802         (726te_alu_op): Likewise.
4803         (726te_alu_shift_op): Likewise.
4804         (726te_alu_shift_reg_op): Likewise.
4805         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4806         (mp626_alu_shift_op): Likewise.
4807         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4808         (pj4_alu_e1_conds): Likewise.
4809         (pj4_alu): Likewise.
4810         (pj4_alu_conds): Likewise.
4811         (pj4_shift): Likewise.
4812         (pj4_shift_conds): Likewise.
4813         (pj4_alu_shift): Likewise.
4814         (pj4_alu_shift_conds): Likewise.
4816 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4818         * config/arm/predicates.md (shiftable_operator_strict_it):
4819         New predicate.
4820         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4821         Disable cond_exec version for arm_restrict_it.
4822         (thumb2_smaxsi3): Convert to generate cond_exec.
4823         (thumb2_sminsi3): Likewise.
4824         (thumb32_umaxsi3): Likewise.
4825         (thumb2_uminsi3): Likewise.
4826         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4827         (thumb2_neg_abssi2): Likewise.
4828         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4829         (thumb2_movsicc_insn): Adjust alternatives.
4830         (thumb2_mov_negscc): Disable for arm_restrict_it.
4831         (thumb2_mov_negscc_strict_it): New pattern.
4832         (thumb2_mov_notscc_strict_it): New pattern.
4833         (thumb2_mov_notscc): Disable for arm_restrict_it.
4834         (thumb2_ior_scc): Likewise.
4835         (thumb2_ior_scc_strict_it): New pattern.
4836         (thumb2_cond_move): Adjust for arm_restrict_it.
4837         (thumb2_cond_arith): Disable for arm_restrict_it.
4838         (thumb2_cond_arith_strict_it): New pattern.
4839         (thumb2_cond_sub): Adjust for arm_restrict_it.
4840         (thumb2_movcond): Likewise.
4841         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4842         (thumb2_zero_extendhisi2_v6): Likewise.
4843         (thumb2_zero_extendqisi2_v6): Likewise.
4844         (orsi_notsi_si): Likewise.
4845         (orsi_not_shiftsi_si): Likewise.
4847 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4849         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4850         instruction sequence is 1 byte shorter.
4852 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4854         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4855         it is not needed after split.
4857 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4859         PR target/51784
4860         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4861         second label for nonlocal goto receivers. Don't output pic base labels
4862         unless we're producing PIC; mark that action unreachable().
4863         (ix86_save_reg): If the function contains a nonlocal label, save the
4864         PIC base reg.
4865         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4866         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4867         (update_pic_label_number_if_needed): New.
4868         (machopic_output_function_base_name): Adjust for nonlocal receiver
4869         case.
4870         (machopic_should_output_picbase_label): New.
4871         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4872         (nonlocal_goto_receiver): New insn and split.
4874 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4876         * config/aarch64/aarch64-builtins.c
4877         (aarch64_fold_builtin): Fold abs in all modes.
4878         * config/aarch64/aarch64-simd-builtins.def
4879         (abs): Enable for all modes.
4880         * config/aarch64/arm_neon.h
4881         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4882         (vabs_f64): Add missing intrinsic.
4884 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4886         * config/aarch64/arm_neon.h (vabs_s64): New function
4888 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4890         PR target/57516
4891         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4892         * config/avr/avr.md (adjust_len): Add `round'.
4893         * config/avr/avr-protos.h (avr_out_round): New prototype.
4894         (avr_out_plus): Add `out_label' argument.
4895         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4896         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4897         Handle the case where `insn' is just a pattern.
4898         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4899         (avr_out_round): New function.
4900         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4902 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
4904         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4905         Add check for flag_stack_usage to handle -fstack-usage support
4907 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
4909         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4910         interaction for new Power8 flags and VSX.
4912 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4914         PR middle-end/57698
4915         * tree-inline.c (expand_call_inline): Emit errors during
4916         early_inlining only if optimization is not turned on.
4918 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4920         * passes.def: New.
4922         * passes.c (init_optimization_passes): Move the construction of
4923         the pass hierarchy into a new passes.def file.
4925         * Makefile.in (passes.o): Add dependency on passes.def.
4927 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4929         * passes.c (init_optimization_passes): Introduce macros for
4930         constructing the tree of passes (INSERT_PASSES_AFTER,
4931         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4932         TERMINATE_PASS_LIST).
4934 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4935             Wei Mi  <wmi@google.com>
4937         PR rtl-optimization/57878
4938         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4939         top.
4940         (reload_pseudo_compare_func): Check nregs first for reload
4941         pseudos.
4943 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4945         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4947 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4949         * read-rtl.c (validate_const_int): Once an invalid character is
4950         seen, quit the loop.
4952         * gengtype.c (write_roots): Similarly once we find the "deletable"
4953         or "if_marked" option.
4955 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4957         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4958         "xtab" and "sat".  Move value "clz" from here to ...
4959         (attriubte "type"): ... here.
4960         (satsi_<SAT:code>): Delete "insn" attribute.
4961         (satsi_<SAT:code>_shift): Likewise.
4962         (arm_zero_extendqisi2addsi): Likewise.
4963         (arm_extendqisi2addsi): Likewise.
4964         (clzsi2): Update for attribute changes.
4965         (rbitsi2): Likewise.
4966         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4967         attribute.
4968         (arm_usatsihi): Likewise.
4969         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4971 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4973         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4974         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4975         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4976         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4977         in alphabetical order.
4978         (attribute "core_cycles"): Update for attribute changes.
4979         (arm_addsi3): Likewise.
4980         (addsi3_compare0): Likewise.
4981         (addsi3_compare0_scratch): Likewise.
4982         (addsi3_compare_op1): Likewise.
4983         (addsi3_compare_op2): Likewise.
4984         (compare_addsi2_op0): Likewise.
4985         (compare_addsi2_op1): Likewise.
4986         (addsi3_carryin_shift_<optab>): Likewise.
4987         (subsi3_carryin_shift): Likewise.
4988         (rsbsi3_carryin_shift): Likewise.
4989         (arm_subsi3_insn): Likewise.
4990         (subsi3_compare0): Likewise.
4991         (subsi3_compare): Likewise.
4992         (arm_andsi3_insn): Likewise.
4993         (thumb1_andsi3_insn): Likewise.
4994         (andsi3_compare0): Likewise.
4995         (andsi3_compare0_scratch): Likewise.
4996         (zeroextractsi_compare0_scratch
4997         (andsi_not_shiftsi_si): Likewise.
4998         (iorsi3_insn): Likewise.
4999         (iorsi3_compare0): Likewise.
5000         (iorsi3_compare0_scratch): Likewise.
5001         (arm_xorsi3): Likewise.
5002         (thumb1_xorsi3_insn): Likewise.
5003         (xorsi3_compare0): Likewise.
5004         (xorsi3_compare0_scratch): Likewise.
5005         (satsi_<SAT:code>_shift): Likewise.
5006         (rrx): Likewise.
5007         (arm_shiftsi3): Likewise.
5008         (shiftsi3_compare0): Likewise.
5009         (not_shiftsi): Likewise.
5010         (not_shiftsi_compare0): Likewise.
5011         (not_shiftsi_compare0_scratch): Likewise.
5012         (arm_one_cmplsi2): Likewise.
5013         (thumb_one_complsi2): Likewise.
5014         (notsi_compare0): Likewise.
5015         (notsi_compare0_scratch): Likewise.
5016         (thumb1_zero_extendhisi2): Likewise.
5017         (arm_zero_extendhisi2): Likewise.
5018         (arm_zero_extendhisi2_v6): Likewise.
5019         (arm_zero_extendhisi2addsi): Likewise.
5020         (thumb1_zero_extendqisi2): Likewise.
5021         (thumb1_zero_extendqisi2_v6): Likewise.
5022         (arm_zero_extendqisi2): Likewise.
5023         (arm_zero_extendqisi2_v6): Likewise.
5024         (arm_zero_extendqisi2addsi): Likewise.
5025         (thumb1_extendhisi2): Likewise.
5026         (arm_extendhisi2): Likewise.
5027         (arm_extendhisi2_v6): Likewise.
5028         (arm_extendqisi): Likewise.
5029         (arm_extendqisi_v6): Likewise.
5030         (arm_extendqisi2addsi): Likewise.
5031         (thumb1_extendqisi2): Likewise.
5032         (thumb1_movdi_insn): Likewise.
5033         (arm_movsi_insn): Likewise.
5034         (movsi_compare0): Likewise.
5035         (movhi_insn_arch4): Likewise.
5036         (movhi_bytes): Likewise.
5037         (arm_movqi_insn): Likewise.
5038         (thumb1_movqi_insn): Likewise.
5039         (arm32_movhf): Likewise.
5040         (thumb1_movhf): Likewise.
5041         (arm_movsf_soft_insn): Likewise.
5042         (thumb1_movsf_insn): Likewise.
5043         (movdf_soft_insn): Likewise.
5044         (thumb_movdf_insn): Likewise.
5045         (arm_cmpsi_insn): Likewise.
5046         (cmpsi_shiftsi): Likewise.
5047         (cmpsi_shiftsi_swp): Likewise.
5048         (arm_cmpsi_negshiftsi_si): Likewise.
5049         (movsicc_insn): Likewise.
5050         (movsfcc_soft_insn): Likewise.
5051         (arith_shiftsi): Likewise.
5052         (arith_shiftsi_compare0
5053         (arith_shiftsi_compare0_scratch
5054         (sub_shiftsi): Likewise.
5055         (sub_shiftsi_compare0
5056         (sub_shiftsi_compare0_scratch
5057         (and_scc): Likewise.
5058         (cond_move): Likewise.
5059         (if_plus_move): Likewise.
5060         (if_move_plus): Likewise.
5061         (if_move_not): Likewise.
5062         (if_not_move): Likewise.
5063         (if_shift_move): Likewise.
5064         (if_move_shift): Likewise.
5065         (if_shift_shift): Likewise.
5066         (if_not_arith): Likewise.
5067         (if_arith_not): Likewise.
5068         (cond_move_not): Likewise.
5069         (thumb1_ashlsi3): Set type attribute.
5070         (thumb1_ashrsi3): Likewise.
5071         (thumb1_lshrsi3): Likewise.
5072         (thumb1_rotrsi3): Likewise.
5073         (shiftsi3_compare0_scratch): Likewise.
5074         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5075         (neon_mov<mode>): Likewise.
5076         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5077         attribute changes.
5078         (thumb2_movsi_insn): Likewise.
5079         (thumb2_cmpsi_neg_shiftsi): Likewise.
5080         (thumb2_extendqisi_v6): Likewise.
5081         (thumb2_zero_extendhisi2_v6): Likewise.
5082         (thumb2_zero_extendqisi2_v6): Likewise.
5083         (thumb2_shiftsi3_short): Likewise.
5084         (thumb2_addsi3_compare0_scratch): Likewise.
5085         (orsi_not_shiftsi_si): Likewise.
5086         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5087         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5088         changes.
5089         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5090         (1020alu_shift_op): Likewise.
5091         (1020alu_shift_reg_op): Likewise.
5092         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5093         (alu_shift_op): Likewise.
5094         (alu_shift_reg_op): Likewise.
5095         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5096         (11_alu_shift_op): Likewise.
5097         (11_alu_shift_reg_op): Likewise.
5098         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5099         (9_alu_shift_reg_op): Likewise.
5100         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5101         changes.
5102         (cortex_a15_alu_shift): Likewise.
5103         (cortex_a15_alu_shift_reg): Likewise.
5104         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5105         changes.
5106         (cortex_a5_alu_shift): Likewise.
5107         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5108         changes.
5109         (cortex_a53_alu_shift): Likewise.
5110         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5111         changes.
5112         (cortex_a7_alu_reg): Likewise.
5113         (cortex_a7_alu_shift): Likewise.
5114         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5115         changes.
5116         (cortex_a8_alu_shift): Likewise.
5117         (cortex_a8_alu_shift_reg): Likewise.
5118         (cortex_a8_mov): Likewise.
5119         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5120         (cortex_a9_dp_shift): Likewise.
5121         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5122         changes.
5123         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5124         changes.
5125         (cortex_r4_mov): Likewise.
5126         (cortex_r4_alu_shift): Likewise.
5127         (cortex_r4_alu_shift_reg): Likewise.
5128         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5129         (526_alu_shift_op): Likewise.
5130         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5131         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5132         (626te_alu_shift_op): Likewise.
5133         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5134         (726te_alu_op): Likewise.
5135         (726te_alu_shift_op): Likewise.
5136         (726te_alu_shift_reg_op): Likewise.
5137         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5138         (mp626_alu_shift_op): Likewise.
5139         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5140         (pj4_alu_e1_conds): Likewise.
5141         (pj4_alu): Likewise.
5142         (pj4_alu_conds): Likewise.
5143         (pj4_shift): Likewise.
5144         (pj4_shift_conds): Likewise.
5145         (pj4_alu_shift): Likewise.
5146         (pj4_alu_shift_conds): Likewise.
5147         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5148         changes.
5149         (cortexa7_older_only): Likewise.
5150         (cortexa7_younger): Likewise.
5152 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5154         * ipa-pure-const.c (generate_summary): Rename to...
5155         (pure_const_generate_summary): ... this.
5157 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
5159         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5161 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
5163         PR target/57909
5164         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5165         usage in HI mode.
5167 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5169         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5170         enabled without -march=zEC12.
5171         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5172         flags to be set.
5174 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5176         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5177         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5178         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5179         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5180         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5181         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5182         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5183         ISA_HAS_FP4.
5184         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5185         and ISA_HAS_NMADD3_NMSUB3.
5186         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5187         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5188         (nmsub4<mode>, nmsub3<mode>): Likewise.
5189         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5191 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5193         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5194         TARGET_MIPS5400 checking.
5196 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
5197             Peter Bergner  <bergner@vnet.ibm.com>
5199         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5200         registers in the comment.
5201         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5202         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5203         rather than FIRST_PSEUDO_REGISTERS.
5205 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
5207         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5208         enable extra ISA flags with TARGET_HTM.
5210 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5212         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5213         Fix comment typos.
5215 2013-07-15  Cong Hou  <congh@google.com>
5217         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5218         in compare function for sorting.
5220 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
5222         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5223         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5224         * config/rs6000/rs6000.opt: Add -mhtm option.
5225         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5226         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5227         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5228         __HTM__ if the HTM instructions are available.
5229         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5230         htm_spr_reg_operand): New define_predicates.
5231         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5232         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5233         Include htm.md.
5234         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5235         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5236         HTM builtin functions.
5237         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5238         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5239         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5240         (rs6000_builtin_mask_calculate): Likewise.
5241         (rs6000_option_override_internal): Likewise.
5242         (bdesc_htm): Add new HTM builtin support.
5243         (htm_spr_num): New function.
5244         (htm_spr_regno): Likewise.
5245         (rs6000_htm_spr_icode): Likewise.
5246         (htm_expand_builtin): Likewise.
5247         (htm_init_builtins): Likewise.
5248         (rs6000_expand_builtin): Add support for HTM builtin functions.
5249         (rs6000_init_builtins): Likewise.
5250         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5251         option.
5252         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5253         (TARGET_HTM, MASK_HTM): Define macros.
5254         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5255         (FIXED_REGISTERS): Likewise.
5256         (CALL_USED_REGISTERS): Likewise.
5257         (CALL_REALLY_USED_REGISTERS): Likewise.
5258         (REG_ALLOC_ORDER): Likewise.
5259         (enum reg_class): Likewise.
5260         (REG_CLASS_NAMES): Likewise.
5261         (REG_CLASS_CONTENTS): Likewise.
5262         (REGISTER_NAMES): Likewise.
5263         (ADDITIONAL_REGISTER_NAMES): Likewise.
5264         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5265         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5266         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5267         * config/rs6000/htm.md: New file.
5268         * config/rs6000/htmintrin.h: New file.
5269         * config/rs6000/htmxlintrin.h: New file.
5271 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5273         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5274         Define SYMBOL_TINY_GOT, update comment.
5275         * config/aarch64/aarch64.c
5276         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5277         (aarch64_expand_mov_immediate): Likewise.
5278         (aarch64_print_operand): Likewise.
5279         (aarch64_classify_symbol): Likewise.
5280         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5281         (ldr_got_tiny): Define.
5283 2013-07-13  Tobias Grosser  <tobias@grosser.es>
5285         PR tree-optimization/54094
5286         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5287           scheduling dimension for the parallelism check from the polyhedral
5288           information in the AST.
5289         * graphite-dependences.c (carries_deps): Do not assume the schedule is
5290           in 2D + 1 form.
5292 2013-07-13  Jason Merrill  <jason@redhat.com>
5294         * print-tree.c (debug_vec_tree): Use debug_raw.
5295         (debug_raw (vec<tree, va_gc> &)): New.
5296         (debug_raw (vec<tree, va_gc> *)): New.
5297         * tree.h: Declare them.
5299 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
5301         * ifcvt.c (ifcvt_after_combine): New static variable.
5302         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5303         for size.
5304         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
5305         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5306         rest_of_handle_if_after_reload): Pass new argument for if_convert.
5308 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
5310         * config/mips/mips.c (mips_expand_call): Remove empty statement.
5312 2013-07-12  Michael Matz  <matz@suse.de>
5314         PR middle-end/55771
5315         * convert.c (convert_to_real): Reject non-float inner types.
5317 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
5319         * config/aarch64/aarch64-protos.h
5320         (aarch64_simd_immediate_valid_for_move): Remove.
5321         * config/aarch64/aarch64.c (simd_immediate_info): New member.
5322         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5323         cases.
5324         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5326 2013-07-11  Steve Ellcey  <sellcey@mips.com>
5328         * config/mips/mips.c (mips_conditional_register_usage): Do not
5329         use t[0-7] registers in MIPS16 mode when optimizing for size.
5331 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
5333         * config/i386/i386.c (dispatch_function_versions): Fix array
5334         indexing of function_version_info to match actual_versions.
5336 2013-07-11  Teresa Johnson  <tejohnson@google.com>
5338         * vec.h (struct va_gc): Move release out-of-line.
5339         (va_gc::release): Call ggc_free on released vec.
5341 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5343         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5344         Require GOT register as additional operand in UNSPEC.
5345         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5346         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5347         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5348         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5349         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5350         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5351         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5353 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
5355         PR target/57631
5356         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5357         name seen by assembler/linker rather if available.
5359 2013-07-11  Andreas Schwab  <schwab@suse.de>
5361         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5363 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
5365         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5367 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
5369         * doc/tm.texi.in: Move hook documentation to ....
5370         * target.def: ... here.
5372         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5373         text on @hook line.
5374         * doc/tm.texi: Regenerate.
5376 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
5378         PR c++/57869
5379         * doc/invoke.texi: Document -Wconditionally-supported.
5381 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5383         PR target/57844
5384         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5385         of my_fp.
5387 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5389         PR target/57506
5390         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5391         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5392         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5393         Remove duplicate devices.
5394         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5395         * config/avr/t-multilib: Regenerate.
5396         * config/avr/avr-tables.opt: Regenerate.
5397         * doc/avr-mmcu.texi: Regenerate.
5399 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5401         PR target/56987
5402         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5404 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
5406         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5407         the cost of MULT when optimizing for size.
5409 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5411         * config/cr16/cr16-protos.h: Don't include target.h.
5413 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
5415         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5416         adjust register size for TDmode and TFmode for VSX registers.
5418 2013-07-08  Kai Tietz  <ktietz@redhat.com>
5420         PR target/56892
5421         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5422         hook_bool_const_tree_true.
5424 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5426         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5427         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5428         * config/s390/s390.md: Define FPR*_REGNUM constants.
5429         Fix FPR2_REGNUM constant (18 -> 17).
5430         ("*trunc<BFP:mode><DFP_ALL:mode>2")
5431         ("*trunc<DFP_ALL:mode><BFP:mode>2")
5432         ("trunc<BFP:mode><DFP_ALL:mode>2")
5433         ("trunc<DFP_ALL:mode><BFP:mode>2")
5434         ("*extend<BFP:mode><DFP_ALL:mode>2")
5435         ("*extend<DFP_ALL:mode><BFP:mode>2")
5436         ("extend<BFP:mode><DFP_ALL:mode>2")
5437         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5438         FPR4_REGNUM.
5440 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
5442         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5444 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5446         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5447         and cfun_fpr_bit_p to cfun_fpr_save_p.
5448         (s390_frame_area, s390_register_info, s390_frame_info)
5449         (s390_emit_prologue, s390_emit_epilogue)
5450         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5451         register numbers.
5452         * config/s390/s390.h: Define *_REGNUM macros for floating point
5453         register numbers.
5455 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
5457         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5459 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
5461         PR rtl-optimization/57786
5462         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5463         and break out of the loop when it is set to false.
5465 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
5467         PR target/57819
5468         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5469         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5470         (const_int 63)) 0)).
5471         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5472         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5473         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5475         PR rtl-optimization/57829
5476         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5477         mask bits outside of mode are just sign-extension from mode to HWI.
5479 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5481         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5482         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5483         adjust_address instead of change_address to keep info about alignment.
5484         (emit_strmov): Remove.
5485         (emit_memmov): New function.
5486         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5487         (expand_movmem_epilogue): Likewise and return updated rtx for
5488         destination.
5489         (expand_constant_movmem_prologue): Likewise and return updated rtx for
5490         destination and source.
5491         (decide_alignment): Refactor, handle vector_loop.
5492         (ix86_expand_movmem): Likewise.
5493         (ix86_expand_setmem): Likewise.
5494         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5496 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
5498         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5499         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5501 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5503         * config/i386/sse.md (sse_movlhps): Change alternative 3
5504         of operand 2 to "m".
5506 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5508         PR target/57807
5509         * config/i386/sse.md (iptr): New mode attribute.
5510         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5511         (<sse>_vm<plusminus_insn><mode>3): Ditto.
5512         (<sse>_vmmul<mode>3): Ditto.
5513         (<sse>_vmdiv<mode>3): Ditto.
5514         (sse_vmrcpv4sf2): Ditto.
5515         (<sse>_vmsqrt<mode>2): Ditto.
5516         (sse_vmrsqrtv4sf2): Ditto.
5517         (<sse>_vm<code><mode>3): Ditto.
5518         (avx_vmcmp<mode>3): Ditto.
5519         (<sse>_vmmaskcmp<mode>3): Ditto.
5520         (<sse>_comi): Ditto.
5521         (<sse>_ucomi): Ditto.
5522         (*xop_vmfrcz_<mode>): Ditto.
5523         (*fmai_fmadd_<mode>): Ditto.
5524         (*fmai_fmsub_<mode>): Ditto.
5525         (*fmai_fnmadd_<mode>): Ditto.
5526         (*fmai_fnmsub_<mode>): Ditto.
5527         (*fma4i_vmfmadd_<mode>): Ditto.
5528         (*fma4i_vmfmsub_<mode>): Ditto.
5529         (*fma4i_vmfnmadd_<mode>): Ditto.
5530         (*fma4i_vmfnmsub_<mode>): Ditto.
5531         (*xop_vmfrcz_<mode>): Ditto.
5532         (sse_cvtps2pi): Ditto.
5533         (sse_cvttps2pi): Ditto.
5534         (sse_cvtss2si): Ditto.
5535         (sse_cvtss2si_2): Ditto.
5536         (sse_cvtss2siq_2): Ditto.
5537         (sse_cvttss2si): Ditto.
5538         (sse_cvttss2siq): Ditto.
5539         (sse_cvtsd2si): Ditto.
5540         (sse_cvtsd2si_2): Ditto.
5541         (sse_cvtsd2siq_2): Ditto.
5542         (sse_cvttsd2si): Ditto.
5543         (sse_cvttsd2siq): Ditto.
5544         (sse_cvtsd2ss): Ditto.
5545         (sse_cvtss2sd): Ditto.
5546         (avx2_pbroadcast<mode>): Ditto.
5547         (avx2_pbroadcast<mode>_1): Ditto.
5548         (*avx_vperm_broadcast_v4sf): Ditto.
5550         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5551         (sse_movlhps): Ditto.
5552         (sse_storehps): Ditto.
5553         (sse_loadhps): Ditto.
5554         (sse_storelps): Ditto.
5555         (sse_loadlps): Ditto.
5556         (*vec_concatv4sf): Ditto.
5557         (*vec_interleave_highv2df): Ditto.
5558         (*vec_interleave_lowv2df): Ditto.
5559         (*vec_extractv2df_1_sse): Ditto.
5560         (*vec_extractv2df_0_sse): Ditto.
5561         (sse2_storelpd): Ditto.
5562         (sse2_loadlpd): Ditto.
5563         (sse2_movsd): Ditto.
5564         (*vec_concatv4si): Ditto.
5565         (vec_concatv2di): Ditto.
5567         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5568         for Intel asm dialect.
5569         (mmx_punpcklwd): Ditto.
5570         (mmx_punpckldq): Ditto.
5572         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5573         for intel assembler dialect.
5575 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
5577         PR target/29776
5578         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5579         for BUILT_IN_C{LZ,LRSB}*.
5580         * tree.h (CASE_INT_FN): Add FN##IMAX case.
5581         * tree-vrp.c (extract_range_basic): Handle
5582         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
5583         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5584         fall thru to code calling set_value*.
5585         * builtins.c (expand_builtin): Remove *IMAX cases.
5586         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5587         if width is bigger than 2*HWI.
5589 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
5591         PR rtl-optimization/55342
5592         * lra-int.h (lra_subreg_reload_pseudos): New.
5593         * lra.c: Add undoing optional reloads to the block diagram.
5594         (lra_subreg_reload_pseudos): New.
5595         (lra_optional_reload_pseudos): Change comments.
5596         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
5597         lra_optional_reload_pseudos after undo transformations.
5598         * lra-assigns.c (pseudo_prefix_title): New.
5599         (lra_setup_reg_renumber): Use it.
5600         (spill_for): Ditto.  Check subreg reload pseudos too.
5601         (assign_by_spills): Consider subreg reload pseudos too.
5602         * lra-constraints.c (simplify_operand_subreg): Use
5603         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5604         (curr_insn_transform): Recognize and do optional reloads.
5605         (undo_optional_reloads): New.
5606         (lra_undo_inheritance): Call undo_optional_reloads.
5608 2013-07-05  Thomas Quinot  <quinot@adacore.com>
5610         * tree-complex.c (expand_complex_operations_1): Fix typo.
5612 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
5614         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5615         (tune_params): New member 'const vec_costs'.
5616         * config/aarch64/aarch64.c (generic_vector_cost): New.
5617         (generic_tunings): New member 'generic_vector_cost'.
5618         (aarch64_builtin_vectorization_cost): New.
5619         (aarch64_add_stmt_cost): New.
5620         (TARGET_VECTORIZE_ADD_STMT_COST): New.
5621         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5623 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
5625         PR target/57777
5626         * config/i386/predicates.md (vsib_address_operand): Disallow
5627         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5629 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
5631         PR middle-end/55030
5632         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5633         expand_builtin_setjmp_receiver.
5634         (expand_label): Adjust, call expand_builtin_setjmp_receiver
5635         with NULL for the label parameter.
5636         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5637         the frame-pointer.  Adjust comments.
5638         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5639         only if LABEL is non-NULL.
5641 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
5643         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5644         (ARM_ABI_AAPCS64): Ditto.
5645         (arm_abi): Ditto.
5646         (ARM_DEFAULT_ABI): Ditto.
5648 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
5650         * config/aarch64/aarch64-builtins.c
5651         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5652         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5653         (st1): Likewise.
5654         * config/aarch64/aarch64-simd.md
5655         (aarch64_ld1<VALL:mode>): New.
5656         (aarch64_st1<VALL:mode>): Likewise.
5657         * config/aarch64/arm_neon.h
5658         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5660 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
5662         * config/i386/i386.c (gate_insert_vzeroupper): Check if
5663         target ISA is AVX.
5664         (ix86_option_override_internal):Turn on all -mavx target flags by
5665         default as they are dependent on AVX anyway.
5667 2013-07-02  Cary Coutant  <ccoutant@google.com>
5669         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5670         deterministic hash.
5671         (loc_checksum_ordered): Likewise.
5672         (hash_loc_operands): Remove inline keyword.
5674 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
5676         PR tree-optimization/57741
5677         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5678         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5679         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5680         Allow REAL_CST step_exprs if flag_associative_math.
5681         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5683 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5685         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5687 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5689         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5691 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5693         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5694         encoding.
5695         (iorsi3_insn): Likewise.
5696         (arm_xorsi3): Likewise.
5698 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
5700         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
5701         to ...
5702         (attribute "type"): ... here, and prefix with "wmmx_".
5703         (attribute "core_cycles"): Update for attribute changes.
5704         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5705         (tbcstv4hi): Likewise.
5706         (tbcstv2si): Likewise.
5707         (iwmmxt_iordi3): Likewise.
5708         (iwmmxt_xordi3): Likewise.
5709         (iwmmxt_anddi3): Likewise.
5710         (iwmmxt_nanddi3): Likewise.
5711         (iwmmxt_arm_movdi): Likewise.
5712         (iwmmxt_movsi_insn): Likewise.
5713         (mov<mode>_internal): Likewise.
5714         (and<mode>3_iwmmxt): Likewise.
5715         (ior<mode>3_iwmmxt): Likewise.
5716         (xor<mode>3_iwmmxt): Likewise.
5717         (add<mode>3_iwmmxt): Likewise.
5718         (ssaddv8qi3): Likewise.
5719         (ssaddv4hi3): Likewise.
5720         (ssaddv2si3): Likewise.
5721         (usaddv8qi3): Likewise.
5722         (usaddv4hi3): Likewise.
5723         (usaddv2si3): Likewise.
5724         (sub<mode>3_iwmmxt): Likewise.
5725         (sssubv8qi3): Likewise.
5726         (sssubv4hi3): Likewise.
5727         (sssubv2si3): Likewise.
5728         (ussubv8qi3): Likewise.
5729         (ussubv4hi3): Likewise.
5730         (ussubv2si3): Likewise.
5731         (mulv4hi3_iwmmxt): Likewise.
5732         (smulv4hi3_highpart): Likewise.
5733         (umulv4hi3_highpart): Likewise.
5734         (iwmmxt_wmacs): Likewise.
5735         (iwmmxt_wmacsz): Likewise.
5736         (iwmmxt_wmacu): Likewise.
5737         (iwmmxt_wmacuz): Likewise.
5738         (iwmmxt_clrdi): Likewise.
5739         (iwmmxt_clrv8qi): Likewise.
5740         (iwmmxt_clr4hi): Likewise.
5741         (iwmmxt_clr2si): Likewise.
5742         (iwmmxt_uavgrndv8qi3): Likewise.
5743         (iwmmxt_uavgrndv4hi3): Likewise.
5744         (iwmmxt_uavgv8qi3): Likewise.
5745         (iwmmxt_uavgv4hi3): Likewise.
5746         (iwmmxt_tinsrb): Likewise.
5747         (iwmmxt_tinsrh): Likewise.
5748         (iwmmxt_tinsrw): Likewise.
5749         (iwmmxt_textrmub): Likewise.
5750         (iwmmxt_textrmsb): Likewise.
5751         (iwmmxt_textrmuh): Likewise.
5752         (iwmmxt_textrmsh): Likewise.
5753         (iwmmxt_textrmw): Likewise.
5754         (iwmxxt_wshufh): Likewise.
5755         (eqv8qi3): Likewise.
5756         (eqv4hi3): Likewise.
5757         (eqv2si3): Likewise.
5758         (gtuv8qi3): Likewise.
5759         (gtuv4hi3): Likewise.
5760         (gtuv2si3): Likewise.
5761         (gtv8qi3): Likewise.
5762         (gtv4hi3): Likewise.
5763         (gtv2si3): Likewise.
5764         (smax<mode>3_iwmmxt): Likewise.
5765         (umax<mode>3_iwmmxt): Likewise.
5766         (smin<mode>3_iwmmxt): Likewise.
5767         (umin<mode>3_iwmmxt): Likewise.
5768         (iwmmxt_wpackhss): Likewise.
5769         (iwmmxt_wpackwss): Likewise.
5770         (iwmmxt_wpackdss): Likewise.
5771         (iwmmxt_wpackhus): Likewise.
5772         (iwmmxt_wpackwus): Likewise.
5773         (iwmmxt_wpackdus): Likewise.
5774         (iwmmxt_wunpckihb): Likewise.
5775         (iwmmxt_wunpckihh): Likewise.
5776         (iwmmxt_wunpckihw): Likewise.
5777         (iwmmxt_wunpckilb): Likewise.
5778         (iwmmxt_wunpckilh): Likewise.
5779         (iwmmxt_wunpckilw): Likewise.
5780         (iwmmxt_wunpckehub): Likewise.
5781         (iwmmxt_wunpckehuh): Likewise.
5782         (iwmmxt_wunpckehuw): Likewise.
5783         (iwmmxt_wunpckehsb): Likewise.
5784         (iwmmxt_wunpckehsh): Likewise.
5785         (iwmmxt_wunpckehsw): Likewise.
5786         (iwmmxt_wunpckelub): Likewise.
5787         (iwmmxt_wunpckeluh): Likewise.
5788         (iwmmxt_wunpckeluw): Likewise.
5789         (iwmmxt_wunpckelsb): Likewise.
5790         (iwmmxt_wunpckelsh): Likewise.
5791         (iwmmxt_wunpckelsw): Likewise.
5792         (ror<mode>3): Likewise.
5793         (ashr<mode>3_iwmmxt): Likewise.
5794         (lshr<mode>3_iwmmxt): Likewise.
5795         (ashl<mode>3_iwmmxt): Likewise.
5796         (ror<mode>3_di): Likewise.
5797         (ashr<mode>3_di): Likewise.
5798         (lshr<mode>3_di): Likewise.
5799         (ashl<mode>3_di): Likewise.
5800         (iwmmxt_wmadds): Likewise.
5801         (iwmmxt_wmaddu): Likewise.
5802         (iwmmxt_tmia): Likewise.
5803         (iwmmxt_tmiaph): Likewise.
5804         (iwmmxt_tmiabb): Likewise.
5805         (iwmmxt_tmiatb): Likewise.
5806         (iwmmxt_tmiabt): Likewise.
5807         (iwmmxt_tmiatt): Likewise.
5808         (iwmmxt_tmovmskb): Likewise.
5809         (iwmmxt_tmovmskh): Likewise.
5810         (iwmmxt_tmovmskw): Likewise.
5811         (iwmmxt_waccb): Likewise.
5812         (iwmmxt_wacch): Likewise.
5813         (iwmmxt_waccw): Likewise.
5814         (iwmmxt_waligni): Likewise.
5815         (iwmmxt_walignr): Likewise.
5816         (iwmmxt_walignr0): Likewise.
5817         (iwmmxt_walignr1): Likewise.
5818         (iwmmxt_walignr2): Likewise.
5819         (iwmmxt_walignr3): Likewise.
5820         (iwmmxt_wsadb): Likewise.
5821         (iwmmxt_wsadh): Likewise.
5822         (iwmmxt_wsadbz): Likewise.
5823         (iwmmxt_wsadhz): Likewise.
5824         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5825         (iwmmxt_wabsdiffb): Likewise.
5826         (iwmmxt_wabsdiffh): Likewise.
5827         (iwmmxt_wabsdiffw): Likewise.
5828         (iwmmxt_waddsubhx): Likewise
5829         (iwmmxt_wsubaddhx): Likewise.
5830         (addc<mode>3): Likewise.
5831         (iwmmxt_avg4): Likewise.
5832         (iwmmxt_avg4r): Likewise.
5833         (iwmmxt_wmaddsx): Likewise.
5834         (iwmmxt_wmaddux): Likewise.
5835         (iwmmxt_wmaddsn): Likewise.
5836         (iwmmxt_wmaddun): Likewise.
5837         (iwmmxt_wmulwsm): Likewise.
5838         (iwmmxt_wmulwum): Likewise.
5839         (iwmmxt_wmulsmr): Likewise.
5840         (iwmmxt_wmulumr): Likewise.
5841         (iwmmxt_wmulwsmr): Likewise.
5842         (iwmmxt_wmulwumr): Likewise.
5843         (iwmmxt_wmulwl): Likewise.
5844         (iwmmxt_wqmulm): Likewise.
5845         (iwmmxt_wqmulwm): Likewise.
5846         (iwmmxt_wqmulmr): Likewise.
5847         (iwmmxt_wqmulwmr): Likewise.
5848         (iwmmxt_waddbhusm): Likewise.
5849         (iwmmxt_waddbhusl): Likewise.
5850         (iwmmxt_wqmiabb): Likewise.
5851         (iwmmxt_wqmiabt): Likewise.
5852         (iwmmxt_wqmiatb): Likewise.
5853         (iwmmxt_wqmiatt): Likewise.
5854         (iwmmxt_wqmiabbn): Likewise.
5855         (iwmmxt_wqmiabtn): Likewise.
5856         (iwmmxt_wqmiatbn): Likewise.
5857         (iwmmxt_wqmiattn): Likewise.
5858         (iwmmxt_wmiabb): Likewise.
5859         (iwmmxt_wmiabt): Likewise.
5860         (iwmmxt_wmiatb): Likewise.
5861         (iwmmxt_wmiatt): Likewise.
5862         (iwmmxt_wmiabbn): Likewise.
5863         (iwmmxt_wmiabtn): Likewise.
5864         (iwmmxt_wmiatbn): Likewise.
5865         (iwmmxt_wmiattn): Likewise.
5866         (iwmmxt_wmiawbb): Likewise.
5867         (iwmmxt_wmiawbt): Likewise.
5868         (iwmmxt_wmiawtb): Likewise.
5869         (iwmmxt_wmiawtt): Likewise.
5870         (iwmmxt_wmiawbbn): Likewise.
5871         (iwmmxt_wmiawbtn): Likewise.
5872         (iwmmxt_wmiawtbn): Likewise.
5873         (iwmmxt_wmiawttn): Likewise.
5874         (iwmmxt_wmerge): Likewise.
5875         (iwmmxt_tandc<mode>3): Likewise.
5876         (iwmmxt_torc<mode>3): Likewise.
5877         (iwmmxt_torvsc<mode>3): Likewise.
5878         (iwmmxt_textrc<mode>3): Likewise.
5879         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5880         (wmmxt_pack): Likewise.
5881         (wmmxt_mult_c1): Likewise.
5882         (wmmxt_mult_c2): Likewise.
5883         (wmmxt_alu_c1): Likewise.
5884         (wmmxt_alu_c2): Likewise.
5885         (wmmxt_alu_c3): Likewise.
5886         (wmmxt_transfer_c1): Likewise.
5887         (wmmxt_transfer_c2): Likewise.
5888         (wmmxt_transfer_c3): Likewise.
5889         (marvell_f_iwmmxt_wstr): Likewise.
5890         (marvell_f_iwmmxt_wldr): Likewise.
5892 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5894         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5896 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5898         Revert:
5899         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5900         * lra-constraints.c (need_for_split_p): Check call used hard regs
5901         living through calls.
5903         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5904         call used regs for call insn.
5906 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5908         PR target/57736
5909         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5910         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5912 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5914         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5916 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5918         * lra-constraints.c (need_for_split_p): Check call used hard regs
5919         living through calls.
5921 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5923         PR target/57744
5924         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5925         to tie with any other modes.  Eliminate Altivec vector mode tests,
5926         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5927         code, to return 0 if testing MODE2 for a condition, if we've
5928         already tested MODE1 for the same condition.
5930 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5932         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5933         layout.
5935 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5937         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5938         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5940 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5942         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5943         Define.
5944         (aarch64_symbolic_constant_p): Remove.
5945         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5946         static.  Fix line length and white space.
5947         (aarch64_symbolic_constant_p): Remove.
5948         * config/aarch64/predicates.md (aarch64_valid_symref):
5949         Use aarch64_classify_symbol_expression.
5951 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5953         * config/arm/constraints.md (Ts): New constraint.
5954         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5955         16-bit encodings.
5956         (compare_scc): Use "Ts" constraint for operand 0.
5957         (ior_scc_scc): Likewise.
5958         (and_scc_scc): Likewise.
5959         (and_scc_scc_nodom): Likewise.
5960         (ior_scc_scc_cmp): Likewise for operand 7.
5961         (and_scc_scc_cmp): Likewise.
5962         * config/arm/thumb2.md (thumb2_movsi_insn):
5963         Add alternatives for 16-bit encodings.
5964         (thumb2_movhi_insn): Likewise.
5965         (thumb2_movsicc_insn): Likewise.
5966         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5967         (thumb2_negscc): Use "Ts" constraint.
5968         Move mvn instruction outside cond_exec block.
5969         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5970         for 16-bit encodings.
5972 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5974         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5975         encoding.
5976         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5977         (mulsi3subsi): Likewise.
5978         (mulsidi3adddi): Likewise.
5979         (mulsidi3_v6): Likewise.
5980         (umulsidi3_v6): Likewise.
5981         (umulsidi3adddi_v6): Likewise.
5982         (smulsi3_highpart_v6): Likewise.
5983         (umulsi3_highpart_v6): Likewise.
5984         (mulhisi3tb): Likewise.
5985         (mulhisi3bt): Likewise.
5986         (mulhisi3tt): Likewise.
5987         (maddhisi4): Likewise.
5988         (maddhisi4tb): Likewise.
5989         (maddhisi4tt): Likewise.
5990         (maddhidi4): Likewise.
5991         (maddhidi4tb): Likewise.
5992         (maddhidi4tt): Likewise.
5993         (zeroextractsi_compare0_scratch): Likewise.
5994         (insv_zero): Likewise.
5995         (insv_t2): Likewise.
5996         (anddi_notzesidi_di): Likewise.
5997         (anddi_notsesidi_di): Likewise.
5998         (andsi_notsi_si): Likewise.
5999         (iordi_zesidi_di): Likewise.
6000         (xordi_zesidi_di): Likewise.
6001         (andsi_iorsi3_notsi): Likewise.
6002         (smax_0): Likewise.
6003         (smax_m1): Likewise.
6004         (smin_0): Likewise.
6005         (not_shiftsi): Likewise.
6006         (unaligned_loadsi): Likewise.
6007         (unaligned_loadhis): Likewise.
6008         (unaligned_loadhiu): Likewise.
6009         (unaligned_storesi): Likewise.
6010         (unaligned_storehi): Likewise.
6011         (extv_reg): Likewise.
6012         (extzv_t2): Likewise.
6013         (divsi3): Likewise.
6014         (udivsi3): Likewise.
6015         (arm_zero_extendhisi2addsi): Likewise.
6016         (arm_zero_extendqisi2addsi): Likewise.
6017         (compareqi_eq0): Likewise.
6018         (arm_extendhisi2_v6): Likewise.
6019         (arm_extendqisi2addsi): Likewise.
6020         (arm_movt): Likewise.
6021         (thumb2_ldrd): Likewise.
6022         (thumb2_ldrd_base): Likewise.
6023         (thumb2_ldrd_base_neg): Likewise.
6024         (thumb2_strd): Likewise.
6025         (thumb2_strd_base): Likewise.
6026         (thumb2_strd_base_neg): Likewise.
6027         (arm_negsi2): Add alternative for 16-bit encoding.
6028         (arm_one_cmplsi2): Likewise.
6030 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6032         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6033         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6034         (movdfcc): Likewise.
6035         * config/arm/vfp.md (*thumb2_movsf_vfp):
6036         Disable predication for arm_restrict_it.
6037         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6038         (*thumb2_movdfcc_vfp): Likewise.
6039         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6040         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6041         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6042         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6043         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6044         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6045         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6046         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6047         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6048         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6049         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6050         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6051         Disable predication for arm_restrict_it.
6053 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
6055         * config/i386/bmiintrin.h (_bextr_u32): New.
6056         (_bextr_u64): Ditto.
6058 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
6060         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6061         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6062         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6063         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6064         * config/mips/n32-elf.h: ...this new file.
6066 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
6068         PR target/57224
6069         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6070         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6072 2013-06-27  Catherine Moore  <clm@codesourcery.com>
6074         * config/mips/mips-tables.opt: Regenerate.
6075         * config/mips/mips-cpus.def: Add m14ke and m14kec.
6076         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6077         * doc/invoke.texi: Add -m14kc.
6079 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6081         PR target/57623
6082         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6083         constraints of operand 1 and 2.
6085         PR target/57623
6086         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6087         to match RTL canonicalization.  Swap predicates and
6088         constraints of operand 1 and 2.
6090 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
6092         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6093         Process OP_INOUT regs for splitting too.
6095 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6097         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6098         decl before the loop, initialize to NULL.
6099         (vectorizable_load): Initialize ptr_incr to NULL.
6101 2013-06-27  Martin Jambor  <mjambor@suse.cz>
6103         PR lto/57208
6104         * ipa-ref.h (ipa_maybe_record_reference): Declare.
6105         * ipa-ref.c (ipa_maybe_record_reference): New function.
6106         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6107         * ipa-cp.c (create_specialized_node): Record potential references from
6108         aggvals.
6109         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6111 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6113         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6114         parameter 'mode' of type 'enum machine_mode mode'; change to pass
6115         'mode' to force_reg.
6116         (aarch64_add_offset): Update calls to aarch64_force_temporary.
6117         (aarch64_expand_mov_immediate): Likewise.
6119 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6121         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6122         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6124 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6126         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6127         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6128         (struct machine_function): Add tbegin_p.
6129         (s390_canonicalize_comparison): Fold CC mode compares to
6130         conditional jump if possible.
6131         (s390_emit_jump): Return the emitted jump.
6132         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6133         Handle CCRAWmode compares.
6134         (s390_option_override): Default to -mhtm if available.
6135         (s390_reg_clobbered_rtx): Handle floating point regs as well.
6136         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6137         FPRs instead of df_regs_ever_live_p.
6138         (s390_optimize_nonescaping_tx): New function.
6139         (s390_init_frame_layout): Extend clobbered_regs array to cover
6140         FPRs as well.
6141         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6142         (s390_expand_tbegin): New function.
6143         (enum s390_builtin): New enum definition.
6144         (code_for_builtin): New array definition.
6145         (s390_init_builtins): New function.
6146         (s390_expand_builtin): New function.
6147         (TARGET_INIT_BUILTINS): Define.
6148         (TARGET_EXPAND_BUILTIN): Define.
6149         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6150         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6151         (s390_alc_comparison): Likewise.
6152         * config/s390/s390-modes.def: Add CCRAWmode.
6153         * config/s390/s390.h (processor_flags): Add PF_TX.
6154         (TARGET_CPU_HTM): Define macro.
6155         (TARGET_HTM): Define macro.
6156         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6157         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6158         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6159         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6160         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6161         (TBEGIN_MASK, TBEGINC_MASK): New constants.
6162         ("*cc_to_int"): Move up.
6163         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6164         constants other than 0.
6165         ("*ccraw_to_int"): New insn and splitter definition.
6166         ("tbegin", "tbegin_nofloat", "tbegin_retry")
6167         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6168         ("tx_assist"): New expander.
6169         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6170         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6171         * config/s390/s390.opt: Add -mhtm option.
6172         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6173         * config/s390/htmxlintrin.h: New file.
6174         * config/s390/htmintrin.h: New file.
6175         * config/s390/s390intrin.h: New file.
6176         * doc/extend.texi: Document htm builtins.
6177         * config.gcc: Add the new header files to extra_headers.
6179 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
6181         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6182         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6184 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
6185             Pat Haugen  <pthaugen@us.ibm.com>
6186             Peter Bergner  <bergner@vnet.ibm.com>
6188         * config/rs6000/power8.md: New.
6189         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6190         setting for power8 entry.
6191         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6192         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6193         test for Power4/Power5 only.
6194         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6195         support.
6196         (force_new_group): Adjust comment.
6197         * config/rs6000/rs6000.md: Include power8.md.
6199 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
6201         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6202         * config/arm/arm-protos.h (arm_max_conditional_execute): New
6203         declaration.
6204         (tune_params): Update comment.
6205         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6206         (arm_max_conditional_execute): New function.
6207         (thumb2_final_prescan_insn): Use max_insn_skipped and
6208         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6210 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
6212         PR tree-optimization/57705
6213         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6214         SSA_NAME step, provided that it is not defined inside the loop.
6215         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6216         (get_initial_def_for_induction): Handle SSA_NAME IV step.
6218 2013-06-25  Martin Jambor  <mjambor@suse.cz>
6220         PR middle-end/57670
6221         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6222         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6223         calls in the dump.
6224         (ipa_note_param_call): Initialize member_ptr flag.
6225         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6226         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6227         (ipa_write_indirect_edge_info): Stream member_ptr flag.
6228         (ipa_read_indirect_edge_info): Likewise.
6230 2013-06-25  Richard Biener  <rguenther@suse.de>
6232         PR middle-end/56977
6233         * passes.c (init_optimization_passes): Move pass_fold_builtins
6234         and pass_dce earlier with -Og.
6236 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
6238         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6239         <BIT_FIELD_REF>: Remove trailing TAB.
6240         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6241         remove blank line.
6243 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6245         PR tree-optimization/57358
6246         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6247         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6248         (ipa_analyze_params_uses): Generate pessimistic info when true.
6250 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6252         PR tree-optimization/57539
6253         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6254         global.inlined_to of the new node to it.  All callers changed.
6255         * ipa-inline-transform.c (clone_inlined_nodes): New variable
6256         inlining_into, pass it to cgraph_clone_node.
6257         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6258         ipa_free_edge_args_substructures.
6259         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6260         rdesc linked list.  Do not assert rdesc edges have inlined caller.
6261         Assert we have found an rdesc in the rdesc list.
6263 2013-06-24  Richard Biener  <rguenther@suse.de>
6265         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6266         (pointer_set_lookup): Declare.
6267         (class pointer_map): New template class implementing a
6268         generic pointer to T map.
6269         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6270         pointer_map<T>::contains, pointer_map<T>::insert,
6271         pointer_map<T>::traverse): New functions.
6272         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6273         (pointer_set_lookup): New function.
6274         (pointer_set_contains): Use pointer_set_lookup.
6275         (pointer_set_insert): Likewise.
6276         (insert_aux): Remove.
6277         (struct pointer_map_t): Embed a pointer_set_t.
6278         (pointer_map_create): Adjust.
6279         (pointer_map_destroy): Likewise.
6280         (pointer_map_contains): Likewise.
6281         (pointer_map_insert): Likewise.
6282         (pointer_map_traverse): Likewise.
6283         * tree-streamer.h (struct streamer_tree_cache_d): Use a
6284         pointer_map<unsigned> instead of a pointer_map_t.
6285         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6286         (streamer_tree_cache_lookup): Likewise.
6287         (streamer_tree_cache_create): Likewise.
6288         (streamer_tree_cache_delete): Likewise.
6289         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6290         pointer_map<unsigned> instead of a pointer_map_t.
6291         (lto_init_tree_ref_encoder): Adjust.
6292         (lto_destroy_tree_ref_encoder): Likewise.
6293         * lto-section-out.c (lto_output_decl_index): Likewise.
6294         (lto_record_function_out_decl_state): Likewise.
6295         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6297 2013-06-24  Richard Biener  <rguenther@suse.de>
6299         PR tree-optimization/57488
6300         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6302 2013-06-24  Alan Modra  <amodra@gmail.com>
6304         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6305         (gen_easy_altivec_constant): Likewise.
6306         * config/rs6000/predicates.md (easy_vector_constant_add_self,
6307         easy_vector_constant_msb): Likewise.
6309 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
6311         PR target/57688
6312         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6313         add missing return true.
6315 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
6317         PR target/52483
6318         * config/sh/predicates.md (general_extend_operand): Invoke
6319         general_movsrc_operand for memory operands.
6320         (general_movsrc_operand): Allow reg+reg addressing, do not use
6321         general_operand for memory operands.
6323 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
6325         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
6326         when current target options does not apply.
6327         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6328         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6329         * config/i386/bmiintrin.h: Pass appropriate target
6330         attributes to header.
6331         * config/i386/mmintrin.h: Ditto.
6332         * config/i386/nmmintrin.h: Ditto.
6333         * config/i386/avx2intrin.h: Ditto.
6334         * config/i386/fxsrintrin.h: Ditto.
6335         * config/i386/tbmintrin.h: Ditto.
6336         * config/i386/xsaveintrin.h: Ditto.
6337         * config/i386/f16cintrin.h: Ditto.
6338         * config/i386/xtestintrin.h: Ditto.
6339         * config/i386/xsaveoptintrin.h: Ditto.
6340         * config/i386/bmi2intrin.h: Ditto.
6341         * config/i386/lzcntintrin.h: Ditto.
6342         * config/i386/smmintrin.h: Ditto.
6343         * config/i386/wmmintrin.h: Ditto.
6344         * config/i386/x86intrin.h: Remove all header include guards.
6345         * config/i386/prfchwintrin.h: Ditto.
6346         * config/i386/pmmintrin.h: Ditto.
6347         * config/i386/tmmintrin.h: Ditto.
6348         * config/i386/xmmintrin.h: Ditto.
6349         * config/i386/popcntintrin.h: Ditto.
6350         * config/i386/rdseedintrin.h: Ditto.
6351         * config/i386/ammintrin.h: Ditto.
6352         * config/i386/emmintrin.h: Ditto.
6353         * config/i386/immintrin.h: Remove all header include guards.
6354         * config/i386/fma4intrin.h: Ditto.
6355         * config/i386/lwpintrin.h: Ditto.
6356         * config/i386/xopintrin.h: Ditto.
6357         * config/i386/ia32intrin.h: Ditto.
6358         * config/i386/avxintrin.h: Ditto.
6359         * config/i386/rtmintrin.h: Ditto.
6360         * config/i386/fmaintrin.h: Ditto.
6361         * config/i386/mm3dnow.h: Ditto.
6363 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
6365         * common/config/i386/i386-common.c: Handle LZCNT.
6367 2013-06-22  Andi Kleen  <ak@linux.intel.com>
6369         * doc/extend.texi: Use __atomic_store_n instead of
6370         __atomic_store in HLE example.
6372 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
6374         * config/sh/sh.c: Remove <cstdlib> workaround.
6376 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6378         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
6380 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6382         * doc/extend.texi: Document that __atomic_clear and
6383           __atomic_test_and_set should only be used with bool.
6385 2013-06-20  Jan Hubicka  <jh@suse.cz>
6387         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6388         types_same_for_odr.
6389         * tree.c (decls_same_for_odr): New function.
6390         (same_for_edr): New function.
6391         (types_same_for_odr): New function.
6392         (get_binfo_at_offset): Use it.
6393         * tree.h (types_same_for_odr): Declare.
6395 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
6396             Jason Merrill  <jason@redhat.com>
6398         * system.h: Include <cstdlib> as well as <stdlib.h>.
6400 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
6402         PR target/57655
6403         * config/i386/i386.c (construct_container): Report error if
6404         long double is used with disabled x87 float returns.
6406 2013-06-20  Jan Hubicka  <jh@suse.cz>
6408         * lto-cgraph.c (input_symtab): Do not set cgraph state.
6410 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
6412         PR rtl-optimization/57425
6413         PR rtl-optimization/57569
6414         * alias.c (write_dependence_p): Remove parameters mem_mode and
6415         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
6416         Changed all callers.
6417         (canon_anti_dependence): Get comments and semantics in sync.
6418         Add parameter mem_canonicalized.  Changed all callers.
6419         * rtl.h (canon_anti_dependence): Update prototype.
6421 2013-06-20  Richard Biener  <rguenther@suse.de>
6423         * data-streamer-in.c (streamer_read_uhwi): Optimize single
6424         byte case, inline streamer_read_uchar and defer section
6425         overrun check.
6427 2013-06-20  Richard Biener  <rguenther@suse.de>
6429         PR tree-optimization/57584
6430         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6431         SSA names into the expanded expression that take part in
6432         abnormal coalescing.
6434 2013-06-19  Sharad Singhai  <singhai@google.com>
6436         * gcov.c (print_usage): Handle new option.
6437         (process_args): Ditto.
6438         (get_gcov_intermediate_filename): New function.
6439         (output_intermediate_file): New function.
6440         (output_gcov_file): New function
6441         (generate_results): Handle new option.
6442         (release_function): Relase demangled name.
6443         (read_graph_file): Handle demangled name.
6444         (output_lines): Ditto.
6445         * doc/gcov.texi: Document gcov intermediate format.
6447 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
6449         PR bootstrap/57604
6450         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6451         (lra_emit_add): Use the functions.  Add comment about Y as an
6452         address segment.
6454 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
6456         PR driver/57652
6457         * collect2.c (collect_atexit): New.
6458         (collect_exit): Delete.
6459         (main): Register collect_atexit with atexit.
6460         (collect_wait): Change collect_exit to exit.
6461         (do_wait): Same.
6462         * collect2.h (collect_exit): Delete.
6463         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6465 2013-06-19  Wei Mi  <wmi@google.com>
6467         PR rtl-optimization/57518
6468         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6469         if regno is used in paradoxical subreg.
6470         (update_equiv_regs): Check pdx_subregs[regno] before
6471         set a reg to be equivalent with a mem.
6473 2013-06-19  Matthias Klose  <doko@ubuntu.com>
6475         PR driver/57651
6476         * file-find.h (find_a_file): Add a mode parameter.
6477         * file-find.c (find_a_file): Likewise.
6478         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6479         with X_OK for the executables.
6480         * collect2.c (main): Call find_a_file with X_OK.
6482 2013-06-19  Steve Ellcey  <sellcey@mips.com>
6484         PR target/56942
6485         * config/mips/mips.md (casesi_internal_mips16_<mode>):
6486         Use NEXT_INSN instead of next_real_insn.
6488 2013-06-19  Jan Hubicka  <jh@suse.cz>
6490         * cgraph.h (const_value_known_p): Replace by ...
6491         (ctor_for_folding): .. this one.
6492         * cgraphunit.c (process_function_and_variable_attributes): Use it.
6493         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6494         * expr.c (expand_expr_real_1): Likewise.
6495         (string_constant): Likewise.
6496         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6497         * ipa.c (process_references): Likewise.
6498         (symtab_remove_unreachable_nodes): Likewise.
6499         * ipa-inline-analysis.c (param_change_prob): Likewise.
6500         * gimple-fold.c (canonicalize_constructor_val): Likewise.
6501         (get_base_constructor): Likwise.
6502         * varpool.c (varpool_remove_node): Likewise.
