c-objc-common.c (c_tree_printer): Tidy.
[official-gcc.git] / gcc / ChangeLog
blob25e3fa5a5ae13164cd48ef5d8b3e6726c70d3c64
1 2013-09-03  Richard Biener  <rguenther@suse.de>
3         PR lto/58285
4         * tree-streamer-out.c: Include tm.h.
5         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7 2013-09-03  Jan Hubicka  <jh@suse.cz>
9         * tree-profile.c (tree_profiling): Cleanup CFG when done.
11 2013-09-03  Alan Modra  <amodra@gmail.com>
13         * config.gcc (powerpc*-*-linux*): Add support for little-endian
14         multilibs to big-endian target and vice versa.
15         * config/rs6000/t-linux64: Use := assignment on all vars.
16         (MULTILIB_EXTRA_OPTS): Remove fPIC.
17         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
18         * config/rs6000/t-linux64le: New file.
19         * config/rs6000/t-linux64bele: New file.
20         * config/rs6000/t-linux64lebe: New file.
22 2013-09-02  Jan Hubicka  <jh@suse.cz>
24         * ipa-inline-transform.c (inline_transform): Do not
25         optimize_inline_calls when not optimizing.
27 2013-09-02  Jan Hubicka  <jh@suse.cz>
29         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
30         duplicated nodes for assembler names.
31         * symtab.c (symtab_unregister_node): Do not attempt to unlink
32         hard registers from assembler name hash.
34 2013-09-02  Jan Hubicka  <jh@suse.cz>
36         * ipa-split.c (execute_split_functions): Split externally visible
37         functions called once.
39 2013-09-02  Martin Jambor  <mjambor@suse.cz>
41         PR ipa/58106
42         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
43         linked list.  When finding the correct duplicate, also consider also
44         the caller in additon to its inlined_to node.
46 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
48         * config/aarch64/aarch64-simd-builtins.def
49         (dup_lane_scalar): Remove.
50         * config/aarch64/aarch64-simd.md
51         (aarch64_simd_dup): Add 'w->w' alternative.
52         (aarch64_dup_lane<mode>): Allow for VALL.
53         (aarch64_dup_lane_scalar<mode>): Remove.
54         (aarch64_dup_lane_<vswap_width_name><mode>): New.
55         (aarch64_get_lane_signed<mode>): Add w->w altenative.
56         (aarch64_get_lane_unsigned<mode>): Likewise.
57         (aarch64_get_lane<mode>): Likewise.
58         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
59         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
60         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
61         (VCON): Change container of V2SF.
62         (vswap_width_name): Likewise.
63         * config/aarch64/arm_neon.h
64         (__aarch64_vdup_lane_any): New.
65         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
66         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
67         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
69 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
71         PR middle-end/56382
72         * expr.c (emit_move_complex): Do not move complex FP values as parts if
73         the source or the destination is a single hard register.
75 2013-09-02  Richard Biener  <rguenther@suse.de>
77         PR middle-end/57511
78         * tree-scalar-evolution.c (instantiate_scev_name): Allow
79         non-linear SCEVs.
81 2013-09-02  Richard Biener  <rguenther@suse.de>
83         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
84         arithmetic to sizetype.
86 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
88         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
89         Find auto-increment use both before and after candidate.
91 2013-09-02  Marek Polacek  <polacek@redhat.com>
93         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
95 2013-09-01  Jan Hubicka  <jh@suse.cz>
97         * Makefile.in: Add ipa-profile.o
98         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
99         * cgraph.c (struct cgraph_propagate_frequency_data,
100         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
101         ipa-profile.c; replace cgraph_ by ipa_ prefix.
102         * cgraph.h (cgraph_propagate_frequency): Remove.
103         * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
104         (inline_update_callee_summaries): Update.
105         * ipa-profile.c: New file.
106         * ipa-utils.h (ipa_propagate_frequency): Declare.
107         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
108         data-streamer.h, value-prof.h
109         (symtab_remove_unreachable_nodes): Update profile.
110         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
111         account_time_size, cmp_counts, dump_histogram,
112         ipa_profile_generate_summary, ipa_profile_write_summary,
113         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
114         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
115         Move to ipa-profile.c
117 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
119         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
121 2013-09-01  Jan Hubicka  <jh@suse.cz>
123         * common.opt (fdevirtualize-speculatively): New function.
124         * invoke.texi (fdevirtualize-speculatively): Document.
125         * ipa-devirt.c: Include ipa-inline.h
126         (likely_target_p): New function.
127         (ipa_devirt): New function.
128         (gate_ipa_devirt): New function.
129         (pass_data_ipa_devirt): New static var.
130         (pass_ipa_devirt): Likewise.
131         (make_pass_ipa_devirt): New function.
132         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
133         (common_handle_option): Disable devirtualization when
134         value range profiling is available.
135         * passes.def (pass_ipa_devirt): Add.
136         * timever.def (TV_IPA_DEVIRT): New timevar.
137         * tree-pass.h (make_pass_ipa_devirt): 
139 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
141         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
142         include sanitize(undefined).
144 2013-08-31  Diego Novillo  <dnovillo@google.com>
146         * Makefile.in (TREE_CORE_H): Define.
147         (TREE_H): Use.
148         (GTFILES): Add tree-core.h.
149         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
150         size the array.
151         * tree-core.h: New file.
152         Move all data structures, enum, typedefs, global
153         declarations and constants from ...
154         * tree.h: ... here.
156 2013-08-31  Jan Hubicka  <jh@suse.cz>
158         * bulitins.c (expand_builtin): Do not early exit for gcov
159         instrumented functions.
161 2013-08-31  Marek Polacek  <polacek@redhat.com>
163         * ubsan.c: Include tm_p.h.
165 2013-08-31  Jan Hubicka  <jh@suse.cz>
167         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
168         warning.
170         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
171         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
172         * tree-cfg.c (verify_gimple_label): ... here.
173         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
174         (ipa_merge_profiles): New function.
175         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
176         (lto_input_function_body): Likewise.
177         * ipa-utils.h (ipa_merge_profiles): Declare.
178         * lto-streamer.h (lto_input_function_body): Update prototype.
179         (emit_label_in_global_context_p): Remove.
180         * lto-symtab.c: Include ipa-utils.h
181         (lto_cgraph_replace_node): Use ipa_merge_profiles.
183 2013-08-31  Jan Hubicka  <jh@suse.cz>
185         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
187 2013-08-31  Jan Hubicka  <jh@suse.cz>
189         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
191 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
193         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
194         "cmp" RTX before signed_comparison_operator check to account
195         for "code" changes.
197 2013-08-30  Jan Hubicka  <jh@suse.cz>
199         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
200         (detect_type_change_1): Rename to ...
201         (detect_type_change): ... this one; early return on non-polymorphic
202         types.
203         (detect_type_change_ssa): Add comp_type parameter; update
204         use of detect_type_change.
205         (compute_complex_assign_jump_func): Add param_type parameter;
206         update use of detect_type_change_ssa.
207         (compute_complex_ancestor_jump_func): Likewise.
208         (ipa_get_callee_param_type): New function.
209         (ipa_compute_jump_functions_for_edge): Compute parameter type;
210         update calls to the jump function computation functions.
212 2013-08-30  Teresa Johnson  <tejohnson@google.com>
213             Steven Bosscher  <steven@gcc.gnu.org>
215         * cfgrtl.c (fixup_new_cold_bb): New routine.
216         (commit_edge_insertions): Invoke fixup_partitions.
217         (find_partition_fixes): New routine.
218         (fixup_partitions): Ditto.
219         (verify_hot_cold_block_grouping): Update comments.
220         (rtl_verify_edges): Invoke find_partition_fixes.
221         (rtl_verify_bb_pointers): Update comments.
222         (rtl_verify_bb_layout): Ditto.
223         * basic-block.h (probably_never_executed_edge_p): Declare.
224         (fixup_partitions): Ditto.
225         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
226         * bb-reorder.c (sanitize_hot_paths): New function.
227         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
228         sanitize_hot_paths.
229         * predict.c (probably_never_executed_edge_p): New routine.
230         * cfg.c (check_bb_profile): Add partition insanity warnings.
232 2013-08-30  Meador Inge  <meadori@codesourcery.com>
234         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
236 2013-08-30  Marek Polacek  <polacek@redhat.com>
238         * Makefile.in (ubsan.o): Add.
239         (c-family/c-ubsan.o): Add.
240         (builtins.o): Add ubsan.h dependency.
241         * ubsan.h: New file.
242         * ubsan.c: New file.
243         * common.opt: Add -fsanitize=undefined option.
244         (flag_sanitize): Add variable.
245         (fsanitize=): Add option.  Add Driver.
246         (fsanitize=thread): Remove option.
247         (fsanitize=address): Likewise.
248         (static-libubsan): New option.
249         * doc/invoke.texi: Document the new flag and -static-libubsan.
250         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
251         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
252         * builtin-attrs.def (ATTR_COLD): Define.
253         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
254         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
255         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
256         * flag-types.h (sanitize_code): New enum.
257         * opts.c (common_handle_option): Parse command line arguments
258         of -fsanitize=.  Add -fsanitize=unreachable option.
259         * varasm.c (get_variable_section): Adjust.
260         (assemble_noswitch_variable): Likewise.
261         (assemble_variable): Likewise.
262         (output_constant_def_contents): Likewise.
263         (categorize_decl_for_section): Likewise.
264         (place_block_symbol): Likewise.
265         (output_object_block): Likewise.
266         * builtins.def: Likewise.
267         * toplev.c (compile_file): Likewise.
268         (process_options): Likewise.
269         * cppbuiltin.c: Likewise.
270         * tsan.c (tsan_pass): Likewise.
271         (tsan_gate): Likewise.
272         (tsan_gate_O0): Likewise.
273         * cfgexpand.c (partition_stack_vars): Likewise.
274         (expand_stack_vars): Likewise.
275         (defer_stack_allocation): Likewise.
276         (expand_used_vars): Likewise.
277         * cfgcleanup.c (old_insns_match_p): Likewise.
278         * asan.c (asan_finish_file): Likewise.
279         (asan_instrument): Likewise.
280         (gate_asan): Likewise.
281         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
282         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
283         (asan_global_struct): Use pointer_sized_int_node instead
284         calling build_nonstandard_integer_type.
285         (initialize_sanitizer_builtins): Likewise.
286         (asan_finish_file): Likewise.
287         * gcc.c: Document %{%:function(args):X}.
288         (static_spec_functions): Add sanitize.
289         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
290         store funcval != NULL there.
291         (do_spec_1): Adjust handle_spec_function caller.
292         (handle_braces): Allow %:function(args) as condition.
293         (sanitize_spec_function): New function.
294         (ADD_STATIC_LIBUBSAN_LIBS): Define.
295         (LIBUBSAN_SPEC): Likewise.
296         (LIBUBSAN_EARLY_SPEC): Likewise.
297         (SANITIZER_SPEC): Handle libubsan.
298         (SANITIZER_EARLY_SPEC): Likewise.
299         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
300         instead of fsanitize=address.
301         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
302         instead of fsanitize=address*.
303         * builtins.c: Include ubsan.h.
304         (fold_builtin_0): Instrument __builtin_unreachable.
305         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
306         instead of flag_asan.
307         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
308         (pointer_sized_int_node): Define.
309         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
311 2013-08-30  Mike Stump  <mikestump@comcast.net>
313         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
314         with RE patterns.
316 2013-08-29  Jan Hubicka  <jh@suse.cz>
318         * cgraph.c (cgraph_function_body_availability): Handle weakref
319         correctly.
320         * passes.def: Remove pass_fixup_cfg.
321         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
322         track when we need to remove functions.
323         (gate_ipa_inline): Execute inlining always; add comment why.
324         (pass_data_ipa_inline): Remove TODO_remove_functions.
325         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
326         do not produce summaries.
327         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
328         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
329         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
330         constant pool and vtable.
332 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
334         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
335         New arm_neon.h's internal macros to specify 64-bit constants.
336         Avoid using stdint.h's macros.
338 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
340         * recog.c (verify_changes): Verify that changes[i].old is non-zero
341         before applying REG_P.
343 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
345         PR tree-optimization/58277
346         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
347         after seeing too many stmts with vdef in between dombb and current
348         bb, invalidate everything.
350 2013-08-30  Richard Biener  <rguenther@suse.de>
352         * fold-const.c (fold_single_bit_test): Fix overflow test.
354 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
356         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
357         and which can live in a register, always retrieve the value on entry.
358         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
359         passed by invisible reference specially.
360         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
361         (vt_add_function_parameter): Correctly deal with a parameter passed by
362         invisible reference.
364 2013-08-30  Jan Hubicka  <jh@suse.cz>
366         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
368 2013-08-30  Richard Biener  <rguenther@suse.de>
370         PR tree-optimization/58228
371         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
372         allow invariant loads in nested loop vectorization.
374 2013-08-30  Richard Biener  <rguenther@suse.de>
376         PR tree-optimization/58223
377         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
378         (has_anti_or_output_dependence): ... this and adjust to also
379         look for output dependences.
380         (mark_nodes_having_upstream_mem_writes): Adjust.
381         (rdg_flag_uses): Likewise.
383 2013-08-30  Richard Biener  <rguenther@suse.de>
385         PR tree-optimization/58010
386         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
387         assert that we have a loop-closed PHI.
389 2013-08-29  Jan Hubicka  <jh@suse.cz>
391         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
392         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
393         * lto-section-in.c (lto_free_function_in_decl_state): New function.
394         (lto_free_function_in_decl_state_for_node): New function.
396 2013-08-29  Xinliang David Li  <davidxl@google.com>
398         * loop-unroll.c (report_unroll_peel): Minor message change.
399         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
400         Emit alignment peeling message with default -fopt-info.
401         (vect_loop_versioning): Emit loop version info message.
402         * tree-vectorizer.c (vectorize_loops): Minor message change.
403         (execute_vect_slp): Ditto.
405 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
407         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
408         of the clone from the DECL_NAME of the original function.
410 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
412         * passes.c (register_pass): Add overload.
413         * tree-pass.h (register_pass): Forward declare it.  Add comment.
415 2013-08-29  Jan Hubicka  <jh@suse.cz>
417         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
418         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
419         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
420         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
421         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
422         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
423         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
424         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
426 2013-08-29  Teresa Johnson  <tejohnson@google.com>
428         * dumpfile.c (dump_loc): Output column number.
429         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
430         * doc/invoke.texi: Document optall -fopt-info flag.
431         * profile.c (read_profile_edge_counts): Use new dump framework.
432         (compute_branch_probabilities): Ditto.
433         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
434         when pass not in any opt group.
435         * pass_manager.h (pass_manager::get_pass_profile): New method.
436         * value-prof.c (check_counter): Use new dump framework.
437         (check_ic_target): Ditto.
438         * coverage.c (get_coverage_counts): Ditto.
439         (coverage_init): Setup new dump framework.
441 2013-08-29  Richard Biener  <rguenther@suse.de>
443         PR tree-optimization/58246
444         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
445         handle the dominance check inside a basic-block.
447 2013-08-29  Richard Biener  <rguenther@suse.de>
449         PR middle-end/57287
450         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
451         of default defs that appear in abnormal PHI nodes.
453 2013-08-29  Richard Biener  <rguenther@suse.de>
455         PR tree-optimization/57685
456         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
457         single-use operands to avoid exponential complexity.
459 2013-08-28  Dehao Chen  <dehao@google.com>
461         * ipa-inline.c (edge_badness): Fix integer underflow.
463 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
465         * gtm-builtins.def (_ITM_free): Declare leaf.
467 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
469         PR target/58067
470         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
471         (*tls_local_dynamic_base_64_largepic): Likewise.
472         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
473         Remove predicate from call operand.
474         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
475         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
477 2013-08-28  Jeff Law  <law@redhat.com>
479         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
480         checks for the number of predecessors and successors allowed.
481         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
482         which require copying a joiner block if there is a request which
483         is a subpath that requires no joiner block copying.
485 2013-08-28  Jan Hubicka  <jh@suse.cz>
487         * lto-streamer-out.c (DFS_write_tree_body): Drop
488         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
489         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
490         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
491         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
492         Do not read DECL_ERROR_ISSUED.
493         (unpack_ts_decl_with_vis_value_fields): Do not read
494         DECL_DEFER_OUTPUT.
495         (lto_input_ts_binfo_tree_pointers): Do not read
496         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
497         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
498         write DECL_ERROR_ISSUED..
499         (pack_ts_decl_with_vis_value_fields): Do not write
500         DECL_DEFER_OUTPUT.
501         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
502         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
503         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
504         * tree.h (tree_decl_common): Update comment.
505         (DECL_ERROR_ISSUED): Remove.
507 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
509         PR middle-end/58257
510         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
512 2013-08-28  Jan Hubicka  <jh@suse.cz>
514         * builtins.def (free): Declare leaf.
516 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
518         * gdbhooks.py: New.
519         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
520         * configure: Regenerate.
522 2013-08-27  Martin Jambor  <mjambor@suse.cz>
524         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
525         (ipa_ancestor_jf_data): Likewise.
526         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
527         (ipa_get_jf_pass_through_type_preserved): New function.
528         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
529         (ipa_get_jf_ancestor_type_preserved): New function.
530         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
531         (ipa_get_jf_ancestor_result): Likewise.
532         (propagate_vals_accross_pass_through): Use
533         ipa_get_jf_pass_through_result to do all the value mappings.
534         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
535         type_preserved flag.
536         (ipa_set_jf_cst_copy): New function.
537         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
538         (ipa_set_jf_arith_pass_through): Likewise.
539         (ipa_set_ancestor_jf): Likewise.
540         (compute_complex_assign_jump_func): Set type_preserved instead of
541         punting.
542         (ipa_compute_jump_functions_for_edge): Likewise.
543         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
544         (update_jump_functions_after_inlining): Update type_preserved.
545         Explicitely create jump functions when combining one with pass_through.
546         (ipa_write_jump_function): Stream the type_preserved flags.
547         (ipa_read_jump_function): Likewise.
549 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
550             Aldy Hernandez  <aldyh@redhat.com>
552         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
553         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
554         * function.h (struct function): Add has_force_vect_loops and
555         has_simduid_loops.
556         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
557         * gimple.c (gimple_build_omp_critical): Add KIND argument and
558         handle it.
559         * gimple.def: Update CLAUSES comments.
560         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
561         (gimple_build_omp_for): Add argument to prototype.
562         (gimple_omp_for_kind): New.
563         (gimple_omp_for_set_kind): New.
564         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
565         GOVD_DATA_SHARE_CLASS.
566         (enum omp_region_type): Add ORT_SIMD.
567         (gimple_add_tmp_var): Handle ORT_SIMD.
568         (gimplify_var_or_parm_decl): Same.
569         (is_gimple_stmt): Same.
570         (omp_firstprivatize_variable): Same.
571         (omp_add_variable): Only use splay_tree_insert if lookup failed.
572         (omp_notice_variable): Handle ORT_SIMD.
573         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
574         (omp_check_private): Handle ORT_SIMD.
575         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
576         OMP_CLAUSE_SAFELEN.
577         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
578         Handle OMP_CLAUSE_LASTPRIVATE.
579         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
580         OMP_CLAUSE_SAFELEN.
581         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
582         (gimplify_expr): Handle OMP_SIMD.
583         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
584         (expand_GOMP_SIMD_VF): New.
585         (expand_GOMP_SIMD_LAST_LANE): New.
586         * internal-fn.def (GOMP_SIMD_LANE): New.
587         (GOMP_SIMD_VF): New.
588         (GOMP_SIMD_LAST_LANE): New.
589         * omp-low.c: Include target.h.
590         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
591         OMP_CLAUSE_SAFELEN.
592         (check_omp_nesting_restrictions): Same.
593         (omp_max_vf): New.
594         (lower_rec_simd_input_clauses): New.
595         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
596         OMP_CLAUSE_LINEAR.
597         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
598         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
599         (expand_omp_build_assign): New.
600         (expand_omp_for_init_counts): New.
601         (expand_omp_for_init_vars): New.
602         (extract_omp_for_update_vars): New.
603         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
604         and rewrite accordingly.
605         (expand_omp_simd): New.
606         (expand_omp_for): Use expand_omp_simd.
607         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
608         (lower_omp_for): Do not lower the body.
609         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
610         in their own loops.
611         * tree-flow.h (find_omp_clause): Remove prototype.
612         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
613         forcing vectorization of the loop, or if flag_tree_vectorize.
614         (gate_tree_if_conversion): Similarly.
615         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
616         gimple_build_omp_for.
617         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
618         * tree-parloops (create_parallel_loop): Pass kind argument to
619         gimple_build_omp_for.
620         * tree-pretty-print.c (dump_omp_clause): Add cases for
621         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
622         OMP_CLAUSE__SIMDUID_.
623         (dump_generic_node): Handle OMP_SIMD.
624         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
625         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
626         unroll OMP_SIMD loops here.
627         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
628         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
629         loop->safelen.
630         (vect_analyze_data_refs): Handle simd loops.
631         * tree-vect-loop.c (vectorizable_live_operation): Handle
632         IFN_GOMP_SIMD*.
633         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
634         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
635         (vectorizable_load): Same.
636         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
637         (struct simduid_to_vf): New.
638         (simduid_to_vf::hash): New.
639         (simduid_to-vf::equal): New.
640         (struct simd_array_to_simduid): New.
641         (simd_array_to_simduid::hash): New.
642         (simd_array_to_simduid::equal): New.
643         (adjust_simduid_builtins): New.
644         (struct note_simd_array_uses_struct): New.
645         (note_simd_array_uses_cb): New.
646         (note_simd_array_uses): New.
647         (vectorize_loops): Handle simd hints and adjust simd builtins
648         accordingly.
649         * tree-vectorizer.h (struct _stmt_vec_info): Add
650         simd_lane_access_p field.
651         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
652         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
653         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
654         (omp_clause_code_name): Same.
655         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
656         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
657         * tree.def (OMP_SIMD): New entry.
658         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
659         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
660         (OMP_CLAUSE_DECL): Adjust range for new clauses.
661         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
662         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
663         (OMP_CLAUSE_LINEAR_STEP): New.
664         (OMP_CLAUSE_SAFELEN_EXPR): New.
665         (OMP_CLAUSE__SIMDUID__DECL): New.
666         (find_omp_clause): New prototype.
668 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
670         * config/i386/driver-i386.c (host_detect_local_cpu): Update
671         Haswell processor detection.
673 2013-08-27  Christian Widmer  <shadow@umbrox.de>
675         PR target/57927
676         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
677         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
678         AVX2 capable processors.
680 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
682         * config/aarch64/arm_neon.h: Replace all inline asm implementations
683         of vget_low_* with implementations in terms of other intrinsics.
685 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
687         PR middle-end/57219
688         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
689         values to -1, 0 and 1.
691 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
693         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
694         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
695         (<optab><mode>3_insn): Remove.
696         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
697         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
698         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
699         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
700         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
701         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
702         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
703         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
704         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
705         (ror<mode>3_insn): Likewise.
706         * config/aarch64/predicates.md (aarch64_simd_register): New.
708 2013-08-27  Richard Biener  <rguenther@suse.de>
710         PR tree-optimization/57521
711         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
712         one edge is non-critical.
713         (find_phi_replacement_condition): Make sure to use a non-critical
714         edge.  Cleanup and remove old bug workarounds.
715         (bb_postdominates_preds): Remove.
716         (if_convertible_loop_p_1): Do not compute post-dominators.
717         (combine_blocks): Do not free post-dominators.
718         (main_tree_if_conversion): Likewise.
719         (pass_data_if_conversion): Add TODO_verify_ssa.
721 2013-08-27  DJ Delorie  <dj@redhat.com>
723         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
725 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
727         * function.c (assign_parm_find_data_types): Set passed_mode and
728         nominal_mode to the TYPE_MODE of nominal_type for the built
729         pointer type in case of the struct-pass-by-reference.
731 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
733         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
734         (UINT16_TYPE): Change default to "unsigned int".
736         * config/avr/avr.opt (mfract-convert-truncate): New option.
737         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
738         is set, round negative fractional integers according to n1169
739         when converting to integer types.
741 2013-08-26  Jan Hubicka  <jh@suse.cz>
743         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
744         methods can not be called indirectly when their address is not taken.
746 2013-08-26  Jan Hubicka  <jh@suse.cz>
748         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
749         ctor_for_folding.
751 2013-08-26  Jan Hubicka  <jh@suse.cz>
753         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
754         can be unshared.
756 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
758         * reload.c (find_valid_class): Allow classes that do not include
759         FIRST_PSEUDO_REGISTER - 1.
761 2013-08-26  Jan Hubicka  <jh@suse.cz>
763         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
764         fix edge count/frequency when speculation failed; fix type check
765         for the direct call.
767 2013-08-26  Jan Hubicka  <jh@suse.cz>
769         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
771 2013-08-26  Jan Hubicka  <jh@suse.cz>
773         * ipa-inline-transform.c (inline_transform): Be ready for basic block
774         to be changed by edge redirection.
776 2013-08-26  Jan Hubicka  <jh@suse.cz>
778         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
779         formating; add sanity check.
780         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
781         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
782         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
783         (resolve_noninline_speculation): Update callee keys, too.
785 2013-08-26  Jan Hubicka  <jh@suse.cz>
787         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
788         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
790 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
792         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
793         into proper place.
795 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
797         * config/i386/i386.c (ix86_debug_options): Remove prototype.
798         (x86_64_elf_select_section): Ditto.
799         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
800         arguments.
801         (ix86_pass_by_reference): Ditto.
802         (output_set_got): Ditto.
803         (ix86_unary_operator_ok): Ditto.
804         (ix86_expand_builtin): Ditto.
806 2013-08-23  Jan Hubicka  <jh@suse.cz>
808         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
810 2013-08-23  Jan Hubicka  <jh@suse.cz>
812         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
813         (tree_decl_with_vis): Add FINAL field.
815 2013-08-23  Jeff Law  <law@redhat.com>
817         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
818         the debugging dump when the expression is fully redundant.
820 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
822         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
823         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
824         * pretty-print.c (pp_formatted_text_data): Likewise.
825         (pp_write_text_to_stream): Likewise.
826         (pp_write_text_as_dot_label_to_stream): Likewise.
827         (pp_append_r): Likewise.
828         (pp_format): Likewise.
829         (pp_flush): Likewise.
830         (pp_clear_output_area): Likewise.
831         (pp_append_text): Likewise.
832         (pp_formatted_text): Likewise.
833         (pp_remaining_character_count_for_line): Likewise.
834         (pp_newline): Likewise.
835         (pp_character): Likewise.
836         (output_buffer::~output_buffer): Define.
837         (pretty_printer::~pretty_printer): Destruct output buffer.
838         * pretty-print.h (output_buffer::~output_buffer): Declare.
839         (pretty_printer::~pretty_printer): Declare virtual.
841 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
843         PR other/57324
844         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
845         HOST_WIDE_INT_M1U): New macros.
846         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
847         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
848         unsigned -1 for lshift.
849         * cse.c (cse_insn): Likewise.
850         * double-int.c (rshift_double, lshift_double): Likewise.
851         * builtins.c (fold_builtin_bitop): Likewise.
852         * combine.c (force_to_mode): Likewise.
853         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
854         * simplify-rtx.c (simplify_const_unary_operation,
855         simplify_const_binary_operation): Likewise.
856         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
857         check_va_list_escapes): Likewise.
858         * rtlanal.c (nonzero_bits1): Likewise.
859         * expmed.c (expand_smod_pow2): Likewise.
860         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
862 2013-08-23  Jan Hubicka  <jh@suse.cz>
864         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
865         as having address taken.
867 2013-08-23  Jan Hubicka  <jh@suse.cz>
869         * ipa-utils.h (method_class_type): Declare.
870         * ipa-devirt.c (method_class_type): Export.
872         * cgraphunit.c (analyze_functions): Do basic devirtualization;
873         do not walk base classes of anonymous types.
875 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
877         PR rtl-optimization/58220
878         PR regression/58221
879         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
880         handle SEQUENCE insns properly.
882 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
884         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
885         definition.
886         (pp_newline_and_indent): Likewise.
887         (pp_separate_with): Likewise.
888         * pretty-print.c (pp_newline_and_flush): Define.
889         (pp_newline_and_indent): Likewise.
890         (pp_separate_with): Likewise.
892 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
894         PR target/58218
895         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
896         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
898 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
900         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
901         * gcc/config/i386/i386.md (*movti_internal): Use
902         predicate to determine if EVEX is needed.
903         (*movsi_internal): Ditto.
904         (*movdf_internal): Ditto.
905         (*movsf_internal): Ditto.
906         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
908 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
910         PR tree-optimization/58209
911         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
912         (find_tail_calls): Give up for pointer result types if m is non-NULL.
913         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
914         emit POINTER_PLUS_EXPR.
915         (create_tailcall_accumulator): For pointer result type accumulate in
916         sizetype type.
918 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
920         * configure.ac: Add backslashes missing from the last change.
921         * configure: Regenerate.
923 2013-08-22  Jan Hubicka  <jh@susue.cz>
925         * ipa.c (function_and_variable_visibility): First remember function
926         was global and then make it local.
928 2013-08-22  Julian Brown  <julian@codesourcery.com>
930         * configure.ac: Add aarch64 to list of arches which use "nop" in
931         debug_line test.
932         * configure: Regenerate.
934 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
936         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
937         gnu_libc_has_function.
938         * config/s390/tpf.h: Likewise.
940 2013-08-22  Jan Hubicka  <jh@susue.cz>
942         * timevar.c (validate_phases): Add cast.
944 2013-08-22  Jan Hubicka  <jh@susue.cz>
946         * timevar.c (validate_phases): Use size_t for memory.
947         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
949 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
951         * pretty-print.h (output_buffer::output_buffer): Declare.
952         (pretty_printer::pretty_printer): Likewise.
953         (pp_construct): Remove.
954         * pretty-print.c (output_buffer::output_buffer): Define.
955         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
956         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
957         (print_gimple_expr): Likewise.
958         (print_gimple_seq): Likewise.
959         (gimple_dump_bb): Likewise.
960         * sched-vis.c (dump_value_slim): Likewise.
961         (dump_insn_slim): Likewise.
962         (dump_rtl_slim): Likewise.
963         (str_pattern_slim): Likewise.
964         * tree-mudflap.c (mf_varname_tree): Likewise.
965         * graph.c (print_graph_cfg): Likewise.
966         (start_graph_dump): Likewise.
967         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
968         placement-new.
969         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
970         pretty printer initialization.
971         * coretypes.h (diagnostic_context): Remove superflous type alias
972         declaration.
973         (pretty_printer): Likewise.  Declare directly as a class.
974         (pretty_print_info): Remove declaration as class.
975         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
976         and pp_clear_output_area.
977         (asan_add_global): Likewise.
979 2013-08-22  Jan Hubicka  <jh@suse.cz>
981         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
982         * ipa-utils.h (update_type_inheritance_graph): Declare.
983         (possible_polymorphic_call_target_p): Declare.
984         (possible_polymorphic_call_target_p): New.
985         * ipa-devirt.c: Update toplevel comments.
986         (cached_polymorphic_call_targets): Move up.
987         (odr_type_d): Move ID down.
988         (polymorphic_type_binfo_p): Update comment.
989         (odr_hasher::remove): Likewise;
990         (get_odr_type): Set anonymous_namespace.
991         (dump_odr_type): Dump it.
992         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
993         (maybe_record_node): Record node in cached_polymorphic_call_targets.
994         (record_binfo): Add comment.
995         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
996         built.
997         (devirt_node_removal_hook): Do not iCE when cache is freed.
998         (possible_polymorphic_call_target_p): New predicate.
999         (update_type_inheritance_graph): New function.
1001 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1002             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1003             Sergey Lega  <sergey.s.lega@intel.com>
1004             Anna Tikhonova  <anna.tikhonova@intel.com>
1005             Ilya Tocar  <ilya.tocar@intel.com>
1006             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1007             Ilya Verbin  <ilya.verbin@intel.com>
1008             Kirill Yukhin  <kirill.yukhin@intel.com>
1009             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1011         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1012         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1013         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1014         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1015         (OPTION_MASK_ISA_AVX2_UNSET): Update.
1016         (OPTION_MASK_ISA_AVX512F_UNSET): New.
1017         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1018         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1019         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1020         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1021         OPT_mavx512pf, OPT_mavx512er cases.
1022         * config/i386/constraints.md (v): New constraint.
1023         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1024         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1025         (bit_AVX512CD): New.
1026         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1027         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1028         * config/i386/i386-c.c (ix86_target_macros_internal):
1029         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1030         __AVX512PF__.
1031         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1032         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1033         * config/i386/i386.c (regclass_map, dbx_register_map)
1034         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1035         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1036         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1037         -mavx512pf options.
1038         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1039         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
1040         -mavx512pf options.  Fix formatting.
1041         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1042         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
1043         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1044         -mavx512cd, -mavx512pf options.
1045         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1046         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1047         (ix86_preferred_output_reload_class): Replace SSE_REGS with
1048         ALL_SSE_REGS.
1049         (ix86_hard_regno_mode_ok): Support 512-bit registers.
1050         (ix86_set_reg_reg_cost): Ditto.
1051         (x86_order_regs_for_local_alloc): Ditto.
1052         (MAX_VECT_LEN): Extend to 64-byte.
1053         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1054         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1055         (TARGET_AVX512ER, TARGET_AVX512CD): New.
1056         (BIGGEST_ALIGNMENT): Extend to 512-bits.
1057         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1058         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1059         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1060         (SSE_REG_MODE_P): Support new modes.
1061         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1062         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1063         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1064         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1065         (REG_CLASS_CONTENTS): Add new registers.
1066         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1067         (EXT_REX_SSE_REGNO_P): New.
1068         (HI_REGISTER_NAMES): Add new registers.
1069         * config/i386/i386.md: Define constants for new registers.
1070         (mode): Add new 512-bit modes.
1071         (prefix): Support evex prefix.
1072         (isa): Support avx512f, noavx512f, fma_avx512f.
1073         (ssemodesuffix): Add new 512-bit modes.
1074         (movxi): New.
1075         (*movxi_internal_avx512f): Ditto.
1076         (*movdi_internal): Replace constraint "x" with the new constraint "v".
1077         Support MODE_XI.
1078         (*movsi_internal): Likewise.
1079         (*movdf_internal): Likewise.
1080         (*movsf_internal): Likewise.
1081         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1082         (<code><mode>3): Likewise.
1083         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1084         New.
1085         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1086         with the new constraint "v".
1087         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1088         modes.
1089         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1090         with the new constraint "v".
1091         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1092         (<sse2>_storedqu<avxsizesuffix>): Likewise.
1093         (*<plusminus_insn><mode>3): Likewise.
1094         (<sse>_vm<plusminus_insn><mode>3): Likewise.
1095         (*mul<mode>3): Likewise.
1096         (<sse>_vmmul<mode>3): Likewise.
1097         (<sse>_div<mode>3): Likewise.
1098         (<sse>_vmdiv<mode>3): Likewise.
1099         (<sse>_sqrt<mode>2): Likewise.
1100         (<sse>_vmsqrt<mode>2): Likewise.
1101         (*<code><mode>3_finite): Likewise.
1102         (*<code><mode>3) <smaxmin>: Likewise.
1103         (<sse>_vm<code><mode>3): Likewise.
1104         (*<code><mode>3) <any_logic>: Likewise.
1105         (*fma_fmadd_<mode>): Likewise.
1106         (*fma_fmsub_<mode>): Likewise.
1107         (*fma_fnmadd_<mode>): Likewise.
1108         (*fma_fnmsub_<mode>): Likewise.
1109         (*fma_fmaddsub_<mode>): Likewise.
1110         (*fma_fmsubadd_<mode>): Likewise.
1111         (*fmai_fmadd_<mode>): Likewise.
1112         (*fmai_fmsub_<mode>): Likewise.
1113         (*fmai_fnmadd_<mode>): Likewise.
1114         (*fmai_fnmsub_<mode>): Likewise.
1115         (sse_cvtsi2ss): Likewise.
1116         (sse_cvtsi2ssq): Likewise.
1117         (sse_cvtss2si): Likewise.
1118         (sse_cvtss2si_2): Likewise.
1119         (sse_cvtss2siq): Likewise.
1120         (sse_cvtss2siq_2): Likewise.
1121         (sse_cvttss2si): Likewise.
1122         (sse_cvtss2siq_2): Likewise.
1123         (float<sseintvecmodelower><mode>2): Likewise.
1124         (sse2_cvtsd2si_2): Likewise.
1125         (sse2_cvtsd2siq_2): Likewise.
1126         (*<plusminus_insn><mode>3): Likewise.
1127         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1128         (*<sse4_1_avx2>_mul<mode>3): Likewise.
1129         (ashr<mode>3): Likewise.
1130         (<shift_insn><mode>3): Likewise.
1131         (avx2_<code><mode>3): Likewise.
1132         (*avx2_<code><mode>3): Likewise.
1133         (*andnot<mode>3): Likewise.
1134         (*<code><mode>3) <any_logic>: Likewise.
1135         (abs<mode>2): Likewise.
1136         (avx2_permvar<mode>): Likewise.
1137         (avx2_perm<mode>_1): Likewise.
1138         (*avx_vpermilp<mode>): Likewise.
1139         (avx_vpermilvar<mode>3): Likewise.
1140         (avx2_ashrv<mode>): Likewise.
1141         (avx2_<shift_insn>v<mode>): Likewise.
1142         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1143         -mavx512cd.
1144         * doc/rtl.texi: Document XImode.
1146 2013-08-21  Jeff Law  <law@redhat.com>
1148         * tree-flow.h (register_jump_thread): Pass vector of edges
1149         instead of each important edge.
1150         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1151         thread path into a vector and pass that to register_jump_thread.
1152         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1153         passed in edge vector to the current 3-edge form.
1155         Revert:
1156         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1158         * dce.c (fini_dce): Call df_analyze again just in case
1159         delete_unmarked_insns removed anything.
1161 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
1163         * reload.h (struct reg_equivs): Rename to ..
1164         (struct reg_equivs_s): .. this.
1166 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
1168         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1170 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1172         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1174 2013-08-21  Jeff Law  <law@redhat.com>
1176         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1177         simplify assignments too.  If the RHS collapses to a singleton
1178         range, then return the value for the range.
1180 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
1182         * config/i386/sse.md (V16): Rename to...
1183         (VMOVE): this.
1184         (mov<mode>): Update iterator name.
1185         (*mov<mode>_internal): Ditto.
1186         (push<mode>1): Ditto.
1187         (movmisalign<mode>): Ditto.
1189 2013-08-20  Jan Hubicka  <jh@suse.cz>
1191         PR bootstrap/58186
1192         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1193         entry for direct edges.
1194         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1196 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1198         Revert my last two changes, r201865 and r201864:
1200         Revert r201865:
1201         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1203         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1204         instances can own GC refs.
1206         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1207         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1208         (gcc::context::gt_pch_nx): Likewise.
1209         (gcc::context::gt_pch_nx):  Likewise.
1210         * ggc.h (gt_ggc_mx <T>): New.
1211         (gt_pch_nx_with_op <T>): New.
1212         (gt_pch_nx <T>): New.
1213         * passes.c (opt_pass::gt_ggc_mx): New.
1214         (opt_pass::gt_pch_nx): New.
1215         (opt_pass::gt_pch_nx_with_op): New.
1216         (pass_manager::gt_ggc_mx): New.
1217         (pass_manager::gt_pch_nx): New.
1218         (pass_manager::gt_pch_nx_with_op): New.
1219         (pass_manager::operator new): Use
1220         ggc_internal_cleared_alloc_stat rather than xcalloc.
1221         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1222         (pass_manager::gt_ggc_mx): New.
1223         (pass_manager::gt_pch_nx): New.
1224         (pass_manager::gt_pch_nx_with_op): New.
1225         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1226         (opt_pass::operator new): New.
1227         (opt_pass::gt_ggc_mx): New.
1228         (opt_pass::gt_pch_nx): New.
1229         (opt_pass::gt_pch_nx_with_op): New.
1231         Revert r201864:
1232         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1234         * Makefile.in (GTFILES): Add context.h.
1235         * context.c (gcc::context::operator new): New.
1236         (gcc::context::gt_ggc_mx): New.
1237         (gcc::context::gt_pch_nx): New.
1238         (gcc::context::gt_pch_nx): New.
1239         * context.h (gcc::context): Add GTY((user)) marking.
1240         (gcc::context::operator new): New.
1241         (gcc::context::gt_ggc_mx): New.
1242         (gcc::context::gt_pch_nx): New.
1243         (gcc::context::gt_pch_nx): New.
1244         (g): Add GTY marking.
1245         (gt_ggc_mx (gcc::context *)): New.
1246         (gt_pch_nx (gcc::context *)): New.
1247         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1248         void *cookie)): New.
1249         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1251 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1253         * dce.c (fini_dce): Call df_analyze again just in case
1254         delete_unmarked_insns removed anything.
1256 2013-08-20  Teresa Johnson  <tejohnson@google.com>
1258         PR rtl-optimizations/57451
1259         * final.c (reemit_insn_block_notes): Prevent lexical blocks
1260         from crossing split section boundaries.
1262 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1264         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1265         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1266         with MULTLIB_DEFAULTS.
1268 2013-08-20  Nick Clifton  <nickc@redhat.com>
1270         * target.def (narrow_volatile_bitfield): Note that the default
1271         value is false, not !TARGET_STRICT_ALIGN.
1272         * doc/tm.texi: Regenerate.
1274 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
1276         Fix LIB_SPEC for systems without libpthread.
1278         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1279         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1280         for Android.
1281         * config/i386/linux-common.h: Likewise.
1282         * config/mips/linux-common.h: Likewise.
1284 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
1286         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1287         checks.
1289 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1291         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1292         instances can own GC refs.
1294         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1295         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1296         (gcc::context::gt_pch_nx): Likewise.
1297         (gcc::context::gt_pch_nx):  Likewise.
1298         * ggc.h (gt_ggc_mx <T>): New.
1299         (gt_pch_nx_with_op <T>): New.
1300         (gt_pch_nx <T>): New.
1301         * passes.c (opt_pass::gt_ggc_mx): New.
1302         (opt_pass::gt_pch_nx): New.
1303         (opt_pass::gt_pch_nx_with_op): New.
1304         (pass_manager::gt_ggc_mx): New.
1305         (pass_manager::gt_pch_nx): New.
1306         (pass_manager::gt_pch_nx_with_op): New.
1307         (pass_manager::operator new): Use
1308         ggc_internal_cleared_alloc_stat rather than xcalloc.
1309         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1310         (pass_manager::gt_ggc_mx): New.
1311         (pass_manager::gt_pch_nx): New.
1312         (pass_manager::gt_pch_nx_with_op): New.
1313         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1314         (opt_pass::operator new): New.
1315         (opt_pass::gt_ggc_mx): New.
1316         (opt_pass::gt_pch_nx): New.
1317         (opt_pass::gt_pch_nx_with_op): New.
1319 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1321         * Makefile.in (GTFILES): Add context.h.
1322         * context.c (gcc::context::operator new): New.
1323         (gcc::context::gt_ggc_mx): New.
1324         (gcc::context::gt_pch_nx): New.
1325         (gcc::context::gt_pch_nx): New.
1326         * context.h (gcc::context): Add GTY((user)) marking.
1327         (gcc::context::operator new): New.
1328         (gcc::context::gt_ggc_mx): New.
1329         (gcc::context::gt_pch_nx): New.
1330         (gcc::context::gt_pch_nx): New.
1331         (g): Add GTY marking.
1332         (gt_ggc_mx (gcc::context *)): New.
1333         (gt_pch_nx (gcc::context *)): New.
1334         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1335         void *cookie)): New.
1336         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1338 2013-08-20  Alan Modra  <amodra@gmail.com>
1340         PR target/57865
1341         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1342         (rs6000_emit_epilogue): Likewise.
1344 2013-08-19  Dehao Chen  <dehao@google.com>
1346         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1348 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
1349             Jakub Jelinek  <jakub@redhat.com>
1351         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1352         (BUILT_IN_FABSD64): Likewise.
1353         (BUILT_IN_FABSD128): Likewise.
1354         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1355         (fold_builtin_1): Likewise.
1356         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1357         destination and source operands.
1358         (*abstd2_fpr): Likewise.
1359         (*nabstd2_fpr): Likewise.
1361 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
1363         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1364         JUMP_P and INSN_P.
1366 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
1368         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1369         incomplete.
1371 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1373         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1374         * builtins.c (default_libc_has_function): New.
1375         (gnu_libc_has_function): Ditto.
1376         (no_c99_libc_has_function): Ditto.
1377         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1378         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1379         (fold_builtin_sincos): Likewise.
1380         (fold_builtin_cexp): Likewise.
1381         * builtins.def (DEF_C94_BUILTIN): Likewise.
1382         (DEF_C99_BUILTIN): Likewise.
1383         (DEF_C99_C90RES_BUILTIN): Likewise.
1384         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1385         definitions to using this define.
1386         * config/darwin-protos.h (darwin_libc_has_function): New.
1387         * config/darwin.c (darwin_libc_has_function): Ditto.
1388         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1389         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1390         * config/darwin.h: Ditto.
1391         * config/elfos.h: Ditto.
1392         * config/freebsd.h: Ditto.
1393         * config/i386/cygming.h: Ditto.
1394         * config/i386/djgpp.h: Ditto.
1395         * config/i386/i386-interix.h: Ditto.
1396         * config/microblaze/microblaze.h: Ditto.
1397         * config/mmix/mmix.h: Ditto.
1398         * config/gnu-user.h: Ditto.
1399         * config/ia64/hpux.h: Ditto.
1400         * config/pa/pa-hpux.h: Ditto.
1401         * config/pdp11/pdp11.h: Ditto.
1402         * config/picochip/picochip.h: Ditto.
1403         * config/linux.h: Ditto.
1404         * config/netbsd.h: Ditto.
1405         * config/openbsd.h: Ditto.
1406         * config/rs6000/aix43.h: Ditto.
1407         * config/rs6000/aix51.h: Ditto.
1408         * config/rs6000/aix52.h: Ditto.
1409         * config/rs6000/aix53.h: Ditto.
1410         * config/rs6000/aix61.h: Ditto.
1411         * config/rs6000/darwin.h: Ditto.
1412         * config/rs6000/linux.h: Ditto.
1413         * config/rs6000/linux64.h: Ditto.
1414         * config/s390/tpf.h: Ditto.
1415         * config/sol2-10.h: Ditto.
1416         * config/sol2.h: Ditto.
1417         * config/vms/vms.h: Ditto.
1418         * config/vxworks.h: Ditto.
1419         * config/linux-android.c (linux_android_libc_has_function):
1420         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1421         * config/linux-protos.h (linux_android_libc_has_function):
1422         New declaration.
1423         * config/i386/i386.c (ix86_libc_has_function): New.
1424         * config/i386/i386-protos.h
1425         (ix86_libc_has_function): New declaration.
1426         * config/i386/i386.md
1427         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1428         ("isinf<mode>2): Likewise.
1429         * convert.c (convert_to_integer): Using new target hook
1430         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1431         TARGET_C99_FUNCTIONS.
1432         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1433         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1434         * coretypes.h (function_class): New enum for different
1435         classes of functions.
1436         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1437         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1438         (TARGET_HAS_SINCOS): Likewise.
1439         (TARGET_LIBC_HAS_FUNCTION): New.
1440         * doc/tm.texi: Regenerated.
1441         * targhooks.h (default_libc_has_function): New declaration.
1442         (no_c99_libc_has_function): Ditto.
1443         (gnu_libc_has_function): Ditto.
1444         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1445         and TARGET_HAS_SINCOS.
1447 2013-08-18  Jan Hubicka  <jh@suse.cz>
1449         * Makeifle-in (ipa-devirt.o): New.
1450         (GTFILES): Add ipa-utils.h and ipa-devirt.c
1451         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1452         (analyze_functions): Look into possible targets of polymorphic call.
1453         * dumpfile.c (dump_files): Add type-inheritance dump.
1454         * dumpfile.h (TDI_inheritance): New.
1455         * ipa-devirt.c: New file.
1456         * ipa-utils.h (odr_type_d): Forward declare.
1457         (odr_type): New type.
1458         (build_type_inheritance_graph): Declare.
1459         (possible_polymorphic_call_targets): Declare and introduce inline
1460         variant when only edge is pased.
1461         (dump_possible_polymorphic_call_targets): Likewise.
1462         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1463         * tree.c (type_in_anonymous_namespace_p): Break out from ...
1464         (types_same_for_odr): ... here.
1465         * tree.h (type_in_anonymous_namespace_p): Declare.
1467 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
1469         PR tree-optimization/58006
1470         * tree-parloops.c (take_address_of): Don't ICE if get_name
1471         returns NULL.
1472         (eliminate_local_variables_stmt): Remove clobber stmts.
1474 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
1476         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1477         error message is issued for an alias to undefined symbol.
1479 2013-08-18  Jan Hubicka  <jh@suse.cz>
1481         * cgraph.c (cgraph_create_indirect_edge): Discover
1482         polymorphic calls and record basic info into indirect_info.
1483         * gimple-fold.c (gimple_fold_call): When doing BINFO based
1484         devirtualization, ignore objc function calls.
1485         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1486         call with no parm index info.
1487         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1488         * tree.c (virtual_method_call_p): New function.
1489         * tree.h (virtual_method_call_p): Declare.
1491 2013-08-16  Jan Hubicka  <jh@suse.cz>
1493         PR middle-end/58179
1494         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1496 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
1498         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1499         attribute.
1501 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1503         * gengtype.c (type_for_name): Add special-case support for
1504         locating types within the "gcc::" namespace.
1505         (open_base_files): Emit a "using namespace gcc" directive.
1507 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1509         PR target/58160
1510         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1511         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1513         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1514         array instead of each individual operand as a separate argument.
1515         (emit_fusion_gpr_load): Likewise.
1516         (expand_fusion_gpr_load): Add new function declaration.
1518         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1519         signature to have the operands passed as an array, instead of as
1520         separate arguments.  Allow ZERO_EXTEND to be in the memory
1521         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
1522         depend on the register live/dead flags when peepholes are run.
1523         (expand_fusion_gpr_load): New function to be called from the
1524         peephole2 pass, to change the register that addis sets to be the
1525         target register.
1526         (emit_fusion_gpr_load): Change the calling signature to have the
1527         operands passed as an array, instead of as separate arguments.
1528         Allow ZERO_EXTEND to be in the memory address, and also
1529         SIGN_EXTEND if -mpower8-fusion-sign.
1531         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1532         unspec enumeration.
1533         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1534         adjust the register addis loads up in the peephole2 pass.  Do not
1535         depend on the register live/dead state when the peephole pass is done.
1537 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1539         * gengtype.c (create_user_defined_type): Ensure that the kind
1540         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1541         declaration is seen before the GTY((user)) marking.
1543 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1545         PR target/58105
1546         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1548 2013-08-16  Jan Hubicka  <jh@suse.cz>
1550         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1551         arugment expected_type.
1552         (gimple_fold_call): Use it.
1553         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1554         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1555         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1556         (try_make_edge_direct_virtual_call): Likewise.
1557         * tree.c (obj_type_ref_class): New.
1558         * tree.h (obj_type_ref_class): Use it.
1560 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1562         * sched-vis.c (rtl_slim_pp_initialized): Remove.
1563         (rtl_slim_pp): Likewise.
1564         (init_rtl_slim_pretty_print): Likewise.
1565         (dump_value_slim):  Don't call it.  Use local pretty printer.
1566         (dump_insn_slim): Likewise.
1567         (dump_rtl_slim): Likewise.
1568         (str_pattern_slim): Likewise.
1569         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1570         Simplify.
1572 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
1574         PR tree-optimization/58164
1575         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1576         walk gimple_goto_dest of GIMPLE_GOTO.
1578         PR tree-optimization/58165
1579         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1580         bi_call must be the last stmt in a bb, don't split_block, instead
1581         use fallthru edge from it and give up if there is none.
1582         Release conds vector when returning early.
1584 2013-08-14  Xinliang David Li  <davidxl@google.com>
1586         * config/i386/i386.c (ix86_option_override_internal):
1587         Remove unused variable and field.
1589 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1591         PR target/57949
1592         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1593         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1594         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1595         and Linux, correct BLKmode alignment when 128-bit alignment is
1596         required and compatibility flag is not set.
1597         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1598         for zero-size arguments when compatibility flag is not set.
1600 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1602         PR tree-optimization/58145
1603         * tree-sra.c (build_ref_for_offset): If prev_base has
1604         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1606 2013-08-14  Xinliang David Li  <davidxl@google.com>
1608         * config/i386/i386.c (ix86_option_override_internal):
1609         Fix uninitialized variable error.
1611 2013-08-14  Xinliang David Li  <davidxl@google.com>
1613         * config/i386/i386.opt: Define two new options.
1614         * config/i386/x86-tune.def: Add arch selector field in macros.
1615         * config/i386/i386.h: Adjust macro definition.
1616         * config/i386/i386.c (ix86_option_override_internal):
1617         Refactor the code.
1618         (parse_mtune_ctrl_str): New function.
1619         (set_ix86_tune_features): New function.
1620         (ix86_function_specific_restore): Call the new helper function.
1622 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
1624         PR rtl-optimization/57662
1625         * sel-sched.c (code_motion_process_successors): When the current insn
1626         is removed after the recursive traversal, break from the loop.
1627         Add comments and debug printouts.
1629 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1630             Alexandre Oliva  <aoliva@redhat.com>
1632         PR target/58067
1633         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1634         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1635         there also UNSPEC_PLTOFF.
1637 2013-08-14  Marek Polacek  <polacek@redhat.com>
1639         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1640         NUM_CONDITIONS bit positions.
1642 2013-08-13  Cary Coutant  <ccoutant@google.com>
1644         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1645         (attr_checksum): Hash vector contents instead of pointer.
1646         (attr_checksum_ordered): Likewise.
1648 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
1650         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1651         when Pmode != word_mode.  Add length_address attribute.
1652         (sse3_monitor_<mode>): Merge from sse3_monitor and
1653         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
1654         Pmode != word_mode.  Update insn length attribute.
1655         * config/i386/i386.c (ix86_option_override_internal): Update
1656         ix86_gen_monitor selection for merged sse3_monitor insn.
1658 2013-08-13  Julian Brown  <julian@codesourcery.com>
1660         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1661         perform invalid legitimization on greater-than-word-size modes for
1662         TARGET_E500_DOUBLE.
1664 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
1666         * ira.c (setup_class_translate_array): Use aclass instead of cl
1667         for classes not fully covered by allocno classes.
1669 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
1671         PR tree-optimization/57661
1672         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1673         * tree-inline.c (tree_function_versioning): Initialize it.
1674         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1675         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1676         that is not being copied.
1678         PR sanitizer/56417
1679         * asan.c (instrument_strlen_call): Fix typo in comment.
1680         Use char * type even for the lhs of POINTER_PLUS_EXPR.
1682 2013-08-13  Steve Ellcey  <sellcey@mips.com>
1684         * config/mips/mips.md (prefetch): Use lw instead of ld on
1685         loongson in 32bit mode.
1687 2013-08-13  Nick Clifton  <nickc@redhat.com>
1689         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1691 2013-08-13  Jan Hubicka  <jh@suse.cz>
1693         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1694         introduced edge; fix typo in sanity check.
1695         (cgraph_resolve_speculation): Export; improve diagnostic.
1696         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1697         speculation at type mismatch.
1698         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1699         (cgraph_resolve_speculation): Declare.
1700         (symtab_can_be_discarded): New function.
1701         * value-prof.c (gimple_ic_transform): Remove actual transform code.
1702         * ipa-inline-transform.c (speculation_removed): New global var.
1703         (clone_inlined_nodes): See if speculation can be removed.
1704         (inline_call): If speculations was removed, we growths may not match.
1705         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1706         (speculation_useful_p): New function.
1707         (resolve_noninline_speculation): New function.
1708         (inline_small_functions): Resolve useless speculations.
1709         * ipa-inline.h (speculation_useful_p): Declare
1710         * ipa.c (can_replace_by_local_alias): Simplify.
1711         (ipa_profile): Produce speculative calls in non-lto, too;
1712         add simple cost model; produce local aliases.
1714 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
1716         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1717         PASS_MANAGER_H.
1719 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
1721         * config/i386/i386.c (ix86_function_versions): Use error + inform.
1723 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
1725         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1726         iterator instead of X87MODEF.
1728 2013-08-12  Perez Read  <netfirewall@gmail.com>
1730         PR target/58132
1731         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
1732         operand 0 for intel asm alternative.
1733         (*movabs<mode>_2): Ditto for operand 1.
1735 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
1737         * config/aarch64/arm_none.h
1738         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1740 2013-08-12  Nick Clifton  <nickc@redhat.com>
1742         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1744 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1746         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1747         expand for QI/HImode operand to produce more effictive code for
1748         unsigned char(short) --> float(double) conversion.
1750 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1752         * doc/invoke.texi: Mention that -ftls-model does not force the final
1753         model.
1755 2013-08-12  Marek Polacek  <polacek@redhat.com>
1756             Marc Glisse  <marc.glisse@inria.fr>
1758         PR tree-optimization/57980
1759         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1760         when creating -1 constant.
1762 2013-08-10  Jan Hubicka  <jh@suse.cz>
1764         Workaround binutils PR14342.
1765         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1766         (gimple_init_edge_profiler): Likewise.
1767         (gimple_gen_ic_func_profiler): Likewise.
1769 2013-08-09  Jan Hubicka  <jh@suse.cz>
1771         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1773 2013-08-09  Xinliang David Li  <davidxl@google.com>
1775         * config/i386/stringop.def: New file.
1776         * config/i386/stringop.opt: New file.
1777         * config/i386/i386-opts.h: Include stringopt.def.
1778         * config/i386/i386.opt: Include stringopt.opt.
1779         * config/i386/i386.c (ix86_option_override_internal):
1780         Override default size based stringop inline strategies with options.
1781         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1782         New function.
1784 2013-08-09  Jan Hubicka  <jh@suse.cz>
1786         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1788 2013-08-09  Jan Hubicka  <jh@suse.cz>
1790         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1791         CGRAPH_FREQ_MAX.
1792         (dump_cgraph_node): Dump profile-id.
1793         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1794         and common_target_probability.
1795         * lto-cgraph.c (lto_output_edge): Stream common targets.
1796         (lto_output_node): Stream profile ids.
1797         (input_node): Stream profile ids.
1798         (input_edge): Stream common targets.
1799         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1800         * ipa.c: Include value-prof.h
1801         (ipa_profile_generate_summary): Turn indirect call statement histograms
1802         into common targets.
1803         (ipa_profile): Turn common targets into speculative edges.
1805 2013-08-09  Jan Hubicka  <jh@suse.cz>
1807         * cgraph.h (cgraph_node): Add profile_id.
1808         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1809         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1810         (del_node_map): Update.
1811         (find_func_by_funcdef_no): Replace by ...
1812         (find_func_by_profile_id): ... this one.
1813         (gimple_ic_transform): Do not remove useful histograms when
1814         speculation is not done; dump info when indirect call removal
1815         can happen at LTO.
1816         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1817         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1818         (__gcov_indirect_call_profiler_v2): .. this one.
1819         * profile.h (init_node_map): Update.
1820         * coverage.c (coverage_compute_profile_id): New function.
1821         * coverage.h (coverage_compute_profile_id): Declare.
1822         * tree-profile.c (init_ic_make_global_vars): Make
1823         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1824         (gimple_init_edge_profiler): Update prototype of
1825         __gcov_indirect_call_profiler.
1826         (gimple_gen_ic_func_profiler): Simplify.
1827         (tree_profiling): Use init_node_map
1829 2013-08-09  Jan Hubicka  <jh@suse.cz>
1831         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1832         non-speculative refs.
1833         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1834         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1835         (cgraph_set_call_stmt): Likewise.
1836         (cgraph_create_edge_1): Fix release checking compilatoin;
1837         clear lto_stmt_uid.
1838         (cgraph_free_edge): Free indirect info.
1839         (cgraph_turn_edge_to_speculative): New function.
1840         (cgraph_speculative_call_info): New function.
1841         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1842         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1843         (dump_cgraph_node): Dump speculation.
1844         (verify_edge_count_and_frequency): Accept speculative edges.
1845         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1846         (verify_cgraph_node): Handle speculation.
1847         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1848         (cgraph_set_call_stmt): Update prototype.
1849         (cgraph_make_edge_direct): Update prototype.
1850         (cgraph_speculative_call_info): Declare.
1851         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1852         to change; update call of ipa_find_references.
1853         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1854         lto_stmt_uid and speculative flags.
1855         (ipa_dump_references): Dump speculation.
1856         (ipa_clone_references): Clone speculative flag.
1857         (ipa_clone_referring): Likewise.
1858         (ipa_clone_ref): New function.
1859         (ipa_find_reference): Look into lto_stmt_uids
1860         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1861         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1862         (ipa_find_reference): Update declaration.
1863         (ipa_clone_ref): Declare.
1864         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1865         stream speculative flag.
1866         (lto_output_ref): Stream statements uids and speculation.
1867         (input_ref): Likewise.
1868         (input_edge): Stream speuclation.
1869         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1870         (cgraph_set_call_stmt_including_clones): Handle speculation.
1871         * ipa-inline.c (heap_edge_removal_hook): New function.
1872         (inline_small_functions): Register it.
1873         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1874         also initialize refs.
1875         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1876         edge to change.
1877         (try_make_edge_direct_simple_call): Likewise.
1878         (try_make_edge_direct_simple_call): Likewise.
1879         (update_indirect_edges_after_inlining): Likewise.
1880         (remove_described_reference): Look proper lto_stmt_uid.
1881         (propagate_controlled_uses): Likewise.
1882         (propagate_controlled_uses): Liekwise.
1883         * tree-inline.c (copy_bb): Copy speculative edges.
1884         (redirect_all_calls): New function.
1885         (copy_cfg_body): Do redirection after loop info is updated.
1886         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1888 2013-08-09  Jan Hubicka  <jh@suse.cz>
1890         * lto-streamer-out.c (output_function): Renumber PHIs.
1891         * lto-streamer-in.c (input_function): Likewise.
1893 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1895         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1896         (get_lane_unsigned): Likewise.
1897         (dup_lane_scalar): Likewise.
1898         (get_lane): enable for VALL.
1899         * config/aarch64/aarch64-simd.md
1900         (aarch64_dup_lane_scalar<mode>): Remove.
1901         (aarch64_get_lane_signed<mode>): Likewise.
1902         (aarch64_get_lane_unsigned<mode>): Likewise.
1903         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1904         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1905         (aarch64_get_lane<mode>): Enable for all vector modes.
1906         (aarch64_get_lanedi): Remove misleading constraints.
1907         * config/aarch64/arm_neon.h
1908         (__aarch64_vget_lane_any): Define.
1909         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1910         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1911         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1912         * config/aarch64/iterators.md (VDQQH): New.
1913         (VDQQHS): Likewise.
1914         (vwcore): Likewise.
1916 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1918         * configure.ac: Add GAS check for LEON instructions on SPARC.
1919         * configure: Regenerate.
1920         * config.in: Likewise.
1921         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1922         sparc*-*-* block.
1923         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1924         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1925         for LEON or LEON3.
1926         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1927         (AS_LEON_FLAG): New macro.
1928         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1929         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1930         Deal with LEON and LEON3 for the memory model.
1931         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1932         (atomic_compare_and_swap<mode>_1): Likewise.
1933         (*atomic_compare_and_swap<mode>_1): Likewise.
1935 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1937         * config/arm/neon.md (vcond): Fix floating-point vector
1938         comparisons against 0.
1940 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1942         * lra-constraints.c (emit_spill_move): Remove assert.
1943         (process_alt_operands): Add more debugging
1944         output.  Increase reject for spilling into memory.  Decrease
1945         reject for reloading scratch.
1946         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1948 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1950         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1951         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1952         micromips incompatible.  Add nan2008.
1953         (MULTILIB_DIRNAMES): Add nan2008.
1954         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1955         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1956         and micromips incompatible.  Add nan2008.
1957         (MULTILIB_DIRNAMES): Add nan2008.
1958         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1960 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1962         PR rtl-optimization/58079
1963         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1964         simplify_comparison has widened a comparison with an integer.
1966 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1968         * config/arm/neon.md (movmisalign<mode>): Disable when we
1969         don't allow unaligned accesses.
1970         (*movmisalign<mode>_neon_store): Likewise.
1971         (*movmisalign<mode>_neon_load): Likewise.
1972         (*movmisalign<mode>_neon_store): Likewise.
1973         (*movmisalign<mode>_neon_load): Likewise.
1975 2013-08-08  Jan Hubicka  <jh@suse.cz>
1977         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1978         (make_pass_rebuild_cgraph_edges): Also clear references.
1979         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1980         * ipa-inline-transform.c (inline_transform): Remove all references
1981         after inlining.
1982         * cgraphunit.c (expand_function): Remove all references after
1983         expansion.
1984         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1985         (ipa_find_reference): Rewrite to iterator.
1986         (remove_stmt_references): Likewise.
1987         (ipa_clear_stmts_in_references): New function.
1988         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1989         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1990         clear references.
1991         * ipa-split.c (split_function): Remove references in split function.
1993 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1995         PR target/57431
1996         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1997         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1999 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2001         PR target/56979
2002         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2003         suggested mode for the assignment isn't compatible with the
2004         registers required.
2006 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2008         PR target/58065
2009         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2011 2013-08-07  Xinliang David Li  <davidxl@google.com>
2013         * config/i386/i386.opt: New option -mtune-ctrl=.
2014         * config/i386/x86-tune.def: New file.
2015         * config/i386/i386.h: include x86-tune.def.
2016         * config/i386/i386.c (ix86_option_override_internal):
2017         Parsing -mtune-ctrl= option and set tune features.
2019 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
2021         PR other/12081
2022         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2023         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2024         to gen_2arg_fn_t.
2026 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2028         * rtl.h (update_alignments): Declare.
2029         * final.c (grow_label_align): New function extracted from...
2030         (shorten_branches): ...here.  Call it.
2031         (update_alignments): New function.
2032         * reorg.c (sibling_labels): New variable.
2033         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
2034         the new label along with it onto the sibling_labels vector.
2035         (fill_simple_delay_slots): Adjust call to get_label_before.
2036         (fill_slots_from_thread): Likewise.
2037         (relax_delay_slots): Likewise.
2038         (make_return_insns): Likewise.
2039         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2041 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2043         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2044         document its semantics.
2045         (diagnostic_report_diagnostic): Adjust accordingly.
2047 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2049         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2050         (sparc_option_override): ...and port to new C++ pass API.
2051         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2053 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
2055         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2057 2013-08-06  Caroline Tice  <cmtice@google.com>
2059         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2060         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2061         * tree-pass.h: Add pass_vtable_verify.
2062         * varasm.c (assemble_variable): Add code to properly set the comdat
2063         section and name for the .vtable_map_vars section.
2064         (assemble_vtyv_preinit_initializer): New function.
2065         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
2066         LINK_ONCE flag.
2067         * output.h: Add function decl for assemble_vtv_preinit_initializer.
2068         * vtable-verify.c: New file.
2069         * vtable-verify.h: New file.
2070         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2071         initialiation levels.
2072         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2073         * passes.def: Insert pass_vtable_verify.
2074         * aclocal.m4: Reorder includes.
2075         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
2076         -fvtv-counts options.
2077         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2078         as appropriate, if -fvtable-verify=... is used.
2079         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2080         -fvtable-verify=... is used.
2081         * Makefile.in (OBJS):  Add vtable-verify.o to list.
2082         (vtable-verify.o): Add new build rule.
2083         (GTFILES): Add vtable-verify.c to list.
2084         * common.opt (fvtable-verify=): New flag.
2085         (vtv_priority): Values for fvtable-verify= flag.
2086         (fvtv-counts): New flag.
2087         (fvtv-debug): New flag.
2088         * tree.h (save_vtable_map_decl): New extern function decl.
2090 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2092         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2093         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2094         (pass_data_rl78_devirt): ...new pass_data instance and...
2095         (make_pass_rl78_devirt): ...new function.
2096         (rl78_asm_file_start): Port pass registration to new C++ API.
2098 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2100         * coretypes.h (rtl_opt_pass): Add.
2101         (gcc::context): Add.
2102         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2103         (epiphany_init): Port to new C++ pass API.
2104         (epiphany_optimize_mode_switching): Likewise.
2105         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2106         (pass_manager::get_pass_mode_switching): New.
2107         (pass_manager::get_pass_peephole2): New.
2108         * mode-switching.c (pass_mode_switching): Add clone method.
2109         * recog.c (pass_peephole2): Add clone method.
2110         (pass_split_all_insns): Add clone method.
2112 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
2114         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2115         (mips_option_override): ...here, porting to new C++ API for passes.
2117 2013-08-06  Jan Hubicka  <jh@suse.cz>
2119         * cgraph.c (cgraph_get_body): New function based on lto.c
2120         implementation.
2121         * cgraph.h (cgraph_get_body): Declare.
2122         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2123         LTO paths.
2124         * cgraphunit.c (expand_function): Get body prior expanding.
2125         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2126         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2127         really need.
2128         * passes.c (do_per_function_toporder): Get body.
2129         * tree-inline.c (expand_call_inline): Get body prior inlining it.
2130         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2132 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2134         PR fortran/57987
2135         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2136         is not re-finalized.  Rename second parameter to no_collect.
2138 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2140         PR middle-end/58041
2141         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2142         MEM_REF has proper alignment information.
2144 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
2146         PR other/12081
2147         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2148         class insn_gen_fn.
2149         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2150         rtx (*) (rtx, ...) with insn_gen_fn.
2151         * genoutput.c (output_insn_data): Cast gen_? function pointers to
2152         insn_gen_fn::stored_funcptr.  Add initializer braces.
2154 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2156         Rewrite how instances of passes are cloned to remove assumptions
2157         about their sizes (thus allowing pass subclasses to have
2158         additional data fields, albeit non-GC-managed ones at this point).
2160         * passes.c (make_pass_instance): Now that passes have clone
2161         methods, rewrite this function to eliminate XNEW and memcpy
2162         calls that used hardcoded sizes.  Since this function no longer
2163         creates pass instances, rename it to...
2164         (add_pass_instance): ...this.  Document the old way that passes were
2165         numbered and flagged, and rework this function to continue using it.
2166         (next_pass_1): Add an initial_pass argument for use by
2167         add_pass_instance.
2168         (position_pass): When adding multiple instances of a pass, use
2169         the pass's clone method, rather than relying on the XNEW/memcpy
2170         within the former make_pass_instance (now add_pass_instance).
2171         (pass_manager::pass_manager): When invoking next_pass_1, also supply
2172         the initial instance of the current pass within the pass manager.
2174 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2176         This is the automated part of the conversion of passes from C
2177         structs to C++ classes.
2179         Patch autogenerated by refactor_passes.py from
2180         https://github.com/davidmalcolm/gcc-refactoring-scripts
2181         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2183         * asan.c (pass_asan): Convert from a global struct to a subclass of
2184         gimple_opt_pass along with...
2185         (pass_data_asan): ...new pass_data instance and...
2186         (make_pass_asan): ...new function.
2187         (pass_asan_O0): Convert from a global struct to a subclass of
2188         gimple_opt_pass along with...
2189         (pass_data_asan_O0): ...new pass_data instance and...
2190         (make_pass_asan_O0): ...new function.
2191         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2192         subclass of rtl_opt_pass along with...
2193         (pass_data_inc_dec): ...new pass_data instance and...
2194         (make_pass_inc_dec): ...new function.
2195         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2196         a subclass of rtl_opt_pass along with...
2197         (pass_data_reorder_blocks): ...new pass_data instance and...
2198         (make_pass_reorder_blocks): ...new function.
2199         (pass_duplicate_computed_gotos): Convert from a global struct to a
2200         subclass of rtl_opt_pass along with...
2201         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2202         (make_pass_duplicate_computed_gotos): ...new function.
2203         (pass_partition_blocks): Convert from a global struct to a subclass of
2204         rtl_opt_pass along with...
2205         (pass_data_partition_blocks): ...new pass_data instance and...
2206         (make_pass_partition_blocks): ...new function.
2207         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2208         struct to a subclass of rtl_opt_pass along with...
2209         (pass_data_branch_target_load_optimize1): ...new pass_data instance
2210         and...
2211         (make_pass_branch_target_load_optimize1): ...new function.
2212         (pass_branch_target_load_optimize2): Convert from a global struct to a
2213         subclass of rtl_opt_pass along with...
2214         (pass_data_branch_target_load_optimize2): ...new pass_data instance
2215         and...
2216         (make_pass_branch_target_load_optimize2): ...new function.
2217         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2218         of rtl_opt_pass along with...
2219         (pass_data_jump): ...new pass_data instance and...
2220         (make_pass_jump): ...new function.
2221         (pass_jump2): Convert from a global struct to a subclass of
2222         rtl_opt_pass along with...
2223         (pass_data_jump2): ...new pass_data instance and...
2224         (make_pass_jump2): ...new function.
2225         * cfgexpand.c (pass_expand): Convert from a global struct to a
2226         subclass of rtl_opt_pass along with...
2227         (pass_data_expand): ...new pass_data instance and...
2228         (make_pass_expand): ...new function.
2229         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2230         of rtl_opt_pass along with...
2231         (pass_data_free_cfg): ...new pass_data instance and...
2232         (make_pass_free_cfg): ...new function.
2233         (pass_into_cfg_layout_mode): Convert from a global struct to a
2234         subclass of rtl_opt_pass along with...
2235         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2236         (make_pass_into_cfg_layout_mode): ...new function.
2237         (pass_outof_cfg_layout_mode): Convert from a global struct to a
2238         subclass of rtl_opt_pass along with...
2239         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2240         (make_pass_outof_cfg_layout_mode): ...new function.
2241         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2242         struct to a subclass of gimple_opt_pass along with...
2243         (pass_data_build_cgraph_edges): ...new pass_data instance and...
2244         (make_pass_build_cgraph_edges): ...new function.
2245         (pass_rebuild_cgraph_edges): Convert from a global struct to a
2246         subclass of gimple_opt_pass along with...
2247         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2248         (make_pass_rebuild_cgraph_edges): ...new function.
2249         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2250         subclass of gimple_opt_pass along with...
2251         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2252         and...
2253         (make_pass_remove_cgraph_callee_edges): ...new function.
2254         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2255         struct to a subclass of rtl_opt_pass along with...
2256         (pass_data_stack_adjustments): ...new pass_data instance and...
2257         (make_pass_stack_adjustments): ...new function.
2258         * combine.c (pass_combine): Convert from a global struct to a subclass
2259         of rtl_opt_pass along with...
2260         (pass_data_combine): ...new pass_data instance and...
2261         (make_pass_combine): ...new function.
2262         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2263         global struct to a subclass of rtl_opt_pass along with...
2264         (pass_data_compare_elim_after_reload): ...new pass_data instance
2265         and...
2266         (make_pass_compare_elim_after_reload): ...new function.
2267         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2268         of rtl_opt_pass along with...
2269         (pass_data_rtl_cprop): ...new pass_data instance and...
2270         (make_pass_rtl_cprop): ...new function.
2271         * cse.c (pass_cse): Convert from a global struct to a subclass of
2272         rtl_opt_pass along with...
2273         (pass_data_cse): ...new pass_data instance and...
2274         (make_pass_cse): ...new function.
2275         (pass_cse2): Convert from a global struct to a subclass of
2276         rtl_opt_pass along with...
2277         (pass_data_cse2): ...new pass_data instance and...
2278         (make_pass_cse2): ...new function.
2279         (pass_cse_after_global_opts): Convert from a global struct to a
2280         subclass of rtl_opt_pass along with...
2281         (pass_data_cse_after_global_opts): ...new pass_data instance and...
2282         (make_pass_cse_after_global_opts): ...new function.
2283         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2284         of rtl_opt_pass along with...
2285         (pass_data_ud_rtl_dce): ...new pass_data instance and...
2286         (make_pass_ud_rtl_dce): ...new function.
2287         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2288         rtl_opt_pass along with...
2289         (pass_data_fast_rtl_dce): ...new pass_data instance and...
2290         (make_pass_fast_rtl_dce): ...new function.
2291         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2292         a subclass of rtl_opt_pass along with...
2293         (pass_data_df_initialize_opt): ...new pass_data instance and...
2294         (make_pass_df_initialize_opt): ...new function.
2295         (pass_df_initialize_no_opt): Convert from a global struct to a
2296         subclass of rtl_opt_pass along with...
2297         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2298         (make_pass_df_initialize_no_opt): ...new function.
2299         (pass_df_finish): Convert from a global struct to a subclass of
2300         rtl_opt_pass along with...
2301         (pass_data_df_finish): ...new pass_data instance and...
2302         (make_pass_df_finish): ...new function.
2303         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2304         rtl_opt_pass along with...
2305         (pass_data_rtl_dse1): ...new pass_data instance and...
2306         (make_pass_rtl_dse1): ...new function.
2307         (pass_rtl_dse2): Convert from a global struct to a subclass of
2308         rtl_opt_pass along with...
2309         (pass_data_rtl_dse2): ...new pass_data instance and...
2310         (make_pass_rtl_dse2): ...new function.
2311         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2312         subclass of rtl_opt_pass along with...
2313         (pass_data_dwarf2_frame): ...new pass_data instance and...
2314         (make_pass_dwarf2_frame): ...new function.
2315         * except.c (pass_set_nothrow_function_flags): Convert from a global
2316         struct to a subclass of rtl_opt_pass along with...
2317         (pass_data_set_nothrow_function_flags): ...new pass_data instance
2318         and...
2319         (make_pass_set_nothrow_function_flags): ...new function.
2320         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2321         subclass of rtl_opt_pass along with...
2322         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2323         and...
2324         (make_pass_convert_to_eh_region_ranges): ...new function.
2325         * final.c (pass_compute_alignments): Convert from a global struct to a
2326         subclass of rtl_opt_pass along with...
2327         (pass_data_compute_alignments): ...new pass_data instance and...
2328         (make_pass_compute_alignments): ...new function.
2329         (pass_final): Convert from a global struct to a subclass of
2330         rtl_opt_pass along with...
2331         (pass_data_final): ...new pass_data instance and...
2332         (make_pass_final): ...new function.
2333         (pass_shorten_branches): Convert from a global struct to a subclass of
2334         rtl_opt_pass along with...
2335         (pass_data_shorten_branches): ...new pass_data instance and...
2336         (make_pass_shorten_branches): ...new function.
2337         (pass_clean_state): Convert from a global struct to a subclass of
2338         rtl_opt_pass along with...
2339         (pass_data_clean_state): ...new pass_data instance and...
2340         (make_pass_clean_state): ...new function.
2341         * function.c (pass_instantiate_virtual_regs): Convert from a global
2342         struct to a subclass of rtl_opt_pass along with...
2343         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2344         (make_pass_instantiate_virtual_regs): ...new function.
2345         (pass_leaf_regs): Convert from a global struct to a subclass of
2346         rtl_opt_pass along with...
2347         (pass_data_leaf_regs): ...new pass_data instance and...
2348         (make_pass_leaf_regs): ...new function.
2349         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2350         subclass of rtl_opt_pass along with...
2351         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2352         and...
2353         (make_pass_thread_prologue_and_epilogue): ...new function.
2354         (pass_match_asm_constraints): Convert from a global struct to a
2355         subclass of rtl_opt_pass along with...
2356         (pass_data_match_asm_constraints): ...new pass_data instance and...
2357         (make_pass_match_asm_constraints): ...new function.
2358         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2359         subclass of rtl_opt_pass along with...
2360         (pass_data_rtl_fwprop): ...new pass_data instance and...
2361         (make_pass_rtl_fwprop): ...new function.
2362         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2363         rtl_opt_pass along with...
2364         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2365         (make_pass_rtl_fwprop_addr): ...new function.
2366         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2367         rtl_opt_pass along with...
2368         (pass_data_rtl_pre): ...new pass_data instance and...
2369         (make_pass_rtl_pre): ...new function.
2370         (pass_rtl_hoist): Convert from a global struct to a subclass of
2371         rtl_opt_pass along with...
2372         (pass_data_rtl_hoist): ...new pass_data instance and...
2373         (make_pass_rtl_hoist): ...new function.
2374         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2375         subclass of gimple_opt_pass along with...
2376         (pass_data_lower_cf): ...new pass_data instance and...
2377         (make_pass_lower_cf): ...new function.
2378         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2379         from a global struct to a subclass of gimple_opt_pass along with...
2380         (pass_data_strength_reduction): ...new pass_data instance and...
2381         (make_pass_strength_reduction): ...new function.
2382         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2383         of rtl_opt_pass along with...
2384         (pass_data_rtl_ifcvt): ...new pass_data instance and...
2385         (make_pass_rtl_ifcvt): ...new function.
2386         (pass_if_after_combine): Convert from a global struct to a subclass of
2387         rtl_opt_pass along with...
2388         (pass_data_if_after_combine): ...new pass_data instance and...
2389         (make_pass_if_after_combine): ...new function.
2390         (pass_if_after_reload): Convert from a global struct to a subclass of
2391         rtl_opt_pass along with...
2392         (pass_data_if_after_reload): ...new pass_data instance and...
2393         (make_pass_if_after_reload): ...new function.
2394         * init-regs.c (pass_initialize_regs): Convert from a global struct to
2395         a subclass of rtl_opt_pass along with...
2396         (pass_data_initialize_regs): ...new pass_data instance and...
2397         (make_pass_initialize_regs): ...new function.
2398         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2399         of ipa_opt_pass_d along with...
2400         (pass_data_ipa_cp): ...new pass_data instance and...
2401         (make_pass_ipa_cp): ...new function.
2402         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2403         global struct to a subclass of gimple_opt_pass along with...
2404         (pass_data_inline_parameters): ...new pass_data instance and...
2405         (make_pass_inline_parameters): ...new function.
2406         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2407         subclass of gimple_opt_pass along with...
2408         (pass_data_early_inline): ...new pass_data instance and...
2409         (make_pass_early_inline): ...new function.
2410         (pass_ipa_inline): Convert from a global struct to a subclass of
2411         ipa_opt_pass_d along with...
2412         (pass_data_ipa_inline): ...new pass_data instance and...
2413         (make_pass_ipa_inline): ...new function.
2414         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2415         struct to a subclass of gimple_opt_pass along with...
2416         (pass_data_local_pure_const): ...new pass_data instance and...
2417         (make_pass_local_pure_const): ...new function.
2418         (pass_ipa_pure_const): Convert from a global struct to a subclass of
2419         ipa_opt_pass_d along with...
2420         (pass_data_ipa_pure_const): ...new pass_data instance and...
2421         (make_pass_ipa_pure_const): ...new function.
2422         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2423         to a subclass of ipa_opt_pass_d along with...
2424         (pass_data_ipa_reference): ...new pass_data instance and...
2425         (make_pass_ipa_reference): ...new function.
2426         * ipa-split.c (pass_split_functions): Convert from a global struct to
2427         a subclass of gimple_opt_pass along with...
2428         (pass_data_split_functions): ...new pass_data instance and...
2429         (make_pass_split_functions): ...new function.
2430         (pass_feedback_split_functions): Convert from a global struct to a
2431         subclass of gimple_opt_pass along with...
2432         (pass_data_feedback_split_functions): ...new pass_data instance and...
2433         (make_pass_feedback_split_functions): ...new function.
2434         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2435         global struct to a subclass of simple_ipa_opt_pass along with...
2436         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2437         instance and...
2438         (make_pass_ipa_function_and_variable_visibility): ...new function.
2439         (pass_ipa_free_inline_summary): Convert from a global struct to a
2440         subclass of simple_ipa_opt_pass along with...
2441         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2442         (make_pass_ipa_free_inline_summary): ...new function.
2443         (pass_ipa_whole_program_visibility): Convert from a global struct to a
2444         subclass of ipa_opt_pass_d along with...
2445         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2446         and...
2447         (make_pass_ipa_whole_program_visibility): ...new function.
2448         (pass_ipa_profile): Convert from a global struct to a subclass of
2449         ipa_opt_pass_d along with...
2450         (pass_data_ipa_profile): ...new pass_data instance and...
2451         (make_pass_ipa_profile): ...new function.
2452         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2453         ipa_opt_pass_d along with...
2454         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2455         (make_pass_ipa_cdtor_merge): ...new function.
2456         * ira.c (pass_ira): Convert from a global struct to a subclass of
2457         rtl_opt_pass along with...
2458         (pass_data_ira): ...new pass_data instance and...
2459         (make_pass_ira): ...new function.
2460         (pass_reload): Convert from a global struct to a subclass of
2461         rtl_opt_pass along with...
2462         (pass_data_reload): ...new pass_data instance and...
2463         (make_pass_reload): ...new function.
2464         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2465         subclass of rtl_opt_pass along with...
2466         (pass_data_cleanup_barriers): ...new pass_data instance and...
2467         (make_pass_cleanup_barriers): ...new function.
2468         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2469         of rtl_opt_pass along with...
2470         (pass_data_loop2): ...new pass_data instance and...
2471         (make_pass_loop2): ...new function.
2472         (pass_rtl_loop_init): Convert from a global struct to a subclass of
2473         rtl_opt_pass along with...
2474         (pass_data_rtl_loop_init): ...new pass_data instance and...
2475         (make_pass_rtl_loop_init): ...new function.
2476         (pass_rtl_loop_done): Convert from a global struct to a subclass of
2477         rtl_opt_pass along with...
2478         (pass_data_rtl_loop_done): ...new pass_data instance and...
2479         (make_pass_rtl_loop_done): ...new function.
2480         (pass_rtl_move_loop_invariants): Convert from a global struct to a
2481         subclass of rtl_opt_pass along with...
2482         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2483         (make_pass_rtl_move_loop_invariants): ...new function.
2484         (pass_rtl_unswitch): Convert from a global struct to a subclass of
2485         rtl_opt_pass along with...
2486         (pass_data_rtl_unswitch): ...new pass_data instance and...
2487         (make_pass_rtl_unswitch): ...new function.
2488         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2489         subclass of rtl_opt_pass along with...
2490         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2491         and...
2492         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2493         (pass_rtl_doloop): Convert from a global struct to a subclass of
2494         rtl_opt_pass along with...
2495         (pass_data_rtl_doloop): ...new pass_data instance and...
2496         (make_pass_rtl_doloop): ...new function.
2497         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2498         a subclass of rtl_opt_pass along with...
2499         (pass_data_lower_subreg): ...new pass_data instance and...
2500         (make_pass_lower_subreg): ...new function.
2501         (pass_lower_subreg2): Convert from a global struct to a subclass of
2502         rtl_opt_pass along with...
2503         (pass_data_lower_subreg2): ...new pass_data instance and...
2504         (make_pass_lower_subreg2): ...new function.
2505         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2506         struct to a subclass of ipa_opt_pass_d along with...
2507         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2508         (make_pass_ipa_lto_gimple_out): ...new function.
2509         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2510         of ipa_opt_pass_d along with...
2511         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2512         (make_pass_ipa_lto_finish_out): ...new function.
2513         * mode-switching.c (pass_mode_switching): Convert from a global struct
2514         to a subclass of rtl_opt_pass along with...
2515         (pass_data_mode_switching): ...new pass_data instance and...
2516         (make_pass_mode_switching): ...new function.
2517         * modulo-sched.c (pass_sms): Convert from a global struct to a
2518         subclass of rtl_opt_pass along with...
2519         (pass_data_sms): ...new pass_data instance and...
2520         (make_pass_sms): ...new function.
2521         * omp-low.c (pass_expand_omp): Convert from a global struct to a
2522         subclass of gimple_opt_pass along with...
2523         (pass_data_expand_omp): ...new pass_data instance and...
2524         (make_pass_expand_omp): ...new function.
2525         (pass_lower_omp): Convert from a global struct to a subclass of
2526         gimple_opt_pass along with...
2527         (pass_data_lower_omp): ...new pass_data instance and...
2528         (make_pass_lower_omp): ...new function.
2529         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2530         of gimple_opt_pass along with...
2531         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2532         (make_pass_diagnose_omp_blocks): ...new function.
2533         * passes.c (pass_early_local_passes): Convert from a global struct to
2534         a subclass of simple_ipa_opt_pass along with...
2535         (pass_data_early_local_passes): ...new pass_data instance and...
2536         (make_pass_early_local_passes): ...new function.
2537         (pass_all_early_optimizations): Convert from a global struct to a
2538         subclass of gimple_opt_pass along with...
2539         (pass_data_all_early_optimizations): ...new pass_data instance and...
2540         (make_pass_all_early_optimizations): ...new function.
2541         (pass_all_optimizations): Convert from a global struct to a subclass
2542         of gimple_opt_pass along with...
2543         (pass_data_all_optimizations): ...new pass_data instance and...
2544         (make_pass_all_optimizations): ...new function.
2545         (pass_all_optimizations_g): Convert from a global struct to a subclass
2546         of gimple_opt_pass along with...
2547         (pass_data_all_optimizations_g): ...new pass_data instance and...
2548         (make_pass_all_optimizations_g): ...new function.
2549         (pass_rest_of_compilation): Convert from a global struct to a subclass
2550         of rtl_opt_pass along with...
2551         (pass_data_rest_of_compilation): ...new pass_data instance and...
2552         (make_pass_rest_of_compilation): ...new function.
2553         (pass_postreload): Convert from a global struct to a subclass of
2554         rtl_opt_pass along with...
2555         (pass_data_postreload): ...new pass_data instance and...
2556         (make_pass_postreload): ...new function.
2557         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2558         subclass of rtl_opt_pass along with...
2559         (pass_data_gcse2): ...new pass_data instance and...
2560         (make_pass_gcse2): ...new function.
2561         * postreload.c (pass_postreload_cse): Convert from a global struct to
2562         a subclass of rtl_opt_pass along with...
2563         (pass_data_postreload_cse): ...new pass_data instance and...
2564         (make_pass_postreload_cse): ...new function.
2565         * predict.c (pass_profile): Convert from a global struct to a subclass
2566         of gimple_opt_pass along with...
2567         (pass_data_profile): ...new pass_data instance and...
2568         (make_pass_profile): ...new function.
2569         (pass_strip_predict_hints): Convert from a global struct to a subclass
2570         of gimple_opt_pass along with...
2571         (pass_data_strip_predict_hints): ...new pass_data instance and...
2572         (make_pass_strip_predict_hints): ...new function.
2573         * recog.c (pass_peephole2): Convert from a global struct to a subclass
2574         of rtl_opt_pass along with...
2575         (pass_data_peephole2): ...new pass_data instance and...
2576         (make_pass_peephole2): ...new function.
2577         (pass_split_all_insns): Convert from a global struct to a subclass of
2578         rtl_opt_pass along with...
2579         (pass_data_split_all_insns): ...new pass_data instance and...
2580         (make_pass_split_all_insns): ...new function.
2581         (pass_split_after_reload): Convert from a global struct to a subclass
2582         of rtl_opt_pass along with...
2583         (pass_data_split_after_reload): ...new pass_data instance and...
2584         (make_pass_split_after_reload): ...new function.
2585         (pass_split_before_regstack): Convert from a global struct to a
2586         subclass of rtl_opt_pass along with...
2587         (pass_data_split_before_regstack): ...new pass_data instance and...
2588         (make_pass_split_before_regstack): ...new function.
2589         (pass_split_before_sched2): Convert from a global struct to a subclass
2590         of rtl_opt_pass along with...
2591         (pass_data_split_before_sched2): ...new pass_data instance and...
2592         (make_pass_split_before_sched2): ...new function.
2593         (pass_split_for_shorten_branches): Convert from a global struct to a
2594         subclass of rtl_opt_pass along with...
2595         (pass_data_split_for_shorten_branches): ...new pass_data instance
2596         and...
2597         (make_pass_split_for_shorten_branches): ...new function.
2598         * ree.c (pass_ree): Convert from a global struct to a subclass of
2599         rtl_opt_pass along with...
2600         (pass_data_ree): ...new pass_data instance and...
2601         (make_pass_ree): ...new function.
2602         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2603         subclass of rtl_opt_pass along with...
2604         (pass_data_stack_regs): ...new pass_data instance and...
2605         (make_pass_stack_regs): ...new function.
2606         (pass_stack_regs_run): Convert from a global struct to a subclass of
2607         rtl_opt_pass along with...
2608         (pass_data_stack_regs_run): ...new pass_data instance and...
2609         (make_pass_stack_regs_run): ...new function.
2610         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2611         subclass of rtl_opt_pass along with...
2612         (pass_data_cprop_hardreg): ...new pass_data instance and...
2613         (make_pass_cprop_hardreg): ...new function.
2614         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2615         subclass of rtl_opt_pass along with...