6503         (varpool_remove_initializer): LIkewise.
6504         (dump_varpool_node): LIkwise.
6505         (const_value_known_p): Rewrite to ...
6506         (ctor_for_folding): ... this one.
6508 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
6510         PR driver/57651
6511         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6512         PERSONALITY in $PATH derived prefixes.
6514 2013-06-19  Jeff Law  <law@redhat.com>
6516         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6517         in comment.
6519         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6520         (simplify_bitwise_binary): Use it to simpify certain binary ops on
6521         booleans.
6523 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
6525         * config/arm/vfp.md: Move VFP instruction classification documentation
6526         to ...
6527         * config/arm/arm.md: ... here.  Update instruction classification
6528         documentation.
6530 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
6532         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6533         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6534         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
6536 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
6538         PR target/57609
6539         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6540         with NEXT_INSN.  Use tablejump_p to check for jump table data
6541         insns.
6543 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
6545         PR c++/56544
6546         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6547         that now in C++ the value is correct per the C++ standards.
6549 2013-06-19  Richard Biener  <rguenther@suse.de>
6551         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6552         for global context.
6554 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6556         Revert:
6557         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6559         PR target/57609
6560         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6561         with next_active_insn.
6563 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
6565         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6566         functions are inlined during failures to flag an error.
6567         * tree-inline.c (expand_call_inline): Allow the error to be flagged
6568         in early inline pass.
6570 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
6572         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6573         in comments.
6575 2013-06-18  Julian Brown  <julian@codesourcery.com>
6577         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6578         Permit virtual register pre-reload if !strict.
6579         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6580         change.
6581         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6582         prototype.
6583         * config/arm/neon.md (movmisalign<mode>): Use
6584         neon_perm_struct_or_reg_operand instead of
6585         neon_struct_or_register_operand.
6586         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6587         neon_permissive_struct_operand instead of neon_struct_operand.
6588         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6589         neon_vector_mem_operand.
6590         * config/arm/predicates.md (neon_struct_operand): Adjust call to
6591         neon_vector_mem_operand.
6592         (neon_permissive_struct_operand): New.
6593         (neon_struct_or_register_operand): Rename to...
6594         (neon_perm_struct_or_reg_operand): This. Adjust call to
6595         neon_vector_mem_operand.
6597 2013-06-18  Richard Biener  <rguenther@suse.de>
6599         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6600         * lto-streamer.h: Include pointer-set.h.
6601         (struct lto_decl_slot): Remove.
6602         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6603         Remove next_index entry.
6604         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6605         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6606         (lto_init_tree_ref_encoder): Adjust.
6607         (lto_destroy_tree_ref_encoder): Likewise.
6608         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6609         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6610         (lto_output_decl_index): Adjust.
6611         (lto_new_out_decl_state): Likewise.
6612         (lto_record_function_out_decl_state): Likewise.
6613         * lto-streamer-out.c (copy_function): Likewise.
6615 2013-06-18  Richard Biener  <rguenther@suse.de>
6617         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6618         * cgraphunit.c: Include cfgloop.h.
6619         (init_lowered_empty_function): Initialize the loop tree.
6620         (assemble_thunk): Insert new BBs into loops.
6622 2013-06-18  Richard Biener  <rguenther@suse.de>
6624         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6625         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6626         the map from cache entry to cache index optional.
6627         (streamer_tree_cache_replace_tree): Adjust accordingly.
6628         (streamer_tree_cache_append): Likewise.
6629         (streamer_tree_cache_delete): Likewise.
6630         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6631         streamer cache map from cache entry to cache index.
6632         * lto-streamer-out.c (create_output_block): Adjust.
6634 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
6636         * config/arm/arm.md (attribute "insn"): Move multiplication and
6637         division attributes to...
6638         (attribute "type"): ... here.  Remove mult.
6639         (attribute "mul32"): New attribute.
6640         (attribute "mul64"): Add umaal.
6641         (*arm_mulsi3): Update attributes.
6642         (*arm_mulsi3_v6): Likewise.
6643         (*thumb_mulsi3): Likewise.
6644         (*thumb_mulsi3_v6): Likewise.
6645         (*mulsi3_compare0): Likewise.
6646         (*mulsi3_compare0_v6): Likewise.
6647         (*mulsi_compare0_scratch): Likewise.
6648         (*mulsi_compare0_scratch_v6): Likewise.
6649         (*mulsi3addsi): Likewise.
6650         (*mulsi3addsi_v6): Likewise.
6651         (*mulsi3addsi_compare0): Likewise.
6652         (*mulsi3addsi_compare0_v6): Likewise.
6653         (*mulsi3addsi_compare0_scratch): Likewise.
6654         (*mulsi3addsi_compare0_scratch_v6): Likewise.
6655         (*mulsi3subsi): Likewise.
6656         (*mulsidi3adddi): Likewise.
6657         (*mulsi3addsi_v6): Likewise.
6658         (*mulsidi3adddi_v6): Likewise.
6659         (*mulsidi3_nov6): Likewise.
6660         (*mulsidi3_v6): Likewise.
6661         (*umulsidi3_nov6): Likewise.
6662         (*umulsidi3_v6): Likewise.
6663         (*umulsidi3adddi): Likewise.
6664         (*umulsidi3adddi_v6): Likewise.
6665         (*smulsi3_highpart_nov6): Likewise.
6666         (*smulsi3_highpart_v6): Likewise.
6667         (*umulsi3_highpart_nov6): Likewise.
6668         (*umulsi3_highpart_v6): Likewise.
6669         (mulhisi3): Likewise.
6670         (*mulhisi3tb): Likewise.
6671         (*mulhisi3bt): Likewise.
6672         (*mulhisi3tt): Likewise.
6673         (maddhisi4): Likewise.
6674         (*maddhisi4tb): Likewise.
6675         (*maddhisi4tt): Likewise.
6676         (maddhidi4): Likewise.
6677         (*maddhidi4tb): Likewise.
6678         (*maddhidi4tt): Likewise.
6679         (divsi3): Likewise.
6680         (udivsi3): Likewise.
6681         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6682         (thumb2_mulsi_short_compare0): Likewise.
6683         (thumb2_mulsi_short_compare0_scratch): Likewise.
6684         * config/arm/arm1020e.md (1020mult1): Update attribute change.
6685         (1020mult2): Likewise.
6686         (1020mult3): Likewise.
6687         (1020mult4): Likewise.
6688         (1020mult5): Likewise.
6689         (1020mult6): Likewise.
6690         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6691         change.
6692         (cortex_a15_mult64): Likewise.
6693         (cortex_a15_sdiv): Likewise.
6694         (cortex_a15_udiv): Likewise.
6695         * config/arm/arm1026ejs.md (mult1): Update attribute change.
6696         (mult2): Likewise.
6697         (mult3): Likewise.
6698         (mult4): Likewise.
6699         (mult5): Likewise.
6700         (mult6): Likewise.
6701         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6702         (pj4_ir_div): Likewise.
6703         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6704         (11_mult2): Likewise.
6705         (11_mult3): Likewise.
6706         (11_mult4): Likewise.
6707         (11_mult5): Likewise.
6708         (11_mult6): Likewise.
6709         (11_mult7): Likewise.
6710         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6711         (cortex_a8_mla): Likewise.
6712         (cortex_a8_mull): Likewise.
6713         (cortex_a8_smulwy): Likewise.
6714         (cortex_a8_smlald): Likewise.
6715         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6716         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6717         (cortex_r4_mul_3): Likewise.
6718         (cortex_r4_mla_4): Likewise.
6719         (cortex_r4_mla_3): Likewise.
6720         (cortex_r4_smlald): Likewise.
6721         (cortex_r4_mull): Likewise.
6722         (cortex_r4_sdiv): Likewise.
6723         (cortex_r4_udiv): Likewise.
6724         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6725         (cortex_a7_idiv): Likewise.
6726         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6727         (9_mult2): Likewise.
6728         (9_mult3): Likewise.
6729         (9_mult4): Likewise.
6730         (9_mult5): Likewise.
6731         (9_mult6): Likewise.
6732         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6733         (cortex_a53_sdiv): Likewise.
6734         (cortex_a53_udiv): Likewise.
6735         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6736         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6737         (mp626_mult2): Likewise.
6738         (mp626_mult3): Likewise.
6739         (mp626_mult4): Likewise.
6740         * config/arm/fa526.md (526_mult1): Update attribute change.
6741         (526_mult2): Likewise.
6742         * config/arm/arm-generic.md (mult): Update attribute change.
6743         (mult_ldsched_strongarm): Likewise.
6744         (mult_ldsched): Likewise.
6745         (multi_cycle): Likewise.
6746         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6747         * config/arm/fa606te.md (606te_mult1): Update attribute change.
6748         (606te_mult2): Likewise.
6749         (606te_mult3): Likewise.
6750         (606te_mult4): Likewise.
6751         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6752         (cortex_a9_mac16): Likewise.
6753         (cortex_a9_multiply): Likewise.
6754         (cortex_a9_mac): Likewise.
6755         (cortex_a9_multiply_long): Likewise.
6756         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6757         (626te_mult2): Likewise.
6758         (626te_mult3): Likewise.
6759         (626te_mult4): Likewise.
6761 2013-06-18  Richard Biener  <rguenther@suse.de>
6763         PR lto/57334
6764         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6766 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6768         PR target/57609
6769         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6770         with next_active_insn.
6772 2013-06-18  Alan Modra  <amodra@gmail.com>
6774         * config/rs6000/rs6000.h (enum data_align): New.
6775         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6776         (DATA_ABI_ALIGNMENT): Define.
6777         (CONSTANT_ALIGNMENT): Correct comment.
6778         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6779         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6781 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6783         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6784         ATTRIBUTE_UNUSED marking.
6786 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6788         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6789         alternative and update.
6790         (aarch64_dup_lanedi): Delete.
6791         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6792         * config/aarch64/aarch64-simd-builtins.def: Update.
6794 2013-06-17  Richard Biener  <rguenther@suse.de>
6796         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6797         (lto_input_scc): Declare.
6798         (lto_input_tree_1): Likewise.
6799         (struct lto_stats_d): Add num_tree_bodies_output and
6800         num_pickle_refs_output.
6801         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6802         (lto_read_tree_1): Split out from ...
6803         (lto_read_tree): ... this.
6804         (lto_input_scc): New function.
6805         (lto_input_tree_1): Split out from ...
6806         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6807         (lto_data_in_create): Create the streamer cache without hashes.
6808         * lto-streamer-out.c (create_output_block): Create the streamer
6809         cache with hashes when not doing WPA.
6810         (lto_write_tree_1): Split out from ...
6811         (lto_write_tree): ... this.
6812         (get_symbol_initial_value): New function.
6813         (lto_output_tree_1): Split out from ...
6814         (lto_output_tree): ... this.  Write trees as series of SCCs
6815         using a DFS walk via DFS_write_tree.
6816         (struct sccs, struct scc_entry): New types.
6817         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6818         (DFS_write_tree_body): New function.
6819         (DFS_write_tree): Likewise.
6820         (hash_tree): Likewise.
6821         (scc_entry_compare): Likewise.
6822         (hash_scc): Likewise.
6823         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6824         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6825         TREE_CHAIN as regular reference.
6826         (streamer_read_integer_cst): Remove.
6827         (streamer_get_pickled_tree): Adjust.
6828         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6829         of DECL_EXTERNALs in BLOCK_VARS for now.
6830         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6831         reference.
6832         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6833         Add hash value argument and record that if hashes are recorded
6834         in the cache.
6835         (streamer_tree_cache_insert_1): Adjust.
6836         (streamer_tree_cache_insert): Likewise.
6837         (streamer_tree_cache_insert_at): Rename to ...
6838         (streamer_tree_cache_replace_tree): ... this and adjust.
6839         (streamer_tree_cache_append): Adjust.
6840         (record_common_node): Likewise.
6841         (streamer_tree_cache_create): Add argument whether to
6842         record hash values together with trees.
6843         (streamer_tree_cache_delete): Adjust.
6844         * tree-streamer.h (struct streamer_tree_cache_d): Add
6845         vector of hashes.
6846         (streamer_read_integer_cst): Remove.
6847         (streamer_tree_cache_insert): Adjust.
6848         (streamer_tree_cache_append): Likewise.
6849         (streamer_tree_cache_insert_at): Rename to ...
6850         (streamer_tree_cache_replace_tree): ... this and adjust.
6851         (streamer_tree_cache_create): Add argument whether to record hashes.
6852         (streamer_tree_cache_get): Rename to ...
6853         (streamer_tree_cache_get_tree): ... this.
6854         (streamer_tree_cache_get_hash): New function.
6855         * tree.c (cache_integer_cst): New function.
6856         * tree.h (cache_integer_cst): Declare.
6857         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6858         * lto-symtab.c (lto_varpool_replace_node): Only release
6859         DECL_INITIAL of non-prevailing decls.
6860         * varpool.c (varpool_remove_initializer): Do not release
6861         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6863 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6865         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6866         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6867         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6868         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6869         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6870         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6871         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6872         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6873         instead of TARGET_64BIT.
6874         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6875         Require ISA_HAS_<D>DIV.
6877 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6879         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6880         (mips*-*-linux*): Move default with_llsc setting to where other
6881         defaults are set.
6882         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6883         with_arch block.
6884         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6885         Likewise.  Remove default with_tune setting.  Move default float
6886         setting to its own block.  Handle with_llsc in the same block as above.
6888 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
6890         PR rtl-optimization/57425
6891         PR rtl-optimization/57569
6892         * alias.c (write_dependence_p): Add new parameters mem_mode,
6893         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6894         Changed all callers.
6895         (canon_anti_dependence): New function.
6896         * cse.c (check_dependence): Use canon_anti_dependence.
6897         * cselib.c (cselib_invalidate_mem): Likewise.
6898         * rtl.h (canon_anti_dependence): Declare.
6900 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6902         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6903         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6904         ".set mips3" for 64-bit targets.
6906 2013-06-15  Dehao Chen  <dehao@google.com>
6908         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6909         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6910         (gimple_check_call_args): Likewise.
6911         * value-prof.c (check_ic_target): Likewise.
6912         * ipa-inline.c (early_inliner): Likewise.
6913         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6914         * cgraph.c (cgraph_create_edge_1): Likewise.
6915         (cgraph_make_edge_direct): Likewise.
6917 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6919         PR target/57615
6920         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6921         rs6000_output_move_128bit to handle emitting quad memory
6922         operations.  Set attribute length to 8 bytes.
6924 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
6926         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6927         New pattern.
6928         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6929         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6930         (aarch64_<su>mlsl<mode>): Likewise.
6932 2013-06-14  Mike Stump  <mikestump@comcast.net>
6934         * Makefile.in (TARGET_H): Add insn-codes.h.
6936 2013-06-14  Alan Modra  <amodra@gmail.com>
6938         PR middle-end/57134
6939         PR middle-end/57586
6940         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6941         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6942         bitfield expansion when EXPAND_MEMORY.
6943         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6945 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6947         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6948         test for clearing quad memory on 32-bit later.
6950 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6952         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6953         (fold_negate_expr): Likewise.
6954         (fold_real_zero_addition_p): Handle vectors.
6955         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6957 2013-06-14  Alan Modra  <amodra@gmail.com>
6959         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6961 2013-06-13  Jan Hubicka  <jh@suse.cz>
6963         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6964         Local comdats are not externally visible.
6965         * symtab.c (dump_symtab_base): Dump externally visible.
6966         (verify_symtab_base): Verify back links in the symtab hash.
6968 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6970         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6971         CONVERT_EXPR as equal nodes.
6973 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6975         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6977 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6979         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6980         Generalize to complex and vector.
6981         * tree.c (build_all_ones_cst): New function.
6982         * tree.h (build_all_ones_cst): Declare it.
6984 2013-06-13  Alan Modra  <amodra@gmail.com>
6986         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6987         * config/rs6000/rs6000.md (signbittf2): New insn.
6988         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6989         (abstf2_internal, cmptf_internal2): Likewise.
6990         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6992 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6993             Pat Haugen  <pthaugen@us.ibm.com>
6994             Peter Bergner  <bergner@vnet.ibm.com>
6996         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6997         power8 byte, half-word, and quad-word atomic instructions.
6998         (emit_store_conditional): Likewise.
6999         (rs6000_expand_atomic_compare_and_swap): Likewise.
7000         (rs6000_expand_atomic_op): Likewise.
7002         * config/rs6000/sync.md (larx): Add new modes for power8.
7003         (stcx): Likewise.
7004         (AINT): New mode iterator to include TImode as well as normal
7005         integer modes on power8.
7006         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
7007         that VSX registers are not considered.  Use AINT mode iterator
7008         instead of INT1 to allow inclusion of quad word atomic operations
7009         on power8.
7010         (load_locked<mode>): Likewise.
7011         (store_conditional<mode>): Likewise.
7012         (atomic_compare_and_swap<mode>): Likewise.
7013         (atomic_exchange<mode>): Likewise.
7014         (atomic_nand<mode>): Likewise.
7015         (atomic_fetch_<fetchop_name><mode>): Likewise.
7016         (atomic_nand_fetch<mode>): Likewise.
7017         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7018         each type.
7019         (ATOMIC): On power8, add QImode, HImode modes.
7020         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7021         modes that promote to SImode.
7022         (load_lockedti): Convert TImode arguments to PTImode, so that we
7023         get a guaranteed even/odd register pair.
7024         (load_lockedpti): Likewise.
7025         (store_conditionalti): Likewise.
7026         (store_conditionalpti): Likewise.
7028         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7029         atomic load/store instructions.
7030         (HSI): Likewise.
7032 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
7034         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7035         loads.
7036         (insn_count): New attribute, with most cases extracted from...
7037         (length): ...here.  Redefine most cases in terms of insn_count.
7038         (single_insn): Delete.
7039         (can_delay): Use insn_count to check for single instructions.
7040         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7041         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7042         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7043         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7044         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7045         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7046         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7047         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7048         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7049         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7050         rather than "length".
7051         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
7052         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7053         Use "insn_count" rather than "length".
7054         * config/mips/mips-dsp.md
7055         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7056         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7057         length attributes.
7059 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
7061         PR tree-optimization/57361
7062         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7064 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
7066         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7067         to split.
7068         (aarch64_simd_combine<mode>): New instruction expansion.
7069         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7070         function prototype.
7071         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7072         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7074 2013-06-12  Jan Hubicka  <jh@suse.cz>
7076         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7077         decl has when in streaming stage.
7078         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7079         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7081 2013-06-12  Roland Stigge  <stigge@antcom.de>
7083         PR target/57578
7084         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7086 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7088         PR tree-optimization/57537
7089         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7090         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7092 2013-06-12  Richard Biener  <rguenther@suse.de>
7094         * data-streamer.h (streamer_write_char_stream): CSE
7095         obs->current_pointer.
7096         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7097         streamer_write_char_stream manually and optimize the resulting loop.
7098         (streamer_write_hwi_stream): Likewise.
7100 2013-06-12  Jan Hubicka  <jh@suse.cz>
7102         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7103         * cgraph.h (varpool_create_empty_node): Declare.
7104         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7105         duplicated nodes.
7106         * symtab.c (symtab_unregister_node): Be lax about missin entries
7107         in node hash.
7108         (symtab_get_node): Update comment.
7109         * varpool.c (varpool_create_empty_node): Break out from ...
7110         (varpool_node_for_decl): ... here.
7111         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7113 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
7115         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7116         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7117         part.  Use straight-line flow at the end.
7118         <COMPONENT_REF>: Remove superfluous else.
7119         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7121 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7123         PR target/56564
7124         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7125         target hook even for !TREE_PUBLIC decls.  If no resolution info
7126         is available, return false for common and external decls.
7128 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7130         * config/rl78/constraints.md (U): New constraint.
7131         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7132         valloc attribute.
7134 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7136         PR target/57589
7137         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7138         to allow returning address to AT_PLATFORM name.
7140 2013-06-11  Jan Hubicka  <jh@suse.cz>
7142         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7143         * cgraph.h (symtab_node_base): Add weakref flag.
7144         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7145         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7146         (output_weakrefs): Use weakref flag.
7147         * fold-const.c (simple_operand_p): Handle WEAK.
7148         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7149         * ipa.c (varpool_externally_visible_p): Drop weakref.
7150         (function_and_variable_visibility): Update comment; fix weakref
7151         sanity checks; do not clear DECL_WEAK on them.
7152         * lto-cgraph.c (lto_output_node): update.
7153         (lto_output_varpool_node): Update.
7154         (input_overwrite_node): Update.
7155         (input_node): Update.
7156         (input_varpool_node): Update.
7157         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7158         (lto_symtab_merge_symbols): Add sanity check.
7159         (lto_symtab_prevailing_decl): Do not special case weakrefs.
7160         * passes.c (rest_of_decl_compilation): Set static flag, too.
7161         * symtab.c (dump_symtab_base): Dump weakref.
7162         (verify_symtab_base): Sanity check weakrefs.
7163         (symtab_make_decl_local): Remove duplicated code.
7164         (symtab_alias_ultimate_target): Simplify.
7165         * varpool.c (varpool_create_variable_alias): Set weakref flag.
7167 2013-06-11  DJ Delorie  <dj@redhat.com>
7169         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7170         (rl78_unwind_word_mode): New.
7172 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7174         * final.c (debug_prefix_maps): Make static.
7176 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7178         * function.c (initial_trampoline): Remove stray copy.
7180 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
7182         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7184 2013-06-11  Martin Jambor  <mjambor@suse.cz>
7186         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7187         within bounds at the beginning of the function.
7189 2013-06-11  Alan Modra  <amodra@gmail.com>
7191         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
7192         reporting.
7193         (get_named_section): Don't NULL !DECL_P decl.
7195 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
7197         * doc/invoke.texi (core-avx2): Document.
7198         (slm): Likewise.
7199         (atom): Updated with MOVBE.
7201 2013-06-11  Richard Biener  <rguenther@suse.de>
7203         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7205 2013-06-11  Anton Blanchard  <anton@samba.org>
7207         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7208         correct shift value in little-endian mode.
7210 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
7212         PR target/56564
7213         * varasm.c (get_variable_align): Move #endif to the right place.
7215 2013-06-10  Cary Coutant  <ccoutant@google.com>
7217         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7218         for hash so that hash table traversal order is deterministic.
7220 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
7221             Pat Haugen  <pthaugen@us.ibm.com>
7222             Peter Bergner  <bergner@vnet.ibm.com>
7224         * config/rs6000/vector.md (GPR move splitter): Do not split moves
7225         of vectors in GPRS if they are direct moves or quad word load or
7226         store moves.
7228         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7229         declaration.
7230         (direct_move_p): Likewise.
7231         (quad_load_store_p): Likewise.
7233         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7234         classes into bins based on the physical register type.
7235         (reg_class_to_reg_type): Likewise.
7236         (IS_STD_REG_TYPE): Likewise.
7237         (IS_FP_VECT_REG_TYPE): Likewise.
7238         (reload_fpr_gpr): Arrays to determine what insn to use if we can
7239         use direct move instructions.
7240         (reload_gpr_vsx): Likewise.
7241         (reload_vsx_gpr): Likewise.
7242         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7243         information that is a simplification of register classes.  Also
7244         precalculate direct move reload helpers.
7245         (direct_move_p): New function to return true if the operation can
7246         be done as a direct move instruciton.
7247         (quad_load_store_p): New function to return true if the operation
7248         is a quad memory operation.
7249         (rs6000_legitimize_address): If quad memory, only allow register
7250         indirect for TImode addresses.
7251         (rs6000_legitimate_address_p): Likewise.
7252         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7253         (rs6000_reload_register_type): Likewise.
7254         (register_to_reg_type): Return register type.
7255         (rs6000_secondary_reload_simple_move): New helper function for
7256         secondary reload and secondary memory needed to identify anything
7257         that is a simple move, and does not need reloading.
7258         (rs6000_secondary_reload_direct_move): New helper function for
7259         secondary reload to identify cases that can be done with several
7260         instructions via the direct move instructions.
7261         (rs6000_secondary_reload_move): New helper function for secondary
7262         reload to identify moves between register types that can be done.
7263         (rs6000_secondary_reload): Add support for quad memory operations
7264         and for direct move.
7265         (rs6000_secondary_memory_needed): Likewise.
7266         (rs6000_debug_secondary_memory_needed): Change argument names.
7267         (rs6000_output_move_128bit): New function to return the move to
7268         use for 128-bit moves, including knowing about the various
7269         limitations of quad memory operations.
7271         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7272         memory operations.  call rs6000_output_move_128bit for the actual
7273         instruciton(s) to generate.
7274         (vsx_movti_64bit): Likewise.
7276         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7277         (UNSPEC_P8V_MTVSRWZ): Likewise.
7278         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7279         (UNSPEC_P8V_MTVSRD): Likewise.
7280         (UNSPEC_P8V_XXPERMDI): Likewise.
7281         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7282         (UNSPEC_FUSION_GPR): Likewise.
7283         (FMOVE128_GPR): New iterator for direct move.
7284         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7285         (f32_sv): Likewise.
7286         (f32_dm): Likewise.
7287         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7288         loads and direct move instructions.
7289         (zero_extendsidi2_lfiwzx): Likewise.
7290         (extendsidi2_lfiwax): Likewise.
7291         (extendsidi2_nocell): Likewise.
7292         (floatsi<mode>2_lfiwax): Likewise.
7293         (lfiwax): Likewise.
7294         (floatunssi<mode>2_lfiwzx): Likewise.
7295         (lfiwzx): Likewise.
7296         (fix_trunc<mode>_stfiwx): Likewise.
7297         (fixuns_trunc<mode>_stfiwx): Likewise.
7298         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7299         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7300         (parity<mode>2_cmpb): Set length/type attr.
7301         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7302         for 'mr.' to fast_compare.
7303         (bpermd_<mode>): Change type attr to popcnt.
7304         (p8_fmrgow_<mode>): New insns for power8 direct move support.
7305         (p8_mtvsrwz_1): Likewise.
7306         (p8_mtvsrwz_2): Likewise.
7307         (reload_fpr_from_gpr<mode>): Likewise.
7308         (p8_mtvsrd_1): Likewise.
7309         (p8_mtvsrd_2): Likewise.
7310         (p8_xxpermdi_<mode>): Likewise.
7311         (reload_vsx_from_gpr<mode>): Likewise.
7312         (reload_vsx_from_gprsf): Likewise.
7313         (p8_mfvsrd_3_<mode>): LIkewise.
7314         (reload_gpr_from_vsx<mode>): Likewise.
7315         (reload_gpr_from_vsxsf): Likewise.
7316         (p8_mfvsrd_4_disf): Likewise.
7317         (multi-word GPR splits): Do not split direct moves or quad memory
7318         operations.
7320 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
7322         * tree-into-ssa.c (interesting_blocks): Make static.
7324 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
7326         PR target/56564
7327         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7328         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7329         Use DATA_ABI_ALIGNMENT for that case instead if defined.
7330         (get_variable_align): New function.
7331         (get_variable_section, emit_bss, emit_common,
7332         assemble_variable_contents, place_block_symbol): Use
7333         get_variable_align instead of DECL_ALIGN.
7334         (assemble_noswitch_variable): Add align argument, use it
7335         instead of DECL_ALIGN.
7336         (assemble_variable): Adjust caller.  Use get_variable_align
7337         instead of DECL_ALIGN.
7338         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7339         caller.
7340         (DATA_ABI_ALIGNMENT): Define.
7341         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7342         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
7343         opt is false, only return the psABI mandated alignment increase.
7344         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7345         (DATA_ABI_ALIGNMENT): ... this.
7346         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7347         (DATA_ABI_ALIGNMENT): ... this.
7348         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7349         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7350         (DATA_ABI_ALIGNMENT): ... this.
7351         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7352         * doc/tm.texi: Regenerated.
7354 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
7356         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7357         cmp_code to construct REG_EQUAL note.
7359 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
7361         PR target/57568
7362         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7363         that operands[2] doesn't overlap with operands[0].
7365 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
7366             Jan Hubicka  <jh@suse.cz>
7368         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7369         hack to mark symbols as used.
7371 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
7373         PR rtl-optimization/57559
7374         * lra-constraints.c (process_alt_operands): Don't discourage
7375         memory with known offset for offsetable memory constraint.
7376         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7378 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
7380         * varasm.c (struct oc_local_state): Reorder fields.
7381         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7382         and adjust accordingly.
7383         (output_constructor): Reorder initialization code and adjust call to
7384         output_constructor_bitfield.
7386 2013-06-07  Jan Hubicka  <jh@suse.cz>
7388         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7390 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
7392         * tree-object-size.c (unknown): Make const.
7394 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7396         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7397         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7398         for last alternative in the cpu_facility attribute.
7400 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7402         PR target/56315
7403         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7404         (xordi3): Change operand 2 constraint to arm_xordi_operand.
7405         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7406         * config/arm/constraints.md (Dg): New constraint.
7407         * config/arm/neon.md (xordi3_neon): Remove.
7408         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7409         * config/arm/predicates.md (arm_xordi_operand): New predicate.
7411 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7413         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7414         Clean up alternatives.
7416 2013-06-07  Alan Modra  <amodra@gmail.com>
7418         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7419         va_list_gpr_size.
7421 2013-06-07  Alan Modra  <amodra@gmail.com>
7423         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7425 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7427         * config/arm/constraints.md (Df): New constraint.
7428         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7429         Correct length attribute for last two alternatives.
7431 2013-06-07  Alan Modra  <amodra@gmail.com>
7433         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7434         override user -mfp-in-toc.
7435         (offsettable_ok_by_alignment): Consider just the current access
7436         rather than the whole object, unless BLKmode.  Handle
7437         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7438         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7439         for -mcmodel=medium.
7440         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7441         override user -mfp-in-toc or -msum-in-toc.  Default to
7442         -mno-fp-in-toc for -mcmodel=medium.
7444 2013-06-06  DJ Delorie  <dj@redhat.com>
7446         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7447         TARGET_VALID_POINTER_MODE.
7449 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
7450             Pat Haugen  <pthaugen@us.ibm.com>
7451             Peter Bergner  <bergner@vnet.ibm.com>
7453         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7454         Document new power8 builtins.
7456         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7457         condition code register, to allow 128-bit logical operations to be
7458         done in the VSX or GPR registers.
7459         (nor<mode>3): Use the canonical form for nor.
7460         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7461         vclz*, and vpopcnt* vector instructions.
7462         (nand<mode>3): Likewise.
7463         (orc<mode>3): Likewise.
7464         (clz<mode>2): LIkewise.
7465         (popcount<mode>2): Likewise.
7467         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7468         that only the GPRs are recognized.
7470         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7471         support for new power8 builtins.
7473         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7474         builtin functions.
7475         (xscvdpspn): Likewise.
7476         (vclz): Likewise.
7477         (vclzb): Likewise.
7478         (vclzh): Likewise.
7479         (vclzw): Likewise.
7480         (vclzd): Likewise.
7481         (vpopcnt): Likewise.
7482         (vpopcntb): Likewise.
7483         (vpopcnth): Likewise.
7484         (vpopcntw): Likewise.
7485         (vpopcntd): Likewise.
7486         (vgbbd): Likewise.
7487         (vmrgew): Likewise.
7488         (vmrgow): Likewise.
7489         (eqv): Likewise.
7490         (eqv_v16qi3): Likewise.
7491         (eqv_v8hi3): Likewise.
7492         (eqv_v4si3): Likewise.
7493         (eqv_v2di3): Likewise.
7494         (eqv_v4sf3): Likewise.
7495         (eqv_v2df3): Likewise.
7496         (nand): Likewise.
7497         (nand_v16qi3): Likewise.
7498         (nand_v8hi3): Likewise.
7499         (nand_v4si3): Likewise.
7500         (nand_v2di3): Likewise.
7501         (nand_v4sf3): Likewise.
7502         (nand_v2df3): Likewise.
7503         (orc): Likewise.
7504         (orc_v16qi3): Likewise.
7505         (orc_v8hi3): Likewise.
7506         (orc_v4si3): Likewise.
7507         (orc_v2di3): Likewise.
7508         (orc_v4sf3): Likewise.
7509         (orc_v2df3): Likewise.
7511         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7512         allow power8 quad mode in 64-bit.
7513         (rs6000_builtin_vectorized_function): Add support to vectorize
7514         ISA 2.07 count leading zeros, population count builtins.
7515         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7516         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7517         (builtin_function_type): Add vgbbd builtin function which takes an
7518         unsigned argument.
7519         (altivec_expand_vec_perm_const): Add support for new power8 merge
7520         instructions.
7522         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7523         that does not include TImdoe for use with 32-bit.
7524         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7525         instructions.
7526         (UNSPEC_VSX_CVDPSPN): Likewise.
7527         (vsx_xscvdpspn): Likewise.
7528         (vsx_xscvspdpn): Likewise.
7529         (vsx_xscvdpspn_scalar): Likewise.
7530         (vsx_xscvspdpn_directmove): Likewise.
7531         (vsx_and<mode>3): Split logical operations into 32-bit and
7532         64-bit. Add support to do logical operations on TImode as well as
7533         VSX vector types.  Allow logical operations to be done in either
7534         VSX registers or in general purpose registers in 64-bit mode.  Add
7535         splitters if GPRs were used. For AND, add clobber of CCmode to
7536         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
7537         encoding.
7538         (vsx_and<mode>3_32bit): Likewise.
7539         (vsx_and<mode>3_64bit): Likewise.
7540         (vsx_ior<mode>3): Likewise.
7541         (vsx_ior<mode>3_32bit): Likewise.
7542         (vsx_ior<mode>3_64bit): Likewise.
7543         (vsx_xor<mode>3): Likewise.
7544         (vsx_xor<mode>3_32bit): Likewise.
7545         (vsx_xor<mode>3_64bit): Likewise.
7546         (vsx_one_cmpl<mode>2): Likewise.
7547         (vsx_one_cmpl<mode>2_32bit): Likewise.
7548         (vsx_one_cmpl<mode>2_64bit): Likewise.
7549         (vsx_nor<mode>3): Likewise.
7550         (vsx_nor<mode>3_32bit): Likewise.
7551         (vsx_nor<mode>3_64bit): Likewise.
7552         (vsx_andc<mode>3): Likewise.
7553         (vsx_andc<mode>3_32bit): Likewise.
7554         (vsx_andc<mode>3_64bit): Likewise.
7555         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7556         and xxlorc instructions.
7557         (vsx_eqv<mode>3_64bit): Likewise.
7558         (vsx_nand<mode>3_32bit): Likewise.
7559         (vsx_nand<mode>3_64bit): Likewise.
7560         (vsx_orc<mode>3_32bit): Likewise.
7561         (vsx_orc<mode>3_64bit): Likewise.
7563         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7565         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7566         instruction.
7567         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7568         (p8_vmrgow): Likewise.
7569         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7570         GPRs to be split under VSX.
7571         (p8v_clz<mode>2): Add power8 count leading zero support.
7572         (p8v_popcount<mode>2): Add power8 population count support.
7573         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7574         support.
7576         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7577         instruction.
7579         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7580         builtin functions.
7581         (vec_nand): Likewise.
7582         (vec_vclz): Likewise.
7583         (vec_vclzb): Likewise.
7584         (vec_vclzd): Likewise.
7585         (vec_vclzh): Likewise.
7586         (vec_vclzw): Likewise.
7587         (vec_vgbbd): Likewise.
7588         (vec_vmrgew): Likewise.
7589         (vec_vmrgow): Likewise.
7590         (vec_vpopcnt): Likewise.
7591         (vec_vpopcntb): Likewise.
7592         (vec_vpopcntd): Likewise.
7593         (vec_vpopcnth): Likewise.
7594         (vec_vpopcntw): Likewise.
7596 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7598         PR rtl-optimization/57468
7599         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7600         spilled pseudos.
7602 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7604         PR rtl-optimization/57459
7605         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7606         type when setting live regs.
7608 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7610         * config/s390/s390.opt (mlra): New option.
7611         * config/s390/s390.c (s390_decompose_address): Check displacement
7612         for all registers for LRA.
7613         (s390_secondary_reload): Don't used secondary reloads for LRA.
7614         (s390_lra_p): New function.
7615         (TARGET_LRA_P): Define.
7616         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7617         of attribute cpu_facility to zarch for the last alternative.
7618         (*cmpmem_short): Ditto.
7620 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
7622         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7623         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
7624         (arm_expand_prologue): Likewise.
7626 2013-06-06  Teresa Johnson  <tejohnson@google.com>
7628         PR c++/53743
7629         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7630         as this is now done by redirect_edge_and_branch_force.
7631         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7632         barriers, and fix interaction with splitting.
7633         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7634         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7635         reflect changes made in the routine.
7636         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7637         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7638         since this is called in cfglayout mode, and replace partition fixup
7639         with assert as that is now done by force_nonfallthru_and_redirect.
7640         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7641         already be marked with region crossing note.
7642         (insert_section_boundary_note): Make non-static, gate on flag
7643         has_bb_partition, rewrite to also check for multiple partitions.
7644         (rest_of_handle_reorder_blocks): Remove call to
7645         insert_section_boundary_note, now done later during free_cfg.
7646         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7647         * bb-reorder.h (insert_section_boundary_note): Declare.
7648         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7649         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7650         invoke insert_section_boundary_note.
7651         (try_redirect_by_replacing_jump): Remove unnecessary
7652         check for region crossing note.
7653         (fixup_partition_crossing): New function.
7654         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7655         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7656         in non-cfglayout mode.
7657         (force_nonfallthru_and_redirect): Fixup partition boundaries,
7658         remove old code that tried to do this. Emit barrier correctly
7659         when we are in cfglayout mode.
7660         (last_bb_in_partition): New function.
7661         (rtl_split_edge): Correctly fixup partition boundaries.
7662         (commit_one_edge_insertion): Remove old code that tried to
7663         fixup region crossing edge since this is now handled in
7664         split_block, and set up insertion point correctly since
7665         block may now end in a jump.
7666         (verify_hot_cold_block_grouping): Guard against checking when not in
7667         linearized RTL mode.
7668         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7669         notes.
7670         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7671         rtl_verify_flow_info, so not called in cfglayout mode.
7672         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7673         (fixup_reorder_chain): Remove old code that attempted to fixup region
7674         crossing note as this is now handled in force_nonfallthru_and_redirect.
7675         (duplicate_insn_chain): Don't duplicate switch section notes.
7676         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7677         note.
7678         * basic-block.h (emit_barrier_after_bb): Declare.
7680 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7682         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7683         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7684         arm_usatsihi): Adjust alternatives for arm_restrict_it.
7686 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7688         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7689         where appropriate.
7690         * config/arm/ldmstm.md: Regenerate.
7692 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7694         * config/arm/sync.md (atomic_loaddi_1):
7695         Disable predication for arm_restrict_it.
7696         (arm_load_exclusive<mode>): Likewise.
7697         (arm_load_exclusivesi): Likewise.
7698         (arm_load_exclusivedi): Likewise.
7699         (arm_load_acquire_exclusive<mode>): Likewise.
7700         (arm_load_acquire_exclusivesi): Likewise.
7701         (arm_load_acquire_exclusivedi): Likewise.
7702         (arm_store_exclusive<mode>): Likewise.
7703         (arm_store_exclusive<mode>): Likewise.
7704         (arm_store_release_exclusivedi): Likewise.
7705         (arm_store_release_exclusive<mode>): Likewise.
7707 2013-06-06  Richard Biener  <rguenther@suse.de>
7709         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7710         after LTO_null.
7711         (lto_tag_is_tree_code_p): Adjust.
7712         (lto_tag_is_gimple_code_p): Likewise.
7713         (lto_gimple_code_to_tag): Likewise.
7714         (lto_tag_to_gimple_code): Likewise.
7715         (lto_tree_code_to_tag): Likewise.
7716         (lto_tag_to_tree_code): Likewise.
7717         * data-streamer.h (streamer_write_hwi_in_range): Use
7718         uhwi streaming to stream the normalized range.
7719         (streamer_read_hwi_in_range): Likewise.
7721 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7723         * config/arm/arm.md (enabled_for_depr_it): New attribute.
7724         (predicable_short_it): Likewise.
7725         (predicated): Likewise.
7726         (enabled): Handle above.
7727         (define_cond_exec): Set predicated attribute to yes.
7729 2013-06-05  Mike Stump  <mikestump@comcast.net>
7731         * gdbinit.in (__FUNCTION__): Add.
7733 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
7735         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7736         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7738 2013-06-05  Jan Hubicka  <jh@suse.cz>
7740         * varasm.c (mark_decl_referenced): Revert the removal until targets
7741         are fixed.
7743 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
7745         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7746         instead of mark_decl_referenced.
7748 2013-06-05  Jan Hubicka  <jh@suse.cz>
7750         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7751         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7752         and symtab_used_from_object_file_p.
7753         (cgraph_make_node_local_1): Clear forced_by_abi.
7754         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7755         * cgraph.h (symtab_node_base): Add forced_by_abi.
7756         (decide_is_variable_needed): Remove.
7757         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7758         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7759         (decide_is_symbol_needed): ... this one; handle symbols in general;
7760         always analyze virtuals; honnor forced_by_abi.
7761         (cgraph_finalize_function): Update.
7762         (varpool_finalize_decl): Update.
7763         (symbol_defined_and_needed): Remove.
7764         (analyze_functions): Update.
7765         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7766         output_refs, input_overwrite_node): Handle forced_by_abi.
7767         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7768         (address_taken_from_non_vtable_p): ... this one.
7769         (comdat_can_be_unshared_p_1): New function.
7770         (cgraph_comdat_can_be_unshared_p): Rename to ...
7771         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7772         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7773         (function_and_variable_visibility): Clear forced_by_abi as needed.
7774         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7775         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7776         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7777         * varpool.c (decide_is_variable_needed): Remove.
7779 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7781         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7782         (arm_option_override): Override arm_restrict_it where appropriate.
7783         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7784         * config/arm/arm.opt (mrestrict-it): New command-line option.
7785         * doc/invoke.texi: Document -mrestrict-it.
7787 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7789         * tsan.c (tsan_atomic_table): Make const.
7791 2013-06-05  Richard Biener  <rguenther@suse.de>
7793         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7794         index associated with the tree we are supposed to replace.
7795         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7796         TREE_ASM_WRITTEN as zero for everything but SSA names.
7798 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7800         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7802 2013-06-04  Jan Hubicka  <jh@suse.cz>
7804         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7805         (input_node, input_varpool_node): Handle correctly external same
7806         body aliases.
7807         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7808         nodes at ltrans stage.
7810 2013-06-04  Jan Hubicka  <jh@suse.cz>
7812         * ipa-inline.c (update_caller_keys): Fix availability test.
7813         (update_callee_keys): Likewise.
7814         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7815         to follow ELF standard.
7817 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7819         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7820         (mips64r5900el-*-elf*): New configurations.
7821         * config/mips/mips-cpus.def (r5900): New processor.
7822         * config/mips/mips-tables.opt: Regenerate.
7823         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7824         (mips_issue_rate): Handle PROCESSOR_R5900.
7825         (mips_reorg_process_insns): Force reorder mode for the R5900.
7826         * config/mips/mips.h (TARGET_MIPS5900): Define.
7827         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7828         TARGET_MIPS5900.
7829         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7830         TARGET_MIPS5900.
7831         * config/mips/mips.md (processor): Add r5900.
7832         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7834 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7836         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7837         into function to generate MOVI instruction.
7838         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7839         (aarch64_preferred_simd_mode): Turn into wrapper.
7840         (aarch64_output_scalar_simd_mov_immediate): New function.
7841         * config/aarch64/aarch64-protos.h: Add prototype for above.
7843 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7845         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7846         element_char member.
7847         (sizetochar): Return signed char.
7848         (aarch64_simd_valid_immediate): Remove elchar and other
7849         unnecessary variables.
7850         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7851         Calculate element_char as required.
7852         * config/aarch64/aarch64-protos.h: Update and move prototype
7853         for aarch64_output_simd_mov_immediate.
7854         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7855         Update arguments.
7857 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7859         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7860         information completed by aarch64_simd_valid_immediate.
7861         (aarch64_legitimate_constant_p): Update arguments.
7862         (aarch64_simd_valid_immediate): Work with struct rather than many
7863         pointers.
7864         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7865         (aarch64_simd_make_constant): Update arguments.
7866         (aarch64_output_simd_mov_immediate): Work with struct rather than
7867         many pointers.  Output immediate directly rather than as operand.
7868         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7869         Update prototype.
7870         * config/aarch64/constraints.md (Dn): Update arguments.
7872 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7874         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7875         longer static.
7876         (aarch64_simd_immediate_valid_for_move): Remove.
7877         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7878         (aarch64_simd_make_constant): Update call.
7879         (aarch64_output_simd_mov_immediate): Update call.
7880         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7881         Add prototype.
7882         * config/aarch64/constraints.md (Dn): Update call.
7884 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7886         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7887         return type to bool for prototype.
7888         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7889         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7890         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7892 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7894         * config/mips/mips.opt (meva): New.
7895         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7896         (ASM_SPEC): Handle -meva.
7897         * doc/invoke.texi (meva):  Document.
7899 2013-06-04  Alan Modra  <amodra@gmail.com>
7901         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7902         constant output.
7904 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7906         * rtl.def: Add extra fourth optional field to define_cond_exec.
7907         * gensupport.c (process_one_cond_exec): Process attributes from
7908         define_cond_exec.
7909         * doc/md.texi: Document fourth field in define_cond_exec.
7911 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7913         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7914         out the processing order as in store_bit_field_1.
7916 2013-06-04  Jan Hubicka  <jh@suse.cz>
7918         PR middle-end/57500
7919         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7920         non-VAR_DECL node if it does not exist yet.
7922 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7924         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7925         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7926         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7927         target_cpu_default setting.
7929 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7931         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7932         default to -fopt-info=optimized instead of all.
7933         * doc/invoke.texi: Ditto.
7934         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7935         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7936         (execute_vect_slp): Emit BB vectorization success under
7937         MSG_OPTIMIZED_LOCATIONS.
7938         * tree-vect-slp.c (vect_slp_transform_bb): Change
7939         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7940         * tree-vect-loop.c (vect_transform_loop): Ditto.
7942 2013-06-03  Jason Merrill  <jason@redhat.com>
7944         PR c++/57415
7945         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7946         Use TARGET_EXPR for C++.
7948 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7950         PR rtl-optimization/57268
7951         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7952         if DEBUG_INSN_P (insn).
7954         Reapply
7955         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7957         PR rtl-optimization/57268
7958         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7959         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7961 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7963         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7964         (ix86_avoid_lea_for_addr): Likewise.
7965         (exact_dependency_1): Likewise.
7966         (ix86_adjust_cost): Likewise.
7967         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7968         removed.
7969         (do_reorder_for_imul): Fix typo, formatting and
7970         !reload_completed check removed.
7971         (ix86_sched_reorder): Fix typo and formatting.
7972         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7973         list.
7975 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7977         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7979 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7981         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7982         <STRING_CST>: Likewise.
7983         <VECTOR_CST>: Likewise.
7985 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7986             Mikael Morin  <mikael@gcc.gnu.org>
7988         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7989         * config.in: Regenerated.
7990         * configure: Regenerated.
7992 2013-06-01  Jan Hubicka  <jh@suse.cz>
7994         PR middle-end/57366
7995         * cgraphunit.c (compile): When weakref is not supported,
7996         set up transparent aliases before final output pass.
7997         * varasm.c (assemble_alias): Do not try to do it here.
7999 2013-06-01  Jan Hubicka  <jh@suse.cz>
8001         PR middle-end/57467
8002         * passes.c (for_per_function): Skip unanalyzed functions.
8004 2013-06-01  Jan Hubicka  <jh@suse.cz>
8006         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
8007         (lto_symtab_merge_symbols_1): ... this one.
8008         (lto_symtab_merge_cgraph_nodes): Rename to ...
8009         (lto_symtab_merge_symbols): ... this one; simplify.
8010         * cgraph.c (same_body_aliases_done): Rename to ...
8011         (cpp_implicit_aliases_done): ... this one.
8012         (cgraph_create_function_alias): Update.
8013         (cgraph_same_body_alias): Update.
8014         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8015         (verify_edge_corresponds_to_fndecl): Simplify.
8016         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8017         (cgraph_node): Remove same_body_alias.
8018         (varpool_node): Remove alias_of and extra_name_alias.
8019         (same_body_aliases_done): Rename to ..
8020         (cpp_implicit_aliases_done): ... this one.
8021         (symtab_alias_ultimate_target): Add default parameter.
8022         (symtab_resolve_alias): New function.
8023         (fixup_same_cpp_alias_visibility): Declare.
8024         (cgraph_function_node): Add default parameter.
8025         (cgraph_node_asm_name): Likewise.
8026         (cgraph_function_or_thunk_node): Add default parameter; do
8027         not ICE when it is NULL.
8028         (varpool_variable_node): Likewise.
8029         * tree-emutls.c (create_emultls_var): Update.
8030         (ipa_lower_emutls): Update.
8031         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8032         (cgraph_reset_node): Reset alias info.
8033         (cgraph_finalize_function): Update.
8034         (fixup_same_cpp_alias_visibility): Move to symtab.c.
8035         (analyze_function): Simplify.
8036         (cgraph_process_same_body_aliases): Simplify.
8037         (analyze_functions): Fixup same body aliases.
8038         (handle_alias_pairs): Simplify.
8039         (assemble_thunk): Update.
8040         (assemble_thunks_and_aliases): Update.
8041         (output_weakrefs): Rewrite.
8042         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8043         (lto_output_varpool_node): Likewise.
8044         (compute_ltrans_boundary): Remve assert.
8045         (get_alias_symbol): New functoin.
8046         (input_node): Rewrite alias handling.
8047         (input_varpool_node): Likewise.
8048         * ipa-pure-const.c (propagate_pure_const): Fix formating.
8049         * ipa.c (process_references): Handle weakrefs correctly.
8050         (symtab_remove_unreachable_nodes): Likewise.
8051         * trans-mem.c (get_cg_data): Update.
8052         (ipa_tm_create_version_alias): Update.
8053         (ipa_tm_execute): Update.
8054         * symtab.c (dump_symtab_base): Dump aliases.
8055         (verify_symtab_base): Verify aliases.
8056         (symtab_node_availability): New function.
8057         (symtab_alias_ultimate_target): Simplify.
8058         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8059         handle all the fixup cases.
8060         (symtab_resolve_alias): New function.
8061         * passes.c (ipa_write_summaries): Handle weakrefs.
8062         * varpool.c (varpool_analyze_node): Simplify.
8063         (assemble_aliases): Update.
8064         (varpool_create_variable_alias): Simplify.
8065         (varpool_extra_name_alias): Simplify.
8066         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8067         (lto_symtab_merge_symbols): ... this one.
8069 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
8071         Revert
8072         PR rtl-optimization/57268
8073         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8074         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8076 2013-06-01  Tobias Burnus  <burnus@net-b.de>
8078         Partially reverted:
8079         2013-05-31  Tobias Burnus  <burnus@net-b.de>
8081         PR middle-end/57073
8082         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8083         further up.
8085 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8087         PR rtl-optimization/57268
8088         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8089         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8091 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8093         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8094         unordered comparison operators when -fno-trapping-math is in effect
8095         on the e500.
8096         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8097         and implement unordered comparison operators properly on the e500.
8099 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8101         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8102         for constant scalar integers.
8103         (simplify_relational_operation_1): Likewise.
8105 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
8107         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8108         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
8109         Fix comment.
8111 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8112             Igor Zamyatin  <igor.zamyatin@intel.com>
8114         Silvermont (SLM) architecture performance tuning.
8115         * config/i386/i386.h (enum ix86_tune_indices): Add
8116         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8117         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8119         * config/i386/i386.c (initial_ix86_tune_features)
8120         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8121         (ix86_lea_outperforms): Handle Silvermont tuning.
8122         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8123         call.
8124         (ix86_use_lea_for_mov): Likewise.
8125         (ix86_avoid_lea_for_addr): Likewise.
8126         (ix86_lea_for_add_ok): Likewise.
8127         (exact_dependency_1): New function.
8128         (exact_store_load_dependency): Likewise.
8129         (ix86_adjust_cost): Handle Silvermont tuning.
8130         (do_reoder_for_imul): Likewise.
8131         (swap_top_of_ready_list): New function.
8132         (ix86_sched_reorder): Changed to handle Silvermont tuning.
8134         * config/i386/i386.md (peepholes that split memory operand in fp
8135         converts): New.
8137 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8139         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8140         Remove un-necessary braces.
8142 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8144         * config/aarch64/aarch64.c (aarch64_classify_symbol):
8145         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8147 2013-05-31  Tobias Burnus  <burnus@net-b.de>
8149         PR middle-end/57073
8150         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8152 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8154         PR target/56315
8155         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8156         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8157         * config/arm/neon.md (iordi3_neon): Remove.
8158         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8159         * config/arm/predicates.md (imm_for_neon_logic_operand):
8160         Move to earlier in the file.
8161         (neon_logic_op2): Likewise.
8162         (arm_iordi_operand_neon): New predicate.
8164 2013-05-31  Richard Biener  <rguenther@suse.de>
8166         PR tree-optimization/57478
8167         PR tree-optimization/57453
8168         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8169         are life as well.
8171 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8173         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8174         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8176 2013-05-30  Tobias Burnus  <burnus@net-b.de>
8177             Thomas Koenig  <tkoenig@gcc.gnu.org>
8179         PR middle-end/57073
8180         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8181         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8183 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
8185         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8187 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
8189         * target.def (register_usage_leveling_p): New hook.
8190         * targhooks.c (default_register_usage_leveling_p): New.
8191         * targhooks.h (default_register_usage_leveling_p): New prototype.
8192         * lra-assigns.c (register_usage_leveling_p): Use the hook.
8193         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8194         * doc/tm.texi: Update.
8195         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8197 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
8199         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8200         (*insv_reg<mode>): New define_insn.
8202 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8204         PR rtl-optimization/57439
8205         * postreload.c (move2add_valid_value_p): Check that we have
8206         a zero subreg_regno_offset when accessing the register in
8207         the requested mode.
8209 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8210             Igor Zamyatin  <igor.zamyatin@intel.com>
8212         Silvermont (SLM) architecture pipeline model, tuning and
8213         insn selection.
8214         * config.gcc: Add slm config options and target.
8216         * config/i386/slm.md: New.
8218         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8220         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
8221         PROCESSOR_SLM.
8222         (ix86_target_macros_internal): Likewise.
8224         * gcc/config/i386/i386.c (slm_cost): New cost.
8225         (m_SLM): New macro flag.
8226         (initial_ix86_tune_features): Set m_SLM.
8227         (x86_accumulate_outgoing_args): Likewise.
8228         (x86_arch_always_fancy_math_387): Likewise.
8229         (processor_target_table): Add slm cost.
8230         (cpu_names): Add slm cpu name.
8231         (x86_option_override_internal): Set SLM ISA.
8232         (ix86_issue_rate): New case PROCESSOR_SLM.
8233         (ia32_multipass_dfa_lookahead): Likewise.
8234         (fold_builtin_cpu): Add slm.
8236         * config/i386/i386.h (TARGET_SLM): New target macro.
8237         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8238         (processor_type): Add PROCESSOR_SLM.
8240         * config/i386/i386.md (cpu): Add new value "slm".
8241         (slm.md): Include slm.md.
8243 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
8244             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8246         * config/arm/arm-protos.h: Add and update function protos.
8247         * config/arm/arm.c (use_simple_return_p): New added.
8248         (thumb2_expand_return): Check simple_return flag.
8249         * config/arm/arm.md: Add simple_return and conditional simple_return.
8250         * config/arm/iterators.md: Add iterator for return and simple_return.
8252 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8254         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8255         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8256         (arm_emit_vfp_multi_reg_pop): Likewise.
8257         (thumb2_emit_ldrd_pop): Likewise.
8258         (arm_expand_epilogue): Add misc REG_CFA notes.
8259         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8261 2013-05-29  Lawrence Crowl  <crowl@google.com>
8263         * config/arm/t-arm: Update for below.
8265         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8266         Change type to hash_table.  Update dependent calls and types.
8268         * config/i386/t-cygming: Update for below.
8270         * config/i386/t-interix: Update for below.
8272         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8273         Change type to hash_table.  Update dependent calls and types.
8274         (i386_find_on_wrapper_list::wrappers): Likewise.
8276         * config/ia64/t-ia64: Update for below.
8278         * config/ia64/ia64.c (bundle_state_table):
8279         Change type to hash_table.  Update dependent calls and types.
8281         * config/mips/mips.c (mips_reorg_process_insns::htab):
8282         Change type to hash_table.  Update dependent calls and types.
8284         * config/sol2.c (solaris_comdat_htab):
8285         Change type to hash_table.  Update dependent calls and types.
8287         * config/t-sol2: Update for above.
8289 2013-05-29  Teresa Johnson  <tejohnson@google.com>
8291         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8292         functions are not yet marked as defined.
8294 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
8295             Pat Haugen  <pthaugen@us.ibm.com>
8296             Peter Bergner  <bergner@vnet.ibm.com>
8298         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8299         instructions.
8300         (VEC_A): Likewise.
8301         (VEC_C): Likewise.
8302         (vrotl<mode>3): Likewise.
8303         (vashl<mode>3): Likewise.
8304         (vlshr<mode>3): Likewise.
8305         (vashr<mode>3): Likewise.
8307         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8308         support for power8 V2DI builtins.
8310         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8311         power8 V2DI builtins.
8312         (vupkhsw): Likewise.
8313         (vupklsw): Likewise.
8314         (vaddudm): Likewise.
8315         (vminsd): Likewise.
8316         (vmaxsd): Likewise.
8317         (vminud): Likewise.
8318         (vmaxud): Likewise.
8319         (vpkudum): Likewise.
8320         (vpksdss): Likewise.
8321         (vpkudus): Likewise.
8322         (vpksdus): Likewise.
8323         (vrld): Likewise.
8324         (vsld): Likewise.
8325         (vsrd): Likewise.
8326         (vsrad): Likewise.
8327         (vsubudm): Likewise.
8328         (vcmpequd): Likewise.
8329         (vcmpgtsd): Likewise.
8330         (vcmpgtud): Likewise.
8331         (vcmpequd_p): Likewise.
8332         (vcmpgtsd_p): Likewise.
8333         (vcmpgtud_p): Likewise.
8334         (vupkhsw): Likewise.
8335         (vupklsw): Likewise.
8336         (vaddudm): Likewise.
8337         (vmaxsd): Likewise.
8338         (vmaxud): Likewise.
8339         (vminsd): Likewise.
8340         (vminud): Likewise.
8341         (vpksdss): Likewise.
8342         (vpksdus): Likewise.
8343         (vpkudum): Likewise.
8344         (vpkudus): Likewise.
8345         (vrld): Likewise.
8346         (vsld): Likewise.
8347         (vsrad): Likewise.
8348         (vsrd): Likewise.
8349         (vsubudm): Likewise.
8351         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8352         support for power8 V2DI instructions.
8354         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8355         power8 V2DI instructions.  Combine pack and unpack insns to use an
8356         iterator for each mode.  Check whether a particular mode supports
8357         Altivec instructions instead of just checking TARGET_ALTIVEC.
8358         (UNSPEC_VPKUWUM): Likewise.
8359         (UNSPEC_VPKSHSS): Likewise.
8360         (UNSPEC_VPKSWSS): Likewise.
8361         (UNSPEC_VPKUHUS): Likewise.
8362         (UNSPEC_VPKSHUS): Likewise.
8363         (UNSPEC_VPKUWUS): Likewise.
8364         (UNSPEC_VPKSWUS): Likewise.
8365         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8366         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8367         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8368         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8369         (UNSPEC_VUPKHSB): Likewise.
8370         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8371         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8372         (UNSPEC_VUPKHSH): Likewise.
8373         (UNSPEC_VUPKLSB): Likewise.
8374         (UNSPEC_VUPKLSH): Likewise.
8375         (VI2): Likewise.
8376         (VI_char): Likewise.
8377         (VI_scalar): Likewise.
8378         (VI_unit): Likewise.
8379         (VP): Likewise.
8380         (VP_small): Likewise.
8381         (VP_small_lc): Likewise.
8382         (VU_char): Likewise.
8383         (add<mode>3): Likewise.
8384         (altivec_vaddcuw): Likewise.
8385         (altivec_vaddu<VI_char>s): Likewise.
8386         (altivec_vadds<VI_char>s): Likewise.
8387         (sub<mode>3): Likewise.
8388         (altivec_vsubcuw): Likewise.
8389         (altivec_vsubu<VI_char>s): Likewise.
8390         (altivec_vsubs<VI_char>s): Likewise.
8391         (altivec_vavgs<VI_char>): Likewise.
8392         (altivec_vcmpbfp): Likewise.
8393         (altivec_eq<mode>): Likewise.
8394         (altivec_gt<mode>): Likewise.
8395         (altivec_gtu<mode>): Likewise.
8396         (umax<mode>3): Likewise.
8397         (smax<mode>3): Likewise.
8398         (umin<mode>3): Likewise.
8399         (smin<mode>3): Likewise.
8400         (altivec_vpkuhum): Likewise.
8401         (altivec_vpkuwum): Likewise.
8402         (altivec_vpkshss): Likewise.
8403         (altivec_vpkswss): Likewise.
8404         (altivec_vpkuhus): Likewise.
8405         (altivec_vpkshus): Likewise.
8406         (altivec_vpkuwus): Likewise.
8407         (altivec_vpkswus): Likewise.
8408         (altivec_vpks<VI_char>ss): Likewise.
8409         (altivec_vpks<VI_char>us): Likewise.
8410         (altivec_vpku<VI_char>us): Likewise.
8411         (altivec_vpku<VI_char>um): Likewise.
8412         (altivec_vrl<VI_char>): Likewise.
8413         (altivec_vsl<VI_char>): Likewise.
8414         (altivec_vsr<VI_char>): Likewise.
8415         (altivec_vsra<VI_char>): Likewise.
8416         (altivec_vsldoi_<mode>): Likewise.
8417         (altivec_vupkhsb): Likewise.
8418         (altivec_vupkhs<VU_char>): Likewise.
8419         (altivec_vupkls<VU_char>): Likewise.
8420         (altivec_vupkhsh): Likewise.
8421         (altivec_vupklsb): Likewise.
8422         (altivec_vupklsh): Likewise.
8423         (altivec_vcmpequ<VI_char>_p): Likewise.
8424         (altivec_vcmpgts<VI_char>_p): Likewise.
8425         (altivec_vcmpgtu<VI_char>_p): Likewise.
8426         (abs<mode>2): Likewise.
8427         (vec_unpacks_hi_v16qi): Likewise.
8428         (vec_unpacks_hi_v8hi): Likewise.
8429         (vec_unpacks_lo_v16qi): Likewise.
8430         (vec_unpacks_hi_<VP_small_lc>): Likewise.
8431         (vec_unpacks_lo_v8hi): Likewise.
8432         (vec_unpacks_lo_<VP_small_lc>): Likewise.
8433         (vec_pack_trunc_v8h): Likewise.
8434         (vec_pack_trunc_v4si): Likewise.
8435         (vec_pack_trunc_<mode>): Likewise.
8437         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8438         V2DI builtins.
8439         (vec_vmaxsd): Likewise.
8440         (vec_vmaxud): Likewise.
8441         (vec_vminsd): Likewise.
8442         (vec_vminud): Likewise.
8443         (vec_vpksdss): Likewise.
8444         (vec_vpksdus): Likewise.
8445         (vec_vpkudum): Likewise.
8446         (vec_vpkudus): Likewise.
8447         (vec_vrld): Likewise.
8448         (vec_vsld): Likewise.
8449         (vec_vsrad): Likewise.
8450         (vec_vsrd): Likewise.
8451         (vec_vsubudm): Likewise.
8452         (vec_vupkhsw): Likewise.
8453         (vec_vupklsw): Likewise.
8455 2013-05-29  Jan Hubicka  <jh@suse.cz>
8457         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8458         flags; reorder rest of fields in more consistent way.
8459         (varpool_node): Remove analyzed, finalized and alias.
8460         (cgraph_ndoe): Likewise.
8461         (symtab_alias_ultimate_target): New function.
8462         (cgraph_function_node): Move offline.
8463         (cgraph_reset_node): Declare.
8464         (cgraph_comdat_can_be_unshared_p): Remove.
8465         (varpool_remove_initializer): Declare.
8466         (varpool_first_defined_variable, varpool_next_defined_variable
8467         cgraph_first_defined_function, cgraph_next_defined_function): Update.
8468         (cgraph_function_with_gimple_body_p): Update.
8469         (varpool_all_refs_explicit_p): Update.
8470         (symtab_alias_target): New function.
8471         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8472         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8473         (cgraph_function_or_thunk_node): Simplify using
8474         symtab_alias_ultimate_target.
8475         (varpool_variable_node): Likewise.
8476         * cgraph.c (cgraph_create_function_alias): Update.
8477         (cgraph_add_thunk): Update.
8478         (cgraph_remove_node): Update.
8479         (dump_cgraph_node): Do not dump removed flags.
8480         (cgraph_function_body_availability): Update.
8481         (cgraph_propagate_frequency): Update.
8482         (verify_cgraph_node): Check sanity of local flag.
8483         (cgraph_function_node): Move here from cgraph.h; revamp for
8484         cgraph_function_or_thunk_node.
8485         * lto-symtab.c (lto_varpool_replace_node): Update.
8486         (lto_symtab_resolve_can_prevail_p): Update.
8487         (lto_symtab_merge_cgraph_nodes): Update.
8488         * ipa-cp.c (determine_versionability, initialize_node_lattices,
8489         propagate_constants_accross_call, devirtualization_time_bonus,
8490         ipcp_propagate_stage): Update.
8491         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8492         * ipa-inline-transform.c (clone_inlined_nodes,
8493         preserve_function_body_p): Update.
8494         * ipa-reference.c (propagate): Update.
8495         (write_node_summary_p): Update.
8496         * toplev.c (wrapup_global_declaration_2): Update.
8497         * cgraphunit.c (cgraph_analyze_function): Rename to ...
8498         (analyze_function) ... this one.
8499         (cgraph_process_new_functions): Update.
8500         (cgraph_reset_node): Export.
8501         (cgraph_finalize_function): Update.
8502         (cgraph_add_new_function): Update.
8503         (process_function_and_variable_attributes): Update.
8504         (varpool_finalize_decl): Update.
8505         (symbol_finalized): Remove.
8506         (symbol_finalized_and_needed): Rename to ...
8507         (symbol_defined_and_needed): ... update.
8508         (cgraph_analyze_functions): Update.
8509         (handle_alias_pairs): Update.
8510         (mark_functions_to_output): Update.
8511         (assemble_thunk): Update.
8512         (output_in_order): Update.
8513         (output_weakrefs): Update.
8514         (finalize_compilation_unit): Update.
8515         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8516         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8517         input_node, input_varpool_node): Update.
8518         * dbxout.c (dbxout_expand_expr): Update.
8519         * cgraphclones.c (cgraph_clone_node): Update.
8520         (cgraph_copy_node_for_versioning): Update.
8521         (cgraph_materialize_clone): Update.
8522         (cgraph_materialize_all_clones): Update.
8523         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8524         propagate_pure_const, propagate_nothrow): Update.
8525         * lto-streamer-out.c (lto_output, write_symbol): Update.
8526         * ipa-utils.c (ipa_reverse_postorder): Update.
8527         * ipa-inline.c (can_inline_edge_p): Update.
8528         (update_caller_keys, ipa_inline): Update.
8529         * dwarf2out.c (reference_to_unused,
8530         premark_types_used_by_global_vars_helper): Update.
8531         * tree-eh.c (tree_could_trap_p): Update.
8532         * ipa-split.c (consider_split, execute_split_functions): Update.
8533         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8534         has_addr_references_p): Update; move ahead in file for better
8535         readability.
8536         (process_references): Simplify.
8537         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8538         bodies are removed.
8539         (cgraph_comdat_can_be_unshared_p): Make static.
8540         (cgraph_externally_visible_p): Update.
8541         (varpool_externally_visible_p): Update.
8542         (function_and_variable_visibility): Update.
8543         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8544         ipa_tm_mark_force_output_node): Update.
8545         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8546         estimate_edge_devirt_benefit, inline_generate_summary,
8547         inline_write_summary): Update.
8548         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8549         * ipa-prop.c (ipa_compute_jump_functions): Update.
8550         (ipa_print_node_params, ipa_prop_read_section,
8551         ipa_update_after_lto_read, read_replacements_section): Update.
8552         * varasm.c (mark_decl_referenced): Update.
8553         (assemble_alias, dump_tm_clone_pairs): Update.
8554         * tree-inline.c (copy_bb): Update.
8555         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8556         Update.
8557         * symtab.c (dump_symtab_base): Print new flags.
8558         (verify_symtab_base): Verify new flags.
8559         (symtab_alias_ultimate_target): New function.
8560         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8561         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8562         Update.
8563         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8564         Update.
8565         * i386.c (ix86_get_function_versions_dispatcher,
8566         ix86_generate_version_dispatcher_body): Update.
8567         (fold_builtin_cpu): Use varpool_add_new_variable.
8568         * varpool.c (varpool_remove_initializer): Break out from ...
8569         (varpool_remove_node): ... this one.
8570         (dump_varpool_node, varpool_node_for_asm,
8571         cgraph_variable_initializer_availability, varpool_analyze_node,
8572         varpool_assemble_decl, varpool_remove_unreferenced_decls,
8573         varpool_finalize_named_section_flags, varpool_create_variable_alias):
8574         Update.
8576 2013-05-29  Jan Hubicka  <jh@suse.cz>
8578         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8580 2013-05-29  Easwaran Raman  <eraman@google.com>
8582         PR tree-optimization/57442
8583         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8584         when control exits the main loop.
8586 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
8588         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8589         and RX600.
8590         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8591         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8592         * rx/t-rx: Add rx100 under multi library matches option for nofpu
8593         option.
8595 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8597         PR tree-optimization/57441
8598         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8599         Don't limit size of incr_vec to number of candidates.
8601 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
8603         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8604         and mips16 directories.
8605         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8606         (MULTILIB_DIRNAMES): Ditto.
8607         (MULTILIB_EXCEPTIONS): Add new exceptions.
8608         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8609         (MULTILIB_DIRNAMES): Ditto.
8610         (MULTILIB_EXCEPTIONS): Add new exceptions.
8612 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8613             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8615         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8616         SYMBOL_TINY_ABSOLUTE.
8617         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8618         SYMBOL_TINY_ABSOLUTE.
8619         (aarch64_expand_mov_immediate): Likewise.
8620         (aarch64_classify_symbol): Likewise.
8621         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8622         Permit SYMBOL_TINY_ABSOLUTE.
8623         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8625 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8626             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8628         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8629         Refactor if/switch.  Replace gcc_assert with if.
8631 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8633         * config/i386/i386.c (initial_ix86_tune_features): Enable
8634         FP Reassociation for AMD bdver1 and bdver2.
8636 2013-05-29  Martin Jambor  <mjambor@suse.cz>
8638         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8639         and IMAGPART_EXPR do not occur within other handled_components.
8641 2013-05-29  Richard Biener  <rguenther@suse.de>
8643         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8644         access on whether the use is in the BB we currently try to
8645         vectorize.
8646         (vect_bb_vectorization_profitable_p): Pass the BB we currently
8647         vectorize to vect_bb_slp_scalar_cost.
8649 2013-05-29  Richard Biener  <rguenther@suse.de>
8651         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8652         computing scalar cost offsetted by stmts that are kept live
8653         by scalar uses.
8654         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8655         for computation of scalar cost.
8657 2013-05-28  Steve Ellcey  <sellcey@mips.com>
8659         * config/mips/mips-cpus.def (mips32r2): Change processor type.
8661 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8663         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8664         array notation built-in reduction functions.
8665         * doc/passes.texi (Passes): Added documentation about changes done
8666         for Cilk Plus.
8667         * doc/invoke.texi (C Dialect Options): Added documentation about
8668         the -fcilkplus flag.
8669         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8670         (BUILTINS_DEF): Depend on cilkplus.def.
8671         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
8672         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8673         * cilkplus.def: New file.
8675 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
8677         PR rtl-optimization/57439
8678         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8680 2013-05-28  Easwaran Raman  <eraman@google.com>
8682         PR tree-optimization/57337
8683         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8684         (find_insert_point): Correctly identify the insertion point
8685         when two statements with the same UID is compared.
8687 2013-05-28  Richard Biener  <rguenther@suse.de>
8689         PR tree-optimization/56787
8690         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8691         from the list of data references.
8692         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8693         clobbers.
8694         (vect_analyze_loop_operations): Likewise.
8695         (vect_transform_loop): Remove clobbers.
8697 2013-05-28  Martin Jambor  <mjambor@suse.cz>
8699         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8700         and REALPART_EXPRs have scalar type.
8702 2013-05-28  Richard Biener  <rguenther@suse.de>
8704         PR tree-optimization/57411
8705         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8706         virtual operands.
8707         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8708         virtual operand propagation.
8710 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8712         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8713         destination register for bmasksi_vis.
8714         (vector_init_bshuffle): Likewise.
8715         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8717 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8719         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8720         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8721         mode if the instruction isn't available in the original mode.
8722         * config/sparc/sparc.opt (mfix-ut699): New option.
8723         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8724         (divdf3): Turn into expander.
8725         (divdf3_nofix): New insn.
8726         (divdf3_fix): Likewise.
8727         (divsf3): Disable if -mfix-ut699.
8728         (sqrtdf2): Turn into expander.
8729         (sqrtdf2_nofix): New insn.
8730         (sqrtdf2_fix): Likewise.
8731         (sqrtsf2): Disable if -mfix-ut699.
8733 2013-05-27  Richard Biener  <rguenther@suse.de>
8735         PR middle-end/57412
8736         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8737         block for the new loop.
8739 2013-05-27  Richard Biener  <rguenther@suse.de>
8741         PR tree-optimization/57343
8742         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8743         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8744         (number_of_iterations_cond): Do not build the folded tree.
8746 2013-05-27  Richard Biener  <rguenther@suse.de>
8748         Revert
8749         PR middle-end/57381
8750         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8751         OEP_CONSTANT_ADDRESS_OF retained.
8753         PR tree-optimization/57417
8754         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8755         for unchanged base.
8756         (set_ssa_val_to): Compare addresses using
8757         get_addr_base_and_unit_offset.
8759 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
8761         PR rtl-optimization/56833
8762         * postreload.c (move2add_record_mode): New function.
8763         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8764         (move2add_use_add2_insn): Use move2add_record_sym_value.
8765         (move2add_use_add3_insn): Likewise.
8766         (reload_cse_move2add): Use move2add_valid_value_p and
8767         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8768         affected regs by setting reg_mode to VOIDmode.
8769         (move2add_note_store): Don't pretend the inside of a SUBREG is
8770         the actual destination.  Invalidate single/leading registers by
8771         setting reg_mode to VOIDmode.
8772         Use move2add_record_sym_value, move2add_valid_value_p and
8773         move2add_record_mode.
8775 2013-05-27  Richard Biener  <rguenther@suse.de>
8777         PR tree-optimization/57396
8778         * tree-affine.c (double_int_constant_multiple_p): Properly
8779         return false for val == 0 and div != 0.
8781 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8783         * config/mips/mips.h: Use #elif in preprocessor conditions.
8785 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8787         PR target/53916
8788         * config/mips/constraints.md (kl): New constraint.
8789         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8790         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8791         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8792         from LO for MIPS16.
8793         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8795 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8797         PR target/55777
8798         * config/mips/mips.c (mips_can_inline_p): New function.
8799         (TARGET_CAN_INLINE_P): Define.
8801 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8803         * sched-int.h (ds_t, dw_t): Make unsigned int.
8804         Fix documentation that describes how all the ds_t bits are used.
8805         Reserve the last bit for delayed-branch scheduling.
8806         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8807         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8808         (gen_dep_weak_1): Remove prototype.
8809         * sched-deps.c (get_dep_weak_1): Make static.
8810         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8811         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8812         * doc/tm.texi: Regenerate.
8813         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8815 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8817         PR debug/56950
8818         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8820 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8821             Sandra Loosemore  <sandra@codesourcery.com>
8823         * config.gcc (powerpc-*): Allow native for with-cpu.
8825 2013-05-24  Jeff Law  <law@redhat.com>
8827         PR tree-optimization/57124
8828         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8829         conversion feeding a condition if the range has an overflow
8830         if -fstrict-overflow.  Add warnings for when we do make the
8831         transformation.
8833 2013-05-24  Dehao Chen  <dehao@google.com>
8835         * tree-cfg.c (locus_discrim_map): Fix the typo.
8836         (locus_discrim_hasher): Likewise.
8837         (locus_discrim_hasher::hash): Likewise.
8838         (locus_discrim_hasher::equal): Likewise.
8840 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8842         PR tree-optimization/57294
8843         * cgraph.h (ipa_record_stmt_references): Declare.
8844         * cgraphbuild.c (ipa_record_stmt_references): New function.
8845         (build_cgraph_edges): Use ipa_record_stmt_references.
8846         (rebuild_cgraph_edges): Likewise.
8847         (cgraph_rebuild_references): Likewise.
8848         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8849         associated with the old statement and build references from the
8850         newly built statements.
8851         * ipa-ref.c (ipa_remove_stmt_references): New function.
8852         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8854 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8856         * lra-constraints.c (emit_spill_move): Use smaller mode for
8857         mem-mem moves.
8858         (check_and_process_move): Consider mem-reg moves for secondary
8859         too.
8860         (curr_insn_transform): Don't lose insns emitted before for
8861         secondary memory moves.
8862         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8863         reg set up in the current insn.
8865 2013-05-24  Dehao Chen  <dehao@google.com>
8867         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8868         hash function.
8869         (locus_descrim_hasher::equal): Likewise.
8870         (build_gimple_cfg): New discriminator assignment algorithm.
8871         (make_edges): Likewise.
8872         (next_discriminator_for_locus): Likewise.
8873         (same_line_p): Likewise.
8874         (assign_discriminators): Likewise.
8875         (make_cond_expr_edges): Likewise.
8876         (make_gimple_switch_edges): Likewise.
8877         (make_goto_expr_edges): Likewise.
8878         (make_gimple_asm_edges): Likewise.
8880 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8882         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8883         X format specifier to only display bottom 16 bits.
8884         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8885         immediate to match for operand 2, since it will be masked.
8887 2013-05-24  Richard Biener  <rguenther@suse.de>
8889         PR tree-optimization/57287
8890         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8891         all SSA names that occur in abnormal PHIs.
8893 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8895         PR tree-ssa/57385
8896         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8897         that index is not negative.
8899 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8901         PR rtl-optimization/55177
8902         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8903         (simplify_byte_swapping_operation): New.
8904         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8905         (simplify_relational_operation_1): Deal with BSWAP.
8907 2013-05-23  Richard Henderson  <rth@redhat.com>
8909         PR target/56742
8910         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8911         (ix86_reorg): Call it.
8913 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8915         PR target/57379
8916         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8917         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8918         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8920 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8922         PR debug/57351
8923         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8925 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8926             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8928         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8929         * config/aarch64/constraints.md (Usa): Remove.
8930         * doc/md.texi (AArch64 Usa): Remove.
8932 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8933             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8935         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8936         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8937         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8938         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8940 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
8942         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8943         instruction (AdvSIMD).
8944         * config/aarch64/aarch64-builtins.c
8945         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8946         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8948 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8950         PR middle-end/57347
8951         * tree.h (contains_bitfld_component_ref_p): Declare.
8952         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8953         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8954         caller.
8955         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8956         not access a bit-field.  Assert all final offsets are byte-aligned.
8958 2013-05-23  Richard Biener  <rguenther@suse.de>
8960         PR tree-optimization/57380
8961         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8962         least one invariant or re-used load.
8963         * passes.c (init_optimization_passes): Move pass_phiprop before
8964         pass_forwprop.
8966 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8968         * config/aarch64/aarch64-simd.md
8969         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8971 2013-05-23  Richard Biener  <rguenther@suse.de>
8973         PR middle-end/57381
8974         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8975         OEP_CONSTANT_ADDRESS_OF retained.
8977 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8979         PR middle-end/57344
8980         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8981         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8983 2013-05-23  Richard Biener  <rguenther@suse.de>
8985         PR rtl-optimization/57341
8986         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8987         instead of true_dependence.
8989 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8991         * bb-reorder.c (branch_threshold): Make const.
8992         (exec_threshold): Ditto.
8994 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8995             Pat Haugen  <pthaugen@us.ibm.com>
8996             Peter Bergner  <bergner@vnet.ibm.com>
8998         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8999         documentation for the power8 crypto builtins.
9001         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
9003         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
9004         macros for defining power8 builtin functions.
9005         (BU_P8V_AV_2): Likewise.
9006         (BU_P8V_AV_P): Likewise.
9007         (BU_P8V_VSX_1): Likewise.
9008         (BU_P8V_OVERLOAD_1): Likewise.
9009         (BU_P8V_OVERLOAD_2): Likewise.
9010         (BU_CRYPTO_1): Likewise.
9011         (BU_CRYPTO_2): Likewise.
9012         (BU_CRYPTO_3): Likewise.
9013         (BU_CRYPTO_OVERLOAD_1): Likewise.
9014         (BU_CRYPTO_OVERLOAD_2): Likewise.
9015         (XSCVSPDP): Fix typo, point to the correct instruction.
9016         (VCIPHER): Add power8 crypto builtins.
9017         (VCIPHERLAST): Likewise.
9018         (VNCIPHER): Likewise.
9019         (VNCIPHERLAST): Likewise.
9020         (VPMSUMB): Likewise.
9021         (VPMSUMH): Likewise.
9022         (VPMSUMW): Likewise.
9023         (VPERMXOR_V2DI): Likewise.
9024         (VPERMXOR_V4SI: Likewise.
9025         (VPERMXOR_V8HI: Likewise.
9026         (VPERMXOR_V16QI: Likewise.
9027         (VSHASIGMAW): Likewise.
9028         (VSHASIGMAD): Likewise.
9029         (VPMSUM): Likewise.
9030         (VPERMXOR): Likewise.
9031         (VSHASIGMA): Likewise.
9033         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9034         __CRYPTO__ if the crypto instructions are available.
9035         (altivec_overloaded_builtins): Add support for overloaded power8
9036         builtins.
9038         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9039         support for power8 crypto builtins.
9040         (builtin_function_type): Likewise.
9041         (altivec_init_builtins): Add support for builtins that take vector
9042         long long (V2DI) arguments.
9044         * config/rs6000/crypto.md: New file, define power8 crypto
9045         instructions.
9047 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9048             Pat Haugen  <pthaugen@us.ibm.com>
9049             Peter Bergner  <bergner@vnet.ibm.com>
9051         * doc/invoke.texi (Option Summary): Add power8 options.
9052         (RS/6000 and PowerPC Options): Likewise.
9054         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9055         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
9056         wm, wn, wr documentation.
9058         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9059         registers if direct move instructions are enabled.
9060         (wn): New constraint for no registers.
9061         (wq): New constraint for quad word even GPR registers.
9062         (wr): New constraint if 64-bit instructions are enabled.
9063         (wv): New constraint if power8 vector instructions are enabled.
9064         (wQ): New constraint for quad word memory locations.
9066         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9067         constraint for 0..15 for crypto instructions.
9068         (gpc_reg_operand): If VSX allow registers in VSX registers as well
9069         as GPR and floating point registers.
9070         (int_reg_operand): New predicate to match only GPR registers.
9071         (base_reg_operand): New predicate to match base registers.
9072         (quad_int_reg_operand): New predicate to match even GPR registers
9073         for quad memory operations.
9074         (vsx_reg_or_cint_operand): New predicate to allow vector logical
9075         operations in both GPR and VSX registers.
9076         (quad_memory_operand): New predicate for quad memory operations.
9077         (reg_or_indexed_operand): New predicate for direct move support.
9079         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9080         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9081         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9082         (POWERPC_MASKS): Add power8 options.
9083         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9084         various options.
9086         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9087         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9089         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9090         (-mpower8-fusion): New power8 options.
9091         (-mpower8-fusion-sign): Likewise.
9092         (-mpower8-vector): Likewise.
9093         (-mcrypto): Likewise.
9094         (-mdirect-move): Likewise.
9095         (-mquad-memory): Likewise.
9097         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9098         power8.
9099         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9100         registers.
9101         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9102         (rs6000_debug_vector_unit): Add p8_vector.
9103         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9104         definitions.  Also print fusion state.
9105         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9106         (rs6000_builtin_mask_calculate): Add power8 builtin support.
9107         (rs6000_option_override_internal): Add support for power8.
9108         (rs6000_common_init_builtins): Add debugging for skipped builtins
9109         if -mdebug=builtin.
9110         (rs6000_adjust_cost): Add power8 support.
9111         (rs6000_issue_rate): Likewise.
9112         (insn_must_be_first_in_group): Likewise.
9113         (insn_must_be_last_in_group): Likewise.
9114         (force_new_group): Likewise.
9115         (rs6000_register_move_cost): Likewise.
9116         (rs6000_opt_masks): Likewise.
9118         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9119         power8 capable assembler, default to power7 options.
9120         (TARGET_DIRECT_MOVE): Likewise.
9121         (TARGET_CRYPTO): Likewise.
9122         (TARGET_P8_VECTOR): Likewise.
9123         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9124         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9125         (VECTOR_MEM_P8_VECTOR_P): Likewise.
9126         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9127         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9128         (TARGET_XSCVDPSPN): Likewise.
9129         (TARGET_XSCVSPDPN): Likewsie.
9130         (TARGET_SYNC_HI_QI): Likewise.
9131         (TARGET_SYNC_TI): Likewise.
9132         (MASK_CRYPTO): Likewise.
9133         (MASK_DIRECT_MOVE): Likewise.
9134         (MASK_P8_FUSION): Likewise.
9135         (MASK_P8_VECTOR): Likewise.
9136         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9137         temporary used by some of the direct move instructions to get two FP
9138         temporary registers does not force creation of a stack frame.
9139         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9140         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9141         that any VSX registers are tieable, even if they are also an
9142         Altivec vector mode.
9143         (r6000_reg_class_enum): Add wm, wr, wv constraints.
9144         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9145         (RS6000_BTM_CRYPTO): Likewise.
9146         (RS6000_BTM_COMMON): Likewise.
9148         * config/rs6000/rs6000.md (cpu attribute): Add power8.
9149         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9150         (enum rs6000_vector): Add power8 vector support.
9152 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9154         PR target/19599
9155         PR target/57340
9156         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9157         (any_sibcall_could_use_r3): this and handle indirect calls.
9158         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9160 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9162         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9164 2013-05-22  Richard Biener  <rguenther@suse.de>
9166         PR middle-end/57349
9167         * profile.c (branch_prob): Do not split blocks that are
9168         abnormally receiving from ECF_RETURNS_TWICE functions.
9170 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9172         * recog.c (offsettable_address_addr_space_p): Fix calculation of
9173         address mode.  Move pointer mode initialization to the same place.
9175 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9177         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9178         while it has any effect.
9180 2013-05-21  Easwaran Raman  <eraman@google.com>
9182         PR tree-optimization/57322
9183         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9184         UID of the statement added to the BB to be 1.
9186 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
9188         PR tree-optimization/57331
9189         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9190         of conversion from pointer type to integral type with integer.
9192 2013-05-21  Martin Jambor  <mjambor@suse.cz>
9194         PR lto/57289
9195         * ipa-prop.c (ipa_read_node_info): Process param_used and
9196         controlled_uses in the same order as when writing.
9198 2013-05-21  Magnus Granberg  <baldrick@free.fr>
9200         PR plugins/56754
9201         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9203 2013-05-21  Richard Biener  <rguenther@suse.de>
9205         PR tree-optimization/57318
9206         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9207         estimate stmts with side-effects as likely eliminated.
9209 2013-05-21  Richard Biener  <rguenther@suse.de>
9211         PR tree-optimization/57330
9212         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9213         preserve the call stmts fntype.
9215 2013-05-21  Richard Biener  <rguenther@suse.de>
9217         PR tree-optimization/57303
9218         * tree-ssa-sink.c (statement_sink_location): Improve killing
9219         stmt detection and properly handle self-assignments.
9221 2013-05-21  Christian Bruel  <christian.bruel@st.com>
9223         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9224         spanning registers. LEAF_REG_REMAP is supported only for contiguous
9225         registers. Set register size out of the PARALLEL loop.
9227 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
9229         PR target/56547
9230         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9231         (*fmasf4, *fmasf4_media): New insns.
9233 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9235         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9236         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9237         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9238         (mips_idiv_insns): Update the comments to say that the returned
9239         instruction counts are in units of BASE_INSN_LENGTH.
9240         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9241         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
9242         using 2 rather than 4 as the length of indirect MIPS16 and
9243         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
9244         length of a NOP.  Don't divide MIPS16 lengths by 2.
9245         (mips16_split_long_branches): Assume a branch is long if the
9246         length is greater than 4 rather than 8.
9247         * config/mips/mips.md (length): Give MIPS16 lengths directly,
9248         rather than multiplying them by 2.  Multiply instruction counts
9249         by BASE_INSN_LENGTH rather than 4.
9250         (*jump_mips16, tls_get_tp_mips16_<mode>)
9251         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9253 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9255         * config/mips/mips.md (extended_mips16): Remove branch case.
9256         (length): Remove duplicated extended_mips16 test.
9258 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9260         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9262 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
9264         * recog.h (Recog_data): Rename to...
9265         (recog_data_d): ...this.
9266         (recog_data): Update accordingly.
9267         * recog.c (recog_data): Likewise.
9268         * reload.c (save_recog_data): Likewise.
9269         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9270         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9272 2013-05-17  Julian Brown  <julian@codesourcery.com>
9274         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9275         found in a REG_EQUAL note, invalidate it.
9277 2013-05-17  Easwaran Raman  <eraman@google.com>
9279         * tree-ssa-reassoc.c (find_insert_point): New function.
9280         (insert_stmt_after): Likewise.
9281         (get_def_stmt): Likewise.
9282         (ensure_ops_are_available): Likewise.
9283         (not_dominated_by): Likewise.
9284         (rewrite_expr_tree): Do not move statements beyond what is
9285         necessary. Remove call to swap_ops_for_binary_stmt...
9286         (reassociate_bb): ... and move it here.
9287         (build_and_add_sum): Assign UIDs for new statements.
9288         (linearize_expr): Likewise.
9289         (do_reassoc): Renumber gimple statement UIDs.
9291 2013-05-17  Jan Hubicka  <jh@suse.cz>
9293         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9294         weakrefs.
9295         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9296         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9297         weakrefs.
9298         (output_weakrefs): Update.
9300 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
9301             Martin Jambor  <mjambor@suse.cz>
9303         PR middle-end/57276
9304         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9305         value that corresponds to the given aggval is found in values vector.
9307 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
9309         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9310         sse, sse2, sse3, ssse3 and sse4a flags to options.
9312 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
9314         * gengtype-state.c: (s_expr_writer): New class, to handle
9315         prettifying of output layout of s-expressions.
9316         (state_writer): New class, to write out gtype.state.
9317         (state_written_type_count): Move this variable into member data of
9318         state_writer.
9319         (s_expr_writer::s_expr_writer): New code: constructor for new class
9320         (state_writer::state_writer(): ditto
9321         (s_expr_writer::write_new_line): New function
9322         (s_expr_writer::write_any_indent): ditto
9323         (s_expr_writer::begin_s_expr): ditto
9324         (s_expr_writer::end_s_expr): ditto
9325         (write_state_fileloc): convert to method of state_writer...
9326         (state_writer:: write_state_fileloc): ...and use methods of
9327         s_expr_writer to write indentation into the gtype.state output file
9328         to visually represent the hierarchical structure of the list
9329         structures
9330         (write_state_fields): ditto, renaming to...
9331         (state_writer::write_state_fields)
9332         (write_state_a_string): ditto, renaming to...
9333         (state_writer::write_state_a_string)
9334         (write_state_string_option): ditto, renaming to...
9335         (state_writer::write_state_string_option)
9336         (write_state_type_option): ditto, renaming to...
9337         (state_writer::write_state_type_option)
9338         (write_state_nested_option): ditto, renaming to...
9339         (state_writer::write_state_nested_option)
9340         (write_state_option): ditto, renaming to...
9341         (state_writer::write_state_option)
9342         (write_state_options): ditto, renaming to...
9343         (state_writer::write_state_options)
9344         (write_state_lang_bitmap): ditto, renaming to...
9345         (state_writer::write_state_lang_bitmap)
9346         (write_state_version): ditto, renaming to...
9347         (state_writer::write_state_version)
9348         (write_state_scalar_type): ditto, renaming to...
9349         (state_writer::write_state_scalar_type)
9350         (write_state_string_type): ditto, renaming to...
9351         (state_writer::write_state_string_type)
9352         (write_state_undefined_type): ditto, renaming to...
9353         (state_writer::write_state_undefined_type)
9354         (write_state_struct_union_type): ditto, renaming to...
9355         (state_writer::write_state_struct_union_type)
9356         (write_state_struct_type): ditto, renaming to...
9357         (state_writer::write_state_struct_type)
9358         (write_state_user_struct_type): ditto, renaming to...
9359         (state_writer::write_state_user_struct_type)
9360         (write_state_lang_struct_type): ditto, renaming to...
9361         (state_writer::write_state_lang_struct_type)
9362         (write_state_param_struct_type): ditto, renaming to...
9363         (state_writer::write_state_param_struct_type)
9364         (write_state_pointer_type): ditto, renaming to...
9365         (state_writer::write_state_pointer_type)
9366         (write_state_array_type): ditto, renaming to...
9367         (state_writer::write_state_array_type)
9368         (write_state_gc_used): ditto, renaming to...
9369         (state_writer::write_state_gc_used)
9370         (write_state_common_type_content): ditto, renaming to...
9371         (state_writer::write_state_common_type_content)
9372         (write_state_type): ditto, renaming to...
9373         (state_writer::write_state_type)
9374         (write_state_pair_list): ditto, renaming to...
9375         (state_writer::write_state_pair_list)
9376         (write_state_pair): ditto, renaming to...
9377         (state_writer::write_state_pair)
9378         (write_state_typedefs): ditto, renaming to...
9379         (state_writer::write_state_typedefs)
9380         (write_state_structures): ditto, renaming to...
9381         (state_writer::write_state_structures)
9382         (write_state_param_structs): ditto, renaming to...
9383         (state_writer::write_state_param_structs)
9384         (write_state_variables): ditto, renaming to...
9385         (state_writer::write_state_variables)
9386         (write_state_srcdir): ditto, renaming to...
9387         (state_writer::write_state_srcdir)
9388         (write_state_files_list): ditto, renaming to...
9389         (state_writer::write_state_files_list)
9390         (write_state_languages): ditto, renaming to...
9391         (state_writer::write_state_languages)
9392         (write_state): create a state_writer instance and use it when
9393         writing out the state file
9395 2013-05-17  Mike Stump  <mikestump@comcast.net>
9397         PR rtl-optimization/57304
9398         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9399         accessing DF_REF_REAL_LOC.
9401 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9403         PR rtl-optimization/57281
9404         PR rtl-optimization/57300
9405         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9406         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9407         what the other splitter did if the registers are dead.
9409 2013-05-17  Richard Biener  <rguenther@suse.de>
9411         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9412         MEM_REF offsets.
9414 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9416         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9417         linking.
9419 2013-05-17  Marek Polacek  <polacek@redhat.com>
9421         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9422         length when doing non-zero store of storing '\0' to '\0'.
9424 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9426         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9427         vect_external_def oprnd1 with loop_vinfo, try to emit
9428         optional cast, negation and and stmts on the loop preheader
9429         edge instead of into the pattern def seq.
9431         PR tree-optimization/57051
9432         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9433         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9435 2013-05-16  Nick Clifton  <nickc@redhat.com>
9437         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9438         (rl78_is_naked_func): New function.
9439         (rl78_expand_prologue): Skip prologue generation for naked functions.
9440         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9441         * doc/extend.texi (naked): Add RL78 to the list of processors
9442         that supports this attribute.
9444 2013-05-16  Jeff Law  <law@redhat.com>
9446         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9448 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9450         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9451         cache parameters using detect_caches_amd also for CYRIX,
9452         NSC and TM2 signatures.
9454 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9455             Dzianis Kahanovich  <mahatma@eu.by>
9457         PR target/45359
9458         PR target/46396
9459         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9460         VIA/Centaur processors and determine their cache parameters
9461         using detect_caches_amd.
9463 2013-05-16  Teresa Johnson  <tejohnson@google.com>
9465         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9466         (rtl_verify_edges): New function.
9467         (rtl_verify_bb_insns): Ditto.
9468         (rtl_verify_bb_pointers): Ditto.
9469         (rtl_verify_bb_insn_chain): Ditto.
9470         (rtl_verify_fallthru): Ditto.
9471         (rtl_verify_bb_layout): Ditto.
9472         (rtl_verify_flow_info_1): Outline checks into new functions.
9473         (rtl_verify_flow_info): Ditto.
9475 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
9477         * cfghooks.c (copy_bbs): Add update_dominance argument.
9478         * cfghooks.h (copy_bbs): Update prototype.
9479         * tree-cfg.c (gimple_duplicate_sese_region):
9480         Add update_dominance argument.
9481         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9482         * tree-ssa-loop-ch.c (copy_loop_headers): Update
9483         gimple_duplicate_sese_region call.
9484         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9485         Update copy_bbs call.
9486         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9487         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9489 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9491         * tree-vectorizer.h (NUM_PATTERNS): Increment.
9492         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9493         vect_recog_rotate_pattern.
9494         (vect_recog_rotate_pattern): New function.
9496 2013-05-16  Jason Merrill  <jason@redhat.com>
9498         * Makefile.in (LLINKER): New variable.
9499         (mostlyclean): Remove link mutex.
9500         * configure.ac: Handle --enable-link-mutex.
9501         * lock-and-run.sh: New script.
9503 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9505         PR target/19599
9506         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9507         for NULL decl.
9509 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9511         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9513 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
9515         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9516         * config/arm/arm.c (next_consecutive_mem): New function.
9517         (gen_movmem_ldrd_strd): Likewise.
9518         * config/arm/arm.md (movmemqi): Update condition and code.
9519         (unaligned_loaddi, unaligned_storedi): New patterns.
9521 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9523         * config.gcc: Obsolete *-*-solaris2.9*.
9524         * doc/install.texi (Specific, *-*-solaris2*): Document it.
9526 2013-05-16  Richard Biener  <rguenther@suse.de>
9528         * passes.c (init_optimization_passes): Move pass_parallelize_loops
9529         earlier, after GRAPHITE transforms and IV canonicalization.
9531 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9533         * omp-low.c (extract_omp_for_data): For collapsed loops,
9534         if at least one of the loops is known at compile time to
9535         iterate zero times, set count to 0.
9536         (expand_omp_regimplify_p): New function.
9537         (expand_omp_for_generic): For collapsed loops, if at least
9538         one of the loops isn't known to iterate at least once,
9539         add runtime check with setting count to 0.
9540         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9541         For unsigned types if it isn't known at compile time that
9542         the loop will iterate at least once, add runtime check to bypass
9543         the whole loop if initial condition isn't true.
9545 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
9547         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9549 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
9551         PR middle-end/57286
9552         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9553         transformations to avoid an infinite loop.
9555 2013-05-16  Marek Polacek  <polacek@redhat.com>
9557         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9559 2013-05-15  Leif Ekblad  <leif@rdos.net>
9561         * config/i386/i386.c (ix86_decompose_address): Use
9562         DEFAULT_TLS_SEG_REG to access TLS segment register.
9563         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9564         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9565         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9567 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
9569         PR target/57260
9570         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9571         sibling calls to functions that would normally be lazily bound,
9572         unless $gp is call-clobbered.
9574 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
9576         * config/i386/i386.c (ix86_option_override_internal): Update
9577         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
9578         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9579         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9580         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9581         of TARGET_3DNOW.
9582         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9584 2013-05-15  Andreas Schwab  <schwab@suse.de>
9586         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9587         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
9588         third operand.
9590 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9592         * loop-unroll.c (report_unroll_peel): Check decision before
9593         emitting unroll/peel message.
9595 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9597         * function.h (has_bb_partition): New rtl_data flag.
9598         (bb_reorder_complete): Ditto.
9599         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9600         instead of flag_reorder_blocks_and_partition.
9601         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9602         with some enhancements.
9603         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9604         * bb-reorder.c (connect_traces): Check for has_bb_partition
9605         instead of flag_reorder_blocks_and_partition.
9606         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9607         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9608         verify_hot_cold_block_grouping.
9609         (partition_hot_cold_basic_blocks): Set has_bb_partition.
9611 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9613         PR target/19599
9614         * config/arm/predicates.md (call_insn_operand): New predicate.
9615         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
9616         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9617         if insn is not a tail call.
9618         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9619         registers.
9620         * config/arm/arm.h (enum reg_class): New caller save register class.
9621         (REG_CLASS_NAMES): Likewise.
9622         (REG_CLASS_CONTENTS): Likewise.
9623         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9624         without decls.
9626 2013-05-15  Richard Biener  <rguenther@suse.de>
9628         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9629         of MSG_OPTIMIZED_LOCATIONS.
9630         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9631         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9632         message.
9633         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9634         of MSG_OPTIMIZED_LOCATIONS.
9635         (execute_vect_slp): Likewise.
9636         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9637         (vect_create_cond_for_alias_checks): Likewise.
9638         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9639         (vect_recog_widen_mult_pattern): Likewise.
9640         (vect_recog_widen_sum_pattern): Likewise.
9641         (vect_recog_over_widening_pattern): Likewise.
9642         (vect_recog_widen_shift_pattern): Likewise.
9643         (vect_recog_vector_vector_shift_pattern): Likewise.
9644         (vect_recog_divmod_pattern): Likewise.
9645         (vect_recog_mixed_size_cond_pattern): Likewise.
9646         (vect_recog_bool_pattern): Likewise.
9647         (vect_pattern_recog_1): Likewise.
9649 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9651         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9652         non-functions to builtin_unreachable.
9653         * ipa-inline-transform.c (inline_call): Do not assert estimates were
9654         correct when new direct edges were discovered.
9656 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9658         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9659         header, print symbol order instead of node uid, print more information
9660         about indirect edge targets.
9661         (ipa_make_edge_direct_to_target): Print symbol order instead of node
9662         uids.
9663         (ipa_make_edge_direct_to_target): Likewise.
9664         (remove_described_reference): Likewise.
9665         (propagate_controlled_uses): Likewise.
9666         (ipa_print_node_params): Also print symbol order.
9667         (ipcp_transform_function): Print symbol order instead of node uids.
9668         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9669         (cgraph_get_create_real_symbol_node): Likewise.
9670         * ipa-cp.c (print_lattice): Likewise.
9671         (print_all_lattices): Likewise.
9672         (determine_versionability): Likewise.
9673         (initialize_node_lattices): Likewise.
9674         (estimate_local_effects): Likewise.
9675         (update_profiling_info): Likewise.
9676         (create_specialized_node): Likewise.
9677         (perhaps_add_new_callers): Likewise.
9678         (decide_about_value): Likewise.
9679         (decide_whether_version_node): Likewise.
9680         (identify_dead_nodes): Likewise.
9681         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9682         (dump_inline_summary): Likewise.
9683         (estimate_node_size_and_time): Likewise.
9684         (inline_analyze_function): Likewise.
9685         * ipa-inline.c (report_inline_failed_reason): Likewise.
9686         (want_early_inline_function_p): Likewise.
9687         (edge_badness): Likewise.
9688         (update_edge_key): Likewise.
9689         (inline_small_functions): Likewise.  Add dumping of order to two other
9690         dumps.
9691         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9692         instead of node uids.
9693         (propagate_pure_const): Likewise.
9694         (propagate_pure_const): Likewise.
9695         * ipa-utils.c (dump_cgraph_node_set): Likewise.
9696         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9697         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9698         of node uids.
9699         * tree-pretty-print.c (dump_function_header): Likewise.
9700         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9701         Print symbol order instead of node uids.
9703 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9705         * config/s390/s390.c (s390_register_move_cost): Don't impose the
9706         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9708 2013-05-15  Richard Biener  <rguenther@suse.de>
9710         PR tree-optimization/57275
9711         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9712         return value for fail to do runtime alias checks for gather loads.
9714 2013-05-15  Jan Hubicka  <jh@suse.cz>
9716         PR lto/57038
9717         PR lto/47375
9718         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9719         weakrefs are not external.
9720         (lto_symtab_merge_decls): Fix thinko when dealing with
9721         non-lto_symtab decls.
9722         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9723         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9724         * varpool.c (dump_varpool_node): Dump more flags.
9726 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9728         * config/i386/i386.c (processor_alias_table): Add instruction
9729         FSGSBASE for AMD bdver3 architecture.
9731 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9733         * tree.c (warn_deprecated_use): Print file:line using locus color.
9734         * diagnostic.c (diagnostic_report_current_module): Print file:line
9735         and file:line:column using locus color.
9737 2013-05-14  Mike Stump  <mikestump@comcast.net>
9739         * gdbinit.in: Add __null.
9741 2013-05-14  Mike Stump  <mikestump@comcast.net>
9743         * recog.h: Rename struct recog_data to Recog_data.
9744         * recog.c: Likewise.
9745         * reload.c (can_reload_into): Likewise.
9746         * config/picochip/picochip.c: Likewise.
9748 2013-05-14  Mike Stump  <mikestump@comcast.net>
9750         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9752 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9754         * resource.h (struct resources): Remove unch_memory member.
9755         (CLEAR_RESOURCE): Don't clear unch_memory.
9756         * resource.c (mark_referenced_resources): Don't set it.
9757         (mark_set_resources): Likewise.
9758         (mark_target_live_regs): Don't clear it.
9759         (init_resource_info): Likewise.
9760         * reorg.c (resource_conflicts_p): Don't compare it.
9761         (redundant_insn): Don't set it.
9763         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9764         Remove prototypes.
9765         * emit-rtl.c (next_label): Remove unused function.
9766         (skip_consecutive_labels, link_cc0_insns): Move to ...
9767         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9768         only place where these functions are used, and make them static.
9770 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9772         * fold-const.c (fold_negate_expr): Handle vectors.
9773         (fold_truth_not_expr): Make it static.
9774         (fold_invert_truthvalue): New static function.
9775         (invert_truthvalue_loc): Handle vectors. Do not call
9776         fold_truth_not_expr directly.
9777         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9778         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9779         (fold_comparison): Handle vector constants.
9780         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9781         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9782         * tree.h (fold_truth_not_expr): Remove declaration.
9784 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9786         * config/aarch64/aarch64-simd.md
9787         (aarch64_vcond_internal<mode>): Rename to...
9788         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9789         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9790         float modes. Clarify all iterator modes.
9791         (vcond<mode><mode>): Use new name for vcond expanders.
9792         (vcond<v_cmp_result><mode>): Likewise.
9793         (vcondu<mode><mode>: Likewise.
9794         * config/aarch64/iterators.md (VDQF_COND): New.
9796 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9798         PR bootstrap/57266
9799         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9800         variable for the shift amount. Check that we shift by non-negative
9801         amounts.
9803 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9805         PR target/42017
9806         * config/arm/arm.h (EPILOGUE_USES): Only return true
9807         for LR_REGNUM after epilogue_completed.
9809 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
9811         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9812         is error_mark_node.
9814 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9816         PR target/57261
9817         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9818         and Solaris 11+/x86 with gld.
9819         * configure: Regenerate.
9821 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9823         * expmed.c (expand_shift_1): Canonicalize rotates by
9824         constant bitsize / 2 to bitsize - 1.
9825         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9826         case ROTATERT>: Likewise.
9828         Revert:
9829         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9831         * config/i386/i386.md (rotateinv): New code attr.
9832         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9833         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9834         roll $31, %eax, etc.
9836 2013-05-14  Richard Biener  <rguenther@suse.de>
9838         PR middle-end/57235
9839         * tree-eh.c (sink_clobbers): Give up for successors with
9840         multiple predecessors and no virtual uses.
9842 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9844         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9845         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9847 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9849         PR middle-end/57251
9850         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9851         the case when both op0 and op1 have VOIDmode.
9853 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9855         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9856         in multiply-accumulate mode.
9858 2013-05-13  Guozhi Wei  <carrot@google.com>
9860         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9862 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9864         PR target/56975
9865         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9866         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9867         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9868         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9869         (ix86_expand_prologue): Likewise.
9870         (ix86_expand_split_stack_prologue): Likewise.
9871         (legitimate_pic_address_disp_p): Likewise.
9872         (legitimize_pic_address): Likewise.
9873         (legitimize_tls_address): Likewise.
9874         (legitimize_pe_coff_symbol): Likewise.
9875         (output_pic_addr_const): Likewise.
9876         (construct_plt_address): Likewise.
9877         (ix86_expand_call): Likewise.
9878         (x86_output_mi_thunk): Likewise.
9879         (x86_function_profiler): Likewise.
9881 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9883         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9884         similar switch cases.
9885         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9886         (aarch64_simd_mov_to_<mode>low): Delete.
9887         (aarch64_simd_mov_to_<mode>high): Delete.
9888         (move_lo_quad_<mode>): Add w<-r alternative.
9889         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9890         (aarch64_simd_mov_from_*): Update type attribute.
9891         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9892         statement.
9894 2013-05-13  Jan Hubicka  <jh@suse.cz>
9896         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9897         * config/i386/i386.c (ix86_compute_frame_layout,
9898         ix86_expand_epilogue, emit_i387_cw_initialization,
9899         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9900         ix86_local_alignment): Fix use of size/speed predicates.
9902 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9904         PR tree-optimization/45216
9905         PR tree-optimization/57157
9906         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9907         the (-Y) & (B - 1) variant if OP is |.
9908         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9909         return shifted.  Use (-op1) & (prec - 1) as other_amount
9910         instead of prec - op1.
9912 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9914         PR middle-end/42371
9915         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9916         (ipa_constant_data): New type.
9917         (ipa_jump_func): Use ipa_constant_data to hold information about
9918         constant jump functions.
9919         (ipa_get_jf_constant): Adjust to jump function type changes.
9920         (ipa_get_jf_constant_rdesc): New function.
9921         (ipa_param_descriptor): New field controlled_uses.
9922         (ipa_get_controlled_uses): New function.
9923         (ipa_set_controlled_uses): Likewise.
9924         * ipa-ref.h (ipa_find_reference): Declare.
9925         * ipa-prop.c (ipa_cst_ref_desc): New type.
9926         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9927         changes.
9928         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9929         New parameter cs.  Adjust all callers.
9930         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9931         (remove_described_reference): New function.
9932         (jfunc_rdesc_usable): Likewise.
9933         (try_make_edge_direct_simple_call): Decrement controlled use count,
9934         attempt to remove reference if it hits zero.
9935         (combine_controlled_uses_counters): New function.
9936         (propagate_controlled_uses): Likewise.
9937         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9938         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9939         (ipa_print_node_params): Print described use counter.
9940         (ipa_write_jump_function): Adjust to jump function type changes.
9941         (ipa_read_jump_function): New parameter CS, pass it to
9942         ipa_set_jf_constant.  Adjust caller.
9943         (ipa_write_node_info): Stream controlled use count
9944         (ipa_read_node_info): Likewise.
9945         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9946         asserting.
9947         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9948         count.  Remove cloning-added reference if it reaches zero.
9949         * ipa-ref.c (ipa_find_reference): New function.
9951 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9953         * config/i386/i386.c (processor_target_table): Modified default
9954         alignment values for AMD BD and BT architectures.
9956 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9958         * tree-vect-generic.c (uniform_vector_p): Move ...
9959         * tree.c (uniform_vector_p): ... here.
9960         * tree.h (uniform_vector_p): Declare it.
9961         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9962         into a scalar.
9964 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9966         PR tree-optimization/57230
9967         * tree-ssa-strlen.c (handle_char_store): Record length for
9968         array store from STRING_CST.
9970         PR tree-optimization/57230
9971         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9972         check.
9974 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9976         * config/epiphany/epiphany.c (epiphany_init): Check size of
9977         NUM_MODES_FOR_MODE_SWITCHING.
9978         (epiphany_expand_prologue):
9979         Remove CONFIG_REGNUM initial value handling code.
9980         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9981         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9982         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9983         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9984         Don't return 1 for FP_MODE_NONE.
9985         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9986         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9987         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9988         * config/epiphany/epiphany.md (save_config): New pattern.
9990 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9992         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9994 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9996         * config/i386/i386.md (memory): Handle sseishft1.
9997         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9998         (*vec_extractv2di_1): Ditto.
10000 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
10002         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
10003         saved registers.
10005 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
10007         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
10008         Add mthumb/march=armv7-a multilib.
10009         Add mthumb/march=armv7-r multilib.
10010         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10012 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
10014         * config/v850/t-rtems: Add more multilibs.
10016 2013-05-10  Richard Biener  <rguenther@suse.de>
10018         PR tree-optimization/57214
10019         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10020         not propagate from SSA names that occur in abnormal PHI nodes.
10022 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
10024         * stor-layout.c (element_precision): New function.
10025         * machmode.h (element_precision): Declare it.
10026         * tree.c (build_minus_one_cst): New function.
10027         (element_precision): Likewise.
10028         * tree.h (build_minus_one_cst): Declare new function.
10029         (element_precision): Likewise.
10030         * fold-const.c (operand_equal_p): Use element_precision.
10031         (fold_binary_loc): Handle vector types.
10032         * convert.c (convert_to_integer): Use element_precision.
10033         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10034         separately.
10036 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
10038         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10039         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10040         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10041         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10042         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10043         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10044         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10045         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10046         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10047         (Uuw8): New constraints.
10048         (Usb4): Move into alphabetical order.
10049         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10050         (sd8_operand, ub8_operand, uw8_operand): New predicates.
10051         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10052         previously unnamed patterns.
10053         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10054         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10055         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10056         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10057         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
10058         of specific lengths.
10060 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
10062         * config/i386/i386.md (rotateinv): New code attr.
10063         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10064         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10065         roll $31, %eax, etc.
10067         PR tree-optimization/45216
10068         PR tree-optimization/57157
10069         * tree-ssa-forwprop.c (simplify_rotate): New function.
10070         (ssa_forward_propagate_and_combine): Call it.
10072 2013-05-10  Richard Biener  <rguenther@suse.de>
10074         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10075         disable peeling when we version for aliasing.
10076         (vector_alignment_reachable_p): Honor explicit user alignment.
10077         (vect_supportable_dr_alignment): Likewise.
10078         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10079         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10080         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10081         then peeling to arrange for the cost-model check to come first.
10083 2013-05-10  Alan Modra  <amodra@gmail.com>
10085         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10086         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10087         * configure: Regenerate.
10089 2013-05-10  Alan Modra  <amodra@gmail.com>
10091         PR target/55033
10092         * varasm.c (default_elf_select_section): Move !DECL_P check..
10093         (get_named_section): ..to here before calling get_section_name.
10094         Adjust assertion.
10095         (default_section_type_flags): Add DECL_P check.
10096         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10097         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10099 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
10101         * config/epiphany/epiphany.c (epiphany_expand_prologue):
10102         When using gen_stack_adjust_str with a register offset, add a
10103         REG_FRAME_RELATED_EXPR note.
10105 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
10107         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10108         (*vec_extractv4si_zext_mem): Ditto.
10109         (*vec_extractv2di): Add 0->x and x->x alternatives.
10110         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10111         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10113 2013-05-09  Jason Merrill  <jason@redhat.com>
10115         N3639 C++1y VLA support
10116         * gimplify.c (gimplify_vla_decl): Don't touch an existing
10117         DECL_VALUE_EXPR.
10119         * tree.c (build_constructor_va): New.
10120         * tree.h: Declare it.
10122 2013-05-09  Martin Jambor  <mjambor@suse.cz>
10124         PR lto/57084
10125         * gimple-fold.c (canonicalize_constructor_val): Call
10126         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10128 2013-05-09  Jan Hubicka  <jh@suse.cz>
10129             Richard Biener  <rguenther@suse.de>
10131         PR lto/54095
10132         * symtab.c (symtab_make_decl_local): Do not add private names.
10134 2013-05-09  Jan Hubicka  <jh@suse.cz>
10136         PR lto/54095
10137         * symtab.c (insert_to_assembler_name_hash): Handle clones.
10138         (unlink_from_assembler_name_hash): Likewise.