2616         (pass_data_reginfo_init): ...new pass_data instance and...
2617         (make_pass_reginfo_init): ...new function.
2618         * regmove.c (pass_regmove): Convert from a global struct to a subclass
2619         of rtl_opt_pass along with...
2620         (pass_data_regmove): ...new pass_data instance and...
2621         (make_pass_regmove): ...new function.
2622         * regrename.c (pass_regrename): Convert from a global struct to a
2623         subclass of rtl_opt_pass along with...
2624         (pass_data_regrename): ...new pass_data instance and...
2625         (make_pass_regrename): ...new function.
2626         * reorg.c (pass_delay_slots): Convert from a global struct to a
2627         subclass of rtl_opt_pass along with...
2628         (pass_data_delay_slots): ...new pass_data instance and...
2629         (make_pass_delay_slots): ...new function.
2630         (pass_machine_reorg): Convert from a global struct to a subclass of
2631         rtl_opt_pass along with...
2632         (pass_data_machine_reorg): ...new pass_data instance and...
2633         (make_pass_machine_reorg): ...new function.
2634         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2635         of rtl_opt_pass along with...
2636         (pass_data_sched): ...new pass_data instance and...
2637         (make_pass_sched): ...new function.
2638         (pass_sched2): Convert from a global struct to a subclass of
2639         rtl_opt_pass along with...
2640         (pass_data_sched2): ...new pass_data instance and...
2641         (make_pass_sched2): ...new function.
2642         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2643         to a subclass of rtl_opt_pass along with...
2644         (pass_data_stack_ptr_mod): ...new pass_data instance and...
2645         (make_pass_stack_ptr_mod): ...new function.
2646         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2647         to a subclass of rtl_opt_pass along with...
2648         (pass_data_rtl_store_motion): ...new pass_data instance and...
2649         (make_pass_rtl_store_motion): ...new function.
2650         * tracer.c (pass_tracer): Convert from a global struct to a subclass
2651         of gimple_opt_pass along with...
2652         (pass_data_tracer): ...new pass_data instance and...
2653         (make_pass_tracer): ...new function.
2654         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2655         to a subclass of gimple_opt_pass along with...
2656         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2657         (make_pass_diagnose_tm_blocks): ...new function.
2658         (pass_lower_tm): Convert from a global struct to a subclass of
2659         gimple_opt_pass along with...
2660         (pass_data_lower_tm): ...new pass_data instance and...
2661         (make_pass_lower_tm): ...new function.
2662         (pass_tm_init): Convert from a global struct to a subclass of
2663         gimple_opt_pass along with...
2664         (pass_data_tm_init): ...new pass_data instance and...
2665         (make_pass_tm_init): ...new function.
2666         (pass_tm_mark): Convert from a global struct to a subclass of
2667         gimple_opt_pass along with...
2668         (pass_data_tm_mark): ...new pass_data instance and...
2669         (make_pass_tm_mark): ...new function.
2670         (pass_tm_edges): Convert from a global struct to a subclass of
2671         gimple_opt_pass along with...
2672         (pass_data_tm_edges): ...new pass_data instance and...
2673         (make_pass_tm_edges): ...new function.
2674         (pass_tm_memopt): Convert from a global struct to a subclass of
2675         gimple_opt_pass along with...
2676         (pass_data_tm_memopt): ...new pass_data instance and...
2677         (make_pass_tm_memopt): ...new function.
2678         (pass_ipa_tm): Convert from a global struct to a subclass of
2679         simple_ipa_opt_pass along with...
2680         (pass_data_ipa_tm): ...new pass_data instance and...
2681         (make_pass_ipa_tm): ...new function.
2682         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2683         subclass of gimple_opt_pass along with...
2684         (pass_data_call_cdce): ...new pass_data instance and...
2685         (make_pass_call_cdce): ...new function.
2686         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2687         subclass of gimple_opt_pass along with...
2688         (pass_data_build_cfg): ...new pass_data instance and...
2689         (make_pass_build_cfg): ...new function.
2690         (pass_split_crit_edges): Convert from a global struct to a subclass of
2691         gimple_opt_pass along with...
2692         (pass_data_split_crit_edges): ...new pass_data instance and...
2693         (make_pass_split_crit_edges): ...new function.
2694         (pass_warn_function_return): Convert from a global struct to a
2695         subclass of gimple_opt_pass along with...
2696         (pass_data_warn_function_return): ...new pass_data instance and...
2697         (make_pass_warn_function_return): ...new function.
2698         (pass_warn_function_noreturn): Convert from a global struct to a
2699         subclass of gimple_opt_pass along with...
2700         (pass_data_warn_function_noreturn): ...new pass_data instance and...
2701         (make_pass_warn_function_noreturn): ...new function.
2702         (pass_warn_unused_result): Convert from a global struct to a subclass
2703         of gimple_opt_pass along with...
2704         (pass_data_warn_unused_result): ...new pass_data instance and...
2705         (make_pass_warn_unused_result): ...new function.
2706         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2707         a subclass of gimple_opt_pass along with...
2708         (pass_data_merge_phi): ...new pass_data instance and...
2709         (make_pass_merge_phi): ...new function.
2710         * tree-complex.c (pass_lower_complex): Convert from a global struct to
2711         a subclass of gimple_opt_pass along with...
2712         (pass_data_lower_complex): ...new pass_data instance and...
2713         (make_pass_lower_complex): ...new function.
2714         (pass_lower_complex_O0): Convert from a global struct to a subclass of
2715         gimple_opt_pass along with...
2716         (pass_data_lower_complex_O0): ...new pass_data instance and...
2717         (make_pass_lower_complex_O0): ...new function.
2718         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2719         subclass of gimple_opt_pass along with...
2720         (pass_data_lower_eh): ...new pass_data instance and...
2721         (make_pass_lower_eh): ...new function.
2722         (pass_refactor_eh): Convert from a global struct to a subclass of
2723         gimple_opt_pass along with...
2724         (pass_data_refactor_eh): ...new pass_data instance and...
2725         (make_pass_refactor_eh): ...new function.
2726         (pass_lower_resx): Convert from a global struct to a subclass of
2727         gimple_opt_pass along with...
2728         (pass_data_lower_resx): ...new pass_data instance and...
2729         (make_pass_lower_resx): ...new function.
2730         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2731         of gimple_opt_pass along with...
2732         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2733         (make_pass_lower_eh_dispatch): ...new function.
2734         (pass_cleanup_eh): Convert from a global struct to a subclass of
2735         gimple_opt_pass along with...
2736         (pass_data_cleanup_eh): ...new pass_data instance and...
2737         (make_pass_cleanup_eh): ...new function.
2738         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2739         to a subclass of simple_ipa_opt_pass along with...
2740         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2741         (make_pass_ipa_lower_emutls): ...new function.
2742         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2743         a subclass of gimple_opt_pass along with...
2744         (pass_data_if_conversion): ...new pass_data instance and...
2745         (make_pass_if_conversion): ...new function.
2746         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2747         subclass of gimple_opt_pass along with...
2748         (pass_data_build_ssa): ...new pass_data instance and...
2749         (make_pass_build_ssa): ...new function.
2750         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2751         global struct to a subclass of gimple_opt_pass along with...
2752         (pass_data_loop_distribution): ...new pass_data instance and...
2753         (make_pass_loop_distribution): ...new function.
2754         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2755         subclass of gimple_opt_pass along with...
2756         (pass_data_mudflap_1): ...new pass_data instance and...
2757         (make_pass_mudflap_1): ...new function.
2758         (pass_mudflap_2): Convert from a global struct to a subclass of
2759         gimple_opt_pass along with...
2760         (pass_data_mudflap_2): ...new pass_data instance and...
2761         (make_pass_mudflap_2): ...new function.
2762         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2763         subclass of gimple_opt_pass along with...
2764         (pass_data_mudflap_1): ...new pass_data instance and...
2765         (make_pass_mudflap_1): ...new function.
2766         (pass_mudflap_2): Convert from a global struct to a subclass of
2767         gimple_opt_pass along with...
2768         (pass_data_mudflap_2): ...new pass_data instance and...
2769         (make_pass_mudflap_2): ...new function.
2770         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2771         gimple_opt_pass along with...
2772         (pass_data_nrv): ...new pass_data instance and...
2773         (make_pass_nrv): ...new function.
2774         (pass_return_slot): Convert from a global struct to a subclass of
2775         gimple_opt_pass along with...
2776         (pass_data_return_slot): ...new pass_data instance and...
2777         (make_pass_return_slot): ...new function.
2778         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2779         to a subclass of gimple_opt_pass along with...
2780         (pass_data_object_sizes): ...new pass_data instance and...
2781         (make_pass_object_sizes): ...new function.
2782         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2783         global struct to a subclass of gimple_opt_pass along with...
2784         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2785         and...
2786         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2787         (pass_fixup_cfg): Convert from a global struct to a subclass of
2788         gimple_opt_pass along with...
2789         (pass_data_fixup_cfg): ...new pass_data instance and...
2790         (make_pass_fixup_cfg): ...new function.
2791         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2792         (make_pass_mudflap_1): ...new function.
2793         (pass_mudflap_2): Replace declaration with that of...
2794         (make_pass_mudflap_2): ...new function.
2795         (pass_asan): Replace declaration with that of...
2796         (make_pass_asan): ...new function.
2797         (pass_asan_O0): Replace declaration with that of...
2798         (make_pass_asan_O0): ...new function.
2799         (pass_tsan): Replace declaration with that of...
2800         (make_pass_tsan): ...new function.
2801         (pass_tsan_O0): Replace declaration with that of...
2802         (make_pass_tsan_O0): ...new function.
2803         (pass_lower_cf): Replace declaration with that of...
2804         (make_pass_lower_cf): ...new function.
2805         (pass_refactor_eh): Replace declaration with that of...
2806         (make_pass_refactor_eh): ...new function.
2807         (pass_lower_eh): Replace declaration with that of...
2808         (make_pass_lower_eh): ...new function.
2809         (pass_lower_eh_dispatch): Replace declaration with that of...
2810         (make_pass_lower_eh_dispatch): ...new function.
2811         (pass_lower_resx): Replace declaration with that of...
2812         (make_pass_lower_resx): ...new function.
2813         (pass_build_cfg): Replace declaration with that of...
2814         (make_pass_build_cfg): ...new function.
2815         (pass_early_tree_profile): Replace declaration with that of...
2816         (make_pass_early_tree_profile): ...new function.
2817         (pass_cleanup_eh): Replace declaration with that of...
2818         (make_pass_cleanup_eh): ...new function.
2819         (pass_sra): Replace declaration with that of...
2820         (make_pass_sra): ...new function.
2821         (pass_sra_early): Replace declaration with that of...
2822         (make_pass_sra_early): ...new function.
2823         (pass_early_ipa_sra): Replace declaration with that of...
2824         (make_pass_early_ipa_sra): ...new function.
2825         (pass_tail_recursion): Replace declaration with that of...
2826         (make_pass_tail_recursion): ...new function.
2827         (pass_tail_calls): Replace declaration with that of...
2828         (make_pass_tail_calls): ...new function.
2829         (pass_tree_loop): Replace declaration with that of...
2830         (make_pass_tree_loop): ...new function.
2831         (pass_tree_loop_init): Replace declaration with that of...
2832         (make_pass_tree_loop_init): ...new function.
2833         (pass_lim): Replace declaration with that of...
2834         (make_pass_lim): ...new function.
2835         (pass_tree_unswitch): Replace declaration with that of...
2836         (make_pass_tree_unswitch): ...new function.
2837         (pass_predcom): Replace declaration with that of...
2838         (make_pass_predcom): ...new function.
2839         (pass_iv_canon): Replace declaration with that of...
2840         (make_pass_iv_canon): ...new function.
2841         (pass_scev_cprop): Replace declaration with that of...
2842         (make_pass_scev_cprop): ...new function.
2843         (pass_empty_loop): Replace declaration with that of...
2844         (make_pass_empty_loop): ...new function.
2845         (pass_record_bounds): Replace declaration with that of...
2846         (make_pass_record_bounds): ...new function.
2847         (pass_graphite): Replace declaration with that of...
2848         (make_pass_graphite): ...new function.
2849         (pass_graphite_transforms): Replace declaration with that of...
2850         (make_pass_graphite_transforms): ...new function.
2851         (pass_if_conversion): Replace declaration with that of...
2852         (make_pass_if_conversion): ...new function.
2853         (pass_loop_distribution): Replace declaration with that of...
2854         (make_pass_loop_distribution): ...new function.
2855         (pass_vectorize): Replace declaration with that of...
2856         (make_pass_vectorize): ...new function.
2857         (pass_slp_vectorize): Replace declaration with that of...
2858         (make_pass_slp_vectorize): ...new function.
2859         (pass_complete_unroll): Replace declaration with that of...
2860         (make_pass_complete_unroll): ...new function.
2861         (pass_complete_unrolli): Replace declaration with that of...
2862         (make_pass_complete_unrolli): ...new function.
2863         (pass_parallelize_loops): Replace declaration with that of...
2864         (make_pass_parallelize_loops): ...new function.
2865         (pass_loop_prefetch): Replace declaration with that of...
2866         (make_pass_loop_prefetch): ...new function.
2867         (pass_iv_optimize): Replace declaration with that of...
2868         (make_pass_iv_optimize): ...new function.
2869         (pass_tree_loop_done): Replace declaration with that of...
2870         (make_pass_tree_loop_done): ...new function.
2871         (pass_ch): Replace declaration with that of...
2872         (make_pass_ch): ...new function.
2873         (pass_ccp): Replace declaration with that of...
2874         (make_pass_ccp): ...new function.
2875         (pass_phi_only_cprop): Replace declaration with that of...
2876         (make_pass_phi_only_cprop): ...new function.
2877         (pass_build_ssa): Replace declaration with that of...
2878         (make_pass_build_ssa): ...new function.
2879         (pass_build_alias): Replace declaration with that of...
2880         (make_pass_build_alias): ...new function.
2881         (pass_build_ealias): Replace declaration with that of...
2882         (make_pass_build_ealias): ...new function.
2883         (pass_dominator): Replace declaration with that of...
2884         (make_pass_dominator): ...new function.
2885         (pass_dce): Replace declaration with that of...
2886         (make_pass_dce): ...new function.
2887         (pass_dce_loop): Replace declaration with that of...
2888         (make_pass_dce_loop): ...new function.
2889         (pass_cd_dce): Replace declaration with that of...
2890         (make_pass_cd_dce): ...new function.
2891         (pass_call_cdce): Replace declaration with that of...
2892         (make_pass_call_cdce): ...new function.
2893         (pass_merge_phi): Replace declaration with that of...
2894         (make_pass_merge_phi): ...new function.
2895         (pass_split_crit_edges): Replace declaration with that of...
2896         (make_pass_split_crit_edges): ...new function.
2897         (pass_pre): Replace declaration with that of...
2898         (make_pass_pre): ...new function.
2899         (pass_profile): Replace declaration with that of...
2900         (make_pass_profile): ...new function.
2901         (pass_strip_predict_hints): Replace declaration with that of...
2902         (make_pass_strip_predict_hints): ...new function.
2903         (pass_lower_complex_O0): Replace declaration with that of...
2904         (make_pass_lower_complex_O0): ...new function.
2905         (pass_lower_complex): Replace declaration with that of...
2906         (make_pass_lower_complex): ...new function.
2907         (pass_lower_vector): Replace declaration with that of...
2908         (make_pass_lower_vector): ...new function.
2909         (pass_lower_vector_ssa): Replace declaration with that of...
2910         (make_pass_lower_vector_ssa): ...new function.
2911         (pass_lower_omp): Replace declaration with that of...
2912         (make_pass_lower_omp): ...new function.
2913         (pass_diagnose_omp_blocks): Replace declaration with that of...
2914         (make_pass_diagnose_omp_blocks): ...new function.
2915         (pass_expand_omp): Replace declaration with that of...
2916         (make_pass_expand_omp): ...new function.
2917         (pass_expand_omp_ssa): Replace declaration with that of...
2918         (make_pass_expand_omp_ssa): ...new function.
2919         (pass_object_sizes): Replace declaration with that of...
2920         (make_pass_object_sizes): ...new function.
2921         (pass_strlen): Replace declaration with that of...
2922         (make_pass_strlen): ...new function.
2923         (pass_fold_builtins): Replace declaration with that of...
2924         (make_pass_fold_builtins): ...new function.
2925         (pass_stdarg): Replace declaration with that of...
2926         (make_pass_stdarg): ...new function.
2927         (pass_early_warn_uninitialized): Replace declaration with that of...
2928         (make_pass_early_warn_uninitialized): ...new function.
2929         (pass_late_warn_uninitialized): Replace declaration with that of...
2930         (make_pass_late_warn_uninitialized): ...new function.
2931         (pass_cse_reciprocals): Replace declaration with that of...
2932         (make_pass_cse_reciprocals): ...new function.
2933         (pass_cse_sincos): Replace declaration with that of...
2934         (make_pass_cse_sincos): ...new function.
2935         (pass_optimize_bswap): Replace declaration with that of...
2936         (make_pass_optimize_bswap): ...new function.
2937         (pass_optimize_widening_mul): Replace declaration with that of...
2938         (make_pass_optimize_widening_mul): ...new function.
2939         (pass_warn_function_return): Replace declaration with that of...
2940         (make_pass_warn_function_return): ...new function.
2941         (pass_warn_function_noreturn): Replace declaration with that of...
2942         (make_pass_warn_function_noreturn): ...new function.
2943         (pass_cselim): Replace declaration with that of...
2944         (make_pass_cselim): ...new function.
2945         (pass_phiopt): Replace declaration with that of...
2946         (make_pass_phiopt): ...new function.
2947         (pass_forwprop): Replace declaration with that of...
2948         (make_pass_forwprop): ...new function.
2949         (pass_phiprop): Replace declaration with that of...
2950         (make_pass_phiprop): ...new function.
2951         (pass_tree_ifcombine): Replace declaration with that of...
2952         (make_pass_tree_ifcombine): ...new function.
2953         (pass_dse): Replace declaration with that of...
2954         (make_pass_dse): ...new function.
2955         (pass_nrv): Replace declaration with that of...
2956         (make_pass_nrv): ...new function.
2957         (pass_rename_ssa_copies): Replace declaration with that of...
2958         (make_pass_rename_ssa_copies): ...new function.
2959         (pass_sink_code): Replace declaration with that of...
2960         (make_pass_sink_code): ...new function.
2961         (pass_fre): Replace declaration with that of...
2962         (make_pass_fre): ...new function.
2963         (pass_check_data_deps): Replace declaration with that of...
2964         (make_pass_check_data_deps): ...new function.
2965         (pass_copy_prop): Replace declaration with that of...
2966         (make_pass_copy_prop): ...new function.
2967         (pass_vrp): Replace declaration with that of...
2968         (make_pass_vrp): ...new function.
2969         (pass_uncprop): Replace declaration with that of...
2970         (make_pass_uncprop): ...new function.
2971         (pass_return_slot): Replace declaration with that of...
2972         (make_pass_return_slot): ...new function.
2973         (pass_reassoc): Replace declaration with that of...
2974         (make_pass_reassoc): ...new function.
2975         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2976         (make_pass_rebuild_cgraph_edges): ...new function.
2977         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2978         (make_pass_remove_cgraph_callee_edges): ...new function.
2979         (pass_build_cgraph_edges): Replace declaration with that of...
2980         (make_pass_build_cgraph_edges): ...new function.
2981         (pass_local_pure_const): Replace declaration with that of...
2982         (make_pass_local_pure_const): ...new function.
2983         (pass_tracer): Replace declaration with that of...
2984         (make_pass_tracer): ...new function.
2985         (pass_warn_unused_result): Replace declaration with that of...
2986         (make_pass_warn_unused_result): ...new function.
2987         (pass_diagnose_tm_blocks): Replace declaration with that of...
2988         (make_pass_diagnose_tm_blocks): ...new function.
2989         (pass_lower_tm): Replace declaration with that of...
2990         (make_pass_lower_tm): ...new function.
2991         (pass_tm_init): Replace declaration with that of...
2992         (make_pass_tm_init): ...new function.
2993         (pass_tm_mark): Replace declaration with that of...
2994         (make_pass_tm_mark): ...new function.
2995         (pass_tm_memopt): Replace declaration with that of...
2996         (make_pass_tm_memopt): ...new function.
2997         (pass_tm_edges): Replace declaration with that of...
2998         (make_pass_tm_edges): ...new function.
2999         (pass_split_functions): Replace declaration with that of...
3000         (make_pass_split_functions): ...new function.
3001         (pass_feedback_split_functions): Replace declaration with that of...
3002         (make_pass_feedback_split_functions): ...new function.
3003         (pass_strength_reduction): Replace declaration with that of...
3004         (make_pass_strength_reduction): ...new function.
3005         (pass_ipa_lower_emutls): Replace declaration with that of...
3006         (make_pass_ipa_lower_emutls): ...new function.
3007         (pass_ipa_function_and_variable_visibility): Replace declaration with
3008         that of...
3009         (make_pass_ipa_function_and_variable_visibility): ...new function.
3010         (pass_ipa_tree_profile): Replace declaration with that of...
3011         (make_pass_ipa_tree_profile): ...new function.
3012         (pass_early_local_passes): Replace declaration with that of...
3013         (make_pass_early_local_passes): ...new function.
3014         (pass_ipa_whole_program_visibility): Replace declaration with that
3015         of...
3016         (make_pass_ipa_whole_program_visibility): ...new function.
3017         (pass_ipa_lto_gimple_out): Replace declaration with that of...
3018         (make_pass_ipa_lto_gimple_out): ...new function.
3019         (pass_ipa_increase_alignment): Replace declaration with that of...
3020         (make_pass_ipa_increase_alignment): ...new function.
3021         (pass_ipa_inline): Replace declaration with that of...
3022         (make_pass_ipa_inline): ...new function.
3023         (pass_ipa_free_lang_data): Replace declaration with that of...
3024         (make_pass_ipa_free_lang_data): ...new function.
3025         (pass_ipa_free_inline_summary): Replace declaration with that of...
3026         (make_pass_ipa_free_inline_summary): ...new function.
3027         (pass_ipa_cp): Replace declaration with that of...
3028         (make_pass_ipa_cp): ...new function.
3029         (pass_ipa_reference): Replace declaration with that of...
3030         (make_pass_ipa_reference): ...new function.
3031         (pass_ipa_pure_const): Replace declaration with that of...
3032         (make_pass_ipa_pure_const): ...new function.
3033         (pass_ipa_pta): Replace declaration with that of...
3034         (make_pass_ipa_pta): ...new function.
3035         (pass_ipa_lto_finish_out): Replace declaration with that of...
3036         (make_pass_ipa_lto_finish_out): ...new function.
3037         (pass_ipa_tm): Replace declaration with that of...
3038         (make_pass_ipa_tm): ...new function.
3039         (pass_ipa_profile): Replace declaration with that of...
3040         (make_pass_ipa_profile): ...new function.
3041         (pass_ipa_cdtor_merge): Replace declaration with that of...
3042         (make_pass_ipa_cdtor_merge): ...new function.
3043         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3044         of...
3045         (make_pass_cleanup_cfg_post_optimizing): ...new function.
3046         (pass_init_datastructures): Replace declaration with that of...
3047         (make_pass_init_datastructures): ...new function.
3048         (pass_fixup_cfg): Replace declaration with that of...
3049         (make_pass_fixup_cfg): ...new function.
3050         (pass_expand): Replace declaration with that of...
3051         (make_pass_expand): ...new function.
3052         (pass_instantiate_virtual_regs): Replace declaration with that of...
3053         (make_pass_instantiate_virtual_regs): ...new function.
3054         (pass_rtl_fwprop): Replace declaration with that of...
3055         (make_pass_rtl_fwprop): ...new function.
3056         (pass_rtl_fwprop_addr): Replace declaration with that of...
3057         (make_pass_rtl_fwprop_addr): ...new function.
3058         (pass_jump): Replace declaration with that of...
3059         (make_pass_jump): ...new function.
3060         (pass_jump2): Replace declaration with that of...
3061         (make_pass_jump2): ...new function.
3062         (pass_lower_subreg): Replace declaration with that of...
3063         (make_pass_lower_subreg): ...new function.
3064         (pass_cse): Replace declaration with that of...
3065         (make_pass_cse): ...new function.
3066         (pass_fast_rtl_dce): Replace declaration with that of...
3067         (make_pass_fast_rtl_dce): ...new function.
3068         (pass_ud_rtl_dce): Replace declaration with that of...
3069         (make_pass_ud_rtl_dce): ...new function.
3070         (pass_rtl_dce): Replace declaration with that of...
3071         (make_pass_rtl_dce): ...new function.
3072         (pass_rtl_dse1): Replace declaration with that of...
3073         (make_pass_rtl_dse1): ...new function.
3074         (pass_rtl_dse2): Replace declaration with that of...
3075         (make_pass_rtl_dse2): ...new function.
3076         (pass_rtl_dse3): Replace declaration with that of...
3077         (make_pass_rtl_dse3): ...new function.
3078         (pass_rtl_cprop): Replace declaration with that of...
3079         (make_pass_rtl_cprop): ...new function.
3080         (pass_rtl_pre): Replace declaration with that of...
3081         (make_pass_rtl_pre): ...new function.
3082         (pass_rtl_hoist): Replace declaration with that of...
3083         (make_pass_rtl_hoist): ...new function.
3084         (pass_rtl_store_motion): Replace declaration with that of...
3085         (make_pass_rtl_store_motion): ...new function.
3086         (pass_cse_after_global_opts): Replace declaration with that of...
3087         (make_pass_cse_after_global_opts): ...new function.
3088         (pass_rtl_ifcvt): Replace declaration with that of...
3089         (make_pass_rtl_ifcvt): ...new function.
3090         (pass_into_cfg_layout_mode): Replace declaration with that of...
3091         (make_pass_into_cfg_layout_mode): ...new function.
3092         (pass_outof_cfg_layout_mode): Replace declaration with that of...
3093         (make_pass_outof_cfg_layout_mode): ...new function.
3094         (pass_loop2): Replace declaration with that of...
3095         (make_pass_loop2): ...new function.
3096         (pass_rtl_loop_init): Replace declaration with that of...
3097         (make_pass_rtl_loop_init): ...new function.
3098         (pass_rtl_move_loop_invariants): Replace declaration with that of...
3099         (make_pass_rtl_move_loop_invariants): ...new function.
3100         (pass_rtl_unswitch): Replace declaration with that of...
3101         (make_pass_rtl_unswitch): ...new function.
3102         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3103         (make_pass_rtl_unroll_and_peel_loops): ...new function.
3104         (pass_rtl_doloop): Replace declaration with that of...
3105         (make_pass_rtl_doloop): ...new function.
3106         (pass_rtl_loop_done): Replace declaration with that of...
3107         (make_pass_rtl_loop_done): ...new function.
3108         (pass_web): Replace declaration with that of...
3109         (make_pass_web): ...new function.
3110         (pass_cse2): Replace declaration with that of...
3111         (make_pass_cse2): ...new function.
3112         (pass_df_initialize_opt): Replace declaration with that of...
3113         (make_pass_df_initialize_opt): ...new function.
3114         (pass_df_initialize_no_opt): Replace declaration with that of...
3115         (make_pass_df_initialize_no_opt): ...new function.
3116         (pass_reginfo_init): Replace declaration with that of...
3117         (make_pass_reginfo_init): ...new function.
3118         (pass_inc_dec): Replace declaration with that of...
3119         (make_pass_inc_dec): ...new function.
3120         (pass_stack_ptr_mod): Replace declaration with that of...
3121         (make_pass_stack_ptr_mod): ...new function.
3122         (pass_initialize_regs): Replace declaration with that of...
3123         (make_pass_initialize_regs): ...new function.
3124         (pass_combine): Replace declaration with that of...
3125         (make_pass_combine): ...new function.
3126         (pass_if_after_combine): Replace declaration with that of...
3127         (make_pass_if_after_combine): ...new function.
3128         (pass_ree): Replace declaration with that of...
3129         (make_pass_ree): ...new function.
3130         (pass_partition_blocks): Replace declaration with that of...
3131         (make_pass_partition_blocks): ...new function.
3132         (pass_match_asm_constraints): Replace declaration with that of...
3133         (make_pass_match_asm_constraints): ...new function.
3134         (pass_regmove): Replace declaration with that of...
3135         (make_pass_regmove): ...new function.
3136         (pass_split_all_insns): Replace declaration with that of...
3137         (make_pass_split_all_insns): ...new function.
3138         (pass_fast_rtl_byte_dce): Replace declaration with that of...
3139         (make_pass_fast_rtl_byte_dce): ...new function.
3140         (pass_lower_subreg2): Replace declaration with that of...
3141         (make_pass_lower_subreg2): ...new function.
3142         (pass_mode_switching): Replace declaration with that of...
3143         (make_pass_mode_switching): ...new function.
3144         (pass_sms): Replace declaration with that of...
3145         (make_pass_sms): ...new function.
3146         (pass_sched): Replace declaration with that of...
3147         (make_pass_sched): ...new function.
3148         (pass_ira): Replace declaration with that of...
3149         (make_pass_ira): ...new function.
3150         (pass_reload): Replace declaration with that of...
3151         (make_pass_reload): ...new function.
3152         (pass_clean_state): Replace declaration with that of...
3153         (make_pass_clean_state): ...new function.
3154         (pass_branch_prob): Replace declaration with that of...
3155         (make_pass_branch_prob): ...new function.
3156         (pass_value_profile_transformations): Replace declaration with that
3157         of...
3158         (make_pass_value_profile_transformations): ...new function.
3159         (pass_postreload_cse): Replace declaration with that of...
3160         (make_pass_postreload_cse): ...new function.
3161         (pass_gcse2): Replace declaration with that of...
3162         (make_pass_gcse2): ...new function.
3163         (pass_split_after_reload): Replace declaration with that of...
3164         (make_pass_split_after_reload): ...new function.
3165         (pass_branch_target_load_optimize1): Replace declaration with that
3166         of...
3167         (make_pass_branch_target_load_optimize1): ...new function.
3168         (pass_thread_prologue_and_epilogue): Replace declaration with that
3169         of...
3170         (make_pass_thread_prologue_and_epilogue): ...new function.
3171         (pass_stack_adjustments): Replace declaration with that of...
3172         (make_pass_stack_adjustments): ...new function.
3173         (pass_peephole2): Replace declaration with that of...
3174         (make_pass_peephole2): ...new function.
3175         (pass_if_after_reload): Replace declaration with that of...
3176         (make_pass_if_after_reload): ...new function.
3177         (pass_regrename): Replace declaration with that of...
3178         (make_pass_regrename): ...new function.
3179         (pass_cprop_hardreg): Replace declaration with that of...
3180         (make_pass_cprop_hardreg): ...new function.
3181         (pass_reorder_blocks): Replace declaration with that of...
3182         (make_pass_reorder_blocks): ...new function.
3183         (pass_branch_target_load_optimize2): Replace declaration with that
3184         of...
3185         (make_pass_branch_target_load_optimize2): ...new function.
3186         (pass_leaf_regs): Replace declaration with that of...
3187         (make_pass_leaf_regs): ...new function.
3188         (pass_split_before_sched2): Replace declaration with that of...
3189         (make_pass_split_before_sched2): ...new function.
3190         (pass_compare_elim_after_reload): Replace declaration with that of...
3191         (make_pass_compare_elim_after_reload): ...new function.
3192         (pass_sched2): Replace declaration with that of...
3193         (make_pass_sched2): ...new function.
3194         (pass_stack_regs): Replace declaration with that of...
3195         (make_pass_stack_regs): ...new function.
3196         (pass_stack_regs_run): Replace declaration with that of...
3197         (make_pass_stack_regs_run): ...new function.
3198         (pass_df_finish): Replace declaration with that of...
3199         (make_pass_df_finish): ...new function.
3200         (pass_compute_alignments): Replace declaration with that of...
3201         (make_pass_compute_alignments): ...new function.
3202         (pass_duplicate_computed_gotos): Replace declaration with that of...
3203         (make_pass_duplicate_computed_gotos): ...new function.
3204         (pass_variable_tracking): Replace declaration with that of...
3205         (make_pass_variable_tracking): ...new function.
3206         (pass_free_cfg): Replace declaration with that of...
3207         (make_pass_free_cfg): ...new function.
3208         (pass_machine_reorg): Replace declaration with that of...
3209         (make_pass_machine_reorg): ...new function.
3210         (pass_cleanup_barriers): Replace declaration with that of...
3211         (make_pass_cleanup_barriers): ...new function.
3212         (pass_delay_slots): Replace declaration with that of...
3213         (make_pass_delay_slots): ...new function.
3214         (pass_split_for_shorten_branches): Replace declaration with that of...
3215         (make_pass_split_for_shorten_branches): ...new function.
3216         (pass_split_before_regstack): Replace declaration with that of...
3217         (make_pass_split_before_regstack): ...new function.
3218         (pass_convert_to_eh_region_ranges): Replace declaration with that
3219         of...
3220         (make_pass_convert_to_eh_region_ranges): ...new function.
3221         (pass_shorten_branches): Replace declaration with that of...
3222         (make_pass_shorten_branches): ...new function.
3223         (pass_set_nothrow_function_flags): Replace declaration with that of...
3224         (make_pass_set_nothrow_function_flags): ...new function.
3225         (pass_dwarf2_frame): Replace declaration with that of...
3226         (make_pass_dwarf2_frame): ...new function.
3227         (pass_final): Replace declaration with that of...
3228         (make_pass_final): ...new function.
3229         (pass_rtl_seqabstr): Replace declaration with that of...
3230         (make_pass_rtl_seqabstr): ...new function.
3231         (pass_release_ssa_names): Replace declaration with that of...
3232         (make_pass_release_ssa_names): ...new function.
3233         (pass_early_inline): Replace declaration with that of...
3234         (make_pass_early_inline): ...new function.
3235         (pass_inline_parameters): Replace declaration with that of...
3236         (make_pass_inline_parameters): ...new function.
3237         (pass_update_address_taken): Replace declaration with that of...
3238         (make_pass_update_address_taken): ...new function.
3239         (pass_convert_switch): Replace declaration with that of...
3240         (make_pass_convert_switch): ...new function.
3241         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3242         to a subclass of simple_ipa_opt_pass along with...
3243         (pass_data_ipa_tree_profile): ...new pass_data instance and...
3244         (make_pass_ipa_tree_profile): ...new function.
3245         * tree-sra.c (pass_sra_early): Convert from a global struct to a
3246         subclass of gimple_opt_pass along with...
3247         (pass_data_sra_early): ...new pass_data instance and...
3248         (make_pass_sra_early): ...new function.
3249         (pass_sra): Convert from a global struct to a subclass of
3250         gimple_opt_pass along with...
3251         (pass_data_sra): ...new pass_data instance and...
3252         (make_pass_sra): ...new function.
3253         (pass_early_ipa_sra): Convert from a global struct to a subclass of
3254         gimple_opt_pass along with...
3255         (pass_data_early_ipa_sra): ...new pass_data instance and...
3256         (make_pass_early_ipa_sra): ...new function.
3257         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3258         subclass of gimple_opt_pass along with...
3259         (pass_data_ccp): ...new pass_data instance and...
3260         (make_pass_ccp): ...new function.
3261         (pass_fold_builtins): Convert from a global struct to a subclass of
3262         gimple_opt_pass along with...
3263         (pass_data_fold_builtins): ...new pass_data instance and...
3264         (make_pass_fold_builtins): ...new function.
3265         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3266         subclass of gimple_opt_pass along with...
3267         (pass_data_copy_prop): ...new pass_data instance and...
3268         (make_pass_copy_prop): ...new function.
3269         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3270         global struct to a subclass of gimple_opt_pass along with...
3271         (pass_data_rename_ssa_copies): ...new pass_data instance and...
3272         (make_pass_rename_ssa_copies): ...new function.
3273         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3274         subclass of gimple_opt_pass along with...
3275         (pass_data_dce): ...new pass_data instance and...
3276         (make_pass_dce): ...new function.
3277         (pass_dce_loop): Convert from a global struct to a subclass of
3278         gimple_opt_pass along with...
3279         (pass_data_dce_loop): ...new pass_data instance and...
3280         (make_pass_dce_loop): ...new function.
3281         (pass_cd_dce): Convert from a global struct to a subclass of
3282         gimple_opt_pass along with...
3283         (pass_data_cd_dce): ...new pass_data instance and...
3284         (make_pass_cd_dce): ...new function.
3285         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3286         subclass of gimple_opt_pass along with...
3287         (pass_data_dominator): ...new pass_data instance and...
3288         (make_pass_dominator): ...new function.
3289         (pass_phi_only_cprop): Convert from a global struct to a subclass of
3290         gimple_opt_pass along with...
3291         (pass_data_phi_only_cprop): ...new pass_data instance and...
3292         (make_pass_phi_only_cprop): ...new function.
3293         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3294         subclass of gimple_opt_pass along with...
3295         (pass_data_dse): ...new pass_data instance and...
3296         (make_pass_dse): ...new function.
3297         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3298         a subclass of gimple_opt_pass along with...
3299         (pass_data_forwprop): ...new pass_data instance and...
3300         (make_pass_forwprop): ...new function.
3301         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3302         struct to a subclass of gimple_opt_pass along with...
3303         (pass_data_tree_ifcombine): ...new pass_data instance and...
3304         (make_pass_tree_ifcombine): ...new function.
3305         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3306         subclass of gimple_opt_pass along with...
3307         (pass_data_ch): ...new pass_data instance and...
3308         (make_pass_ch): ...new function.
3309         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3310         subclass of gimple_opt_pass along with...
3311         (pass_data_tree_loop): ...new pass_data instance and...
3312         (make_pass_tree_loop): ...new function.
3313         (pass_tree_loop_init): Convert from a global struct to a subclass of
3314         gimple_opt_pass along with...
3315         (pass_data_tree_loop_init): ...new pass_data instance and...
3316         (make_pass_tree_loop_init): ...new function.
3317         (pass_lim): Convert from a global struct to a subclass of
3318         gimple_opt_pass along with...
3319         (pass_data_lim): ...new pass_data instance and...
3320         (make_pass_lim): ...new function.
3321         (pass_tree_unswitch): Convert from a global struct to a subclass of
3322         gimple_opt_pass along with...
3323         (pass_data_tree_unswitch): ...new pass_data instance and...
3324         (make_pass_tree_unswitch): ...new function.
3325         (pass_predcom): Convert from a global struct to a subclass of
3326         gimple_opt_pass along with...
3327         (pass_data_predcom): ...new pass_data instance and...
3328         (make_pass_predcom): ...new function.
3329         (pass_vectorize): Convert from a global struct to a subclass of
3330         gimple_opt_pass along with...
3331         (pass_data_vectorize): ...new pass_data instance and...
3332         (make_pass_vectorize): ...new function.
3333         (pass_graphite): Convert from a global struct to a subclass of
3334         gimple_opt_pass along with...
3335         (pass_data_graphite): ...new pass_data instance and...
3336         (make_pass_graphite): ...new function.
3337         (pass_graphite_transforms): Convert from a global struct to a subclass
3338         of gimple_opt_pass along with...
3339         (pass_data_graphite_transforms): ...new pass_data instance and...
3340         (make_pass_graphite_transforms): ...new function.
3341         (pass_check_data_deps): Convert from a global struct to a subclass of
3342         gimple_opt_pass along with...
3343         (pass_data_check_data_deps): ...new pass_data instance and...
3344         (make_pass_check_data_deps): ...new function.
3345         (pass_iv_canon): Convert from a global struct to a subclass of
3346         gimple_opt_pass along with...
3347         (pass_data_iv_canon): ...new pass_data instance and...
3348         (make_pass_iv_canon): ...new function.
3349         (pass_scev_cprop): Convert from a global struct to a subclass of
3350         gimple_opt_pass along with...
3351         (pass_data_scev_cprop): ...new pass_data instance and...
3352         (make_pass_scev_cprop): ...new function.
3353         (pass_record_bounds): Convert from a global struct to a subclass of
3354         gimple_opt_pass along with...
3355         (pass_data_record_bounds): ...new pass_data instance and...
3356         (make_pass_record_bounds): ...new function.
3357         (pass_complete_unroll): Convert from a global struct to a subclass of
3358         gimple_opt_pass along with...
3359         (pass_data_complete_unroll): ...new pass_data instance and...
3360         (make_pass_complete_unroll): ...new function.
3361         (pass_complete_unrolli): Convert from a global struct to a subclass of
3362         gimple_opt_pass along with...
3363         (pass_data_complete_unrolli): ...new pass_data instance and...
3364         (make_pass_complete_unrolli): ...new function.
3365         (pass_parallelize_loops): Convert from a global struct to a subclass
3366         of gimple_opt_pass along with...
3367         (pass_data_parallelize_loops): ...new pass_data instance and...
3368         (make_pass_parallelize_loops): ...new function.
3369         (pass_loop_prefetch): Convert from a global struct to a subclass of
3370         gimple_opt_pass along with...
3371         (pass_data_loop_prefetch): ...new pass_data instance and...
3372         (make_pass_loop_prefetch): ...new function.
3373         (pass_iv_optimize): Convert from a global struct to a subclass of
3374         gimple_opt_pass along with...
3375         (pass_data_iv_optimize): ...new pass_data instance and...
3376         (make_pass_iv_optimize): ...new function.
3377         (pass_tree_loop_done): Convert from a global struct to a subclass of
3378         gimple_opt_pass along with...
3379         (pass_data_tree_loop_done): ...new pass_data instance and...
3380         (make_pass_tree_loop_done): ...new function.
3381         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3382         struct to a subclass of gimple_opt_pass along with...
3383         (pass_data_cse_reciprocals): ...new pass_data instance and...
3384         (make_pass_cse_reciprocals): ...new function.
3385         (pass_cse_sincos): Convert from a global struct to a subclass of
3386         gimple_opt_pass along with...
3387         (pass_data_cse_sincos): ...new pass_data instance and...
3388         (make_pass_cse_sincos): ...new function.
3389         (pass_optimize_bswap): Convert from a global struct to a subclass of
3390         gimple_opt_pass along with...
3391         (pass_data_optimize_bswap): ...new pass_data instance and...
3392         (make_pass_optimize_bswap): ...new function.
3393         (pass_optimize_widening_mul): Convert from a global struct to a
3394         subclass of gimple_opt_pass along with...
3395         (pass_data_optimize_widening_mul): ...new pass_data instance and...
3396         (make_pass_optimize_widening_mul): ...new function.
3397         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3398         subclass of gimple_opt_pass along with...
3399         (pass_data_phiopt): ...new pass_data instance and...
3400         (make_pass_phiopt): ...new function.
3401         (pass_cselim): Convert from a global struct to a subclass of
3402         gimple_opt_pass along with...
3403         (pass_data_cselim): ...new pass_data instance and...
3404         (make_pass_cselim): ...new function.
3405         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3406         subclass of gimple_opt_pass along with...
3407         (pass_data_phiprop): ...new pass_data instance and...
3408         (make_pass_phiprop): ...new function.
3409         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3410         subclass of gimple_opt_pass along with...
3411         (pass_data_pre): ...new pass_data instance and...
3412         (make_pass_pre): ...new function.
3413         (pass_fre): Convert from a global struct to a subclass of
3414         gimple_opt_pass along with...
3415         (pass_data_fre): ...new pass_data instance and...
3416         (make_pass_fre): ...new function.
3417         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3418         subclass of gimple_opt_pass along with...
3419         (pass_data_reassoc): ...new pass_data instance and...
3420         (make_pass_reassoc): ...new function.
3421         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3422         subclass of gimple_opt_pass along with...
3423         (pass_data_sink_code): ...new pass_data instance and...
3424         (make_pass_sink_code): ...new function.
3425         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3426         subclass of gimple_opt_pass along with...
3427         (pass_data_strlen): ...new pass_data instance and...
3428         (make_pass_strlen): ...new function.
3429         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3430         struct to a subclass of gimple_opt_pass along with...
3431         (pass_data_build_alias): ...new pass_data instance and...
3432         (make_pass_build_alias): ...new function.
3433         (pass_build_ealias): Convert from a global struct to a subclass of
3434         gimple_opt_pass along with...
3435         (pass_data_build_ealias): ...new pass_data instance and...
3436         (make_pass_build_ealias): ...new function.
3437         (pass_ipa_pta): Convert from a global struct to a subclass of
3438         simple_ipa_opt_pass along with...
3439         (pass_data_ipa_pta): ...new pass_data instance and...
3440         (make_pass_ipa_pta): ...new function.
3441         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3442         subclass of gimple_opt_pass along with...
3443         (pass_data_uncprop): ...new pass_data instance and...
3444         (make_pass_uncprop): ...new function.
3445         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3446         global struct to a subclass of gimple_opt_pass along with...
3447         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3448         (make_pass_late_warn_uninitialized): ...new function.
3449         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3450         to a subclass of gimple_opt_pass along with...
3451         (pass_data_init_datastructures): ...new pass_data instance and...
3452         (make_pass_init_datastructures): ...new function.
3453         (pass_early_warn_uninitialized): Convert from a global struct to a
3454         subclass of gimple_opt_pass along with...
3455         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3456         (make_pass_early_warn_uninitialized): ...new function.
3457         (pass_update_address_taken): Convert from a global struct to a
3458         subclass of gimple_opt_pass along with...
3459         (pass_data_update_address_taken): ...new pass_data instance and...
3460         (make_pass_update_address_taken): ...new function.
3461         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3462         struct to a subclass of gimple_opt_pass along with...
3463         (pass_data_release_ssa_names): ...new pass_data instance and...
3464         (make_pass_release_ssa_names): ...new function.
3465         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3466         subclass of gimple_opt_pass along with...
3467         (pass_data_stdarg): ...new pass_data instance and...
3468         (make_pass_stdarg): ...new function.
3469         * tree-switch-conversion.c (pass_convert_switch): Convert from a
3470         global struct to a subclass of gimple_opt_pass along with...
3471         (pass_data_convert_switch): ...new pass_data instance and...
3472         (make_pass_convert_switch): ...new function.
3473         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3474         to a subclass of gimple_opt_pass along with...
3475         (pass_data_tail_recursion): ...new pass_data instance and...
3476         (make_pass_tail_recursion): ...new function.
3477         (pass_tail_calls): Convert from a global struct to a subclass of
3478         gimple_opt_pass along with...
3479         (pass_data_tail_calls): ...new pass_data instance and...