10139         (symtab_prevail_in_asm_name_hash, symtab_register_node,
10140         symtab_unregister_node, symtab_initialize_asm_name_hash,
10141         change_decl_assembler_name): Update.
10143 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
10145         * config/aarch64/aarch64.md: New movtf split.
10146         (*movtf_aarch64): Update.
10147         (aarch64_movdi_tilow): Handle TF modes and rename to
10148         aarch64_movdi_<mode>low.
10149         (aarch64_movdi_tihigh): Handle TF modes and rename to
10150         aarch64_movdi_<mode>high
10151         (aarch64_movtihigh_di): Handle TF modes and rename to
10152         aarch64_mov<mode>high_di
10153         (aarch64_movtilow_di): Handle TF modes and rename to
10154         aarch64_mov<mode>low_di
10155         (aarch64_movtilow_tilow): Remove spurious whitespace.
10156         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10157         splits.
10158         (aarch64_print_operand): Update.
10160 2013-05-09  Alan Modra  <amodra@gmail.com>
10162         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10163         powerpc64le.
10164         * configure: Regenerate.
10166 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10168         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10169         splitter preparation statements.
10170         * config/i386/sse.md (*vec_extract* splitters): Ditto.
10171         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10172         adjust_address_nv.
10174 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10176         * gimple-ssa-strength-reduction.c (count_candidates): Change
10177         return value to int.
10178         (analyze_candidates_and_replace): Change type of length to int.
10180 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10182         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10183         (*vec_extract<mode>): Use VI12_128 mode iterator.
10184         (*vec_extract<mode>_mem): Ditto.
10185         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10186         attribute.
10188 2013-05-08  Diego Novillo  <dnovillo@google.com>
10190         PR bootstrap/54659
10192         Revert:
10193         2012-08-17  Diego Novillo  <dnovillo@google.com>
10195         PR bootstrap/54281
10196         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10197         * config.in: Regenerate.
10198         * configure: Regenerate.
10199         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10201 2013-05-08  Jan Hubicka  <jh@suse.cz>
10203         PR lto/54095
10204         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10205         * cgraph.h (symtab_node_base): Add unique_name.
10206         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10207         input_overwrite_node, input_varpool_node): Stream unique_name.
10208         * cgraphclones.c (cgraph_create_virtual_clone,
10209         cgraph_function_versioning): Set unique_name.
10210         * ipa.c (function_and_variable_visibility): Set unique_name.
10212 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10214         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10215         (alloc_cand_and_find_basis): Restrict conditional candidate
10216         processing to CAND_MULTs.
10218 2013-05-08  Jan Hubicka  <jh@suse.cz>
10220         PR lto/54095
10221         lto-symtab.c (lto_symtab_symbol_p): New function.
10222         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10223         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10224         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10225         Skip static symbols.
10227 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
10229         PR tree-optimization/57200
10230         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10231         Only call inform if the preceding warning_at returns true.
10233 2013-05-07  Han Shen  <shenhan@google.com>
10235         * cfgexpand.c (record_or_union_type_has_array_p): New function.
10236         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10237         * common.opt (fstack-protector-strong): New option.
10238         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10239         * doc/invoke.texi (Optimization Options): Document
10240         "-fstack-protector-strong".
10241         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10243 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10245         * config/mips/mips.c (mips_machine_reorg2): Return 0.
10247 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
10249         * ira.c (update_equiv_regs): Add insn having equiv memory even if
10250         it is not lhs of the insn.
10251         (setup_reg_equiv): Remove insn having equiv memory which it is not
10252         lhs of the insn.
10253         * lra-constraints.c (process_address): Try to improve generation
10254         code for address base + disp.
10255         (lra_constraints): Make correct the code for checking insn setting
10256         up backward equivalence.  Remove insn only if it is in the init
10257         insn list.
10258         * lra-eliminations.c (update_reg_eliminate): Change return value.
10259         (lra_eliminate): Use the result.
10261 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
10263         * config/i386/sse.md (ssescalarnummask): New mode attribute.
10264         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10265         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10266         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10267         register target operands.
10268         (*vec_extractv8hi_sse2): New pattern.
10269         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10270         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10271         (*vec_extract<mode>_mem): New insn and split pattern.
10273 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
10275         * config/arm/arm.c (arm_asan_shadow_offset): New function.
10276         (TARGET_ASAN_SHADOW_OFFSET): Define.
10277         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10278         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10280 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10282         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10283         (incr_vec_index): Return -1 if increment not found.
10284         (create_add_on_incoming_edge): Assert if increment not found.
10285         (record_increment): Limit number of increments recorded.
10286         (all_phi_incrs_profitable): Return false if an increment not found.
10287         (replace_profitable_candidates): Don't process increments that were
10288         not recorded.
10289         (analyze_candidates_and_replace): Limit size of incr_vec.
10291 2013-05-07  Richard Biener  <rguenther@suse.de>
10293         * calls.c (special_function_p): setjmp-like functions are leaf.
10294         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10295         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10297 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
10299         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10300         (aarch64_simd_mov<mode>): New expander.
10301         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10302         (aarch64_simd_mov_to_<mode>high): Likewise.
10303         (aarch64_simd_mov_from_<mode>low): Likewise.
10304         (aarch64_simd_mov_from_<mode>high): Likewise.
10305         (aarch64_dup_lane<mode>): Update.
10306         (aarch64_dup_lanedi): New instruction pattern.
10307         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10308         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10310 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10312         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10313         (replace_mult_candidate): Remove unnecessary argument; remove
10314         unnecessary parameter from call to introduce_cast_before_cand.
10315         (replace_unconditional_candidate): Remove unnecessary parameter
10316         from call to replace_mult_candidate.
10317         (replace_conditional_candidate): Likewise.
10318         (insert_initializers): Use make_temp_ssa_name.
10319         (introduce_cast_before_cand): Remove unnecessary argument; use
10320         make_temp_ssa_name.
10321         (replace_one_candidate): Remove unnecessary argument; remove
10322         unnecessary parameter from calls to introduce_cast_before_cand.
10323         (replace_profitable_candidates): Remove unnecessary parameters
10324         from calls to replace_one_candidate.
10326 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10328         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10329         phi def as possibly hiding a basis for a CAND_ADD whose operands
10330         have been commuted in the analysis.
10331         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10333 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10335         * config/aarch64/aarch64.md
10336         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10337         shift value between 0-4.
10339 2013-05-07  Richard Biener  <rguenther@suse.de>
10341         * double-int.h (rshift): New overload.
10342         * double-int.c (rshift): New function.
10343         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10344         (create_reference_ops_from_ref): Remove.
10345         (vn_reference_insert): Use shared ops for constructing the
10346         reference and copy it.
10348 2013-05-07  Richard Biener  <rguenther@suse.de>
10350         PR middle-end/57190
10351         * tree-eh.c (sink_clobbers): Properly propagate
10352         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10354 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
10356         PR tree-optimization/57149
10357         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10358         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10359         collect_phi_def_edges, execute_late_warn_uninitialized): Use
10360         uninit_undefined_value_p instead of ssa_undefined_value_p.
10362         PR debug/57184
10363         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10364         for modifier == EXPAND_INITIALIZER.
10366 2013-05-07  Anton Blanchard  <anton@samba.org>
10368         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10369         for powerpc64 little endian.
10370         * configure: Regenerate.
10372 2013-05-06  Graham Stott  <grahams@btinternet.com>
10374         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10375         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10376         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10377         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10379 2013-05-06  Graham Stott  <grahams@btinternet.com>
10381         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10382         codes which allow non-lvalues.
10384 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
10386         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10387         components are all 1s.
10388         (integer_minus_onep): New function.
10389         * tree.h (integer_minus_onep): Declare it.
10390         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10391         integer_minus_onep instead of integer_all_onesp.
10393 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10395         PR target/52933
10396         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10397         variations of these patterns.
10399 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10401         * config/i386/i386.md (isa): Add x64_sse4 member.
10402         (enabled): Handle x64_sse4.
10403         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10404         instruction for 64bit SSE4_1 targets.  Update insn attributes.
10405         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10406         instruction for SSE4_1 targets.  Update insn attributes.
10407         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10408         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10409         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10410         const_1 selector.
10411         (*vec_extractv4si): Rename from *sse4_1_pextrd.
10412         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10413         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10415 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10417         PR target/57108
10418         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10420 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
10422         * final.c (do_assembler_dialects): Don't handle curly braces and
10423         vertical bar escaped by % as dialect delimiters.
10424         (output_asm_insn): Print curly braces and vertical bar if escaped
10425         by % and ASSEMBLER_DIALECT defined.
10426         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10427         * doc/tm.texi: Regenerated.
10429 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10431         * config/mips/mips.c: Include tree-pass.h.
10432         (mips_reorg): Split in pre- and post-dbr_schedule parts.
10433         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10434         (pass_mips_machine_reorg2): New machine specific pass.
10435         (insert_pass_mips_machine_reorg2): New pass plugin definition.
10436         (mips_option_override): Register the new pass.
10437         * rtl.h (cleanup_barriers): Remove prototype.
10438         (dbr_schedule): Likewise.
10439         * jump.c (cleanup_barriers): Make static.
10440         * reorg.c (dbr_schedule): Likewise.
10442 2013-05-06  Richard Biener  <rguenther@suse.de>
10444         PR tree-optimization/57185
10445         * tree-parloops.c (add_field_for_reduction): Handle anonymous
10446         SSA names properly.
10448 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10450         PR target/57106
10451         * config/i386/i386.c (add_parameter_dependencies): Add dependence
10452         between "first_arg" and "insn", not "last" and "insn".
10454 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10456         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10457         (find_candidates_in_block): Re-enable slsr_process_phi.
10458         (create_phi_basis): Fix double counting of candidate adjustment.
10460 2013-05-06  Richard Biener  <rguenther@suse.de>
10462         PR middle-end/57147
10463         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10464         the edge is also fallthru, preserve it and just clear the
10465         abnormal flag.
10466         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10467         also complex, preserve that and just clear the fallthru flag.
10468         * tree-inline.c (update_ssa_across_abnormal_edges): Also
10469         update virtual operands.
10471 2013-05-06  Alan Modra  <amodra@gmail.com>
10473         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10474         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10475         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10476         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10477         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10478         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10480 2013-05-06  Alan Modra  <amodra@gmail.com>
10482         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10483         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
10484         (DEFAULT_ASM_ENDIAN): Define.
10485         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10486         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10487         Update -K PIC clause from sysv4.h.
10488         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10489         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10491 2013-05-06  Alan Modra  <amodra@gmail.com>
10493         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10494         twice for little-endian.
10495         (ashrdi3_no_power, ashrdi3): Support little-endian.
10497 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10499         PR target/55303
10500         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10501         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10502         related expanders.
10503         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10504         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10505         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10506         New predicates.
10508 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
10509             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
10511         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10512         * config/pa/pa.opt: Make mbig-switch a no-op.
10513         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10514         (CASE_VECTOR_MODE): Always return SImode.
10515         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10516         for the !TARGET_BIG_SWITCH case.
10517         * config/pa/pa-linux.h: Likewise.
10518         * config/pa/pa-openbsd.h: Likewise.
10519         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10520         * config/pa/pa.md (short_jump): Remove define_insn.
10521         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10522         (casesi0): Remove define_insn.
10523         (type): Remove btable_branch.
10524         (pa_combine_type): Likewise.
10525         (in_nullified_branch_delay): Likewise.
10526         (in_call_delay): Likewise.
10527         (define_delay): Likewise.
10528         (define_insn_reservation "Z3"): Likewise.
10529         (define_insn_reservation "Z4"): Likewise.
10530         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10531         (pa_adjust_insn_length): Remove adjustment for btable branches.
10532         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10533         and mno-big-switch
10535 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
10537         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10538         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10539         Add m->r,x alternatives.
10540         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10541         splitters using SWI48x mode iterator.
10542         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
10543         TARGET_64BIT.  Add m->x alternative.
10544         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10545         Add o->x alternative.  Enable for TARGET_SSE.
10546         (sse_storeq): Remove expander.
10547         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10548         with memory input operand.
10549         (*vec_extractv2di_1 splitter): New.
10550         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10551         * config/i386/i386.md (ssevecmodelower): New mode attribute.
10553 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10555         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
10556         (INT_LOWPART): Delete.
10557         (extract_MB): Adjust.
10558         (extract_ME): Adjust.
10559         (print_operand): Adjust.
10561 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10563         * config/rs6000/predicates.md (reg_or_add_cint_operand,
10564         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10565         (reg_or_logical_cint_operand, easy_fp_constant,
10566         logical_const_operand): Delete "CONST_DOUBLE" case.
10567         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10568         "HOST_BITS_PER_WIDE_INT == 64" test.
10569         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
10570         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10571         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10572         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10573         test.
10574         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10575         CONST_DOUBLE DImode/VOIDmode case.
10576         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10577         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10578         CONST_DOUBLE VOIDmode case.
10579         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10580         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10581         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10582         Delete CONST_DOUBLE case.
10583         (splitters for mov FMOVE64 const_double): Delete
10584         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10585         "HOST_BITS_PER_WIDE_INT >= 64" test.
10586         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10587         case.
10588         (mov DI const_double): Delete.
10590 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
10592         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10593         on op shows all bits zero in mode of a lowpart subreg, return zero.
10595 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
10597         PR target/57150
10598         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10599         to save TFmode registers and DImode to save TImode registers for
10600         caller save operations.
10601         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10602         mark being partially clobbered since they only use the first
10603         double word.
10605         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10606         and TDmode only use the upper 64-bits of each VSX register.
10608 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10610         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10611         (find_candidates_in_block): Disable slsr_process_phi.
10613 2013-05-03  Guozhi Wei  <carrot@google.com>
10615         * coverage.c (coverage_obj_init): Move the construction of gcov
10616         constructor to ...
10617         (build_init_ctor): ... here.
10619 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10621         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10622         (slsr_cand_d): Redefine def_phi.
10623         (stride_status, phi_adjust_status, count_phis_status): New enums.
10624         (find_phi_def): New.
10625         (find_basis_for_base_expr): New.
10626         (find_basis_for_candidate): Handle hidden bases.
10627         (alloc_cand_and_find_basis): Handle phi candidates.
10628         (slsr_process_phi): New.
10629         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10630         (create_mul_imm_cand): Likewise.
10631         (create_add_ssa_cand): Exclude phi base candidates.
10632         (create_add_imm_cand): Likewise.
10633         (slsr_process_cast): Likewise.
10634         (slsr_process_copy): Likewise.
10635         (find_candidates_in_block): Handle phi candidates.
10636         (dump_candidate): Likewise.
10637         (unconditional_cands): Delete.
10638         (unconditional_cands_with_known_stride_p): Delete.
10639         (phi_dependent_cand_p): New.
10640         (cand_increment): Handle phi-dependent candidates.
10641         (replace_dependent): Delete.
10642         (replace_mult_candidate): New.
10643         (replace_unconditional_candidate): New.
10644         (incr_vec_index): Move to avoid forward reference.
10645         (create_add_on_incoming_edge): New.
10646         (create_phi_basis): New.
10647         (replace_dependents): Delete.
10648         (replace_conditional_candidate): New.
10649         (phi_add_costs): New.
10650         (replace_uncond_cands_and_profitable_phis): New.
10651         (record_increment): Handle phi adjustments.
10652         (record_phi_increments): New.
10653         (record_increments): Handle phi adjustments.
10654         (phi_incr_cost): New.
10655         (lowest_cost_path): Handle phis.
10656         (total_savings): Likewise.
10657         (analyze_increments): Likewise.
10658         (ncd_with_phi): New.
10659         (ncd_of_cand_and_phis): New.
10660         (nearest_common_dominator_for_cands): Handle phi increments.
10661         (all_phi_incrs_profitable): New.
10662         (replace_profitable_candidates): Handle phi-dependent candidates.
10663         (analyze_candidates_and_replace): Likewise.
10665 2013-05-03  Teresa Johnson  <tejohnson@google.com>
10667         PR bootstrap/57154
10668         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10669         do not exceed REG_BR_PROB_BASE.
10671 2013-05-03  Jeff Law  <law@redhat.com>
10673         PR tree-optimization/57144
10674         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10675         operand of the condition will bit into the new type when eliminating
10676         a cast feeding a condition.
10678 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
10680         PR rtl-optimization/57130
10681         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10682         of COMPARE as in_code to the recursive call if needed.
10684 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
10686         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10687         (enabled): Handle new members.
10688         * config/i386/sse.md (*vec_concatv2si): Merge from
10689         *vec_concatv2si_sse2 and vec_concatv2si_sse.
10690         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10692 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
10694         PR tree-optimization/57027
10695         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10696         for fnms opportunity, check we got the prerequisite kind
10697         of tree / gimple before using accessor functions.
10699 2013-05-03  Richard Biener  <rguenther@suse.de>
10701         * double-int.h (lshift): New overload without precision
10702         and arith argument.
10703         (operator *=, operator +=, operator -=): Move ...
10704         * double-int.c (operator *=, operator +=, operator -=): ... here
10705         and implement more efficiently.
10706         (mul_double_with_sign): Remove.
10707         (lshift_double): Adjust to take unsinged shift argument, push
10708         dispatching code to callers.
10709         (mul_double_wide_with_sign): Add early out for callers that
10710         are not interested in high parts or overflow.
10711         (lshift): New function.
10712         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10713         dispatch code here.
10714         (lrotate, rrotate): Use logical shifts.
10715         * expr.c (get_inner_reference): Use lshift.
10716         * fixed-value.c (do_fixed_divide): Likewise.
10717         * tree-dfa.c (get_ref_base_and_extent): Likewise.
10718         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10719         (indirect_refs_may_alias_p): Likewise.
10720         (stmt_kills_ref_p_1): Likewise.
10722 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10724         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10726 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10728         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10729         scalar form of FABD instruction.
10731 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
10733         * lra-constraints.c (process_alt_operands): Add checking alt
10734         number to choose the best alternative.
10736 2013-05-02  Richard Biener  <rguenther@suse.de>
10738         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10739         bitmap and its handling.
10740         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10742 2013-05-02  Richard Biener  <rguenther@suse.de>
10744         PR middle-end/57140
10745         * tree-inline.c (copy_loops): Properly handle removed loops.
10746         (copy_cfg_body): Mark destination loops for fixup if source
10747         loops needed fixup.
10749 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
10751         PR target/56732
10752         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10753         generating simple_return for naked functions.
10755 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10757         PR middle-end/56988
10758         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10759         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10760         flags match.
10761         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10762         ipa_agg_replacement_value structures.
10763         (known_aggs_to_agg_replacement_list): Likewise.
10764         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10765         (read_agg_replacement_chain): Likewise.
10766         (ipcp_transform_function): Also check that by_ref flags match.
10768 2013-05-02  Richard Biener  <rguenther@suse.de>
10770         * graphds.h (struct graph): Add obstack member.
10771         * graphds.c (new_graph): Initialize obstack and allocate
10772         vertices from it.
10773         (add_edge): Allocate edge from the obstack.
10774         (free_graph): Free the obstack instead of all edges and vertices.
10776 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10778         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10779         divides.
10780         * cfg.c (update_bb_profile_for_threading): Ditto.
10781         * tree-inline.c (copy_bb): Ditto.
10782         (copy_edges_for_bb): Ditto.
10783         (initialize_cfun): Ditto.
10784         (copy_cfg_body): Ditto.
10785         (expand_call_inline): Ditto.
10786         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10787         (estimate_node_size_and_time): Ditto.
10788         (inline_merge_summary): Ditto.
10789         * cgraphclones.c (cgraph_clone_edge): Ditto.
10790         (cgraph_clone_node): Ditto.
10791         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10792         (compute_trg_info): Ditto.
10794 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10796         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10797         S reg when fp attribute set.
10798         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10800 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10802         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10803         New pattern.
10804         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10805         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10806         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10808 2013-05-02  Richard Biener  <rguenther@suse.de>
10810         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10811         (struct instantiate_cache_entry): New type.
10812         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10813         (struct instantiate_cache_type): New type.
10814         (set_instantiated_value, get_instantiated_value): Remove.
10815         (get_instantiated_value_entry): New function.
10816         (instantiate_scev_name): Use the new cache and adjust.
10817         (instantiate_scev_poly): Adjust.
10818         (instantiate_scev_binary): Likewise.
10819         (instantiate_array_ref): Likewise.
10820         (instantiate_scev_convert): Likewise.
10821         (instantiate_scev_not): Likewise.
10822         (instantiate_scev_3): Likewise.
10823         (instantiate_scev_2): Likewise.
10824         (instantiate_scev_r): Likewise.
10825         (instantiate_scev): Likewise.
10826         (resolve_mixers): Likewise.
10828 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10830         PR target/57091
10831         * lra-constraints.c (best_small_class_operands_num): Remove.
10832         (process_alt_operands): Remove small_class_operands_num.  Take
10833         small classes operands into losers and only if the operand is not
10834         matched.  Modify debugging output.
10835         (curr_insn_transform): Remove best_small_class_operands_num.
10836         Print insn name.
10838 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10840         * config/aarch64/aarch64-builtins.c
10841         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10842         * config/aarch64/aarch64-simd-builtins.def
10843         (reduc_splus_): Add new modes.
10844         (reduc_uplus_): New.
10845         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10846         (reduc_uplus_v4sf): Likewise.
10847         (reduc_splus_v4sf): Likewise.
10848         (aarch64_addv<mode>): Likewise.
10849         (reduc_uplus_<mode>): Likewise.
10850         (reduc_splus_<mode>): Likewise.
10851         (aarch64_addvv2di): Likewise.
10852         (reduc_uplus_v2di): Likewise.
10853         (reduc_splus_v2di): Likewise.
10854         (aarch64_addvv2si): Likewise.
10855         (reduc_uplus_v2si): Likewise.
10856         (reduc_splus_v2si): Likewise.
10857         (reduc_<sur>plus_<mode>): New.
10858         (reduc_<sur>plus_v2di): Likewise.
10859         (reduc_<sur>plus_v2si): Likewise.
10860         (reduc_<sur>plus_v4sf): Likewise.
10861         (aarch64_addpv4sf): Likewise.
10862         * config/aarch64/arm_neon.h
10863         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10864         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10865         add UNSPEC_SADDV, UNSPEC_UADDV.
10866         (SUADDV): New.
10867         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10869 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10871         * config/aarch64/arm_neon.h
10872         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10874 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10876         * config/aarch64/aarch64-builtins
10877         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10879 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10881         * config/aarch64/aarch64-simd-builtins.def
10882         (reduc_smax_): New.
10883         (reduc_smin_): Likewise.
10884         (reduc_umax_): Likewise.
10885         (reduc_umin_): Likewise.
10886         (reduc_smax_nan_): Likewise.
10887         (reduc_smin_nan_): Likewise.
10888         (fmax): Remove.
10889         (fmin): Likewise.
10890         (smax): Update for V2SF, V4SF and V2DF modes.
10891         (smin): Likewise.
10892         (smax_nan): New.
10893         (smin_nan): Likewise.
10894         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10895         (<su><maxmin><mode>3): ...This, refactor.
10896         (s<maxmin><mode>3): New.
10897         (<maxmin_uns><mode>3): Likewise.
10898         (reduc_<maxmin_uns>_<mode>): Refactor.
10899         (reduc_<maxmin_uns>_v4sf): Likewise.
10900         (reduc_<maxmin_uns>_v2si): Likewise.
10901         (aarch64_<fmaxmin><mode>: Remove.
10902         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10903         new builtin names.
10904         (vmin<q>_f<32,64>): Likewise.
10905         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10906         (FMAXMIN): New.
10907         (su): Add mappings for smax, smin, umax, umin.
10908         (maxmin): New.
10909         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10910         (FMAXMIN): Rename as...
10911         (FMAXMIN_UNS): ...This.
10912         (maxminv): Remove.
10913         (fmaxminv): Likewise.
10914         (fmaxmin): Likewise.
10915         (maxmin_uns): New.
10916         (maxmin_uns_op): Likewise.
10918 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10920         * config/aarch64/arm_neon.h
10921         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10922         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10923         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10925 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10927         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10928         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10930 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10932         * config/aarch64/aarch64-simd.md
10933         (vcond<mode>_internal): Handle special cases for constant masks.
10934         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10935         (vcondu<mode><mode>): Likewise.
10936         (vcond<v_cmp_result><mode>): New.
10938 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10940         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10941         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10942         * config/aarch64/aarch64-simd-builtins.def
10943         (cmeq): Update to BUILTIN_VALLDI.
10944         (cmgt): Likewise.
10945         (cmge): Likewise.
10946         (cmle): Likewise.
10947         (cmlt): Likewise.
10948         * config/aarch64/arm_neon.h
10949         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10950         to builtins or C as appropriate.
10952 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10954         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10955         (cmgeu): ...This.
10956         (cmhi): Rename to...
10957         (cmgtu): ...This.
10958         * config/aarch64/aarch64-simd.md
10959         (simd_mode): Add SF.
10960         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10961         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10962         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10963         (cstore<mode>_neg): ...This.
10964         * config/aarch64/iterators.md
10965         (VALLF): new.
10966         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10967         (COMPARISONS): New.
10968         (UCOMPARISONS): Likewise.
10969         (optab): Add missing comparisons.
10970         (n_optab): New.
10971         (cmp_1): Likewise.
10972         (cmp_2): Likewise.
10973         (CMP): Likewise.
10974         (cmp): Remove.
10975         (VCMP_S): Likewise.
10976         (VCMP_U): Likewise.
10977         (V_cmp_result): Add DF, SF modes.
10978         (v_cmp_result): Likewise.
10979         (v): Likewise.
10980         (vmtype): Likewise.
10981         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10983 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10985         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10986         define_insn to define_insn_and_split.
10987         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10988         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10989         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10990         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10991         (thumb2_negscc): Likewise.
10993 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10995         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10997 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10999         * config/arm/thumb2.md: Remove trailing whitespaces.
11001 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11003         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
11004         Use gen_int_mode rather than GEN_INT.
11006 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
11008         * value-prof.c (stream_in_histogram_value): Remove the strayed
11009         debug_gimple_stmt.
11011 2013-04-30  Richard Biener  <rguenther@suse.de>
11013         PR middle-end/57122
11014         * cfghooks.c (split_edge): Properly check for the loop latch edge.
11016 2013-04-30  Richard Biener  <rguenther@suse.de>
11018         PR middle-end/57107
11019         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11021 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
11023         PR rtl-optimization/56957
11024         PR rtl-optimization/57105
11025         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11026         variable.  Use just INSN_UID for determining whether an insn
11027         should be only disconnected from the insn stream.
11028         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11030 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
11032         PR tree-optimization/57104
11033         * tsan.c (instrument_expr): Don't instrument accesses to
11034         DECL_HARD_REGISTER VAR_DECLs.
11036 2013-04-30  Richard Biener  <rguenther@suse.de>
11038         * function.h (loops_for_fn): New inline function.
11039         (set_loops_for_fn): Likewise.
11040         * cfgloop.h (place_new_loop): Add struct function parameter.
11041         (get_loop): Likewise.
11042         (get_loops): Likewise.
11043         (number_of_loops): Likewise.
11044         (fel_next): Adjust.
11045         (fel_init): Likewise.
11046         * cfg.c (get_loop_copy): Adjust.
11047         * cfgloop.c (flow_loops_dump): Likewise.
11048         (record_loop_exits): Likewise.
11049         (verify_loop_structure): Likewise.
11050         * cfgloopanal.c (mark_irreducible_loops): Likewise.
11051         (estimate_reg_pressure_cost): Likewise.
11052         (mark_loop_exit_edges): Likewise.
11053         * cfgloopmanip.c (place_new_loop): Likewise.
11054         (add_loop): Likewise.
11055         (duplicate_loop): Likewise.
11056         * graph.c (draw_cfg_nodes): Likewise.
11057         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11058         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11059         (extract_affine_chrec): Likewise.
11060         (build_scop_iteration_domain): Likewise.
11061         * graphite.c (graphite_initialize): Likewise.
11062         * ira-build.c (create_loop_tree_nodes): Likewise.
11063         (more_one_region_p): Likewise.
11064         (rebuild_regno_allocno_maps): Likewise.
11065         (mark_loops_for_removal): Likewise.
11066         (mark_all_loops_for_removal): Likewise.
11067         (remove_unnecessary_regions): Likewise.
11068         (ira_build): Likewise.
11069         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11070         * loop-init.c (fix_loop_structure): Likewise.
11071         (gate_rtl_move_loop_invariants): Likewise.
11072         (gate_rtl_unswitch): Likewise.
11073         (gate_rtl_unroll_and_peel_loops): Likewise.
11074         (rtl_doloop): Likewise.
11075         * lto-streamer-in.c (input_cfg): Likewise.
11076         * lto-streamer-out.c (output_cfg): Likewise.
11077         * modulo-sched.c (sms_schedule): Likewise.
11078         * predict.c (tree_estimate_probability): Likewise.
11079         (tree_estimate_probability_driver): Likewise.
11080         (estimate_loops): Likewise.
11081         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11082         (move_sese_region_to_fn): Likewise.
11083         (debug_loop_num): Likewise.
11084         * tree-chrec.c (chrec_evaluate): Likewise.
11085         (hide_evolution_in_other_loops_than_loop): Likewise.
11086         (chrec_component_in_loop_num): Likewise.
11087         (reset_evolution_in_loop): Likewise.
11088         (evolution_function_is_invariant_rec_p): Likewise.
11089         * tree-if-conv.c (main_tree_if_conversion): Likewise.
11090         * tree-inline.c (copy_loops): Likewise.
11091         (copy_cfg_body): Likewise.
11092         (tree_function_versioning): Likewise.
11093         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11094         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11095         Likewise.
11096         (add_to_evolution_1): Likewise.
11097         (scev_const_prop): Likewise.
11098         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11099         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11100         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11101         (tree_ssa_lim_initialize): Likewise.
11102         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11103         (verify_loop_closed_ssa): Likewise.
11104         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
11105         (tree_ssa_loop_im): Likewise.
11106         (tree_ssa_loop_unswitch): Likewise.
11107         (tree_vectorize): Likewise.
11108         (check_data_deps): Likewise.
11109         (tree_ssa_loop_ivcanon): Likewise.
11110         (tree_ssa_loop_bounds): Likewise.
11111         (tree_complete_unroll): Likewise.
11112         (tree_complete_unroll_inner): Likewise.
11113         (tree_parallelize_loops): Likewise.
11114         (tree_ssa_loop_prefetch): Likewise.
11115         (tree_ssa_loop_ivopts): Likewise.
11116         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11117         * tree-vectorizer.c (vectorize_loops): Likewise.
11119 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
11121         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11122         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11123         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11124         with EABI_LINK_SPEC.
11126 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11128         PR target/44578
11129         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11130         alternative.
11132 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
11134         PR target/57097
11135         * lra-constraints.c (process_alt_operands): Discourage a bit more
11136         using memory for pseudos.  Print cost dump for alternatives.
11137         Modify cost values for conflicts with early clobbers.
11138         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11140 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11142         PR target/57098
11143         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11145 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11147         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11148         from/to S register.
11149         (movdi_aarch64): Support LDR/STR from/to D register.
11151 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11153         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11154         or higher by default.
11156 2013-04-29  Richard Biener  <rguenther@suse.de>
11158         PR middle-end/57075
11159         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11160         even if not adding abnormal edges for calls that can make
11161         abnormal gotos.
11163 2013-04-29  Richard Biener  <rguenther@suse.de>
11165         PR middle-end/57103
11166         * tree-cfg.c (move_stmt_op): Fix condition under which to update
11167         TREE_BLOCK.
11168         (move_stmt_r): Remove redundant checking.
11170 2013-04-29  Teresa Johnson  <tejohnson@google.com>
11172         PR bootstrap/57077
11173         * basic-block.h (apply_scale): New function.
11174         (apply_probability): Use apply_scale.
11175         * gimple-streamer-in.c (input_bb): Ditto.
11176         * lto-streamer-in.c (input_cfg): Ditto.
11177         * lto-cgraph.c (merge_profile_summaries): Ditto.
11178         * tree-optimize.c (execute_fixup_cfg): Ditto.
11179         * tree-inline.c (copy_bb): Update comment to use apply_scale.
11180         (copy_edges_for_bb): Ditto.
11181         (copy_cfg_body): Ditto.
11183 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11185         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11186         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11187         (tail_merge_optimize): Handle current_loops == NULL.
11189 2013-04-26  Jeff Law  <law@redhat.com>
11191         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11192         (simplify_cond_using_ranges): Generalize code to simplify
11193         COND_EXPRs where one argument is a constant and the other
11194         is an SSA_NAME created by an integral type conversion.
11196 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11198         * config/arm/arm.md (store_minmaxsi): Use only when
11199         optimize_insn_for_size_p.
11201 2013-04-29  Christian Bruel  <christian.bruel@st.com>
11203         PR target/57108
11204         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11206 2013-04-29  Richard Biener  <rguenther@suse.de>
11208         PR middle-end/57089
11209         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11210         loop tree make sure to schedule a fixup for the child as well.
11211         (expand_omp_for_generic): Properly add loops.
11212         (expand_omp_for_static_nochunk): Likewise.
11213         (expand_omp_for_static_chunk): Likewise.
11214         (expand_omp_for): For the degenerate case fixup loops.
11215         (expand_omp_sections): Fix default bb placement in loops.
11216         (expand_omp_atomic_pipeline): Properly add loops.
11218 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11220         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11222 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11224         * tree-ssa-tail-merge.c: Update header comment.
11226 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11228         * config/aarch64/arm_neon.h
11229         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11230         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11231         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11232         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11233         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11234         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11235         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11236         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11238 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11240         * config/aarch64/aarch64-simd.md
11241         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11242         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11243         fix_trunc, fixuns_trunc.
11244         (ftrunc<VDQF:mode>2): New.
11245         * config/aarch64/iterators.md (optab): Add fix, fixuns.
11246         (fix_trunc_optab): New.
11248 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11250         * config/aarch64/aarch64-builtins.c
11251         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11252         iceilf, lround, iroundf.
11254 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11256         PR target/54349
11257         * config/i386/i386.h (enum ix86_tune_indices)
11258         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11259         New, split from X86_TUNE_INTER_UNIT_MOVES.
11260         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11261         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11262         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11263         (TARGET_INTER_UNIT_MOVES): Remove.
11264         * config/i386/i386.c (initial_ix86_tune_features): Update.
11265         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11266         (ix86_expand_convert_uns_didf_sse): Use
11267         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11268         (ix86_expand_vector_init_one_nonzero): Ditto.
11269         (ix86_expand_vector_init_interleave): Ditto.
11270         (inline_secondary_memory_needed): Return true for moves from SSE class
11271         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11272         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11273         * config/i386/constraints.md (Yi, Ym): Depend on
11274         TARGET_INTER_UNIT_MOVES_TO_VEC.
11275         (Yj, Yn): New constraints.
11276         * config/i386/i386.md (*movdi_internal): Change constraints of
11277         operand 1 from Yi to Yj and from Ym to Yn.
11278         (*movsi_internal): Ditto.
11279         (*movdf_internal): Ditto.
11280         (*movsf_internal): Ditto.
11281         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11282         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11283         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11284         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11285         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11286         * config/i386/sse.md (movdi_to_sse): Ditto.
11287         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11288         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11289         TARGET_INTER_UNIT_MOVES.
11290         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11291         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11292         instead of TARGET_INTER_UNIT_MOVES.
11293         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11294         operand 1 from Yi to Yj and from Ym to Yn.
11296 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11298         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11299         (float_truncate_hi_): Likewise.
11300         (float_extend_lo_): Likewise.
11301         (float_truncate_lo_): Likewise.
11302         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11303         (aarch64_float_extend_lo_v2df): Likewise.
11304         (vec_unpacks_hi_v4sf): Likewise.
11305         (aarch64_float_truncate_lo_v2sf): Likewise.
11306         (aarch64_float_truncate_hi_v4sf): Likewise.
11307         (vec_pack_trunc_v2df): Likewise.
11308         (vec_pack_trunc_df): Likewise.
11310 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11312         * config/aarch64/aarch64-builtins.c
11313         (aarch64_fold_builtin): Fold float conversions.
11314         * config/aarch64/aarch64-simd-builtins.def
11315         (floatv2si, floatv4si, floatv2di): New.
11316         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11317         * config/aarch64/aarch64-simd.md
11318         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11319         * config/aarch64/iterators.md (FLOATUORS): New.
11320         (optab): Add float, floatuns.
11321         (su_optab): Likewise.
11323 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11325         * config/aarch64/aarch64-builtins.c
11326         (aarch64_builtin_vectorized_function): Use new names for
11327         fcvt builtins.
11328         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11329         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11330         (fcvtzu): Split as...
11331         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11332         (fcvtas): Split as...
11333         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11334         (fcvtau): Split as...
11335         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11336         (fcvtps): Split as...
11337         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11338         (fcvtpu): Split as...
11339         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11340         (fcvtms): Split as...
11341         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11342         (fcvtmu): Split as...
11343         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11344         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11345         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11346         (lfrintnusf, lfrintnudf): Likewise.
11347         * config/aarch64/aarch64-simd.md
11348         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11349         define_insn.
11350         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11351         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11352         (fcvt_pattern): Likewise.
11354 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11356         * config/aarch64/aarch64-simd.md
11357         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11358         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11360 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11362         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11363         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11364         (vrnd<a,m,n,p>_f32): Implement using builtins.
11365         (vrnd<i,x><q>_f<32, 64>): New.
11367 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11369         * config/aarch64/aarch64-builtins.c
11370         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11371         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11372         (frintz): Rename to...
11373         (btrunc): ...this.
11374         (frintp): Rename to...
11375         (ceil): ...this.
11376         (frintm): Rename to...
11377         (floor): ...this.
11378         (frinti): Rename to...
11379         (nearbyint): ...this.
11380         (frintx): Rename to...
11381         (rint): ...this.
11382         (frinta): Rename to...
11383         (round): ...this.
11384         * config/aarch64/aarch64-simd.md
11385         (aarch64_frint<frint_suffix><mode>): Delete.
11386         (<frint_pattern><mode>2): Convert to insn.
11387         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11388         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11389         (frint_pattern): Likewise.
11390         (frint_suffix): Likewise.
11392 2013-04-29  Richard Biener  <rguenther@suse.de>
11394         PR tree-optimization/57081
11395         * loop-init.c: Include tree-flow.h.
11396         (loop_optimizer_finalize): Free number of iteration estimates.
11397         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11399 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
11401         PR tree-optimization/57083
11402         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11403         non-singleton shift count range, zero extend low_bound for uns case.
11405         * config/i386/predicates.md (general_vector_operand): New predicate.
11406         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11407         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11408         if they aren't nonimmediate operands.  If their original values
11409         satisfy const_vector_equal_evenodd_p, don't shift them.
11410         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11411         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
11412         if not nonimmediate_operand.
11413         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11414         instead of register_operand.
11415         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11417 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
11419         * stor-layout.c (finalize_size_functions): Allocate a structure and
11420         reset cfun before dumping the functions.
11422 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
11424         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11426         PR target/56866
11427         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11428         use xop_pmacsdqh if uns_p.
11429         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11430         the immediate rotate count.
11432 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
11434         * rtl.h (struct rtx_def): Add comment for field jump.
11435         (LRA_SUBREG_P): New macro.
11436         * recog.c (register_operand): Check LRA_SUBREG_P.
11437         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11438         stack frame.
11439         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11440         (lra_final_code_change): Skip subreg change for operators.
11441         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11442         if there are no operand changes.
11443         * lra-constraints.c (curr_insn_set): New.
11444         (match_reload): Set LRA_SUBREG_P.
11445         (emit_spill_move): Ditto.
11446         (check_and_process_move): Use curr_insn_set. Process only single
11447         set insns.  Don't initialize sec_mem_p and change_p.
11448         (simplify_operand_subreg): Use LRA_SUBREG_P.
11449         (reg_in_class_p): New function.
11450         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11451         of #ifdef.  Add code to remove cycling.
11452         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11453         non-null disp.  Reload inner instead of disp when base and index
11454         are null.  Try to put lo_sum into register.
11455         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11456         (check_and_process_move): Move code for move cost check to
11457         simple_move_p.  Remove equiv_substitution.
11458         (simple_move_p): New function.
11459         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11460         curr_insn_set.  Call check_and_process_move only for single set
11461         insns.  Use the new function.  Move call of check_and_process_move
11462         after operand equiv substitution and address process.
11464 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11466         PR go/57045
11467         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11468         with nonlocal goto receivers or returns twice calls, ignore
11469         unininitialized values from abnormal edges to nl goto receiver
11470         or returns twice call.
11472 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11474         PR tree-optimization/57051
11475         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11476         and VEC_RSHIFT_EXPR if shift count is a multiple of element
11477         bitsize.
11479 2013-04-26  Richard Biener  <rguenther@suse.de>
11481         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11482         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
11483         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11484         (fixup_loop_arrays_after_move): New function.
11485         (move_sese_region_to_fn): Properly outline the loop tree parts
11486         of the SESE region.
11488 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
11490         * config/i386/i386.md (type, unit): Fix long lines.
11492 2013-04-26  Richard Biener  <rguenther@suse.de>
11494         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11495         (lto-streamer-out.o): Likewise.
11496         * cfgloop.c (init_loops_structure): Export, add struct function
11497         argument and adjust.
11498         (flow_loops_find): Adjust.
11499         * cfgloop.h (enum loop_estimation): Add EST_LAST.
11500         (init_loops_structure): Declare.
11501         * lto-streamer-in.c: Include cfgloop.h.
11502         (input_cfg): Input the loop tree.
11503         * lto-streamer-out.c: Include cfgloop.h.
11504         (output_cfg): Output the loop tree.
11505         (output_struct_function_base): Do not drop PROP_loops.
11507 2013-03-26  Richard Biener  <rguenther@suse.de>
11509         * tree-cfg.c (execute_build_cfg): Build the loop tree.
11510         (pass_build_cfg): Provide PROP_loops.
11511         (move_sese_region_to_fn): Remove loops that are outlined into fn
11512         for now.
11513         * tree-inline.c: Include cfgloop.h.
11514         (initialize_cfun): Do not drop PROP_loops.
11515         (copy_loops): New function.
11516         (copy_cfg_body): Copy loop structure.
11517         (tree_function_versioning): Initialize destination loop tree.
11518         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11519         (pass_parallelize_loops): Do IL verification.
11520         * loop-init.c (loop_optimizer_init): Fixup loops if required.
11521         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11522         the CFG make sure we fixup loops as well.
11523         * tree-ssa-tail-merge.c: Include cfgloop.h.
11524         (replace_block_by): When merging loop latches mark loops for fixup.
11525         * lto-streamer-out.c (output_struct_function_base): Drop
11526         PROP_loops for now.
11527         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11528         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11529         * ipa-split.c: Include cfgloop.h.
11530         (split_function): Add the new return block to the loop tree root.
11531         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11532         whether we have removed the forwarder block.
11533         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11534         * cfgloop.h (place_new_loop): Declare.
11535         * cfgloopmanip.c (place_new_loop): Export.
11536         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11537         (tree-switch-conversion.o): Likewise.
11538         (tree-complex.o): Likewise.
11539         (tree-inline.o): Likewise.
11540         (tree-ssa-tailmerge.o): Likewise.
11541         (ipa-split.o): Likewise.
11542         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11543         (tree-ssa-copy.o): Likewise.
11544         * tree-switch-conversion.c: Include cfgloop.h
11545         (process_switch): If we emit a bit-test cascade, schedule loops
11546         for fixup.
11547         * tree-complex.c: Include cfgloop.h.
11548         (expand_complex_div_wide): Properly add new basic-blocks to loops.
11549         * asan.c: Include cfgloop.h.
11550         (create_cond_insert_point): Properly add new basic-blocks to
11551         loops, schedule loop fixup.
11552         * cfgloop.c (verify_loop_structure): Check that looks are not
11553         marked for fixup.
11554         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11555         to loops.
11556         (expand_omp_for_generic): Likewise.
11557         (expand_omp_sections): Likewise.
11558         (expand_omp_atomic_pipeline): Schedule loops for fixup.
11559         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11560         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11561         is initialized, not when loops are present.
11562         * tree-parloops.c (parallelize_loops): Remove checking here.
11563         * passes.c (init_optimization_passes): Schedule a copy-propagation
11564         pass before complete unrolling of inner loops.
11566 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11568         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11569         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11570         (colorize_init): Add argument to _WIN32 version.
11571         * toplev.c: Include diagnostic-color.h.
11572         (process_options): Default to -fdiagnostics-color=auto if
11573         GCC_COLORS env var is in the environment.
11574         * common.opt (fdiagnostics-color=): Add Var and Init.
11575         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11576         env var is in the environment, the default is auto rather than never.
11578         * diagnostic.h (file_name_as_prefix): Add context argument.
11579         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
11580         the string as locus.
11581         * langhooks.c (lhd_print_error_function): Adjust caller.
11583 2013-04-25  Lawrence Crowl  <crowl@google.com>
11585         * var-tracking.c (shared_hash_def::htab):
11586         Change type to hash_table.  Update dependent calls and types.
11588 2013-04-25  Lawrence Crowl  <crowl@google.com>
11590         * Makefile.in: Update as needed below.
11592         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11593         Move declaration to after the type's method definitons.
11595         * attribs.c (htab_t scoped_attributes::attribute_hash):
11596         Change type to hash_table.  Update dependent calls and types.
11598         * bitmap.c (htab_t bitmap_desc_hash):
11599         Change type to hash_table.  Update dependent calls and types.
11601         * cselib.c (htab_t cselib_hash_table):
11602         Change type to hash_table.  Update dependent calls and types.
11604         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11605         (hash_string_slot_node): Move implementation into lto-streamer.h
11606         struct string_slot_hasher.
11607         (eq_string_slot_node): Likewise.
11609         * data-streamer-out.c: Update output_block::string_hash_table
11610         dependent calls and types.
11612         * dwarf2cfi.c (htab_t trace_index):
11613         Change type to hash_table.  Update dependent calls and types.
11615         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11616         Change type to hash_table.  Update dependent calls and types.
11617         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11618         (htab_t optimize_external_refs::map): Likewise.
11619         (htab_t output_comp_unit::extern_map): Likewise.
11620         (htab_t output_comdat_type_unit::extern_map): Likewise.
11621         (htab_t output_macinfo::macinfo_htab): Likewise.
11622         (htab_t optimize_location_lists::htab): Likewise.
11623         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11625         * except.c (htab_t ehspec_hash_type):
11626         Change type to hash_table.  Update dependent calls and types.
11627         (assign_filter_values::ttypes): Likewise.
11628         (assign_filter_values::ehspec): Likewise.
11629         (sjlj_assign_call_site_values::ar_hash): Likewise.
11630         (convert_to_eh_region_ranges::ar_hash): Likewise.
11632         * gcse.c (htab_t pre_ldst_table):
11633         Change type to hash_table.  Update dependent calls and types.
11635         * ggc-common.c (htab_t saving_htab):
11636         Change type to hash_table.  Update dependent calls and types.
11637         (htab_t loc_hash): Likewise.
11638         (htab_t ptr_hash): Likewise.
11639         (call_count): Rename ggc_call_count.
11640         (call_alloc): Rename ggc_call_alloc.
11641         (loc_descriptor): Rename make_loc_descriptor.
11642         (add_statistics): Rename ggc_add_statistics.
11644         * ggc-common.c (saving_htab):
11645         Change type to hash_table.  Update dependent calls and types.
11647         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11648         (push_gimplify_context): Likewise.
11649         (pop_gimplify_context): Likewise.
11650         (struct gimple_temp_hash_elt): Added.
11651         (struct gimplify_hasher): Likewise.
11652         (struct gimplify_ctx.temp_htab):
11653         Change type to hash_table.  Update dependent calls and types.
11655         * gimple-fold.c: Include gimplify-ctx.h.
11657         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11658         Change type to hash_table.  Update dependent calls and types.
11659         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11660         avoid potential global name collision.
11662         * gimplify.c: Include gimplify-ctx.h.
11663         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11664         (htab_t gimplify_ctx::temp_htab):
11665         Update dependent calls and types for new type hash_table.
11666         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11667         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11669         * gimplify-ctx.h: New.
11670         (struct gimple_temp_hash_elt): Move from gimplify.c.
11671         (class gimplify_hasher): New.
11672         (struct gimplify_ctx): Move from gimple.h.
11673         (htab_t gimplify_ctx::temp_htab):
11674         Change type to hash_table.  Update dependent calls and types.
11676         * graphite-clast-to-gimple.c: Include graphite-htab.h.
11677         (htab_t ivs_params::newivs_index):
11678         Change type to hash_table.  Update dependent calls and types.
11679         (htab_t ivs_params::params_index): Likewise.
11680         (htab_t print_generated_program::params_index): Likewise.
11681         (htab_t gloog::newivs_index): Likewise.
11682         (htab_t gloog::params_index): Likewise.
11684         * graphite.c: Include graphite-htab.h.
11685         4htab_t graphite_transform_loops::bb_pbb_mapping):
11686         Change type to hash_table.  Update dependent calls and types.
11688         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11689         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11690         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11692         * graphite-dependences.c: Include graphite-htab.h.
11693         (loop_is_parallel_p): Change hash table type of parameter.
11695         * graphite-htab.h: New.
11696         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11697         (extern find_pbb_via_hash): Move from graphite-poly.h.
11698         (extern loop_is_parallel_p): Move from graphite-poly.h.
11699         (extern get_loop_body_pbbs): Move from graphite-poly.h.
11701         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11702         (extern loop_is_parallel_p): Move to graphite-htab.h.
11703         (extern get_loop_body_pbbs): Move to graphite-htab.h.
11705         * haifa-sched.c (htab_t delay_htab):
11706         Change type to hash_table.  Update dependent calls and types.
11707         (htab_t delay_htab_i2): Likewise.
11709         * ira-color.c (htab_t allocno_hard_regs_htab):
11710         Change type to hash_table.  Update dependent calls and types.
11712         * ira-costs.c (htab_t cost_classes_htab):
11713         Change type to hash_table.  Update dependent calls and types.
11715         * loop-invariant.c (htab_t merge_identical_invariants::eq):
11716         Change type to hash_table.  Update dependent calls and types.
11718         * loop-iv.c (htab_t bivs):
11719         Change type to hash_table.  Update dependent calls and types.
11721         * loop-unroll.c (htab_t opt_info::insns_to_split):
11722         Change type to hash_table.  Update dependent calls and types.
11723         (htab_t opt_info::insns_with_var_to_expand): Likewise.
11725         * lto-streamer.h (struct string_slot): Move from data-streamer.h
11726         (struct string_slot_hasher): New.
11727         (htab_t output_block::string_hash_table):
11728         Change type to hash_table.  Update dependent calls and types.
11730         * lto-streamer-in.c (freeing_string_slot_hasher): New.
11731         (htab_t file_name_hash_table):
11732         Change type to hash_table.  Update dependent calls and types.
11734         * lto-streamer-out.c: Update output_block::string_hash_table dependent
11735         calls and types.
11737         * lto-streamer.c (htab_t tree_htab):
11738         Change type to hash_table.  Update dependent calls and types.
11740         * omp-low.c: Include gimplify-ctx.h.
11742         * passes.c (htab_t name_to_pass_map):
11743         Change type to hash_table.  Update dependent calls and types.
11744         (pass_traverse): Rename to passes_pass_traverse.
11746         * plugin.c (htab_t event_tab):
11747         Change type to hash_table.  Update dependent calls and types.
11749         * postreload-gcse.c (htab_t expr_table):
11750         Change type to hash_table.  Update dependent calls and types.
11751         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11753         * sese.c (debug_rename_map_1): Make extern.
11754         (htab_t copy_bb_and_scalar_dependences::rename_map):
11755         Change type to hash_table.  Update dependent calls and types.
11757         * sese.h (extern debug_rename_map): Move to .c file.
11759         * store-motion.c (htab_t store_motion_mems_table):
11760         Change type to hash_table.  Update dependent calls and types.
11762         * trans-mem.c (htab_t tm_new_mem_hash):
11763         Change type to hash_table.  Update dependent calls and types.
11765         * tree-browser.c (htab_t TB_up_ht):
11766         Change type to hash_table.  Update dependent calls and types.
11768         * tree-cfg.c (htab_t discriminator_per_locus):
11769         Change type to hash_table.  Update dependent calls and types.
11771         * tree-complex.c: Include tree-hasher.h
11772         (htab_t complex_variable_components):
11773         Change type to hash_table.  Update dependent calls and types.
11775         * tree-eh.c (htab_t finally_tree):
11776         Change type to hash_table.  Update dependent calls and types.
11778         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11779         struct int_tree_hasher.
11780         (extern int_tree_map_eq): Likewise.
11781         (uid_decl_map_hash): Removed.
11782         (extern decl_tree_map_eq): Likewise.
11784         * tree-hasher.h: New.
11785         (struct int_tree_hasher): New.
11786         (typedef int_tree_htab_type): New.
11788         * tree-inline.c: Include gimplify-ctx.h.
11790         * tree-mudflap.c: Include gimplify-ctx.h.
11792         * tree-parloops.c: Include tree-hasher.h.
11793         (htab_t eliminate_local_variables_stmt::decl_address):
11794         Change type to hash_table.  Update dependent calls and types.
11795         (htab_t separate_decls_in_region::decl_copies): Likewise.
11797         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11798         Change type to hash_table.  Update dependent calls and types.
11800         * tree-sra.c (candidates):
11801         Change type to hash_table.  Update dependent calls and types.
11803         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11804         in tree-flow.h.
11805         (int_tree_map_hash): Likewise.
11807         * tree-ssa-dom.c (htab_t avail_exprs):
11808         Change type to hash_table.  Update dependent calls and types.
11810         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11811         Change type to hash_table.  Update dependent calls and types.
11813         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11814         Change type to hash_table.  Update dependent calls and types.
11816         * tree-ssa-phiopt.c (seen_ssa_names):
11817         Change type to hash_table.  Update dependent calls and types.
11819         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11820         Change type to hash_table.  Update dependent calls and types.
11822         * tree-ssa-uncprop.c (equiv):
11823         Change type to hash_table.  Update dependent calls and types.
11825 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11827         PR rtl-optimization/57003
11828         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11829         call note_stores with kill_clobbered_value callback again after
11830         killing regs_invalidated_by_call.
11832 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11834         * config/aarch64/aarch64-simd.md
11835         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11836         (aarch64_simd_bsl<mode>): Likewise.
11837         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11839 2013-04-25  Marek Polacek  <polacek@redhat.com>
11841         PR tree-optimization/57066
11842         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11844 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11846         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11848 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11850         * config/aarch64/aarch64-builtins.c
11851         (aarch64_fold_builtin): New.
11852         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11853         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11854         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11855         * config/aarch64/arm_neon.h
11856         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11858 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11859             Tejas Belagod  <tejas.belagod@arm.com>
11861         * config/aarch64/aarch64-builtins.c
11862         (aarch64_gimple_fold_builtin): New.
11863         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11864         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11865         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11866         (addvv4sf): Update.
11867         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11869 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11871         * config/aarch64/aarch64.md
11872         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11874 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11876         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11877         (*ngcsi_uxtw): New pattern.
11879 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11880             Julian Brown  <julian@codesourcery.com>
11882         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11883         (TB_DREG): Add T_V4HF.
11884         (v4hf_UP): New macro.
11885         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11886         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11887         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11888         built-ins.
11889         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11890         (arm_vector_mode_supported_p): Handle V4HF.
11891         (arm_mangle_map): Handle V4HFmode.
11892         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11893         * config/arm/arm_neon_builtins.def: Add entries for
11894         vcvtv4hfv4sf, vcvtv4sfv4hf.
11895         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11896         (neon_vcvtv4hfv4sf): Likewise.
11897         * config/arm/neon-gen.ml: Handle half-precision floating point
11898         features.
11899         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11900         * config/arm/arm_neon.h: Regenerate.
11901         * config/arm/neon.ml (type elts): Add F16.
11902         (type vectype): Add T_float16x4, T_floatHF.
11903         (type vecmode): Add V4HF.
11904         (type features): Add Requires_FP_bit feature.
11905         (elt_width): Handle F16.
11906         (elt_class): Likewise.
11907         (elt_of_class_width): Likewise.
11908         (mode_of_elt): Refactor.
11909         (type_for_elt): Handle F16, fix error messages.
11910         (vectype_size): Handle T_float16x4.
11911         (vcvt_sh): New function.
11912         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11913         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11914         (string_of_mode): Handle V4HF.
11915         * doc/arm-neon-intrinsics.texi: Regenerate.
11917 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11919         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11920         format specifier in 'X' case.
11922 2013-04-25  Alan Modra  <amodra@gmail.com>
11924         PR target/57052
11925         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11926         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11927         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11928         Repeat for many other rotate/shift and mask patterns using subregs.
11929         Name lshiftrt insns.
11930         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11931         on WORDS_BIG_ENDIAN.
11933 2013-04-25  Alan Modra  <amodra@gmail.com>
11935         * config.gcc: Support little-endian powerpc-linux targets.
11936         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11937         (LINK_OS_LINUX_SPEC): Define.
11938         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11939         Preserve MASK_LITTLE_ENDIAN.
11940         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11941         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11942         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11943         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11944         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11945         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11946         Correct fp word order for little-endian.  Don't shift toc entries
11947         smaller than a word for little-endian.
11948         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11949         (bswapdi2 splits): Correct low-part subreg for little-endian.
11950         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11951         low/high where such is correct only for be.
11952         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11953         little-endian for -mcall-aixdesc.
11955 2013-04-25  Alan Modra  <amodra@gmail.com>
11957         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11958         replace_equiv_address_nv.
11960 2013-04-25  Alan Modra  <amodra@gmail.com>
11962         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11964 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11966         Revert:
11967         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11968         * rtl.h (struct rtx_def): ...
11970 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11972         PR rtl-optimizations/57046
11973         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11974         for multi-reg splits.
11976 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11978         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11980 2013-04-24  Sterling Augustine  <saugustine@google.com>
11982         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11983         (comp_dir_string, debug_str_dwo_section): New.
11984         (DEBUG_STR_DWO_SECTION): Rename to ...
11985         (DEBUG_DWO_STR_SECTION): ... this.
11986         (DEBUG_NORM_STR_SECTION): Delete.