3480         (make_pass_tail_calls): ...new function.
3481         * tree-vect-generic.c (pass_lower_vector): Convert from a global
3482         struct to a subclass of gimple_opt_pass along with...
3483         (pass_data_lower_vector): ...new pass_data instance and...
3484         (make_pass_lower_vector): ...new function.
3485         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3486         gimple_opt_pass along with...
3487         (pass_data_lower_vector_ssa): ...new pass_data instance and...
3488         (make_pass_lower_vector_ssa): ...new function.
3489         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3490         to a subclass of gimple_opt_pass along with...
3491         (pass_data_slp_vectorize): ...new pass_data instance and...
3492         (make_pass_slp_vectorize): ...new function.
3493         (pass_ipa_increase_alignment): Convert from a global struct to a
3494         subclass of simple_ipa_opt_pass along with...
3495         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3496         (make_pass_ipa_increase_alignment): ...new function.
3497         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3498         gimple_opt_pass along with...
3499         (pass_data_vrp): ...new pass_data instance and...
3500         (make_pass_vrp): ...new function.
3501         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3502         subclass of simple_ipa_opt_pass along with...
3503         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3504         (make_pass_ipa_free_lang_data): ...new function.
3505         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3506         gimple_opt_pass along with...
3507         (pass_data_tsan): ...new pass_data instance and...
3508         (make_pass_tsan): ...new function.
3509         (pass_tsan_O0): Convert from a global struct to a subclass of
3510         gimple_opt_pass along with...
3511         (pass_data_tsan_O0): ...new pass_data instance and...
3512         (make_pass_tsan_O0): ...new function.
3513         * var-tracking.c (pass_variable_tracking): Convert from a global
3514         struct to a subclass of rtl_opt_pass along with...
3515         (pass_data_variable_tracking): ...new pass_data instance and...
3516         (make_pass_variable_tracking): ...new function.
3517         * web.c (pass_web): Convert from a global struct to a subclass of
3518         rtl_opt_pass along with...
3519         (pass_data_web): ...new pass_data instance and...
3520         (make_pass_web): ...new function.
3521         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3522         declaration with that of...
3523         (make_pass_mode_switch_use): ...new function.
3524         (pass_resolve_sw_modes): Replace declaration with that of...
3525         (make_pass_resolve_sw_modes): ...new function.
3526         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3527         from a global struct to a subclass of rtl_opt_pass along with...
3528         (pass_data_mode_switch_use): ...new pass_data instance and...
3529         (make_pass_mode_switch_use): ...new function.
3530         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3531         from a global struct to a subclass of rtl_opt_pass along with...
3532         (pass_data_resolve_sw_modes): ...new pass_data instance and...
3533         (make_pass_resolve_sw_modes): ...new function.
3534         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3535         struct to a subclass of rtl_opt_pass along with...
3536         (pass_data_insert_vzeroupper): ...new pass_data instance and...
3537         (make_pass_insert_vzeroupper): ...new function.
3538         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3539         global struct to a subclass of rtl_opt_pass along with...
3540         (pass_data_work_around_errata): ...new pass_data instance and...
3541         (make_pass_work_around_errata): ...new function.
3542         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3543         struct to a subclass of rtl_opt_pass along with...
3544         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3545         (make_pass_mips_machine_reorg2): ...new function.
3547 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3549         * passes.c (pass_manager::operator new): New.
3551 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3553         Handwritten part of conversion of passes to C++ classes.
3555         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3556         (toplev.o): Add dep on PASS_MANAGER_H.
3557         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3558         of early local pases to reflect this moving from a global to a
3559         member of gcc::pass_manager.
3560         (cgraph_add_new_function): Likewise.
3561         * lto-cgraph.c (lto_output_node): Update for conversion of
3562         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3563         * passes.c (opt_pass::clone): New.
3564         (opt_pass::gate): New.
3565         (opt_pass::execute): New.
3566         (opt_pass::opt_pass): New.
3567         (pass_manager::execute_early_local_passes): New.
3568         (pass_manager::execute_pass_mode_switching): new.
3569         (finish_optimization_passes): Convert to...
3570         (pass_manager::finish_optimization_passes): ...this.
3571         (finish_optimization_passes): Update for conversion of passes to
3572         C++ classes.
3573         (register_dump_files_1): Use has_gate since we cannot portably
3574         check a vtable entry against NULL.
3575         (dump_one_pass): Likewise.
3576         (ipa_write_summaries_2): Likewise.
3577         (ipa_write_optimization_summaries_1): Likewise.
3578         (ipa_read_summaries_1): Likewise.
3579         (ipa_read_optimization_summaries_1): Likewise.
3580         (execute_ipa_stmt_fixups): Likewise.
3581         (pass_manager::pass_manager): Rewrite pass-creation, invoking
3582         pass-creation functions rather than wiring up globals, and
3583         storing the results in fields of pass_manager generated using
3584         pass-instances.def.
3585         (pass_manager::dump_profile_report): Update for conversion of
3586         passes to C++ classes.
3587         (pass_manager::execute_ipa_summary_passes): Likewise.
3588         (execute_one_ipa_transform_pass): Likewise.
3589         (execute_one_pass): Use has_gate and has_execute since we cannot
3590         portably check a vtable entry against NULL.
3591         * pass_manager.h (pass_manager::finish_optimization_passes): New.
3592         (pass_manager): Use pass-instances.def to add fields for the
3593         various pass instances.
3594         * toplev.c (finalize): Update for move of
3595         finish_optimization_passes to a method of gcc::pass_manager.
3596         * toplev.h (finish_optimization_passes): Move to method of class
3597         pass_manager.
3598         * tree-pass.h (struct pass_data): New.
3599         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3600         (opt_pass::gate): Convert to virtual function.
3601         (opt_pass::~opt_pass): New.
3602         (opt_pass::clone): New.
3603         (opt_pass::execute): Convert to virtual function.
3604         (opt_pass::opt_pass): New.
3605         (opt_pass::ctxt_): new.
3606         (gimple_opt_pass): Convert to subclass of opt_pass.
3607         (gimple_opt_pass::gimple_opt_pass): New.
3608         (rtl_opt_pass): Convert to subclass of opt_pass.
3609         (rtl_opt_pass::rtl_opt_pass): New.
3610         (ipa_opt_pass_d): Convert to subclass of opt_pass.
3611         (ipa_opt_pass_d::ipa_opt_pass_d): New.
3612         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3613         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3614         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3615         invocation of pass_mode_switching to reflect this moving from a
3616         global to a member of gcc::pass_manager.
3617         (ix86_option_override): Rework how pass_insert_vzeroupper is
3618         added to the pass_manager to reflect autogenerated changes.
3619         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3621 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
3623         PR rtl-optimization/57708
3624         * recog.c (peep2_find_free_register): Validate all regs in a
3625         multi-reg mode.
3627 2013-08-05  Jan Hubicka  <jh@suse.cz>
3629         PR lto/57602
3630         * cgraph.c (verify_cgraph_node): Accept local flags from other
3631         partitions.
3632         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3633         (function_and_variable_visibility): Likewise.
3634         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3636 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3638         * graph.c (init_graph_slim_pretty_print): Remove.
3639         (print_graph_cfg): Do not call it.  Use local pretty printer.
3640         (start_graph_dump): Likewise.
3642 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3644         * gimple-pretty-print.c (buffer): Remove.
3645         (initialized): Likewise.
3646         (maybe_init_pretty_print): Likewise.
3647         (print_gimple_stmt): Do not call it.  Use non-static local
3648         pretty_printer variable.
3649         (print_gimple_expr): Likewise.
3650         (print_gimple_seq): Likewise.
3651         (gimple_dump_bb): Likewise.
3653 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3655         * asan.c (asan_pp): Remove.
3656         (asan_pp_initialized): Likewise.
3657         (asan_pp_initialize): Likewise.
3658         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
3659         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
3660         (asan_add_global): Likewise.
3662 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3664         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
3665         * diagnostic.h (diagnostic_flush_buffer): Adjust.
3666         * pretty-print.c (pp_formatted_text_data): Likewise.
3667         (pp_indent): Rename from pp_base_indent.
3668         (pp_format): Rename from pp_base_format.
3669         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3670         (pp_format_verbatim): Rename from pp_base_format_verbatim.
3671         (pp_flush): Rename from pp_base_flush.
3672         (pp_set_line_maximum_length): Rename from
3673         pp_base_set_line_maximum_length.
3674         (pp_clear_output_area): Rename from pp_base_clear_output_area.
3675         (pp_set_prefix): Rename from pp_base_set_prefix.
3676         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3677         (pp_emit_prefix): Rename from pp_base_emit_prefix.
3678         (pp_append_text): Rename from pp_base_append_text.
3679         (pp_formatted_text): Rename from pp_base_formatted_text.
3680         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3681         (pp_remaining_character_count_for_line): Rename from
3682         pp_base_remaining_character_count_for_line.
3683         (pp_newline): Rename from pp_base_newline.
3684         (pp_character): Rename from pp_base_character.
3685         (pp_string): Rename from pp_base_string.
3686         (pp_maybe_space): Rename from pp_base_maybe_space.
3687         * asan.c (asan_pp_string): Adjust.
3688         (asan_emit_stack_protection): Likewise.
3689         (asan_add_global): Likewise.
3690         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3691         * tree-mudflap.c (mf_varname_tree): Likewise.
3692         * tree-pretty-print.c (pp_tree_identifier): Rename from
3693         pp_base_tree_identifier.
3694         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3695         Declare as function.
3697 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3699         * pretty-print.h (pp_bar_bar): New.
3700         (pp_ampersand_ampersand): Likewise.
3701         (pp_less_equal): Likewise.
3702         (pp_greater_equal): Likewise.
3703         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3704         printer functions instead of pp_string or operators and punctuators.
3705         (dump_gimple_call): Likewise.
3706         (dump_gimple_omp_for): Likewise.
3707         (dump_gimple_transaction): Likewise.
3708         (dump_gimple_phi): Likewise.
3709         (pp_gimple_stmt_1): Likewise.
3710         * sched-vis.c (print_insn): Likewise.
3711         * tree-mudflap.c (mf_varname_tree): Likewise.
3712         * tree-pretty-print.c (dump_block_node): Likewise.
3713         (dump_generic_node): Likewise.
3715 2013-08-02  Jan Hubicka  <jh@suse.cz>
3717         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3718         boundaries.
3719         * lto-streamer-out.c (tree_is_indexable): Results decls and
3720         parm decls are not indexable.
3721         (DFS_write_tree_body): Do not follow args and results.
3722         (hash_tree): Likewise.
3723         (output_functions): Rearrange so struct function is needed
3724         only when real body is output; be able to also ouptut abstract
3725         functions; output DECL_ARGUMENTS and DECL_RESULT.
3726         (lto_output): When not in WPA, ale store abstract functions.
3727         (write_symbol): Do not care about RESULT_DECL.
3728         (output_symbol_p): Handle correctly sbtract decls.
3729         * lto-streamer-in.c (input_function): Rearrange so struct
3730         function can be NULL at entry; allow streaming of
3731         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3732         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3733         sanity check during LTO.
3734         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3735         RESULT_DECl and DECL_ARGUMENTS.
3736         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3737         Likewise.
3739 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3741         * pretty-print.h (pp_underscore): New.
3742         (pp_comma): Tidy.
3743         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3744         printer functions instead of pp_character.
3745         (dump_binary_rhs): Likewise.
3746         (dump_ternary_rhs): Likewise.
3747         (dump_gimple_call_args): Likewise.
3748         (pp_points_to_solution): Likewise.
3749         (dump_gimple_call): Likewise.
3750         (dump_gimple_switch): Likewise.
3751         (dump_gimple_cond): Likewise.
3752         (dump_gimple_bind): Likewise.
3753         (dump_gimple_try): Likewise.
3754         (dump_gimple_omp_for): Likewise.
3755         (dump_gimple_omp_continue): Likewise.
3756         (dump_gimple_omp_single): Likewise.
3757         (dump_gimple_omp_sections): Likewise.
3758         (dump_gimple_omp_block): Likewise.
3759         (dump_gimple_omp_critical): Likewise.
3760         (dump_gimple_transaction): Likewise.
3761         (dump_gimple_asm): Likewise.
3762         (dump_gimple_phi): Likewise.
3763         (dump_gimple_omp_parallel): Likewise.
3764         (dump_gimple_omp_task): Likewise.
3765         (dump_gimple_omp_atomic_load): Likewise.
3766         (dump_gimple_omp_atomic_store): Likewise.
3767         (dump_gimple_mem_ops): Likewise.
3768         (pp_gimple_stmt_1): Likewise.
3769         (pp_cfg_jump): Likewise.
3770         (dump_implicit_edges): Likewise.
3771         (gimple_dump_bb_for_graph): Likewise.
3772         * graph.c (draw_cfg_node): Likewise.
3773         * langhooks.c (lhd_print_error_function): Likewise.
3774         * sched-vis.c (print_exp): Likewise.
3775         (print_value): Likewise.
3776         (print_pattern): Likewise.
3777         (print_insn): Likewise.
3778         (rtl_dump_bb_for_graph): Likewise.
3779         * tree-pretty-print.c (dump_function_declaration): Likewise.
3780         (dump_array_domain): Likewise.
3781         (dump_omp_clause): Likewise.
3782         (dump_location): Likewise.
3783         (dump_generic_node): Likewise.
3784         (print_struct_decl): Likewise.
3785         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3787 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3789         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3790         candidate table when replacing a candidate statement.
3791         (replace_rhs_if_not_dup): Likewise.
3792         (replace_one_candidate): Likewise.
3794 2013-08-02  Jan Hubicka  <jh@suse.cz>
3795             Martin Liska  <marxin.liska@gmail.com>
3797         * cgraphunit.c (add_new_function): Fix logic when adding from
3798         late IPA pass.
3799         (assemble_thunk): Rename to ...
3800         (expand_thunk); .. this one; export; get it working with
3801         general functions; make produced gimple valid.
3802         * cgraph.h (expand_thunk): Declare.
3804 2013-08-02  Jan Hubicka  <jh@suse.cz>
3806         * ipa-cp.c (gather_context_independent_values): Use
3807         ipa_get_param_move_cost.
3808         (get_replacement_map): Remove PARAM; move parameter folding
3809         into tree-inline.c
3810         (create_specialized_node): Update.
3811         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3812         assert that we have gimple body; update move_cost.
3813         (count_formal_params): Assert that we have gimple body.
3814         (ipa_dump_param): New function.
3815         (ipa_alloc_node_params): Break out from ...
3816         (ipa_initialize_node_params): ... here.
3817         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3818         (ipa_write_node_info): Stream move costs.
3819         (ipa_read_node_info): Read move costs.
3820         (ipa_update_after_lto_read): Do not recompute node params.
3821         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3822         (ipa_get_param): Check we are not in WPA.
3823         (ipa_get_param_move_cost): New.
3824         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3825         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3826         parm numbers to be present.
3828 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3830         PR rtl-optimization/58048
3831         * lra-constraints.c (process_alt_operands): Don't check asm
3832         operand on register.
3834 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3836         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3837         the implied StoreLoad barrier for atomic operations if before.
3839 2013-08-02  Jan Hubicka  <jh@suse.cz>
3840             Martin Liska  <marxin.liska@gmail.com>
3842         * cgraph.c (cgraph_function_body_availability): Do not check
3843         cgraph flags.
3844         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3845         symtab_node_availability): Declare.
3846         * ipa.c (can_replace_by_local_alias): New.
3847         (function_and_variable_visibility): Use it.
3848         * symtab.c (symtab_for_node_and_aliases,
3849         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3851 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3853         PR rtl-optimization/57963
3854         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3855         (lra_constraints): Use them.
3857 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3859         * config/arm/types.md (define_attr "type"): Add "load_acq"
3860         and "store_rel".
3861         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3862         changes.
3863         (cortex_a53_store1): Likewise.
3865 2013-08-01  Jan Hubicka  <jh@suse.cz>
3867         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3868         partitions are not needed.
3870 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3872         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3873         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3874         MAYBE_NON_Q_CLASS_P where appropriate.
3876 2013-08-01  Jan Hubicka  <jh@suse.cz>
3878         * cgraph.h (release_function_body): Declare.
3879         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3880         of unused delcarations.
3882 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3884         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3885         RTL form when subtracting a constant.
3887 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3889         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3890         Generate canonical plus rtx with negated immediate instead of minus
3891         where appropriate.
3892         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3894 2013-08-01  Jan Hubicka  <jh@suse.cz>
3896         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3897         (cgraph_release_function_body): Likewise.
3898         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3899         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3900         to used_as_abstract_origin.
3901         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3902         symbols used as abstract origins.
3903         * cgraphunit.c (analyze_functions): Update.
3904         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3905         used_as_abstract_origin.
3906         * tree-inline.c (tree_function_versioning): Update
3907         used_as_abstract_origin; be ready for DECL_RESULT and
3908         DECL_ARGUMENTS to be NULL.
3910         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3911         for abstract functions.
3912         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3913         real symbols.
3915 2013-08-01  Jan Hubicka  <jh@suse.cz>
3917         * profile.c (compute_value_histograms): Fix thinko.
3919 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3921         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3922         aarch-common-protos.h to extra_headers.
3923         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3924         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3925         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3927 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3929         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3930         Include "../arm/types.md".  Define "type" attribute for all patterns.
3931         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3932         attribute changes.
3934 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3936         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3937         to support power8 load fusion.
3938         (fusion_gpr_mem_load): Likewise.
3940         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3942         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3943         declarations for power8 load fusion.
3944         (emit_fusion_gpr_load): Likewise.
3946         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3947         tuning for power8, turn on fusion mode by default.  Turn on sign
3948         extending fusion mode if normal fusion mode is on, and we are at
3949         -O2 or -O3.
3950         (fusion_gpr_load_p): New function, return true if we can fuse an
3951         addis instruction with a dependent load to a GPR.
3952         (emit_fusion_gpr_load): Emit the instructions for power8 load
3953         fusion to GPRs.
3955         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3956         (VSX load fusion peepholes): New peepholes to fuse together an
3957         addi instruction with a VSX load instruction.
3959         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3960         peepholes to fuse an addis instruction with a load to a GPR base
3961         register.  If we are supporting sign extending fusions, convert
3962         sign extending loads to zero extending loads and add an explicit
3963         sign extension.
3965 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3967         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3968         aarch-common-protos.h to extra_headers.
3969         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3970         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3971         (arm_early_store_addr_dep): Likewise.
3972         (arm_no_early_alu_shift_dep): Likewise.
3973         (arm_no_early_alu_shift_value_dep): Likewise.
3974         (arm_no_early_mul_dep): Likewise.
3975         (arm_no_early_store_addr_dep): Likewise.
3976         (arm_mac_accumulator_is_mul_result): Likewise.
3977         (arm_mac_accumulator_is_result): Likewise.
3978         * config/arm/aarch-common.c: ... here.  New file.
3979         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3980         here to ...
3981         (arm_early_store_addr_dep): Likewise.
3982         (arm_no_early_alu_shift_dep): Likewise.
3983         (arm_no_early_alu_shift_value_dep): Likewise.
3984         (arm_no_early_mul_dep): Likewise.
3985         (arm_no_early_store_addr_dep): Likewise.
3986         (arm_mac_accumulator_is_mul_result): Likewise.
3987         (arm_mac_accumulator_is_result): Likewise.
3988         * config/arm/aarch-common-protos.h: ... here.  New file.
3989         * config/arm/t-arm (aarch-common.o): Define.
3991 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3993         * config/arm/arm.md: Include new file "types.md".
3994         (define_attr "type"): Move from here to ...
3995         (define_attr "mul32"): Likewise.
3996         (define_attr "mul64"): Likewise.
3997         * config/arm/types.md: ... here.  New file.
3999 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4001         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4002         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4004 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4006         * gen-pass-instances.awk: Fix offset of substr().
4008 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
4010         * Makefile.in (pass-instances.def): New.
4011         (passes.o): Replace dependency on passes.def with one on
4012         pass-instances.def
4014         * gen-pass-instances.awk: New.
4016         * passes.c (pass_manager::pass_manager): Use pass-instances.def
4017         rather than passes.def, updating local definition of NEXT_PASS
4018         macro to add an extra NUM parameter (currently unused).
4020 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
4022         * Makefile.in (PASS_MANAGER_H): New.
4023         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4024         (passes.o): Likewise.
4025         (statistics.o): Likewise.
4026         (cgraphunit.o): Likewise.
4027         (context.o): Depend on PASS_MANAGER_H.
4029         * pass_manager.h: New.
4031         * cgraphunit.c (cgraph_add_new_function): Update for moves
4032         of globals to fields of pass_manager.
4033         (analyze_function): Likewise.
4034         (expand_function): Likewise.
4035         (ipa_passes): Likewise.
4036         (compile): Likewise.
4038         * context.c (context::context): New.
4039         * context.h  (context::context): New.
4040         (context::get_passes): New.
4041         (context::passes_): New.
4043         * lto-cgraph.c (input_node): Update for moves of globals to
4044         fields of pass_manager.
4046         * passes.c (all_passes): Remove, in favor of a field of the
4047         same name within the new class pass_manager.
4048         (all_small_ipa_passes): Likewise.
4049         (all_lowering_passes): Likewise.
4050         (all_regular_ipa_passes): Likewise.
4051         (all_late_ipa_passes): Likewise.
4052         (all_lto_gen_passes): Likewise.
4053         (passes_by_id): Likewise.
4054         (passes_by_id_size): Likewise.
4055         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4056         the new class pass_manager.
4057         (set_pass_for_id): Convert to...
4058         (pass_manager::set_pass_for_id): ...method.
4059         (get_pass_for_id): Convert to...
4060         (pass_manager::get_pass_for_id): ...method.
4061         (register_one_dump_file): Move body of implementation into...
4062         (pass_manager::register_one_dump_file): ...here.
4063         (register_dump_files_1): Convert to...
4064         (pass_manager::register_dump_files_1): ...method.
4065         (register_dump_files): Convert to...
4066         (pass_manager::register_dump_files): ...method.
4067         (create_pass_tab): Update for moves of globals to fields of
4068         pass_manager.
4069         (dump_passes): Move body of implementation into...
4070         (pass_manager::dump_passes): ...here.
4071         (register_pass): Move body of implementation into...
4072         (pass_manager::register_pass): ...here.
4073         (init_optimization_passes): Convert into...
4074         (pass_manager::pass_manager): ...constructor for new
4075         pass_manager class, and initialize the pass_lists array.
4076         (check_profile_consistency): Update for moves of globals to
4077         fields of pass_manager.
4078         (dump_profile_report): Move body of implementation into...
4079         (pass_manager::dump_profile_report): ...here.
4080         (ipa_write_summaries_1): Update for moves of pass lists from
4081         being globals to fields of pass_manager.
4082         (ipa_write_optimization_summaries): Likewise.
4083         (ipa_read_summaries):  Likewise.
4084         (ipa_read_optimization_summaries): Likewise.
4085         (execute_all_ipa_stmt_fixups): Likewise.
4087         * statistics.c (statistics_fini): Update for moves of globals to
4088         fields of pass_manager.
4090         * toplev.c (general_init): Replace call to
4091         init_optimization_passes with construction of the pass_manager
4092         instance.
4094         * tree-pass.h (all_passes): Remove, in favor of a field of the
4095         same name within the new class pass_manager.
4096         (all_small_ipa_passes): Likewise.
4097         (all_lowering_passes): Likewise.
4098         (all_regular_ipa_passes): Likewise.
4099         (all_lto_gen_passes): Likewise.
4100         (all_late_ipa_passes): Likewise.
4101         (passes_by_id): Likewise.
4102         (passes_by_id_size): Likewise.
4103         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4104         the new class pass_manager.
4105         (get_pass_for_id): Remove.
4107 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4109         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4110         configs.
4112 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4114         * arm.md (mulhi3): New expand pattern.
4116 2013-07-30  Jan Hubicka  <jh@suse.cz>
4117             Martin Liska  <marxin.liska@gmail.com>
4119         * profile.c (compute_value_histograms): Do not ICE when
4120         there is mismatch only on some counters.
4122 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4124         PR rtl-optimization/57637
4125         * function.c (move_insn_for_shrink_wrap): Also check the
4126         GEN set of the LIVE problem for the liveness analysis
4127         if it exists, otherwise give up.
4129 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4131         PR tree-optimization/57993
4132         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4133         replaced statement in the candidate table.
4134         (phi_add_costs): Return infinite cost when the hidden basis does
4135         not dominate all phis on which the candidate is dependent.
4136         (replace_one_candidate): Record replaced statement in the
4137         candidate table.
4139 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
4141         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4142         (ashlv2si3): New expander.
4143         (*ashlv2si3_i): New define_insn_and_split.
4144         * predicates.md (float_operation): Allow patterns with three
4145         basic sub-patterns.
4147         PR rtl-optimization/58021
4148         * mode-switching.c (create_pre_exit): Always split off preceding
4149         insns if we are not at the basic block head.
4151 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
4153         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4154         (UCLIBC_DYNAMIC_LINKER): New macro.
4155         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4156         `-mnan=2008'.
4157         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4158         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
4159         `-mnan=2008'.
4160         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4161         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4162         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4163         for SF and DF modes.  Use ieee_quad_format for TF mode.
4164         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4165         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4166         (mips_option_override): Handle `-mnan=legacy'.
4167         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4168         `-mabs=2008' and `-mnan=2008'.
4169         (OPTION_DEFAULT_SPECS): Add "nan" default.
4170         (ASM_SPEC): Handle `-mnan='.
4171         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4172         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4173         comment accordingly.
4174         (neg<mode>2): Likewise.
4175         * config/mips/mips.opt (mabs, mnan): New options.
4176         * doc/install.texi (Configuration): Document `--with-nan=' option.
4177         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4178         `-mnan=' options.
4179         (MIPS Options): Document them.
4180         * config.gcc <mips*-*-*>: Handle `--with-nan='.
4181         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4182         * configure: Regenerate.
4183         * config.in: Regenerate.
4185 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4187         * config/i386/i386.md (float post-reload splitters): Do not check
4188         for subregs of SSE registers.
4190 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4191             H.J. Lu  <hongjiu.lu@intel.com>
4193         PR target/57954
4194         PR target/57988
4195         * config/i386/i386.md (post-reload splitter
4196         to avoid partial SSE reg dependency stalls): New pattern.
4198 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4200         * config/s390/s390.md ("movcc"): Swap load and store instructions.
4202 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
4204         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4205         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4207 2013-07-26  Cary Coutant  <ccoutant@google.com>
4209         * dwarf2out.c (die_checksum_ordered): Don't include template
4210         instantiations in signature.
4211         (is_template_parameter): New function.
4212         (is_template_instantiation): New function.
4213         (generate_skeleton_bottom_up): Don't include template instantiations
4214         in type unit DIE.
4215         (generate_skeleton): Likewise.
4216         (break_out_comdat_types): Move recursive call to break out nested
4217         types earlier.
4218         (prune_unused_types_mark_generic_parms_dies): Call
4219         is_template_parameter.
4221 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
4223         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4224         uses vector registers.
4225         * config/aarch64/iterators.md: Add attributes rtn and vas.
4227 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4228             Richard Earnshaw  <richard.earnshaw@arm.com>
4230         * combine.c (simplify_comparison): Re-canonicalize operands
4231         where appropriate.
4232         * config/arm/arm.md (movcond_addsi): New splitter.
4234 2013-07-25  Sterling Augustine  <saugustine@google.com>
4236         * dwarf2out.c (size_of_pubnames): Move code to...
4237         (include_pubname_in_output): ...here.  New.
4238         (want_pubnames): Rearrange.
4239         (output_pubnames): Call include_pubname_in_output.  Move assertion.
4241 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4243         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4245 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4247         PR target/38836
4248         * doc/extend.texi: Remove obsolete builtins. Fix
4249         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4251 2013-07-25  Jan Hubicka  <jh@suse.cz>
4253         * cgraph.c (release_function_body): Break out from ...
4254         (cgraph_release_function_body): ... this one; also release DECL_RESULT
4255         and DECL_ARGUMENTS.
4256         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4257         old_tree in the map.
4258         (create_specialized_node): Update.
4259         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4260         into index.
4261         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4262         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4263         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4264         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4265         DECL_RESULT.
4267 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4269         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4270         addsi3_carryin_alt2_<optab>): Correct output template.
4272 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4274         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4275         Adjust for arm_restrict_it.
4276         Remove trailing whitespace.
4278 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
4280         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4281         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4283         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4285 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
4287         PR rtl-optimization/57960
4288         * lra-constraints.c (process_alt_operands): Use the right mode
4289         when checking strict_low.
4291 2013-07-25  Jan Hubicka  <jh@suse.cz>
4293         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4294         * cgraph.c (cgraph_remove_node): Do not release function body
4295         when in cgraph streaming.
4296         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4297         in other partitions are not considered reachable; fix handling of
4298         clones.
4300 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4302         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4304 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4306         PR target/19599
4307         PR target/57731
4308         PR target/57837
4309         * config/arm/arm.md ("*sibcall_insn): Replace use of
4310         Ss with US. Adjust output for v5 and v4t.
4311         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4313         * config/arm/constraints.md ("Ss"): Rename to US.
4315 2013-07-25  Terry Guo  <terry.guo@arm.com>
4317         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4318         shift_add/shift_sub0/shift_sub1 RTXs.
4320 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
4321             Anton Blanchard  <anton@au1.ibm.com>
4323         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4324         (altivec_vpks<VI_char>ss): Likewise.
4325         (altivec_vpks<VI_char>us): Likewise.
4326         (altivec_vpku<VI_char>us): Likewise.
4327         (altivec_vpku<VI_char>um): Likewise.
4329 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
4331         Introduce context class.
4333         * Makefile.in (CONTEXT_H): New.
4334         (OBJS): Add context.o.
4335         (toplev.o): Add CONTEXT_H to dependencies.
4336         (context.o): New.
4338         * toplev.c (general_init): Create the singleton gcc::context instance.
4340         * context.c: New.
4342         * context.h: New.
4344 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
4346         PR rtl-optimization/57968
4347         * mode-switching.c (create_pre_exit): Allow instructions that
4348         don't set a return register to need a non-exit mode.
4350 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4351             Anton Blanchard  <anton@au1.ibm.com>
4353         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4354         operands to vperm for little endian.
4355         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4356         of lvsl to create the control mask for a vperm for little endian.
4358 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4359             Anton Blanchard  <anton@au1.ibm.com>
4361         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4362         two operands for little-endian.
4364 2013-07-23  Steve Ellcey  <sellcey@mips.com>
4366         * config/mips/mips.c (mips_case_values_threshold): New.
4367         (TARGET_CASE_VALUES_THRESHOLD): Define.
4369 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4370             Anton Blanchard  <anton@au1.ibm.com>
4372         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4373         selection of field for vector splat in little endian mode.
4375 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4377         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4378         expanders to rs6000.md.
4379         (ior<mode>3): Likewise.
4380         (and<mode>3): Likewise.
4381         (one_cmpl<mode>2): Likewise.
4382         (nor<mode>3): Likewise.
4383         (andc<mode>3): Likewise.
4384         (eqv<mode>3): Likewise.
4385         (nand<mode>3): Likewise.
4386         (orc<mode>3): Likewise.
4388         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4389         declaration.
4391         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4392         to split multi-word logical operations.
4393         (rs6000_split_logical_di): Likewise.
4394         (rs6000_split_logical): Likewise.
4396         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4397         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4398         and allow TImode operations in 32-bit.
4399         (vsx_and<mode>3_64bit): Likewise.
4400         (vsx_ior<mode>3_32bit): Likewise.
4401         (vsx_ior<mode>3_64bit): Likewise.
4402         (vsx_xor<mode>3_32bit): Likewise.
4403         (vsx_xor<mode>3_64bit): Likewise.
4404         (vsx_one_cmpl<mode>2_32bit): Likewise.
4405         (vsx_one_cmpl<mode>2_64bit): Likewise.
4406         (vsx_nor<mode>3_32bit): Likewise.
4407         (vsx_nor<mode>3_64bit): Likewise.
4408         (vsx_andc<mode>3_32bit): Likewise.
4409         (vsx_andc<mode>3_64bit): Likewise.
4410         (vsx_eqv<mode>3_32bit): Likewise.
4411         (vsx_eqv<mode>3_64bit): Likewise.
4412         (vsx_nand<mode>3_32bit): Likewise.
4413         (vsx_nand<mode>3_64bit): Likewise.
4414         (vsx_orc<mode>3_32bit): Likewise.
4415         (vsx_orc<mode>3_64bit): Likewise.
4417         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4418         logical types in GPRs.
4420         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4421         logical insns to rs6000.md, and allow TImode operations in
4422         32-bit.
4423         (altivec_ior<mode>3): Likewise.
4424         (altivec_xor<mode>3): Likewise.
4425         (altivec_one_cmpl<mode>2): Likewise.
4426         (altivec_nor<mode>3): Likewise.
4427         (altivec_andc<mode>3): Likewise.
4429         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4430         attributes for moving the 128-bit logical operations into
4431         rs6000.md.
4432         (BOOL_REGS_OUTPUT): Likewise.
4433         (BOOL_REGS_OP1): Likewise.
4434         (BOOL_REGS_OP2): Likewise.
4435         (BOOL_REGS_UNARY): Likewise.
4436         (BOOL_REGS_AND_CR0): Likewise.
4437         (one_cmpl<mode>2): Add support for DI logical operations on
4438         32-bit, splitting the operations to 32-bit.
4439         (anddi3): Likewise.
4440         (iordi3): Likewise.
4441         (xordi3): Likewise.
4442         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4443         changes to combine the 32/64-bit code, allow logical operations on
4444         TI mode in 32-bit, and to use similar match_operator patterns like
4445         scalar mode uses.  Combine the Altivec and VSX code for logical
4446         operations, and move it here.
4447         (ior<mode>3, 128-bit types): Likewise.
4448         (xor<mode>3, 128-bit types): Likewise.
4449         (one_cmpl<mode>3, 128-bit types): Likewise.
4450         (nor<mode>3, 128-bit types): Likewise.
4451         (andc<mode>3, 128-bit types): Likewise.
4452         (eqv<mode>3, 128-bit types): Likewise.
4453         (nand<mode>3, 128-bit types): Likewise.
4454         (orc<mode>3, 128-bit types): Likewise.
4455         (and<mode>3_internal): Likewise.
4456         (bool<mode>3_internal): Likewise.
4457         (boolc<mode>3_internal1): Likewise.
4458         (boolc<mode>3_internal2): Likewise.
4459         (boolcc<mode>3_internal1): Likewise.
4460         (boolcc<mode>3_internal2): Likewise.
4461         (eqv<mode>3_internal1): Likewise.
4462         (eqv<mode>3_internal2): Likewise.
4463         (one_cmpl1<mode>3_internal): Likewise.
4465 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4467         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4468         Rename flag_stack_usage to flag_stack_usage_info.
4470 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4472         * config/microblaze/sync.md: New file.
4473         * config/microblaze/microblaze.md: Include sync.md
4474         * config/microblaze/microblaze.c: Add print_operand 'y'.
4475         * config/microblaze/constraints.md: Add memory_contraint
4476         'Q' which is a single register.
4478 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
4480         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4482 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4484         * reload.c (find_reloads): Exit loop once we find this operand
4485         cannot be reloaded somehow for this alternative.
4487         * reload.c (find_reloads): Exit loop once we find a hard register.
4489         * rtlanal.c (computed_jump_p): Exit loop once we find label
4490         reference is used.
4492         * i386.c (ix86_pad_returns): Exit loop after setting replace.
4494         * cfgloopmanip.c (remove_path): Exit loop after setting
4495         irred_invalidated.
4497         * gensupport.c (subst_dup): Avoid loop if code is not
4498         MATCH_DUP nor MATCH_OP_DUP.
4500 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
4502         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4504 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4506         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4507         true for SP_REGNUM if mode == ptr_mode.
4508         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4509         with value R0_REGNUM + 31.
4511 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4513         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4514         pad pointer-typed argument downward.
4516 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4518         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4519         and __ILP32__ when the ILP32 model is in use.
4521 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4523         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4524         (aarch64_load_symref_appropriately): In the case of
4525         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4526         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4527         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4528         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4529         if 'mode' doesn't equal to 'ptr_mode'.
4530         (aarch64_output_mi_thunk): Add an assertion on the alignment of
4531         'vcall_offset'; change to call aarch64_emit_move differently depending
4532         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4533         to calculate the upper bound of 'vcall_offset'.
4534         (aarch64_cannot_force_const_mem): Change to also return true if
4535         mode != ptr_mode.
4536         (aarch64_legitimize_reload_address): In the case of large
4537         displacements, add new local variable 'xmode' and an assertion
4538         based on it; change to use 'xmode' to generate the new rtx and
4539         reload.
4540         (aarch64_asm_trampoline_template): Change to generate the template
4541         differently depending on TARGET_ILP32 or not; change to use
4542         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4543         (aarch64_trampoline_size): Removed.
4544         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4545         and replace immediate literals with it.  Change to use 'ptr_mode'
4546         instead of 'DImode' and call convert_memory_address if the mode
4547         of 'fnaddr' doesn't equal to 'ptr_mode'.
4548         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4549         to output symbol.
4550         (aarch64_elf_asm_destructor): Likewise.
4551         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4552         on TARGET_ILP32 instead of aarch64_trampoline_size.
4553         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4554         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4555         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4556         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
4557         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4558         (storewb_pair<GPI:mode>_<P:mode>): ... this.
4559         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4560         depending on the value of 'mode'.
4561         (add_losym_<mode>): New.
4562         (ldr_got_small_<mode>): New, based on ldr_got_small.
4563         (ldr_got_small): Remove.
4564         (ldr_got_small_sidi): New.
4565         * config/aarch64/iterators.md (P): New.
4566         (PTR): Change to 'ptr_mode' in the condition.
4568 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4570         * config.gcc (aarch64*-*-*): Support --with-abi.
4571         (aarch64*-*-elf): Support --with-multilib-list.
4572         (aarch64*-*-linux*): Likewise.
4573         (supported_defaults): Add abi to aarch64*-*-*.
4574         * configure.ac: Mention AArch64 for --with-multilib-list.
4575         * configure: Re-generated.
4576         * config/aarch64/biarchilp32.h: New file.
4577         * config/aarch64/biarchlp64.h: New file.
4578         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4579         (ABI_SPEC): Ditto.
4580         (MULTILIB_DEFAULTS): Ditto.
4581         (DRIVER_SELF_SPECS): Ditto.
4582         (ASM_SPEC): Update to also substitute -mabi.
4583         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4584         file whose name depends on -mabi= and -mbig-endian.
4585         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4586         TARGET_ILP32.
4587         (POINTER_SIZE): New define.
4588         (POINTERS_EXTEND_UNSIGNED): Ditto.
4589         (enum aarch64_abi_type): New enumeration tag.
4590         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4591         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4592         (TARGET_ILP32): New define.
4593         * config/aarch64/aarch64.opt (mabi): New.
4594         (aarch64_abi): New.
4595         (ilp32, lp64): New values for -mabi.
4596         * config/aarch64/t-aarch64 (comma): New define.
4597         (MULTILIB_OPTIONS): Ditto.
4598         (MULTILIB_DIRNAMES): Ditto.
4599         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4600         * doc/invoke.texi: Document -mabi for AArch64.
4602 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
4604         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4606 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4607             Anton Blanchard  <anton@au1.ibm.com>
4609         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4610         endianness when selecting field to splat.
4612 2013-07-22  Eric Christopher  <echristo@gmail.com>
4614         * dwarf2out.c (die_odr_checksum): New function to use
4615         CHECKSUM_ macros and ULEB128 for DIE tag.
4616         (generate_type_signature): Use.
4618 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
4620         * config.gcc (sparc*-*-*): Accept leon3 processor.
4621         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4622         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4623         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4624         * config/sparc/sparc.opt (enum processor_type): Add leon3.
4625         (mfix-ut699): Adjust comment.
4626         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4627         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4628         (CPP_CPU_SPEC): Likewise.
4629         (ASM_CPU_SPEC): Likewise.
4630         * config/sparc/sparc.c (leon3_cost): New constant.
4631         (sparc_option_override): Add leon3 support.
4632         (mem_ref): New function.
4633         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4634         (sparc_do_work_around_errata): Look into the instruction in the delay
4635         slot and adjust accordingly.  Add fix for the data cache nullify issues
4636         of the UT699.  Change insertion position for the NOP.
4637         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4638         (leon3_load): New reservation.
4639         (leon_store): Bump latency to 2.
4640         (grfpu): New automaton.
4641         (grfpu_alu): New unit.
4642         (grfpu_ds): Likewise.
4643         (leon_fp_alu): Adjust.
4644         (leon_fp_mult): Delete.
4645         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4646         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4647         * config/sparc/sparc.md (cpu): Add leon3.
4648         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4649         (swapsi): Likewise.
4650         (atomic_test_and_set): Likewise.
4651         (ldstub): Likewise.
4653 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
4655         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
4656         default for R5900 targets.
4657         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4658         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4659         * config/mips/mips.c (mips_option_override): Report an error for
4660         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
4661         for -march=r5900 -mhard-float.
4663 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4665         * df-problems.c (can_move_insns_across): Exit loop once we
4666         find a non-fixed, non-global register.
4668         * ipa-pure-const.c (propagate_nothrow): Exit loop after
4669         setting can_throw.
4671         * omega.c (omega_eliminate_red): Break after setting red_found.
4672         (omega_problem_has_red_equations): Similarly after setting found.
4673         (omega_query_variable): Similarly after setting coupled.
4675 2013-07-22  Marek Polacek  <polacek@redhat.com>
4677         * gimplify.c: Don't include gimple.h twice.
4679 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4681         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4682         instead of TARGET_THUMB1.
4683         (Pz): New constraint.
4684         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4685         encodings.
4686         (compare_negsi_si): Likewise.
4687         (compare_addsi2_op0): Likewise.
4688         (compare_addsi2_op1): Likewise.
4689         (addsi3_carryin_<optab>): Likewise.
4690         (addsi3_carryin_alt2_<optab>): Likewise.
4691         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4692         for arm_restrict_it.
4693         (subsi3_carryin): Likewise.
4694         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4695         (minmax_arithsi): Disable for arm_restrict_it.
4696         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4697         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4698         (satsi_<SAT:code>_shift): Likewise.
4699         (arm_shiftsi3): Add alternative for 16-bit encoding.
4700         (arm32_movhf): Disable for arm_restrict_it.
4701         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4702         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4704 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
4706         * config/arm/arm.md (attribute "insn"): Delete.
4707         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4708         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4709         (not_shiftsi): Update for attribute change.
4710         (not_shiftsi_compare0): Likewise.
4711         (not_shiftsi_compare0_scratch): Likewise.
4712         (arm_one_cmplsi2): Likewise.
4713         (thumb1_one_cmplsi2): Likewise.
4714         (notsi_compare0): Likewise.
4715         (notsi_compare0_scratch): Likewise.
4716         (thumb1_movdi_insn): Likewise.
4717         (arm_movsi_insn): Likewise.
4718         (movhi_insn_arch4): Likewise.
4719         (movhi_bytes): Likewise.
4720         (arm_movqi_insn): Likewise.
4721         (thumb1_movqi_insn): Likewise.
4722         (arm32_movhf): Likewise.
4723         (thumb1_movhf): Likewise.
4724         (arm_movsf_soft_insn): Likewise.
4725         (thumb1_movsf_insn): Likewise.
4726         (thumb_movdf_insn): Likewise.
4727         (movsicc_insn): Likewise.
4728         (movsfcc_soft_insn): Likewise.
4729         (and_scc): Likewise.
4730         (cond_move): Likewise.
4731         (if_move_not): Likewise.
4732         (if_not_move): Likewise.
4733         (if_shift_move): Likewise.
4734         (if_move_shift): Likewise.
4735         (if_shift_shift): Likewise.
4736         (if_not_arith): Likewise.
4737         (if_arith_not): Likewise.
4738         (cond_move_not): Likewise.
4739         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4740         (neon_mov<mode>): Likewise.
4741         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4742         (thumb2_movsi_vfp): Likewise.
4743         (movsf_vfp): Likewise.
4744         (thumb2_movsf_vfp): Likewise.
4745         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4746         change.
4747         (cortexa7_older_only): Likewise.
4748         (cortexa7_younger): Likewise.
4749         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4750         (1020alu_shift_op): Likewise.
4751         (1020alu_shift_reg_op): Likewise.
4752         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4753         (alu_shift_op): Likewise.
4754         (alu_shift_reg_op): Likewise.
4755         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4756         (11_alu_shift_op): Likewise.
4757         (11_alu_shift_reg_op): Likewise.
4758         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4759         (9_alu_shift_reg_op): Likewise.
4760         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4761         change.
4762         (cortex_a15_alu_shift): Likewise.
4763         (cortex_a15_alu_shift_reg): Likewise.
4764         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4765         (cortex_a5_alu_shift): Likewise.
4766         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4767         change.
4768         (cortex_a53_alu_shift): Likewise.
4769         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4770         change.
4771         (cortex_a7_alu_reg): Likewise.
4772         (cortex_a7_alu_shift): Likewise.
4773         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4774         (cortex_a8_alu_shift): Likewise.
4775         (cortex_a8_alu_shift_reg): Likewise.
4776         (cortex_a8_mov): Likewise.
4777         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4778         (cortex_a9_dp_shift): Likewise.
4779         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4780         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4781         (cortex_r4_mov): Likewise.
4782         (cortex_r4_alu_shift): Likewise.
4783         (cortex_r4_alu_shift_reg): Likewise.
4784         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4785         (526_alu_shift_op): Likewise.
4786         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4787         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4788         (626te_alu_shift_op): Likewise.
4789         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4790         (726te_alu_op): Likewise.
4791         (726te_alu_shift_op): Likewise.
4792         (726te_alu_shift_reg_op): Likewise.
4793         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4794         (mp626_alu_shift_op): Likewise.
4795         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4796         (pj4_alu_e1_conds): Likewise.
4797         (pj4_alu): Likewise.
4798         (pj4_alu_conds): Likewise.
4799         (pj4_shift): Likewise.
4800         (pj4_shift_conds): Likewise.
4801         (pj4_alu_shift): Likewise.
4802         (pj4_alu_shift_conds): Likewise.
4804 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4806         * config/arm/predicates.md (shiftable_operator_strict_it):
4807         New predicate.
4808         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4809         Disable cond_exec version for arm_restrict_it.
4810         (thumb2_smaxsi3): Convert to generate cond_exec.