11987         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11988         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11989         (find_AT_string): Move most logic to ...
11990         (find_AT_string_in_table): ... here.  New.
11991         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11992         add_skeleton_AT_string.  Delete logic.
11993         (output_skeleton_debug_sections): Remove call to
11994         add_top_level_skeleton_die_attrs.
11995         (add_comp_dir_attribute): Move logic to comp_dir_string.
11996         (dwarf2out_init): Initialize debug_str_dwo_section.
11997         (output_indirect_string): Call find_string_form.
11998         (output_indirect_strings): Rewrite.
11999         (prune_unused_types): Empty skeleton_debug_str_hash.
12000         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
12001         (dwarf2out_finish):  Call output_indirect_strings.
12003 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12005         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
12007 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12009         * rtl.h (struct rtx_def): Add comment for field jump.
12010         (LRA_SUBREG_P): New macro.
12011         * recog.c (register_operand): Check LRA_SUBREG_P.
12012         * lra.c (lra): Add note at the end of RTL code. Align non-empty
12013         stack frame.
12014         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12015         (lra_final_code_change): Skip subreg change for operators.
12016         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12017         if there are no operand changes.
12018         * lra-constraints.c (curr_insn_set): New.
12019         (match_reload): Set LRA_SUBREG_P.
12020         (emit_spill_move): Ditto.
12021         (check_and_process_move): Use curr_insn_set. Process only single
12022         set insns.  Don't initialize sec_mem_p and change_p.
12023         (simplify_operand_subreg): Use LRA_SUBREG_P.
12024         (reg_in_class_p): New function.
12025         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
12026         of #ifdef.  Add code to remove cycling.
12027         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12028         non-null disp.  Reload inner instead of disp when base and index
12029         are null.  Try to put lo_sum into register.
12030         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12031         (check_and_process_move): Move code for move cost check to
12032         simple_move_p.  Remove equiv_substitution.
12033         (simple_move_p): New function.
12034         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
12035         curr_insn_set.  Call check_and_process_move only for single set
12036         insns.  Use the new function.  Move call of check_and_process_move
12037         after operand equiv substitution and address process.
12039 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
12041         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12042         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12043         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12045 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12047         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12049 2013-04-24  Marek Polacek  <polacek@redhat.com>
12051         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12052         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12053         (select_loops_exit_conditions): Likewise.
12054         (number_of_iterations_for_all_loops): Likewise.
12055         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12056         (scev_analysis): Likewise.
12058 2013-04-02  Catherine Moore  <clm@codesourcery.com>
12059             Chao-ying Fu  <fu@mips.com>
12061         * config/mips/micromips.md (jraddiusp): New pattern.
12062         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12063         instruction if possible.
12065 2013-04-24  Alan Modra  <amodra@gmail.com>
12067         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12069 2013-04-24  Julian Brown  <julian@codesourcery.com>
12070             Chung-Lin Tang  <cltang@codesourcery.com>
12072         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12073         dependency behavior in enumeration type DIE generation. Add TODO note
12074         to comments about future DW_FORM_sdata/udata re-work of related code.
12076 2013-04-23  Lawrence Crowl  <crowl@google.com>
12078         * Makefile.in: Update as needed below.
12080         * hash-table.h (class hash_table):
12081         Correct many methods with parameter types compare_type to the correct
12082         value_type.  (Correct code was unlikely to notice the change.)
12083         (hash_table::elements_with_deleted) New.
12084         (class hashtable::iterator): New.
12085         (hashtable::begin()): New.
12086         (hashtable::end()): New.
12087         (FOR_EACH_HASH_TABLE_ELEMENT): New.
12089         * statistics.c (statistics_hashes):
12090         Change type to hash_table.  Update dependent calls and types.
12092         * tree-into-ssa.c (var_infos):
12093         Change type to hash_table.  Update dependent calls and types.
12095         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12096         Change type to hash_table.  Update dependent calls and types.
12098         * tree-ssa-loop-im.c (struct mem_ref.refs):
12099         Change type to hash_table.  Update dependent calls and types.
12101         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12102         Change type to hash_table.  Update dependent calls and types.
12104         * tree-ssa-sccvn.c (vn_tables_s::nary):
12105         Change type to hash_table.  Update dependent calls and types.
12106         (vn_tables_s::phis): Likewise.
12107         (vn_tables_s::references): Likewise.
12109         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12110         (vn_reference_eq): Update parameter and return types.
12112         * tree-ssa-structalias.c (pointer_equiv_class_table):
12113         Change type to hash_table.  Update dependent calls and types.
12114         (location_equiv_class_table): Likewise.
12116         * tree-vect-data-refs.c: Consequential changes for making
12117         peeling a hash_table.
12119         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12120         (destroy_loop_vec_info): Dependent hash_table update.
12122         * tree-vectorizer.h (peeling_htab):
12123         Change type to hash_table.  Update dependent calls and types.
12125 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
12127         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12128         to check the register content is equal or not.
12129         * lra-constraints.c (match_reload): Use lra_assign_reg_val
12130         to assign register content record.
12131         * lra-eliminations.c (update_reg_eliminate): Use
12132         lra_update_reg_val_offset to update register content offset.
12133         * lra-int.h (struct lra_reg): Add offset member.
12134         (lra_reg_val_equal_p): New static inline function.
12135         (lra_update_reg_val_offset): New static inline function.
12136         (lra_assign_reg_val): New static inline function.
12137         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12138         to assign register content record.
12139         (initialize_lra_reg_info_element): Initial offset to zero.
12141 2013-04-23  Catherine Moore  <clm@codesourcery.com>
12143         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12144         operands.  Record compression.
12146 2013-04-23  Xinliang David Li  <davidxl@google.com>
12148         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12150 2013-04-23  Richard Biener  <rguenther@suse.de>
12152         PR middle-end/57036
12153         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12154         parameter, only add abnormal goto edges from the copied body
12155         if the call could perform abnormal gotos.
12156         (copy_cfg_body): Adjust.
12158 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
12160         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12162 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
12164         * coretypes.h (gimple_stmt_iterator): Add struct to make
12165         compatible with C.
12167 2013-04-23  Richard Biener  <rguenther@suse.de>
12169         PR tree-optimization/57026
12170         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12171         from SSA names occuring in abnormal PHI nodes.
12173 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12175         * lto/lto.c (print_lto_report_1): Fix LTO report names.
12177 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12179         * lto/lto.c (print_lto_report_1): Declare early.
12180         (read_cgraph_and_symbols): Call print_lto_report_1 early.
12182 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12184         * common.opt (-flto-report-wpa): Add.
12185         * doc/invoke.texi (-flto-report-wpa): Add.
12186         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12187         (lto_main): dito.
12189 2013-04-22  Xinliang David Li  <davidxl@google.com>
12191         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12192         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12193         * Makefile.in: New dependency
12195             David Daney <ddaney.cavm@gmail.com>
12197         * configure.ac (gcc_cv_as_micromips_support):  Use the
12198         --fatal-warnings option.
12199         * configure: Regenerate.
12201 2013-04-22  Marek Polacek  <polacek@redhat.com>
12203         PR sanitizer/56990
12204         * tsan.c (instrument_expr): Don't instrument expression
12205         in case its size is zero.
12207 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
12209         PR target/57032
12210         Revert:
12211         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
12213         * config/alpha/alpha.c (TARGET_LRA_P): New define.
12215 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12217         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12218         (gimple_stmt_iterator): New typedef.
12219         * gimple.h (gimple_stmt_iterator): Rename to...
12220         (gimple_stmt_iterator_d): ... This.
12221         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12222         trees be valid for GIMPLE and GENERIC.
12223         (TARGET_GIMPLE_FOLD_BUILTIN): New.
12224         * gimple-fold.c (gimple_fold_call): Call target hook
12225         gimple_fold_builtin.
12226         * hooks.c (hook_bool_gsiptr_false): New.
12227         * hooks.h (hook_bool_gsiptr_false): New.
12228         * target.def (fold_stmt): New.
12229         * doc/tm.texi: Regenerate.
12231 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
12233         PR target/57018
12234         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12235         a set sp if no stack realignment.
12237 2013-04-22  Nick Clifton  <nickc@redhat.com>
12239         * config.gcc (tilegx-linux): Extend extra_objs rather than
12240         overwriting it.
12241         (tilepro-linux): Likewise.
12243 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12245         * config/aarch64/aarch64-builtins.c
12246         (CF): Remove.
12247         (CF0, CF1, CF2, CF3, CF4, CF10): New.
12248         (VAR<1-12>): Add MAP parameter.
12249         (BUILTIN_*): Likewise.
12250         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12251         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12252         (aarch64_ushl_n<mode>): Likewise.
12253         (aarch64_sshr_n<mode>): Likewise.
12254         (aarch64_ushr_n<mode>): Likewise.
12255         (aarch64_<maxmin><mode>): Likewise.
12256         (aarch64_sqrt<mode>): Likewise.
12257         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12258         (vshr<q>_n_*): Likewise.
12260 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12262         * config/aarch64/aarch64-builtins.c
12263         (aarch64_simd_builtin_type_mode): Handle SF types.
12264         (sf_UP): Define.
12265         (BUILTIN_GPF): Define.
12266         (aarch64_init_simd_builtins): Handle SF types.
12267         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12268         (frecps): Likewise.
12269         (frecpx): Likewise.
12270         * config/aarch64/aarch64-simd.md
12271         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12272         (aarch64_frecpe<mode>): New.
12273         (aarch64_frecps<mode>): Likewise.
12274         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12275         (v8type): Add frecp<esx>.
12276         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12277         (aarch64_frecps<mode>): Likewise.
12278         * config/aarch64/iterators.md (FRECP): New.
12279         (frecp_suffix): Likewise.
12280         * config/aarch64/arm_neon.h
12281         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12283 2013-04-22  Christian Bruel  <christian.bruel@st.com>
12285         PR target/56995
12286         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12287         (REG_CLASS_NAMES): Idem.
12288         (REG_CLASS_CONTENTS): Idem.
12289         (REGCLASS_HAS_FP_REG): Idem.
12290         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12291         (sh_conditional_register_usage): Idem.
12293 2013-04-21  Jeff Law  <law@redhat.com>
12295         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12296         (ssa_forward_propagate_and_combine): Use it.
12298 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12300         * lra.c: Update the flow chart diagram.
12302 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12304         PR rtl-optimization/56847
12305         * lra-constraints.c (process_alt_operands): Discourage alternative
12306         with non-matche doffsettable memory constraint fro memory with
12307         known offset.
12309 2013-04-19  Richard Biener  <rguenther@suse.de>
12311         PR tree-optimization/56982
12312         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12313         function.
12314         * gimplify.c (gimplify_call_expr): Notice special calls.
12315         (gimplify_modify_expr): Likewise.
12316         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12317         abnormal control flow receivers.
12318         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12319         in the same way as cfun->has_nonlocal_labels.
12320         (gimple_purge_dead_abnormal_call_edges): Likewise.
12321         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12322         receivers start a basic-block.
12324 2013-04-19  Richard Biener  <rguenther@suse.de>
12326         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12327         member ...
12328         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
12329         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12330         (SLP_TREE_LOAD_PERMUTATION): Add.
12331         (vect_transform_slp_perm_load): Adjust prototype.
12332         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12333         (vect_free_slp_instance): Likewise.
12334         (vect_create_new_slp_node): Likewise.
12335         (vect_supported_slp_permutation_p): Remove.
12336         (vect_slp_rearrange_stmts): Adjust.
12337         (vect_supported_load_permutation_p): Likewise.  Inline
12338         vect_supported_slp_permutation_p here.
12339         (vect_analyze_slp_instance): Compute load permutations per
12340         slp node instead of per instance.
12341         (vect_get_slp_defs): Adjust.
12342         (vect_transform_slp_perm_load): Likewise.
12343         (vect_schedule_slp_instance): Remove redundant code.
12344         (vect_schedule_slp): Remove hack for PR56270, add it ...
12345         * tree-vect-stmts.c (vectorizable_load): ... here, do not
12346         CSE loads for SLP.  Adjust.
12348 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12350         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12351         spelling in two comments.
12353 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12355         PR target/56797
12356         * config/arm/arm.c (load_multiple_sequence): Require SP
12357         as base register for loads if SP is in the register list.
12359 2013-04-19  Martin Jambor  <mjambor@suse.cz>
12361         PR tree-optimization/56718
12362         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12363         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12364         and made public.  Adjusted all callers.
12365         (ipa_intraprocedural_devirtualization): New function.
12366         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12367         (ipa_intraprocedural_devirtualization): Likewise.
12368         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12370 2013-04-19  Richard Biener  <rguenther@suse.de>
12372         PR tree-optimization/57000
12373         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12375 2013-04-19  Terry Guo  <terry.guo@arm.com>
12377         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12378         Replace with ...
12379         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
12380         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12381         (cortex_m4_fmacs): Use new reservations.
12382         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12384 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
12386         PR rtl-optimization/56999
12387         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12388         related code.
12389         (lra_coalesce): Remove split_origin_bitmap and related code.
12390         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
12391         ranges if necessary.
12393 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
12395         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12396         New array.
12397         (ix86_expand_call): Remove clobbered_registers array and use
12398         x86_64_ms_sysv_extra_clobbered_registers instead.
12399         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12400         Declare here.
12401         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12402         predicate.
12403         * config/i386/i386.md (*call_rex64_ms_sysv): Use
12404         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
12405         (*call_value_rex64_ms_sysv): Ditto.
12407 2013-04-18  Cary Coutant  <ccoutant@google.com>
12409         * dwarf2out.c (output_pubnames): Check die_perennial_p of
12410         parent instead of die_mark.
12412 2013-04-18  Diego Novillo  <dnovillo@google.com>
12414         * gimple.c (create_gimple_tmp): New.
12415         (get_expr_type): New.
12416         (build_assign): New.
12417         (build_type_cast): New.
12418         * gimple.h (enum ssa_mode): Define.
12419         (gimple_seq_set_location): New.
12420         * asan.c (build_check_stmt): Change some gimple_build_* calls
12421         to use build_assign and build_type_cast.
12423 2013-04-18  Richard Biener  <rguenther@suse.de>
12425         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12426         handle negative step.  Remove redundant checks.
12427         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12428         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12429         for negative step and grouped loads fail to vectorize.
12431 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12433         * emit-rtl.c (reset_insn_used_flags): New function.
12434         (reset_all_used_flags): Use it.
12435         (verify_insn_sharing): New function.
12436         (verify_rtl_sharing): Fix verification for SEQUENCEs.
12438 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
12440         PR tree-optimization/56984
12441         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12442         and (x >> M) >= N don't register any assertion if N << M is the
12443         minimum value.
12445 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12447         * lower-subreg.c (resolve_simple_move): If called self-recursive,
12448         do not delete_insn insns that have not yet been emitted, only
12449         unlink them with remove_insn.
12450         * df-scan.c (df_insn_delete): Revert r197492.
12452 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
12454         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12455         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12457 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12459         * config/arm/arm.md (movsicc_insn): Convert define_insn into
12460         define_insn_and_split.
12461         (and_scc,ior_scc,negscc): Likewise.
12462         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12464 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12466         * config/arm/arm.c (use_return_insn): Return 0 for targets that
12467         can benefit from using a sequence of LDRD instructions in epilogue
12468         instead of a single LDM instruction.
12470 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12472         PR 45688
12473         * doc/extend.texi: Fix typo.
12475 2013-04-17  Richard Biener  <rguenther@suse.de>
12477         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12478         (vect_build_slp_tree): ... here.
12479         (vect_build_slp_tree_1): Compute which stmts of the SLP group
12480         match.  Remove special-casing of mismatched complex loads.
12481         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12482         re-try the match with swapped commutative operands.
12483         (vect_supported_load_permutation_p): Remove special-casing of
12484         mismatched complex loads.
12485         (vect_analyze_slp_instance): Adjust.
12487 2013-04-17  Richard Biener  <rguenther@suse.de>
12489         PR rtl-optimization/56921
12490         * cfgloop.h (struct loop): Add simple_loop_desc member.
12491         (struct niter_desc): Mark with GTY(()).
12492         (simple_loop_desc): Do not use aux field but simple_loop_desc.
12493         * loop-iv.c (get_simple_loop_desc): Likewise.
12494         (free_simple_loop_desc): Likewise.
12496         Revert
12497         2013-04-16  Richard Biener  <rguenther@suse.de>
12499         PR rtl-optimization/56921
12500         * loop-init.c (pass_rtl_move_loop_invariants): Add
12501         TODO_do_not_ggc_collect to todo_flags_finish.
12502         (pass_rtl_unswitch): Same.
12503         (pass_rtl_unroll_and_peel_loops): Same.
12504         (pass_rtl_doloop): Same.
12506 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
12508         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12509         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12510         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12511         references.
12512         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12513         * tree-streamer.c (record_common_node): Adjust reference in comment.
12515 2013-04-17  Terry Guo  <terry.guo@arm.com>
12517         * config/arm/cortex-m4.md: Add a new bypass.
12519 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12521         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12522         New pattern.
12523         (*subs_<optab><mode>_multp2): New pattern.
12524         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12525         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12527 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12529         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12530         (*subs_mul_imm_<mode>): New pattern.
12532 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
12534         PR target/56948
12535         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12536         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
12537         (vsx_movti_32bit): Change j->wa to O->wa.
12539 2013-04-16  Richard Biener  <rguenther@suse.de>
12541         PR rtl-optimization/56921
12542         * loop-init.c (pass_rtl_move_loop_invariants): Add
12543         TODO_do_not_ggc_collect to todo_flags_finish.
12544         (pass_rtl_unswitch): Same.
12545         (pass_rtl_unroll_and_peel_loops): Same.
12546         (pass_rtl_doloop): Same.
12548 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12550         * config/arm/arm.c (emit_multi_reg_push): New declaration
12551         for an existing function.
12552         (arm_emit_strd_push): New function.
12553         (arm_expand_prologue): Used here.
12554         (arm_emit_ldrd_pop): New function.
12555         (arm_expand_epilogue): Used here.
12556         (arm_get_frame_offsets): Update condition.
12557         (arm_emit_multi_reg_pop): Add a special case for load of a single
12558         register with writeback.
12560 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
12562         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12563         description.
12565 2013-04-16  Richard Biener  <rguenther@suse.de>
12567         PR tree-optimization/56756
12568         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12569         (first_mem_ref_loc): New.
12570         (execute_sm): Place the load temporarily before a previous
12571         access instead of in the latch edge to ensure its SSA dependencies
12572         are defined at points dominating the load.
12574 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
12576         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12577         correct fix by moving header and footer insn to the footer of
12578         the merged basic block.  Clear BB_END of the merged-away block.
12580         PR middle-end/43631
12581         * emit-rtl.c (make_note_raw): New function.
12582         (link_insn_into_chain): New static inline function.
12583         (add_insn): Use it.
12584         (add_insn_before, add_insn_after): Factor insn chain linking code...
12585         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12586         using link_insn_into_chain.
12587         (note_outside_basic_block_p): New helper function for emit_note_after
12588         and emit_note_before.
12589         (emit_note_after): Use nobb variant of add_insn_after if the note
12590         should not be contained in a basic block.
12591         (emit_note_before): Use nobb variant of add_insn_before if the note
12592         should not be contained in a basic block.
12593         (emit_note_copy): Use make_note_raw.
12594         (emit_note): Likewise.
12595         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12596         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12597         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12598         the moved barrier the tail of the basic block it follows.
12599         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12601 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
12603         PR tree-optimization/56962
12604         * gimple-ssa-strength-reduction.c (record_increment): Only set
12605         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12606         either rhs1 or rhs2 is equal to c->base_expr.
12608 2013-04-15  Richard Biener  <rguenther@suse.de>
12610         PR tree-optimization/56933
12611         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12612         member.
12613         (GROUP_READ_WRITE_DEPENDENCE): Remove.
12614         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12615         * tree-vect-data-refs.c (vect_analyze_group_access): Move
12616         dependence check ...
12617         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12618         ... here.
12619         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12620         GROUP_READ_WRITE_DEPENDENCE.
12622 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12624         * emit-rtl.c (reset_all_used_flags): New function.
12625         (verify_rtl_sharing): Call reset_all_used_flags before and after
12626         performing the checks.
12628 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12630         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12631         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12632         * config/arm/constraints.md (De): New constraint.
12633         * config/arm/neon.md (anddi3_neon): Delete.
12634         (neon_vand<mode>): Expand to standard anddi3 pattern.
12635         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12636         Move earlier in the file.
12637         (neon_inv_logic_op2): Likewise.
12638         (arm_anddi_operand_neon): New predicate.
12640 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12642         * configure.ac (gcc_cv_ld_as_needed): Set
12643         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12644         Use -z ignore, -z record on *-*-solaris2*.
12645         (HAVE_LD_AS_NEEDED): Update comment.
12646         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12647         * configure: Regenerate.
12648         * config.in: Regenerate.
12649         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12650         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12651         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12652         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12653         equivalents.  Fix markup.
12654         * doc/tm.texi: Regenerate.
12656 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
12658         * config/i386/i386.opt: New option mstack-protector-guard=.
12659         * config/i386/i386-opts.h: Add enum stack_protector_guard.
12660         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12661         TARGET_SSP_TLS_GUARD.
12662         * config/i386/i386.c (ix86_option_override_internal): Set
12663         ix86_stack_protector_guard.
12664         * config/i386/i386.md (stack_protect_set): Enable for
12665         TARGET_SSP_TLS_GUARD only.
12666         (stack_protect_set_<mode>): Ditto.
12667         (stack_protect_test): Ditto.
12668         (stack_protect_test_<mode>): Ditto.
12669         * doc/invoke.texi (i386 Option): Document.
12671 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
12673         PR target/56890
12674         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12675         (S_MODES): Set H_MODE bit.
12676         (SF_MODES): Set only S_MODE and SF_MODE bits.
12677         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12678         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12679         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12680         <MODE_FLOAT>: Likewise.
12682 2013-04-15  Joey Ye  <joey.ye@arm.com>
12684         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12686 2013-04-15  Joey Ye  <joey.ye@arm.com>
12688         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12689         for real far jump.
12690         (thumb_far_jump_used_p): Count instruction size and set
12691         far_jump_used.
12693 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
12695         * reorg.c (fill_simple_delay_slots): Reindent block of code.
12696         * resource.c (mark_target_live_regs): Reformat conditional block.
12698 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12700         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12701         notes, they are emitted only just before final.
12702         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12704 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12706         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12707         * cfgrtl.c (delete_insn): Call it here instead.
12708         * lra-spills.c (lra_final_code_change): Use delete_insn.
12709         * haifa-sched.c (sched_remove_insn): Likewise.
12710         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12711         returning to the nop pool.
12712         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12713         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
12715 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12717         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12718         * doc/tm.texi: Regenerated.
12720 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
12722         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12723         QImode checks.
12725 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12727         * df-core.c (df_find_def): Compare register numbers.
12728         (df_find_use): Likewise.
12730 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
12732         PR target/56903
12733         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12734         lra_in_progress for return.
12736 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12738         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12739         define_insn into define_insn_and_split and emit movsicc patterns.
12741 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12743         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12745 2013-04-12  Richard Biener  <rguenther@suse.de>
12747         * tree-pass.h (TODO_do_not_ggc_collect): New.
12748         * passes.c (execute_one_ipa_transform_pass): Honor
12749         TODO_do_not_ggc_collect.
12750         (execute_one_pass): Likewise.
12752         Revert
12753         2013-04-10  Richard Biener  <rguenther@suse.de>
12755         * passes.c (init_optimization_passes): Remove reload pass.
12756         * ira.c (do_reload): Merge into ...
12757         (ira): ... this.
12758         (rest_of_handle_reload): Remove.
12759         (pass_reload): Likewise.
12760         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12761         of reload for vzeroupper pass placement.
12763 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12765         PR tree-optimization/56918
12766         PR tree-optimization/56920
12767         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12768         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12769         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12770         use wide_mul_with_sign method.
12772 2013-04-12  Richard Biener  <rguenther@suse.de>
12774         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12775         not be considered a gimple constant.
12777 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12779         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12780         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12781         operations.
12783 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12784             Jakub Jelinek  <jakub@redhat.com>
12786         * opts.c: Include diagnostic-color.h.
12787         (common_handle_option): Handle OPT_fdiagnostics_color_.
12788         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12789         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12790         (diagnostic-color.o): New.
12791         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12792         (diagnostic_color_rule): New enum.
12793         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12794         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12795         the location string.
12796         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12797         either NULL, or color kind.
12798         * diagnostic-color.c: New file.
12799         * diagnostic-color.h: New file.
12800         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12801         arguments.
12802         * doc/invoke.texi (-fdiagnostics-color): Document.
12803         * pretty-print.h (pp_show_color): Define.
12804         (struct pretty_print_info): Add show_color field.
12805         * diagnostic.c: Include diagnostic-color.h.
12806         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12807         macros.  Colorize error:, warning: etc. strings and also the location
12808         string.
12809         (diagnostic_show_locus): Colorize the caret line.
12810         * pretty-print.c: Include diagnostic-color.h.
12811         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12812         inside of %< %> quotes or quoted through q format modifier.
12814 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12816         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12818 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12820         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12821         code in CC_NZ mode.
12822         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12823         pattern.
12825 2013-04-11  Marek Polacek  <polacek@redhat.com>
12827         PR tree-optimization/48184
12828         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12830 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12832         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12833         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12834         (skip_simple_arithmetic): Tidy up.
12835         * tree.h (skip_simple_constant_arithmetic): Declare.
12837 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12839         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12841 2013-04-11  Richard Biener  <rguenther@suse.de>
12843         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12844         generate vector constants.
12846 2013-04-11  Richard Biener  <rguenther@suse.de>
12848         PR tree-optimization/56878
12849         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12850         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12851         New function.
12852         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12853         Prefer to align the DR with the most invariant base address.
12855 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12857         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12858         comment.
12860 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12862         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12863         floating-point vector comparisons against 0.
12865 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12867         PR tree-optimization/56899
12868         * fold-const.c (extract_muldiv_1): Apply distributive law
12869         only if TYPE_OVERFLOW_WRAPS (ctype).
12871 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12873         PR target/56124
12874         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12875         loading has side effect.
12877 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12879         * config/sparc/sparc.c: Include tree-pass.h.
12880         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12881         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12882         head of file.  Change return type.  Split off gate function.
12883         (sparc_gate_work_around_errata): New function.
12884         (pass_work_around_errata): New pass definition.
12885         (insert_pass_work_around_errata) New pass insert definition to
12886         insert pass_work_around_errata just after delayed-branch scheduling.
12887         (sparc_option_override): Insert the pass.
12888         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12890 2013-04-10  David S. Miller  <davem@davemloft.net>
12892         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12893         or -mcpu=hypersparc.
12895         * target.def (cstore_mode): New hook.
12896         * target.h: Include insn-codes.h
12897         * targhooks.c: Likewise.
12898         (default_cstore_mode): New function.
12899         * targhooks.h: Declare it.
12900         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12901         * doc/tm.texi: Rebuild.
12902         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12903         target hook, rather than inspecting the insn_data.
12904         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12905         (TARGET_CSTORE_MODE): Redefine.
12906         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12907         result patterns.
12908         * config/sparc/predicates.md (cstore_result_operand): New special
12909         predicate.
12910         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12911         Use it for operand 0.
12912         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12913         (*snesi_special): Likewise.
12914         (*snesi_zero): Likewise.
12915         (*seqsi_zero): Likewise.
12916         (*sltu_insn): Likewise.
12917         (*sgeu_insn): Likewise.
12918         (*seqdi_special): Make operand 0 and comparison operation be of
12919         DImode.
12920         (*snedi_special): Likewise.
12921         (*snedi_special_vis3): Likewise.
12922         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12923         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12924         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12925         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12926         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12927         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12928         (*sltu_extend_sp64): Likewise.
12929         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12930         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12931         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12932         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12933         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12935 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12937         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12938         (aarch64_start_file): Use the new function.
12940 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12941             Jason Merrill  <jason@redhat.com>
12943         * common.opt: Add -gdwarf.
12944         * opts.c (common_handle_option): Handle it.
12945         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12947 2013-04-10  Richard Biener  <rguenther@suse.de>
12949         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12950         (execute_one_ipa_transform_pass): But unconditionally here.
12951         (execute_one_pass): And here.
12952         (init_optimization_passes): Remove reload pass.
12953         * tree-pass.h (TODO_ggc_collect): Remove.
12954         (pass_reload): Likewise.
12955         * ira.c (do_reload): Merge into ...
12956         (ira): ... this.
12957         (rest_of_handle_reload): Remove.
12958         (pass_reload): Likewise.
12959         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12960         of reload for vzeroupper pass placement.
12961         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12962         and todo_flags_finish of all passes.
12964 2013-04-10  Richard Biener  <rguenther@suse.de>
12966         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12967         first_const_oprnd field, rename first_def_type to first_op_type.
12968         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12969         (vect_get_and_check_slp_defs): Always use the type of the
12970         operand.  Allow mixed vect_external_def, vect_constant_def types.
12971         (vect_get_constant_vectors): Handle mixed vect_external_def,
12972         vect_constant_def types.
12974 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
12976         PR tree-optimization/55524
12977         * tree-ssa-math-opts.c
12978         (convert_mult_to_fma): Don't use an fms construct
12979         when we don't have an fms operation, but fnma, and it looks
12980         likely that we'll be able to use the latter.
12982 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
12984         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12985         function.
12986         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12987         inline fail caused by overwritable functions.
12989 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12991         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12992         unnecessary bits in the constant power of two case.
12994 2013-04-10  Richard Biener  <rguenther@suse.de>
12996         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12997         broken code swapping operands.
12998         (vect_build_slp_tree): Do not compute load permutations here.
12999         (vect_analyze_slp_instance): Compute load permutations here,
13000         after building the SLP tree.
13002 2013-04-09  Christian Bruel  <christian.bruel@st.com>
13004         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
13005         of next/prev_real_insn.
13007 2013-04-09  Jan Hubicka  <jh@suse.cz>
13009         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
13010         Drop aliased parameter.
13011         (function_and_variable_visibility): Do not handle alias pairs.
13012         * cgraph.c (varpool_externally_visible_p): Update prototype.
13013         * varpool.c (varpool_add_new_variable): Update.
13015 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13017         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13019 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
13021         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13023         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13025 2013-04-09  Marek Polacek  <polacek@redhat.com>
13027         PR tree-optimization/48762
13028         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13030 2013-04-09  Richard Biener  <rguenther@suse.de>
13032         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13033         dealing with cost.
13034         (vect_build_slp_tree): Likewise.
13035         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13036         calculating the cost of a SLP instance.
13037         (vect_analyze_slp_instance): Use it from here, after building
13038         the SLP tree.
13040 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13042         PR middle-end/56883
13043         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13044         expand_omp_for_static_chunk): Use simple_p = true in
13045         force_gimple_operand_gsi calls when assigning to addressable decls.
13047 2013-04-09  Jeff Law  <law@redhat.com>
13049         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13050         when the boolean was created by converting a wider object which
13051         had a boolean range.
13053 2013-04-09  Richard Biener  <rguenther@suse.de>
13055         * tree-vectorizer.h (slp_void_p): Remove.
13056         (slp_tree): Typedef before _slp_tree declaration.
13057         (struct _slp_tree): Use a vector of slp_tree as children.
13058         (vect_get_place_in_interleaving_chain): Remove.
13059         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13060         Move ...
13061         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13062         and make static.
13063         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13064         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13065         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13066         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13067         Use slp_node instead of slp_void_p and adjust.
13069 2013-04-09  Richard Biener  <rguenther@suse.de>
13071         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13072         work that is not necessary.
13074 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13076         PR tree-optimization/56854
13077         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13078         forward into clobber stmts if it would change MEM_REF lhs into
13079         non-MEM_REF.
13081 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
13083         * tree.c (type_hash_lookup, type_hash_add): Make static.
13084         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13086 2013-04-09  Richard Biener  <rguenther@suse.de>
13088         * tree.h (unsave_expr_now): Remove.
13089         * tree-inline.c (mark_local_for_remap_r): Remove.
13090         (unsave_expr_1): Likewise.
13091         (unsave_r): Likewise.
13092         (unsave_expr_now): Likewise.
13093         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13094         (propagate_tree_value): Likewise.
13096 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
13098         * doc/rtl.texi (sequence): Rewrite documentation to match the
13099         current use of SEQUENCE rtl objects.
13100         * rtl.def (SEQUENCE): Likewise.
13102         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13103         Update documentation.
13104         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13105         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13107         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13109 2013-04-08  Teresa Johnson  <tejohnson@google.com>
13111         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13112         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13113         methods.
13114         (estimate_edge_size_and_time): Add comment to suggest using rounding
13115         methods.
13116         (estimate_node_size_and_time): Ditto.
13117         (remap_edge_change_prob): Use helper rounding divide methods.
13118         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13119         (gimple_mod_pow2_value_transform): Ditto.
13120         (gimple_mod_subtract_transform): Ditto.
13121         (gimple_ic_transform): Ditto.
13122         (gimple_stringops_transform): Ditto.
13123         * stmt.c (conditional_probability): Ditto.
13124         (emit_case_dispatch_table): Ditto.
13125         * lto-cgraph.c (merge_profile_summaries): Ditto.
13126         * tree-optimize.c (execute_fixup_cfg): Ditto.
13127         * cfgcleanup.c (try_forward_edges): Ditto.
13128         * cfgloopmanip.c (scale_loop_profile): Ditto.
13129         (loopify): Ditto.
13130         (duplicate_loop_to_header_edge): Ditto.
13131         (lv_adjust_loop_entry_edge): Ditto.
13132         * tree-vect-loop.c (vect_transform_loop): Ditto.
13133         * profile.c (compute_branch_probabilities): Ditto.
13134         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13135         * lto-streamer-in.c (input_cfg): Ditto.
13136         * gimple-streamer-in.c (input_bb): Ditto.
13137         * ipa-cp.c (update_profiling_info): Ditto.
13138         (update_specialized_profile): Ditto.
13139         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13140         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13141         rounding methods.
13142         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13143         (compute_trg_info): Ditto.
13144         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13145         (purge_dead_edges): Ditto.
13146         * loop-unswitch.c (unswitch_loop): Ditto.
13147         * cgraphclones.c (cgraph_clone_edge): Ditto.
13148         (cgraph_clone_node): Ditto.
13149         * tree-inline.c (copy_bb): Ditto.
13150         (copy_edges_for_bb): Ditto.
13151         (initialize_cfun): Ditto.
13152         (copy_cfg_body): Ditto.
13153         (expand_call_inline): Ditto.
13155 2013-04-08  Kai Tietz  <ktietz@redhat.com>
13157         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13158         TARGET_CYGWIN64 by TARGET_64BIT.
13160 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13162         * config/epiphany/epiphany.md (GPR_1): New constant.
13163         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13164         * config/epiphany/epiphany.c (gen_compare_reg):
13165         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13166         is already in place.
13167         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13168         Don't require being called during rtl expansion; If y operlaps r0,
13169         return 0.
13170         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13171         (epiphany_expand_epilogue): Likewise.
13173         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13174         Don't use CC_FPmode for ORDERED / UNORDERED.
13175         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13177         * config/epiphany/constraints.md (CnL): New constraint.
13178         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13179         * config/epiphany/predicates.md (add_operand): Allow 1024.
13181         * config/epiphany/epiphany.md (logical_op): New code iterator.
13182         (op_mnc): New code attribute.
13183         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13184         (mov_f+1, mov_f+2): New peephole2 patterns.
13186         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13187         (cstoresi4): Also allow re-use of zero result when doing a NE
13188         comparison to a non-zero operand.
13189         Use (clobber (scratch)) for first insn if the gpr output is not needed.
13191         * config/epiphany/epiphany.md (<insn_opname>v2si3):
13192         Use gen_addsi3_i / gen_subsi3_i.
13194 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13196         PR c++/34949
13197         PR c++/50243
13198         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13199         contain anything but clobbers, at most one __builtin_stack_restore,
13200         optionally debug stmts and final resx, and if it has at least one
13201         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
13202         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13203         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13204         which isn't defaut definition, remove them.
13205         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13206         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13207         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13208         with MEM_REF LHS with SSA_NAME address.
13210 2013-04-08  Jeff Law  <law@redhat.com>
13212         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13214 2013-04-08  Richard Biener  <rguenther@suse.de>
13216         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13217         extra newline.
13218         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13219         determined vector type.
13220         (vect_analyze_data_refs): Likewise.
13221         (vect_get_new_vect_var): Adjust.
13222         (vect_create_destination_var): Preserve SSA name versions.
13223         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13224         not dump anything here.
13226 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13228         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13229         Add member lr_slot_known.
13230         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13231         if necessary.
13232         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13233         Remove code that sets lr_slot_offset according to what a previous
13234         version of epiphany_emit_save_restore used to do.
13235         (epiphany_emit_save_restore): When doing an lr save or restore,
13236         set/verify lr_slot_known and lr_slot_offset.
13238 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
13240         PR target/54338
13241         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13242         in ALL_REGS.
13244 2013-04-08  Richard Biener  <rguenther@suse.de>
13246         * alias.c (find_base_term): Fix thinko in previous change.
13248 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13250         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13251         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13252         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
13253         if possible to compute val.
13254         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13255         For QImode integers don't require anything about precision.  Use
13256         const_with_all_bytes_same to find out if the constant doesn't have
13257         repeated bytes in it.
13259 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13261         * config/s390/s390.c (s390_expand_insv): Only accept insertions
13262         within mode size.
13264 2013-04-08  Marek Polacek  <polacek@redhat.com>
13266         PR rtl-optimization/48182
13267         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13268         value to 1.
13270 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13272         PR target/55487
13273         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13274         nuses, make sure we have a label.
13276 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13278         PR target/56843
13279         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13280         (rs6000_emit_swdiv_low_precision): Remove.
13281         (rs6000_emit_swdiv): Rewrite to handle between one and four
13282         iterations of Newton-Raphson generally; modify required number of
13283         iterations for some cases.
13284         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13286 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
13288         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13289         set-but-unused variable.
13291         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13292         basic blocks of released function bodies garbage-collectable.
13294         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13295         (struct rtl_opt_pass): Add TODO_df_finish.
13297         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13299 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13301         * config/arm/constraints.md (q): New constraint.
13302         * config/arm/ldrdstrd.md: New file.
13303         * config/arm/arm.md (ldrdstrd.md) New include.
13304         (arm_movdi): Use "q" instead of "r" constraint
13305         for double-word memory access.
13306         (movdf_soft_insn): Likewise.
13307         * config/arm/vfp.md (movdi_vfp): Likewise.
13308         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13309         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13310         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13311         (mem_ok_for_ldrd_strd): Likewise.
13312         (output_move_double): Update assertion.
13314 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13316         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13318 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13320         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13321         define_insn_and_split.
13322         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13324 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13326         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13327         define_insn_and_split.
13328         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13329         (shiftsi3_compare): New pattern.
13330         (rrx): New pattern.
13331         * config/arm/unspecs.md (UNSPEC_RRX): New.
13333 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13335         * config/arm/arm.md (negdi_extendsidi): New pattern.
13336         (negdi_zero_extendsidi): Likewise.
13338 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13340         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13341         define_insn_and_split.
13342         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13343         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13345 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13347         * config/arm/arm.md (arm_subdi3): Convert define_insn into
13348         define_insn_and_split.
13349         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13350         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13352 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13354         * config/arm/arm.md (subsi3_carryin): New pattern.
13355         (subsi3_carryin_const): Likewise.
13356         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13357         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13359 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13361         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13363 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13365         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13366         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13368 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13370         * config/arm/arm.c (arm_expand_builtin): Change fcode
13371         type to unsigned int.
13373 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13375         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13377 2013-04-04  Ian Lance Taylor  <iant@google.com>
13379         * doc/standards.texi (Standards): The Go frontend supports the Go 1
13380         language standard.
13382 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
13384         PR middle-end/56729
13385         * df-scan.c (df_insn_delete): Disable failing assert.
13387 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13389         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13390         New function prototype.
13391         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13392         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13393         (arm_builtin_vectorized_function): New function.
13395 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13397         * config/arm/arm_neon_builtins.def: New file.
13398         * config/arm/arm.c (neon_builtin_data): Move contents to
13399         arm_neon_builtins.def.
13400         (enum arm_builtins): Include neon builtin definitions.
13401         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13402         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13404 2013-04-04  Marek Polacek  <polacek@redhat.com>
13406         PR tree-optimization/48186
13407         * predict.c (maybe_hot_frequency_p): Return false if
13408         HOT_BB_FREQUENCY_FRACTION is 0.
13409         (cgraph_maybe_hot_edge_p): Likewise.
13411 2013-04-04  Richard Biener  <rguenther@suse.de>
13413         PR tree-optimization/56826
13414         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13415         more accurately.
13417 2013-04-04  Richard Biener  <rguenther@suse.de>
13419         PR tree-optimization/56213
13420         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13421         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13423 2013-04-04  Richard Biener  <rguenther@suse.de>
13425         PR tree-optimization/56837
13426         * tree-loop-distribution.c (classify_partition): For non-zero
13427         values require that the value has the same precision as its
13428         mode to be useful as memset value.
13430 2013-04-03  Nick Clifton  <nickc@redhat.com>
13432         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13433         (fmssf4): Use fmsf.s on E3V5 architectures.
13434         (fnmasf4): Use fnmaf.s on E3V5 architectures.
13435         (fnmssf4): Use fnmsf.s on E3V5 architectures.
13437 2013-04-03  Jeff Law  <law@redhat.com>
13439         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13440         (lra-eliminations.o): Likewise.
13442 2013-04-03  Teresa Johnson  <tejohnson@google.com>
13444         * gcov-io.c (compute_working_sets): Moved most of body of old
13445         compute_working_sets here from profile.c.
13446         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13447         (gcov_working_set_t): Moved typedef here from basic-block.h
13448         (compute_working_set): Declare.
13449         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13450         (get_working_sets): Renamed from compute_working_set,
13451         replace most of body with call to new compute_working_sets.
13452         (get_exec_counts): Replace call to compute_working_sets
13453         to get_working_sets.
13454         * profile.h (get_working_sets): Renamed from compute_working_set.
13455         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13456         to get_working_sets.
13457         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13458         * gcov-dump.c (dump_working_sets): New function.
13460 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
13462         * hwint.c (sext_hwi, zext_hwi): New functions.
13463         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13464         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13465         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13466         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13467         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13468         (sext_hwi, zext_hwi): New functions.
13470 2013-04-03  Jeff Law  <law@redhat.com>
13472         PR tree-optimization/56799
13473         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13474         back test for widening conversion erroneously dropped in prior change.
13476 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13478         PR target/56809
13479         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13480         instead of next_real_insn.
13482 2013-04-03  Marek Polacek  <polacek@redhat.com>
13484         PR sanitizer/55702
13485         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13487 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13489         PR target/56809
13490         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13491         next_real_insn.
13492         (thumb1_output_casesi): Likewise.
13493         (thumb2_output_casesi): Likewise.
13495 2013-04-03  Richard Biener  <rguenther@suse.de>
13497         PR tree-optimization/56817
13498         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13499         Split out ...
13500         (tree_unroll_loops_completely_1): ... new function to manually
13501         walk the loop tree, properly defering outer loops of unrolled
13502         loops to later iterations.
13504 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13506         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13507         (vectorizable_load): Likewise.
13508         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13509         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13511 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13513         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13514         BIT_FIELD_REF.
13516 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13518         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13520 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
13522         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13524 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13526         PR tree-optimization/56790
13527         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13528         folding.
13530 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13532         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13533         Handle VEC_MERGE.
13534         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13535         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13536         equal arguments.
13538 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
13540         PR c/19449
13541         * tree.h (force_folding_builtin_constant_p): New decl.
13542         * builtins.c (force_folding_builtin_constant_p): New variable.
13543         (fold_builtin_constant_p): Fold immediately also if
13544         force_folding_builtin_constant_p.
13546 2013-04-03  Richard Biener  <rguenther@suse.de>
13548         PR tree-optimization/56812
13549         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13550         DRs of the same interleaving chain are independent.
13552 2013-04-02  Jason Merrill  <jason@redhat.com>
13554         * gdbinit.in (pbb): Use debug fn.
13556 2013-04-02  Lawrence Crowl  <crowl@google.com>
13558         * sese.h (struct ivtype_map_elt_s): Remove unused.
13559         (extern debug_ivtype_map): Remove unused.
13560         (extern eq_ivtype_map_elts): Remove unused.
13561         * sese.c (debug_ivtype_map): Removed unused.
13562         (debug_ivtype_map_1): Removed unused.
13563         (debug_ivtype_elt): Remove unused.
13564         (eq_ivtype_map_elts): Remove unused.
13567 2013-04-02  Kai Tietz  <ktietz@redhat.com>
13569         PR target/52790
13570         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13571         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13572         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13573         function.
13574         (legitimize_pe_coff_symbol): Likewise.
13575         (is_imported_p): New helper-function.
13576         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13577         for Windows x64 targets.
13578         (ix86_expand_prologue): Optimize for pe-coff targets.
13579         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13580         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13581         medium/large code-model.
13582         (legitimize_pic_address): Likewise.
13583         (legitimize_tls_address): Likewise.
13584         (ix86_expand_call): Likewise.
13585         (x86_output_mi_thunk): Likewise.
13586         (get_dllimport_decl): Add new beimport argument.
13587         (construct_plt_address): Don't assert for x64 pe-coff targets.
13588         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13589         targets.
13590         (SYMBOL_FLAG_STUBVAR): New macro.
13591         (SYMBOL_REF_STUBVAR_P): Likewise.
13592         * config/i386/winnt.c (stub_list): New structure.
13593         (stub_head): New local variable.
13594         (i386_pe_record_stub): New function.
13595         (i386_pe_file_end): Emit refptr-stubs.
13597 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
13599         PR rtl-optimization/56745
13600         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13601         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13603         PR c++/34949
13604         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13605         and both of them are MEM_REFs, just compare first argument for
13606         equality and attempt to deal even with differing offsets.
13608         PR c++/34949
13609         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13610         of gimple_clobber_p to be MEM_REF.
13611         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13612         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
13613         after gimplification.
13614         * asan.c (get_mem_ref_of_assignment): Don't instrument
13615         gimple_clobber_p stmts.
13616         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13617         gimple_clobber_p stmt if they have MEM_REF lhs and
13618         are dead because of another gimple_clobber_p stmt.
13619         * tree-ssa-live.c (clear_unused_block_pointer): Treat
13620         gimple_clobber_p stmts like debug stmts.
13621         (remove_unused_locals): Remove clobbers with MEM_REF lhs
13622         that refer to unused VAR_DECLs or uninitialized values.
13623         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13624         gimple_clobber_p stmts if they refer to removed parameters.
13625         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13626         formatting.
13628 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
13630         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13631         using SWI48 mode attribute.
13633 2013-04-02  Wei Mi  <wmi@google.com>
13635         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13636         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13637         *<rotate_insn><mode>3_mask in i386.md.
13639 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13641         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13643 2013-04-02  Richard Biener  <rguenther@suse.de>
13645         PR tree-optimization/56778
13646         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13647         Runtime alias tests are not supported for gather loads.
13648         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13649         stmts referenced from SSA operands before updating SSA form.
13651 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
13652             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13654         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13655         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13656         * config/arm/cortex-a53.md: New file.
13657         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13658         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13659         * config/arm/arm.c (arm_issue_rate): Likewise.
13660         * config/arm/arm-tune.md: Regenerate
13661         * config/arm/arm-tables.opt: Regenerate.
13662         * config/arm/arm-cores.def: Add cortex-a53.
13664 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
13666         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13667         non-static link.
13669 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13671         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13672         scalar load/store operations using B/H registers.
13673         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13675 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13677         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13678         scalar move.
13679         * config/aarch64/aarch64.c
13680         (aarch64_simd_scalar_immediate_valid_for_move): New.
13681         * config/aarch64/aarch64-protos.h
13682         (aarch64_simd_scalar_immediate_valid_for_move): New.
13683         * config/aarch64/constraints.md (Dh, Dq): New.
13684         * config/aarch64/iterators.md (hq): New.
13686 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
13688         * reorg.c (get_branch_condition): Deal with conditional returns.
13689         (fill_simple_delay_slots): Remove dead code dealing with jumps.
13691 2013-04-01  Wei Mi  <wmi@google.com>
13693         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13694         Truncate operand 2 using %b asm operand modifier.
13695         (*<shift_insn><mode>3_mask): Ditto.
13696         (*<rotate_insn><mode>3_mask): Ditto.
13698 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
13700         PR middle-end/56798
13701         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13703 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
13705         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13706         of next_real_insn.
13707         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13709 2013-03-30  Lawrence Crowl  <crowl@google.com>
13711         * dse.c (clear_alias_sets): Remove never set.
13712         (disqualified_clear_alias_sets): Remove never set.
13713         (clear_alias_mode_pool): Remove never set.
13714         (dse_step0): Remove condition that is never true.
13715         (canon_address): Remove condition that is never true.
13716         (dse_step7): Remove condition that is never true.
13717         (rest_of_handle_dse): Remove condition that is never true.
13718         (rest_of_handle_dse::did_global): Remove never read from above.
13719         (dse_step2_spill): Remove never called from above.
13720         (dse_step5_spill): Remove never called from above.
13722 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
13724         * doc/md.texi (Standard Names) <casesi>: Update documentation for
13725         JUMP_TABLE_DATA changes.
13726         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13727         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13728         (Insns) <jump_table_data>: New entry.
13729         * doc/tm.texi: Regenerate.
13731         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13733         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13734         for table jump at the end of a basic block using tablejump_p.
13735         * targhooks.c (default_invalid_within_doloop): Likewise.
13736         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13737         target hook implementation that is identical to the default hook.
13738         (rs6000_invalid_within_doloop): Remove.
13740         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13741         unused variable from tablejump_p call.
13743         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13744         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13745         (INSN_DELETED_P): Likewise.
13746         (emit_jump_table_data): New prototype.
13747         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13748         after 4th as unused.
13749         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13750         * sched-vis.c (print_insn): Likewise.
13751         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13752         insn for compatibility with back ends that use next_active_insn to
13753         identify jump table data.
13754         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13755         (remove_insn): Likewise.
13756         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13757         to be emitted.
13758         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13759         (emit_jump_table_data): New function.
13761         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13762         basic block, a JUMP_TABLE_DATA never is.
13763         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13764         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13765         off from code handling real insns.
13766         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13767         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13768         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13769         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13770         is not a NONDEBUG_INSN_P.
13771         * ira-costs.c (scan_one_insn): Likewise.
13772         * jump.c (mark_all_labels): Likewise.
13773         (mark_jump_label_1): Likewise.
13774         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13775         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13776         a basic block.
13777         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13778         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13779         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13780         JUMP_TABLE_DATA_P insns.
13781         (calculate_elim_costs_all_insns): Likewise.
13782         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13783         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13784         (delete_output_reload): Code style fixups.
13785         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13786         insn flags on this non-insn.
13787         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13788         as scheduling barriers, for pre-change compatibility.
13789         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13790         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13792         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13793         redundant JUMP_TABLE_DATA_P test.
13794         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13795         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13796         (frv_for_each_packet): Likewise.
13797         * config/i386/i386.c (min_insn_size): Likewise.
13798         (ix86_avoid_jump_mispredicts): Likewise.
13799         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13800         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13801         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13802         (mips16_insn_length): Robustify.
13803         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13804         (mips16_split_long_branches): Likewise.
13805         * config/pa/pa.c (pa_combine_instructions): Likewise.
13806         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13807         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13808         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13809         as contributing to pool range lengths.
13810         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13811         Remove redundant JUMP_TABLE_DATA_P test.
13812         (sh_loop_align): Likewise.
13813         (split_branches): Likewise.
13814         (sh_insn_length_adjustment): Likewise.
13815         * config/spu/spu.c (get_branch_target): Likewise.
13817 2013-03-29  Jan Hubicka  <jh@suse.cz>
13819         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13820         gcov streaming; stream hot bb threshold to ltrans.
13821         * predict.c (get_hot_bb_threshold): Break out from ....
13822         (maybe_hot_count_p): ... here.
13823         (set_hot_bb_threshold): New function.
13824         * lto-section-in.c (lto_section_name): Add profile.
13825         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13826         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13827         and data-streamer.h
13828         (histogram_entry): New structure.
13829         (histogram, histogram_pool): New global vars.
13830         (histogram_hash): New structure.
13831         (histogram_hash::hash): New method.
13832         (histogram_hash::equal): Likewise.
13833         (account_time_size): New function.
13834         (cmp_counts): New function.
13835         (dump_histogram): New function.
13836         (ipa_profile_generate_summary): New function.
13837         (ipa_profile_write_summary): New function.
13838         (ipa_profile_read_summary): New function.
13839         (ipa_profile): Decide on threshold.
13840         (pass_ipa_profile): Add ipa_profile_write_summary and
13841         ipa_profile_read_summary.
13842         * Makefile.in (ipa.o): Update dependencies.
13843         * lto-streamer.h (LTO_section_ipa_profile): New section.
13845 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13847         * tree.h (VAR_P): New.
13849 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13851         PR lto/56777
13852         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13854 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13856         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13857         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13858         (control_flow_insn_p): Likewise.
13859         * cfgrtl.c (duplicate_insn_chain): Likewise.
13860         * final.c (get_attr_length_1): Likewise.
13861         (shorten_branches): Likewise.
13862         (final_scan_insn): Likewise.
13863         * function.c (instantiate_virtual_regs): Likewise.
13864         * gcse.c (insert_insn_end_basic_block): Likewise.
13865         * ira-costs.c (scan_one_insn): Likewise.
13866         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13867         * lra.c (check_rtl): Likewise.
13868         * reload1.c (elimination_costs_in_insn): Likewise.
13869         * reorg.c (follow_jumps): Likewise.
13871         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13872         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13873         (thumb_far_jump_used_p): Likewise.
13874         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13875         (workaround_speculation): Likewise.
13876         (add_sched_insns_for_speculation): Likewise.
13877         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13878         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13879         (frv_for_each_packet): Likewise.
13880         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13881         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13882         (final_emit_insn_group_barriers): Likewise.
13883         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13884         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13885         (mips16_insn_length): Likewise.
13886         * config/pa/pa.c (pa_reorg): Likewise.
13887         (pa_combine_instructions): Likewise.
13888         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13889         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13890         (sh_reorg): Likewise.
13891         (split_branches): Likewise.
13892         * config/spu/spu.c (get_branch_target): Likewise.
13894         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13895         JUMP_TABLE_DATA_P.
13897 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13899         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13900         Fix declaration name.
13902 2013-03-28  Lawrence Crowl  <crowl@google.com>
13904         * graphds.h (struct graph.indicies): Remove unused.
13905         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13906         (SCOP_ORIGINAL_PDDRS): Remove unused.
13907         * sese.h (extern insert_loop_close_phis): Removed unused.
13908         (extern insert_guard_phis): Removed unused.
13909         (extern ivtype_map_elt_info): Removed unused.
13910         (new_ivtype_map_elt): Removed unused.
13911         * sese.c (ivtype_map_elt_info): Removed unused.
13913 2013-03-28  Lawrence Crowl  <crowl@google.com>
13915         * Makefile.in: Add several missing include dependences.
13916         (DUMPFILE_H): New.
13917         (test-dump.o): New.  This object is not added to any executable,
13918         but is present for ad-hoc testing.
13919         * bitmap.c
13920         (debug (const bitmap_head_def &)): New.
13921         (debug (const bitmap_head_def *)): New.
13922         * bitmap.h
13923         (extern debug (const bitmap_head_def &)): New.
13924         (extern debug (const bitmap_head_def *)): New.
13925         * cfg.c
13926         (debug (edge_def &)): New.
13927         (debug (edge_def *)): New.
13928         * cfghooks.c
13929         (debug (basic_block_def &)): New.
13930         (debug (basic_block_def *)): New.
13931         * dumpfile.h
13932         (dump_node (const_tree, int, FILE *)): Correct source file.
13933         * dwarf2out.c
13934         (debug (die_struct &)): New.
13935         (debug (die_struct *)): New.
13936         * dwarf2out.h
13937         (extern debug (die_struct &)): New.
13938         (extern debug (die_struct *)): New.
13939         * gimple-pretty-print.c
13940         (debug (gimple_statement_d &)): New.
13941         (debug (gimple_statement_d *)): New.
13942         * gimple-pretty-print.h
13943         (extern debug (gimple_statement_d &)): New.
13944         (extern debug (gimple_statement_d *)): New.
13945         * ira-build.c
13946         (debug (ira_allocno_copy &)): New.
13947         (debug (ira_allocno_copy *)): New.
13948         (debug (ira_allocno &)): New.
13949         (debug (ira_allocno *)): New.
13950         * ira-int.h
13951         (extern debug (ira_allocno_copy &)): New.
13952         (extern debug (ira_allocno_copy *)): New.
13953         (extern debug (ira_allocno &)): New.
13954         (extern debug (ira_allocno *)): New.
13955         * ira-lives.c
13956         (debug (live_range &)): New.
13957         (debug (live_range *)): New.
13958         * lra-int.h
13959         (debug (lra_live_range &)): New.
13960         (debug (lra_live_range *)): New.
13961         * lra-lives.c
13962         (debug (lra_live_range &)): New.
13963         (debug (lra_live_range *)): New.
13964         * omega.c
13965         (debug (omega_pb_d &)): New.
13966         (debug (omega_pb_d *)): New.
13967         * omega.h
13968         (extern debug (omega_pb_d &)): New.
13969         (extern debug (omega_pb_d *)): New.
13970         * print-rtl.c
13971         (debug (const rtx_def &)): New.
13972         (debug (const rtx_def *)): New.
13973         * print-tree.c
13974         (debug_tree (tree): Move within file.
13975         (debug_raw (const tree_node &)): New.
13976         (debug_raw (const tree_node *)): New.
13977         (dump_tree_via_hooks (const tree_node *, int)): New.
13978         (debug (const tree_node &)): New.
13979         (debug (const tree_node *)): New.
13980         (debug_verbose (const tree_node &)): New.
13981         (debug_verbose (const tree_node *)): New.
13982         (debug_head (const tree_node &)): New.
13983         (debug_head (const tree_node *)): New.
13984         (debug_body (const tree_node &)): New.