4811         (thumb2_sminsi3): Likewise.
4812         (thumb32_umaxsi3): Likewise.
4813         (thumb2_uminsi3): Likewise.
4814         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4815         (thumb2_neg_abssi2): Likewise.
4816         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4817         (thumb2_movsicc_insn): Adjust alternatives.
4818         (thumb2_mov_negscc): Disable for arm_restrict_it.
4819         (thumb2_mov_negscc_strict_it): New pattern.
4820         (thumb2_mov_notscc_strict_it): New pattern.
4821         (thumb2_mov_notscc): Disable for arm_restrict_it.
4822         (thumb2_ior_scc): Likewise.
4823         (thumb2_ior_scc_strict_it): New pattern.
4824         (thumb2_cond_move): Adjust for arm_restrict_it.
4825         (thumb2_cond_arith): Disable for arm_restrict_it.
4826         (thumb2_cond_arith_strict_it): New pattern.
4827         (thumb2_cond_sub): Adjust for arm_restrict_it.
4828         (thumb2_movcond): Likewise.
4829         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4830         (thumb2_zero_extendhisi2_v6): Likewise.
4831         (thumb2_zero_extendqisi2_v6): Likewise.
4832         (orsi_notsi_si): Likewise.
4833         (orsi_not_shiftsi_si): Likewise.
4835 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4837         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4838         instruction sequence is 1 byte shorter.
4840 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4842         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4843         it is not needed after split.
4845 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4847         PR target/51784
4848         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4849         second label for nonlocal goto receivers. Don't output pic base labels
4850         unless we're producing PIC; mark that action unreachable().
4851         (ix86_save_reg): If the function contains a nonlocal label, save the
4852         PIC base reg.
4853         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4854         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4855         (update_pic_label_number_if_needed): New.
4856         (machopic_output_function_base_name): Adjust for nonlocal receiver
4857         case.
4858         (machopic_should_output_picbase_label): New.
4859         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4860         (nonlocal_goto_receiver): New insn and split.
4862 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4864         * config/aarch64/aarch64-builtins.c
4865         (aarch64_fold_builtin): Fold abs in all modes.
4866         * config/aarch64/aarch64-simd-builtins.def
4867         (abs): Enable for all modes.
4868         * config/aarch64/arm_neon.h
4869         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4870         (vabs_f64): Add missing intrinsic.
4872 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4874         * config/aarch64/arm_neon.h (vabs_s64): New function
4876 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4878         PR target/57516
4879         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4880         * config/avr/avr.md (adjust_len): Add `round'.
4881         * config/avr/avr-protos.h (avr_out_round): New prototype.
4882         (avr_out_plus): Add `out_label' argument.
4883         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4884         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4885         Handle the case where `insn' is just a pattern.
4886         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4887         (avr_out_round): New function.
4888         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4890 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
4892         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4893         Add check for flag_stack_usage to handle -fstack-usage support
4895 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
4897         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4898         interaction for new Power8 flags and VSX.
4900 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4902         PR middle-end/57698
4903         * tree-inline.c (expand_call_inline): Emit errors during
4904         early_inlining only if optimization is not turned on.
4906 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4908         * passes.def: New.
4910         * passes.c (init_optimization_passes): Move the construction of
4911         the pass hierarchy into a new passes.def file.
4913         * Makefile.in (passes.o): Add dependency on passes.def.
4915 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4917         * passes.c (init_optimization_passes): Introduce macros for
4918         constructing the tree of passes (INSERT_PASSES_AFTER,
4919         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4920         TERMINATE_PASS_LIST).
4922 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4923             Wei Mi  <wmi@google.com>
4925         PR rtl-optimization/57878
4926         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4927         top.
4928         (reload_pseudo_compare_func): Check nregs first for reload
4929         pseudos.
4931 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4933         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4935 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4937         * read-rtl.c (validate_const_int): Once an invalid character is
4938         seen, quit the loop.
4940         * gengtype.c (write_roots): Similarly once we find the "deletable"
4941         or "if_marked" option.
4943 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4945         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4946         "xtab" and "sat".  Move value "clz" from here to ...
4947         (attriubte "type"): ... here.
4948         (satsi_<SAT:code>): Delete "insn" attribute.
4949         (satsi_<SAT:code>_shift): Likewise.
4950         (arm_zero_extendqisi2addsi): Likewise.
4951         (arm_extendqisi2addsi): Likewise.
4952         (clzsi2): Update for attribute changes.
4953         (rbitsi2): Likewise.
4954         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4955         attribute.
4956         (arm_usatsihi): Likewise.
4957         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4959 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4961         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4962         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4963         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4964         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4965         in alphabetical order.
4966         (attribute "core_cycles"): Update for attribute changes.
4967         (arm_addsi3): Likewise.
4968         (addsi3_compare0): Likewise.
4969         (addsi3_compare0_scratch): Likewise.
4970         (addsi3_compare_op1): Likewise.
4971         (addsi3_compare_op2): Likewise.
4972         (compare_addsi2_op0): Likewise.
4973         (compare_addsi2_op1): Likewise.
4974         (addsi3_carryin_shift_<optab>): Likewise.
4975         (subsi3_carryin_shift): Likewise.
4976         (rsbsi3_carryin_shift): Likewise.
4977         (arm_subsi3_insn): Likewise.
4978         (subsi3_compare0): Likewise.
4979         (subsi3_compare): Likewise.
4980         (arm_andsi3_insn): Likewise.
4981         (thumb1_andsi3_insn): Likewise.
4982         (andsi3_compare0): Likewise.
4983         (andsi3_compare0_scratch): Likewise.
4984         (zeroextractsi_compare0_scratch
4985         (andsi_not_shiftsi_si): Likewise.
4986         (iorsi3_insn): Likewise.
4987         (iorsi3_compare0): Likewise.
4988         (iorsi3_compare0_scratch): Likewise.
4989         (arm_xorsi3): Likewise.
4990         (thumb1_xorsi3_insn): Likewise.
4991         (xorsi3_compare0): Likewise.
4992         (xorsi3_compare0_scratch): Likewise.
4993         (satsi_<SAT:code>_shift): Likewise.
4994         (rrx): Likewise.
4995         (arm_shiftsi3): Likewise.
4996         (shiftsi3_compare0): Likewise.
4997         (not_shiftsi): Likewise.
4998         (not_shiftsi_compare0): Likewise.
4999         (not_shiftsi_compare0_scratch): Likewise.
5000         (arm_one_cmplsi2): Likewise.
5001         (thumb_one_complsi2): Likewise.
5002         (notsi_compare0): Likewise.
5003         (notsi_compare0_scratch): Likewise.
5004         (thumb1_zero_extendhisi2): Likewise.
5005         (arm_zero_extendhisi2): Likewise.
5006         (arm_zero_extendhisi2_v6): Likewise.
5007         (arm_zero_extendhisi2addsi): Likewise.
5008         (thumb1_zero_extendqisi2): Likewise.
5009         (thumb1_zero_extendqisi2_v6): Likewise.
5010         (arm_zero_extendqisi2): Likewise.
5011         (arm_zero_extendqisi2_v6): Likewise.
5012         (arm_zero_extendqisi2addsi): Likewise.
5013         (thumb1_extendhisi2): Likewise.
5014         (arm_extendhisi2): Likewise.
5015         (arm_extendhisi2_v6): Likewise.
5016         (arm_extendqisi): Likewise.
5017         (arm_extendqisi_v6): Likewise.
5018         (arm_extendqisi2addsi): Likewise.
5019         (thumb1_extendqisi2): Likewise.
5020         (thumb1_movdi_insn): Likewise.
5021         (arm_movsi_insn): Likewise.
5022         (movsi_compare0): Likewise.
5023         (movhi_insn_arch4): Likewise.
5024         (movhi_bytes): Likewise.
5025         (arm_movqi_insn): Likewise.
5026         (thumb1_movqi_insn): Likewise.
5027         (arm32_movhf): Likewise.
5028         (thumb1_movhf): Likewise.
5029         (arm_movsf_soft_insn): Likewise.
5030         (thumb1_movsf_insn): Likewise.
5031         (movdf_soft_insn): Likewise.
5032         (thumb_movdf_insn): Likewise.
5033         (arm_cmpsi_insn): Likewise.
5034         (cmpsi_shiftsi): Likewise.
5035         (cmpsi_shiftsi_swp): Likewise.
5036         (arm_cmpsi_negshiftsi_si): Likewise.
5037         (movsicc_insn): Likewise.
5038         (movsfcc_soft_insn): Likewise.
5039         (arith_shiftsi): Likewise.
5040         (arith_shiftsi_compare0
5041         (arith_shiftsi_compare0_scratch
5042         (sub_shiftsi): Likewise.
5043         (sub_shiftsi_compare0
5044         (sub_shiftsi_compare0_scratch
5045         (and_scc): Likewise.
5046         (cond_move): Likewise.
5047         (if_plus_move): Likewise.
5048         (if_move_plus): Likewise.
5049         (if_move_not): Likewise.
5050         (if_not_move): Likewise.
5051         (if_shift_move): Likewise.
5052         (if_move_shift): Likewise.
5053         (if_shift_shift): Likewise.
5054         (if_not_arith): Likewise.
5055         (if_arith_not): Likewise.
5056         (cond_move_not): Likewise.
5057         (thumb1_ashlsi3): Set type attribute.
5058         (thumb1_ashrsi3): Likewise.
5059         (thumb1_lshrsi3): Likewise.
5060         (thumb1_rotrsi3): Likewise.
5061         (shiftsi3_compare0_scratch): Likewise.
5062         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5063         (neon_mov<mode>): Likewise.
5064         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5065         attribute changes.
5066         (thumb2_movsi_insn): Likewise.
5067         (thumb2_cmpsi_neg_shiftsi): Likewise.
5068         (thumb2_extendqisi_v6): Likewise.
5069         (thumb2_zero_extendhisi2_v6): Likewise.
5070         (thumb2_zero_extendqisi2_v6): Likewise.
5071         (thumb2_shiftsi3_short): Likewise.
5072         (thumb2_addsi3_compare0_scratch): Likewise.
5073         (orsi_not_shiftsi_si): Likewise.
5074         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5075         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5076         changes.
5077         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5078         (1020alu_shift_op): Likewise.
5079         (1020alu_shift_reg_op): Likewise.
5080         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5081         (alu_shift_op): Likewise.
5082         (alu_shift_reg_op): Likewise.
5083         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5084         (11_alu_shift_op): Likewise.
5085         (11_alu_shift_reg_op): Likewise.
5086         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5087         (9_alu_shift_reg_op): Likewise.
5088         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5089         changes.
5090         (cortex_a15_alu_shift): Likewise.
5091         (cortex_a15_alu_shift_reg): Likewise.
5092         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5093         changes.
5094         (cortex_a5_alu_shift): Likewise.
5095         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5096         changes.
5097         (cortex_a53_alu_shift): Likewise.
5098         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5099         changes.
5100         (cortex_a7_alu_reg): Likewise.
5101         (cortex_a7_alu_shift): Likewise.
5102         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5103         changes.
5104         (cortex_a8_alu_shift): Likewise.
5105         (cortex_a8_alu_shift_reg): Likewise.
5106         (cortex_a8_mov): Likewise.
5107         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5108         (cortex_a9_dp_shift): Likewise.
5109         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5110         changes.
5111         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5112         changes.
5113         (cortex_r4_mov): Likewise.
5114         (cortex_r4_alu_shift): Likewise.
5115         (cortex_r4_alu_shift_reg): Likewise.
5116         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5117         (526_alu_shift_op): Likewise.
5118         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5119         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5120         (626te_alu_shift_op): Likewise.
5121         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5122         (726te_alu_op): Likewise.
5123         (726te_alu_shift_op): Likewise.
5124         (726te_alu_shift_reg_op): Likewise.
5125         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5126         (mp626_alu_shift_op): Likewise.
5127         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5128         (pj4_alu_e1_conds): Likewise.
5129         (pj4_alu): Likewise.
5130         (pj4_alu_conds): Likewise.
5131         (pj4_shift): Likewise.
5132         (pj4_shift_conds): Likewise.
5133         (pj4_alu_shift): Likewise.
5134         (pj4_alu_shift_conds): Likewise.
5135         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5136         changes.
5137         (cortexa7_older_only): Likewise.
5138         (cortexa7_younger): Likewise.
5140 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5142         * ipa-pure-const.c (generate_summary): Rename to...
5143         (pure_const_generate_summary): ... this.
5145 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
5147         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5149 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
5151         PR target/57909
5152         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5153         usage in HI mode.
5155 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5157         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5158         enabled without -march=zEC12.
5159         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5160         flags to be set.
5162 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5164         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5165         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5166         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5167         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5168         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5169         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5170         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5171         ISA_HAS_FP4.
5172         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5173         and ISA_HAS_NMADD3_NMSUB3.
5174         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5175         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5176         (nmsub4<mode>, nmsub3<mode>): Likewise.
5177         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5179 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5181         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5182         TARGET_MIPS5400 checking.
5184 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
5185             Peter Bergner  <bergner@vnet.ibm.com>
5187         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5188         registers in the comment.
5189         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5190         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5191         rather than FIRST_PSEUDO_REGISTERS.
5193 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
5195         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5196         enable extra ISA flags with TARGET_HTM.
5198 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5200         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5201         Fix comment typos.
5203 2013-07-15  Cong Hou  <congh@google.com>
5205         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5206         in compare function for sorting.
5208 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
5210         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5211         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5212         * config/rs6000/rs6000.opt: Add -mhtm option.
5213         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5214         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5215         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5216         __HTM__ if the HTM instructions are available.
5217         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5218         htm_spr_reg_operand): New define_predicates.
5219         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5220         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5221         Include htm.md.
5222         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5223         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5224         HTM builtin functions.
5225         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5226         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5227         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5228         (rs6000_builtin_mask_calculate): Likewise.
5229         (rs6000_option_override_internal): Likewise.
5230         (bdesc_htm): Add new HTM builtin support.
5231         (htm_spr_num): New function.
5232         (htm_spr_regno): Likewise.
5233         (rs6000_htm_spr_icode): Likewise.
5234         (htm_expand_builtin): Likewise.
5235         (htm_init_builtins): Likewise.
5236         (rs6000_expand_builtin): Add support for HTM builtin functions.
5237         (rs6000_init_builtins): Likewise.
5238         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5239         option.
5240         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5241         (TARGET_HTM, MASK_HTM): Define macros.
5242         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5243         (FIXED_REGISTERS): Likewise.
5244         (CALL_USED_REGISTERS): Likewise.
5245         (CALL_REALLY_USED_REGISTERS): Likewise.
5246         (REG_ALLOC_ORDER): Likewise.
5247         (enum reg_class): Likewise.
5248         (REG_CLASS_NAMES): Likewise.
5249         (REG_CLASS_CONTENTS): Likewise.
5250         (REGISTER_NAMES): Likewise.
5251         (ADDITIONAL_REGISTER_NAMES): Likewise.
5252         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5253         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5254         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5255         * config/rs6000/htm.md: New file.
5256         * config/rs6000/htmintrin.h: New file.
5257         * config/rs6000/htmxlintrin.h: New file.
5259 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5261         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5262         Define SYMBOL_TINY_GOT, update comment.
5263         * config/aarch64/aarch64.c
5264         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5265         (aarch64_expand_mov_immediate): Likewise.
5266         (aarch64_print_operand): Likewise.
5267         (aarch64_classify_symbol): Likewise.
5268         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5269         (ldr_got_tiny): Define.
5271 2013-07-13  Tobias Grosser  <tobias@grosser.es>
5273         PR tree-optimization/54094
5274         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5275           scheduling dimension for the parallelism check from the polyhedral
5276           information in the AST.
5277         * graphite-dependences.c (carries_deps): Do not assume the schedule is
5278           in 2D + 1 form.
5280 2013-07-13  Jason Merrill  <jason@redhat.com>
5282         * print-tree.c (debug_vec_tree): Use debug_raw.
5283         (debug_raw (vec<tree, va_gc> &)): New.
5284         (debug_raw (vec<tree, va_gc> *)): New.
5285         * tree.h: Declare them.
5287 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
5289         * ifcvt.c (ifcvt_after_combine): New static variable.
5290         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5291         for size.
5292         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
5293         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5294         rest_of_handle_if_after_reload): Pass new argument for if_convert.
5296 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
5298         * config/mips/mips.c (mips_expand_call): Remove empty statement.
5300 2013-07-12  Michael Matz  <matz@suse.de>
5302         PR middle-end/55771
5303         * convert.c (convert_to_real): Reject non-float inner types.
5305 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
5307         * config/aarch64/aarch64-protos.h
5308         (aarch64_simd_immediate_valid_for_move): Remove.
5309         * config/aarch64/aarch64.c (simd_immediate_info): New member.
5310         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5311         cases.
5312         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5314 2013-07-11  Steve Ellcey  <sellcey@mips.com>
5316         * config/mips/mips.c (mips_conditional_register_usage): Do not
5317         use t[0-7] registers in MIPS16 mode when optimizing for size.
5319 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
5321         * config/i386/i386.c (dispatch_function_versions): Fix array
5322         indexing of function_version_info to match actual_versions.
5324 2013-07-11  Teresa Johnson  <tejohnson@google.com>
5326         * vec.h (struct va_gc): Move release out-of-line.
5327         (va_gc::release): Call ggc_free on released vec.
5329 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5331         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5332         Require GOT register as additional operand in UNSPEC.
5333         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5334         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5335         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5336         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5337         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5338         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5339         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5341 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
5343         PR target/57631
5344         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5345         name seen by assembler/linker rather if available.
5347 2013-07-11  Andreas Schwab  <schwab@suse.de>
5349         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5351 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
5353         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5355 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
5357         * doc/tm.texi.in: Move hook documentation to ....
5358         * target.def: ... here.
5360         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5361         text on @hook line.
5362         * doc/tm.texi: Regenerate.
5364 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
5366         PR c++/57869
5367         * doc/invoke.texi: Document -Wconditionally-supported.
5369 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5371         PR target/57844
5372         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5373         of my_fp.
5375 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5377         PR target/57506
5378         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5379         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5380         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5381         Remove duplicate devices.
5382         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5383         * config/avr/t-multilib: Regenerate.
5384         * config/avr/avr-tables.opt: Regenerate.
5385         * doc/avr-mmcu.texi: Regenerate.
5387 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5389         PR target/56987
5390         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5392 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
5394         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5395         the cost of MULT when optimizing for size.
5397 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5399         * config/cr16/cr16-protos.h: Don't include target.h.
5401 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
5403         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5404         adjust register size for TDmode and TFmode for VSX registers.
5406 2013-07-08  Kai Tietz  <ktietz@redhat.com>
5408         PR target/56892
5409         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5410         hook_bool_const_tree_true.
5412 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5414         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5415         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5416         * config/s390/s390.md: Define FPR*_REGNUM constants.
5417         Fix FPR2_REGNUM constant (18 -> 17).
5418         ("*trunc<BFP:mode><DFP_ALL:mode>2")
5419         ("*trunc<DFP_ALL:mode><BFP:mode>2")
5420         ("trunc<BFP:mode><DFP_ALL:mode>2")
5421         ("trunc<DFP_ALL:mode><BFP:mode>2")
5422         ("*extend<BFP:mode><DFP_ALL:mode>2")
5423         ("*extend<DFP_ALL:mode><BFP:mode>2")
5424         ("extend<BFP:mode><DFP_ALL:mode>2")
5425         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5426         FPR4_REGNUM.
5428 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
5430         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5432 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5434         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5435         and cfun_fpr_bit_p to cfun_fpr_save_p.
5436         (s390_frame_area, s390_register_info, s390_frame_info)
5437         (s390_emit_prologue, s390_emit_epilogue)
5438         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5439         register numbers.
5440         * config/s390/s390.h: Define *_REGNUM macros for floating point
5441         register numbers.
5443 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
5445         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5447 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
5449         PR rtl-optimization/57786
5450         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5451         and break out of the loop when it is set to false.
5453 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
5455         PR target/57819
5456         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5457         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5458         (const_int 63)) 0)).
5459         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5460         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5461         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5463         PR rtl-optimization/57829
5464         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5465         mask bits outside of mode are just sign-extension from mode to HWI.
5467 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5469         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5470         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5471         adjust_address instead of change_address to keep info about alignment.
5472         (emit_strmov): Remove.
5473         (emit_memmov): New function.
5474         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5475         (expand_movmem_epilogue): Likewise and return updated rtx for
5476         destination.
5477         (expand_constant_movmem_prologue): Likewise and return updated rtx for
5478         destination and source.
5479         (decide_alignment): Refactor, handle vector_loop.
5480         (ix86_expand_movmem): Likewise.
5481         (ix86_expand_setmem): Likewise.
5482         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5484 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
5486         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5487         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5489 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5491         * config/i386/sse.md (sse_movlhps): Change alternative 3
5492         of operand 2 to "m".
5494 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5496         PR target/57807
5497         * config/i386/sse.md (iptr): New mode attribute.
5498         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5499         (<sse>_vm<plusminus_insn><mode>3): Ditto.
5500         (<sse>_vmmul<mode>3): Ditto.
5501         (<sse>_vmdiv<mode>3): Ditto.
5502         (sse_vmrcpv4sf2): Ditto.
5503         (<sse>_vmsqrt<mode>2): Ditto.
5504         (sse_vmrsqrtv4sf2): Ditto.
5505         (<sse>_vm<code><mode>3): Ditto.
5506         (avx_vmcmp<mode>3): Ditto.
5507         (<sse>_vmmaskcmp<mode>3): Ditto.
5508         (<sse>_comi): Ditto.
5509         (<sse>_ucomi): Ditto.
5510         (*xop_vmfrcz_<mode>): Ditto.
5511         (*fmai_fmadd_<mode>): Ditto.
5512         (*fmai_fmsub_<mode>): Ditto.
5513         (*fmai_fnmadd_<mode>): Ditto.
5514         (*fmai_fnmsub_<mode>): Ditto.
5515         (*fma4i_vmfmadd_<mode>): Ditto.
5516         (*fma4i_vmfmsub_<mode>): Ditto.
5517         (*fma4i_vmfnmadd_<mode>): Ditto.
5518         (*fma4i_vmfnmsub_<mode>): Ditto.
5519         (*xop_vmfrcz_<mode>): Ditto.
5520         (sse_cvtps2pi): Ditto.
5521         (sse_cvttps2pi): Ditto.
5522         (sse_cvtss2si): Ditto.
5523         (sse_cvtss2si_2): Ditto.
5524         (sse_cvtss2siq_2): Ditto.
5525         (sse_cvttss2si): Ditto.
5526         (sse_cvttss2siq): Ditto.
5527         (sse_cvtsd2si): Ditto.
5528         (sse_cvtsd2si_2): Ditto.
5529         (sse_cvtsd2siq_2): Ditto.
5530         (sse_cvttsd2si): Ditto.
5531         (sse_cvttsd2siq): Ditto.
5532         (sse_cvtsd2ss): Ditto.
5533         (sse_cvtss2sd): Ditto.
5534         (avx2_pbroadcast<mode>): Ditto.
5535         (avx2_pbroadcast<mode>_1): Ditto.
5536         (*avx_vperm_broadcast_v4sf): Ditto.
5538         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5539         (sse_movlhps): Ditto.
5540         (sse_storehps): Ditto.
5541         (sse_loadhps): Ditto.
5542         (sse_storelps): Ditto.
5543         (sse_loadlps): Ditto.
5544         (*vec_concatv4sf): Ditto.
5545         (*vec_interleave_highv2df): Ditto.
5546         (*vec_interleave_lowv2df): Ditto.
5547         (*vec_extractv2df_1_sse): Ditto.
5548         (*vec_extractv2df_0_sse): Ditto.
5549         (sse2_storelpd): Ditto.
5550         (sse2_loadlpd): Ditto.
5551         (sse2_movsd): Ditto.
5552         (*vec_concatv4si): Ditto.
5553         (vec_concatv2di): Ditto.
5555         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5556         for Intel asm dialect.
5557         (mmx_punpcklwd): Ditto.
5558         (mmx_punpckldq): Ditto.
5560         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5561         for intel assembler dialect.
5563 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
5565         PR target/29776
5566         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5567         for BUILT_IN_C{LZ,LRSB}*.
5568         * tree.h (CASE_INT_FN): Add FN##IMAX case.
5569         * tree-vrp.c (extract_range_basic): Handle
5570         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
5571         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5572         fall thru to code calling set_value*.
5573         * builtins.c (expand_builtin): Remove *IMAX cases.
5574         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5575         if width is bigger than 2*HWI.
5577 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
5579         PR rtl-optimization/55342
5580         * lra-int.h (lra_subreg_reload_pseudos): New.
5581         * lra.c: Add undoing optional reloads to the block diagram.
5582         (lra_subreg_reload_pseudos): New.
5583         (lra_optional_reload_pseudos): Change comments.
5584         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
5585         lra_optional_reload_pseudos after undo transformations.
5586         * lra-assigns.c (pseudo_prefix_title): New.
5587         (lra_setup_reg_renumber): Use it.
5588         (spill_for): Ditto.  Check subreg reload pseudos too.
5589         (assign_by_spills): Consider subreg reload pseudos too.
5590         * lra-constraints.c (simplify_operand_subreg): Use
5591         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5592         (curr_insn_transform): Recognize and do optional reloads.
5593         (undo_optional_reloads): New.
5594         (lra_undo_inheritance): Call undo_optional_reloads.
5596 2013-07-05  Thomas Quinot  <quinot@adacore.com>
5598         * tree-complex.c (expand_complex_operations_1): Fix typo.
5600 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
5602         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5603         (tune_params): New member 'const vec_costs'.
5604         * config/aarch64/aarch64.c (generic_vector_cost): New.
5605         (generic_tunings): New member 'generic_vector_cost'.
5606         (aarch64_builtin_vectorization_cost): New.
5607         (aarch64_add_stmt_cost): New.
5608         (TARGET_VECTORIZE_ADD_STMT_COST): New.
5609         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5611 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
5613         PR target/57777
5614         * config/i386/predicates.md (vsib_address_operand): Disallow
5615         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5617 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
5619         PR middle-end/55030
5620         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5621         expand_builtin_setjmp_receiver.
5622         (expand_label): Adjust, call expand_builtin_setjmp_receiver
5623         with NULL for the label parameter.
5624         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5625         the frame-pointer.  Adjust comments.
5626         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5627         only if LABEL is non-NULL.
5629 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
5631         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5632         (ARM_ABI_AAPCS64): Ditto.
5633         (arm_abi): Ditto.
5634         (ARM_DEFAULT_ABI): Ditto.
5636 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
5638         * config/aarch64/aarch64-builtins.c
5639         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5640         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5641         (st1): Likewise.
5642         * config/aarch64/aarch64-simd.md
5643         (aarch64_ld1<VALL:mode>): New.
5644         (aarch64_st1<VALL:mode>): Likewise.
5645         * config/aarch64/arm_neon.h
5646         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5648 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
5650         * config/i386/i386.c (gate_insert_vzeroupper): Check if
5651         target ISA is AVX.
5652         (ix86_option_override_internal):Turn on all -mavx target flags by
5653         default as they are dependent on AVX anyway.
5655 2013-07-02  Cary Coutant  <ccoutant@google.com>
5657         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5658         deterministic hash.
5659         (loc_checksum_ordered): Likewise.
5660         (hash_loc_operands): Remove inline keyword.
5662 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
5664         PR tree-optimization/57741
5665         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5666         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5667         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5668         Allow REAL_CST step_exprs if flag_associative_math.
5669         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5671 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5673         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5675 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5677         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5679 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5681         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5682         encoding.
5683         (iorsi3_insn): Likewise.
5684         (arm_xorsi3): Likewise.
5686 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
5688         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
5689         to ...
5690         (attribute "type"): ... here, and prefix with "wmmx_".
5691         (attribute "core_cycles"): Update for attribute changes.
5692         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5693         (tbcstv4hi): Likewise.
5694         (tbcstv2si): Likewise.
5695         (iwmmxt_iordi3): Likewise.
5696         (iwmmxt_xordi3): Likewise.
5697         (iwmmxt_anddi3): Likewise.
5698         (iwmmxt_nanddi3): Likewise.
5699         (iwmmxt_arm_movdi): Likewise.
5700         (iwmmxt_movsi_insn): Likewise.
5701         (mov<mode>_internal): Likewise.
5702         (and<mode>3_iwmmxt): Likewise.
5703         (ior<mode>3_iwmmxt): Likewise.
5704         (xor<mode>3_iwmmxt): Likewise.
5705         (add<mode>3_iwmmxt): Likewise.
5706         (ssaddv8qi3): Likewise.
5707         (ssaddv4hi3): Likewise.
5708         (ssaddv2si3): Likewise.
5709         (usaddv8qi3): Likewise.
5710         (usaddv4hi3): Likewise.
5711         (usaddv2si3): Likewise.
5712         (sub<mode>3_iwmmxt): Likewise.
5713         (sssubv8qi3): Likewise.
5714         (sssubv4hi3): Likewise.
5715         (sssubv2si3): Likewise.
5716         (ussubv8qi3): Likewise.
5717         (ussubv4hi3): Likewise.
5718         (ussubv2si3): Likewise.
5719         (mulv4hi3_iwmmxt): Likewise.
5720         (smulv4hi3_highpart): Likewise.
5721         (umulv4hi3_highpart): Likewise.
5722         (iwmmxt_wmacs): Likewise.
5723         (iwmmxt_wmacsz): Likewise.
5724         (iwmmxt_wmacu): Likewise.
5725         (iwmmxt_wmacuz): Likewise.
5726         (iwmmxt_clrdi): Likewise.
5727         (iwmmxt_clrv8qi): Likewise.
5728         (iwmmxt_clr4hi): Likewise.
5729         (iwmmxt_clr2si): Likewise.
5730         (iwmmxt_uavgrndv8qi3): Likewise.
5731         (iwmmxt_uavgrndv4hi3): Likewise.
5732         (iwmmxt_uavgv8qi3): Likewise.
5733         (iwmmxt_uavgv4hi3): Likewise.
5734         (iwmmxt_tinsrb): Likewise.
5735         (iwmmxt_tinsrh): Likewise.
5736         (iwmmxt_tinsrw): Likewise.
5737         (iwmmxt_textrmub): Likewise.
5738         (iwmmxt_textrmsb): Likewise.
5739         (iwmmxt_textrmuh): Likewise.
5740         (iwmmxt_textrmsh): Likewise.
5741         (iwmmxt_textrmw): Likewise.
5742         (iwmxxt_wshufh): Likewise.
5743         (eqv8qi3): Likewise.
5744         (eqv4hi3): Likewise.
5745         (eqv2si3): Likewise.
5746         (gtuv8qi3): Likewise.
5747         (gtuv4hi3): Likewise.
5748         (gtuv2si3): Likewise.
5749         (gtv8qi3): Likewise.
5750         (gtv4hi3): Likewise.
5751         (gtv2si3): Likewise.
5752         (smax<mode>3_iwmmxt): Likewise.
5753         (umax<mode>3_iwmmxt): Likewise.
5754         (smin<mode>3_iwmmxt): Likewise.
5755         (umin<mode>3_iwmmxt): Likewise.
5756         (iwmmxt_wpackhss): Likewise.
5757         (iwmmxt_wpackwss): Likewise.
5758         (iwmmxt_wpackdss): Likewise.
5759         (iwmmxt_wpackhus): Likewise.
5760         (iwmmxt_wpackwus): Likewise.
5761         (iwmmxt_wpackdus): Likewise.
5762         (iwmmxt_wunpckihb): Likewise.
5763         (iwmmxt_wunpckihh): Likewise.
5764         (iwmmxt_wunpckihw): Likewise.
5765         (iwmmxt_wunpckilb): Likewise.
5766         (iwmmxt_wunpckilh): Likewise.
5767         (iwmmxt_wunpckilw): Likewise.
5768         (iwmmxt_wunpckehub): Likewise.
5769         (iwmmxt_wunpckehuh): Likewise.
5770         (iwmmxt_wunpckehuw): Likewise.
5771         (iwmmxt_wunpckehsb): Likewise.
5772         (iwmmxt_wunpckehsh): Likewise.
5773         (iwmmxt_wunpckehsw): Likewise.
5774         (iwmmxt_wunpckelub): Likewise.
5775         (iwmmxt_wunpckeluh): Likewise.
5776         (iwmmxt_wunpckeluw): Likewise.
5777         (iwmmxt_wunpckelsb): Likewise.
5778         (iwmmxt_wunpckelsh): Likewise.
5779         (iwmmxt_wunpckelsw): Likewise.
5780         (ror<mode>3): Likewise.
5781         (ashr<mode>3_iwmmxt): Likewise.
5782         (lshr<mode>3_iwmmxt): Likewise.
5783         (ashl<mode>3_iwmmxt): Likewise.
5784         (ror<mode>3_di): Likewise.
5785         (ashr<mode>3_di): Likewise.
5786         (lshr<mode>3_di): Likewise.
5787         (ashl<mode>3_di): Likewise.
5788         (iwmmxt_wmadds): Likewise.
5789         (iwmmxt_wmaddu): Likewise.
5790         (iwmmxt_tmia): Likewise.
5791         (iwmmxt_tmiaph): Likewise.
5792         (iwmmxt_tmiabb): Likewise.
5793         (iwmmxt_tmiatb): Likewise.
5794         (iwmmxt_tmiabt): Likewise.
5795         (iwmmxt_tmiatt): Likewise.
5796         (iwmmxt_tmovmskb): Likewise.
5797         (iwmmxt_tmovmskh): Likewise.
5798         (iwmmxt_tmovmskw): Likewise.
5799         (iwmmxt_waccb): Likewise.
5800         (iwmmxt_wacch): Likewise.
5801         (iwmmxt_waccw): Likewise.
5802         (iwmmxt_waligni): Likewise.
5803         (iwmmxt_walignr): Likewise.
5804         (iwmmxt_walignr0): Likewise.
5805         (iwmmxt_walignr1): Likewise.
5806         (iwmmxt_walignr2): Likewise.
5807         (iwmmxt_walignr3): Likewise.
5808         (iwmmxt_wsadb): Likewise.
5809         (iwmmxt_wsadh): Likewise.
5810         (iwmmxt_wsadbz): Likewise.
5811         (iwmmxt_wsadhz): Likewise.
5812         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5813         (iwmmxt_wabsdiffb): Likewise.
5814         (iwmmxt_wabsdiffh): Likewise.
5815         (iwmmxt_wabsdiffw): Likewise.
5816         (iwmmxt_waddsubhx): Likewise
5817         (iwmmxt_wsubaddhx): Likewise.
5818         (addc<mode>3): Likewise.
5819         (iwmmxt_avg4): Likewise.
5820         (iwmmxt_avg4r): Likewise.
5821         (iwmmxt_wmaddsx): Likewise.
5822         (iwmmxt_wmaddux): Likewise.
5823         (iwmmxt_wmaddsn): Likewise.
5824         (iwmmxt_wmaddun): Likewise.
5825         (iwmmxt_wmulwsm): Likewise.
5826         (iwmmxt_wmulwum): Likewise.
5827         (iwmmxt_wmulsmr): Likewise.
5828         (iwmmxt_wmulumr): Likewise.
5829         (iwmmxt_wmulwsmr): Likewise.
5830         (iwmmxt_wmulwumr): Likewise.
5831         (iwmmxt_wmulwl): Likewise.
5832         (iwmmxt_wqmulm): Likewise.
5833         (iwmmxt_wqmulwm): Likewise.
5834         (iwmmxt_wqmulmr): Likewise.
5835         (iwmmxt_wqmulwmr): Likewise.
5836         (iwmmxt_waddbhusm): Likewise.
5837         (iwmmxt_waddbhusl): Likewise.
5838         (iwmmxt_wqmiabb): Likewise.
5839         (iwmmxt_wqmiabt): Likewise.
5840         (iwmmxt_wqmiatb): Likewise.
5841         (iwmmxt_wqmiatt): Likewise.
5842         (iwmmxt_wqmiabbn): Likewise.
5843         (iwmmxt_wqmiabtn): Likewise.
5844         (iwmmxt_wqmiatbn): Likewise.
5845         (iwmmxt_wqmiattn): Likewise.
5846         (iwmmxt_wmiabb): Likewise.
5847         (iwmmxt_wmiabt): Likewise.
5848         (iwmmxt_wmiatb): Likewise.
5849         (iwmmxt_wmiatt): Likewise.
5850         (iwmmxt_wmiabbn): Likewise.
5851         (iwmmxt_wmiabtn): Likewise.
5852         (iwmmxt_wmiatbn): Likewise.
5853         (iwmmxt_wmiattn): Likewise.
5854         (iwmmxt_wmiawbb): Likewise.
5855         (iwmmxt_wmiawbt): Likewise.
5856         (iwmmxt_wmiawtb): Likewise.
5857         (iwmmxt_wmiawtt): Likewise.
5858         (iwmmxt_wmiawbbn): Likewise.
5859         (iwmmxt_wmiawbtn): Likewise.
5860         (iwmmxt_wmiawtbn): Likewise.
5861         (iwmmxt_wmiawttn): Likewise.
5862         (iwmmxt_wmerge): Likewise.
5863         (iwmmxt_tandc<mode>3): Likewise.
5864         (iwmmxt_torc<mode>3): Likewise.
5865         (iwmmxt_torvsc<mode>3): Likewise.
5866         (iwmmxt_textrc<mode>3): Likewise.
5867         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5868         (wmmxt_pack): Likewise.
5869         (wmmxt_mult_c1): Likewise.
5870         (wmmxt_mult_c2): Likewise.
5871         (wmmxt_alu_c1): Likewise.
5872         (wmmxt_alu_c2): Likewise.
5873         (wmmxt_alu_c3): Likewise.
5874         (wmmxt_transfer_c1): Likewise.
5875         (wmmxt_transfer_c2): Likewise.
5876         (wmmxt_transfer_c3): Likewise.
5877         (marvell_f_iwmmxt_wstr): Likewise.
5878         (marvell_f_iwmmxt_wldr): Likewise.
5880 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5882         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5884 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5886         Revert:
5887         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5888         * lra-constraints.c (need_for_split_p): Check call used hard regs
5889         living through calls.
5891         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5892         call used regs for call insn.
5894 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5896         PR target/57736
5897         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5898         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5900 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5902         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5904 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5906         * lra-constraints.c (need_for_split_p): Check call used hard regs
5907         living through calls.
5909 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5911         PR target/57744
5912         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5913         to tie with any other modes.  Eliminate Altivec vector mode tests,
5914         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5915         code, to return 0 if testing MODE2 for a condition, if we've
5916         already tested MODE1 for the same condition.
5918 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5920         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5921         layout.
5923 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5925         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5926         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5928 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5930         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5931         Define.
5932         (aarch64_symbolic_constant_p): Remove.
5933         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5934         static.  Fix line length and white space.
5935         (aarch64_symbolic_constant_p): Remove.
5936         * config/aarch64/predicates.md (aarch64_valid_symref):
5937         Use aarch64_classify_symbol_expression.
5939 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5941         * config/arm/constraints.md (Ts): New constraint.
5942         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5943         16-bit encodings.
5944         (compare_scc): Use "Ts" constraint for operand 0.
5945         (ior_scc_scc): Likewise.
5946         (and_scc_scc): Likewise.
5947         (and_scc_scc_nodom): Likewise.
5948         (ior_scc_scc_cmp): Likewise for operand 7.
5949         (and_scc_scc_cmp): Likewise.
5950         * config/arm/thumb2.md (thumb2_movsi_insn):
5951         Add alternatives for 16-bit encodings.
5952         (thumb2_movhi_insn): Likewise.
5953         (thumb2_movsicc_insn): Likewise.
5954         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5955         (thumb2_negscc): Use "Ts" constraint.
5956         Move mvn instruction outside cond_exec block.
5957         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5958         for 16-bit encodings.
5960 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5962         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5963         encoding.
5964         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5965         (mulsi3subsi): Likewise.
5966         (mulsidi3adddi): Likewise.
5967         (mulsidi3_v6): Likewise.
5968         (umulsidi3_v6): Likewise.
5969         (umulsidi3adddi_v6): Likewise.
5970         (smulsi3_highpart_v6): Likewise.
5971         (umulsi3_highpart_v6): Likewise.
5972         (mulhisi3tb): Likewise.
5973         (mulhisi3bt): Likewise.
5974         (mulhisi3tt): Likewise.
5975         (maddhisi4): Likewise.
5976         (maddhisi4tb): Likewise.
5977         (maddhisi4tt): Likewise.
5978         (maddhidi4): Likewise.
5979         (maddhidi4tb): Likewise.
5980         (maddhidi4tt): Likewise.
5981         (zeroextractsi_compare0_scratch): Likewise.
5982         (insv_zero): Likewise.
5983         (insv_t2): Likewise.
5984         (anddi_notzesidi_di): Likewise.
5985         (anddi_notsesidi_di): Likewise.
5986         (andsi_notsi_si): Likewise.
5987         (iordi_zesidi_di): Likewise.
5988         (xordi_zesidi_di): Likewise.
5989         (andsi_iorsi3_notsi): Likewise.
5990         (smax_0): Likewise.
5991         (smax_m1): Likewise.
5992         (smin_0): Likewise.
5993         (not_shiftsi): Likewise.
5994         (unaligned_loadsi): Likewise.
5995         (unaligned_loadhis): Likewise.
5996         (unaligned_loadhiu): Likewise.
5997         (unaligned_storesi): Likewise.
5998         (unaligned_storehi): Likewise.
5999         (extv_reg): Likewise.
6000         (extzv_t2): Likewise.
6001         (divsi3): Likewise.
6002         (udivsi3): Likewise.
6003         (arm_zero_extendhisi2addsi): Likewise.
6004         (arm_zero_extendqisi2addsi): Likewise.
6005         (compareqi_eq0): Likewise.
6006         (arm_extendhisi2_v6): Likewise.
6007         (arm_extendqisi2addsi): Likewise.
6008         (arm_movt): Likewise.
6009         (thumb2_ldrd): Likewise.
6010         (thumb2_ldrd_base): Likewise.
6011         (thumb2_ldrd_base_neg): Likewise.
6012         (thumb2_strd): Likewise.
6013         (thumb2_strd_base): Likewise.
6014         (thumb2_strd_base_neg): Likewise.
6015         (arm_negsi2): Add alternative for 16-bit encoding.
6016         (arm_one_cmplsi2): Likewise.
6018 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6020         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6021         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6022         (movdfcc): Likewise.
6023         * config/arm/vfp.md (*thumb2_movsf_vfp):
6024         Disable predication for arm_restrict_it.
6025         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6026         (*thumb2_movdfcc_vfp): Likewise.
6027         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6028         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6029         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6030         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6031         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6032         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6033         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6034         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6035         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6036         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6037         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6038         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6039         Disable predication for arm_restrict_it.
6041 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
6043         * config/i386/bmiintrin.h (_bextr_u32): New.
6044         (_bextr_u64): Ditto.
6046 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
6048         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6049         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6050         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6051         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6052         * config/mips/n32-elf.h: ...this new file.
6054 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
6056         PR target/57224
6057         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6058         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6060 2013-06-27  Catherine Moore  <clm@codesourcery.com>
6062         * config/mips/mips-tables.opt: Regenerate.
6063         * config/mips/mips-cpus.def: Add m14ke and m14kec.
6064         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6065         * doc/invoke.texi: Add -m14kc.
6067 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6069         PR target/57623
6070         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6071         constraints of operand 1 and 2.
6073         PR target/57623
6074         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6075         to match RTL canonicalization.  Swap predicates and
6076         constraints of operand 1 and 2.
6078 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
6080         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6081         Process OP_INOUT regs for splitting too.
6083 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6085         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6086         decl before the loop, initialize to NULL.
6087         (vectorizable_load): Initialize ptr_incr to NULL.
6089 2013-06-27  Martin Jambor  <mjambor@suse.cz>
6091         PR lto/57208
6092         * ipa-ref.h (ipa_maybe_record_reference): Declare.
6093         * ipa-ref.c (ipa_maybe_record_reference): New function.
6094         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6095         * ipa-cp.c (create_specialized_node): Record potential references from
6096         aggvals.
6097         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6099 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6101         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6102         parameter 'mode' of type 'enum machine_mode mode'; change to pass
6103         'mode' to force_reg.
6104         (aarch64_add_offset): Update calls to aarch64_force_temporary.
6105         (aarch64_expand_mov_immediate): Likewise.
6107 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6109         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6110         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6112 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6114         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6115         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6116         (struct machine_function): Add tbegin_p.
6117         (s390_canonicalize_comparison): Fold CC mode compares to
6118         conditional jump if possible.
6119         (s390_emit_jump): Return the emitted jump.
6120         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6121         Handle CCRAWmode compares.
6122         (s390_option_override): Default to -mhtm if available.
6123         (s390_reg_clobbered_rtx): Handle floating point regs as well.
6124         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6125         FPRs instead of df_regs_ever_live_p.
6126         (s390_optimize_nonescaping_tx): New function.
6127         (s390_init_frame_layout): Extend clobbered_regs array to cover
6128         FPRs as well.
6129         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6130         (s390_expand_tbegin): New function.
6131         (enum s390_builtin): New enum definition.
6132         (code_for_builtin): New array definition.
6133         (s390_init_builtins): New function.
6134         (s390_expand_builtin): New function.
6135         (TARGET_INIT_BUILTINS): Define.
6136         (TARGET_EXPAND_BUILTIN): Define.
6137         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6138         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6139         (s390_alc_comparison): Likewise.
6140         * config/s390/s390-modes.def: Add CCRAWmode.
6141         * config/s390/s390.h (processor_flags): Add PF_TX.
6142         (TARGET_CPU_HTM): Define macro.
6143         (TARGET_HTM): Define macro.
6144         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6145         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6146         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6147         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6148         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6149         (TBEGIN_MASK, TBEGINC_MASK): New constants.
6150         ("*cc_to_int"): Move up.
6151         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6152         constants other than 0.
6153         ("*ccraw_to_int"): New insn and splitter definition.
6154         ("tbegin", "tbegin_nofloat", "tbegin_retry")
6155         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6156         ("tx_assist"): New expander.
6157         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6158         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6159         * config/s390/s390.opt: Add -mhtm option.
6160         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6161         * config/s390/htmxlintrin.h: New file.
6162         * config/s390/htmintrin.h: New file.
6163         * config/s390/s390intrin.h: New file.
6164         * doc/extend.texi: Document htm builtins.
6165         * config.gcc: Add the new header files to extra_headers.
6167 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
6169         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6170         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6172 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
6173             Pat Haugen  <pthaugen@us.ibm.com>
6174             Peter Bergner  <bergner@vnet.ibm.com>
6176         * config/rs6000/power8.md: New.
6177         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6178         setting for power8 entry.
6179         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6180         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6181         test for Power4/Power5 only.
6182         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6183         support.
6184         (force_new_group): Adjust comment.
6185         * config/rs6000/rs6000.md: Include power8.md.
6187 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
6189         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6190         * config/arm/arm-protos.h (arm_max_conditional_execute): New
6191         declaration.
6192         (tune_params): Update comment.
6193         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6194         (arm_max_conditional_execute): New function.
6195         (thumb2_final_prescan_insn): Use max_insn_skipped and
6196         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6198 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
6200         PR tree-optimization/57705
6201         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6202         SSA_NAME step, provided that it is not defined inside the loop.
6203         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6204         (get_initial_def_for_induction): Handle SSA_NAME IV step.
6206 2013-06-25  Martin Jambor  <mjambor@suse.cz>
6208         PR middle-end/57670
6209         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6210         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6211         calls in the dump.
6212         (ipa_note_param_call): Initialize member_ptr flag.
6213         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6214         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6215         (ipa_write_indirect_edge_info): Stream member_ptr flag.
6216         (ipa_read_indirect_edge_info): Likewise.
6218 2013-06-25  Richard Biener  <rguenther@suse.de>
6220         PR middle-end/56977
6221         * passes.c (init_optimization_passes): Move pass_fold_builtins
6222         and pass_dce earlier with -Og.
6224 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
6226         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6227         <BIT_FIELD_REF>: Remove trailing TAB.
6228         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6229         remove blank line.
6231 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6233         PR tree-optimization/57358
6234         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6235         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6236         (ipa_analyze_params_uses): Generate pessimistic info when true.
6238 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6240         PR tree-optimization/57539
6241         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6242         global.inlined_to of the new node to it.  All callers changed.
6243         * ipa-inline-transform.c (clone_inlined_nodes): New variable
6244         inlining_into, pass it to cgraph_clone_node.
6245         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6246         ipa_free_edge_args_substructures.
6247         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6248         rdesc linked list.  Do not assert rdesc edges have inlined caller.
6249         Assert we have found an rdesc in the rdesc list.
6251 2013-06-24  Richard Biener  <rguenther@suse.de>
6253         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6254         (pointer_set_lookup): Declare.
6255         (class pointer_map): New template class implementing a
6256         generic pointer to T map.
6257         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6258         pointer_map<T>::contains, pointer_map<T>::insert,
6259         pointer_map<T>::traverse): New functions.
6260         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6261         (pointer_set_lookup): New function.
6262         (pointer_set_contains): Use pointer_set_lookup.
6263         (pointer_set_insert): Likewise.
6264         (insert_aux): Remove.
6265         (struct pointer_map_t): Embed a pointer_set_t.
6266         (pointer_map_create): Adjust.
6267         (pointer_map_destroy): Likewise.
6268         (pointer_map_contains): Likewise.
6269         (pointer_map_insert): Likewise.
6270         (pointer_map_traverse): Likewise.
6271         * tree-streamer.h (struct streamer_tree_cache_d): Use a
6272         pointer_map<unsigned> instead of a pointer_map_t.
6273         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6274         (streamer_tree_cache_lookup): Likewise.
6275         (streamer_tree_cache_create): Likewise.
6276         (streamer_tree_cache_delete): Likewise.
6277         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6278         pointer_map<unsigned> instead of a pointer_map_t.
6279         (lto_init_tree_ref_encoder): Adjust.
6280         (lto_destroy_tree_ref_encoder): Likewise.
6281         * lto-section-out.c (lto_output_decl_index): Likewise.
6282         (lto_record_function_out_decl_state): Likewise.
6283         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6285 2013-06-24  Richard Biener  <rguenther@suse.de>
6287         PR tree-optimization/57488
6288         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6290 2013-06-24  Alan Modra  <amodra@gmail.com>
6292         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6293         (gen_easy_altivec_constant): Likewise.
6294         * config/rs6000/predicates.md (easy_vector_constant_add_self,
6295         easy_vector_constant_msb): Likewise.
6297 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
6299         PR target/57688
6300         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6301         add missing return true.
6303 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
6305         PR target/52483
6306         * config/sh/predicates.md (general_extend_operand): Invoke
6307         general_movsrc_operand for memory operands.
6308         (general_movsrc_operand): Allow reg+reg addressing, do not use
6309         general_operand for memory operands.
6311 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
6313         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
6314         when current target options does not apply.
6315         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6316         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6317         * config/i386/bmiintrin.h: Pass appropriate target
6318         attributes to header.
6319         * config/i386/mmintrin.h: Ditto.
6320         * config/i386/nmmintrin.h: Ditto.
6321         * config/i386/avx2intrin.h: Ditto.
6322         * config/i386/fxsrintrin.h: Ditto.
6323         * config/i386/tbmintrin.h: Ditto.
6324         * config/i386/xsaveintrin.h: Ditto.
6325         * config/i386/f16cintrin.h: Ditto.
6326         * config/i386/xtestintrin.h: Ditto.
6327         * config/i386/xsaveoptintrin.h: Ditto.
6328         * config/i386/bmi2intrin.h: Ditto.
6329         * config/i386/lzcntintrin.h: Ditto.
6330         * config/i386/smmintrin.h: Ditto.
6331         * config/i386/wmmintrin.h: Ditto.
6332         * config/i386/x86intrin.h: Remove all header include guards.
6333         * config/i386/prfchwintrin.h: Ditto.
6334         * config/i386/pmmintrin.h: Ditto.
6335         * config/i386/tmmintrin.h: Ditto.
6336         * config/i386/xmmintrin.h: Ditto.
6337         * config/i386/popcntintrin.h: Ditto.
6338         * config/i386/rdseedintrin.h: Ditto.
6339         * config/i386/ammintrin.h: Ditto.
6340         * config/i386/emmintrin.h: Ditto.
6341         * config/i386/immintrin.h: Remove all header include guards.
6342         * config/i386/fma4intrin.h: Ditto.
6343         * config/i386/lwpintrin.h: Ditto.
6344         * config/i386/xopintrin.h: Ditto.
6345         * config/i386/ia32intrin.h: Ditto.
6346         * config/i386/avxintrin.h: Ditto.
6347         * config/i386/rtmintrin.h: Ditto.
6348         * config/i386/fmaintrin.h: Ditto.
6349         * config/i386/mm3dnow.h: Ditto.
6351 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
6353         * common/config/i386/i386-common.c: Handle LZCNT.
6355 2013-06-22  Andi Kleen  <ak@linux.intel.com>
6357         * doc/extend.texi: Use __atomic_store_n instead of
6358         __atomic_store in HLE example.
6360 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
6362         * config/sh/sh.c: Remove <cstdlib> workaround.
6364 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6366         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
6368 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6370         * doc/extend.texi: Document that __atomic_clear and
6371           __atomic_test_and_set should only be used with bool.
6373 2013-06-20  Jan Hubicka  <jh@suse.cz>
6375         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6376         types_same_for_odr.
6377         * tree.c (decls_same_for_odr): New function.
6378         (same_for_edr): New function.
6379         (types_same_for_odr): New function.
6380         (get_binfo_at_offset): Use it.
6381         * tree.h (types_same_for_odr): Declare.
6383 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
6384             Jason Merrill  <jason@redhat.com>
6386         * system.h: Include <cstdlib> as well as <stdlib.h>.
6388 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
6390         PR target/57655
6391         * config/i386/i386.c (construct_container): Report error if
6392         long double is used with disabled x87 float returns.
6394 2013-06-20  Jan Hubicka  <jh@suse.cz>
6396         * lto-cgraph.c (input_symtab): Do not set cgraph state.
6398 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
6400         PR rtl-optimization/57425
6401         PR rtl-optimization/57569
6402         * alias.c (write_dependence_p): Remove parameters mem_mode and
6403         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
6404         Changed all callers.
6405         (canon_anti_dependence): Get comments and semantics in sync.
6406         Add parameter mem_canonicalized.  Changed all callers.
6407         * rtl.h (canon_anti_dependence): Update prototype.
6409 2013-06-20  Richard Biener  <rguenther@suse.de>
6411         * data-streamer-in.c (streamer_read_uhwi): Optimize single
6412         byte case, inline streamer_read_uchar and defer section
6413         overrun check.
6415 2013-06-20  Richard Biener  <rguenther@suse.de>
6417         PR tree-optimization/57584
6418         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6419         SSA names into the expanded expression that take part in
6420         abnormal coalescing.
6422 2013-06-19  Sharad Singhai  <singhai@google.com>
6424         * gcov.c (print_usage): Handle new option.
6425         (process_args): Ditto.
6426         (get_gcov_intermediate_filename): New function.
6427         (output_intermediate_file): New function.
6428         (output_gcov_file): New function
6429         (generate_results): Handle new option.
6430         (release_function): Relase demangled name.
6431         (read_graph_file): Handle demangled name.
6432         (output_lines): Ditto.
6433         * doc/gcov.texi: Document gcov intermediate format.
6435 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
6437         PR bootstrap/57604
6438         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6439         (lra_emit_add): Use the functions.  Add comment about Y as an
6440         address segment.
6442 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
6444         PR driver/57652
6445         * collect2.c (collect_atexit): New.
6446         (collect_exit): Delete.
6447         (main): Register collect_atexit with atexit.
6448         (collect_wait): Change collect_exit to exit.
6449         (do_wait): Same.
6450         * collect2.h (collect_exit): Delete.
6451         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6453 2013-06-19  Wei Mi  <wmi@google.com>
6455         PR rtl-optimization/57518
6456         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6457         if regno is used in paradoxical subreg.
6458         (update_equiv_regs): Check pdx_subregs[regno] before
6459         set a reg to be equivalent with a mem.
6461 2013-06-19  Matthias Klose  <doko@ubuntu.com>
6463         PR driver/57651
6464         * file-find.h (find_a_file): Add a mode parameter.
6465         * file-find.c (find_a_file): Likewise.
6466         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6467         with X_OK for the executables.
6468         * collect2.c (main): Call find_a_file with X_OK.
6470 2013-06-19  Steve Ellcey  <sellcey@mips.com>
6472         PR target/56942
6473         * config/mips/mips.md (casesi_internal_mips16_<mode>):
6474         Use NEXT_INSN instead of next_real_insn.
6476 2013-06-19  Jan Hubicka  <jh@suse.cz>
6478         * cgraph.h (const_value_known_p): Replace by ...
6479         (ctor_for_folding): .. this one.
6480         * cgraphunit.c (process_function_and_variable_attributes): Use it.
6481         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6482         * expr.c (expand_expr_real_1): Likewise.
6483         (string_constant): Likewise.
6484         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6485         * ipa.c (process_references): Likewise.
6486         (symtab_remove_unreachable_nodes): Likewise.
6487         * ipa-inline-analysis.c (param_change_prob): Likewise.
6488         * gimple-fold.c (canonicalize_constructor_val): Likewise.
6489         (get_base_constructor): Likwise.
6490         * varpool.c (varpool_remove_node): Likewise.
6491         (varpool_remove_initializer): LIkewise.
6492         (dump_varpool_node): LIkwise.
6493         (const_value_known_p): Rewrite to ...
6494         (ctor_for_folding): ... this one.
6496 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
6498         PR driver/57651
6499         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6500         PERSONALITY in $PATH derived prefixes.
6502 2013-06-19  Jeff Law  <law@redhat.com>
6504         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6505         in comment.
6507         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6508         (simplify_bitwise_binary): Use it to simpify certain binary ops on
6509         booleans.
6511 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
6513         * config/arm/vfp.md: Move VFP instruction classification documentation
6514         to ...
6515         * config/arm/arm.md: ... here.  Update instruction classification
6516         documentation.
6518 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
6520         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6521         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6522         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
6524 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
6526         PR target/57609
6527         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6528         with NEXT_INSN.  Use tablejump_p to check for jump table data
6529         insns.
6531 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
6533         PR c++/56544
6534         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6535         that now in C++ the value is correct per the C++ standards.
6537 2013-06-19  Richard Biener  <rguenther@suse.de>
6539         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6540         for global context.
6542 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6544         Revert:
6545         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6547         PR target/57609
6548         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6549         with next_active_insn.
6551 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
6553         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6554         functions are inlined during failures to flag an error.
6555         * tree-inline.c (expand_call_inline): Allow the error to be flagged
6556         in early inline pass.
6558 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
6560         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6561         in comments.
6563 2013-06-18  Julian Brown  <julian@codesourcery.com>
6565         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6566         Permit virtual register pre-reload if !strict.
6567         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6568         change.
6569         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6570         prototype.
6571         * config/arm/neon.md (movmisalign<mode>): Use
6572         neon_perm_struct_or_reg_operand instead of
6573         neon_struct_or_register_operand.
6574         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6575         neon_permissive_struct_operand instead of neon_struct_operand.
6576         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6577         neon_vector_mem_operand.
6578         * config/arm/predicates.md (neon_struct_operand): Adjust call to
6579         neon_vector_mem_operand.
6580         (neon_permissive_struct_operand): New.
6581         (neon_struct_or_register_operand): Rename to...
6582         (neon_perm_struct_or_reg_operand): This. Adjust call to
6583         neon_vector_mem_operand.
6585 2013-06-18  Richard Biener  <rguenther@suse.de>
6587         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6588         * lto-streamer.h: Include pointer-set.h.
6589         (struct lto_decl_slot): Remove.
6590         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6591         Remove next_index entry.
6592         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6593         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6594         (lto_init_tree_ref_encoder): Adjust.
6595         (lto_destroy_tree_ref_encoder): Likewise.
6596         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6597         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6598         (lto_output_decl_index): Adjust.
6599         (lto_new_out_decl_state): Likewise.
6600         (lto_record_function_out_decl_state): Likewise.
6601         * lto-streamer-out.c (copy_function): Likewise.
6603 2013-06-18  Richard Biener  <rguenther@suse.de>
6605         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6606         * cgraphunit.c: Include cfgloop.h.
6607         (init_lowered_empty_function): Initialize the loop tree.
6608         (assemble_thunk): Insert new BBs into loops.
6610 2013-06-18  Richard Biener  <rguenther@suse.de>
6612         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6613         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6614         the map from cache entry to cache index optional.
6615         (streamer_tree_cache_replace_tree): Adjust accordingly.
6616         (streamer_tree_cache_append): Likewise.
6617         (streamer_tree_cache_delete): Likewise.
6618         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6619         streamer cache map from cache entry to cache index.
6620         * lto-streamer-out.c (create_output_block): Adjust.
6622 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
6624         * config/arm/arm.md (attribute "insn"): Move multiplication and
6625         division attributes to...
6626         (attribute "type"): ... here.  Remove mult.
6627         (attribute "mul32"): New attribute.
6628         (attribute "mul64"): Add umaal.
6629         (*arm_mulsi3): Update attributes.
6630         (*arm_mulsi3_v6): Likewise.
6631         (*thumb_mulsi3): Likewise.
6632         (*thumb_mulsi3_v6): Likewise.
6633         (*mulsi3_compare0): Likewise.
6634         (*mulsi3_compare0_v6): Likewise.
6635         (*mulsi_compare0_scratch): Likewise.
6636         (*mulsi_compare0_scratch_v6): Likewise.
6637         (*mulsi3addsi): Likewise.
6638         (*mulsi3addsi_v6): Likewise.
6639         (*mulsi3addsi_compare0): Likewise.
6640         (*mulsi3addsi_compare0_v6): Likewise.
6641         (*mulsi3addsi_compare0_scratch): Likewise.
6642         (*mulsi3addsi_compare0_scratch_v6): Likewise.
6643         (*mulsi3subsi): Likewise.
6644         (*mulsidi3adddi): Likewise.
6645         (*mulsi3addsi_v6): Likewise.
6646         (*mulsidi3adddi_v6): Likewise.
6647         (*mulsidi3_nov6): Likewise.
6648         (*mulsidi3_v6): Likewise.
6649         (*umulsidi3_nov6): Likewise.
6650         (*umulsidi3_v6): Likewise.
6651         (*umulsidi3adddi): Likewise.
6652         (*umulsidi3adddi_v6): Likewise.
6653         (*smulsi3_highpart_nov6): Likewise.
6654         (*smulsi3_highpart_v6): Likewise.
6655         (*umulsi3_highpart_nov6): Likewise.
6656         (*umulsi3_highpart_v6): Likewise.
6657         (mulhisi3): Likewise.
6658         (*mulhisi3tb): Likewise.
6659         (*mulhisi3bt): Likewise.
6660         (*mulhisi3tt): Likewise.
6661         (maddhisi4): Likewise.
6662         (*maddhisi4tb): Likewise.
6663         (*maddhisi4tt): Likewise.
6664         (maddhidi4): Likewise.
6665         (*maddhidi4tb): Likewise.
6666         (*maddhidi4tt): Likewise.
6667         (divsi3): Likewise.
6668         (udivsi3): Likewise.
6669         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6670         (thumb2_mulsi_short_compare0): Likewise.
6671         (thumb2_mulsi_short_compare0_scratch): Likewise.
6672         * config/arm/arm1020e.md (1020mult1): Update attribute change.
6673         (1020mult2): Likewise.
6674         (1020mult3): Likewise.
6675         (1020mult4): Likewise.
6676         (1020mult5): Likewise.
6677         (1020mult6): Likewise.
6678         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6679         change.
6680         (cortex_a15_mult64): Likewise.
6681         (cortex_a15_sdiv): Likewise.
6682         (cortex_a15_udiv): Likewise.
6683         * config/arm/arm1026ejs.md (mult1): Update attribute change.
6684         (mult2): Likewise.
6685         (mult3): Likewise.
6686         (mult4): Likewise.
6687         (mult5): Likewise.
6688         (mult6): Likewise.
6689         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6690         (pj4_ir_div): Likewise.
6691         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6692         (11_mult2): Likewise.
6693         (11_mult3): Likewise.
6694         (11_mult4): Likewise.
6695         (11_mult5): Likewise.
6696         (11_mult6): Likewise.
6697         (11_mult7): Likewise.
6698         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6699         (cortex_a8_mla): Likewise.
6700         (cortex_a8_mull): Likewise.
6701         (cortex_a8_smulwy): Likewise.
6702         (cortex_a8_smlald): Likewise.
6703         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6704         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6705         (cortex_r4_mul_3): Likewise.
6706         (cortex_r4_mla_4): Likewise.
6707         (cortex_r4_mla_3): Likewise.
6708         (cortex_r4_smlald): Likewise.
6709         (cortex_r4_mull): Likewise.
6710         (cortex_r4_sdiv): Likewise.
6711         (cortex_r4_udiv): Likewise.
6712         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6713         (cortex_a7_idiv): Likewise.
6714         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6715         (9_mult2): Likewise.
6716         (9_mult3): Likewise.
6717         (9_mult4): Likewise.
6718         (9_mult5): Likewise.
6719         (9_mult6): Likewise.
6720         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6721         (cortex_a53_sdiv): Likewise.
6722         (cortex_a53_udiv): Likewise.
6723         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6724         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6725         (mp626_mult2): Likewise.
6726         (mp626_mult3): Likewise.
6727         (mp626_mult4): Likewise.
6728         * config/arm/fa526.md (526_mult1): Update attribute change.
6729         (526_mult2): Likewise.
6730         * config/arm/arm-generic.md (mult): Update attribute change.
6731         (mult_ldsched_strongarm): Likewise.
6732         (mult_ldsched): Likewise.
6733         (multi_cycle): Likewise.
6734         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6735         * config/arm/fa606te.md (606te_mult1): Update attribute change.
6736         (606te_mult2): Likewise.
6737         (606te_mult3): Likewise.
6738         (606te_mult4): Likewise.
6739         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6740         (cortex_a9_mac16): Likewise.
6741         (cortex_a9_multiply): Likewise.
6742         (cortex_a9_mac): Likewise.
6743         (cortex_a9_multiply_long): Likewise.
6744         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6745         (626te_mult2): Likewise.
6746         (626te_mult3): Likewise.
6747         (626te_mult4): Likewise.
6749 2013-06-18  Richard Biener  <rguenther@suse.de>
6751         PR lto/57334
6752         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6754 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6756         PR target/57609
6757         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6758         with next_active_insn.
6760 2013-06-18  Alan Modra  <amodra@gmail.com>
6762         * config/rs6000/rs6000.h (enum data_align): New.
6763         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6764         (DATA_ABI_ALIGNMENT): Define.
6765         (CONSTANT_ALIGNMENT): Correct comment.
6766         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6767         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6769 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6771         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6772         ATTRIBUTE_UNUSED marking.
6774 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6776         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6777         alternative and update.
6778         (aarch64_dup_lanedi): Delete.
6779         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6780         * config/aarch64/aarch64-simd-builtins.def: Update.
6782 2013-06-17  Richard Biener  <rguenther@suse.de>
6784         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6785         (lto_input_scc): Declare.
6786         (lto_input_tree_1): Likewise.
6787         (struct lto_stats_d): Add num_tree_bodies_output and
6788         num_pickle_refs_output.
6789         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6790         (lto_read_tree_1): Split out from ...
6791         (lto_read_tree): ... this.
6792         (lto_input_scc): New function.
6793         (lto_input_tree_1): Split out from ...
6794         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6795         (lto_data_in_create): Create the streamer cache without hashes.
6796         * lto-streamer-out.c (create_output_block): Create the streamer
6797         cache with hashes when not doing WPA.
6798         (lto_write_tree_1): Split out from ...
6799         (lto_write_tree): ... this.
6800         (get_symbol_initial_value): New function.
6801         (lto_output_tree_1): Split out from ...
6802         (lto_output_tree): ... this.  Write trees as series of SCCs
6803         using a DFS walk via DFS_write_tree.
6804         (struct sccs, struct scc_entry): New types.
6805         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6806         (DFS_write_tree_body): New function.
6807         (DFS_write_tree): Likewise.
6808         (hash_tree): Likewise.
6809         (scc_entry_compare): Likewise.
6810         (hash_scc): Likewise.
6811         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6812         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6813         TREE_CHAIN as regular reference.
6814         (streamer_read_integer_cst): Remove.
6815         (streamer_get_pickled_tree): Adjust.
6816         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6817         of DECL_EXTERNALs in BLOCK_VARS for now.
6818         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6819         reference.
6820         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6821         Add hash value argument and record that if hashes are recorded
6822         in the cache.
6823         (streamer_tree_cache_insert_1): Adjust.
6824         (streamer_tree_cache_insert): Likewise.
6825         (streamer_tree_cache_insert_at): Rename to ...
6826         (streamer_tree_cache_replace_tree): ... this and adjust.
6827         (streamer_tree_cache_append): Adjust.
6828         (record_common_node): Likewise.
6829         (streamer_tree_cache_create): Add argument whether to
6830         record hash values together with trees.
6831         (streamer_tree_cache_delete): Adjust.
6832         * tree-streamer.h (struct streamer_tree_cache_d): Add
6833         vector of hashes.
6834         (streamer_read_integer_cst): Remove.
6835         (streamer_tree_cache_insert): Adjust.
6836         (streamer_tree_cache_append): Likewise.
6837         (streamer_tree_cache_insert_at): Rename to ...
6838         (streamer_tree_cache_replace_tree): ... this and adjust.
6839         (streamer_tree_cache_create): Add argument whether to record hashes.
6840         (streamer_tree_cache_get): Rename to ...
6841         (streamer_tree_cache_get_tree): ... this.
6842         (streamer_tree_cache_get_hash): New function.
6843         * tree.c (cache_integer_cst): New function.
6844         * tree.h (cache_integer_cst): Declare.
6845         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6846         * lto-symtab.c (lto_varpool_replace_node): Only release
6847         DECL_INITIAL of non-prevailing decls.
6848         * varpool.c (varpool_remove_initializer): Do not release
6849         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6851 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6853         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6854         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6855         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6856         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6857         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6858         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6859         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6860         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6861         instead of TARGET_64BIT.
6862         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6863         Require ISA_HAS_<D>DIV.
6865 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6867         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6868         (mips*-*-linux*): Move default with_llsc setting to where other
6869         defaults are set.
6870         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6871         with_arch block.
6872         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6873         Likewise.  Remove default with_tune setting.  Move default float
6874         setting to its own block.  Handle with_llsc in the same block as above.
6876 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
6878         PR rtl-optimization/57425
6879         PR rtl-optimization/57569
6880         * alias.c (write_dependence_p): Add new parameters mem_mode,
6881         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6882         Changed all callers.
6883         (canon_anti_dependence): New function.
6884         * cse.c (check_dependence): Use canon_anti_dependence.
6885         * cselib.c (cselib_invalidate_mem): Likewise.
6886         * rtl.h (canon_anti_dependence): Declare.
6888 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6890         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6891         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6892         ".set mips3" for 64-bit targets.
6894 2013-06-15  Dehao Chen  <dehao@google.com>
6896         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6897         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6898         (gimple_check_call_args): Likewise.
6899         * value-prof.c (check_ic_target): Likewise.
6900         * ipa-inline.c (early_inliner): Likewise.
6901         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6902         * cgraph.c (cgraph_create_edge_1): Likewise.
6903         (cgraph_make_edge_direct): Likewise.
6905 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6907         PR target/57615
6908         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6909         rs6000_output_move_128bit to handle emitting quad memory
6910         operations.  Set attribute length to 8 bytes.
6912 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
6914         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6915         New pattern.
6916         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6917         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6918         (aarch64_<su>mlsl<mode>): Likewise.
6920 2013-06-14  Mike Stump  <mikestump@comcast.net>
6922         * Makefile.in (TARGET_H): Add insn-codes.h.
6924 2013-06-14  Alan Modra  <amodra@gmail.com>
6926         PR middle-end/57134
6927         PR middle-end/57586
6928         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6929         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6930         bitfield expansion when EXPAND_MEMORY.
6931         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6933 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6935         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6936         test for clearing quad memory on 32-bit later.
6938 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6940         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6941         (fold_negate_expr): Likewise.
6942         (fold_real_zero_addition_p): Handle vectors.
6943         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6945 2013-06-14  Alan Modra  <amodra@gmail.com>
6947         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6949 2013-06-13  Jan Hubicka  <jh@suse.cz>
6951         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6952         Local comdats are not externally visible.
6953         * symtab.c (dump_symtab_base): Dump externally visible.
6954         (verify_symtab_base): Verify back links in the symtab hash.
6956 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6958         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6959         CONVERT_EXPR as equal nodes.
6961 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6963         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6965 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6967         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6968         Generalize to complex and vector.
6969         * tree.c (build_all_ones_cst): New function.
6970         * tree.h (build_all_ones_cst): Declare it.
6972 2013-06-13  Alan Modra  <amodra@gmail.com>
6974         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6975         * config/rs6000/rs6000.md (signbittf2): New insn.
6976         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6977         (abstf2_internal, cmptf_internal2): Likewise.
6978         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6980 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6981             Pat Haugen  <pthaugen@us.ibm.com>
6982             Peter Bergner  <bergner@vnet.ibm.com>
6984         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6985         power8 byte, half-word, and quad-word atomic instructions.
6986         (emit_store_conditional): Likewise.
6987         (rs6000_expand_atomic_compare_and_swap): Likewise.
6988         (rs6000_expand_atomic_op): Likewise.
6990         * config/rs6000/sync.md (larx): Add new modes for power8.
6991         (stcx): Likewise.
6992         (AINT): New mode iterator to include TImode as well as normal
6993         integer modes on power8.
6994         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6995         that VSX registers are not considered.  Use AINT mode iterator
6996         instead of INT1 to allow inclusion of quad word atomic operations
6997         on power8.
6998         (load_locked<mode>): Likewise.
6999         (store_conditional<mode>): Likewise.
7000         (atomic_compare_and_swap<mode>): Likewise.
7001         (atomic_exchange<mode>): Likewise.
7002         (atomic_nand<mode>): Likewise.
7003         (atomic_fetch_<fetchop_name><mode>): Likewise.
7004         (atomic_nand_fetch<mode>): Likewise.
7005         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7006         each type.
7007         (ATOMIC): On power8, add QImode, HImode modes.
7008         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7009         modes that promote to SImode.
7010         (load_lockedti): Convert TImode arguments to PTImode, so that we
7011         get a guaranteed even/odd register pair.
7012         (load_lockedpti): Likewise.
7013         (store_conditionalti): Likewise.
7014         (store_conditionalpti): Likewise.
7016         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7017         atomic load/store instructions.
7018         (HSI): Likewise.
7020 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
7022         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7023         loads.
7024         (insn_count): New attribute, with most cases extracted from...
7025         (length): ...here.  Redefine most cases in terms of insn_count.
7026         (single_insn): Delete.
7027         (can_delay): Use insn_count to check for single instructions.
7028         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7029         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7030         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7031         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7032         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7033         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7034         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7035         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7036         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7037         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7038         rather than "length".
7039         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
7040         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7041         Use "insn_count" rather than "length".
7042         * config/mips/mips-dsp.md
7043         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7044         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7045         length attributes.
7047 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
7049         PR tree-optimization/57361
7050         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7052 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
7054         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7055         to split.
7056         (aarch64_simd_combine<mode>): New instruction expansion.
7057         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7058         function prototype.
7059         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7060         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7062 2013-06-12  Jan Hubicka  <jh@suse.cz>
7064         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7065         decl has when in streaming stage.
7066         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7067         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7069 2013-06-12  Roland Stigge  <stigge@antcom.de>
7071         PR target/57578
7072         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7074 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7076         PR tree-optimization/57537
7077         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7078         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7080 2013-06-12  Richard Biener  <rguenther@suse.de>
7082         * data-streamer.h (streamer_write_char_stream): CSE
7083         obs->current_pointer.
7084         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7085         streamer_write_char_stream manually and optimize the resulting loop.
7086         (streamer_write_hwi_stream): Likewise.
7088 2013-06-12  Jan Hubicka  <jh@suse.cz>
7090         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7091         * cgraph.h (varpool_create_empty_node): Declare.
7092         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7093         duplicated nodes.
7094         * symtab.c (symtab_unregister_node): Be lax about missin entries
7095         in node hash.
7096         (symtab_get_node): Update comment.
7097         * varpool.c (varpool_create_empty_node): Break out from ...
7098         (varpool_node_for_decl): ... here.
7099         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7101 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
7103         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7104         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7105         part.  Use straight-line flow at the end.
7106         <COMPONENT_REF>: Remove superfluous else.
7107         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7109 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7111         PR target/56564
7112         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7113         target hook even for !TREE_PUBLIC decls.  If no resolution info
7114         is available, return false for common and external decls.
7116 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7118         * config/rl78/constraints.md (U): New constraint.
7119         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7120         valloc attribute.
7122 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7124         PR target/57589
7125         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7126         to allow returning address to AT_PLATFORM name.
7128 2013-06-11  Jan Hubicka  <jh@suse.cz>
7130         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7131         * cgraph.h (symtab_node_base): Add weakref flag.
7132         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7133         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7134         (output_weakrefs): Use weakref flag.
7135         * fold-const.c (simple_operand_p): Handle WEAK.
7136         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7137         * ipa.c (varpool_externally_visible_p): Drop weakref.
7138         (function_and_variable_visibility): Update comment; fix weakref
7139         sanity checks; do not clear DECL_WEAK on them.
7140         * lto-cgraph.c (lto_output_node): update.
7141         (lto_output_varpool_node): Update.
7142         (input_overwrite_node): Update.
7143         (input_node): Update.
7144         (input_varpool_node): Update.
7145         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7146         (lto_symtab_merge_symbols): Add sanity check.
7147         (lto_symtab_prevailing_decl): Do not special case weakrefs.
7148         * passes.c (rest_of_decl_compilation): Set static flag, too.
7149         * symtab.c (dump_symtab_base): Dump weakref.
7150         (verify_symtab_base): Sanity check weakrefs.
7151         (symtab_make_decl_local): Remove duplicated code.
7152         (symtab_alias_ultimate_target): Simplify.
7153         * varpool.c (varpool_create_variable_alias): Set weakref flag.
7155 2013-06-11  DJ Delorie  <dj@redhat.com>
7157         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7158         (rl78_unwind_word_mode): New.
7160 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7162         * final.c (debug_prefix_maps): Make static.
7164 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7166         * function.c (initial_trampoline): Remove stray copy.
7168 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
7170         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7172 2013-06-11  Martin Jambor  <mjambor@suse.cz>
7174         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7175         within bounds at the beginning of the function.
7177 2013-06-11  Alan Modra  <amodra@gmail.com>
7179         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
7180         reporting.
7181         (get_named_section): Don't NULL !DECL_P decl.
7183 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
7185         * doc/invoke.texi (core-avx2): Document.
7186         (slm): Likewise.
7187         (atom): Updated with MOVBE.
7189 2013-06-11  Richard Biener  <rguenther@suse.de>
7191         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7193 2013-06-11  Anton Blanchard  <anton@samba.org>
7195         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7196         correct shift value in little-endian mode.
7198 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
7200         PR target/56564
7201         * varasm.c (get_variable_align): Move #endif to the right place.
7203 2013-06-10  Cary Coutant  <ccoutant@google.com>
7205         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7206         for hash so that hash table traversal order is deterministic.
7208 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
7209             Pat Haugen  <pthaugen@us.ibm.com>
7210             Peter Bergner  <bergner@vnet.ibm.com>
7212         * config/rs6000/vector.md (GPR move splitter): Do not split moves
7213         of vectors in GPRS if they are direct moves or quad word load or
7214         store moves.
7216         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7217         declaration.
7218         (direct_move_p): Likewise.
7219         (quad_load_store_p): Likewise.
7221         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7222         classes into bins based on the physical register type.
7223         (reg_class_to_reg_type): Likewise.
7224         (IS_STD_REG_TYPE): Likewise.
7225         (IS_FP_VECT_REG_TYPE): Likewise.
7226         (reload_fpr_gpr): Arrays to determine what insn to use if we can
7227         use direct move instructions.
7228         (reload_gpr_vsx): Likewise.
7229         (reload_vsx_gpr): Likewise.
7230         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7231         information that is a simplification of register classes.  Also
7232         precalculate direct move reload helpers.
7233         (direct_move_p): New function to return true if the operation can
7234         be done as a direct move instruciton.
7235         (quad_load_store_p): New function to return true if the operation
7236         is a quad memory operation.
7237         (rs6000_legitimize_address): If quad memory, only allow register
7238         indirect for TImode addresses.
7239         (rs6000_legitimate_address_p): Likewise.
7240         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7241         (rs6000_reload_register_type): Likewise.
7242         (register_to_reg_type): Return register type.
7243         (rs6000_secondary_reload_simple_move): New helper function for
7244         secondary reload and secondary memory needed to identify anything
7245         that is a simple move, and does not need reloading.
7246         (rs6000_secondary_reload_direct_move): New helper function for
7247         secondary reload to identify cases that can be done with several
7248         instructions via the direct move instructions.
7249         (rs6000_secondary_reload_move): New helper function for secondary
7250         reload to identify moves between register types that can be done.
7251         (rs6000_secondary_reload): Add support for quad memory operations
7252         and for direct move.
7253         (rs6000_secondary_memory_needed): Likewise.
7254         (rs6000_debug_secondary_memory_needed): Change argument names.
7255         (rs6000_output_move_128bit): New function to return the move to
7256         use for 128-bit moves, including knowing about the various
7257         limitations of quad memory operations.
7259         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7260         memory operations.  call rs6000_output_move_128bit for the actual
7261         instruciton(s) to generate.
7262         (vsx_movti_64bit): Likewise.
7264         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7265         (UNSPEC_P8V_MTVSRWZ): Likewise.
7266         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7267         (UNSPEC_P8V_MTVSRD): Likewise.
7268         (UNSPEC_P8V_XXPERMDI): Likewise.
7269         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7270         (UNSPEC_FUSION_GPR): Likewise.
7271         (FMOVE128_GPR): New iterator for direct move.
7272         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7273         (f32_sv): Likewise.
7274         (f32_dm): Likewise.
7275         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7276         loads and direct move instructions.
7277         (zero_extendsidi2_lfiwzx): Likewise.
7278         (extendsidi2_lfiwax): Likewise.
7279         (extendsidi2_nocell): Likewise.
7280         (floatsi<mode>2_lfiwax): Likewise.
7281         (lfiwax): Likewise.
7282         (floatunssi<mode>2_lfiwzx): Likewise.
7283         (lfiwzx): Likewise.
7284         (fix_trunc<mode>_stfiwx): Likewise.
7285         (fixuns_trunc<mode>_stfiwx): Likewise.
7286         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7287         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7288         (parity<mode>2_cmpb): Set length/type attr.
7289         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7290         for 'mr.' to fast_compare.
7291         (bpermd_<mode>): Change type attr to popcnt.
7292         (p8_fmrgow_<mode>): New insns for power8 direct move support.
7293         (p8_mtvsrwz_1): Likewise.
7294         (p8_mtvsrwz_2): Likewise.
7295         (reload_fpr_from_gpr<mode>): Likewise.
7296         (p8_mtvsrd_1): Likewise.
7297         (p8_mtvsrd_2): Likewise.
7298         (p8_xxpermdi_<mode>): Likewise.
7299         (reload_vsx_from_gpr<mode>): Likewise.
7300         (reload_vsx_from_gprsf): Likewise.
7301         (p8_mfvsrd_3_<mode>): LIkewise.
7302         (reload_gpr_from_vsx<mode>): Likewise.
7303         (reload_gpr_from_vsxsf): Likewise.
7304         (p8_mfvsrd_4_disf): Likewise.
7305         (multi-word GPR splits): Do not split direct moves or quad memory
7306         operations.
7308 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
7310         * tree-into-ssa.c (interesting_blocks): Make static.
7312 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
7314         PR target/56564
7315         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7316         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7317         Use DATA_ABI_ALIGNMENT for that case instead if defined.
7318         (get_variable_align): New function.
7319         (get_variable_section, emit_bss, emit_common,
7320         assemble_variable_contents, place_block_symbol): Use
7321         get_variable_align instead of DECL_ALIGN.
7322         (assemble_noswitch_variable): Add align argument, use it
7323         instead of DECL_ALIGN.
7324         (assemble_variable): Adjust caller.  Use get_variable_align
7325         instead of DECL_ALIGN.
7326         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7327         caller.
7328         (DATA_ABI_ALIGNMENT): Define.
7329         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7330         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
7331         opt is false, only return the psABI mandated alignment increase.
7332         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7333         (DATA_ABI_ALIGNMENT): ... this.
7334         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7335         (DATA_ABI_ALIGNMENT): ... this.
7336         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7337         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7338         (DATA_ABI_ALIGNMENT): ... this.
7339         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7340         * doc/tm.texi: Regenerated.
7342 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
7344         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7345         cmp_code to construct REG_EQUAL note.
7347 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
7349         PR target/57568
7350         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7351         that operands[2] doesn't overlap with operands[0].
7353 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
7354             Jan Hubicka  <jh@suse.cz>
7356         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7357         hack to mark symbols as used.
7359 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
7361         PR rtl-optimization/57559
7362         * lra-constraints.c (process_alt_operands): Don't discourage
7363         memory with known offset for offsetable memory constraint.
7364         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7366 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
7368         * varasm.c (struct oc_local_state): Reorder fields.
7369         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7370         and adjust accordingly.
7371         (output_constructor): Reorder initialization code and adjust call to
7372         output_constructor_bitfield.
7374 2013-06-07  Jan Hubicka  <jh@suse.cz>
7376         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7378 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
7380         * tree-object-size.c (unknown): Make const.
7382 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7384         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7385         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7386         for last alternative in the cpu_facility attribute.
7388 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7390         PR target/56315
7391         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7392         (xordi3): Change operand 2 constraint to arm_xordi_operand.
7393         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7394         * config/arm/constraints.md (Dg): New constraint.
7395         * config/arm/neon.md (xordi3_neon): Remove.
7396         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7397         * config/arm/predicates.md (arm_xordi_operand): New predicate.
7399 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7401         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7402         Clean up alternatives.
7404 2013-06-07  Alan Modra  <amodra@gmail.com>
7406         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7407         va_list_gpr_size.
7409 2013-06-07  Alan Modra  <amodra@gmail.com>
7411         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7413 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7415         * config/arm/constraints.md (Df): New constraint.
7416         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7417         Correct length attribute for last two alternatives.
7419 2013-06-07  Alan Modra  <amodra@gmail.com>
7421         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7422         override user -mfp-in-toc.
7423         (offsettable_ok_by_alignment): Consider just the current access
7424         rather than the whole object, unless BLKmode.  Handle
7425         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7426         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7427         for -mcmodel=medium.
7428         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7429         override user -mfp-in-toc or -msum-in-toc.  Default to
7430         -mno-fp-in-toc for -mcmodel=medium.
7432 2013-06-06  DJ Delorie  <dj@redhat.com>
7434         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7435         TARGET_VALID_POINTER_MODE.
7437 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
7438             Pat Haugen  <pthaugen@us.ibm.com>
7439             Peter Bergner  <bergner@vnet.ibm.com>
7441         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7442         Document new power8 builtins.
7444         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7445         condition code register, to allow 128-bit logical operations to be
7446         done in the VSX or GPR registers.
7447         (nor<mode>3): Use the canonical form for nor.
7448         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7449         vclz*, and vpopcnt* vector instructions.
7450         (nand<mode>3): Likewise.
7451         (orc<mode>3): Likewise.
7452         (clz<mode>2): LIkewise.
7453         (popcount<mode>2): Likewise.
7455         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7456         that only the GPRs are recognized.
7458         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7459         support for new power8 builtins.
7461         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7462         builtin functions.
7463         (xscvdpspn): Likewise.
7464         (vclz): Likewise.
7465         (vclzb): Likewise.
7466         (vclzh): Likewise.
7467         (vclzw): Likewise.
7468         (vclzd): Likewise.
7469         (vpopcnt): Likewise.
7470         (vpopcntb): Likewise.
7471         (vpopcnth): Likewise.
7472         (vpopcntw): Likewise.
7473         (vpopcntd): Likewise.
7474         (vgbbd): Likewise.
7475         (vmrgew): Likewise.
7476         (vmrgow): Likewise.
7477         (eqv): Likewise.
7478         (eqv_v16qi3): Likewise.
7479         (eqv_v8hi3): Likewise.
7480         (eqv_v4si3): Likewise.
7481         (eqv_v2di3): Likewise.
7482         (eqv_v4sf3): Likewise.
7483         (eqv_v2df3): Likewise.
7484         (nand): Likewise.
7485         (nand_v16qi3): Likewise.
7486         (nand_v8hi3): Likewise.
7487         (nand_v4si3): Likewise.
7488         (nand_v2di3): Likewise.
7489         (nand_v4sf3): Likewise.
7490         (nand_v2df3): Likewise.
7491         (orc): Likewise.
7492         (orc_v16qi3): Likewise.
7493         (orc_v8hi3): Likewise.
7494         (orc_v4si3): Likewise.
7495         (orc_v2di3): Likewise.
7496         (orc_v4sf3): Likewise.
7497         (orc_v2df3): Likewise.
7499         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7500         allow power8 quad mode in 64-bit.
7501         (rs6000_builtin_vectorized_function): Add support to vectorize
7502         ISA 2.07 count leading zeros, population count builtins.
7503         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7504         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7505         (builtin_function_type): Add vgbbd builtin function which takes an
7506         unsigned argument.
7507         (altivec_expand_vec_perm_const): Add support for new power8 merge
7508         instructions.
7510         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7511         that does not include TImdoe for use with 32-bit.
7512         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7513         instructions.
7514         (UNSPEC_VSX_CVDPSPN): Likewise.
7515         (vsx_xscvdpspn): Likewise.
7516         (vsx_xscvspdpn): Likewise.
7517         (vsx_xscvdpspn_scalar): Likewise.
7518         (vsx_xscvspdpn_directmove): Likewise.
7519         (vsx_and<mode>3): Split logical operations into 32-bit and
7520         64-bit. Add support to do logical operations on TImode as well as
7521         VSX vector types.  Allow logical operations to be done in either
7522         VSX registers or in general purpose registers in 64-bit mode.  Add
7523         splitters if GPRs were used. For AND, add clobber of CCmode to
7524         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
7525         encoding.
7526         (vsx_and<mode>3_32bit): Likewise.
7527         (vsx_and<mode>3_64bit): Likewise.
7528         (vsx_ior<mode>3): Likewise.
7529         (vsx_ior<mode>3_32bit): Likewise.
7530         (vsx_ior<mode>3_64bit): Likewise.
7531         (vsx_xor<mode>3): Likewise.
7532         (vsx_xor<mode>3_32bit): Likewise.
7533         (vsx_xor<mode>3_64bit): Likewise.
7534         (vsx_one_cmpl<mode>2): Likewise.
7535         (vsx_one_cmpl<mode>2_32bit): Likewise.
7536         (vsx_one_cmpl<mode>2_64bit): Likewise.
7537         (vsx_nor<mode>3): Likewise.
7538         (vsx_nor<mode>3_32bit): Likewise.
7539         (vsx_nor<mode>3_64bit): Likewise.
7540         (vsx_andc<mode>3): Likewise.
7541         (vsx_andc<mode>3_32bit): Likewise.
7542         (vsx_andc<mode>3_64bit): Likewise.
7543         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7544         and xxlorc instructions.
7545         (vsx_eqv<mode>3_64bit): Likewise.
7546         (vsx_nand<mode>3_32bit): Likewise.
7547         (vsx_nand<mode>3_64bit): Likewise.
7548         (vsx_orc<mode>3_32bit): Likewise.
7549         (vsx_orc<mode>3_64bit): Likewise.
7551         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7553         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7554         instruction.
7555         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7556         (p8_vmrgow): Likewise.
7557         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7558         GPRs to be split under VSX.
7559         (p8v_clz<mode>2): Add power8 count leading zero support.
7560         (p8v_popcount<mode>2): Add power8 population count support.
7561         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7562         support.
7564         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7565         instruction.
7567         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7568         builtin functions.
7569         (vec_nand): Likewise.
7570         (vec_vclz): Likewise.
7571         (vec_vclzb): Likewise.
7572         (vec_vclzd): Likewise.
7573         (vec_vclzh): Likewise.
7574         (vec_vclzw): Likewise.
7575         (vec_vgbbd): Likewise.
7576         (vec_vmrgew): Likewise.
7577         (vec_vmrgow): Likewise.