13985         (debug_body (const tree_node *)): New.
13986         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13987         (debug (vec<tree, va_gc> &)): New.
13988         (debug (vec<tree, va_gc> *)): New.
13989         * rtl.h
13990         (extern debug (const rtx_def &)): New.
13991         (extern debug (const rtx_def *)): New.
13992         * sbitmap.c
13993         (debug_raw (simple_bitmap_def &)): New.
13994         (debug_raw (simple_bitmap_def *)): New.
13995         (debug (simple_bitmap_def &)): New.
13996         (debug (simple_bitmap_def *)): New.
13997         * sbitmap.h
13998         (extern debug (simple_bitmap_def &)): New.
13999         (extern debug (simple_bitmap_def *)): New.
14000         (extern debug_raw (simple_bitmap_def &)): New.
14001         (extern debug_raw (simple_bitmap_def *)): New.
14002         * sel-sched-dump.c
14003         (debug (vinsn_def &)): New.
14004         (debug (vinsn_def *)): New.
14005         (debug_verbose (vinsn_def &)): New.
14006         (debug_verbose (vinsn_def *)): New.
14007         (debug (expr_def &)): New.
14008         (debug (expr_def *)): New.
14009         (debug_verbose (expr_def &)): New.
14010         (debug_verbose (expr_def *)): New.
14011         (debug (vec<rtx> &)): New.
14012         (debug (vec<rtx> *)): New.
14013         * sel-sched-dump.h
14014         (extern debug (vinsn_def &)): New.
14015         (extern debug (vinsn_def *)): New.
14016         (extern debug_verbose (vinsn_def &)): New.
14017         (extern debug_verbose (vinsn_def *)): New.
14018         (extern debug (expr_def &)): New.
14019         (extern debug (expr_def *)): New.
14020         (extern debug_verbose (expr_def &)): New.
14021         (extern debug_verbose (expr_def *)): New.
14022         (extern debug (vec<rtx> &)): New.
14023         (extern debug (vec<rtx> *)): New.
14024         * sel-sched-ir.h
14025         (_list_iter_cond_expr): Make inline instead of static.
14026         * sreal.c
14027         (debug (sreal &)): New.
14028         (debug (sreal *)): New.
14029         * sreal.h
14030         (extern debug (sreal &)): New.
14031         (extern debug (sreal *)): New.
14032         * tree.h
14033         (extern debug_raw (const tree_node &)): New.
14034         (extern debug_raw (const tree_node *)): New.
14035         (extern debug (const tree_node &)): New.
14036         (extern debug (const tree_node *)): New.
14037         (extern debug_verbose (const tree_node &)): New.
14038         (extern debug_verbose (const tree_node *)): New.
14039         (extern debug_head (const tree_node &)): New.
14040         (extern debug_head (const tree_node *)): New.
14041         (extern debug_body (const tree_node &)): New.
14042         (extern debug_body (const tree_node *)): New.
14043         (extern debug (vec<tree, va_gc> &)): New.
14044         (extern debug (vec<tree, va_gc> *)): New.
14045         * tree-cfg.c
14046         (debug (struct loop &)): New.
14047         (debug (struct loop *)): New.
14048         (debug_verbose (struct loop &)): New.
14049         (debug_verbose (struct loop *)): New.
14050         * tree-dump.c: Add header dependence.
14051         * tree-flow.h
14052         (extern debug (struct loop &)): New.
14053         (extern debug (struct loop *)): New.
14054         (extern debug_verbose (struct loop &)): New.
14055         (extern debug_verbose (struct loop *)): New.
14056         * tree-data-ref.c
14057         (debug (data_reference &)): New.
14058         (debug (data_reference *)): New.
14059         (debug (vec<data_reference_p> &)): New.
14060         (debug (vec<data_reference_p> *)): New.
14061         (debug (vec<ddr_p> &)): New.
14062         (debug (vec<ddr_p> *)): New.
14063         * tree-data-ref.h
14064         (extern debug (data_reference &)): New.
14065         (extern debug (data_reference *)): New.
14066         (extern debug (vec<data_reference_p> &)): New.
14067         (extern debug (vec<data_reference_p> *)): New.
14068         (extern debug (vec<ddr_p> &)): New.
14069         (extern debug (vec<ddr_p> *)): New.
14070         * tree-ssa-alias.c
14071         (debug (pt_solution &)): New.
14072         (debug (pt_solution *)): New.
14073         * tree-ssa-alias.h
14074         (extern debug (pt_solution &)): New.
14075         (extern debug (pt_solution *)): New.
14076         * tree-ssa-alias.c
14077         (debug (_var_map &)): New.
14078         (debug (_var_map *)): New.
14079         (debug (tree_live_info_d &)): New.
14080         (debug (tree_live_info_d *)): New.
14081         * tree-ssa-alias.h
14082         (extern debug (_var_map &)): New.
14083         (extern debug (_var_map *)): New.
14084         (extern debug (tree_live_info_d &)): New.
14085         (extern debug (tree_live_info_d *)): New.
14087 2013-03-28  Jan Hubicka  <jh@suse.cz>
14089         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14091 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
14093         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14094         record only when desired or required.
14096 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
14098         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14099         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
14101 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14103         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14104         (*andsi3_compare0_uxtw): New pattern.
14105         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14106         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14108 2013-03-28  Jan Hubicka  <jh@suse.cz>
14110         * data-streamer-in.c (streamer_read_gcov_count): New function.
14111         * gimple-streamer-out.c: Include value-prof.h.
14112         (output_gimple_stmt): Output histogram.
14113         (output_bb): Use streamer_write_gcov_count.
14114         * value-prof.c: Include data-streamer.h
14115         (dump_histogram_value): Add HIST_TYPE_MAX.
14116         (stream_out_histogram_value): New function.
14117         (stream_in_histogram_value): New function.
14118         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14119         (stream_out_histogram_value, stream_in_histogram_value): Declare.
14120         * data-streamer-out.c (streamer_write_gcov_count): New function.
14121         (streamer_write_gcov_count_stream): New function.
14122         * lto-cgraph.c (lto_output_edge): Update counter streaming.
14123         (lto_output_node): Likewise.
14124         (input_node, input_edge): Likewise.
14125         * lto-streamer-out.c (output_cfg): Update streaming.
14126         * lto-streamer-in.c (input_cfg): Likewise.
14127         * data-streamer.h (streamer_write_gcov_count,
14128         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14129         * gimple-streamer-in.c: Include value-prof.h
14130         (input_gimple_stmt): Input histograms.
14131         (input_bb): Update profile streaming.
14133 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14135         * genmodes.c (emit_max_int): New function.
14136         (emit_insn_modes_h): Added call to emit_max_function.
14137         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14138         Added doc.
14139         * machmode.def: Fixed comment.
14141 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14143         * combine.c (try_combine): Removed useless assert.
14144         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14146 2013-03-28  Marek Polacek  <polacek@redhat.com>
14147             Richard Biener  <rguenther@suse.de>
14149         PR tree-optimization/56695
14150         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14151         build signed result of a vector comparison.
14152         * tree-cfg.c (verify_gimple_comparison): Check that a result
14153         of a vector comparison has signed type.
14155 2013-03-28  Richard Biener  <rguenther@suse.de>
14157         PR tree-optimization/37021
14158         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14159         do not restrict gaps between groups.
14160         * tree-vect-stmts.c (vectorizable_load): Properly account for
14161         a gap between groups.
14163 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
14165         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14166         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14167         is not enabled.
14169 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
14171         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14172         * doc/extend.texi (Named Address Spaces): Ditto.
14173         (Variable Attributes): Ditto.
14175 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14177         * config.build: Add support for cygwin x64 target.
14178         * config.gcc: Likewise.
14179         * config.host: Likewise.
14180         * configure.ac: Likewise
14181         * configure: Regenerated.
14183 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14185         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14186         * config/i386/t-cygwin-w64: New file.
14187         * config/i386/cygwin-w64.h: New file.
14188         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14189         and add support for x64-cygwin target.
14190         (CPP_SPEC): Likewise.
14191         (CXX_WRAP_SPEC_LIST): Undefine before define.
14192         (LIBGCJ_SONAME): Use 15 as version.
14194 2013-03-27  Richard Biener  <rguenther@suse.de>
14196         PR tree-optimization/56716
14197         * tree-ssa-structalias.c (perform_var_substitution): Adjust
14198         dumping for ref nodes.
14200 2013-03-27  Martin Jambor  <mjambor@suse.cz>
14202         PR tree-optimization/55334
14203         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14204         restricted pointers to arrays.
14206 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14208         * Makefile.in (.SUFFIXES): Add .cc.
14209         (.c.o): Apply same recipe for implicit rule .cc.o.
14211 2013-03-27  Richard Biener  <rguenther@suse.de>
14213         PR tree-optimization/37021
14214         * tree-vect-data-refs.c (vect_check_strided_load): Allow
14215         REALPART/IMAGPART_EXPRs around the supported refs.
14216         * tree-ssa-structalias.c (find_func_aliases): Assume that
14217         floating-point values are not used to transfer pointers.
14219 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
14221         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14222         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14223         * gcc/doc/tm.texi: Regenerate.
14224         * gcc/targhooks.h (default_has_ifunc_p): New.
14225         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14226         * gcc/config/linux-protos.h: New file.
14227         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14228         this hook for linux which disables support of indirect functions in
14229         android.
14230         * gcc/config/linux-android.c: New file.
14231         * gcc/config/t-linux-android.c: Ditto.
14232         * gcc/config.gcc: Added new object file linux-android.o.
14233         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14234         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14235         * gcc/varasm.c (do_assemble_alias): Likewise.
14236         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14237         doesn't support indirect functions.
14238         * configure: Regenerate.
14240 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
14242         PR target/56102
14243         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14244         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14245         mult-word mode.
14247 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14249         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14251 2013-03-27  Terry Guo  <terry.guo@arm.com>
14253         * config/arm/arm-cores.def: Added core cortex-r7.
14254         * config/arm/arm-tune.md: Regenerated.
14255         * config/arm/arm-tables.opt: Regenerated.
14256         * doc/invoke.texi: Added entry for core cortex-r7.
14258 2013-03-27  Walter Lee  <walt@tilera.com>
14260         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14261         double-decrement of next_scratch_regno.
14263 2013-03-27  Walter Lee  <walt@tilera.com>
14265         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14266         input operands.
14267         (insn_v1mulus): Ditto.
14268         (insn_v2muls): Ditto.
14270 2013-03-27  Walter Lee  <walt@tilera.com>
14272         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14273         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14275 2013-03-27  Walter Lee  <walt@tilera.com>
14277         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14278         (*sibcall_value): Ditto.
14280 2013-03-27  Walter Lee  <walt@tilera.com>
14282         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14283         (insn_mnz_v8qi): ... this ...
14284         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
14285         vector equivalent.
14286         (insn_v<n>mnz): Replaced by ...
14287         (insn_v1mnz): ... this ...
14288         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
14289         equivalent.
14290         (insn_mz_<mode>): Replaced by ...
14291         (insn_mz_v8qi): ... this ...
14292         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
14293         vector equivalent.
14294         (insn_v<n>mz): Replaced by ...
14295         (insn_v1mz): ... this ...
14296         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
14297         equivalent.
14299 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
14301         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14303 2013-03-26  Roland McGrath  <mcgrathr@google.com>
14305         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14306         than fprintf with a non-constant, non-format string.
14308 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14310         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14311         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
14312         operand 0 predicate.
14313         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14314         attribute.  Use general_x64nomem_operand as operand 1 predicate.
14315         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14316         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14317         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14318         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14319         (mov<mode>_insv_1): Remove expander.  Merge insn with
14320         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14321         Use general_x64nomem_operand as operand 1 predicate.
14322         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14323         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14324         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14325         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14326         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
14327         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14328         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14329         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14330         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
14331         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14332         (general_x64nomem_operand): Ditto.
14334 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14336         * config/rtems.opt: Add -pthread option.
14338 2013-03-26  Richard Biener  <rguenther@suse.de>
14340         * alias.c (find_base_term): Avoid redundant and not used recursion.
14341         (base_alias_check): Get the initial base term from the caller.
14342         (true_dependence_1): Compute and pass base terms to base_alias_check.
14343         (write_dependence_p): Likewise.
14344         (may_alias_p): Likewise.
14346 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
14348         * config/aarch64/aarch64.c (aarch64_classify_address): Support
14349         PC-relative load in SI modes and above only.
14351 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
14353         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14354         * config/arm/iwmmxt.md (WCGR0): Update.
14355         (WCGR1, WCGR2, WCGR3): Likewise.
14357 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14359         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14360         Use x64 and nox64 isa attributes.
14362 2013-03-26  Richard Biener  <rguenther@suse.de>
14364         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14365         alignment computations and rely on get_object_alignment_1
14366         for the !TYPE_P case.
14367         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14369 2013-03-26  Walter Lee  <walt@tilera.com>
14371         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14372         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14374 2013-03-25  Jeff Law  <law@redhat.com>
14376         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14377         check for INTEGRAL_TYPE_P that was missing due to checking in
14378         wrong version of prior patch.
14380 2013-03-25  Walter Lee  <walt@tilera.com>
14382         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14383         TILEGX_INSN_SHUFFLEBYTES1.
14384         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14385         shufflebytes1.
14386         (tilegx_builtins): Ditto.
14387         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14389 2013-03-25  Walter Lee  <walt@tilera.com>
14391         * config/tilegx/tilegx.md (floatsisf2): New pattern.
14392         (floatunssisf2): New pattern.
14393         (floatsidf2): New pattern.
14394         (floatunssidf2): New pattern.
14396 2013-03-25  Walter Lee  <walt@tilera.com>
14398         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14399         tests for constraint J, K, N, P.
14401 2013-03-25  Walter Lee  <walt@tilera.com>
14403         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14404         Use indirect/pcrel encoding.
14405         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14406         Ditto.
14408 2013-03-25  Steve Ellcey  <sellcey@mips.com>
14410         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14411         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14412         * config/mips/mips.c (mips_option_override): Set IMADD default.
14413         * config/mips/mips.h (PTF_AVOID_IMADD): New.
14414         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14415         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14416         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14417         * doc/invoke.texi (-mimadd/-mno-imadd): New.
14419 2013-03-25  Jeff Law  <law@redhat.com>
14421         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14422         slightly to avoid creating and folding useless trees.  Simplify
14423         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14425 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
14427         * config/i386/i386.md (*zero_extendsidi2): Merge with
14428         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
14429         * config/i386/predicates.md (x86_64_zext_operand): Rename from
14430         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
14431         targets.  Clarify comment.
14433 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14435         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14436         pass-through jump functions differently.
14437         (ipa_read_jump_function): Likewise.  Also use setter functions to set
14438         up jump functions.
14440 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14442         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14443         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14444         process it.
14445         (ipa_get_indirect_edge_target): New function.
14446         (devirtualization_time_bonus): New parameter known_aggs, pass it to
14447         ipa_get_indirect_edge_target.  Update all callers.
14448         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
14449         ipa_get_indirect_edge_target_1 instead of calling
14450         ipa_get_indirect_edge_target.
14451         (create_specialized_node): Pass aggvlas to
14452         ipcp_discover_new_direct_edges.
14454 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14456         * config/arm/arm.md (f_sels, f_seld): New types.
14457         (*cmov<mode>): New pattern.
14458         * config/arm/predicates.md (arm_vsel_comparison_operator): New
14459         predicate.
14461 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14463         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14464         POSIX-printf for mingw-hosted  builds.
14466 2013-03-25  Richard Biener  <rguenther@suse.de>
14468         PR middle-end/56694
14469         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14470         must-not-throw stmt location.
14472 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14474         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14475         Emit load-acquire versions when acq is true.
14476         (arm_emit_store_exclusive): Add rel parameter.
14477         Emit store-release versions when rel is true.
14478         (arm_split_compare_and_swap): Use acquire-release instructions
14479         instead.
14480         of barriers when appropriate.
14481         (arm_split_atomic_op): Likewise.
14482         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14483         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14484         (VUNSPEC_SLX): Likewise.
14485         (VUNSPEC_LDA): Likewise.
14486         (VUNSPEC_STL): Likewise.
14487         * config/arm/sync.md (atomic_load<mode>): New pattern.
14488         (atomic_store<mode>): Likewise.
14489         (arm_load_acquire_exclusive<mode>): Likewise.
14490         (arm_load_acquire_exclusivesi): Likewise.
14491         (arm_load_acquire_exclusivedi): Likewise.
14492         (arm_store_release_exclusive<mode>): Likewise.
14494 2013-03-25  Catherine Moore  <clm@codesourcery.com>
14496         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14497         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14498         * config/mip/predicates.md (lwsp_swsp_operand,
14499         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14500         sb16_operand, db4_operand, db7_operand, ib3_operand,
14501         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14502         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14503         andi16_operand): New predicates.
14504         * config/mips/mips.md (compression): New attribute.
14505         (enabled): New attribute.
14506         (length): Consider compression in computing length.
14507         (shift_compression): New code attribute.
14508         (*add<mode>3): New operands. Record compression.
14509         (sub<mode>3): Likewise.
14510         (one_cmpl<mode>2): Likewise.
14511         (*and<mode>3): Likewise.
14512         (*ior<mode>3): Likewise.
14513         (unnamed pattern for xor): Likewise.
14514         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14515         (*<optab><mode>3): Likewise.
14516         (*mov<mode>_internal: Likewise.
14517         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14518         (mips_unsigned_immediate_p): New.
14519         (umips_lwsp_swsp_address_p): New.
14520         (m16_based_address_p): New.
14521         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14522         (mips_unsigned_immediate_p): New prototype.
14523         (lwsp_swsp_address_p): New prototype.
14524         (m16_based_address_p): New prototype.
14525         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14526         (mips_signed_immediate_p): New function.
14527         (m16_based_address_p): New function.
14528         (lwsp_swsp_address_p): New function.
14529         (mips_print_operand_punctuation): Recognize short delay slot insns
14530         for microMIPS.add<mode>3"
14532 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14534         PR target/56720
14535         * config/arm/iterators.md (v_cmp_result): New mode attribute.
14536         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14538 2013-03-25  Richard Biener  <rguenther@suse.de>
14540         PR tree-optimization/56689
14541         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14542         any edge.
14544 2013-03-25  Richard Biener  <rguenther@suse.de>
14546         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14547         of bitmap.
14548         (memory_references): Likewise.
14549         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14550         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14551         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14552         (gather_mem_refs_in_loops): Fold into ...
14553         (analyze_memory_references): ... this.  Move initialization
14554         to tree_ssa_lim_initialize.
14555         (fill_always_executed_in): Rename to ...
14556         (fill_always_executed_in_1): ... this.
14557         (fill_always_executed_in): Move contains_call computation to
14558         this new function from ...
14559         (tree_ssa_lim_initialize): ... here.
14560         (tree_ssa_lim): Call fill_always_executed_in.
14562 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14564         * postreload.c (reload_combine): Fix code detecting returns.
14566 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14568         * function.c (emit_use_return_register_into_block): On cc0 targets,
14569         do not emit the sequence between cc0 setter and user.
14571 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14573         * config/i386/predicates.md (local_symbolic_operand): Interpret
14574         dll-imported symbols as none-local.
14576 2013-03-25  Richard Biener  <rguenther@suse.de>
14578         * tree-ssa-loop-im.c (struct depend): Remove.
14579         (struct lim_aux_data): Make depends a vec of gimples.
14580         (free_lim_aux_data): Adjust.
14581         (add_dependency): Likewise.
14582         (set_level): Likewise.
14584 2013-03-25  Richard Biener  <rguenther@suse.de>
14586         PR middle-end/56434
14587         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14588         the pointer returned by calls with ECF_MALLOC set.
14590 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14592         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14594 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14596         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14597         using MMXMODE mode iterator.
14598         (*move<mode>_internal): Merge with *movv2sf_internal and
14599         *movv2sf_internal_rex64 using MMXMODE mode iterator.
14601 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14603         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14604         (record_last_mem_set_info): Likewise.
14606         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14607         of XNEWVEC followed by memset.
14608         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14610 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14612         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14613         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14614         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14615         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14616         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14617         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14618         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14619         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14620         BARRIER_P instead of GET_CODE.
14622 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14624         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14625         inaccuracy in the probing code.
14627         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14628         (ctrapdi4): Likewise.
14630 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14632         * calls.c (expand_call): Add missing guard to code handling return
14633         of non-BLKmode structures in MSB.
14634         * function.c (expand_function_end): Likewise.
14636 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14638         * combine.c (try_combine): Adjust comment.  Do not add the set of
14639         insn #0 if the destination indirectly is set or dies in insn #2.
14640         Tidy up code to distribute a new note.
14642 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14644         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14645         also for alternatives 16 and 17.
14647 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14649         * config/i386/sse.md (*mov<mode>_internal): Merge with
14650         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
14651         Emit insn template depending on type attribute.  Use
14652         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14653         movd instead of movq mnemonic for interunit moves.  Rewrite mode
14654         attribute calculation.  Remove unit attribute calculation.
14655         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14656         Set prefix_data16 attribute for DImode ssemov types.
14657         Use Ym instead of y for SSE-MMX conversion alternatives.
14658         Reorder operand constraints.
14660 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
14662         * df.h (df_insn_delete): Adjust prototype.
14663         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14664         and let it decide whether mark the basic block dirty.
14665         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14666         * df-scan.c (df_insn_info_delete): New helper function, split
14667         off from df_insn_delete.
14668         (df_scan_free_bb_info): Use it.
14669         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14670         Likewise.
14671         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
14672         that the insn is actually an insn and it has a non-NULL basic block.
14673         Do not mark basic block dirty if only deleting a DEBUG_INSN.
14675 2013-03-22  Richard Biener  <rguenther@suse.de>
14677         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14678         dep_ref members.
14679         (mem_ref_alloc): Do not allocate them.
14680         (refs_independent_p): Do not query or maintain a cache.
14682 2013-03-22  Richard Biener  <rguenther@suse.de>
14684         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14685         (gather_mem_refs_in_loops): Do not compute it.
14686         (analyze_memory_references): Do not allocate it.
14687         (tree_ssa_lim_finalize): Do not free it.
14688         (for_all_locs_in_loop): Do not query all_refs_in_loop.
14690 2013-03-22  Richard Biener  <rguenther@suse.de>
14692         * is-a.h (as_a): Use gcc_checking_assert.
14694 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
14696         * config/aarch64/aarch64.c (aarch64_print_operand): New
14697         format specifier for printing a constant in hex.
14698         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14699         format specifier for printing second operand.
14701 2013-03-22  Richard Biener  <rguenther@suse.de>
14703         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14704         bitmaps.
14705         (gather_mem_refs_in_loops): Perform store accumulation here.
14706         (create_vop_ref_mapping_loop): Remove.
14707         (create_vop_ref_mapping): Likewise.
14708         (analyze_memory_references): Initialize refs_stored_in_loop.
14709         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14710         (record_indep_loop): Remove.
14711         (record_dep_loop): New function.
14712         (ref_indep_loop_p_1): Adjust to only walk over references
14713         in the loop, not its subloops.
14714         (ref_indep_loop_p): Rename to ...
14715         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14716         maintaining a more fine-grained cache.
14717         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14718         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14720 2013-03-22  Richard Biener  <rguenther@suse.de>
14722         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14723         (struct mem_ref): Make accesses_in_loop a vec of a vec of
14724         aggregate mem_ref_loc.
14725         (free_mem_ref_locs): Inline into ...
14726         (memref_free): ... this and adjust.
14727         (mem_ref_alloc): Adjust.
14728         (mem_ref_locs_alloc): Remove.
14729         (record_mem_ref_loc): Adjust.
14730         (get_all_locs_in_loop): Rewrite into ...
14731         (for_all_locs_in_loop): ... this iterator.
14732         (rewrite_mem_ref_loc): New functor.
14733         (rewrite_mem_refs): Use for_all_locs_in_loop.
14734         (sm_set_flag_if_changed): New functor.
14735         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14736         (ref_always_accessed): New functor.
14737         (ref_always_accessed_p): Use for_all_locs_in_loop.
14739 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14741         * tree-pass.h (PROP_gimple_lvec): New.
14742         * passes.c (dump_properties): Handle PROP_gimple_lvec.
14743         (init_optimization_passes): Move pass_lower_vector.
14744         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14745         PROP_gimple_lvec.
14746         (pass_lower_vector): Provide PROP_gimple_lvec.
14747         (pass_lower_vector_ssa): Likewise.
14748         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14750 2013-03-21  Mark Wielaard  <mjw@redhat.com>
14752         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14754 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14756         * config/i386/i386.md (*movdi_internal): Disparage slightly
14757         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14758         conversion alternatives.
14760 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14762         PR middle-end/48087
14763         * diagnostic.def (DK_WERROR): New kind.
14764         * diagnostic.h (werrorcount): Define.
14765         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14766         promoted to DK_ERROR, increment DK_WERROR counter instead of
14767         DK_ERROR counter.
14768         * toplev.c (toplev_main): Call print_ignored_options even if
14769         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14770         even if just werrorcount is non-zero.
14772         PR debug/55608
14773         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14774         on failure.
14775         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14776         (string_cst_pool_decl): New function.
14777         (optimize_one_addr_into_implicit_ptr): New function.
14778         (resolve_addr_in_expr): Optimize DWARF location expression
14779         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14780         which doesn't live in memory, but has DW_AT_location or
14781         DW_AT_const_value, or refers to a string literal, into
14782         DW_OP_GNU_implicit_pointer.
14783         (optimize_location_into_implicit_ptr): New function.
14784         (resolve_addr): If removing DW_AT_location of a variable because
14785         it was DW_OP_addr of address of the variable, but the variable doesn't
14786         live in memory, try to emit const value attribute for the initializer.
14788 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14790         * tree.h (VECTOR_TYPE_P): New macro.
14791         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14792         TYPE_MODE): Use it.
14793         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14794         VEC_COND_EXPR cannot be lvalues.
14795         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14797 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14799         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14800         Restrict the transformation to equal modes.
14802 2013-03-21  Richard Biener  <rguenther@suse.de>
14804         PR tree-optimization/39326
14805         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14806         (MEM_ANALYZABLE): Adjust.
14807         (record_mem_ref_loc): Move bitmap ops ...
14808         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14809         unanalyzable refs, do not record locations for it.
14810         (analyze_memory_references): Allocate ref zero as shared
14811         unanalyzable ref.
14812         (refs_independent_p): Do not test for unanalyzed mems here.
14813         (ref_indep_loop_p_1): Special-case disambiguation against
14814         the unanalyzed ref.
14815         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14817 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14819         * config/arm/arm-protos.h (tune_params): Add
14820         prefer_neon_for_64bits field.
14821         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14822         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14823         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14824         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14825         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14826         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14827         (arm_option_override): Handle -mneon-for-64bits new option.
14828         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14829         (prefer_neon_for_64bits): Declare new variable.
14830         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14831         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14832         (arch_enabled): Handle new arch types. Remove support for onlya8
14833         and nota8.
14834         (one_cmpldi2): Use new arch names.
14835         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14836         * config/arm/arm.opt (mneon-for-64bits): Add option.
14837         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14838         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14839         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14840         of onlya8.
14841         * doc/invoke.texi (-mneon-for-64bits): Document.
14843 2013-03-21  Richard Biener  <rguenther@suse.de>
14845         PR tree-optimization/39326
14846         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14847         (sort_bbs_in_loop_postorder_cmp): New function.
14848         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14850 2013-03-21  Richard Biener  <rguenther@suse.de>
14852         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14853         (vect_insert_into_interleaving_chain): Likewise.
14854         (vect_drs_dependent_in_basic_block): Inline ...
14855         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14856         split out from ...
14857         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14858         (vect_check_interleaving): Simplify.
14859         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14860         (vect_slp_analyze_data_ref_dependences): ... this new function.
14861         (dr_group_sort_cmp): New function.
14862         (vect_analyze_data_ref_accesses): Compute data-reference groups
14863         here instead of in vect_analyze_data_ref_dependence.  Use
14864         a more efficient algorithm.
14865         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14866         vect_slp_analyze_data_ref_dependences.  Call
14867         vect_analyze_data_ref_accesses earlier.
14868         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14869         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14870         (vect_slp_analyze_data_ref_dependences): New prototype.
14872 2013-03-21  Richard Biener  <rguenther@suse.de>
14874         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14875         ref is stored in the loop.
14876         (find_refs_for_sm): Walk only over all stores.
14877         (store_motion_loop): Allocate from lim_bitmap_obstack.
14878         (store_motion): Likewise.
14880 2013-03-21  Richard Biener  <rguenther@suse.de>
14882         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14883         Update virtual SSA form.
14885 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14887         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14888         * configure: Regenerate.
14889         * config.in: Regenerate.
14890         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14891         if !HAVE_LD_EH_FRAME_CIEV3.
14893 2013-03-21  Richard Biener  <rguenther@suse.de>
14895         * tree-cfg.c (verify_expr_no_block): New function.
14896         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14897         nor DECL_VALUE_EXPR have locations with associated blocks.
14898         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14899         (clear_unused_block_pointer): Remove code dealing with
14900         blocks in DECL_DEBUG_EXPR locations.
14902 2013-03-21  Richard Biener  <rguenther@suse.de>
14904         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14905         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14906         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14907         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14908         instead of DECL_DEBUG_EXPR_IS_FROM.
14909         * gimplify.c (gimplify_modify_expr): Likewise.
14910         * tree-cfg.c (verify_expr_location_1): Likewise.
14911         * tree-complex.c (create_one_component_var): Likewise.
14912         * tree-sra.c (create_access_replacement): Likewise.
14913         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14914         (clear_unused_block_pointer): Likewise.
14915         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14916         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14917         * var-tracking.c (var_debug_decl): Likewise.
14918         (track_expr_p): Likewise.
14919         * tree-inline.c (add_local_variables): Likewise.  Set
14920         DECL_HAS_DEBUG_EXPR_P after copying it.
14921         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14922         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14924 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14926         PR bootstrap/56656
14927         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14928         * configure: Regenerate.
14929         * config.in: Regenerate.
14930         * config/i386/i386.md (*movdf_internal): Use
14931         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14932         movd instead of movq mnemonic for interunit moves.
14933         (*movdi_internal): Ditto.
14935 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14937         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14938         (abd<mode>_3): New pattern.
14939         (aba<mode>_3): New pattern.
14940         (fabd<mode>_3): New pattern.
14942 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14944         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14945         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14946         occurrence of REGISTER_PREFIX as its empty string.
14948 2013-03-20  Jeff Law  <law@redhat.com>
14950         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14951         addititional equivalences for equality comparisons between an SSA_NAME
14952         and a constant where the SSA_NAME was set from a widening conversion.
14954 2013-03-20  Walter Lee  <walt@tilera.com>
14956         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14958 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14960         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14961         depending on type attribute.
14962         (*movti_internal): Ditto.
14963         (*movtf_internal): Ditto.
14964         (*movxf_internal): Ditto.
14965         (*movdf_internal): Ditto.
14966         (*movsf_internal): Ditto.
14968 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14970         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14971         maybe_vex for sselog1 and ssemov types.
14972         (*movdi_internal): Reorder operand constraints.
14973         (*movsi_internal): Ditto.  Set prefix attribute to
14974         maybe_vex for sselog1 and ssemov types.
14975         (*movtf_internal): Set prefix attribute to maybe_vex
14976         for sselog1 and ssemov types.
14977         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14978         DImode ssemov types.  Reorder operand constraints.
14979         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14980         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14981         attribute for SImode ssemov types.  Reorder operand constraints.
14983 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14985         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14986         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14988 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
14990         * config/rs6000/predicates.md (indexed_address, update_address_mem
14991         update_indexed_address_mem): New predicates.
14992         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14993         attribute for load/store instructions.
14994         * config/rs6000/dfp.md (movsd_store): Likewise.
14995         (movsd_load): Likewise.
14996         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14997         (unnamed HI->DI extend define_insn): Likewise.
14998         (unnamed SI->DI extend define_insn): Likewise.
14999         (unnamed QI->SI extend define_insn): Likewise.
15000         (unnamed QI->HI extend define_insn): Likewise.
15001         (unnamed HI->SI extend define_insn): Likewise.
15002         (unnamed HI->SI extend define_insn): Likewise.
15003         (extendsfdf2_fpr): Likewise.
15004         (movsi_internal1): Likewise.
15005         (movsi_internal1_single): Likewise.
15006         (movhi_internal): Likewise.
15007         (movqi_internal): Likewise.
15008         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
15009         attribute for load/store instructions.
15010         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
15011         instructions.
15012         (mov<mode>_softfloat): Likewise.
15013         (mov<mode>_hardfloat32): Likewise.
15014         (mov<mode>_hardfloat64): Likewise.
15015         (mov<mode>_softfloat64): Likewise.
15016         (movdi_internal32): Likewise.
15017         (movdi_internal64): Likewise.
15018         (probe_stack_<mode>): Likewise.
15020 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
15022         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15023         floating point, and decimal floating point to reload iterator.
15025         * config/rs6000/constraints.md (wl constraint): New constraints to
15026         return FLOAT_REGS if certain options are used to reduce the number
15027         of separate patterns that exist in the file.
15028         (wx constraint): Likewise.
15029         (wz constraint): Likewise.
15031         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15032         -mdebug=reg, print wg, wl, wx, and wz constraints.
15033         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15034         Initialize the reload functions for 64-bit binary/decimal floating
15035         point types.
15036         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15037         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15038         create the buffer on the stack to overcome not having a 32-bit
15039         load and store.
15040         (rs6000_emit_move): Likewise.
15041         (rs6000_secondary_memory_needed_rtx): Likewise.
15042         (rs6000_alloc_sdmode_stack_slot): Likewise.
15043         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15044         via xxlxor, just like DFmode 0.0.
15046         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15047         define as 1 if we are running on a power7 or newer.
15048         (enum r6000_reg_class_enum): Add new constraints.
15050         * config/rs6000/dfp.md (movsd): Delete, combine with binary
15051         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15052         with other moves by using conditional constraits (wg).  Use LFIWZX
15053         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
15054         (movsd splitter): Likewise.
15055         (movsd_hardfloat): Likewise.
15056         (movsd_softfloat): Likewise.
15058         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15059         binary and decimal floating point moves.
15060         (fmove_ok): New attributes to combine binary and decimal floating
15061         point moves, and to combine power6x (mfpgpr) moves along normal
15062         floating moves.
15063         (real_value_to_target): Likewise.
15064         (f32_lr): Likewise.
15065         (f32_lm): Likewise.
15066         (f32_li): Likewise.
15067         (f32_sr): Likewise.
15068         (f32_sm): Likewise.
15069         (f32_si): Likewise.
15070         (movsf): Combine binary and decimal floating point moves.  Combine
15071         power6x (mfpgpr) moves with other moves by using conditional
15072         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
15073         (mov<mode> for SFmode/SDmode); Likewise.
15074         (SFmode/SDmode splitters): Likewise.
15075         (movsf_hardfloat): Likewise.
15076         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15077         (movsf_softfloat): Likewise.
15078         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15080         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15081         wx and wz constraints.
15083         * config/rs6000/constraints.md (wg constraint): New constraint to
15084         return FLOAT_REGS if -mmfpgpr (power6x) was used.
15086         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15087         constraint.
15089         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15090         -mdebug=reg, print wg, wl, wx, and wz constraints.
15091         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15092         Initialize the reload functions for 64-bit binary/decimal floating
15093         point types.
15094         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15095         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15096         create the buffer on the stack to overcome not having a 32-bit
15097         load and store.
15098         (rs6000_emit_move): Likewise.
15099         (rs6000_secondary_memory_needed_rtx): Likewise.
15100         (rs6000_alloc_sdmode_stack_slot): Likewise.
15101         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15102         via xxlxor, just like DFmode 0.0.
15104         * config/rs6000/dfp.md (movdd): Delete, combine with binary
15105         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15106         with other moves by using conditional constraits (wg).  Use LFIWZX
15107         and STFIWX for loading SDmode on power7.
15108         (movdd splitters): Likewise.
15109         (movdd_hardfloat32): Likewise.
15110         (movdd_softfloat32): Likewise.
15111         (movdd_hardfloat64_mfpgpr): Likewise.
15112         (movdd_hardfloat64): Likewise.
15113         (movdd_softfloat64): Likewise.
15115         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15116         64-bit binary and decimal floating point moves.
15117         (FMOVE64X): Likewise.
15118         (movdf): Combine 64-bit binary and decimal floating point moves.
15119         Combine power6x (mfpgpr) moves with other moves by using
15120         conditional constraits (wg).
15121         (mov<mode> for DFmode/DDmode): Likewise.
15122         (DFmode/DDmode splitters): Likewise.
15123         (movdf_hardfloat32): Likewise.
15124         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15125         (movdf_softfloat32): Likewise.
15126         (movdf_hardfloat64_mfpgpr): Likewise.
15127         (movdf_hardfloat64): Likewise.
15128         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15129         (movdf_softfloat64): Likewise.
15130         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15131         (reload_<mode>_load): Move to later in the file so they aren't in
15132         the middle of the floating point move insns.
15133         (reload_<mode>_store): Likewise.
15135         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15136         constraint.
15138         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15139         constraint if -mdebug=reg.
15140         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15141         Enable using dd reload support if needed.
15143         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15144         binary and decimal floating point moves in rs6000.md.
15145         (movtd_internal): Likewise.
15147         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15148         decimal floating point moves.
15149         (movtf): Likewise.
15150         (movtf_internal): Likewise.
15151         (mov<mode>_internal, TDmode/TFmode): Likewise.
15152         (movtf_softfloat): Likewise.
15153         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15155         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15156         movdi_internal64, using wg constraint for move direct operations.
15157         (movdi_internal64): Likewise.
15159         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15160         MODES_TIEABLE_P for selected modes.  Print the numerical value of
15161         the various virtual registers. Use GPR/FPR first/last values,
15162         instead of hard coding the register numbers.  Print which modes
15163         have reload functions registered.
15164         (rs6000_option_override_internal): If -mdebug=reg, trace the options
15165         settings before/after setting cpu, target and subtarget settings.
15166         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15167         and for secondary reload failures in rs6000_secondary_reload_inner.
15168         (rs6000_secondary_reload_fail): Likewise.
15169         (rs6000_secondary_reload_inner): Likewise.
15171         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15172         macros for first/last GPR and FPR registers.
15173         (LAST_GPR_REGNO): Likewise.
15174         (FIRST_FPR_REGNO): Likewise.
15175         (LAST_FPR_REGNO): Likewise.
15177         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15178         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15179         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15180         (vcond<mode><mode>): Likewise.
15181         (vcondu<mode><mode>): Likewise.
15182         (vector_gtu<mode>): Likewise.
15183         (vector_gte<mode>): Likewise.
15184         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15185         to prevent the compiler from converting DImode operations to TImode.
15186         (ior<mode>3): Likewise.
15187         (and<mode>3): Likewise.
15188         (one_cmpl<mode>2): Likewise.
15189         (nor<mode>3): Likewise.
15190         (andc<mode>3): Likewise.
15192         * config/rs6000/constraints.md (wt constraint): New constraint
15193         that returns VSX_REGS if TImode is allowed in VSX registers.
15195         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15196         constant under VSX.
15198         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15199         similar to TImode, but it is restricted to being in the GPRs.
15201         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15202         TImode to occupy a single VSX register.
15204         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15205         -mvsx-timode for power7/power8.
15206         (power7 cpu): Likewise.
15207         (power8 cpu): Likewise.
15209         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15210         sure that TFmode/TDmode take up two registers if they are ever
15211         allowed in the upper VSX registers.
15212         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15213         registers.
15214         (rs6000_init_hard_regno_mode_ok): Likewise.
15215         (rs6000_debug_reg_global): Add debugging for PTImode and wt
15216         constraint.  Print if LRA is turned on.
15217         (rs6000_option_override_internal): Give an error if -mvsx-timode
15218         and VSX is not enabled.
15219         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
15220         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15221         to reg+offset addressing.  Use PTImode when checking offset
15222         addresses for validity.
15223         (reg_offset_addressing_ok_p): Likewise.
15224         (rs6000_legitimate_offset_address_p): Likewise.
15225         (rs6000_legitimize_address): Likewise.
15226         (rs6000_legitimize_reload_address): Likewise.
15227         (rs6000_legitimate_address_p): Likewise.
15228         (rs6000_eliminate_indexed_memrefs): Likewise.
15229         (rs6000_emit_move): Likewise.
15230         (rs6000_secondary_reload): Likewise.
15231         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
15232         reloads to fpr registers to continue to use reg+offset addressing,
15233         but 64-bit reloads to altivec registers need reg+reg addressing.
15234         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15235         it.  Treat LO_SUM like a PLUS operation.
15236         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15237         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15238         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15239         registers to share a register with a smaller sized type, since VSX
15240         puts scalars in the upper 64-bits.
15241         (print_operand): Add support for PTImode.
15242         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15243         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15244         registers, but don't have arithmetic support.
15245         (rs6000_memory_move_cost): Add test for VSX.
15246         (rs6000_opt_masks): Add -mvsx-timode.
15248         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15249         for TImode.
15250         (VSs): Likewise.
15251         (VSr): Use wt constraint for TImode.
15252         (VSv): Drop TImode support.
15253         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15254         (vsx_movti_64bit): Likewise.
15255         (vsx_movti_32bit): Likewise.
15256         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15257         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15258         one '?' on the appropriate output constraint.  Do not allow TImode
15259         logical operations on 32-bit systems.
15260         (vsx_ior<mode>3): Likewise.
15261         (vsx_xor<mode>3): Likewise.
15262         (vsx_one_cmpl<mode>2): Likewise.
15263         (vsx_nor<mode>3): Likewise.
15264         (vsx_andc<mode>3): Likewise.
15265         (vsx_concat_<mode>): Likewise.
15266         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15268         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15269         OPTION_MASK_VSX_TIMODE.
15270         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15271         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15273         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15274         (TI2 iterator): New iterator for TImode, PTImode.
15275         (wd mode attribute): Add values for vector types.
15276         (movti_string): Replace TI move operations with operations for TImode
15277         and PTImode.  Add support for TImode being allowed in VSX registers.
15278         (mov<mode>_string, TImode/PTImode): Likewise.
15279         (movti_ppc64): Likewise.
15280         (mov<mode>_ppc64, TImode/PTImode): Likewise.
15281         (TI mode splitters): Likewise.
15283         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15284         constraint.
15286 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
15288         PR tree-optimization/56355
15289         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15290         Also handle integers with undefined overflow.
15292 2013-03-20  Catherine Moore  <clm@codesourcery.com>
15293             Maciej W. Rozycki  <macro@codesourcery.com>
15294             Tom de Vries  <tom@codesourcery.com>
15295             Nathan Sidwell  <nathan@codesourcery.com>
15296             Iain Sandoe  <iain@codesourcery.com>
15297             Nathan Froyd  <froydnj@codesourcery.com>
15298             Chao-ying Fu  <fu@mips.com>
15300         * doc/extend.texi: (micromips, nomicromips, nocompression):
15301         Document new function attributes.
15302         * doc/invoke.texi (minterlink-compressed, mmicromips,
15303         m14k, m14ke, m14kec): Document new options.
15304         (minterlink-mips16): Update documentation.
15305         * doc/md.texi (ZC, ZD): Document new constraints.
15306         * configure.ac (gcc_cv_as_micromips): Check if linker
15307         supports the .set micromips directive.
15308         * configure: Regenerate.
15309         * config.in: Regenerate.
15310         * config/mips/mips-tables.opt: Regenerate.
15311         * config/mips/micromips.md: New file.
15312         * constraints.md (ZC, ZD): New constraints.
15313         * config/mips/predicates.md (movep_src_register): New predicate.
15314         (movep_src_operand): New predicate.
15315         (non_volatile_mem_operand): New predicate.
15316         * config/mips/mips.md (multimem): New type.
15317         (length): Differentiate between 17-bit and 18-bit branch offsets.
15318         (MOVEP1, MOVEP2): New mode iterator.
15319         (mov_<load>l): Use ZC constraint.
15320         (mov_<load>r): Likewise.
15321         (mov_<store>l): Likewise.
15322         (mov_<store>r): Likewise.
15323         (*branch_equality<mode>_inverted): Add microMIPS support.
15324         (*branch_equality<mode>): Likewise.
15325         (*jump_absolute): Likewise.
15326         (indirect_jump_<mode>): Likewise.
15327         (tablejump_<mode>): Likewise.
15328         (<optab>_internal): Likewise.
15329         (sibcall_internal): Likewise.
15330         (sibcall_value_internal): Likewise.
15331         (prefetch): Use constraint ZD.
15332         * config/mips/mips.opt (minterlink-compressed): New option.
15333         (minterlink-mips16): Now an alias for minterlink-compressed.
15334         (mmicromips): New option.
15335         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15336         (compare_and_swap_12): Likewise.
15337         (sync_add<mode>): Likewise.
15338         (sync_<optab>_12): Likewise.
15339         (sync_old_<optab>_12): Likewise.
15340         (sync_new_<optab>_12): Likewise.
15341         (sync_nand_12): Likewise.
15342         (sync_old_nand_12): Likewise.
15343         (sync_new_nand_12): Likewise.
15344         (sync_sub<mode>): Likewise.
15345         (sync_old_add<mode>): Likewise.
15346         (sync_old_sub<mode>): Likewise.
15347         (sync_new_add<mode>): Likewise.
15348         (sync_new_sub<mode>): Likewise.
15349         (sync_<optab><mode>): Likewise.
15350         (sync_old_<optab><mode>): Likewise.
15351         (sync_new_<optab><mode>): Likewise.
15352         (sync_nand<mode>): Likewise.
15353         (sync_old_nand<mode>): Likewise.
15354         (sync_new_nand<mode>): Likewise.
15355         (sync_lock_test_and_set<mode>): Likewise.
15356         (test_and_set_12): Likewise.
15357         (atomic_compare_and_swap<mode>): Likewise.
15358         (atomic_exchange<mode>_llsc): Likewise.
15359         (atomic_fetch_add<mode>_llsc): Likewise.
15360         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15361         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15362         (umips_save_restore_pattern_p): Likewise.
15363         (umips_load_store_pair_p): Likewise.
15364         (umips_output_load_store_pair): Likewise.
15365         (umips_movep_target_p): Likewise.
15366         (umips_12bit_offset_address_p): Likewise.
15367         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15368         (mips_base_mips16): Rename this...
15369         (mips_base_compression_flags): ...to this. Update all uses.
15370         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15371         (mips_mips16_decl_p): Delete.
15372         (mips_nomips16_decl_p): Delete.
15373         (mips_get_compress_on_flags): New function.
15374         (mips_get_compress_off_flags): New function.
15375         (mips_get_compress_mode): New function.
15376         (mips_get_compress_on_name): New function.
15377         (mips_get_compress_off_name): New function.
15378         (mips_insert_attributes): Support multiple compression types.
15379         (mips_merge_decl_attributes): Likewise.
15380         (umips_12bit_offset_address_p): New function.
15381         (mips_start_function_definition): Emit .set micromips directive.
15382         (mips_call_may_need_jalx_p): New function.
15383         (mips_function_ok_for_sibcall): Add microMIPS support.
15384         (mips_print_operand_punctuation): Support short delay slots and
15385         compact jumps.
15386         (umips_swm_mask, umips_swm_encoding): New.
15387         (umips_build_save_restore): New function.
15388         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15389         (was_mips16_p): Remove.
15390         (old_compression_mode): New.
15391         (mips_set_compression_mode): New function.
15392         (mips_set_current_function): Add microMIPS support.
15393         (mips_option_override): Likewise.
15394         (umips_save_restore_pattern_p): New function.
15395         (umips_output_save_restore): New function.
15396         (umips_load_store_pair_p_1): New function.
15397         (umips_load_store_pair_p): New function.
15398         (umips_output_load_store_pair_1): New function.
15399         (umips_output_load_store_pair): New function.
15400         (umips_movep_target_p) New function.
15401         (mips_prepare_pch_save): Add microMIPS support.
15402         * config/mips/mips.h (TARGET_COMPRESSION): New.
15403         (TARGET_CPU_CPP_BUILTINS): Update macro
15404         to use new compression flags and to support microMIPS.
15405         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15406         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15407         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15408         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15409         (ASM_SPEC): Support mmicromips and mno-micromips.
15410         (M16STORE_REG_P): New macro.
15411         (MIPS_CALL): Support TARGET_MICROMIPS.
15412         (MICROMIPS_J): New macro.
15413         (mips_base_mips16): Rename this...
15414         (mips_base_compression_flags): ...to this.
15415         (UMIPS_12BIT_OFFSET_P): New macro.
15416         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15417         (MULTILIB_DIRNAMES): Likewise.
15418 2013-03-20  Richard Biener  <rguenther@suse.de>
15420         PR tree-optimization/56661
15421         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15422         the result does not have to be distinct.
15424 2013-03-20  Richard Biener  <rguenther@suse.de>
15426         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15427         remap_gimple_op_r.
15429 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15430             Steven Bosscher  <steven@gcc.gnu.org>
15432         PR rtl-optimization/56605
15433         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15435 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15437         PR bootstrap/56656
15438         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15439         that require movd instead of movq.
15441 2013-03-20  Richard Biener  <rguenther@suse.de>
15443         * tree-ssa-structalias.c (struct variable_info): Add pointer
15444         to the first field of an aggregate with sub-vars.  Make
15445         this and the pointer to the next subfield its ID.
15446         (vi_next): New function.
15447         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15448         storedanything_id, integer_id): Increment by one.
15449         (new_var_info, get_call_vi, lookup_call_clobber_vi,
15450         get_call_clobber_vi): Adjust.
15451         (solution_set_expand): Simplify and speedup.
15452         (solution_set_add): Inline into ...
15453         (set_union_with_increment): ... this.  Adjust accordingly.
15454         (do_sd_constraint): Likewise.
15455         (do_ds_constraint): Likewise.
15456         (do_complex_constraint): Simplify.
15457         (build_pred_graph): Adjust.
15458         (solve_graph): Likewise.  Simplify and speedup.
15459         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15460         get_constraint_for_component_ref, get_constraint_for_1,
15461         first_vi_for_offset, first_or_preceding_vi_for_offset,
15462         create_function_info_for, create_variable_info_for_1,
15463         create_variable_info_for, intra_create_variable_infos): Adjust.
15464         (init_base_vars): Push NULL for ID zero.
15465         (compute_points_to_sets): Adjust.
15467 2013-03-20  Richard Biener  <rguenther@suse.de>
15469         * cfgloop.c (verify_loop_structure): Streamline and avoid
15470         ICEing on corrupt loop tree.
15471         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15472         loop tree.
15474 2013-03-20  Richard Biener  <rguenther@suse.de>
15476         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15477         check whether an SSA update is needed.
15479 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
15481         * config/mips/constraints.md (T): Rename to...
15482         (Yf): ...this.
15483         (U): Rename to...
15484         (Yd): ...this.
15485         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15486         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15488 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15490         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15491         (*subsi3_carryin_uxtw): Likewise.
15493 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15495         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15496         (*rorsi3_insn_uxtw): Likewise.
15498 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15500         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15501         (*extrsi5_insn_uxtw): Likewise.
15503 2013-03-19  Richard Biener  <rguenther@suse.de>
15505         PR tree-optimization/56273
15506         * passes.c (init_optimization_passes): Move second VRP after DOM.
15508 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
15510         * config/i386/i386.md (*movti_internal): Merge from
15511         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
15512         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
15513         nox64 isa attributes.
15515 2013-03-18  Richard Biener  <rguenther@suse.de>
15517         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15518         (unite): Likewise.
15519         (merge_node_constraints): Likewise.
15520         (build_succ_graph): Likewise.
15521         (valid_graph_edge): Inline into single caller.
15522         (unify_nodes): Likewise.  Use bitmap_set_bit return value
15523         and cache varinfo.
15524         (scc_visit): Fix formatting and variable use.
15525         (do_sd_constraint): Use gcc_checking_assert.
15526         (do_ds_constraint): Likewise.
15527         (do_complex_constraint): Likewise.
15528         (condense_visit): Likewise.  Cleanup.
15529         (dump_pred_graph): New function.
15530         (perform_var_substitution): Dump the pred-graph before
15531         variable substitution.
15532         (find_equivalent_node): Use gcc_checking_assert.
15533         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15535 2013-03-18  Richard Biener  <rguenther@suse.de>
15537         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15538         Remove cond_expr_stmt_list argument and do not gimplify the
15539         built expression.
15540         (vect_loop_versioning): Adjust.
15541         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15542         Cleanup to use less temporaries.
15543         (vect_create_data_ref_ptr): Cleanup.
15545 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15547         PR tree-optimization/56635
15548         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15549         require types_compatible_p types.
15551 2013-03-18  Nick Clifton  <nickc@redhat.com>
15553         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15554         spurious backslash.
15556         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15557         Add missing line to comment describing function.
15559 2013-03-18  Richard Biener  <rguenther@suse.de>
15561         PR tree-optimization/56210
15562         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15563         Handle string / character search functions.
15564         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15566 2013-03-18  Richard Biener  <rguenther@suse.de>
15568         PR middle-end/56483
15569         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15570         and implement properly.
15571         * gimple.h (gimple_cond_single_var_p): Remove.
15573 2013-03-18  Richard Biener  <rguenther@suse.de>
15575         * tree-data-ref.h (find_data_references_in_loop): Declare.
15576         * tree-data-ref.c (get_references_in_stmt): Use a stack
15577         vector pre-allocated in the callers.
15578         (find_data_references_in_stmt): Adjust.
15579         (graphite_find_data_references_in_stmt): Likewise.
15580         (create_rdg_vertices): Likewise.
15581         (find_data_references_in_loop): Export.
15582         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15583         Compute dependences here...
15584         (vect_analyze_data_refs): ...not here.  When we encounter
15585         a non-vectorizable data reference in basic-block vectorization
15586         truncate the data reference vector.  Do not bother to
15587         fixup data-dependence information for gather loads.
15588         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15589         of data references, as reported.
15591 2013-03-18  Richard Biener  <rguenther@suse.de>
15593         PR tree-optimization/3713
15594         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
15595         has_constants and expr.
15596         (stmt_has_constants): Properly valueize SSA names when deciding
15597         whether the stmt has constants.
15599 2013-03-18  Richard Biener  <rguenther@suse.de>
15601         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15602         whole function when there is nothing to do.
15603         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15604         * tree-vectorizer.c (vectorize_loops): Update virtual and
15605         loop-closed SSA once.
15606         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15608 2013-03-18  Richard Biener  <rguenther@suse.de>
15610         PR middle-end/56113
15611         * domwalk.c (bb_postorder): New global static.
15612         (cmp_bb_postorder): New function.
15613         (walk_dominator_tree): Replace scheme imposing an order for
15614         visiting dominator sons by one sorting them at the time they
15615         are pushed on the stack.
15617 2013-03-18  Richard Biener  <rguenther@suse.de>
15619         PR tree-optimization/39326
15620         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15621         (struct mem_ref): Replace mem member with ao_ref typed member.
15622         (MEM_ANALYZABLE): Adjust.
15623         (memref_eq): Likewise.
15624         (mem_ref_alloc): Likewise.
15625         (gather_mem_refs_stmt): Likewise.
15626         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15627         (execute_sm_if_changed_flag_set): Adjust.
15628         (execute_sm): Likewise.
15629         (ref_always_accessed_p): Likewise.
15630         (refs_independent_p): Likewise.
15631         (can_sm_ref_p): Likewise.
15633 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15635         PR c/56566
15636         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15637         return 1 even for !unsignedp.
15639 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15641         * config/i386/i386.md (isa): Add x64 and nox64.
15642         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15643         (*pushtf): Enable *roF alternative for x64 isa only.
15644         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
15645         mode attribute of integer alternatives to DImode for TARGET_64BIT.
15646         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
15647         (*movtf_internal): Merge from *movtf_internal_rex64 and
15648         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
15649         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
15650         nox64 isa attributes.
15651         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
15652         nox64 isa attributes.
15653         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15655 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15657         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15659 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
15661         PR target/56640
15662         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15663         class names.  Remove trailing comma after "ALL_REGS".
15665 2013-03-16  Jan Hubicka  <jh@suse.cz>
15667         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15668         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15669         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15670         of cgraph_get_create_node.
15671         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15673 2013-03-16  Jason Merrill  <jason@redhat.com>
15675         PR debug/49090
15676         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15677         with DW_AT_default_value.
15679 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
15681         * BASE-VER: Set to 4.9.0.
15683 2013-03-14  Andi Kleen  <ak@linux.intel.com>
15685         PR target/56619
15686         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15687         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15688         Document _x* TSX intrinsics.
15690 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
15691             David Holsgrove  <david.holsgrove@xilinx.com>
15693         * configure.ac: Add MicroBlaze TLS support detection.
15694         * configure: Regenerate.
15695         * config/microblaze/microblaze-protos.h
15696         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15697         symbol_mentioned_p, label_mentioned_p): Add prototypes.
15698         * config/microblaze/microblaze.c (microblaze_address_type): Add
15699         ADDRESS_TLS and tls_reloc address types.
15700         (microblaze_address_info): Add tls_reloc.
15701         (TARGET_HAVE_TLS): Define.
15702         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15703          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15704          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15705         load_tls_operand,  microblaze_call_tls_get_addr,
15706         microblaze_legitimize_tls_address): New functions.
15707         (microblaze_classify_unspec): Handle UNSPEC_TLS.
15708         (get_base_reg): Use microblaze_tls_symbol_p.
15709         (microblaze_classify_address): Handle TLS.
15710         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15711         label_mentioned_p and microblaze_tls_referenced_p.
15712         (microblaze_legitimize_address): Handle TLS.
15713         (microblaze_address_insns): Handle ADDRESS_TLS.
15714         (pic_address_needs_scratch): Handle TLS.
15715         (print_operand_address): Handle TLS.
15716         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15717         (microblaze_expand_move): Handle TLS.
15718         (microblaze_legitimate_constant_p): Check
15719         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15720         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15721         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15722         (PIC_OFFSET_TABLE_REGNUM): Set.
15723         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15724         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15725         (addsi3, movsi_internal2, movdf_internal): Update constraints
15726         * config/microblaze/predicates.md (arith_plus_operand): Define
15727         (move_operand): Redefine as move_src_operand,
15728         check microblaze_tls_referenced_p.
15730 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15732         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15733         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15735 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15737         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15738         CC mode for AND.
15740 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
15742         PR tree-optimization/53265
15743         * common.opt (Waggressive-loop-optimizations): New option.
15744         * tree-ssa-loop-niter.c: Include tree-pass.h.
15745         (do_warn_aggressive_loop_optimizations): New function.
15746         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
15747         if number_of_latch_executions returned constant.
15748         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15749         early.  If number_of_latch_executions returned constant, set
15750         nb_iterations_upper_bound back to it.
15751         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15752         field.
15753         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15754         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15756         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15757         (MULTILIB_OSDIRNAMES): Set.
15758         * genmultilib: If defaultosdirname doesn't start with :: , set
15759         defaultosdirname2 instead, clear it and emit two . multilib_raw
15760         entries instead of just one.
15762 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15764         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15765         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15766         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15767         (SUBTARGET_OVERRIDE_OPTIONS): New.
15769 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15771         PR target/49880
15772         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15773         (musermode): Convert to Var(TARGET_USERMODE).
15774         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15775         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15776         * config/sh/sh.c (sh_option_override): Use
15777         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15778         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15779         condition.
15780         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15781         TARGET_SH4.
15782         (udivsi3_i4_single, divsi3_i4_single): Use
15783         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15785 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15787         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15788         default setting.
15790 2013-03-13  Richard Biener  <rguenther@suse.de>
15792         PR tree-optimization/56608
15793         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15794         calls when vectorizing basic-blocks.
15796 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15798         PR plugins/45078
15799         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15800         tm_file.
15802 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15804         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15806 2013-03-11  Jan Hubicka  <jh@suse.cz>
15808         PR lto/56557
15809         * lto-streamer-out.c (output_symbol_p): Skip references from
15810         constructors of external variables.
15812 2013-03-11  Jan Hubicka  <jh@suse.cz>
15814         PR middle-end/56571
15815         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15816         from pseudos.
15817         * emit-rtl.c (verify_rtx_sharing): Likewise.
15818         (copy_insn_1): Likewise.
15819         * rtl.c (copy_rtx): Likewise.
15821 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15823         PR target/56591
15824         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15825         output_operand_lossage message.
15827 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15829         PR target/56470
15830         * arm.c (shift_op): Validate RTL pattern on the fly.
15831         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15832         the RTL.
15834 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15836         PR target/56347
15837         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15838         new call patterns that clobber %fr12.
15839         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15840         split and postreload patterns.
15841         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15842         registers %fr12 and %fr12R as call used.
15844 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15846         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15847         (canon_address, record_store, replace_read, check_mem_read_rtx,
15848         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15849         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15850         rest_of_handle_dse): Likewise.
15852 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15854         PR middle-end/56524
15855         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15856         Add base_optabs.
15857         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15858         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15859         (save_optabs_if_changed): Replace with...
15860         (init_tree_optimization_optabs): ...this.
15861         * optabs.c (save_optabs_if_changed): Rename to...
15862         (init_tree_optimization_optabs): ...this.  Take the optimization node
15863         as argument.  Do nothing if the base optabs are already correct.
15864         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15865         to recompute optabs.
15866         * function.h (function): Remove optabs field.
15867         * function.c (invoke_set_current_function_hook): Call
15868         init_tree_optimization_optabs.  Use the result to initialize
15869         this_fn_optabs.
15871 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15873         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15874         if GTMA_HAS_NO_INSTRUMENTATION.
15875         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15876         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15877         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15878         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15879         GTMA_HAS_NO_INSTRUMENTATION.
15881 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15883         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15884         libasan_preinit.o.
15886 2013-03-08  Marek Polacek  <polacek@redhat.com>
15887             Jakub Jelinek  <jakub@redhat.com>
15889         PR tree-optimization/56478
15890         * predict.c (is_comparison_with_loop_invariant_p): Change the
15891         type of loop_step to tree.
15892         (predict_loops): Adjust.
15893         (predict_iv_comparison): Perform the computations on double_ints.
15895 2013-03-08  Richard Biener  <rguenther@suse.de>
15897         PR tree-optimization/56570
15898         * tree-cfg.c (verify_expr_location_1): Verify locations for
15899         DECL_DEBUG_EXPR.
15900         * tree-sra.c (create_access_replacement): Strip locations
15901         from DECL_DEBUG_EXPRs.
15903 2013-03-08  Richard Biener  <rguenther@suse.de>
15905         * tree-inline.c (expand_call_inline): Do not associate
15906         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15907         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15909 2013-03-08  Richard Biener  <rguenther@suse.de>
15911         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15912         or block changes with -Og.  Fix for location / block encoding
15913         changes and PHI arguments with locations.
15915 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15917         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15918         for all counters.
15919         (struct output_info): Likewise.
15920         (register_overhead): Remove bad gcc_assert.
15921         (bitmap_find_bit): If there is only a single bitmap element, do not
15922         count a miss as a search.
15923         (print_statistics): Update for counter type changes.
15924         (dump_bitmap_statistics): Likewise.  Print headers such that they
15925         are properly lined up with the printed counters.
15927 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15929         PR tree-optimization/56559
15930         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15931         check that it has only a single use.
15933 2013-03-07  Richard Biener  <rguenther@suse.de>
15935         * doc/invoke.texi (fwhole-program): Discourage use in combination
15936         with -flto.
15938 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15940         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15942         PR tree-optimization/56539
15943         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15944         instead of GSI_CONTINUE_LINKING as last argument to
15945         force_gimple_operand_gsi.  Adjust function comment.
15947         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15948         aarch64-cores.def.
15950         PR middle-end/56548
15951         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15952         promoted mode, convert the result back to the original mode.
15954 2013-03-06  Richard Biener  <rguenther@suse.de>
15956         PR middle-end/56294
15957         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15958         (insert_updated_phi_nodes_compare_uids): New function.
15959         (update_ssa): Sort symbols_to_rename after UID before
15960         traversing it to insert PHI nodes.
15962 2013-03-06  Richard Biener  <rguenther@suse.de>
15964         PR middle-end/50494
15965         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15966         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15968         Revert
15969         2013-02-13  Richard Biener  <rguenther@suse.de>
15971         PR lto/50494
15972         * varasm.c (output_constant_def_1): Get the decl representing
15973         the constant as argument.
15974         (output_constant_def): Wrap output_constant_def_1.
15975         (make_decl_rtl): Use output_constant_def_1 with the decl
15976         representing the constant.
15977         (build_constant_desc): Optionally re-use a decl already
15978         representing the constant.
15979         (tree_output_constant_def): Adjust.
15981 2013-03-06  Joey Ye  <joey.ye@arm.com>
15983         PR lto/50293
15984         * gcc.c (convert_white_space): New function.
15985         (main): Handles white space in function name.
15987 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15989         PR target/56529
15990         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15991         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15992         to SH_DIV_CALL_TABLE for TARGET_SH2.
15993         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15994         list.
15995         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15996         call-table options.
15998 2013-03-05  Sterling Augustine  <saugustine@google.com>
15999             Cary Coutant  <ccoutant@google.com>
16001         PR debug/55364
16002         * dwarf2out.c (resolve_addr): Don't call
16003         remove_loc_list_addr_table_entries a second time for the same
16004         expression.
16006 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16008         PR debug/56510
16009         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
16010         (avoid_complex_debug_insns): New function.
16011         (expand_debug_locations): Call it.
16013         PR rtl-optimization/56484
16014         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16015         lifetimes of hard registers on small register class machines.
16017 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16019         * config/microblaze/microblaze-protos.h: Rename
16020         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16021         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16022         fast_interrupt.
16023         (microblaze_fast_interrupt_function_p): New function.
16024         (microblaze_is_interrupt_handler): Rename to
16025         microblaze_is_interrupt_variant and add fast_interrupt check.
16026         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16027         (save_restore_insns): Likewise.
16028         (compute_frame_size): Likewise.
16029         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16030         (microblaze_globalize_label): Likewise.
16031         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16032         * config/microblaze/microblaze.md: Use wrapper
16033         microblaze_is_interrupt_variant.
16035 2013-03-05  Kai Tietz  <ktietz@redhat.com>
16037         * sdbout.c (sdbout_one_type): Switch to current function's section
16038         supporting cold/hot.
16040 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16042         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16043         -mxl-reorder.
16045 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16047         PR middle-end/56461
16048         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16049         if VALGRIND_GET_VBITS is defined, temporarily make object
16050         memory all defined, and restore previous valgrind addressability
16051         and definability afterwards.  Free this_object at the end.
16053         PR middle-end/56461
16054         * lra.c (lra): Call lra_clear_live_ranges if live_p,
16055         right before calling lra_create_live_ranges, also call it
16056         when clearing live_p.  Only call lra_clear_live_ranges
16057         at the end if live_p.
16059         PR middle-end/56461
16060         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16062 2013-03-05  Richard Biener  <rguenther@suse.de>
16064         PR tree-optimization/56521
16065         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16066         value-id.
16068 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
16070         PR c++/55135
16071         * except.h (remove_unreachable_eh_regions): New prototype.
16072         * except.c (remove_eh_handler_splicer): New function, split out
16073         of remove_eh_handler.
16074         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
16075         warning about running it on many EH regions one at a time.
16076         (remove_unreachable_eh_regions_worker): New function, walk the
16077         EH tree in depth-first order and remove non-marked regions.
16078         (remove_unreachable_eh_regions): New function.
16079         * tree-eh.c (mark_reachable_handlers): New function, split out
16080         from remove_unreachable_handlers.
16081         (remove_unreachable_handlers): Use mark_reachable_handlers and
16082         remove_unreachable_eh_regions.
16083         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16084         and remove_unreachable_eh_regions.
16086 2013-03-05  Richard Biener  <rguenther@suse.de>
16088         PR middle-end/56525
16089         * loop-init.c (fix_loop_structure): Remove loops in two stages,
16090         not freeing them until the end.
16092 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16094         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16096 2013-03-05  Richard Biener  <rguenther@suse.de>
16098         PR tree-optimization/56270
16099         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16100         of loads after scheduling an SLP instance.
16102 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16104         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16105         tic6x.exp.
16106         (check_gcc_parallelize): Run guality.exp as a separate job from
16107         vect.exp with unsorted.exp and $(dg_target_exps) separately from
16108         struct-layout-1.exp with stackalign.exp.
16110         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16112         PR middle-end/56461
16113         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16114         load_index sbitmap even if some bit in it isn't set.
16116         PR middle-end/56461
16117         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16118         (discover_iteration_bound_by_body_walk): Change queues to
16119         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
16120         spelling in comment.  Call safe_push on queues[bound_index] directly.
16121         Release queues[queue_index] in every iteration unconditionally.
16122         Release bounds vector.
16124         PR middle-end/56461
16125         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16126         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16127         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16128         inner_phis vector.
16130 2013-03-05  Richard Biener  <rguenther@suse.de>
16132         PR lto/56515
16133         * tree-inline.c (remap_blocks_to_null): New function.
16134         (expand_call_inline): When expanding a call stmt without
16135         an associated block inline remap all callee blocks to NULL.
16137 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16139         PR rtl-optimization/56494
16140         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16141         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16142         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16144         PR middle-end/56461
16145         * sel-sched-ir.c (free_sched_pools): Release
16146         succs_info_pool.stack[succs_info_pool.max_top] vectors too
16147         if succs_info_pool.max_top isn't -1.
16149         PR bootstrap/56509
16150         * opts.c (opts_obstack, opts_concat): Moved to...
16151         * opts-common.c (opts_obstack, opts_concat): ... here.
16153 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16155         PR middle-end/56461
16156         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16158 2013-03-04  Martin Jambor  <mjambor@suse.cz>
16160         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16161         all appropriate places.
16163 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
16165         PR tree-optimization/56424
16166         * ipa-split.c (split_function): Do not set the RSO flag if result is
16167         not by reference and its type is a register type.
16169 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
16171         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16172         (microblaze_legitimate_pic_operand): Likewise
16173         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16174         new function microblaze_legitimate_pic_operand
16175         * config/microblaze/microblaze-protos.h
16176         (microblaze_legitimate_pic_operand): Declare.
16178 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16180         * config/microblaze/predicates.md (call_insn_simple_operand):
16181         New predicate for supported rtx code types.
16182         * config/microblaze/microblaze.md (call_internal1): Use
16183         call_insn_simple_operand predicate.
16185 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16187         PR middle-end/56461
16188         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16189         partitions.ordered_remove.
16191         PR middle-end/56461
16192         * tree-vect-stmts.c (vectorizable_conversion): Don't call
16193         vec_oprnds0.create (1) for modifier == NONE.
16195         PR middle-end/56461
16196         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16197         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16198         vec_oprnds1 right before pushing anything to it for
16199         scalar_shift_arg.
16201         PR middle-end/56461
16202         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16203         set nbbs to 0 instead of having separate code path.
16204         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16205         instead of false as last argument if returning NULL.
16207 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
16209         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16210         the attribute is now called "target" instead of "option".
16211         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16212         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
16213         attribute/pragma name for TARGET_OPTION_VALID_P and
16214         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
16215         * doc/tm.texi: Regenerated.
16217 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
16219         * config/microblaze/microblaze.c:
16220         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16221         * config/microblaze/microblaze.h: Add -mxl-reorder to
16222         DRIVER_SELF_SPECS.
16223         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16224         instructions emitted if TARGET_REORDER.
16225         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16226         or 0 for -m/-mno case, but initialises as 2 to detect default use case
16227         separately.
16229 2013-03-01  Xinliang David Li  <davidxl@google.com>
16231         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16232         walk length.
16234 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
16236         PR middle-end/56461
16237         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16238         vector even when returning true.  Fix up function comment formatting.
16240         PR middle-end/56461
16241         * ira-build.c (ira_loop_nodes_count): New variable.
16242         (create_loop_tree_nodes): Initialize it.
16243         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16245         PR middle-end/56461
16246         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16247         method on dr_chain and result_chain.
16248         * tree-vect-stmts.c (vectorizable_store): Only call
16249         result_chain.create if j == 0.
16251         PR middle-end/56461
16252         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16253         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16254         before overwriting it.
16256 2013-03-01  Tobias Burnus  <burnus@net-b.de>
16258         * doc/extended.texi (C Extensions): Change order in @menu
16259         to match @node.
16260         (Other MIPS Built-in Functions): Move last MIPS entry before
16261         "picoChip Built-in Functions".
16262         (SH Built-in Functions): Move after RX Built-in Functions.
16263         * doc/gcc.texi (Introduction): Change order in @menu
16264         to match @node.
16265         * doc/md.texi (Constraints): Ditto.
16266         * gty.texi (Type Information): Ditto.
16267         (User-provided marking routines for template types): Make
16268         subsection.
16269         * doc/invoke.texi (AArch64 Options): Move before
16270         "Adapteva Epiphany Options".
16272 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
16273             Jakub Jelinek  <jakub@redhat.com>
16275         PR sanitizer/56454
16276         * asan.c (gate_asan): Lookup no_sanitize_address instead of
16277         no_address_safety_analysis attribute.
16278         * doc/extend.texi (no_address_safety_attribute): Rename to
16279         no_sanitize_address attribute, mention no_address_safety_analysis
16280         attribute as deprecated alias.
16282 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
16284         PR middle-end/56461
16285         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16286         type to vec<vec<tree> > *.
16287         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
16288         to be vec<tree> instead of vec<tree> *, set vec_defs
16289         to vNULL and call vec_defs.create (number_of_vects), adjust other
16290         uses of vec_defs.
16291         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16292         vectorizable_condition): Adjust vect_get_slp_defs callers.
16294 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16296         * config/aarch64/aarch64.c
16297         (aarch64_float_const_representable): Remove unused variable.
16299 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16301         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16303 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16305         * config/aarch64/aarch64-builtins.c
16306         (aarch64_init_simd_builtins): Make static.
16308 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16310         * config/aarch64/aarch64.c
16311         (aarch64_simd_make_constant): Make static.
16313 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16315         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16316         with no initialization to the RHS of debug statements.
16318 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16320         PR tree-optimization/56294
16321         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16322         Adjust dumping.
16323         (get_access_replacement): Do not call create_access_replacement.
16324         Assert a replacement exists.
16325         (get_repl_default_def_ssa_name): Create the replacement declaration
16326         itself.
16328 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16330         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16331         final_end_function.
16333 2013-02-28  Marek Polacek  <polacek@redhat.com>
16335         PR rtl-optimization/56466
16336         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16337         if we're changing a loop.
16338         (peel_loops_completely): Likewise.
16340 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
16342         PR c++/55813
16343         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16345 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16347         PR target/56445
16348         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16349         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16350         INTX_FTYPE_FX, FX_FTYPE_INTX.
16351         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16353 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16355         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16356         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16357         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16358         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16359         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16360         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16361         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16362         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16363         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16364         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16365         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16366         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16367         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16368         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16369         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16370         (avrxmega6): Increase max flash segments from 5 to 6.
16371         * config/avr/t-multilib: Regenerate.
16372         * config/avr/avr-tables.opt: Regenerate.
16373         * doc/avr-mmcu.texi: Regenerate.
16375 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16377         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16378         (avr_device_to_arch): Rename to avr_device_to_ld.
16379         (avr_device_to_as): New prototype.
16380         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16381         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16382         * config/avr/driver-avr.c (avr_device_to_as): New.
16383         (avr_device_to_arch): Rename to avr_device_to_ld.
16385 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16387         PR middle-end/56461
16388         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16389         method on dr_chain and result_chain.
16391         PR middle-end/56461
16392         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16393         pointer_set_destroy on not_executed_last_iteration.
16395         PR middle-end/56461
16396         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16398         PR middle-end/56461
16399         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16400         FOR_EACH_DEFINED_FUNCTION when freeing state.
16402         PR middle-end/56461
16403         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16404         pool_free.
16405         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16406         overwriting it.
16408         PR middle-end/56461
16409         * ipa-cp.c (decide_whether_version_node): Call vec_free on
16410         known_aggs[i].items and release known_aggs vector.
16412         PR middle-end/56461
16413         * ipa-reference.c (propagate): Free node_info even for alias nodes.
16415 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16417         * config/microblaze/microblaze.c (microblaze_emit_compare):
16418         Use xor for EQ/NE comparisions.
16419         * config/microblaze/microblaze.md (cstoresf4): Add constraints
16420         (cbranchsf4): Adjust operator to comparison_operator.
16422 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16424         PR middle-end/56461
16425         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16426         vector.
16427         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16428         vec_safe_push, always update *slot.
16429         (redirect_edge_var_map_clear): Use vec_free.
16430         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16431         (free_var_map_entry): Use vec_free.
16432         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16433         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16435 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
16437         PR middle-end/45472
16438         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16439         when the may_trap_p bit of the exprs being merged differs.
16440         Reorder tests for speculativeness in the logical and operator.
16442 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16444         * incpath.c (add_standard_paths): Use reconcat instead of concat
16445         where appropriate and avoid leaking memory.
16447         * opts.h: Include obstack.h.
16448         (opts_concat): New prototype.
16449         (opts_obstack): New declaration.
16450         * opts.c (opts_concat): New function.
16451         (opts_obstack): New variable.
16452         (init_options_struct): Call gcc_init_obstack on opts_obstack.
16453         (finish_options): Use opts_concat instead of concat
16454         and XOBNEWVEC instead of XNEWVEC.
16455         * opts-common.c (generate_canonical_option, decode_cmdline_option,
16456         generate_option): Likewise.
16457         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16458         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16460         PR target/56455
16461         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16462         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16464 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16466         PR middle-end/56461
16467         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16469 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16471         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16472         (arm_block_move_unaligned_straight): Likewise.
16473         (arm_adjust_block_mem): Likewise.
16475 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16477         PR target/48901
16478         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16479         temp, cond and label.
16480         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16482         PR target/52500
16483         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16484         * config/c6x/c6x.h (dbx_register_map): Update declaration.
16486         PR target/52501
16487         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16488         of prologue/epilogue functions.
16490         PR target/52550
16491         * config/tilegx/tilegx.c (tilegx_expand_prologue):
16492         Remove unused variable cfa_offset.
16493         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16495         PR target/54639
16496         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16497         type promotion to unsigned.
16499         PR target/54640
16500         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16501         for HOST_WIDE_INT of 32 bit / same size as int.
16502         (arm_block_move_unaligned_straight): Likewise.
16503         (arm_adjust_block_mem): Likewise.
16505         PR target/54662
16506         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16507         ALL_CFLAGS.
16509 2013-02-26  Marek Polacek  <polacek@redhat.com>
16511         PR tree-optimization/56426
16512         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16514 2013-02-26  Richard Biener  <rguenther@suse.de>
16516         PR target/56444
16517         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16518         unused variable loops.
16520 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16522         PR tree-optimization/56448
16523         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16524         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16525         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16526         later operands of the references, or even first operand for
16527         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16529         PR tree-optimization/56443
16530         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16531         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16532         to type_for_mode langhook.
16534 2013-02-25  Matt Turner  <mattst88@gmail.com>
16536         * doc/invoke.texi: Document r4700.
16538 2013-02-25  Richard Biener  <rguenther@suse.de>
16540         PR tree-optimization/56175
16541         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16542         split out from ...
16543         (simplify_bitwise_binary): ... here.  Also guard the conversion
16544         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16546 2013-02-25  Catherine Moore  <clm@codesourcery.com>
16548         Revert:
16549         2013-02-24  Catherine Moore  <clm@codesourcery.com>
16550             Maciej W. Rozycki  <macro@codesourcery.com>
16551             Tom de Vries  <tom@codesourcery.com>
16552             Nathan Sidwell  <nathan@codesourcery.com>
16553             Iain Sandoe  <iain@codesourcery.com>
16554             Nathan Froyd  <froydnj@codesourcery.com>
16555             Chao-ying Fu  <fu@mips.com>
16557         * doc/extend.texi: (micromips, nomicromips, nocompression):
16558         Document new function attributes.
16559         * doc/invoke.texi (minterlink-compressed, mmicromips,
16560         m14k, m14ke, m14kec): Document new options.
16561         (minterlink-mips16): Update documentation.
16562         * doc/md.texi (ZC, ZD): Document new constraints.
16563         * configure.ac (gcc_cv_as_micromips): Check if linker
16564         supports the .set micromips directive.
16565         * configure: Regenerate.
16566         * config.in: Regenerate.
16567         * config/mips/mips-tables.opt: Regenerate.
16568         * config/mips/micromips.md: New file.
16569         * constraints.md (ZC, AD): New constraints.
16570         * config/mips/predicates.md (movep_src_register): New predicate.
16571         (movep_src_operand): New predicate.
16572         (non_volatile_mem_operand): New predicate.
16573         * config/mips/mips.md (multimem): New type.
16574         (length): Differentiate between 17-bit and 18-bit branch offsets.
16575         (MOVEP1, MOVEP2): New mode iterator.
16576         (mov_<load>l): Use ZC constraint.
16577         (mov_<load>r): Likewise.
16578         (mov_<store>l): Likewise.
16579         (mov_<store>r): Likewise.
16580         (*branch_equality<mode>_inverted): Add microMIPS support.
16581         (*branch_equality<mode>): Likewise.
16582         (*jump_absolute): Likewise.
16583         (indirect_jump_<mode>): Likewise.
16584         (tablejump_<mode>): Likewise.
16585         (<optab>_internal): Likewise.
16586         (sibcall_internal): Likewise.
16587         (sibcall_value_internal): Likewise.
16588         (prefetch): Use constraint ZD.
16589         * config/mips/mips.opt (minterlink-compressed): New option.
16590         (minterlink-mips16): Now an alias for minterlink-compressed.
16591         (mmicromips): New option.
16592         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16593         (compare_and_swap_12): Likewise.
16594         (sync_add<mode>): Likewise.
16595         (sync_<optab>_12): Likewise.
16596         (sync_old_<optab>_12): Likewise.
16597         (sync_new_<optab>_12): Likewise.
16598         (sync_nand_12): Likewise.
16599         (sync_old_nand_12): Likewise.
16600         (sync_new_nand_12): Likewise.
16601         (sync_sub<mode>): Likewise.
16602         (sync_old_add<mode>): Likewise.
16603         (sync_old_sub<mode>): Likewise.
16604         (sync_new_add<mode>): Likewise.
16605         (sync_new_sub<mode>): Likewise.
16606         (sync_<optab><mode>): Likewise.
16607         (sync_old_<optab><mode>): Likewise.
16608         (sync_new_<optab><mode>): Likewise.
16609         (sync_nand<mode>): Likewise.
16610         (sync_old_nand<mode>): Likewise.
16611         (sync_new_nand<mode>): Likewise.
16612         (sync_lock_test_and_set<mode>): Likewise.
16613         (test_and_set_12): Likewise.
16614         (atomic_compare_and_swap<mode>): Likewise.
16615         (atomic_exchange<mode>_llsc): Likewise.
16616         (atomic_fetch_add<mode>_llsc): Likewise.
16617         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16618         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16619         (umips_save_restore_pattern_p): Likewise.
16620         (umips_load_store_pair_p): Likewise.
16621         (umips_output_load_store_pair): Likewise.
16622         (umips_movep_target_p): Likewise.
16623         (umips_12bit_offset_address_p): Likewise.
16624         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16625         (mips_base_mips16): Rename this...
16626         (mips_base_compression_flags): ...to this. Update all uses.
16627         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16628         (mips_mips16_decl_p): Delete.
16629         (mips_nomips16_decl_p): Delete.
16630         (mips_get_compress_on_flags): New function.
16631         (mips_get_compress_off_flags): New function.
16632         (mips_get_compress_mode): New function.
16633         (mips_get_compress_on_name): New function.
16634         (mips_get_compress_off_name): New function.
16635         (mips_insert_attributes): Support multiple compression types.
16636         (mips_merge_decl_attributes): Likewise.
16637         (umips_12bit_offset_address_p): New function.
16638         (mips_start_function_definition): Emit .set micromips directive.
16639         (mips_call_may_need_jalx_p): New function.
16640         (mips_function_ok_for_sibcall): Add microMIPS support.
16641         (mips_print_operand_punctuation): Support short delay slots and
16642         compact jumps.
16643         (umips_swm_mask, umips_swm_encoding): New.
16644         (umips_build_save_restore): New function.
16645         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16646         (was_mips16_p): Remove.
16647         (old_compression_mode): New.
16648         (mips_set_compression_mode): New function.
16649         (mips_set_current_function): Add microMIPS support.
16650         (mips_option_override): Likewise.
16651         (umips_save_restore_pattern_p): New function.
16652         (umips_output_save_restore): New function.
16653         (umips_load_store_pair_p_1): New function.
16654         (umips_load_store_pair_p): New function.
16655         (umips_output_load_store_pair_1): New function.
16656         (umips_output_load_store_pair): New function.
16657         (umips_movep_target_p) New function.
16658         (mips_prepare_pch_save): Add microMIPS support.
16659         * config/mips/mips.h (TARGET_COMPRESSION): New.
16660         (TARGET_CPU_CPP_BUILTINS): Update macro
16661         to use new compression flags and to support microMIPS.
16662         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16663         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16664         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16665         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16666         (ASM_SPEC): Support mmicromips and mno-micromips.
16667         (M16STORE_REG_P): New macro.
16668         (MIPS_CALL): Support TARGET_MICROMIPS.
16669         (MICROMIPS_J): New macro.
16670         (mips_base_mips16): Rename this...
16671         (mips_base_compression_flags): ...to this.
16672         (UMIPS_12BIT_OFFSET_P): New macro.
16673         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16674         (MULTILIB_DIRNAMES): Likewise.
16676 2013-02-25  Tom de Vries  <tom@codesourcery.com>
16678         PR rtl-optimization/56131
16679         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16680         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16681         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16683 2013-02-25  Tobias Burnus  <burnus@net-b.de>
16685         * doc/invoke.texi (-fsanitize=): Move from optimization
16686         to debugging options.
16688 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16690         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16692 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16693             Alexander Monakov  <amonakov@ispras.ru>
16695         PR middle-end/56077
16696         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16697         flush pending lists also on non-jumps.  Adjust comment.
16699 2013-02-24  Catherine Moore  <clm@codesourcery.com>
16700             Maciej W. Rozycki  <macro@codesourcery.com>
16701             Tom de Vries  <tom@codesourcery.com>
16702             Nathan Sidwell  <nathan@codesourcery.com>
16703             Iain Sandoe  <iain@codesourcery.com>
16704             Nathan Froyd  <froydnj@codesourcery.com>
16705             Chao-ying Fu  <fu@mips.com>
16707         * doc/extend.texi: (micromips, nomicromips, nocompression):
16708         Document new function attributes.
16709         * doc/invoke.texi (minterlink-compressed, mmicromips,
16710         m14k, m14ke, m14kec): Document new options.
16711         (minterlink-mips16): Update documentation.
16712         * doc/md.texi (ZC, ZD): Document new constraints.
16713         * configure.ac (gcc_cv_as_micromips): Check if linker
16714         supports the .set micromips directive.
16715         * configure: Regenerate.
16716         * config.in: Regenerate.
16717         * config/mips/mips-tables.opt: Regenerate.
16718         * config/mips/micromips.md: New file.
16719         * constraints.md (ZC, AD): New constraints.
16720         * config/mips/predicates.md (movep_src_register): New predicate.
16721         (movep_src_operand): New predicate.
16722         (non_volatile_mem_operand): New predicate.
16723         * config/mips/mips.md (multimem): New type.
16724         (length): Differentiate between 17-bit and 18-bit branch offsets.
16725         (MOVEP1, MOVEP2): New mode iterator.
16726         (mov_<load>l): Use ZC constraint.
16727         (mov_<load>r): Likewise.
16728         (mov_<store>l): Likewise.
16729         (mov_<store>r): Likewise.
16730         (*branch_equality<mode>_inverted): Add microMIPS support.
16731         (*branch_equality<mode>): Likewise.
16732         (*jump_absolute): Likewise.
16733         (indirect_jump_<mode>): Likewise.
16734         (tablejump_<mode>): Likewise.
16735         (<optab>_internal): Likewise.
16736         (sibcall_internal): Likewise.
16737         (sibcall_value_internal): Likewise.
16738         (prefetch): Use constraint ZD.
16739         * config/mips/mips.opt (minterlink-compressed): New option.
16740         (minterlink-mips16): Now an alias for minterlink-compressed.
16741         (mmicromips): New option.
16742         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16743         (compare_and_swap_12): Likewise.
16744         (sync_add<mode>): Likewise.
16745         (sync_<optab>_12): Likewise.
16746         (sync_old_<optab>_12): Likewise.
16747         (sync_new_<optab>_12): Likewise.
16748         (sync_nand_12): Likewise.
16749         (sync_old_nand_12): Likewise.
16750         (sync_new_nand_12): Likewise.
16751         (sync_sub<mode>): Likewise.
16752         (sync_old_add<mode>): Likewise.
16753         (sync_old_sub<mode>): Likewise.
16754         (sync_new_add<mode>): Likewise.
16755         (sync_new_sub<mode>): Likewise.
16756         (sync_<optab><mode>): Likewise.
16757         (sync_old_<optab><mode>): Likewise.
16758         (sync_new_<optab><mode>): Likewise.
16759         (sync_nand<mode>): Likewise.
16760         (sync_old_nand<mode>): Likewise.
16761         (sync_new_nand<mode>): Likewise.
16762         (sync_lock_test_and_set<mode>): Likewise.
16763         (test_and_set_12): Likewise.
16764         (atomic_compare_and_swap<mode>): Likewise.
16765         (atomic_exchange<mode>_llsc): Likewise.
16766         (atomic_fetch_add<mode>_llsc): Likewise.
16767         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16768         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16769         (umips_save_restore_pattern_p): Likewise.
16770         (umips_load_store_pair_p): Likewise.
16771         (umips_output_load_store_pair): Likewise.
16772         (umips_movep_target_p): Likewise.
16773         (umips_12bit_offset_address_p): Likewise.
16774         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16775         (mips_base_mips16): Rename this...
16776         (mips_base_compression_flags): ...to this. Update all uses.
16777         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16778         (mips_mips16_decl_p): Delete.
16779         (mips_nomips16_decl_p): Delete.
16780         (mips_get_compress_on_flags): New function.
16781         (mips_get_compress_off_flags): New function.
16782         (mips_get_compress_mode): New function.
16783         (mips_get_compress_on_name): New function.
16784         (mips_get_compress_off_name): New function.
16785         (mips_insert_attributes): Support multiple compression types.
16786         (mips_merge_decl_attributes): Likewise.
16787         (umips_12bit_offset_address_p): New function.
16788         (mips_start_function_definition): Emit .set micromips directive.
16789         (mips_call_may_need_jalx_p): New function.
16790         (mips_function_ok_for_sibcall): Add microMIPS support.
16791         (mips_print_operand_punctuation): Support short delay slots and
16792         compact jumps.
16793         (umips_swm_mask, umips_swm_encoding): New.
16794         (umips_build_save_restore): New function.
16795         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16796         (was_mips16_p): Remove.
16797         (old_compression_mode): New.
16798         (mips_set_compression_mode): New function.
16799         (mips_set_current_function): Add microMIPS support.
16800         (mips_option_override): Likewise.
16801         (umips_save_restore_pattern_p): New function.
16802         (umips_output_save_restore): New function.
16803         (umips_load_store_pair_p_1): New function.
16804         (umips_load_store_pair_p): New function.
16805         (umips_output_load_store_pair_1): New function.
16806         (umips_output_load_store_pair): New function.
16807         (umips_movep_target_p) New function.
16808         (mips_prepare_pch_save): Add microMIPS support.
16809         * config/mips/mips.h (TARGET_COMPRESSION): New.
16810         (TARGET_CPU_CPP_BUILTINS): Update macro
16811         to use new compression flags and to support microMIPS.
16812         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16813         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16814         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16815         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16816         (ASM_SPEC): Support mmicromips and mno-micromips.
16817         (M16STORE_REG_P): New macro.
16818         (MIPS_CALL): Support TARGET_MICROMIPS.
16819         (MICROMIPS_J): New macro.
16820         (mips_base_mips16): Rename this...
16821         (mips_base_compression_flags): ...to this.
16822         (UMIPS_12BIT_OFFSET_P): New macro.
16823         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16824         (MULTILIB_DIRNAMES): Likewise.
16826 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16828         PR target/52555
16829         * target-globals.c (save_target_globals): For init_reg_sets and
16830         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16832 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16834         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16835         * config/aarch64/t-aarch64
16836         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16838 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16840         PR inline-asm/56148
16841         * lra-constraints.c (process_alt_operands): Reload operand
16842         conflicting with earlier clobber only if no more other conflicting
16843         operands.
16845 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16847         PR sanitizer/56393
16848         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16849         if not linking a shared library.
16851 2013-02-22  Seth LaForge  <sethml@google.com>
16853         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16855 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16857         * config/arm/arm.md (split for extendsidi): Update condition.
16858         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16859         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16860         (qhs_zextenddi_cstr): Likewise.
16862 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16864         PR middle-end/56420
16865         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16866         avoid signed wrapping.
16867         (expand_mult): Handle properly multiplication by
16868         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16869         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16870         in the compiler if coeff is HOST_WIDE_INT_MIN.
16871         (expand_divmod): Don't make ext_op1 static, change it's type to
16872         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16874         PR rtl-optimization/50339
16875         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16876         field.
16877         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16878         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16879         into splitting_ashiftrt field.
16880         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16881         ASHIFTRT.
16882         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16883         choices.
16885 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16887         PR middle-end/56108
16888         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16889         are sure to go irrevocable.
16891 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16893         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16894         scalars are valid operands.
16896 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16898         PR tree-optimization/56310
16899         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16900         only matching indices and non-negative final offsets.
16901         (intersect_aggregates_with_edge): Pass src_idx to
16902         agg_replacements_to_vector.  Pass src_idx insstead of index to
16903         intersect_with_agg_replacements.
16905 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16907         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16908         instead of hard-wired defaults.
16910 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16912         * doc/invoke.texi (MIPS Options): Update documentation of the
16913         floating-point multiply-accumulate instruction restrictions.
16915 2013-02-21  Kostya Serebryany  <kcc@google.com>
16917         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16918         asan_shadow_offset on x86_64 linux.
16920 2013-02-21  Richard Biener  <rguenther@suse.de>
16922         PR tree-optimization/56415
16923         Revert
16924         2013-02-11  Richard Biener  <rguenther@suse.de>
16926         PR tree-optimization/56273
16927         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16928         first VRP run.
16930 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16932         PR bootstrap/56258
16933         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16934         instead of @itemx.
16936         PR inline-asm/56405
16937         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16938         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16940 2013-02-20  Jan Hubicka  <jh@suse.cz>
16942         PR tree-optimization/56265
16943         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16944         when target is referenced for first time.
16946 2013-02-20  Richard Biener  <rguenther@suse.de>
16948         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16949         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16950         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16951         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16952         not return anything.
16953         (rename_ssa_copies): Do not remove unused locals.
16954         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16955         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16956         * passes.c (execute_function_todo): Do not schedule unused locals
16957         removal if cleanup_tree_cfg did something.
16958         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16959         about the number of removed locals.
16961 2013-02-20  Richard Biener  <rguenther@suse.de>
16963         PR tree-optimization/56398
16964         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16966 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16968         PR tree-optimization/55334
16969         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16970         restricted pointers to arrays.
16972 2013-02-20  Richard Biener  <rguenther@suse.de>
16973             Jakub Jelinek  <jakub@redhat.com>
16975         PR tree-optimization/56396
16976         * tree-ssa-ccp.c (n_const_val): New static variable.
16977         (get_value): Return NULL for SSA names we don't have a lattice
16978         entry for.
16979         (ccp_initialize): Initialize n_const_val.
16980         * tree-ssa-copy.c (n_copy_of): New static variable.
16981         (init_copy_prop): Initialize n_copy_of.
16982         (get_value): Return NULL_TREE for SSA names we don't have a
16983         lattice entry for.
16985 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16987         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16989 2013-02-20  Richard Biener  <rguenther@suse.de>
16991         * genpreds.c (write_lookup_constraint): Do not compare first
16992         letter of the constraint again.
16994 2013-02-20  Richard Biener  <rguenther@suse.de>
16996         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16997         and ceil_log2.
16998         (get_use_iv_cost): Terminate hashtable walk when coming across
16999         an empty entry.
17001 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
17003         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
17004         reassociation for avx2 targets.
17006 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17008         * config/microblaze/microblaze.c: microblaze_has_clz = 0
17009         Add version check for v8.10.a to enable microblaze_has_clz
17010         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
17011         version and TARGET_PATTERN_COMPARE check
17012         * config/microblaze/microblaze.md: New clzsi2 instruction
17014 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17016         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17017         function before branching.
17019 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17021         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17022         DUMP_INSN_RTX_UID.
17023         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17025 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17027         PR middle-end/55889
17028         * sel-sched.c: Include ira.h.
17029         (implicit_clobber_conflict_p): New function.
17030         (moveup_expr): Use it.
17031         * Makefile.in (sel-sched.o): Depend on ira.h.
17033 2013-02-19  Richard Biener  <rguenther@suse.de>
17035         PR tree-optimization/56384
17036         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17037         (vn_hash_type): Split out from ...
17038         (vn_hash_constant_with_type): ... here.
17039         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17040         (vn_phi_eq): Compare types from vn_phi_s structure.
17041         (vn_phi_lookup): Populate vn_phi_s type.
17042         (vn_phi_insert): Likewise.
17044 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
17046         PR tree-optimization/56350
17047         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17048         if haven't found reduction or nested cycle operand, rather than
17049         asserting we must find it.
17051         PR tree-optimization/56381
17052         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17053         to fold_build3.
17055 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
17056             Jakub Jelinek  <jakub@redhat.com>
17058         PR target/52555
17059         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17060         (swap_optab_enable): Same.
17061         (init_all_optabs): Use argument instead of global.
17062         * tree.h (struct tree_optimization_option): New field target_optabs.
17063         * expr.h (init_all_optabs): Add argument to prototype.
17064         (TREE_OPTIMIZATION_OPTABS): New.
17065         (save_optabs_if_changed): Protoize.
17066         * optabs.h: Declare this_fn_optabs.
17067         * optabs.c (save_optabs_if_changed): New.
17068         Declare this_fn_optabs.
17069         (init_optabs): Add argument to init_all_optabs() call.
17070         * function.c (invoke_set_current_function_hook): Handle per
17071         function optabs.
17072         * function.h (struct function): New field optabs.
17073         * config/mips/mips.c (mips_set_mips16_mode): Handle when
17074         optimization_current_node has changed.
17075         * target-globals.h (save_target_globals_default_opts): Protoize.
17076         * target-globals.c (save_target_globals_default_opts): New.
17078 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17080         PR target/56347
17081         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17082         registers %fr12 and %fr12R as call used.
17084         PR target/56214
17085         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17086         and HImode, require all displacements to be an integer multiple of
17087         their mode size.
17088         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17089         only allow QImode and HImode when reload is in progress and strict is
17090         true.  Likewise for symbolic addresses.  Use base14_operand to check
17091         displacements in REG+BASE addresses.
17093 2013-02-18  Richard Biener  <rguenther@suse.de>
17095         PR tree-optimization/56366
17096         * tree-vect-loop.c (get_initial_def_for_induction): Properly
17097         handle sign-conversion of outer-loop initial induction value.
17099 2013-02-18  Richard Biener  <rguenther@suse.de>
17101         PR middle-end/56349
17102         * cfghooks.c (merge_blocks): If we merge a latch into another
17103         block adjust references to it.
17104         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17105         (verify_loop_structure): Verify that a recorded latch is in fact
17106         a latch.
17108 2013-02-18  Richard Biener  <rguenther@suse.de>
17110         PR tree-optimization/56321
17111         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17112         order SSA name release and virtual operand unlinking.
17114 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17116         * config/microblaze/microblaze.md (save_stack_block): Define.
17117         (restore_stack_block): Likewise.
17119 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17121         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17122         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17123         * config/microblaze/microblaze.c (microblaze_option_override):
17124         Bail out early for PIC modes when target does not support PIC.
17126 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17128         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17129         Replace with a microblaze version.
17130         (microblaze_trampoline_init): Adapt for microblaze.
17131         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17132         microblaze.
17134 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
17135             Dodji Seketeli  <dodji@redhat.com>
17137         PR asan/56330
17138         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17139         (instrument_mem_region_access): Do not forget to always put
17140         instrumentation of the of 'base' and 'base + len' in a "if (len !=
17141         0) statement, even for cases where either 'base' or 'base + len'
17142         are not instrumented -- because they have been previously
17143         instrumented.  Simplify the logic by putting all the statements
17144         instrument 'base + len' inside a sequence, and then insert that
17145         sequence right before the current insertion point.  Then, to
17146         instrument 'base + len', just get an iterator on that statement.
17147         And do not forget to update the pointer to iterator the function
17148         received as argument.
17150 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
17152         PR rtl-optimization/56348
17153         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17155 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
17157         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17158         (clean_graph_dump_file): Pass base to start_graph_dump.
17160 2013-02-14  Richard Henderson  <rth@redhat.com>
17162         PR target/55941
17163         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17165 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
17167         * collect2-aix.h: Define F_LOADONLY.
17169 2013-02-14  Richard Biener  <rguenther@suse.de>
17171         PR lto/50494
17172         * varasm.c (output_constant_def_1): Get the decl representing
17173         the constant as argument.
17174         (output_constant_def): Wrap output_constant_def_1.
17175         (make_decl_rtl): Use output_constant_def_1 with the decl
17176         representing the constant.
17177         (build_constant_desc): Optionally re-use a decl already
17178         representing the constant.
17179         (tree_output_constant_def): Adjust.
17181 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
17183         Fix an asan crash
17184         * asan.c (instrument_builtin_call):  Really put the length of the
17185         second source argument into src1_len.
17187 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17189         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17190         argument.  If it is false, don't create edge from then_bb to
17191         fallthru_bb.
17192         (insert_if_then_before_iter): Pass true to it.
17193         (build_check_stmt): Pass false to it.
17194         (transform_statements): Flush hash table only on extended basic
17195         block boundaries, rather than at the beginning of every bb.
17196         Don't flush hash table on nonfreeing_call_p calls.
17197         * tree-flow.h (nonfreeing_call_p): New prototype.
17198         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17200 2013-02-13  David S. Miller  <davem@davemloft.net>
17202         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17204 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
17206         PR target/56184
17207         * ira.c (max_regno_before_ira): Move from ...
17208         (ira): ... here.
17209         (fix_reg_equiv_init): Use max_regno_before_ira instead of
17210         vec_safe_length.
17212 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17214         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17216 2013-02-13  Richard Biener  <rguenther@suse.de>
17218         PR lto/56295
17219         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17220         globals in MEM_REFs.
17222 2013-02-13  Richard Biener  <rguenther@suse.de>
17224         * loop-init.c (loop_optimizer_init): Clear loop state when
17225         re-initializing preserved loops.
17226         * loop-unswitch.c (unswitch_single_loop): Return whether
17227         we unswitched the loop.  Do not verify loop state here.
17228         (unswitch_loops): When we unswitched a loop discover new loops.
17230 2013-02-13  Kostya Serebryany  <kcc@google.com>
17232         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17233         on x86_64 linux.
17234         * sanitizer.def: Rename __asan_init to __asan_init_v1.
17236 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
17238         Avoid instrumenting duplicated memory access in the same basic block
17239         * Makefile.in (asan.o): Add new dependency on hash-table.h
17240         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17241         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17242         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17243         (free_mem_ref_resources, has_mem_ref_been_instrumented)
17244         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17245         (get_mem_ref_of_assignment): New functions.
17246         (get_mem_refs_of_builtin_call): Extract from
17247         instrument_builtin_call and tweak a little bit to make it fit with
17248         the new signature.
17249         (instrument_builtin_call): Use the new
17250         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
17251         of is_gimple_builtin_call.
17252         (instrument_derefs, instrument_mem_region_access): Insert the
17253         instrumented memory reference into the hash table.
17254         (maybe_instrument_assignment): Renamed instrument_assignment into
17255         this, and change it to advance the iterator when instrumentation
17256         actually happened and return true in that case.  This makes it
17257         homogeneous with maybe_instrument_assignment, and thus give a
17258         chance to callers to be more 'regular'.
17259         (transform_statements): Clear the memory reference hash table
17260         whenever we enter a new BB, when we cross a function call, or when
17261         we are done transforming statements.  Use
17262         maybe_instrument_assignment instead of instrumentation.  No more
17263         need to special case maybe_instrument_assignment and advance the
17264         iterator after calling it; it's now handled just like
17265         maybe_instrument_call.  Update comment.
17267 2013-02-13  Richard Biener  <rguenther@suse.de>
17269         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17270         Fix loop discovery code.
17272 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
17274         PR inline-asm/56148
17275         * lra-constraints.c (process_alt_operands): Match early clobber
17276         operand with itself.  Check conflicts with earlyclobber only if
17277         the operand is not reloaded.  Prefer to reload conflicting operand
17278         if earlyclobber and matching operands are the same.
17280 2013-02-12  Richard Biener  <rguenther@suse.de>
17282         PR lto/56297
17283         * lto-streamer-out.c (write_symbol): Do not output symbols
17284         for hard register variables.
17286 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
17288         PR target/54222
17289         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17290         (umulsidi3_insn, mulsidi3_insn): New insns.
17292 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
17294         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17295         (struct tune_params): Add vec_costs field.
17296         * config/arm/arm.c (arm_builtin_vectorization_cost)
17297         (arm_add_stmt_cost): New functions.
17298         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17299         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17300         (arm_default_vec_cost): New struct of type cpu_vec_costs.
17301         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17302         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17303         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17304         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17306 2013-02-12  Richard Biener  <rguenther@suse.de>
17308         PR lto/56295
17309         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17310         decls again if possible.
17312 2013-02-12  Richard Biener  <rguenther@suse.de>
17314         PR middle-end/56288
17315         * tree-ssa.c (verify_ssa_name): Fix check, move
17316         SSA_NAME_IN_FREE_LIST check up.
17318 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
17319             Steven Bosscher   <steven@gcc.gnu.org>
17321         PR rtl-optimization/56151
17322         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17323         equal to op0 or op1, and last_insn pattern is CODE operation
17324         with MEM dest and one of the operands matches that MEM.
17326 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
17328         * doc/extend.texi: Document Function Multiversioning and "default"
17329         parameter string to target attribute.
17330         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17331         target attribute parameter is "default".
17332         (ix86_compare_version_priority): Remove checks for target attribute.
17333         (ix86_mangle_function_version_assembler_name): Change error to sorry.
17334         Remove check for target attribute equal to NULL. Add assert.
17335         (ix86_generate_version_dispatcher_body): Change error to sorry.
17337 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
17338             Jack Howarth  <howarth@bromo.med.uc.edu>
17339             Patrick Marlier  <patrick.marlier@gmail.com>
17341         PR libitm/55693
17342         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17343         define ENDFILE_SPEC as TM_DESTRUCTOR.
17344         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17346 2013-02-11  Alexander Potapenko  <glider@google.com>
17347             Jack Howarth  <howarth@bromo.med.uc.edu>
17348             Jakub Jelinek  <jakub@redhat.com>
17350         PR sanitizer/55617
17351         * config/darwin.c (cdtor_record): Rename ctor_record.
17352         (sort_cdtor_records): Rename sort_ctor_records.
17353         (finalize_dtors): New routine to sort destructors by
17354         priority before use in assemble_integer.
17355         (machopic_asm_out_destructor): Use finalize_dtors if needed.
17357 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
17359         PR rtl-optimization/56275
17360         * simplify-rtx.c (avoid_constant_pool_reference): Check that
17361         offset is non-negative and less than cmode size before
17362         calling simplify_subreg.
17364 2013-02-11  Richard Biener  <rguenther@suse.de>
17366         PR tree-optimization/56264
17367         * cfgloop.h (fix_loop_structure): Adjust prototype.
17368         * loop-init.c (fix_loop_structure): Return the number of
17369         newly discovered loops.
17370         * tree-cfgcleanup.c (repair_loop_structures): When new loops
17371         are discovered, do a full loop-closed SSA rewrite.
17373 2013-02-11  Richard Biener  <rguenther@suse.de>
17375         PR tree-optimization/56273
17376         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17377         first VRP run.
17378         (check_array_ref): Fix missing newline in dumps.
17379         (search_for_addr_array): Likewise.
17381 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
17383         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17385 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
17387         PR target/56256
17388         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17390 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
17392         PR rtl-optimization/56246
17393         * lra-constraints.c (simplify_operand_subreg): Try to reuse
17394         reload pseudo.
17395         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17396         constraints are satisfied.
17398 2013-02-08  Jeff Law  <law@redhat.com>
17400         PR debug/53948
17401         * emit-rtl.c (reg_is_parm_p): New function.
17402         * regs.h (reg_is_parm_p): New prototype.
17403         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17404         callee-clobbered registers.
17406 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
17408         PR target/56043
17409         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17410         If there is no implicit builtin declaration, just return NULL.
17412 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17414         * config/i386/sse.md (FMAMODEM): New mode iterator.
17415         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17416         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17418 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17420         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17421         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17422         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17424 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17426         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17427         (microblaze*-*-elf): Likewise.
17428         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17429         LINK_SPEC.
17430         * config/microblaze/microblaze-c.c: Add builtin defines for
17431         _LITTLE_ENDIAN and _BIG_ENDIAN.
17432         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17433         add to TARGET_DEFAULT flags.
17434         Expand ASM_SPEC and LINK_SPEC.
17435         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17436         * config/microblaze/microblaze.md: Update extendsidi2 and
17437         movdi_internal instructions to use low-order / high-order reg
17438         print_operands.
17439         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17440         options and inversemask / mask of LITTLE_ENDIAN.
17441         * config/microblaze/t-microblaze: Expand multilib options to
17442         include mlittle-endian (le) and update exceptions patterns.
17444 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
17446         PR rtl-optimization/56195
17447         * lra-constraints.c (get_reload_reg): Don't reuse regs
17448         if they have smaller mode than requested, if they have
17449         wider mode than requested, try to return a SUBREG.
17451         PR tree-optimization/56250
17452         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17453         if type is unsigned and code isn't MULT_EXPR.
17455 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17457         PR tree-optimization/56064
17458         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17459         bits according to mode.
17460         * fixed-value.h (fixed_from_double_int)
17461         (const_fixed_from_double_int): Adjust comments.
17463 2013-02-08  Richard Biener  <rguenther@suse.de>
17465         PR lto/56231
17466         * lto-streamer.h (struct data_in): Remove current_file, current_line
17467         and current_col members.
17468         * lto-streamer-out.c (lto_output_location): Stream changed bits
17469         en-block for efficiency.
17470         * lto-streamer-in.c (clear_line_info): Remove.
17471         (lto_input_location): Cache current file, line and column
17472         globally via local statics.  Read changed bits en-block.
17473         (input_function): Do not call clear_line_info.
17474         (lto_read_body): Likewise.
17475         (lto_input_toplevel_asms): Likewise.
17477 2013-02-08  Michael Matz  <matz@suse.de>
17479         PR tree-optimization/52448
17480         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17481         (nt_call_phase): New static.
17482         (add_or_mark_expr): Only mark accesses with newer phase than any
17483         call seen.
17484         (nonfreeing_call_p): New.
17485         (nt_init_block): Update nt_call_phase, mark blocks as visited.
17486         (nt_fini_block): Keep blocks marked as visited.
17487         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17489 2013-02-08  Richard Biener  <rguenther@suse.de>
17491         * ira.c (ira): Free broken dominator information.
17493 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17495         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17497 2013-02-08  Marek Polacek  <polacek@redhat.com>
17499         * cfgloop.c (verify_loop_structure): Add more checking of headers.
17501 2013-02-08  Richard Biener  <rguenther@suse.de>
17503         PR middle-end/56181
17504         * cfgloop.h (flow_loops_find): Adjust.
17505         (bb_loop_header_p): Declare.
17506         * cfgloop.c (bb_loop_header_p): New function split out from ...
17507         (flow_loops_find): ... here.  Adjust function signature,
17508         support incremental loop structure update.
17509         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
17510         * cfgloopmanip.c (fix_loop_structure): Move ...
17511         * loop-init.c (fix_loop_structure): ... here.
17512         (apply_loop_flags): Split out from ...
17513         (loop_optimizer_init): ... here.
17514         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
17515         in incremental mode, only remove dead loops here.
17517 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17519         PR target/54222
17520         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17521         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17522         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17523         (*round<mode>3.libgcc): New insns for fixed-modes.
17524         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17525         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17526         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17527         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17528         implementations.  Define to __builtin_avr_absFX,
17529         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17530         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17531         __builtin_avr_countlsFX, respectively.
17532         * config/avr/avr-c.c (target.h): Include it.
17533         (enum avr_builtin_id): New enum.
17534         (avr_resolve_overloaded_builtin): New static function.
17535         (avr_register_target_pragmas): Use it to set
17536         targetm.resolve_overloaded_builtin.
17537         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17538         tree nodes used by DEF_BUILTIN.
17539         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17540         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17541         <AVR_BUILTIN_xxBITS>: Same.
17543 2013-02-08  Richard Biener  <rguenther@suse.de>
17545         * cfgloop.c (verify_loop_structure): Properly handle
17546         a loop exiting to another loop header.
17547         * ira-int.h (ira_loops): Remove.
17548         * ira.c (ira_loops): Remove.
17549         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17550         (do_reload): Use loop_optimizer_finalize.
17551         * ira-build.c (create_loop_tree_nodes): Use get_loops and
17552         number_of_loops to access the loop tree.
17553         (more_one_region_p): Likewise.
17554         (finish_loop_tree_nodes): Likewise.
17555         (rebuild_regno_allocno_maps): Likewise.
17556         (mark_loops_for_removal): Likewise.
17557         (mark_all_loops_for_removal): Likewise.
17558         (remove_unnecessary_regions): Likewise.
17559         (ira_build): Likewise.
17560         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17562 2013-02-08  Richard Biener  <rguenther@suse.de>
17564         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17565         * ipa-pure-const.c (analyze_function): Avoid calling
17566         mark_irreducible_loops twice.
17567         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17569 2013-02-07  David S. Miller  <davem@davemloft.net>
17571         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17572         on 'reg'.
17573         * var-tracking.c (vt_add_function_parameter): Test the presence of
17574         HAVE_window_save properly and do not remap argument registers when
17575         we have a leaf function.
17577 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
17579         PR bootstrap/56227
17580         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17581         instead of "ll".
17582         * config/i386/i386.c (ix86_print_operand): Ditto.
17584 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17586         * lra-constraints.c (process_alt_operands): Fix recently added comment.
17588 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17590         PR rtl-optimization/56225
17591         * lra-constraints.c (process_alt_operands): Check that reload hard
17592         reg can hold value for strict_low_part.
17594 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17596         PR debug/56154
17597         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17598         dwarf2out_end_function.
17599         (in_first_function_p, maybe_at_text_label_p,
17600         first_loclabel_num_not_at_text_label): New variables.
17601         (dwarf2out_var_location): In the first function find out
17602         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17603         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17604         functions.
17606 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
17608         PR rtl-optimization/56178
17609         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17610         SUBREG of a register.  Tidy up related block of code.
17611         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17612         note if the source is a register or a SUBREG of a register.
17614 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17616         PR target/56228
17617         * config/rs6000/rs6000.md (ptrm): New mode attr.
17618         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17619         call_value_indirect_aix<pttrsize>,
17620         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17621         m in constraints.
17623 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17625         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17626         if -bnortl. Convert to strcmp and strncmp.
17628 2013-02-07  Alan Modra  <amodra@gmail.com>
17630         PR target/54009
17631         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17632         addresses won't wrap when offsetting.
17633         (rs6000_secondary_reload): Provide secondary reloads needed for
17634         wrapping LO_SUM addresses.
17636 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
17638         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17639         MACH, just __MACH__.
17641 2013-02-06  Richard Biener  <rguenther@suse.de>
17643         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17644         instead of calling fix_loop_structure.
17646 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
17648         PR middle-end/56217
17649         * omp-low.c (use_pointer_for_field): Return false if
17650         lower_send_shared_vars doesn't generate any copy-out code.
17652 2013-02-06  Tom de Vries  <tom@codesourcery.com>
17654         PR rtl-optimization/56131
17655         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17656         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
17657         of the label is NULL.  Add comment.