7578         (vec_vpopcnt): Likewise.
7579         (vec_vpopcntb): Likewise.
7580         (vec_vpopcntd): Likewise.
7581         (vec_vpopcnth): Likewise.
7582         (vec_vpopcntw): Likewise.
7584 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7586         PR rtl-optimization/57468
7587         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7588         spilled pseudos.
7590 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7592         PR rtl-optimization/57459
7593         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7594         type when setting live regs.
7596 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7598         * config/s390/s390.opt (mlra): New option.
7599         * config/s390/s390.c (s390_decompose_address): Check displacement
7600         for all registers for LRA.
7601         (s390_secondary_reload): Don't used secondary reloads for LRA.
7602         (s390_lra_p): New function.
7603         (TARGET_LRA_P): Define.
7604         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7605         of attribute cpu_facility to zarch for the last alternative.
7606         (*cmpmem_short): Ditto.
7608 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
7610         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7611         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
7612         (arm_expand_prologue): Likewise.
7614 2013-06-06  Teresa Johnson  <tejohnson@google.com>
7616         PR c++/53743
7617         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7618         as this is now done by redirect_edge_and_branch_force.
7619         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7620         barriers, and fix interaction with splitting.
7621         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7622         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7623         reflect changes made in the routine.
7624         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7625         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7626         since this is called in cfglayout mode, and replace partition fixup
7627         with assert as that is now done by force_nonfallthru_and_redirect.
7628         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7629         already be marked with region crossing note.
7630         (insert_section_boundary_note): Make non-static, gate on flag
7631         has_bb_partition, rewrite to also check for multiple partitions.
7632         (rest_of_handle_reorder_blocks): Remove call to
7633         insert_section_boundary_note, now done later during free_cfg.
7634         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7635         * bb-reorder.h (insert_section_boundary_note): Declare.
7636         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7637         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7638         invoke insert_section_boundary_note.
7639         (try_redirect_by_replacing_jump): Remove unnecessary
7640         check for region crossing note.
7641         (fixup_partition_crossing): New function.
7642         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7643         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7644         in non-cfglayout mode.
7645         (force_nonfallthru_and_redirect): Fixup partition boundaries,
7646         remove old code that tried to do this. Emit barrier correctly
7647         when we are in cfglayout mode.
7648         (last_bb_in_partition): New function.
7649         (rtl_split_edge): Correctly fixup partition boundaries.
7650         (commit_one_edge_insertion): Remove old code that tried to
7651         fixup region crossing edge since this is now handled in
7652         split_block, and set up insertion point correctly since
7653         block may now end in a jump.
7654         (verify_hot_cold_block_grouping): Guard against checking when not in
7655         linearized RTL mode.
7656         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7657         notes.
7658         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7659         rtl_verify_flow_info, so not called in cfglayout mode.
7660         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7661         (fixup_reorder_chain): Remove old code that attempted to fixup region
7662         crossing note as this is now handled in force_nonfallthru_and_redirect.
7663         (duplicate_insn_chain): Don't duplicate switch section notes.
7664         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7665         note.
7666         * basic-block.h (emit_barrier_after_bb): Declare.
7668 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7670         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7671         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7672         arm_usatsihi): Adjust alternatives for arm_restrict_it.
7674 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7676         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7677         where appropriate.
7678         * config/arm/ldmstm.md: Regenerate.
7680 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7682         * config/arm/sync.md (atomic_loaddi_1):
7683         Disable predication for arm_restrict_it.
7684         (arm_load_exclusive<mode>): Likewise.
7685         (arm_load_exclusivesi): Likewise.
7686         (arm_load_exclusivedi): Likewise.
7687         (arm_load_acquire_exclusive<mode>): Likewise.
7688         (arm_load_acquire_exclusivesi): Likewise.
7689         (arm_load_acquire_exclusivedi): Likewise.
7690         (arm_store_exclusive<mode>): Likewise.
7691         (arm_store_exclusive<mode>): Likewise.
7692         (arm_store_release_exclusivedi): Likewise.
7693         (arm_store_release_exclusive<mode>): Likewise.
7695 2013-06-06  Richard Biener  <rguenther@suse.de>
7697         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7698         after LTO_null.
7699         (lto_tag_is_tree_code_p): Adjust.
7700         (lto_tag_is_gimple_code_p): Likewise.
7701         (lto_gimple_code_to_tag): Likewise.
7702         (lto_tag_to_gimple_code): Likewise.
7703         (lto_tree_code_to_tag): Likewise.
7704         (lto_tag_to_tree_code): Likewise.
7705         * data-streamer.h (streamer_write_hwi_in_range): Use
7706         uhwi streaming to stream the normalized range.
7707         (streamer_read_hwi_in_range): Likewise.
7709 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7711         * config/arm/arm.md (enabled_for_depr_it): New attribute.
7712         (predicable_short_it): Likewise.
7713         (predicated): Likewise.
7714         (enabled): Handle above.
7715         (define_cond_exec): Set predicated attribute to yes.
7717 2013-06-05  Mike Stump  <mikestump@comcast.net>
7719         * gdbinit.in (__FUNCTION__): Add.
7721 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
7723         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7724         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7726 2013-06-05  Jan Hubicka  <jh@suse.cz>
7728         * varasm.c (mark_decl_referenced): Revert the removal until targets
7729         are fixed.
7731 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
7733         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7734         instead of mark_decl_referenced.
7736 2013-06-05  Jan Hubicka  <jh@suse.cz>
7738         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7739         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7740         and symtab_used_from_object_file_p.
7741         (cgraph_make_node_local_1): Clear forced_by_abi.
7742         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7743         * cgraph.h (symtab_node_base): Add forced_by_abi.
7744         (decide_is_variable_needed): Remove.
7745         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7746         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7747         (decide_is_symbol_needed): ... this one; handle symbols in general;
7748         always analyze virtuals; honnor forced_by_abi.
7749         (cgraph_finalize_function): Update.
7750         (varpool_finalize_decl): Update.
7751         (symbol_defined_and_needed): Remove.
7752         (analyze_functions): Update.
7753         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7754         output_refs, input_overwrite_node): Handle forced_by_abi.
7755         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7756         (address_taken_from_non_vtable_p): ... this one.
7757         (comdat_can_be_unshared_p_1): New function.
7758         (cgraph_comdat_can_be_unshared_p): Rename to ...
7759         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7760         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7761         (function_and_variable_visibility): Clear forced_by_abi as needed.
7762         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7763         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7764         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7765         * varpool.c (decide_is_variable_needed): Remove.
7767 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7769         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7770         (arm_option_override): Override arm_restrict_it where appropriate.
7771         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7772         * config/arm/arm.opt (mrestrict-it): New command-line option.
7773         * doc/invoke.texi: Document -mrestrict-it.
7775 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7777         * tsan.c (tsan_atomic_table): Make const.
7779 2013-06-05  Richard Biener  <rguenther@suse.de>
7781         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7782         index associated with the tree we are supposed to replace.
7783         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7784         TREE_ASM_WRITTEN as zero for everything but SSA names.
7786 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7788         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7790 2013-06-04  Jan Hubicka  <jh@suse.cz>
7792         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7793         (input_node, input_varpool_node): Handle correctly external same
7794         body aliases.
7795         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7796         nodes at ltrans stage.
7798 2013-06-04  Jan Hubicka  <jh@suse.cz>
7800         * ipa-inline.c (update_caller_keys): Fix availability test.
7801         (update_callee_keys): Likewise.
7802         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7803         to follow ELF standard.
7805 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7807         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7808         (mips64r5900el-*-elf*): New configurations.
7809         * config/mips/mips-cpus.def (r5900): New processor.
7810         * config/mips/mips-tables.opt: Regenerate.
7811         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7812         (mips_issue_rate): Handle PROCESSOR_R5900.
7813         (mips_reorg_process_insns): Force reorder mode for the R5900.
7814         * config/mips/mips.h (TARGET_MIPS5900): Define.
7815         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7816         TARGET_MIPS5900.
7817         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7818         TARGET_MIPS5900.
7819         * config/mips/mips.md (processor): Add r5900.
7820         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7822 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7824         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7825         into function to generate MOVI instruction.
7826         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7827         (aarch64_preferred_simd_mode): Turn into wrapper.
7828         (aarch64_output_scalar_simd_mov_immediate): New function.
7829         * config/aarch64/aarch64-protos.h: Add prototype for above.
7831 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7833         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7834         element_char member.
7835         (sizetochar): Return signed char.
7836         (aarch64_simd_valid_immediate): Remove elchar and other
7837         unnecessary variables.
7838         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7839         Calculate element_char as required.
7840         * config/aarch64/aarch64-protos.h: Update and move prototype
7841         for aarch64_output_simd_mov_immediate.
7842         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7843         Update arguments.
7845 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7847         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7848         information completed by aarch64_simd_valid_immediate.
7849         (aarch64_legitimate_constant_p): Update arguments.
7850         (aarch64_simd_valid_immediate): Work with struct rather than many
7851         pointers.
7852         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7853         (aarch64_simd_make_constant): Update arguments.
7854         (aarch64_output_simd_mov_immediate): Work with struct rather than
7855         many pointers.  Output immediate directly rather than as operand.
7856         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7857         Update prototype.
7858         * config/aarch64/constraints.md (Dn): Update arguments.
7860 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7862         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7863         longer static.
7864         (aarch64_simd_immediate_valid_for_move): Remove.
7865         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7866         (aarch64_simd_make_constant): Update call.
7867         (aarch64_output_simd_mov_immediate): Update call.
7868         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7869         Add prototype.
7870         * config/aarch64/constraints.md (Dn): Update call.
7872 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7874         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7875         return type to bool for prototype.
7876         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7877         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7878         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7880 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7882         * config/mips/mips.opt (meva): New.
7883         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7884         (ASM_SPEC): Handle -meva.
7885         * doc/invoke.texi (meva):  Document.
7887 2013-06-04  Alan Modra  <amodra@gmail.com>
7889         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7890         constant output.
7892 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7894         * rtl.def: Add extra fourth optional field to define_cond_exec.
7895         * gensupport.c (process_one_cond_exec): Process attributes from
7896         define_cond_exec.
7897         * doc/md.texi: Document fourth field in define_cond_exec.
7899 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7901         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7902         out the processing order as in store_bit_field_1.
7904 2013-06-04  Jan Hubicka  <jh@suse.cz>
7906         PR middle-end/57500
7907         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7908         non-VAR_DECL node if it does not exist yet.
7910 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7912         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7913         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7914         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7915         target_cpu_default setting.
7917 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7919         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7920         default to -fopt-info=optimized instead of all.
7921         * doc/invoke.texi: Ditto.
7922         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7923         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7924         (execute_vect_slp): Emit BB vectorization success under
7925         MSG_OPTIMIZED_LOCATIONS.
7926         * tree-vect-slp.c (vect_slp_transform_bb): Change
7927         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7928         * tree-vect-loop.c (vect_transform_loop): Ditto.
7930 2013-06-03  Jason Merrill  <jason@redhat.com>
7932         PR c++/57415
7933         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7934         Use TARGET_EXPR for C++.
7936 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7938         PR rtl-optimization/57268
7939         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7940         if DEBUG_INSN_P (insn).
7942         Reapply
7943         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7945         PR rtl-optimization/57268
7946         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7947         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7949 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7951         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7952         (ix86_avoid_lea_for_addr): Likewise.
7953         (exact_dependency_1): Likewise.
7954         (ix86_adjust_cost): Likewise.
7955         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7956         removed.
7957         (do_reorder_for_imul): Fix typo, formatting and
7958         !reload_completed check removed.
7959         (ix86_sched_reorder): Fix typo and formatting.
7960         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7961         list.
7963 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7965         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7967 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7969         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7970         <STRING_CST>: Likewise.
7971         <VECTOR_CST>: Likewise.
7973 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7974             Mikael Morin  <mikael@gcc.gnu.org>
7976         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7977         * config.in: Regenerated.
7978         * configure: Regenerated.
7980 2013-06-01  Jan Hubicka  <jh@suse.cz>
7982         PR middle-end/57366
7983         * cgraphunit.c (compile): When weakref is not supported,
7984         set up transparent aliases before final output pass.
7985         * varasm.c (assemble_alias): Do not try to do it here.
7987 2013-06-01  Jan Hubicka  <jh@suse.cz>
7989         PR middle-end/57467
7990         * passes.c (for_per_function): Skip unanalyzed functions.
7992 2013-06-01  Jan Hubicka  <jh@suse.cz>
7994         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7995         (lto_symtab_merge_symbols_1): ... this one.
7996         (lto_symtab_merge_cgraph_nodes): Rename to ...
7997         (lto_symtab_merge_symbols): ... this one; simplify.
7998         * cgraph.c (same_body_aliases_done): Rename to ...
7999         (cpp_implicit_aliases_done): ... this one.
8000         (cgraph_create_function_alias): Update.
8001         (cgraph_same_body_alias): Update.
8002         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8003         (verify_edge_corresponds_to_fndecl): Simplify.
8004         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8005         (cgraph_node): Remove same_body_alias.
8006         (varpool_node): Remove alias_of and extra_name_alias.
8007         (same_body_aliases_done): Rename to ..
8008         (cpp_implicit_aliases_done): ... this one.
8009         (symtab_alias_ultimate_target): Add default parameter.
8010         (symtab_resolve_alias): New function.
8011         (fixup_same_cpp_alias_visibility): Declare.
8012         (cgraph_function_node): Add default parameter.
8013         (cgraph_node_asm_name): Likewise.
8014         (cgraph_function_or_thunk_node): Add default parameter; do
8015         not ICE when it is NULL.
8016         (varpool_variable_node): Likewise.
8017         * tree-emutls.c (create_emultls_var): Update.
8018         (ipa_lower_emutls): Update.
8019         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8020         (cgraph_reset_node): Reset alias info.
8021         (cgraph_finalize_function): Update.
8022         (fixup_same_cpp_alias_visibility): Move to symtab.c.
8023         (analyze_function): Simplify.
8024         (cgraph_process_same_body_aliases): Simplify.
8025         (analyze_functions): Fixup same body aliases.
8026         (handle_alias_pairs): Simplify.
8027         (assemble_thunk): Update.
8028         (assemble_thunks_and_aliases): Update.
8029         (output_weakrefs): Rewrite.
8030         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8031         (lto_output_varpool_node): Likewise.
8032         (compute_ltrans_boundary): Remve assert.
8033         (get_alias_symbol): New functoin.
8034         (input_node): Rewrite alias handling.
8035         (input_varpool_node): Likewise.
8036         * ipa-pure-const.c (propagate_pure_const): Fix formating.
8037         * ipa.c (process_references): Handle weakrefs correctly.
8038         (symtab_remove_unreachable_nodes): Likewise.
8039         * trans-mem.c (get_cg_data): Update.
8040         (ipa_tm_create_version_alias): Update.
8041         (ipa_tm_execute): Update.
8042         * symtab.c (dump_symtab_base): Dump aliases.
8043         (verify_symtab_base): Verify aliases.
8044         (symtab_node_availability): New function.
8045         (symtab_alias_ultimate_target): Simplify.
8046         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8047         handle all the fixup cases.
8048         (symtab_resolve_alias): New function.
8049         * passes.c (ipa_write_summaries): Handle weakrefs.
8050         * varpool.c (varpool_analyze_node): Simplify.
8051         (assemble_aliases): Update.
8052         (varpool_create_variable_alias): Simplify.
8053         (varpool_extra_name_alias): Simplify.
8054         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8055         (lto_symtab_merge_symbols): ... this one.
8057 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
8059         Revert
8060         PR rtl-optimization/57268
8061         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8062         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8064 2013-06-01  Tobias Burnus  <burnus@net-b.de>
8066         Partially reverted:
8067         2013-05-31  Tobias Burnus  <burnus@net-b.de>
8069         PR middle-end/57073
8070         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8071         further up.
8073 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8075         PR rtl-optimization/57268
8076         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8077         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8079 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8081         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8082         unordered comparison operators when -fno-trapping-math is in effect
8083         on the e500.
8084         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8085         and implement unordered comparison operators properly on the e500.
8087 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8089         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8090         for constant scalar integers.
8091         (simplify_relational_operation_1): Likewise.
8093 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
8095         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8096         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
8097         Fix comment.
8099 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8100             Igor Zamyatin  <igor.zamyatin@intel.com>
8102         Silvermont (SLM) architecture performance tuning.
8103         * config/i386/i386.h (enum ix86_tune_indices): Add
8104         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8105         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8107         * config/i386/i386.c (initial_ix86_tune_features)
8108         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8109         (ix86_lea_outperforms): Handle Silvermont tuning.
8110         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8111         call.
8112         (ix86_use_lea_for_mov): Likewise.
8113         (ix86_avoid_lea_for_addr): Likewise.
8114         (ix86_lea_for_add_ok): Likewise.
8115         (exact_dependency_1): New function.
8116         (exact_store_load_dependency): Likewise.
8117         (ix86_adjust_cost): Handle Silvermont tuning.
8118         (do_reoder_for_imul): Likewise.
8119         (swap_top_of_ready_list): New function.
8120         (ix86_sched_reorder): Changed to handle Silvermont tuning.
8122         * config/i386/i386.md (peepholes that split memory operand in fp
8123         converts): New.
8125 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8127         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8128         Remove un-necessary braces.
8130 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8132         * config/aarch64/aarch64.c (aarch64_classify_symbol):
8133         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8135 2013-05-31  Tobias Burnus  <burnus@net-b.de>
8137         PR middle-end/57073
8138         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8140 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8142         PR target/56315
8143         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8144         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8145         * config/arm/neon.md (iordi3_neon): Remove.
8146         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8147         * config/arm/predicates.md (imm_for_neon_logic_operand):
8148         Move to earlier in the file.
8149         (neon_logic_op2): Likewise.
8150         (arm_iordi_operand_neon): New predicate.
8152 2013-05-31  Richard Biener  <rguenther@suse.de>
8154         PR tree-optimization/57478
8155         PR tree-optimization/57453
8156         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8157         are life as well.
8159 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8161         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8162         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8164 2013-05-30  Tobias Burnus  <burnus@net-b.de>
8165             Thomas Koenig  <tkoenig@gcc.gnu.org>
8167         PR middle-end/57073
8168         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8169         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8171 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
8173         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8175 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
8177         * target.def (register_usage_leveling_p): New hook.
8178         * targhooks.c (default_register_usage_leveling_p): New.
8179         * targhooks.h (default_register_usage_leveling_p): New prototype.
8180         * lra-assigns.c (register_usage_leveling_p): Use the hook.
8181         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8182         * doc/tm.texi: Update.
8183         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8185 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
8187         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8188         (*insv_reg<mode>): New define_insn.
8190 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8192         PR rtl-optimization/57439
8193         * postreload.c (move2add_valid_value_p): Check that we have
8194         a zero subreg_regno_offset when accessing the register in
8195         the requested mode.
8197 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8198             Igor Zamyatin  <igor.zamyatin@intel.com>
8200         Silvermont (SLM) architecture pipeline model, tuning and
8201         insn selection.
8202         * config.gcc: Add slm config options and target.
8204         * config/i386/slm.md: New.
8206         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8208         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
8209         PROCESSOR_SLM.
8210         (ix86_target_macros_internal): Likewise.
8212         * gcc/config/i386/i386.c (slm_cost): New cost.
8213         (m_SLM): New macro flag.
8214         (initial_ix86_tune_features): Set m_SLM.
8215         (x86_accumulate_outgoing_args): Likewise.
8216         (x86_arch_always_fancy_math_387): Likewise.
8217         (processor_target_table): Add slm cost.
8218         (cpu_names): Add slm cpu name.
8219         (x86_option_override_internal): Set SLM ISA.
8220         (ix86_issue_rate): New case PROCESSOR_SLM.
8221         (ia32_multipass_dfa_lookahead): Likewise.
8222         (fold_builtin_cpu): Add slm.
8224         * config/i386/i386.h (TARGET_SLM): New target macro.
8225         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8226         (processor_type): Add PROCESSOR_SLM.
8228         * config/i386/i386.md (cpu): Add new value "slm".
8229         (slm.md): Include slm.md.
8231 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
8232             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8234         * config/arm/arm-protos.h: Add and update function protos.
8235         * config/arm/arm.c (use_simple_return_p): New added.
8236         (thumb2_expand_return): Check simple_return flag.
8237         * config/arm/arm.md: Add simple_return and conditional simple_return.
8238         * config/arm/iterators.md: Add iterator for return and simple_return.
8240 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8242         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8243         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8244         (arm_emit_vfp_multi_reg_pop): Likewise.
8245         (thumb2_emit_ldrd_pop): Likewise.
8246         (arm_expand_epilogue): Add misc REG_CFA notes.
8247         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8249 2013-05-29  Lawrence Crowl  <crowl@google.com>
8251         * config/arm/t-arm: Update for below.
8253         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8254         Change type to hash_table.  Update dependent calls and types.
8256         * config/i386/t-cygming: Update for below.
8258         * config/i386/t-interix: Update for below.
8260         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8261         Change type to hash_table.  Update dependent calls and types.
8262         (i386_find_on_wrapper_list::wrappers): Likewise.
8264         * config/ia64/t-ia64: Update for below.
8266         * config/ia64/ia64.c (bundle_state_table):
8267         Change type to hash_table.  Update dependent calls and types.
8269         * config/mips/mips.c (mips_reorg_process_insns::htab):
8270         Change type to hash_table.  Update dependent calls and types.
8272         * config/sol2.c (solaris_comdat_htab):
8273         Change type to hash_table.  Update dependent calls and types.
8275         * config/t-sol2: Update for above.
8277 2013-05-29  Teresa Johnson  <tejohnson@google.com>
8279         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8280         functions are not yet marked as defined.
8282 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
8283             Pat Haugen  <pthaugen@us.ibm.com>
8284             Peter Bergner  <bergner@vnet.ibm.com>
8286         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8287         instructions.
8288         (VEC_A): Likewise.
8289         (VEC_C): Likewise.
8290         (vrotl<mode>3): Likewise.
8291         (vashl<mode>3): Likewise.
8292         (vlshr<mode>3): Likewise.
8293         (vashr<mode>3): Likewise.
8295         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8296         support for power8 V2DI builtins.
8298         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8299         power8 V2DI builtins.
8300         (vupkhsw): Likewise.
8301         (vupklsw): Likewise.
8302         (vaddudm): Likewise.
8303         (vminsd): Likewise.
8304         (vmaxsd): Likewise.
8305         (vminud): Likewise.
8306         (vmaxud): Likewise.
8307         (vpkudum): Likewise.
8308         (vpksdss): Likewise.
8309         (vpkudus): Likewise.
8310         (vpksdus): Likewise.
8311         (vrld): Likewise.
8312         (vsld): Likewise.
8313         (vsrd): Likewise.
8314         (vsrad): Likewise.
8315         (vsubudm): Likewise.
8316         (vcmpequd): Likewise.
8317         (vcmpgtsd): Likewise.
8318         (vcmpgtud): Likewise.
8319         (vcmpequd_p): Likewise.
8320         (vcmpgtsd_p): Likewise.
8321         (vcmpgtud_p): Likewise.
8322         (vupkhsw): Likewise.
8323         (vupklsw): Likewise.
8324         (vaddudm): Likewise.
8325         (vmaxsd): Likewise.
8326         (vmaxud): Likewise.
8327         (vminsd): Likewise.
8328         (vminud): Likewise.
8329         (vpksdss): Likewise.
8330         (vpksdus): Likewise.
8331         (vpkudum): Likewise.
8332         (vpkudus): Likewise.
8333         (vrld): Likewise.
8334         (vsld): Likewise.
8335         (vsrad): Likewise.
8336         (vsrd): Likewise.
8337         (vsubudm): Likewise.
8339         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8340         support for power8 V2DI instructions.
8342         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8343         power8 V2DI instructions.  Combine pack and unpack insns to use an
8344         iterator for each mode.  Check whether a particular mode supports
8345         Altivec instructions instead of just checking TARGET_ALTIVEC.
8346         (UNSPEC_VPKUWUM): Likewise.
8347         (UNSPEC_VPKSHSS): Likewise.
8348         (UNSPEC_VPKSWSS): Likewise.
8349         (UNSPEC_VPKUHUS): Likewise.
8350         (UNSPEC_VPKSHUS): Likewise.
8351         (UNSPEC_VPKUWUS): Likewise.
8352         (UNSPEC_VPKSWUS): Likewise.
8353         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8354         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8355         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8356         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8357         (UNSPEC_VUPKHSB): Likewise.
8358         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8359         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8360         (UNSPEC_VUPKHSH): Likewise.
8361         (UNSPEC_VUPKLSB): Likewise.
8362         (UNSPEC_VUPKLSH): Likewise.
8363         (VI2): Likewise.
8364         (VI_char): Likewise.
8365         (VI_scalar): Likewise.
8366         (VI_unit): Likewise.
8367         (VP): Likewise.
8368         (VP_small): Likewise.
8369         (VP_small_lc): Likewise.
8370         (VU_char): Likewise.
8371         (add<mode>3): Likewise.
8372         (altivec_vaddcuw): Likewise.
8373         (altivec_vaddu<VI_char>s): Likewise.
8374         (altivec_vadds<VI_char>s): Likewise.
8375         (sub<mode>3): Likewise.
8376         (altivec_vsubcuw): Likewise.
8377         (altivec_vsubu<VI_char>s): Likewise.
8378         (altivec_vsubs<VI_char>s): Likewise.
8379         (altivec_vavgs<VI_char>): Likewise.
8380         (altivec_vcmpbfp): Likewise.
8381         (altivec_eq<mode>): Likewise.
8382         (altivec_gt<mode>): Likewise.
8383         (altivec_gtu<mode>): Likewise.
8384         (umax<mode>3): Likewise.
8385         (smax<mode>3): Likewise.
8386         (umin<mode>3): Likewise.
8387         (smin<mode>3): Likewise.
8388         (altivec_vpkuhum): Likewise.
8389         (altivec_vpkuwum): Likewise.
8390         (altivec_vpkshss): Likewise.
8391         (altivec_vpkswss): Likewise.
8392         (altivec_vpkuhus): Likewise.
8393         (altivec_vpkshus): Likewise.
8394         (altivec_vpkuwus): Likewise.
8395         (altivec_vpkswus): Likewise.
8396         (altivec_vpks<VI_char>ss): Likewise.
8397         (altivec_vpks<VI_char>us): Likewise.
8398         (altivec_vpku<VI_char>us): Likewise.
8399         (altivec_vpku<VI_char>um): Likewise.
8400         (altivec_vrl<VI_char>): Likewise.
8401         (altivec_vsl<VI_char>): Likewise.
8402         (altivec_vsr<VI_char>): Likewise.
8403         (altivec_vsra<VI_char>): Likewise.
8404         (altivec_vsldoi_<mode>): Likewise.
8405         (altivec_vupkhsb): Likewise.
8406         (altivec_vupkhs<VU_char>): Likewise.
8407         (altivec_vupkls<VU_char>): Likewise.
8408         (altivec_vupkhsh): Likewise.
8409         (altivec_vupklsb): Likewise.
8410         (altivec_vupklsh): Likewise.
8411         (altivec_vcmpequ<VI_char>_p): Likewise.
8412         (altivec_vcmpgts<VI_char>_p): Likewise.
8413         (altivec_vcmpgtu<VI_char>_p): Likewise.
8414         (abs<mode>2): Likewise.
8415         (vec_unpacks_hi_v16qi): Likewise.
8416         (vec_unpacks_hi_v8hi): Likewise.
8417         (vec_unpacks_lo_v16qi): Likewise.
8418         (vec_unpacks_hi_<VP_small_lc>): Likewise.
8419         (vec_unpacks_lo_v8hi): Likewise.
8420         (vec_unpacks_lo_<VP_small_lc>): Likewise.
8421         (vec_pack_trunc_v8h): Likewise.
8422         (vec_pack_trunc_v4si): Likewise.
8423         (vec_pack_trunc_<mode>): Likewise.
8425         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8426         V2DI builtins.
8427         (vec_vmaxsd): Likewise.
8428         (vec_vmaxud): Likewise.
8429         (vec_vminsd): Likewise.
8430         (vec_vminud): Likewise.
8431         (vec_vpksdss): Likewise.
8432         (vec_vpksdus): Likewise.
8433         (vec_vpkudum): Likewise.
8434         (vec_vpkudus): Likewise.
8435         (vec_vrld): Likewise.
8436         (vec_vsld): Likewise.
8437         (vec_vsrad): Likewise.
8438         (vec_vsrd): Likewise.
8439         (vec_vsubudm): Likewise.
8440         (vec_vupkhsw): Likewise.
8441         (vec_vupklsw): Likewise.
8443 2013-05-29  Jan Hubicka  <jh@suse.cz>
8445         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8446         flags; reorder rest of fields in more consistent way.
8447         (varpool_node): Remove analyzed, finalized and alias.
8448         (cgraph_ndoe): Likewise.
8449         (symtab_alias_ultimate_target): New function.
8450         (cgraph_function_node): Move offline.
8451         (cgraph_reset_node): Declare.
8452         (cgraph_comdat_can_be_unshared_p): Remove.
8453         (varpool_remove_initializer): Declare.
8454         (varpool_first_defined_variable, varpool_next_defined_variable
8455         cgraph_first_defined_function, cgraph_next_defined_function): Update.
8456         (cgraph_function_with_gimple_body_p): Update.
8457         (varpool_all_refs_explicit_p): Update.
8458         (symtab_alias_target): New function.
8459         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8460         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8461         (cgraph_function_or_thunk_node): Simplify using
8462         symtab_alias_ultimate_target.
8463         (varpool_variable_node): Likewise.
8464         * cgraph.c (cgraph_create_function_alias): Update.
8465         (cgraph_add_thunk): Update.
8466         (cgraph_remove_node): Update.
8467         (dump_cgraph_node): Do not dump removed flags.
8468         (cgraph_function_body_availability): Update.
8469         (cgraph_propagate_frequency): Update.
8470         (verify_cgraph_node): Check sanity of local flag.
8471         (cgraph_function_node): Move here from cgraph.h; revamp for
8472         cgraph_function_or_thunk_node.
8473         * lto-symtab.c (lto_varpool_replace_node): Update.
8474         (lto_symtab_resolve_can_prevail_p): Update.
8475         (lto_symtab_merge_cgraph_nodes): Update.
8476         * ipa-cp.c (determine_versionability, initialize_node_lattices,
8477         propagate_constants_accross_call, devirtualization_time_bonus,
8478         ipcp_propagate_stage): Update.
8479         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8480         * ipa-inline-transform.c (clone_inlined_nodes,
8481         preserve_function_body_p): Update.
8482         * ipa-reference.c (propagate): Update.
8483         (write_node_summary_p): Update.
8484         * toplev.c (wrapup_global_declaration_2): Update.
8485         * cgraphunit.c (cgraph_analyze_function): Rename to ...
8486         (analyze_function) ... this one.
8487         (cgraph_process_new_functions): Update.
8488         (cgraph_reset_node): Export.
8489         (cgraph_finalize_function): Update.
8490         (cgraph_add_new_function): Update.
8491         (process_function_and_variable_attributes): Update.
8492         (varpool_finalize_decl): Update.
8493         (symbol_finalized): Remove.
8494         (symbol_finalized_and_needed): Rename to ...
8495         (symbol_defined_and_needed): ... update.
8496         (cgraph_analyze_functions): Update.
8497         (handle_alias_pairs): Update.
8498         (mark_functions_to_output): Update.
8499         (assemble_thunk): Update.
8500         (output_in_order): Update.
8501         (output_weakrefs): Update.
8502         (finalize_compilation_unit): Update.
8503         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8504         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8505         input_node, input_varpool_node): Update.
8506         * dbxout.c (dbxout_expand_expr): Update.
8507         * cgraphclones.c (cgraph_clone_node): Update.
8508         (cgraph_copy_node_for_versioning): Update.
8509         (cgraph_materialize_clone): Update.
8510         (cgraph_materialize_all_clones): Update.
8511         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8512         propagate_pure_const, propagate_nothrow): Update.
8513         * lto-streamer-out.c (lto_output, write_symbol): Update.
8514         * ipa-utils.c (ipa_reverse_postorder): Update.
8515         * ipa-inline.c (can_inline_edge_p): Update.
8516         (update_caller_keys, ipa_inline): Update.
8517         * dwarf2out.c (reference_to_unused,
8518         premark_types_used_by_global_vars_helper): Update.
8519         * tree-eh.c (tree_could_trap_p): Update.
8520         * ipa-split.c (consider_split, execute_split_functions): Update.
8521         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8522         has_addr_references_p): Update; move ahead in file for better
8523         readability.
8524         (process_references): Simplify.
8525         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8526         bodies are removed.
8527         (cgraph_comdat_can_be_unshared_p): Make static.
8528         (cgraph_externally_visible_p): Update.
8529         (varpool_externally_visible_p): Update.
8530         (function_and_variable_visibility): Update.
8531         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8532         ipa_tm_mark_force_output_node): Update.
8533         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8534         estimate_edge_devirt_benefit, inline_generate_summary,
8535         inline_write_summary): Update.
8536         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8537         * ipa-prop.c (ipa_compute_jump_functions): Update.
8538         (ipa_print_node_params, ipa_prop_read_section,
8539         ipa_update_after_lto_read, read_replacements_section): Update.
8540         * varasm.c (mark_decl_referenced): Update.
8541         (assemble_alias, dump_tm_clone_pairs): Update.
8542         * tree-inline.c (copy_bb): Update.
8543         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8544         Update.
8545         * symtab.c (dump_symtab_base): Print new flags.
8546         (verify_symtab_base): Verify new flags.
8547         (symtab_alias_ultimate_target): New function.
8548         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8549         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8550         Update.
8551         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8552         Update.
8553         * i386.c (ix86_get_function_versions_dispatcher,
8554         ix86_generate_version_dispatcher_body): Update.
8555         (fold_builtin_cpu): Use varpool_add_new_variable.
8556         * varpool.c (varpool_remove_initializer): Break out from ...
8557         (varpool_remove_node): ... this one.
8558         (dump_varpool_node, varpool_node_for_asm,
8559         cgraph_variable_initializer_availability, varpool_analyze_node,
8560         varpool_assemble_decl, varpool_remove_unreferenced_decls,
8561         varpool_finalize_named_section_flags, varpool_create_variable_alias):
8562         Update.
8564 2013-05-29  Jan Hubicka  <jh@suse.cz>
8566         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8568 2013-05-29  Easwaran Raman  <eraman@google.com>
8570         PR tree-optimization/57442
8571         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8572         when control exits the main loop.
8574 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
8576         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8577         and RX600.
8578         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8579         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8580         * rx/t-rx: Add rx100 under multi library matches option for nofpu
8581         option.
8583 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8585         PR tree-optimization/57441
8586         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8587         Don't limit size of incr_vec to number of candidates.
8589 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
8591         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8592         and mips16 directories.
8593         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8594         (MULTILIB_DIRNAMES): Ditto.
8595         (MULTILIB_EXCEPTIONS): Add new exceptions.
8596         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8597         (MULTILIB_DIRNAMES): Ditto.
8598         (MULTILIB_EXCEPTIONS): Add new exceptions.
8600 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8601             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8603         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8604         SYMBOL_TINY_ABSOLUTE.
8605         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8606         SYMBOL_TINY_ABSOLUTE.
8607         (aarch64_expand_mov_immediate): Likewise.
8608         (aarch64_classify_symbol): Likewise.
8609         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8610         Permit SYMBOL_TINY_ABSOLUTE.
8611         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8613 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8614             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8616         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8617         Refactor if/switch.  Replace gcc_assert with if.
8619 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8621         * config/i386/i386.c (initial_ix86_tune_features): Enable
8622         FP Reassociation for AMD bdver1 and bdver2.
8624 2013-05-29  Martin Jambor  <mjambor@suse.cz>
8626         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8627         and IMAGPART_EXPR do not occur within other handled_components.
8629 2013-05-29  Richard Biener  <rguenther@suse.de>
8631         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8632         access on whether the use is in the BB we currently try to
8633         vectorize.
8634         (vect_bb_vectorization_profitable_p): Pass the BB we currently
8635         vectorize to vect_bb_slp_scalar_cost.
8637 2013-05-29  Richard Biener  <rguenther@suse.de>
8639         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8640         computing scalar cost offsetted by stmts that are kept live
8641         by scalar uses.
8642         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8643         for computation of scalar cost.
8645 2013-05-28  Steve Ellcey  <sellcey@mips.com>
8647         * config/mips/mips-cpus.def (mips32r2): Change processor type.
8649 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8651         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8652         array notation built-in reduction functions.
8653         * doc/passes.texi (Passes): Added documentation about changes done
8654         for Cilk Plus.
8655         * doc/invoke.texi (C Dialect Options): Added documentation about
8656         the -fcilkplus flag.
8657         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8658         (BUILTINS_DEF): Depend on cilkplus.def.
8659         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
8660         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8661         * cilkplus.def: New file.
8663 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
8665         PR rtl-optimization/57439
8666         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8668 2013-05-28  Easwaran Raman  <eraman@google.com>
8670         PR tree-optimization/57337
8671         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8672         (find_insert_point): Correctly identify the insertion point
8673         when two statements with the same UID is compared.
8675 2013-05-28  Richard Biener  <rguenther@suse.de>
8677         PR tree-optimization/56787
8678         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8679         from the list of data references.
8680         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8681         clobbers.
8682         (vect_analyze_loop_operations): Likewise.
8683         (vect_transform_loop): Remove clobbers.
8685 2013-05-28  Martin Jambor  <mjambor@suse.cz>
8687         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8688         and REALPART_EXPRs have scalar type.
8690 2013-05-28  Richard Biener  <rguenther@suse.de>
8692         PR tree-optimization/57411
8693         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8694         virtual operands.
8695         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8696         virtual operand propagation.
8698 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8700         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8701         destination register for bmasksi_vis.
8702         (vector_init_bshuffle): Likewise.
8703         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8705 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8707         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8708         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8709         mode if the instruction isn't available in the original mode.
8710         * config/sparc/sparc.opt (mfix-ut699): New option.
8711         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8712         (divdf3): Turn into expander.
8713         (divdf3_nofix): New insn.
8714         (divdf3_fix): Likewise.
8715         (divsf3): Disable if -mfix-ut699.
8716         (sqrtdf2): Turn into expander.
8717         (sqrtdf2_nofix): New insn.
8718         (sqrtdf2_fix): Likewise.
8719         (sqrtsf2): Disable if -mfix-ut699.
8721 2013-05-27  Richard Biener  <rguenther@suse.de>
8723         PR middle-end/57412
8724         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8725         block for the new loop.
8727 2013-05-27  Richard Biener  <rguenther@suse.de>
8729         PR tree-optimization/57343
8730         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8731         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8732         (number_of_iterations_cond): Do not build the folded tree.
8734 2013-05-27  Richard Biener  <rguenther@suse.de>
8736         Revert
8737         PR middle-end/57381
8738         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8739         OEP_CONSTANT_ADDRESS_OF retained.
8741         PR tree-optimization/57417
8742         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8743         for unchanged base.
8744         (set_ssa_val_to): Compare addresses using
8745         get_addr_base_and_unit_offset.
8747 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
8749         PR rtl-optimization/56833
8750         * postreload.c (move2add_record_mode): New function.
8751         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8752         (move2add_use_add2_insn): Use move2add_record_sym_value.
8753         (move2add_use_add3_insn): Likewise.
8754         (reload_cse_move2add): Use move2add_valid_value_p and
8755         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8756         affected regs by setting reg_mode to VOIDmode.
8757         (move2add_note_store): Don't pretend the inside of a SUBREG is
8758         the actual destination.  Invalidate single/leading registers by
8759         setting reg_mode to VOIDmode.
8760         Use move2add_record_sym_value, move2add_valid_value_p and
8761         move2add_record_mode.
8763 2013-05-27  Richard Biener  <rguenther@suse.de>
8765         PR tree-optimization/57396
8766         * tree-affine.c (double_int_constant_multiple_p): Properly
8767         return false for val == 0 and div != 0.
8769 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8771         * config/mips/mips.h: Use #elif in preprocessor conditions.
8773 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8775         PR target/53916
8776         * config/mips/constraints.md (kl): New constraint.
8777         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8778         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8779         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8780         from LO for MIPS16.
8781         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8783 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8785         PR target/55777
8786         * config/mips/mips.c (mips_can_inline_p): New function.
8787         (TARGET_CAN_INLINE_P): Define.
8789 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8791         * sched-int.h (ds_t, dw_t): Make unsigned int.
8792         Fix documentation that describes how all the ds_t bits are used.
8793         Reserve the last bit for delayed-branch scheduling.
8794         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8795         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8796         (gen_dep_weak_1): Remove prototype.
8797         * sched-deps.c (get_dep_weak_1): Make static.
8798         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8799         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8800         * doc/tm.texi: Regenerate.
8801         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8803 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8805         PR debug/56950
8806         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8808 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8809             Sandra Loosemore  <sandra@codesourcery.com>
8811         * config.gcc (powerpc-*): Allow native for with-cpu.
8813 2013-05-24  Jeff Law  <law@redhat.com>
8815         PR tree-optimization/57124
8816         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8817         conversion feeding a condition if the range has an overflow
8818         if -fstrict-overflow.  Add warnings for when we do make the
8819         transformation.
8821 2013-05-24  Dehao Chen  <dehao@google.com>
8823         * tree-cfg.c (locus_discrim_map): Fix the typo.
8824         (locus_discrim_hasher): Likewise.
8825         (locus_discrim_hasher::hash): Likewise.
8826         (locus_discrim_hasher::equal): Likewise.
8828 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8830         PR tree-optimization/57294
8831         * cgraph.h (ipa_record_stmt_references): Declare.
8832         * cgraphbuild.c (ipa_record_stmt_references): New function.
8833         (build_cgraph_edges): Use ipa_record_stmt_references.
8834         (rebuild_cgraph_edges): Likewise.
8835         (cgraph_rebuild_references): Likewise.
8836         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8837         associated with the old statement and build references from the
8838         newly built statements.
8839         * ipa-ref.c (ipa_remove_stmt_references): New function.
8840         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8842 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8844         * lra-constraints.c (emit_spill_move): Use smaller mode for
8845         mem-mem moves.
8846         (check_and_process_move): Consider mem-reg moves for secondary
8847         too.
8848         (curr_insn_transform): Don't lose insns emitted before for
8849         secondary memory moves.
8850         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8851         reg set up in the current insn.
8853 2013-05-24  Dehao Chen  <dehao@google.com>
8855         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8856         hash function.
8857         (locus_descrim_hasher::equal): Likewise.
8858         (build_gimple_cfg): New discriminator assignment algorithm.
8859         (make_edges): Likewise.
8860         (next_discriminator_for_locus): Likewise.
8861         (same_line_p): Likewise.
8862         (assign_discriminators): Likewise.
8863         (make_cond_expr_edges): Likewise.
8864         (make_gimple_switch_edges): Likewise.
8865         (make_goto_expr_edges): Likewise.
8866         (make_gimple_asm_edges): Likewise.
8868 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8870         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8871         X format specifier to only display bottom 16 bits.
8872         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8873         immediate to match for operand 2, since it will be masked.
8875 2013-05-24  Richard Biener  <rguenther@suse.de>
8877         PR tree-optimization/57287
8878         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8879         all SSA names that occur in abnormal PHIs.
8881 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8883         PR tree-ssa/57385
8884         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8885         that index is not negative.
8887 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8889         PR rtl-optimization/55177
8890         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8891         (simplify_byte_swapping_operation): New.
8892         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8893         (simplify_relational_operation_1): Deal with BSWAP.
8895 2013-05-23  Richard Henderson  <rth@redhat.com>
8897         PR target/56742
8898         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8899         (ix86_reorg): Call it.
8901 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8903         PR target/57379
8904         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8905         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8906         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8908 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8910         PR debug/57351
8911         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8913 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8914             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8916         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8917         * config/aarch64/constraints.md (Usa): Remove.
8918         * doc/md.texi (AArch64 Usa): Remove.
8920 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8921             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8923         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8924         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8925         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8926         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8928 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
8930         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8931         instruction (AdvSIMD).
8932         * config/aarch64/aarch64-builtins.c
8933         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8934         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8936 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8938         PR middle-end/57347
8939         * tree.h (contains_bitfld_component_ref_p): Declare.
8940         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8941         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8942         caller.
8943         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8944         not access a bit-field.  Assert all final offsets are byte-aligned.
8946 2013-05-23  Richard Biener  <rguenther@suse.de>
8948         PR tree-optimization/57380
8949         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8950         least one invariant or re-used load.
8951         * passes.c (init_optimization_passes): Move pass_phiprop before
8952         pass_forwprop.
8954 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8956         * config/aarch64/aarch64-simd.md
8957         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8959 2013-05-23  Richard Biener  <rguenther@suse.de>
8961         PR middle-end/57381
8962         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8963         OEP_CONSTANT_ADDRESS_OF retained.
8965 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8967         PR middle-end/57344
8968         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8969         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8971 2013-05-23  Richard Biener  <rguenther@suse.de>
8973         PR rtl-optimization/57341
8974         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8975         instead of true_dependence.
8977 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8979         * bb-reorder.c (branch_threshold): Make const.
8980         (exec_threshold): Ditto.
8982 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8983             Pat Haugen  <pthaugen@us.ibm.com>
8984             Peter Bergner  <bergner@vnet.ibm.com>
8986         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8987         documentation for the power8 crypto builtins.
8989         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8991         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8992         macros for defining power8 builtin functions.
8993         (BU_P8V_AV_2): Likewise.
8994         (BU_P8V_AV_P): Likewise.
8995         (BU_P8V_VSX_1): Likewise.
8996         (BU_P8V_OVERLOAD_1): Likewise.
8997         (BU_P8V_OVERLOAD_2): Likewise.
8998         (BU_CRYPTO_1): Likewise.
8999         (BU_CRYPTO_2): Likewise.
9000         (BU_CRYPTO_3): Likewise.
9001         (BU_CRYPTO_OVERLOAD_1): Likewise.
9002         (BU_CRYPTO_OVERLOAD_2): Likewise.
9003         (XSCVSPDP): Fix typo, point to the correct instruction.
9004         (VCIPHER): Add power8 crypto builtins.
9005         (VCIPHERLAST): Likewise.
9006         (VNCIPHER): Likewise.
9007         (VNCIPHERLAST): Likewise.
9008         (VPMSUMB): Likewise.
9009         (VPMSUMH): Likewise.