17659 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
17661         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17663         PR sanitizer/55374
17664         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17665         (STATIC_LIBTSAN_LIBS): Likewise.
17666         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17667         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17668         is defined, don't add anything else beyond that.
17669         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17670         (LINK_COMMAND_SPEC): Use them.
17672         PR tree-optimization/56205
17673         * tree-stdarg.c (check_all_va_list_escapes): Return true if
17674         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17675         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17677 2013-02-05  Richard Biener  <rguenther@suse.de>
17679         PR tree-optimization/53342
17680         PR tree-optimization/53185
17681         * tree-vectorizer.h (vect_check_strided_load): Remove.
17682         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17683         not disallow peeling for vectorized strided loads.
17684         (vect_check_strided_load): Make static and simplify.
17685         (vect_analyze_data_refs): Adjust.
17686         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17687         correctly when vectorizing strided loads.
17689 2013-02-05  Richard Biener  <rguenther@suse.de>
17691         * doc/install.texi: Refer to ISL, not PPL.
17693 2013-02-05  Jan Hubicka  <jh@suse.cz>
17695         PR tree-optimization/55789
17696         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17698 2013-02-05  Jan Hubicka  <jh@suse.cz>
17700         PR tree-optimization/55789
17701         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17702         the dead call anyway.
17704 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
17706         PR sanitizer/55374
17707         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17709 2013-02-04  Alexander Potapenko  <glider@google.com>
17710             Jack Howarth  <howarth@bromo.med.uc.edu>
17711             Jakub Jelinek  <jakub@redhat.com>
17713         PR sanitizer/55617
17714         * config/darwin.c (sort_ctor_records): Stabilized qsort
17715         on constructor priority by using original position.
17716         (finalize_ctors): New routine to sort constructors by
17717         priority before use in assemble_integer.
17718         (machopic_asm_out_constructor): Use finalize_ctors if needed.
17720 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
17722         PR libstdc++/54314
17723         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17724         about visibility on artificial decls.
17725         * config/sol2.c (solaris_assemble_visibility): Likewise.
17727 2013-02-04  Kai Tietz  <ktietz@redhat.com>
17729         PR target/56186
17730         * config/i386/i386.c (function_value_ms_64): Add additional valtype
17731         argument and improve checking of return-argument types for 16-byte
17732         modes.
17733         (ix86_function_value_1): Add additional valtype argument on call
17734         of function_value_64.
17735         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17736         handling infunction_value_64 function.
17738 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
17740         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17742 2013-02-04  Richard Biener  <rguenther@suse.de>
17744         PR tree-optimization/56188
17745         * tree-ssa-structalias.c (label_visit): Consider case with
17746         initially non-empty points-to set.
17747         (perform_var_substitution): Dump node mapping and clean up.
17749 2013-02-04  Richard Guenther  <rguenther@suse.de>
17751         PR lto/56168
17752         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17753         node prevail as last resort.
17754         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17755         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17757 2013-02-04  Richard Biener  <rguenther@suse.de>
17759         PR tree-optimization/56113
17760         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17761         Merge into ...
17762         (equiv_class_lookup_or_add): ... this.
17763         (label_visit): Adjust and fix error in previous patch.
17764         (perform_var_substitution): Adjust.
17766 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17768         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17769         * config/sh/sh4-300.md: Likewise.
17770         * config/sh/sh4a.md: Likewise.
17771         * config/sh/constraints.md: Likewise.
17772         * config/sh/sh.md: Likewise.
17773         * config/sh/netbsd-elf.h: Likewise.
17774         * config/sh/predicates.md: Likewise.
17775         * config/sh/sh-protos.h: Likewise.
17776         * config/sh/ushmedia.h: Likewise.
17777         * config/sh/linux.h: Likewise.
17778         * config/sh/sh.c: Likewise.
17779         * config/sh/superh.h: Likewise.
17780         * config/sh/elf.h: Likewise.
17781         * config/sh/sh4.md: Likewise.
17782         * config/sh/sh.h: Likewise.
17784 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17786         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17787         constraint to match_test floating_point_store_memory_operand().
17788         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17789         (base14_operand): New.
17790         (floating_point_store_memory_operand): New.
17791         (integer_store_memory_operand): Revise to use base14_operand and
17792         reg_plus_base_memory_operand.
17793         (move_dest_operand): Allow symbolic_memory_operands.
17794         (symbolic_memory_operand): Check for LO_SOM.
17795         (symbolic_operand): Change default case to break.
17796         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17797         CONST_DOUBLE values to be reloaded by putting them into memory when
17798         the destination is a floating point register.
17799         (movdf): Remove code to handle CONST_DOUBLE.
17800         (movsf): Likewise.
17801         (reload_indf_r1): New.
17802         (reload_insf_r1): New.
17803         Consistently use "Q" and "T" constraints with integer and floating
17804         point move instructions, respectively.
17805         (movdi): Remove FAIL.
17806         Change predicate for source operand unamed DImode move from
17807         general_operand to move_src_operand.
17808         (umulsidi3): Change predicate for destination operand to
17809         register_operand.
17810         Likewise for similar unamed patterns.
17811         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17812         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17813         (hppa_legitimize_address): Simplify mask calculation.
17814         (pa_emit_move_sequence): Revised handling of secondary reloads from
17815         REG+D addresses for floating point loads and stores.  Directly handle
17816         loading CONST0_RTX (mode) to a floating point register.
17817         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17818         to floating point registers.  Don't restrict secondary reloads to
17819         floating point registers to integer modes.  Revise some comments and
17820         cleanup some code.
17821         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17822         (pa_legitimate_address_p): New.
17823         (pa_legitimize_reload_address): New.
17824         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17825         (STRICT_REG_OK_FOR_BASE_P): New.
17826         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17827         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17829 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17830             Andrew Dixie  <andrewd@gentrack.com>
17832         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17833         flag set.
17835 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17837         * expmed.c (extract_bit_field_1): Pass the full width of the
17838         structure to get_best_reg_extraction_insn.
17840 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17842         PR target/54601
17843         * configure.ac (use_cxa_atexit): Add AIX.
17844         * configure: Regenerate.
17846         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17848 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17850         PR debug/54793
17851         * final.c (need_profile_function): New variable.
17852         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17853         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17854         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17855         notes, targetm.asm_out.function_prologue doesn't emit anything,
17856         HAVE_prologue and profiler should be emitted before prologue,
17857         set need_profile_function instead of emitting it.
17858         (final_scan_insn): If need_profile_function, emit
17859         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17860         NOTE_INSN_FUNCTION_BEG note.
17862 2013-02-01  Richard Henderson  <rth@redhat.com>
17864         * config/rs6000/rs6000.md (smulditi3): New.
17865         (umulditi3): New.
17867         * config/alpha/alpha.md (umulditi3): New.
17869 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17871         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17872         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17874 2013-02-01  Richard Biener  <rguenther@suse.de>
17876         PR tree-optimization/56113
17877         * tree-ssa-structalias.c (label_visit): Reduce work for
17878         single-predecessor nodes.
17880 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17882         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17883         range isn't testing for zero.
17885 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17887         PR middle-end/56113
17888         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17890 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17891             Nick Clifton  <nickc@redhat.com>
17893         * config/v850/constraints.md (Q): Define as a memory constraint.
17894         * config/v850/predicates.md (label_ref_operand): New predicate.
17895         (e3v5_shift_operand): New predicate.
17896         (ior_operator): New predicate.
17897         * config/v850/t-v850: Add e3v5 multilib.
17898         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17899         (v850_gen_movdi): Prototype.
17900         * config/v850/v850.c: Add support for e3v5 architecture.
17901         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17902         TARGET_V850E_UP.
17903         (construct_save_jarl): Add e3v5 long JARL support.
17904         (v850_adjust_insn_length): New function.  Adjust length of call
17905         insns when using e3v5 instructions.
17906         (v850_gen_movdi): New function: Generate instructions to move a
17907         DImode value.
17908         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17909         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17910         (TARGET_USE_FPU): Enable for e3v5.
17911         (CONST_OK_FOR_W): New macro.
17912         (ADJUST_INSN_LENGTH): Define.
17913         * config/v850/v850.md (UNSPEC_LOOP): Define.
17914         (attr cpu): Add v850e3v5.
17915         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17916         (movdi): New pattern.
17917         (movdi_internal): New pattern.
17918         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17919         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17920         (cstoresf4): Likewise.
17921         (cstoredf4): Likewise.
17922         (insv): New pattern.
17923         (rotlso3_a): New pattern.
17924         (rotlsi3_b): New pattern
17925         (rotlsi3_v850e3v5): New pattern.
17926         (doloop_begin): New pattern.
17927         (fix_loop_counter): New pattern.
17928         (doloop_end): New pattern.
17929         (branch_normal): Add e3v5 long branch support.
17930         (branch_invert): Likewise.
17931         (branch_z_normal): Likewise.
17932         (branch_z_invert): Likewise.
17933         (branch_nz_normal): Likewise.
17934         (branch_nz_invert): Likewise.
17935         (call_internal_short): Add e3v5 register-indirect JARL support.
17936         (call_internal_long): Likewise.
17937         (call_value_internal_short): Likewise.
17938         (call_value_internal_long): Likewise.
17939         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17940         (mloop): New option.
17941         * config.gcc: Add support for configuring v840e3v5 target.
17942         * doc/invoke.texi: Document new v850 specific command line options.
17944 2013-01-31  Paul Koning  <ni1d@arrl.net>
17946         PR debug/55059
17947         PR debug/54508
17948         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17949         children if parent is a class.
17950         (prune_unused_types_prune): Don't add DW_AT_declaration.
17952 2013-01-31  Richard Biener  <rguenther@suse.de>
17954         PR tree-optimization/56157
17955         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17956         match up operand with SLP child.
17958 2013-01-31  Jason Merrill  <jason@redhat.com>
17960         PR debug/54410
17961         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17962         parameters the first time.
17963         (gen_scheduled_generic_parms_dies): Check completeness here.
17965 2013-01-31  Richard Biener  <rguenther@suse.de>
17967         PR middle-end/53073
17968         * common.opt (faggressive-loop-optimizations): New flag,
17969         enabled by default.
17970         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17971         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17972         infer_loop_bounds_from_undefined by it.
17974 2013-01-31  Richard Biener  <rguenther@suse.de>
17976         PR tree-optimization/56150
17977         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17978         visit virtual operands.
17979         (find_uses_to_rename_bb): Likewise.
17981 2013-01-31  Richard Biener  <rguenther@suse.de>
17983         PR tree-optimization/56150
17984         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17985         mixed store non-store stmts.
17987 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17989         PR sanitizer/55374
17990         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17991         LIBASAN_EARLY_SPEC is defined.
17992         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17993         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17994         before %o.
17995         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17997         PR c++/55742
17998         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17999         invalid args instead of ICEing on it.
18000         (ix86_valid_target_attribute_tree): Return error_mark_node if
18001         ix86_valid_target_attribute_inner_p failed.
18002         (ix86_valid_target_attribute_p): Return false only if
18003         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
18004         target("default") attribute.
18005         (sorted_attr_string): Change argument from const char * to tree,
18006         merge in all target attribute arguments rather than just one.
18007         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
18008         instead of free.  Avoid using strcat.
18009         (ix86_mangle_function_version_assembler_name): Mangle
18010         target("default") as if no target attribute is present.  Adjust
18011         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
18012         instead of xmalloc and XDELETEVEC instead of free.
18013         (ix86_function_versions): Don't return true if one of the decls
18014         doesn't have target attribute.  If they don't and one of the decls
18015         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
18016         sorted_attr_string caller.  Use XDELETEVEC instead of free.
18017         (ix86_supports_function_versions): Remove.
18018         (make_name): Fix up formatting.
18019         (make_dispatcher_decl): Remove resolver_name and its initialization.
18020         Avoid leaking memory.
18021         (is_function_default_version): Return true if there is
18022         target("default") attribute rather than no target attribute at all.
18023         (make_resolver_func): Avoid leaking memory.
18024         (ix86_generate_version_dispatcher_body): Likewise.
18025         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18026         * target.def (supports_function_versions): Remove.
18027         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18028         * doc/tm.texi: Regenerated.
18030 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
18032         PR rtl-optimization/56144
18033         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18034         for values with side effects.
18036 2013-01-30  Richard Biener  <rguenther@suse.de>
18038         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18039         (sparseset_pop): Likewise.
18040         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
18041         to be able to use quick_push in the worker loop.
18043 2013-01-30  Marek Polacek  <polacek@redhat.com>
18045         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18047 2013-01-30  Richard Biener  <rguenther@suse.de>
18049         PR lto/56147
18050         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18052 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
18054         PR tree-optimization/56064
18055         * fixed-value.c (fixed_from_double_int): New function.
18056         * fixed-value.h (fixed_from_double_int): New prototype.
18057         (const_fixed_from_double_int): New static inline function.
18058         * fold-const.c (native_interpret_fixed): New static function.
18059         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18060         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18061         (native_encode_fixed): New static function.
18062         (native_encode_expr) <FIXED_CST>: Use it.
18063         (native_interpret_int): Move double_int worker code to...
18064         * double-int.c (double_int::from_buffer): ...this new static method.
18065         * double-int.h (double_int::from_buffer): Prototype it.
18067 2013-01-30  Richard Biener  <rguenther@suse.de>
18069         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18070         New pointer-map and obstack.
18071         (init_alias_vars): Allocate pointer-map and obstack.
18072         (delete_points_to_sets): Free them.
18073         (find_what_var_points_to): Cache result.
18074         (find_what_p_points_to): Adjust for changed interface of
18075         find_what_var_points_to.
18076         (compute_points_to_sets): Likewise.
18077         (ipa_pta_execute): Likewise.
18079 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18081         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18082         * configure: Regenerate.
18083         * config.in: Regenerate.
18084         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18085         #nobits/#progbits if supported.
18087 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
18089         PR target/56121
18090         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18091         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18092         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18094 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18096         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18097         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18099 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18101         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18102         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18104 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18106         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18107         declaration.
18108         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18109         * config/arm/cortex-a7.md: New bypasses using
18110         arm_mac_accumulator_is_result.
18112 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18114         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
18115         (cortex_a7_neon_mla): Likewise.
18116         (cortex_a7_fpfmad): New reservation.
18117         (cortex_a7_fpmacs): Use ffmas and update required units.
18118         (cortex_a7_fpmuld): Update required units and latency.
18119         (cortex_a7_fpmacd): Likewise.
18120         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18121         (cortex_a7_neon). Likewise.
18122         (bypass) Update participating units.
18124 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18126         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18127         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18128         from fmac to ffma.
18129         * config/arm/vfp11.md (vfp_farith): Use ffmas.
18130         (vfp_fmul): Use ffmad.
18131         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18132         (cortex_r4_fmacd): Use ffmad.
18133         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18134         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
18135         (cortex_a9_fmacd): Use ffmad.
18136         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18137         (cortex_a8_vfp_macd): Use ffmad.
18138         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18139         (cortex_a5_fpmacd): Use ffmad.
18140         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18141         (cortex_a15_vfp_macd): Use ffmad.
18142         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18144 2013-01-29  Jason Merrill  <jason@redhat.com>
18146         PR libstdc++/54314
18147         * varasm.c (default_assemble_visibility): Don't warn about
18148         visibility on artificial decls.
18150 2013-01-29  Richard Biener  <rguenther@suse.de>
18152         PR tree-optimization/56113
18153         * tree-ssa-structalias.c (equiv_class_lookup): Also return
18154         the bitmap leader.
18155         (label_visit): Free duplicate bitmaps and record the leader instead.
18156         (perform_var_substitution): Adjust.
18158 2013-01-29  Richard Biener  <rguenther@suse.de>
18160         PR tree-optimization/55270
18161         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18162         the CFG, schedule loops for fixup.
18164 2013-01-29  Nick Clifton  <nickc@redhat.com>
18166         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18167         SP_REG.
18169 2013-01-28  Leif Ekblad  <leif@rdos.net>
18171         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18172         * config/i386/i386.h (TARGET_RDOS): New macro.
18173         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18174         * config/i386/i386.c (ix86_option_override_internal): For 64bit
18175         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18176         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18177         DEFAULT_LARGE_SECTION_THRESHOLD.
18178         * config/i386/i386.md (R14_REG, R15_REG): New constants.
18179         * config/i386/rdos.h: New file.
18180         * config/i386/rdos64.h: New file.
18182 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
18184         PR other/54814
18185         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18186         TEST_HARD_REG_BIT.
18188 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18190         PR rtl-optimization/56117
18191         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18192         call cselib_lookup_from_insn on the MEM before calling
18193         add_insn_mem_dependence.
18195 2013-01-28  Richard Biener  <rguenther@suse.de>
18197         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18198         to a stmt that didn't have one.
18199         (copy_phis_for_bb): Likewise for PHI arguments.
18200         (copy_debug_stmt): Likewise for debug stmts.
18202 2013-01-28  Richard Biener  <rguenther@suse.de>
18204         PR tree-optimization/56034
18205         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18206         (partition_builtin_p): Adjust.
18207         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
18208         it is the last partition.
18209         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18210         up the vertex for the definition.
18211         (classify_partition): Classify whether a partition is a
18212         PKIND_REDUCTION, thus has uses outside of the loop.
18213         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18214         Merge all PKIND_REDUCTION partitions into the last partition.
18215         (tree_loop_distribution): Seed partitions from reductions as well.
18217 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18219         PR tree-optimization/56125
18220         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18221         pow(x,c) into sqrt(x) * powi(x, n/2) or
18222         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18223         optimizing for size.
18224         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18225         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18226         integer.
18228         PR tree-optimization/56094
18229         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18230         to UNKNOWN_LOCATION while gimplifying expr.
18232 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
18234         PR target/56114
18235         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18236         operand 0 in movabs insn template for -masm=intel asm alternative.
18237         (*movabs<mode>_2): Ditto for operand 1.
18239 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
18241         PR target/54663
18242         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18243         of microblaze-c.o
18245 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18247         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18248         tm_file.
18250 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18252         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18253         Undef to avoid warning.
18255 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
18257         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18258         * configure: Regenerate.
18260 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
18262         PR tree-optimization/56098
18263         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18264         for stmts with volatile ops.
18265         (cond_store_replacement): Don't optimize if assign has volatile ops.
18266         (cond_if_else_store_replacement_1): Don't optimize if either
18267         then_assign or else_assign have volatile ops.
18268         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18269         volatile ops.
18271 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18273         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18275 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18277         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18278         missing ':' in asm example.
18280 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
18282         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18283         entries into lane and laneq entries.
18284         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18285         Remove AdvSIMD scalar modes.
18286         (aarch64_sq<r>dmulh_laneq<mode>): New.
18287         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18288         modes.
18289         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18290         builtin implementations to relfect changes in RTL in aarch64-simd.md.
18291         * config/aarch64/iterators.md (VCOND): New.
18292         (VCONQ): New.
18294 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18296         PR target/54222
18297         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18298         Add NULL LIBNAME argument to existing definitions.
18299         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18300         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18301         * config/avr/avr.c (DEF_BUILTIN): Same.
18302         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18303         (avr_expand_builtin): Expand to a vanilla call if a libgcc
18304         implementation is available (DECL_ASSEMBLER_NAME is set).
18305         (avr_fold_absfx): New static function.
18306         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18307         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18308         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18309         AVR_BUILTIN_ABSLLK.
18310         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18311         (abshk, absk, abslk, absllk): Provide as static inline functions.
18313 2013-01-25  Marek Polacek  <polacek@redhat.com>
18315         PR tree-optimization/56035
18316         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18318 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
18320         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18321         (*movtf_internal_rex64): Add (!o,C) alternative
18322         (*movxf_internal_rex64): Ditto.
18323         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18325 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
18327         * doc/invoke.texi: fix typo.
18328         * doc/objc.texi: fix typo.
18330 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
18332         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18333         for the first two alternatives.
18335 2013-01-24  Diego Novillo  <dnovillo@google.com>
18337         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
18338         (ggc-zone.o): Remove.
18339         * configure.ac: Remove option --with-gc.
18340         * configure: Re-generate.
18341         * doc/install.texi: Remove documentation for --with-gc.
18342         * gengtype.c (write_enum_defn): Remove.  Update all users.
18343         (write_Types_process_field): Remove generation of gt_e_* argument.
18344         (output_type_enum): Remove.  Update all users.
18345         (write_enum_defn): Remove.  Update all users.
18346         (enum alloc_zone): Remove.  Update all users.
18347         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18348         * ggc-common.c (ggc_splay_alloc): Remove first argument.
18349         Update all callers.
18350         (struct ptr_data): Remove field TYPE.  Update all users.
18351         (gt_pch_note_object): Remove argument TYPE.  Update all users.
18352         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18353         Update all users.
18354         * ggc-none.c (ggc_alloc_typed_stat): Remove.
18355         (struct alloc_zone): Remove.
18356         (ggc_internal_alloc_zone_stat): Remove.
18357         (ggc_internal_cleared_alloc_zone_stat): Remove.
18358         * ggc-page.c (ggc_alloc_typed_stat): Remove.
18359         (ggc_pch_count_object): Remove last argument.  Update all users.
18360         (ggc_pch_alloc_object): Remove last argument.  Update all users.
18361         (struct alloc_zone): Remove.
18362         * ggc-zone.c: Remove.
18363         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
18364         (struct alloc_zone): Remove.
18365         (ggc_alloc_typed_stat): Remove.
18366         (ggc_alloc_typed): Remove.
18367         (ggc_splay_alloc): Remove first argument.
18368         (rtl_zone): Remove.  Update all users.
18369         (tree_zone): Remove.  Update all users.
18370         (tree_id_zone): Remove.  Update all users.
18371         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
18372         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
18373         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
18374         * tree-ssanames.c: Remove references to zone allocator in comments.
18376 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
18378         * config/avr/avr.c (avr_out_fract): Make register numbers that
18379         might be outside of source operand signed.
18381 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
18383         * config/i386/constraints.md (Yf): New constraint.
18384         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18385         of f constraint to conditionaly disable x87 register preferences.
18386         (*movdf_internal): Ditto.
18387         (*movsf_internal): Ditto.
18389 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
18391         PR inline-asm/55934
18392         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18393         that have operands with impossible constraints.
18394         Add a FIXME for a speed-up opportunity.
18395         * lra-constraints.c (process_alt_operands): Verify that a class
18396         selected from constraints on asms is valid for the operand mode.
18397         (curr_insn_transform): Remove incorrect comment.
18399 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
18401         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18402         TOC operand is a valid symbol ref in the constant pool.
18404 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18406         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18408 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
18410         PR target/54222
18411         * config/avr/stdfix.h: New file.
18412         * t-avr (stdfix-gcc.h): New rule to build it.
18413         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18415 2013-01-23  Kostya Serebryany  <kcc@google.com>
18417         * config/darwin.h: remove dependency on
18418         CoreFoundation (asan on Mac OS).
18420 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
18422         PR target/49069
18423         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18424         instead of cmpdi_operand for first comparison operand.
18425         Don't assert that comparison operands aren't both constants.
18427 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18429         * doc/install.texi (Downloading the Source): Update references to
18430         downloading separate components.
18432 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18434         * doc/extend.texi (__int128): Improve grammar.
18436 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
18438         PR target/56028
18439         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18440         alternative to (o,r).
18441         (*movdi_internal_rex64): Remove (!o,n) alternative.
18442         (DImode immediate->memory splitter): Remove.
18443         (DImode immediate->memory peephole2): Remove.
18444         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18445         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18446         alternative to (!o,*r).
18447         (*movtf_internal_sse): New pattern.
18448         (*movxf_internal_rex64): New pattern.
18449         (*movxf_internal): Disable for TARGET_64BIT.
18450         (*movdf_internal_rex64): Remove (!o,F) alternative.
18452 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
18454         PR middle-end/56074
18455         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18456         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18457         * tree-vect-loop-manip.c (find_loop_location): Also ignore
18458         stmt locations where LOCATION_LOCUS of the stmt location is
18459         UNKNOWN_LOCATION or BUILTINS_LOCATION.
18461         PR target/55686
18462         * config/i386/i386.md (UNSPEC_STOS): New.
18463         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18464         *strsetqi_1): Add UNSPEC_STOS.
18466 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
18468         PR c++/56067
18469         * doc/invoke.texi: Remove left over -Wsynth example.
18471 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18473         PR tree-optimization/56051
18474         * fold-const.c (fold_binary_loc): Don't fold
18475         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18476         a narrowing conversion, or widening conversion from signed
18477         to unsigned.
18479 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18481         PR rtl-optimization/56023
18482         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18483         dependent on debug instruction.
18485 2013-01-21  Martin Jambor  <mjambor@suse.cz>
18487         PR middle-end/56022
18488         * function.c (allocate_struct_function): Call
18489         invoke_set_current_function_hook earlier.
18491 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18493         * reload1.c (init_reload): Only initialize reload_obstack
18494         during the first call.
18496 2013-01-21  Marek Polacek  <polacek@redhat.com>
18498         * cfgloop.c (verify_loop_structure): Fix up grammar.
18500 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
18502         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18503         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18505 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18507         PR target/56058
18508         * config/arm/marvell-pj4.md: Update copyright year.
18509         Fix up use of alu to alu_reg and simple_alu_imm.
18511 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18513         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18515 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
18517         PR target/55433
18518         * lra-constraints.c (curr_insn_transform): Don't reuse original
18519         insn for secondary memory move when memory mode should be different.
18521 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18523         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18524         atomic_storedi_1): New patterns.
18526 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18528         btver2 pipeline descriptions.
18529         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18530         descriptions.
18531         * config/i386/i386.md (btver2_decode): New type attributes.
18532         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18533         type attributes.
18534         * config/i386/btver2.md: New file describing btver2 pipelines.
18536 2013-01-19  Andrew Pinski  <apinski@cavium.com>
18538         PR tree-optimization/52631
18539         * tree-ssa-sccvn (visit_use): Before looking up the original
18540         statement, try looking up the simplified expression.
18542 2013-01-19  Anthony Green  <green@moxielogic.com>
18544         * config/moxie/moxie.c (moxie_expand_prologue): Set
18545         current_function_static_stack_size.
18547 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18549         PR tree-optimization/56029
18550         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18551         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18553 2013-01-18  Sharad Singhai  <singhai@google.com>
18555         PR tree-optimization/55995
18556         * dumpfile.c (dump_loc): Print location only if available.
18557         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18559 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
18561         PR target/55433
18562         * lra-constraints.c (curr_insn_transform): Reuse original insn for
18563         secondary memory move.
18564         (inherit_reload_reg): Use rclass instead of cl for
18565         check_secondary_memory_needed_p.
18567 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18569         PR middle-end/56015
18570         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18571         the case where writing real complex part of target modifies op1.
18573 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
18575         * config/aarch64/aarch64-simd.md
18576         (aarch64_vcond_internal<mode>): Handle unordered cases.
18577         * config/aarch64/iterators.md (v_cmp_result): New.
18579 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
18580             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18582         * config/arm/marvell-pj4.md: New file.
18583         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18584         * config/arm/arm.md (generic_sched): Add marvell_pj4.
18585         (generic_vfp): Likewise.
18586         * config/arm/arm-cores.def: Add marvell-pj4.
18587         * config/arm/arm-tune.md: Regenerate.
18588         * config/arm/arm-tables.opt: Regenerate.
18589         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18590         * doc/invoke.texi: Document marvell-pj4.
18592 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
18594         * config/aarch64/arm_neon.h: Map scalar types to standard types.
18596 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18598         PR debug/54114
18599         PR debug/54402
18600         PR debug/49888
18601         * var-tracking.c (negative_power_of_two_p): New.
18602         (global_get_addr_cache, local_get_addr_cache): New.
18603         (get_addr_from_global_cache, get_addr_from_local_cache): New.
18604         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
18605         heading comment.
18606         (vt_stack_offset_p): Remove.
18607         (vt_canon_true_dep): Always canonicalize loc's address.
18608         (clobber_overlapping_mems): Make sure we have a MEM.
18609         (local_get_addr_clear_given_value): New.
18610         (val_reset): Clear local cached entries.
18611         (compute_bb_dataflow): Create and release the local cache.
18612         Disable duplicate MEMs clobbering.
18613         (emit_notes_in_bb): Clobber MEMs likewise.
18614         (vt_emit_notes): Create and release the local cache.
18615         (vt_initialize, vt_finalize): Create and release the global
18616         cache, respectively.
18617         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18619 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18621         PR libmudflap/53359
18622         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18623         not found in the symtab.
18625 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18627         PR debug/56006
18628         PR rtl-optimization/55547
18629         PR rtl-optimization/53827
18630         PR debug/53671
18631         PR debug/49888
18632         * alias.c (offset_overlap_p): New, factored out of...
18633         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
18634         the conservative special case for symbolic constants.  Don't
18635         adjust zero sizes on alignment.
18637 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
18639         PR rtl-optimization/52573
18640         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18641         REG_UNUSED for the same register.
18643 2013-01-17  Richard Biener  <rguenther@suse.de>
18644             Marek Polacek  <polacek@redhat.com>
18646         PR rtl-optimization/55833
18647         * loop-unswitch.c (unswitch_loops): Move loop verification...
18648         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
18649         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18650         Set it to true when we're removing a loop from hierarchy tree in
18651         an irreducible region.
18652         (fix_bb_placements): Adjust caller.
18653         (fix_loop_placements): Likewise.
18655 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
18657         * config/avr/builtins.def (DEF_BUILTIN): Factor out
18658         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18659         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18660         Remove ID.  Adjust comments.
18661         * config/avr/avr-c.c (avr_builtin_name): Remove.
18662         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18663         * config/avr/avr.c (avr_tolower): New static function.
18664         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
18665         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18666         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18667         default expansion.
18669 2013-01-17  Jan Hubicka  <jh@suse.cz>
18671         PR tree-optimization/55273
18672         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18674 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
18676         PR target/55981
18677         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18678         store through atomic_store<mode>_1.
18679         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18681 2013-01-17  Martin Jambor  <mjambor@suse.cz>
18683         PR tree-optimizations/55264
18684         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18685         for virtual methods.
18686         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18687         virtual methods before inlining is over.
18688         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18689         virtual functions.
18690         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18691         non-virtual.
18693 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
18695         PR rtl-optimization/56005
18696         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18697         pending reads for prefetch.
18699 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
18701         * config/aarch64/aarch64.md
18702         (*cstoresi_neg_uxtw): New pattern.
18703         (*cmovsi_insn_uxtw): New pattern.
18704         (*<optab>si3_uxtw): New pattern.
18705         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18706         (*<optab>si3_insn_uxtw): New pattern.
18707         (*bswapsi2_uxtw): New pattern.
18709 2013-01-16  Richard Biener  <rguenther@suse.de>
18711         * tree-inline.c (tree_function_versioning): Remove set but
18712         never used variable.
18714 2013-01-16  Richard Biener  <rguenther@suse.de>
18716         PR tree-optimization/55964
18717         * tree-flow.h (rename_variables_in_loop): Remove.
18718         (rename_variables_in_bb): Likewise.
18719         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18720         (copy_loop_before): Adjust and delete update-ssa status.
18721         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18722         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
18723         (rename_variables_in_loop): Remove.
18724         (slpeel_update_phis_for_duplicate_loop): Likewise.
18725         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18726         use available cfg machinery instead of duplicating it.
18727         Update PHI nodes and perform poor-mans SSA update here.
18728         (slpeel_tree_peel_loop_to_edge): Adjust.
18730 2013-01-16  Richard Biener  <rguenther@suse.de>
18732         PR tree-optimization/54767
18733         PR tree-optimization/53465
18734         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18735         (vrp_visit_phi_node): For PHI arguments coming via backedges
18736         drop all symbolical range information.
18737         (execute_vrp): Compute backedges.
18739 2013-01-16  Richard Biener  <rguenther@suse.de>
18741         * doc/install.texi: Update CLooG and ISL requirements to
18742         0.18.0 and 0.11.1.
18744 2013-01-16  Christian Bruel  <christian.bruel@st.com>
18746         PR target/55301
18747         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18748         (broken_move): Handle UNSPECV_SP_SWITCH_B.
18749         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18751 2013-01-16  DJ Delorie  <dj@redhat.com>
18753         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18754         (UNSPECV_SP_SWITCH_E): New.
18755         (sp_switch_1): Change to an unspec.
18756         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18757         replace $r15.
18759 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18761         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18762         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18763         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18764         (expand_mem_thread_fence): Ditto.
18765         (expand_mem_signal_fence): Ditto.
18766         (expand_atomic_load): Ditto.
18767         (expand_atomic_store): Ditto.
18769 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18771         PR rtl-optimization/55547
18772         PR rtl-optimization/53827
18773         PR debug/53671
18774         PR debug/49888
18775         * alias.c (memrefs_conflict_p): Set sizes to negative after
18776         AND adjustments.
18778 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18780         PR target/55940
18781         * function.c (thread_prologue_and_epilogue_insns): Always
18782         add crtl->drap_reg to set_up_by_prologue.set, even if
18783         stack_realign_drap is false.
18785 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18787         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18788         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18789         *call): Fix indention.
18791 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18793         PR target/55876
18794         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18795         Update comment.
18797 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18799         PR rtl-optimization/55153
18800         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18802 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18804         PR tree-optimization/55920
18805         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18806         accesses as grp_to_be_debug_replaced.
18808 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18810         PR tree-optimization/55920
18811         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18812         there is non-useless type conversion needed from debug rhs to lhs,
18813         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18815 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18816             Mikael Pettersson  <mikpe@it.uu.se>
18818         PR target/43961
18819         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18820         Thumb.
18821         (ASM_OUTPUT_CASE_LABEL): Remove.
18822         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18823         * final.c (shorten_branches): Update alignment of labels before
18824         jump tables if CASE_VECTOR_SHORTEN_MODE.
18826 2013-01-15  Richard Biener  <rguenther@suse.de>
18828         PR bootstrap/55961
18829         * system.h: Do not include gmp.h for building host tools.
18831 2013-01-15  Richard Biener  <rguenther@suse.de>
18833         PR middle-end/55882
18834         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18835         account for bitpos when computing alignment.
18837 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18839         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18840         (ix86_target_macros_internal): Likewise.
18842         * config/i386/i386.c (m_CORE2I7): Removed.
18843         (m_CORE_HASWELL): New macro.
18844         (m_CORE_ALL): Likewise.
18845         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18846         (initial_ix86_arch_features): Likewise.
18847         (processor_target_table): Initializations for Core avx2.
18848         (cpu_names): New names "core-avx2".
18849         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18850         PROCESSOR_CORE_HASWELL.
18851         (ix86_issue_rate): New case.
18852         (ia32_multipass_dfa_lookahead): Likewise.
18853         (ix86_sched_init_global): Likewise.
18855         * config/i386/i386.h (TARGET_HASWELL): New macro.
18856         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18857         (processor_type): New PROCESSOR_HASWELL.
18859 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18861         PR tree-optimization/55955
18862         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18863         *SHIFT_EXPR and *ROTATE_EXPR codes.
18865         PR tree-optimization/48766
18866         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18867         -ftrapv disable -fwrapv.
18869 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18871         PR target/55974
18872         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18873         etc. to 1 and not to __flash.
18874         Use LL suffix for __INT24_MAX__ with -mint8.
18875         Use ULL suffix for __UINT24_MAX__ with -mint8.
18877 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18879         * config/avr/avr-arch.h
18880         (struct base_arch_s): Use typedef avr_arch_t instead.
18881         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18882         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18883         * config/avr/avr.c: Same.
18884         * config/avr/avr-devices.c: Same.
18885         * config/avr/driver-avr.c: Same.
18886         * config/avr/gen-avr-mmcu-texi.c: Same.
18887         * config/avr/avr-mcus.def: Adjust comment.
18889 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18891         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18892         * config/aarch64/iterators.md (VALLDI): New.
18894 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18895             Andi Kleen  <ak@linux.intel.com>
18897         PR target/55948
18898         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18899         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18900         memmodel flag.
18902 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18904         * config/avr/avr-stdint.h: Remove trailing blanks.
18905         * config/avr/avr-log.h: Same.
18906         * config/avr/avr-arch.h: Same.
18907         * config/avr/avr-devices.c: Same.
18908         * config/avr/avr-dimode.md: Same.
18909         * config/avr/predicates.md: Same.
18910         * config/avr/avr-c.c: Same.  And fix typo.
18912         * config/avr/avr-protos.h: Same.  And:
18913         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18914         (init_cumulative_args): Rename to avr_init_cumulative_args.
18915         (expand_prologue): Rename to avr_expand_prologue.
18916         (expand_epilogue): Rename to avr_expand_epilogue.
18917         (adjust_insn_length): Rename to avr_adjust_insn_length.
18918         (notice_update_cc): Rename to avr_notice_update_cc.
18919         (final_prescan_insn): Rename to avr_final_prescan_insn.
18920         * config/avr/avr.c: Same.
18921         * config/avr/avr.h: Same.
18922         * config/avr/avr.md: Remove trailing blanks.
18923         (prologue): Use avr_expand_prologue.
18924         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18926 2013-01-14  Richard Biener  <rguenther@suse.de>
18928         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18929         verify_location, collect_subblocks): New functions.
18930         (verify_gimple_in_cfg): Verify that locations only reference
18931         BLOCKs in the functions BLOCK tree.
18933 2013-01-14  Richard Biener  <rguenther@suse.de>
18935         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18936         PHI argument.
18937         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18938         unshare reference.
18939         (insert_out_of_ssa_copy_on_edge): Likewise.
18940         (rewrite_close_phi_out_of_ssa): Likewise.
18941         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18942         debug expressions.
18943         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18944         propagated constants.
18945         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18946         can not be shared.
18948 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18950         * config/avr/avr-modes.def: Add GPL copyright notice.
18952 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18954         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18955         MEMMODEL_MASK to determine memory model.
18956         (atomic_store<mode>): Ditto from operands[2].
18957         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18959 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18961         PR fortran/55935
18962         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18963         (fold_gimple_assign): Don't call unshare_expr here.
18964         (fold_ctor_reference): Call unshare_expr.
18966 2013-01-13  Terry Guo  <terry.guo@arm.com>
18968         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18969         * doc/fragments.texi: Document MULTILIB_REUSE.
18970         * gcc.c (multilib_reuse): New internal spec.
18971         (set_multilib_dir): Also search multilib from multilib_reuse.
18972         * genmultilib (tmpmultilib3): Refactor code.
18973         (tmpmultilib4): Ditto.
18974         (multilib_reuse): New multilib argument.
18976 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18978         * Makefile.in: Update copyright.
18980 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18982         PR middle-end/55890
18983         * calls.c (expand_call): Check if arg_nr is valid.
18985 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18987         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18988         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18989         documentation.  Add missing '__' in front of
18990         __builtin_ia32_packssdw256.
18992 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18994         PR target/55719
18995         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18996         NO_REGS for larl operands.
18997         (s390_reload_larl_operand): Use s390_load_address instead of
18998         emit_move_insn.
19000 2013-01-11  Richard Biener  <rguenther@suse.de>
19002         * tree-cfg.c (verify_node_sharing_1): Split out from ...
19003         (verify_node_sharing): ... here.
19004         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
19006 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
19008         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
19009         Substitute TREECHECKING.
19010         * configure: Regenerate.
19011         * Makefile.in (TREECHECKING): New.
19013 2013-01-11  Richard Guenther  <rguenther@suse.de>
19015         PR tree-optimization/44061
19016         * tree-vrp.c (extract_range_basic): Compute zero as
19017         value-range for __builtin_constant_p of function parameters.
19019 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
19021         Update copyright years.
19023 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
19025         PR rtl-optimization/55672
19026         * lra-eliminations.c (mark_not_eliminable): Permit addition with
19027         const to be eliminable.
19029 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
19031         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19032         * configure: Regenerate.
19034 2013-01-10  Richard Biener  <rguenther@suse.de>
19036         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19038 2013-01-10  Richard Biener  <rguenther@suse.de>
19040         PR bootstrap/55792
19041         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19042         locations for virtual PHI arguments.
19043         (rewrite_update_phi_arguments): Likewise.
19045 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
19047         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19048         on to assembler.
19050 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
19052         PR tree-optimization/55921
19053         * tree-complex.c (expand_complex_asm): New function.
19054         (expand_complex_operations_1): Call it for GIMPLE_ASM.
19056 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19058         PR target/55718
19059         * config/s390/s390.c (s390_symref_operand_p)
19060         (s390_loadrelative_operand_p): Merge the two functions.
19061         (s390_check_qrst_address, print_operand_address): Add parameters
19062         to s390_loadrelative_operand_p invokation.
19063         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19064         (s390_reload_larl_operand, s390_secondary_reload): Use
19065         s390_loadrelative_operand_p instead of s390_symref_operand_p.
19066         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19068 2013-01-09  Mike Stump  <mikestump@comcast.net>
19070         * dse.c (record_store): Remove unnecessary assert.
19072 2013-01-09  Jan Hubicka  <jh@suse.cz>
19074         PR tree-optimization/55569
19075         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19076         * cfgloop.h (scale_loop_profile): Likewise.
19078 2013-01-09  Jan Hubicka  <jh@suse.cz>
19080         PR lto/45375
19081         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19082         functions.
19083         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19085 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
19087         PR middle-end/55114
19088         * expr.h (maybe_emit_group_store): Declare.
19089         * expr.c (maybe_emit_group_store): New function.
19090         * builtins.c (expand_builtin_int_roundingfn): Call it.
19091         (expand_builtin_int_roundingfn_2): Likewise.
19093 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
19095         PR rtl-optimization/55829
19096         * lra-constraints.c (match_reload): Add code for absent output.
19097         (curr_insn_transform): Add code for reloads of matched inputs
19098         without output.
19100 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
19102         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19103         attribute of movddup insn to DF.
19104         (*vec_interleave_lowv2df): Ditto.
19105         (vec_dupv2df): Ditto.
19107 2013-01-09  Jan Hubicka  <jh@suse.cz>
19109         PR tree-optimiation/55875
19110         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19111         EVERY_ITERATION parameter.
19112         (number_of_iterations_exit): Check if exit is executed every iteration.
19113         (idx_infer_loop_bounds): Similarly here.
19114         (n_of_executions_at_most): Simplify
19115         to only test for cases where statement is dominated by the
19116         particular bound; handle correctly the "postdominance" test.
19117         (scev_probably_wraps_p): Use max loop iterations info
19118         as a global bound first.
19120 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
19121             Nick Clifton  <nickc@redhat.com>
19123         * config/v850/v850.md (cbranchsf4): New pattern.
19124         (cstoresf4): New pattern.
19125         (cbranchdf4): New pattern.
19126         (cstoredf4): New pattern.
19127         (movsicc): Disallow floating point comparisons.
19128         (cmpsf_le_insn): Fix order of operators.
19129         (cmpsf_lt_insn): Likewise.
19130         (cmpsf_eq_insn): Likewise.
19131         (cmpdf_le_insn): Likewise.
19132         (cmpdf_lt_insn): Likewise.
19133         (cmpdf_eq_insn): Likewise.
19134         (cmpsf_ge_insn): Use LE comparison.
19135         (cmpdf_ge_insn): Likewise.
19136         (cmpsf_gt_insn): Use LT comparison.
19137         (cmpdf_gt_insn): Likewise.
19138         (cmpsf_ne_insn): Delete pattern.
19139         (cmpdf_ne_insn): Delete pattern.
19140         * config/v850/v850.c (v850_gen_float_compare): Use
19141         gen_cmpdf_eq_insn for NE comparison.
19142         (v850_float_z_comparison_operator)
19143         (v850_float_nz_comparison_operator): Move from here ...
19144         * config/v850/predicates.md: ... to here.  Move GT and GE
19145         comparisons into v850_float_z_comparison_operator.
19146         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19147         Delete prototype.
19148         (v850_float_nz_comparison_operator): Likewise.
19150 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19152         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19153         with calls to gen_insvsi/gen_insvdi.
19155 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19157         * config/i386/i386.c (initial_ix86_tune_features): Set up
19158         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19160 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
19161             Jakub Jelinek  <jakub@redhat.com>
19163         PR tree-optimization/48189
19164         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19165         If nitercst is 0, don't predict the exit edge.
19167 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19169         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19170         in asm_fprintf with reg_names.
19171         (aarch64_print_operand_address): Likewise.
19172         (aarch64_return_addr): Likewise.
19173         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19175 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19177         * config/pa/pa.h (VAL_U6_BITS_P): Define.
19178         (INT_U6_BITS): Likewise.
19179         * config/pa/predicates.md (uint6_operand): New predicate.
19180         (shift5_operand, shift6_operand): Likewise.
19181         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19182         arith32_operand.
19183         (lshrdi3): Use shift6_operand.
19184         (shrpsi4, shrpdi4): New insn patterns.
19185         (extzv): Delete expander.
19186         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
19187         predicates in unamed zero extract patterns.  Tighten common constraint.
19188         (extv): Delete expander.
19189         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
19190         predicates in unamed sign extract patterns.  Tighten common constraint.
19191         (insv): Delete expander.
19192         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
19193         predicates in unamed insert patterns.  Tighten common constraint.
19194         Change uint32_operand predicate to uint6_operand predicate in unamed
19195         DImode pattern to insert constant values of type 1...1xxxx.
19197 2013-01-04  Jan Hubicka  <jh@suse.cz>
19199         PR tree-optimization/55823
19200         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19201         issue.
19203 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19204             Uros Bizjak  <ubizjak@gmail.com>
19206         PR rtl-optimization/55845
19207         * df-problems.c (can_move_insns_across): Stop scanning at
19208         volatile_insn_p source instruction or give up if
19209         across_from .. across_to range contains any volatile_insn_p
19210         instructions.
19212 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19214         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19215         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19216         Declare.
19217         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19218         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19220 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19222         PR fortran/55341
19223         * asan.c (asan_clear_shadow): New function.
19224         (asan_emit_stack_protection): Use it.
19226 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19228         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19229         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19230         with tab instead of space.
19232 2013-01-08  Nick Clifton  <nickc@redhat.com>
19234         * config/rl78/rl78.c (rl78_expand_prologue): Always select
19235         register bank 0 at the start of an interrupt handler.
19236         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19237         MDBH registers.
19239 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19241         * config/aarch64/aarch64-simd.md
19242         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19243         (aarch64_simd_bsl): Likewise.
19244         (aarch64_vcond_internal<mode>): Likewise.
19245         (vcond<mode><mode>): Likewise.
19246         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19247         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19249 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19251         * config/aarch64/aarch64-builtins.c
19252         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19254 2013-01-08  Martin Jambor  <mjambor@suse.cz>
19256         PR debug/55579
19257         * tree-sra.c (analyze_access_subtree): Return true also after
19258         potentially creating a debug-only replacement.
19260 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19262         PR middle-end/55890
19263         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19265         PR tree-optimization/54120
19266         * tree-vrp.c (range_fits_type_p): Don't allow
19267         src_precision < precision from signed vr to unsigned_p
19268         if vr->min or vr->max is negative.
19269         (simplify_float_conversion_using_ranges): Test can_float_p
19270         against CODE_FOR_nothing.
19272 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19273             Richard Biener  <rguenther@suse.de>
19275         PR middle-end/55851
19276         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19277         types instead of just INTEGER_TYPE types.
19279 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
19281         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19282         TF_SIZE): Define.
19284 2013-01-07  Steve Ellcey  <sellcey@mips.com>
19286         PR target/42661
19287         * config/mips/mips.opt: Change mad to mmad to match documentation.
19289 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19291         PR target/55897
19292         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19293         .progmemx.data now.
19295 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19297         PR target/55897
19298         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19299         (avr_addrspace_t): Add .section_name field.
19300         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19301         array size.
19302         (avr_addrspace): Same.  Initialize .section_name.  Remove last
19303         NULL entry.  Put __memx into .progmemx.data.
19304         (progmem_section_prefix): Remove.
19305         (avr_asm_init_sections): No need to initialize progmem_section.
19306         (avr_asm_named_section): Use avr_addrspace[].section_name to get
19307         section name prefix.
19308         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
19309         retrieve the progmem section.
19310         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19311         boundary to run over avr_addrspace[].
19312         (avr_register_target_pragmas): Ditto.
19314 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
19316         * varasm.c (output_constant_def_contents): For asan_protect_global
19317         protected strings, adjust DECL_ALIGN if needed, before testing for
19318         anchored symbols.
19319         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19320         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
19321         normal decls.
19322         (output_object_block): For asan protected decls, emit asan padding
19323         after their contents.
19324         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19325         (asan_finish_file): Test it here instead.
19327 2013-01-07  Nick Clifton  <nickc@redhat.com>
19328             Matthias Klose  <doko@debian.org>
19329             Doug Kwan  <dougkwan@google.com>
19330             H.J. Lu  <hongjiu.lu@intel.com>
19332         PR driver/55470
19333         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19335         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19337         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19339         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19341         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19343 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19345         PR target/54461
19346         * doc/install.texi (Cross-Compiler-Specific Options): Document
19347         --with-avrlibc.
19349 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
19351         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19352         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19353         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19354         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19355         vqmovun_high_s64): Fix source operand number and update copyright.
19357 2013-01-07  Richard Biener  <rguenther@suse.de>
19359         PR middle-end/55890
19360         * gimple.h (gimple_call_builtin_p): New overload.
19361         * gimple.c (validate_call): New function.
19362         (gimple_call_builtin_p): Likewise.
19363         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19364         Use gimple_call_builtin_p.
19365         (find_func_clobbers): Likewise.
19366         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19367         (strlen_optimize_stmt): Likewise.
19369 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19371         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19372         (vld1q_dup_*): Likewise.
19373         (vld1_*): Likewise.
19374         (vld1q_*): Likewise.
19375         (vld1_lane_*): Likewise.
19376         (vld1q_lane_*): Likewise.
19378 2013-01-07  Richard Biener  <rguenther@suse.de>
19380         * lto-streamer.h (LTO_minor_version): Bump to 2.
19382 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19384         * config/aarch64/aarch64-protos.h
19385         (aarch64_const_double_zero_rtx_p): Rename to...
19386         (aarch64_float_const_zero_rtx_p): ...this.
19387         (aarch64_float_const_representable_p): New.
19388         (aarch64_output_simd_mov_immediate): Likewise.
19389         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19390         move immediate case.
19391         * config/aarch64/aarch64.c
19392         (aarch64_const_double_zero_rtx_p): Rename to...
19393         (aarch64_float_const_zero_rtx_p): ...this.
19394         (aarch64_print_operand): Allow printing of new constants.
19395         (aarch64_valid_floating_const): New.
19396         (aarch64_legitimate_constant_p): Check for valid floating-point
19397         constants.
19398         (aarch64_simd_valid_immediate): Likewise.
19399         (aarch64_vect_float_const_representable_p): New.
19400         (aarch64_float_const_representable_p): Likewise.
19401         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19402         (aarch64_output_simd_mov_immediate): New.
19403         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19404         (*movdf_aarch64): Likewise.
19405         * config/aarch64/constraints.md (Ufc): New.
19406         (Y): call aarch64_float_const_zero_rtx.
19407         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19409 2013-01-07  Richard Biener  <rguenther@suse.de>
19411         PR tree-optimization/55888
19412         PR tree-optimization/55862
19413         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19414         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19415         not if it is contained therein.
19417 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19419         * config/avr/t-avr: Typo.
19421 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19423         PR55243
19424         * config/avr/t-avr: Don't automatically rebuild
19425         $(srcdir)/config/avr/t-multilib
19426         $(srcdir)/config/avr/avr-tables.opt
19427         $(srcdir)/doc/avr-mmcu.texi
19428         (avr-mcus): New phony target to build them on request.
19429         (s-avr-mlib, s-avr-mmcu-texi): Remove.
19430         * avr/avr-mcus.def: Adjust comments.
19432 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
19434         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19436 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19438         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19440 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19442         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19444 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
19446         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19447         to generate profiling.
19448         * config/rs6000/aix64.h (LIB_SPEC): Same.
19450 2013-01-04  Andrew Pinski  <apinski@cavium.com>
19452         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19453         New function.
19454         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19456 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
19458         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19459         unconditionally.
19460         (ix86_expand_move): Ditto.
19461         (ix86_zero_extend_to_Pmode): Ditto.
19462         (ix86_expand_call): Ditto.
19463         (ix86_expand_special_args_builtin): Ditto.
19464         (ix86_expand_builtin): Ditto.
19466 2013-01-04  Richard Biener  <rguenther@suse.de>
19468         PR tree-optimization/55862
19469         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19470         translating them through PHI nodes.
19472 2013-01-04  Martin Jambor  <mjambor@suse.cz>
19474         PR tree-optimization/55755
19475         * tree-sra.c (sra_modify_assign): Do not check that an access has no
19476         children when trying to avoid producing a VIEW_CONVERT_EXPR.
19478 2013-01-04  Marek Polacek  <polacek@redhat.com>
19480         PR middle-end/55859
19481         * opts.c (default_options_optimization): Clarify error message.
19483 2013-01-04  Richard Biener  <rguenther@suse.de>
19485         PR middle-end/55863
19486         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19487         reassociation.
19489 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19491         PR target/53789
19492         * config/pa/pa.md (movsi): Revert previous change.
19493         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19494         references.
19496 2013-01-03  Richard Henderson  <rth@redhat.com>
19498         * config/i386/i386.c (ix86_expand_move): Always assign to op1
19499         after eliminating TLS symbols.
19501 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
19503         PR bootstrap/50167
19504         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19505         * graphite-poly.c (debug_gmp_value): Likewise.
19507 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
19509         PR target/55712
19510         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19511         selected code model, define __code_mode_small__, __code_model_medium__,
19512         __code_model_large__, __code_model_32__ or __code_model_kernel__.
19513         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19514         xchg temporary register with %k.  Declare temporary register as
19515         early clobbered.
19516         [__x86_64__]: For medium and large code models, preserve %rbx register.
19518 2013-01-03  Richard Biener  <rguenther@suse.de>
19520         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19521         (dump_subscript): Adjust.
19522         (finalize_ddr_dependent): Do not dump redundant info.
19523         (analyze_siv_subscript): Adjust.
19524         (subscript_dependence_tester): Likewise.
19525         (compute_affine_dependence): Likewise.
19527 2013-01-03  Richard Biener  <rguenther@suse.de>
19529         Revert
19530         2013-01-03  Richard Biener  <rguenther@suse.de>
19532         PR tree-optimization/55857
19533         * tree-vect-stmts.c (vectorizable_load): Do not setup
19534         re-alignment for invariant loads.
19536         2013-01-02  Richard Biener  <rguenther@suse.de>
19538         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19539         invariant load do not generate a vector load from the scalar location.
19541 2013-01-03  Richard Biener  <rguenther@suse.de>
19543         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19544         for not vectorizing.
19545         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19546         not build INDIRECT_REFs, call get_name once only.
19547         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
19548         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19550 2013-01-03  Richard Biener  <rguenther@suse.de>
19552         PR tree-optimization/55857
19553         * tree-vect-stmts.c (vectorizable_load): Do not setup
19554         re-alignment for invariant loads.
19556 2013-01-03  Richard Biener  <rguenther@suse.de>
19558         PR lto/55848
19559         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19560         prefer a built-in decl.
19562 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19564         * gcc.c (process_command): Update copyright notice dates.
19565         * gcov.c (print_version): Likewise.
19566         * gcov-dump.c (print_version): Likewise.
19568         PR rtl-optimization/55838
19569         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19570         iv0.step, iv1.step and step.
19572 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19573             Marc Glisse  <marc.glisse@inria.fr>
19575         PR tree-optimization/55832
19576         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19577         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19578         integer_{one,zero}_node.
19580 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19582         PR debug/54402
19583         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19584         * var-tracking.c (reverse_op): Don't add reverse ops to
19585         VALUEs that have already
19586         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19588 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
19590         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19592 2013-01-02  Teresa Johnson  <tejohnson@google.com>
19594         * dumpfile.c (dump_loc): Print filename with location.
19595         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19596         new location_t parameter to emit complete unroll message with
19597         new dump framework.
19598         (canonicalize_loop_induction_variables): Compute loops location
19599         and pass to try_unroll_loop_completely.
19600         * loop-unroll.c (report_unroll_peel): New function.
19601         (peel_loops_completely): Use new dump format with location
19602         for main dumpfile message, and invoke report_unroll_peel on success.
19603         (decide_unrolling_and_peeling): Ditto.
19604         (decide_peel_once_rolling): Remove old dumpfile message subsumed
19605         by report_unroll_peel.
19606         (decide_peel_completely): Ditto.
19607         (decide_unroll_constant_iterations): Ditto.
19608         (decide_unroll_runtime_iterations): Ditto.
19609         (decide_peel_simple): Ditto.
19610         (decide_unroll_stupid): Ditto.
19611         * cfgloop.c (get_loop_location): New function.
19612         * cfgloop.h (get_loop_location): Declare.
19614 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19616         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19617         NULL.
19619 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19621         PR middle-end/55198
19622         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19623         BLKmode objects when EXPAND_MEMORY is specified.
19625 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19627         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19628         in loop predicate.
19629         (fold_builtin_cpu): Do not share cpu model decls across statements.
19631 2013-01-02  Jason Merrill  <jason@redhat.com>
19633         PR c++/55804
19634         * tree.c (build_array_type_1): Revert earlier change.
19636 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
19638         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19639         "cortex-a57".
19640         * config/aarch64/aarch64-tune.md: Re-generate.
19642 2013-01-02  Richard Biener  <rguenther@suse.de>
19644         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19645         invariant load do not generate a vector load from the scalar location.
19647 2013-01-02  Richard Biener  <rguenther@suse.de>
19649         PR bootstrap/55784
19650         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19651         * configure: Regenerate.
19653 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19655         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19656         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19657         (expand_builtin_int_roundingfn_2): Keep the original target around
19658         for the fallback case.
19660 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19662         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19663         to be clear for sign changes.
19665 2013-01-01  Jan Hubicka  <jh@suse.cz>
19667         * ipa-inline-analysis.c: Fix formatting.
19669 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
19671         PR tree-optimization/55831
19672         * tree-vect-loop.c (get_initial_def_for_induction): Use
19673         gsi_after_labels instead of gsi_start_bb.
19675 Copyright (C) 2013 Free Software Foundation, Inc.
19677 Copying and distribution of this file, with or without modification,
19678 are permitted in any medium without royalty provided the copyright
19679 notice and this notice are preserved.