9010         (VPMSUMW): Likewise.
9011         (VPERMXOR_V2DI): Likewise.
9012         (VPERMXOR_V4SI: Likewise.
9013         (VPERMXOR_V8HI: Likewise.
9014         (VPERMXOR_V16QI: Likewise.
9015         (VSHASIGMAW): Likewise.
9016         (VSHASIGMAD): Likewise.
9017         (VPMSUM): Likewise.
9018         (VPERMXOR): Likewise.
9019         (VSHASIGMA): Likewise.
9021         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9022         __CRYPTO__ if the crypto instructions are available.
9023         (altivec_overloaded_builtins): Add support for overloaded power8
9024         builtins.
9026         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9027         support for power8 crypto builtins.
9028         (builtin_function_type): Likewise.
9029         (altivec_init_builtins): Add support for builtins that take vector
9030         long long (V2DI) arguments.
9032         * config/rs6000/crypto.md: New file, define power8 crypto
9033         instructions.
9035 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9036             Pat Haugen  <pthaugen@us.ibm.com>
9037             Peter Bergner  <bergner@vnet.ibm.com>
9039         * doc/invoke.texi (Option Summary): Add power8 options.
9040         (RS/6000 and PowerPC Options): Likewise.
9042         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9043         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
9044         wm, wn, wr documentation.
9046         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9047         registers if direct move instructions are enabled.
9048         (wn): New constraint for no registers.
9049         (wq): New constraint for quad word even GPR registers.
9050         (wr): New constraint if 64-bit instructions are enabled.
9051         (wv): New constraint if power8 vector instructions are enabled.
9052         (wQ): New constraint for quad word memory locations.
9054         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9055         constraint for 0..15 for crypto instructions.
9056         (gpc_reg_operand): If VSX allow registers in VSX registers as well
9057         as GPR and floating point registers.
9058         (int_reg_operand): New predicate to match only GPR registers.
9059         (base_reg_operand): New predicate to match base registers.
9060         (quad_int_reg_operand): New predicate to match even GPR registers
9061         for quad memory operations.
9062         (vsx_reg_or_cint_operand): New predicate to allow vector logical
9063         operations in both GPR and VSX registers.
9064         (quad_memory_operand): New predicate for quad memory operations.
9065         (reg_or_indexed_operand): New predicate for direct move support.
9067         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9068         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9069         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9070         (POWERPC_MASKS): Add power8 options.
9071         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9072         various options.
9074         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9075         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9077         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9078         (-mpower8-fusion): New power8 options.
9079         (-mpower8-fusion-sign): Likewise.
9080         (-mpower8-vector): Likewise.
9081         (-mcrypto): Likewise.
9082         (-mdirect-move): Likewise.
9083         (-mquad-memory): Likewise.
9085         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9086         power8.
9087         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9088         registers.
9089         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9090         (rs6000_debug_vector_unit): Add p8_vector.
9091         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9092         definitions.  Also print fusion state.
9093         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9094         (rs6000_builtin_mask_calculate): Add power8 builtin support.
9095         (rs6000_option_override_internal): Add support for power8.
9096         (rs6000_common_init_builtins): Add debugging for skipped builtins
9097         if -mdebug=builtin.
9098         (rs6000_adjust_cost): Add power8 support.
9099         (rs6000_issue_rate): Likewise.
9100         (insn_must_be_first_in_group): Likewise.
9101         (insn_must_be_last_in_group): Likewise.
9102         (force_new_group): Likewise.
9103         (rs6000_register_move_cost): Likewise.
9104         (rs6000_opt_masks): Likewise.
9106         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9107         power8 capable assembler, default to power7 options.
9108         (TARGET_DIRECT_MOVE): Likewise.
9109         (TARGET_CRYPTO): Likewise.
9110         (TARGET_P8_VECTOR): Likewise.
9111         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9112         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9113         (VECTOR_MEM_P8_VECTOR_P): Likewise.
9114         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9115         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9116         (TARGET_XSCVDPSPN): Likewise.
9117         (TARGET_XSCVSPDPN): Likewsie.
9118         (TARGET_SYNC_HI_QI): Likewise.
9119         (TARGET_SYNC_TI): Likewise.
9120         (MASK_CRYPTO): Likewise.
9121         (MASK_DIRECT_MOVE): Likewise.
9122         (MASK_P8_FUSION): Likewise.
9123         (MASK_P8_VECTOR): Likewise.
9124         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9125         temporary used by some of the direct move instructions to get two FP
9126         temporary registers does not force creation of a stack frame.
9127         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9128         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9129         that any VSX registers are tieable, even if they are also an
9130         Altivec vector mode.
9131         (r6000_reg_class_enum): Add wm, wr, wv constraints.
9132         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9133         (RS6000_BTM_CRYPTO): Likewise.
9134         (RS6000_BTM_COMMON): Likewise.
9136         * config/rs6000/rs6000.md (cpu attribute): Add power8.
9137         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9138         (enum rs6000_vector): Add power8 vector support.
9140 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9142         PR target/19599
9143         PR target/57340
9144         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9145         (any_sibcall_could_use_r3): this and handle indirect calls.
9146         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9148 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9150         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9152 2013-05-22  Richard Biener  <rguenther@suse.de>
9154         PR middle-end/57349
9155         * profile.c (branch_prob): Do not split blocks that are
9156         abnormally receiving from ECF_RETURNS_TWICE functions.
9158 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9160         * recog.c (offsettable_address_addr_space_p): Fix calculation of
9161         address mode.  Move pointer mode initialization to the same place.
9163 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9165         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9166         while it has any effect.
9168 2013-05-21  Easwaran Raman  <eraman@google.com>
9170         PR tree-optimization/57322
9171         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9172         UID of the statement added to the BB to be 1.
9174 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
9176         PR tree-optimization/57331
9177         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9178         of conversion from pointer type to integral type with integer.
9180 2013-05-21  Martin Jambor  <mjambor@suse.cz>
9182         PR lto/57289
9183         * ipa-prop.c (ipa_read_node_info): Process param_used and
9184         controlled_uses in the same order as when writing.
9186 2013-05-21  Magnus Granberg  <baldrick@free.fr>
9188         PR plugins/56754
9189         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9191 2013-05-21  Richard Biener  <rguenther@suse.de>
9193         PR tree-optimization/57318
9194         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9195         estimate stmts with side-effects as likely eliminated.
9197 2013-05-21  Richard Biener  <rguenther@suse.de>
9199         PR tree-optimization/57330
9200         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9201         preserve the call stmts fntype.
9203 2013-05-21  Richard Biener  <rguenther@suse.de>
9205         PR tree-optimization/57303
9206         * tree-ssa-sink.c (statement_sink_location): Improve killing
9207         stmt detection and properly handle self-assignments.
9209 2013-05-21  Christian Bruel  <christian.bruel@st.com>
9211         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9212         spanning registers. LEAF_REG_REMAP is supported only for contiguous
9213         registers. Set register size out of the PARALLEL loop.
9215 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
9217         PR target/56547
9218         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9219         (*fmasf4, *fmasf4_media): New insns.
9221 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9223         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9224         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9225         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9226         (mips_idiv_insns): Update the comments to say that the returned
9227         instruction counts are in units of BASE_INSN_LENGTH.
9228         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9229         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
9230         using 2 rather than 4 as the length of indirect MIPS16 and
9231         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
9232         length of a NOP.  Don't divide MIPS16 lengths by 2.
9233         (mips16_split_long_branches): Assume a branch is long if the
9234         length is greater than 4 rather than 8.
9235         * config/mips/mips.md (length): Give MIPS16 lengths directly,
9236         rather than multiplying them by 2.  Multiply instruction counts
9237         by BASE_INSN_LENGTH rather than 4.
9238         (*jump_mips16, tls_get_tp_mips16_<mode>)
9239         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9241 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9243         * config/mips/mips.md (extended_mips16): Remove branch case.
9244         (length): Remove duplicated extended_mips16 test.
9246 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9248         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9250 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
9252         * recog.h (Recog_data): Rename to...
9253         (recog_data_d): ...this.
9254         (recog_data): Update accordingly.
9255         * recog.c (recog_data): Likewise.
9256         * reload.c (save_recog_data): Likewise.
9257         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9258         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9260 2013-05-17  Julian Brown  <julian@codesourcery.com>
9262         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9263         found in a REG_EQUAL note, invalidate it.
9265 2013-05-17  Easwaran Raman  <eraman@google.com>
9267         * tree-ssa-reassoc.c (find_insert_point): New function.
9268         (insert_stmt_after): Likewise.
9269         (get_def_stmt): Likewise.
9270         (ensure_ops_are_available): Likewise.
9271         (not_dominated_by): Likewise.
9272         (rewrite_expr_tree): Do not move statements beyond what is
9273         necessary. Remove call to swap_ops_for_binary_stmt...
9274         (reassociate_bb): ... and move it here.
9275         (build_and_add_sum): Assign UIDs for new statements.
9276         (linearize_expr): Likewise.
9277         (do_reassoc): Renumber gimple statement UIDs.
9279 2013-05-17  Jan Hubicka  <jh@suse.cz>
9281         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9282         weakrefs.
9283         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9284         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9285         weakrefs.
9286         (output_weakrefs): Update.
9288 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
9289             Martin Jambor  <mjambor@suse.cz>
9291         PR middle-end/57276
9292         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9293         value that corresponds to the given aggval is found in values vector.
9295 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
9297         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9298         sse, sse2, sse3, ssse3 and sse4a flags to options.
9300 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
9302         * gengtype-state.c: (s_expr_writer): New class, to handle
9303         prettifying of output layout of s-expressions.
9304         (state_writer): New class, to write out gtype.state.
9305         (state_written_type_count): Move this variable into member data of
9306         state_writer.
9307         (s_expr_writer::s_expr_writer): New code: constructor for new class
9308         (state_writer::state_writer(): ditto
9309         (s_expr_writer::write_new_line): New function
9310         (s_expr_writer::write_any_indent): ditto
9311         (s_expr_writer::begin_s_expr): ditto
9312         (s_expr_writer::end_s_expr): ditto
9313         (write_state_fileloc): convert to method of state_writer...
9314         (state_writer:: write_state_fileloc): ...and use methods of
9315         s_expr_writer to write indentation into the gtype.state output file
9316         to visually represent the hierarchical structure of the list
9317         structures
9318         (write_state_fields): ditto, renaming to...
9319         (state_writer::write_state_fields)
9320         (write_state_a_string): ditto, renaming to...
9321         (state_writer::write_state_a_string)
9322         (write_state_string_option): ditto, renaming to...
9323         (state_writer::write_state_string_option)
9324         (write_state_type_option): ditto, renaming to...
9325         (state_writer::write_state_type_option)
9326         (write_state_nested_option): ditto, renaming to...
9327         (state_writer::write_state_nested_option)
9328         (write_state_option): ditto, renaming to...
9329         (state_writer::write_state_option)
9330         (write_state_options): ditto, renaming to...
9331         (state_writer::write_state_options)
9332         (write_state_lang_bitmap): ditto, renaming to...
9333         (state_writer::write_state_lang_bitmap)
9334         (write_state_version): ditto, renaming to...
9335         (state_writer::write_state_version)
9336         (write_state_scalar_type): ditto, renaming to...
9337         (state_writer::write_state_scalar_type)
9338         (write_state_string_type): ditto, renaming to...
9339         (state_writer::write_state_string_type)
9340         (write_state_undefined_type): ditto, renaming to...
9341         (state_writer::write_state_undefined_type)
9342         (write_state_struct_union_type): ditto, renaming to...
9343         (state_writer::write_state_struct_union_type)
9344         (write_state_struct_type): ditto, renaming to...
9345         (state_writer::write_state_struct_type)
9346         (write_state_user_struct_type): ditto, renaming to...
9347         (state_writer::write_state_user_struct_type)
9348         (write_state_lang_struct_type): ditto, renaming to...
9349         (state_writer::write_state_lang_struct_type)
9350         (write_state_param_struct_type): ditto, renaming to...
9351         (state_writer::write_state_param_struct_type)
9352         (write_state_pointer_type): ditto, renaming to...
9353         (state_writer::write_state_pointer_type)
9354         (write_state_array_type): ditto, renaming to...
9355         (state_writer::write_state_array_type)
9356         (write_state_gc_used): ditto, renaming to...
9357         (state_writer::write_state_gc_used)
9358         (write_state_common_type_content): ditto, renaming to...
9359         (state_writer::write_state_common_type_content)
9360         (write_state_type): ditto, renaming to...
9361         (state_writer::write_state_type)
9362         (write_state_pair_list): ditto, renaming to...
9363         (state_writer::write_state_pair_list)
9364         (write_state_pair): ditto, renaming to...
9365         (state_writer::write_state_pair)
9366         (write_state_typedefs): ditto, renaming to...
9367         (state_writer::write_state_typedefs)
9368         (write_state_structures): ditto, renaming to...
9369         (state_writer::write_state_structures)
9370         (write_state_param_structs): ditto, renaming to...
9371         (state_writer::write_state_param_structs)
9372         (write_state_variables): ditto, renaming to...
9373         (state_writer::write_state_variables)
9374         (write_state_srcdir): ditto, renaming to...
9375         (state_writer::write_state_srcdir)
9376         (write_state_files_list): ditto, renaming to...
9377         (state_writer::write_state_files_list)
9378         (write_state_languages): ditto, renaming to...
9379         (state_writer::write_state_languages)
9380         (write_state): create a state_writer instance and use it when
9381         writing out the state file
9383 2013-05-17  Mike Stump  <mikestump@comcast.net>
9385         PR rtl-optimization/57304
9386         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9387         accessing DF_REF_REAL_LOC.
9389 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9391         PR rtl-optimization/57281
9392         PR rtl-optimization/57300
9393         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9394         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9395         what the other splitter did if the registers are dead.
9397 2013-05-17  Richard Biener  <rguenther@suse.de>
9399         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9400         MEM_REF offsets.
9402 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9404         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9405         linking.
9407 2013-05-17  Marek Polacek  <polacek@redhat.com>
9409         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9410         length when doing non-zero store of storing '\0' to '\0'.
9412 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9414         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9415         vect_external_def oprnd1 with loop_vinfo, try to emit
9416         optional cast, negation and and stmts on the loop preheader
9417         edge instead of into the pattern def seq.
9419         PR tree-optimization/57051
9420         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9421         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9423 2013-05-16  Nick Clifton  <nickc@redhat.com>
9425         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9426         (rl78_is_naked_func): New function.
9427         (rl78_expand_prologue): Skip prologue generation for naked functions.
9428         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9429         * doc/extend.texi (naked): Add RL78 to the list of processors
9430         that supports this attribute.
9432 2013-05-16  Jeff Law  <law@redhat.com>
9434         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9436 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9438         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9439         cache parameters using detect_caches_amd also for CYRIX,
9440         NSC and TM2 signatures.
9442 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9443             Dzianis Kahanovich  <mahatma@eu.by>
9445         PR target/45359
9446         PR target/46396
9447         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9448         VIA/Centaur processors and determine their cache parameters
9449         using detect_caches_amd.
9451 2013-05-16  Teresa Johnson  <tejohnson@google.com>
9453         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9454         (rtl_verify_edges): New function.
9455         (rtl_verify_bb_insns): Ditto.
9456         (rtl_verify_bb_pointers): Ditto.
9457         (rtl_verify_bb_insn_chain): Ditto.
9458         (rtl_verify_fallthru): Ditto.
9459         (rtl_verify_bb_layout): Ditto.
9460         (rtl_verify_flow_info_1): Outline checks into new functions.
9461         (rtl_verify_flow_info): Ditto.
9463 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
9465         * cfghooks.c (copy_bbs): Add update_dominance argument.
9466         * cfghooks.h (copy_bbs): Update prototype.
9467         * tree-cfg.c (gimple_duplicate_sese_region):
9468         Add update_dominance argument.
9469         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9470         * tree-ssa-loop-ch.c (copy_loop_headers): Update
9471         gimple_duplicate_sese_region call.
9472         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9473         Update copy_bbs call.
9474         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9475         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9477 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9479         * tree-vectorizer.h (NUM_PATTERNS): Increment.
9480         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9481         vect_recog_rotate_pattern.
9482         (vect_recog_rotate_pattern): New function.
9484 2013-05-16  Jason Merrill  <jason@redhat.com>
9486         * Makefile.in (LLINKER): New variable.
9487         (mostlyclean): Remove link mutex.
9488         * configure.ac: Handle --enable-link-mutex.
9489         * lock-and-run.sh: New script.
9491 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9493         PR target/19599
9494         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9495         for NULL decl.
9497 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9499         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9501 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
9503         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9504         * config/arm/arm.c (next_consecutive_mem): New function.
9505         (gen_movmem_ldrd_strd): Likewise.
9506         * config/arm/arm.md (movmemqi): Update condition and code.
9507         (unaligned_loaddi, unaligned_storedi): New patterns.
9509 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9511         * config.gcc: Obsolete *-*-solaris2.9*.
9512         * doc/install.texi (Specific, *-*-solaris2*): Document it.
9514 2013-05-16  Richard Biener  <rguenther@suse.de>
9516         * passes.c (init_optimization_passes): Move pass_parallelize_loops
9517         earlier, after GRAPHITE transforms and IV canonicalization.
9519 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9521         * omp-low.c (extract_omp_for_data): For collapsed loops,
9522         if at least one of the loops is known at compile time to
9523         iterate zero times, set count to 0.
9524         (expand_omp_regimplify_p): New function.
9525         (expand_omp_for_generic): For collapsed loops, if at least
9526         one of the loops isn't known to iterate at least once,
9527         add runtime check with setting count to 0.
9528         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9529         For unsigned types if it isn't known at compile time that
9530         the loop will iterate at least once, add runtime check to bypass
9531         the whole loop if initial condition isn't true.
9533 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
9535         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9537 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
9539         PR middle-end/57286
9540         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9541         transformations to avoid an infinite loop.
9543 2013-05-16  Marek Polacek  <polacek@redhat.com>
9545         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9547 2013-05-15  Leif Ekblad  <leif@rdos.net>
9549         * config/i386/i386.c (ix86_decompose_address): Use
9550         DEFAULT_TLS_SEG_REG to access TLS segment register.
9551         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9552         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9553         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9555 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
9557         PR target/57260
9558         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9559         sibling calls to functions that would normally be lazily bound,
9560         unless $gp is call-clobbered.
9562 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
9564         * config/i386/i386.c (ix86_option_override_internal): Update
9565         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
9566         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9567         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9568         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9569         of TARGET_3DNOW.
9570         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9572 2013-05-15  Andreas Schwab  <schwab@suse.de>
9574         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9575         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
9576         third operand.
9578 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9580         * loop-unroll.c (report_unroll_peel): Check decision before
9581         emitting unroll/peel message.
9583 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9585         * function.h (has_bb_partition): New rtl_data flag.
9586         (bb_reorder_complete): Ditto.
9587         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9588         instead of flag_reorder_blocks_and_partition.
9589         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9590         with some enhancements.
9591         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9592         * bb-reorder.c (connect_traces): Check for has_bb_partition
9593         instead of flag_reorder_blocks_and_partition.
9594         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9595         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9596         verify_hot_cold_block_grouping.
9597         (partition_hot_cold_basic_blocks): Set has_bb_partition.
9599 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9601         PR target/19599
9602         * config/arm/predicates.md (call_insn_operand): New predicate.
9603         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
9604         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9605         if insn is not a tail call.
9606         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9607         registers.
9608         * config/arm/arm.h (enum reg_class): New caller save register class.
9609         (REG_CLASS_NAMES): Likewise.
9610         (REG_CLASS_CONTENTS): Likewise.
9611         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9612         without decls.
9614 2013-05-15  Richard Biener  <rguenther@suse.de>
9616         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9617         of MSG_OPTIMIZED_LOCATIONS.
9618         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9619         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9620         message.
9621         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9622         of MSG_OPTIMIZED_LOCATIONS.
9623         (execute_vect_slp): Likewise.
9624         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9625         (vect_create_cond_for_alias_checks): Likewise.
9626         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9627         (vect_recog_widen_mult_pattern): Likewise.
9628         (vect_recog_widen_sum_pattern): Likewise.
9629         (vect_recog_over_widening_pattern): Likewise.
9630         (vect_recog_widen_shift_pattern): Likewise.
9631         (vect_recog_vector_vector_shift_pattern): Likewise.
9632         (vect_recog_divmod_pattern): Likewise.
9633         (vect_recog_mixed_size_cond_pattern): Likewise.
9634         (vect_recog_bool_pattern): Likewise.
9635         (vect_pattern_recog_1): Likewise.
9637 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9639         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9640         non-functions to builtin_unreachable.
9641         * ipa-inline-transform.c (inline_call): Do not assert estimates were
9642         correct when new direct edges were discovered.
9644 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9646         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9647         header, print symbol order instead of node uid, print more information
9648         about indirect edge targets.
9649         (ipa_make_edge_direct_to_target): Print symbol order instead of node
9650         uids.
9651         (ipa_make_edge_direct_to_target): Likewise.
9652         (remove_described_reference): Likewise.
9653         (propagate_controlled_uses): Likewise.
9654         (ipa_print_node_params): Also print symbol order.
9655         (ipcp_transform_function): Print symbol order instead of node uids.
9656         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9657         (cgraph_get_create_real_symbol_node): Likewise.
9658         * ipa-cp.c (print_lattice): Likewise.
9659         (print_all_lattices): Likewise.
9660         (determine_versionability): Likewise.
9661         (initialize_node_lattices): Likewise.
9662         (estimate_local_effects): Likewise.
9663         (update_profiling_info): Likewise.
9664         (create_specialized_node): Likewise.
9665         (perhaps_add_new_callers): Likewise.
9666         (decide_about_value): Likewise.
9667         (decide_whether_version_node): Likewise.
9668         (identify_dead_nodes): Likewise.
9669         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9670         (dump_inline_summary): Likewise.
9671         (estimate_node_size_and_time): Likewise.
9672         (inline_analyze_function): Likewise.
9673         * ipa-inline.c (report_inline_failed_reason): Likewise.
9674         (want_early_inline_function_p): Likewise.
9675         (edge_badness): Likewise.
9676         (update_edge_key): Likewise.
9677         (inline_small_functions): Likewise.  Add dumping of order to two other
9678         dumps.
9679         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9680         instead of node uids.
9681         (propagate_pure_const): Likewise.
9682         (propagate_pure_const): Likewise.
9683         * ipa-utils.c (dump_cgraph_node_set): Likewise.
9684         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9685         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9686         of node uids.
9687         * tree-pretty-print.c (dump_function_header): Likewise.
9688         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9689         Print symbol order instead of node uids.
9691 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9693         * config/s390/s390.c (s390_register_move_cost): Don't impose the
9694         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9696 2013-05-15  Richard Biener  <rguenther@suse.de>
9698         PR tree-optimization/57275
9699         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9700         return value for fail to do runtime alias checks for gather loads.
9702 2013-05-15  Jan Hubicka  <jh@suse.cz>
9704         PR lto/57038
9705         PR lto/47375
9706         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9707         weakrefs are not external.
9708         (lto_symtab_merge_decls): Fix thinko when dealing with
9709         non-lto_symtab decls.
9710         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9711         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9712         * varpool.c (dump_varpool_node): Dump more flags.
9714 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9716         * config/i386/i386.c (processor_alias_table): Add instruction
9717         FSGSBASE for AMD bdver3 architecture.
9719 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9721         * tree.c (warn_deprecated_use): Print file:line using locus color.
9722         * diagnostic.c (diagnostic_report_current_module): Print file:line
9723         and file:line:column using locus color.
9725 2013-05-14  Mike Stump  <mikestump@comcast.net>
9727         * gdbinit.in: Add __null.
9729 2013-05-14  Mike Stump  <mikestump@comcast.net>
9731         * recog.h: Rename struct recog_data to Recog_data.
9732         * recog.c: Likewise.
9733         * reload.c (can_reload_into): Likewise.
9734         * config/picochip/picochip.c: Likewise.
9736 2013-05-14  Mike Stump  <mikestump@comcast.net>
9738         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9740 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9742         * resource.h (struct resources): Remove unch_memory member.
9743         (CLEAR_RESOURCE): Don't clear unch_memory.
9744         * resource.c (mark_referenced_resources): Don't set it.
9745         (mark_set_resources): Likewise.
9746         (mark_target_live_regs): Don't clear it.
9747         (init_resource_info): Likewise.
9748         * reorg.c (resource_conflicts_p): Don't compare it.
9749         (redundant_insn): Don't set it.
9751         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9752         Remove prototypes.
9753         * emit-rtl.c (next_label): Remove unused function.
9754         (skip_consecutive_labels, link_cc0_insns): Move to ...
9755         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9756         only place where these functions are used, and make them static.
9758 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9760         * fold-const.c (fold_negate_expr): Handle vectors.
9761         (fold_truth_not_expr): Make it static.
9762         (fold_invert_truthvalue): New static function.
9763         (invert_truthvalue_loc): Handle vectors. Do not call
9764         fold_truth_not_expr directly.
9765         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9766         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9767         (fold_comparison): Handle vector constants.
9768         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9769         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9770         * tree.h (fold_truth_not_expr): Remove declaration.
9772 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9774         * config/aarch64/aarch64-simd.md
9775         (aarch64_vcond_internal<mode>): Rename to...
9776         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9777         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9778         float modes. Clarify all iterator modes.
9779         (vcond<mode><mode>): Use new name for vcond expanders.
9780         (vcond<v_cmp_result><mode>): Likewise.
9781         (vcondu<mode><mode>: Likewise.
9782         * config/aarch64/iterators.md (VDQF_COND): New.
9784 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9786         PR bootstrap/57266
9787         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9788         variable for the shift amount. Check that we shift by non-negative
9789         amounts.
9791 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9793         PR target/42017
9794         * config/arm/arm.h (EPILOGUE_USES): Only return true
9795         for LR_REGNUM after epilogue_completed.
9797 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
9799         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9800         is error_mark_node.
9802 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9804         PR target/57261
9805         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9806         and Solaris 11+/x86 with gld.
9807         * configure: Regenerate.
9809 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9811         * expmed.c (expand_shift_1): Canonicalize rotates by
9812         constant bitsize / 2 to bitsize - 1.
9813         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9814         case ROTATERT>: Likewise.
9816         Revert:
9817         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9819         * config/i386/i386.md (rotateinv): New code attr.
9820         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9821         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9822         roll $31, %eax, etc.
9824 2013-05-14  Richard Biener  <rguenther@suse.de>
9826         PR middle-end/57235
9827         * tree-eh.c (sink_clobbers): Give up for successors with
9828         multiple predecessors and no virtual uses.
9830 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9832         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9833         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9835 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9837         PR middle-end/57251
9838         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9839         the case when both op0 and op1 have VOIDmode.
9841 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9843         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9844         in multiply-accumulate mode.
9846 2013-05-13  Guozhi Wei  <carrot@google.com>
9848         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9850 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9852         PR target/56975
9853         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9854         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9855         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9856         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9857         (ix86_expand_prologue): Likewise.
9858         (ix86_expand_split_stack_prologue): Likewise.
9859         (legitimate_pic_address_disp_p): Likewise.
9860         (legitimize_pic_address): Likewise.
9861         (legitimize_tls_address): Likewise.
9862         (legitimize_pe_coff_symbol): Likewise.
9863         (output_pic_addr_const): Likewise.
9864         (construct_plt_address): Likewise.
9865         (ix86_expand_call): Likewise.
9866         (x86_output_mi_thunk): Likewise.
9867         (x86_function_profiler): Likewise.
9869 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9871         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9872         similar switch cases.
9873         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9874         (aarch64_simd_mov_to_<mode>low): Delete.
9875         (aarch64_simd_mov_to_<mode>high): Delete.
9876         (move_lo_quad_<mode>): Add w<-r alternative.
9877         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9878         (aarch64_simd_mov_from_*): Update type attribute.
9879         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9880         statement.
9882 2013-05-13  Jan Hubicka  <jh@suse.cz>
9884         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9885         * config/i386/i386.c (ix86_compute_frame_layout,
9886         ix86_expand_epilogue, emit_i387_cw_initialization,
9887         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9888         ix86_local_alignment): Fix use of size/speed predicates.
9890 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9892         PR tree-optimization/45216
9893         PR tree-optimization/57157
9894         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9895         the (-Y) & (B - 1) variant if OP is |.
9896         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9897         return shifted.  Use (-op1) & (prec - 1) as other_amount
9898         instead of prec - op1.
9900 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9902         PR middle-end/42371
9903         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9904         (ipa_constant_data): New type.
9905         (ipa_jump_func): Use ipa_constant_data to hold information about
9906         constant jump functions.
9907         (ipa_get_jf_constant): Adjust to jump function type changes.
9908         (ipa_get_jf_constant_rdesc): New function.
9909         (ipa_param_descriptor): New field controlled_uses.
9910         (ipa_get_controlled_uses): New function.
9911         (ipa_set_controlled_uses): Likewise.
9912         * ipa-ref.h (ipa_find_reference): Declare.
9913         * ipa-prop.c (ipa_cst_ref_desc): New type.
9914         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9915         changes.
9916         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9917         New parameter cs.  Adjust all callers.
9918         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9919         (remove_described_reference): New function.
9920         (jfunc_rdesc_usable): Likewise.
9921         (try_make_edge_direct_simple_call): Decrement controlled use count,
9922         attempt to remove reference if it hits zero.
9923         (combine_controlled_uses_counters): New function.
9924         (propagate_controlled_uses): Likewise.
9925         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9926         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9927         (ipa_print_node_params): Print described use counter.
9928         (ipa_write_jump_function): Adjust to jump function type changes.
9929         (ipa_read_jump_function): New parameter CS, pass it to
9930         ipa_set_jf_constant.  Adjust caller.
9931         (ipa_write_node_info): Stream controlled use count
9932         (ipa_read_node_info): Likewise.
9933         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9934         asserting.
9935         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9936         count.  Remove cloning-added reference if it reaches zero.
9937         * ipa-ref.c (ipa_find_reference): New function.
9939 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9941         * config/i386/i386.c (processor_target_table): Modified default
9942         alignment values for AMD BD and BT architectures.
9944 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9946         * tree-vect-generic.c (uniform_vector_p): Move ...
9947         * tree.c (uniform_vector_p): ... here.
9948         * tree.h (uniform_vector_p): Declare it.
9949         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9950         into a scalar.
9952 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9954         PR tree-optimization/57230
9955         * tree-ssa-strlen.c (handle_char_store): Record length for
9956         array store from STRING_CST.
9958         PR tree-optimization/57230
9959         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9960         check.
9962 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9964         * config/epiphany/epiphany.c (epiphany_init): Check size of
9965         NUM_MODES_FOR_MODE_SWITCHING.
9966         (epiphany_expand_prologue):
9967         Remove CONFIG_REGNUM initial value handling code.
9968         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9969         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9970         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9971         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9972         Don't return 1 for FP_MODE_NONE.
9973         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9974         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9975         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9976         * config/epiphany/epiphany.md (save_config): New pattern.
9978 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9980         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9982 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9984         * config/i386/i386.md (memory): Handle sseishft1.
9985         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9986         (*vec_extractv2di_1): Ditto.
9988 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9990         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9991         saved registers.
9993 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9995         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9996         Add mthumb/march=armv7-a multilib.
9997         Add mthumb/march=armv7-r multilib.
9998         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10000 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
10002         * config/v850/t-rtems: Add more multilibs.
10004 2013-05-10  Richard Biener  <rguenther@suse.de>
10006         PR tree-optimization/57214
10007         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10008         not propagate from SSA names that occur in abnormal PHI nodes.
10010 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
10012         * stor-layout.c (element_precision): New function.
10013         * machmode.h (element_precision): Declare it.
10014         * tree.c (build_minus_one_cst): New function.
10015         (element_precision): Likewise.
10016         * tree.h (build_minus_one_cst): Declare new function.
10017         (element_precision): Likewise.
10018         * fold-const.c (operand_equal_p): Use element_precision.
10019         (fold_binary_loc): Handle vector types.
10020         * convert.c (convert_to_integer): Use element_precision.
10021         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10022         separately.
10024 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
10026         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10027         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10028         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10029         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10030         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10031         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10032         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10033         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10034         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10035         (Uuw8): New constraints.
10036         (Usb4): Move into alphabetical order.
10037         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10038         (sd8_operand, ub8_operand, uw8_operand): New predicates.
10039         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10040         previously unnamed patterns.
10041         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10042         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10043         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10044         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10045         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
10046         of specific lengths.
10048 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
10050         * config/i386/i386.md (rotateinv): New code attr.
10051         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10052         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10053         roll $31, %eax, etc.
10055         PR tree-optimization/45216
10056         PR tree-optimization/57157
10057         * tree-ssa-forwprop.c (simplify_rotate): New function.
10058         (ssa_forward_propagate_and_combine): Call it.
10060 2013-05-10  Richard Biener  <rguenther@suse.de>
10062         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10063         disable peeling when we version for aliasing.
10064         (vector_alignment_reachable_p): Honor explicit user alignment.
10065         (vect_supportable_dr_alignment): Likewise.
10066         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10067         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10068         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10069         then peeling to arrange for the cost-model check to come first.
10071 2013-05-10  Alan Modra  <amodra@gmail.com>
10073         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10074         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10075         * configure: Regenerate.
10077 2013-05-10  Alan Modra  <amodra@gmail.com>
10079         PR target/55033
10080         * varasm.c (default_elf_select_section): Move !DECL_P check..
10081         (get_named_section): ..to here before calling get_section_name.
10082         Adjust assertion.
10083         (default_section_type_flags): Add DECL_P check.
10084         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10085         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10087 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
10089         * config/epiphany/epiphany.c (epiphany_expand_prologue):
10090         When using gen_stack_adjust_str with a register offset, add a
10091         REG_FRAME_RELATED_EXPR note.
10093 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
10095         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10096         (*vec_extractv4si_zext_mem): Ditto.
10097         (*vec_extractv2di): Add 0->x and x->x alternatives.
10098         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10099         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10101 2013-05-09  Jason Merrill  <jason@redhat.com>
10103         N3639 C++1y VLA support
10104         * gimplify.c (gimplify_vla_decl): Don't touch an existing
10105         DECL_VALUE_EXPR.
10107         * tree.c (build_constructor_va): New.
10108         * tree.h: Declare it.
10110 2013-05-09  Martin Jambor  <mjambor@suse.cz>
10112         PR lto/57084
10113         * gimple-fold.c (canonicalize_constructor_val): Call
10114         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10116 2013-05-09  Jan Hubicka  <jh@suse.cz>
10117             Richard Biener  <rguenther@suse.de>
10119         PR lto/54095
10120         * symtab.c (symtab_make_decl_local): Do not add private names.
10122 2013-05-09  Jan Hubicka  <jh@suse.cz>
10124         PR lto/54095
10125         * symtab.c (insert_to_assembler_name_hash): Handle clones.
10126         (unlink_from_assembler_name_hash): Likewise.
10127         (symtab_prevail_in_asm_name_hash, symtab_register_node,
10128         symtab_unregister_node, symtab_initialize_asm_name_hash,
10129         change_decl_assembler_name): Update.
10131 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
10133         * config/aarch64/aarch64.md: New movtf split.
10134         (*movtf_aarch64): Update.
10135         (aarch64_movdi_tilow): Handle TF modes and rename to
10136         aarch64_movdi_<mode>low.
10137         (aarch64_movdi_tihigh): Handle TF modes and rename to
10138         aarch64_movdi_<mode>high
10139         (aarch64_movtihigh_di): Handle TF modes and rename to
10140         aarch64_mov<mode>high_di
10141         (aarch64_movtilow_di): Handle TF modes and rename to
10142         aarch64_mov<mode>low_di
10143         (aarch64_movtilow_tilow): Remove spurious whitespace.
10144         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10145         splits.
10146         (aarch64_print_operand): Update.
10148 2013-05-09  Alan Modra  <amodra@gmail.com>
10150         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10151         powerpc64le.
10152         * configure: Regenerate.
10154 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10156         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10157         splitter preparation statements.
10158         * config/i386/sse.md (*vec_extract* splitters): Ditto.
10159         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10160         adjust_address_nv.
10162 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10164         * gimple-ssa-strength-reduction.c (count_candidates): Change
10165         return value to int.
10166         (analyze_candidates_and_replace): Change type of length to int.
10168 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10170         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10171         (*vec_extract<mode>): Use VI12_128 mode iterator.
10172         (*vec_extract<mode>_mem): Ditto.
10173         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10174         attribute.
10176 2013-05-08  Diego Novillo  <dnovillo@google.com>
10178         PR bootstrap/54659
10180         Revert:
10181         2012-08-17  Diego Novillo  <dnovillo@google.com>
10183         PR bootstrap/54281
10184         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10185         * config.in: Regenerate.
10186         * configure: Regenerate.
10187         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10189 2013-05-08  Jan Hubicka  <jh@suse.cz>
10191         PR lto/54095
10192         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10193         * cgraph.h (symtab_node_base): Add unique_name.
10194         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10195         input_overwrite_node, input_varpool_node): Stream unique_name.
10196         * cgraphclones.c (cgraph_create_virtual_clone,
10197         cgraph_function_versioning): Set unique_name.
10198         * ipa.c (function_and_variable_visibility): Set unique_name.
10200 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10202         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10203         (alloc_cand_and_find_basis): Restrict conditional candidate
10204         processing to CAND_MULTs.
10206 2013-05-08  Jan Hubicka  <jh@suse.cz>
10208         PR lto/54095
10209         lto-symtab.c (lto_symtab_symbol_p): New function.
10210         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10211         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10212         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10213         Skip static symbols.
10215 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
10217         PR tree-optimization/57200
10218         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10219         Only call inform if the preceding warning_at returns true.
10221 2013-05-07  Han Shen  <shenhan@google.com>
10223         * cfgexpand.c (record_or_union_type_has_array_p): New function.
10224         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10225         * common.opt (fstack-protector-strong): New option.
10226         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10227         * doc/invoke.texi (Optimization Options): Document
10228         "-fstack-protector-strong".
10229         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10231 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10233         * config/mips/mips.c (mips_machine_reorg2): Return 0.
10235 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
10237         * ira.c (update_equiv_regs): Add insn having equiv memory even if
10238         it is not lhs of the insn.
10239         (setup_reg_equiv): Remove insn having equiv memory which it is not
10240         lhs of the insn.
10241         * lra-constraints.c (process_address): Try to improve generation
10242         code for address base + disp.
10243         (lra_constraints): Make correct the code for checking insn setting
10244         up backward equivalence.  Remove insn only if it is in the init
10245         insn list.
10246         * lra-eliminations.c (update_reg_eliminate): Change return value.
10247         (lra_eliminate): Use the result.
10249 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
10251         * config/i386/sse.md (ssescalarnummask): New mode attribute.
10252         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10253         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10254         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10255         register target operands.
10256         (*vec_extractv8hi_sse2): New pattern.
10257         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10258         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10259         (*vec_extract<mode>_mem): New insn and split pattern.
10261 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
10263         * config/arm/arm.c (arm_asan_shadow_offset): New function.
10264         (TARGET_ASAN_SHADOW_OFFSET): Define.
10265         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10266         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10268 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10270         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10271         (incr_vec_index): Return -1 if increment not found.
10272         (create_add_on_incoming_edge): Assert if increment not found.
10273         (record_increment): Limit number of increments recorded.
10274         (all_phi_incrs_profitable): Return false if an increment not found.
10275         (replace_profitable_candidates): Don't process increments that were
10276         not recorded.
10277         (analyze_candidates_and_replace): Limit size of incr_vec.
10279 2013-05-07  Richard Biener  <rguenther@suse.de>
10281         * calls.c (special_function_p): setjmp-like functions are leaf.
10282         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10283         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10285 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
10287         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10288         (aarch64_simd_mov<mode>): New expander.
10289         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10290         (aarch64_simd_mov_to_<mode>high): Likewise.
10291         (aarch64_simd_mov_from_<mode>low): Likewise.
10292         (aarch64_simd_mov_from_<mode>high): Likewise.
10293         (aarch64_dup_lane<mode>): Update.
10294         (aarch64_dup_lanedi): New instruction pattern.
10295         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10296         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10298 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10300         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10301         (replace_mult_candidate): Remove unnecessary argument; remove
10302         unnecessary parameter from call to introduce_cast_before_cand.
10303         (replace_unconditional_candidate): Remove unnecessary parameter
10304         from call to replace_mult_candidate.
10305         (replace_conditional_candidate): Likewise.
10306         (insert_initializers): Use make_temp_ssa_name.
10307         (introduce_cast_before_cand): Remove unnecessary argument; use
10308         make_temp_ssa_name.
10309         (replace_one_candidate): Remove unnecessary argument; remove
10310         unnecessary parameter from calls to introduce_cast_before_cand.
10311         (replace_profitable_candidates): Remove unnecessary parameters
10312         from calls to replace_one_candidate.
10314 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10316         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10317         phi def as possibly hiding a basis for a CAND_ADD whose operands
10318         have been commuted in the analysis.
10319         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10321 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10323         * config/aarch64/aarch64.md
10324         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10325         shift value between 0-4.
10327 2013-05-07  Richard Biener  <rguenther@suse.de>
10329         * double-int.h (rshift): New overload.
10330         * double-int.c (rshift): New function.
10331         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10332         (create_reference_ops_from_ref): Remove.
10333         (vn_reference_insert): Use shared ops for constructing the
10334         reference and copy it.
10336 2013-05-07  Richard Biener  <rguenther@suse.de>
10338         PR middle-end/57190
10339         * tree-eh.c (sink_clobbers): Properly propagate
10340         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10342 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
10344         PR tree-optimization/57149
10345         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10346         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10347         collect_phi_def_edges, execute_late_warn_uninitialized): Use
10348         uninit_undefined_value_p instead of ssa_undefined_value_p.
10350         PR debug/57184
10351         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10352         for modifier == EXPAND_INITIALIZER.
10354 2013-05-07  Anton Blanchard  <anton@samba.org>
10356         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10357         for powerpc64 little endian.
10358         * configure: Regenerate.
10360 2013-05-06  Graham Stott  <grahams@btinternet.com>
10362         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10363         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10364         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10365         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10367 2013-05-06  Graham Stott  <grahams@btinternet.com>
10369         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10370         codes which allow non-lvalues.
10372 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
10374         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10375         components are all 1s.
10376         (integer_minus_onep): New function.
10377         * tree.h (integer_minus_onep): Declare it.
10378         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10379         integer_minus_onep instead of integer_all_onesp.
10381 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10383         PR target/52933
10384         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10385         variations of these patterns.
10387 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10389         * config/i386/i386.md (isa): Add x64_sse4 member.
10390         (enabled): Handle x64_sse4.
10391         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10392         instruction for 64bit SSE4_1 targets.  Update insn attributes.
10393         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10394         instruction for SSE4_1 targets.  Update insn attributes.
10395         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10396         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10397         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10398         const_1 selector.
10399         (*vec_extractv4si): Rename from *sse4_1_pextrd.
10400         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10401         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10403 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10405         PR target/57108
10406         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10408 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
10410         * final.c (do_assembler_dialects): Don't handle curly braces and
10411         vertical bar escaped by % as dialect delimiters.
10412         (output_asm_insn): Print curly braces and vertical bar if escaped
10413         by % and ASSEMBLER_DIALECT defined.
10414         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10415         * doc/tm.texi: Regenerated.
10417 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10419         * config/mips/mips.c: Include tree-pass.h.
10420         (mips_reorg): Split in pre- and post-dbr_schedule parts.
10421         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10422         (pass_mips_machine_reorg2): New machine specific pass.
10423         (insert_pass_mips_machine_reorg2): New pass plugin definition.
10424         (mips_option_override): Register the new pass.
10425         * rtl.h (cleanup_barriers): Remove prototype.
10426         (dbr_schedule): Likewise.
10427         * jump.c (cleanup_barriers): Make static.
10428         * reorg.c (dbr_schedule): Likewise.
10430 2013-05-06  Richard Biener  <rguenther@suse.de>
10432         PR tree-optimization/57185
10433         * tree-parloops.c (add_field_for_reduction): Handle anonymous
10434         SSA names properly.
10436 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10438         PR target/57106
10439         * config/i386/i386.c (add_parameter_dependencies): Add dependence
10440         between "first_arg" and "insn", not "last" and "insn".
10442 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10444         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10445         (find_candidates_in_block): Re-enable slsr_process_phi.
10446         (create_phi_basis): Fix double counting of candidate adjustment.
10448 2013-05-06  Richard Biener  <rguenther@suse.de>
10450         PR middle-end/57147
10451         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10452         the edge is also fallthru, preserve it and just clear the
10453         abnormal flag.
10454         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10455         also complex, preserve that and just clear the fallthru flag.
10456         * tree-inline.c (update_ssa_across_abnormal_edges): Also
10457         update virtual operands.
10459 2013-05-06  Alan Modra  <amodra@gmail.com>
10461         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10462         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10463         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10464         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10465         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10466         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10468 2013-05-06  Alan Modra  <amodra@gmail.com>
10470         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10471         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
10472         (DEFAULT_ASM_ENDIAN): Define.
10473         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10474         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10475         Update -K PIC clause from sysv4.h.
10476         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10477         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10479 2013-05-06  Alan Modra  <amodra@gmail.com>
10481         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10482         twice for little-endian.
10483         (ashrdi3_no_power, ashrdi3): Support little-endian.
10485 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10487         PR target/55303
10488         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10489         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10490         related expanders.
10491         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10492         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10493         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10494         New predicates.
10496 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
10497             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
10499         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10500         * config/pa/pa.opt: Make mbig-switch a no-op.
10501         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10502         (CASE_VECTOR_MODE): Always return SImode.
10503         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10504         for the !TARGET_BIG_SWITCH case.
10505         * config/pa/pa-linux.h: Likewise.
10506         * config/pa/pa-openbsd.h: Likewise.
10507         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10508         * config/pa/pa.md (short_jump): Remove define_insn.
10509         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10510         (casesi0): Remove define_insn.
10511         (type): Remove btable_branch.
10512         (pa_combine_type): Likewise.
10513         (in_nullified_branch_delay): Likewise.
10514         (in_call_delay): Likewise.
10515         (define_delay): Likewise.
10516         (define_insn_reservation "Z3"): Likewise.
10517         (define_insn_reservation "Z4"): Likewise.
10518         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10519         (pa_adjust_insn_length): Remove adjustment for btable branches.
10520         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10521         and mno-big-switch
10523 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
10525         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10526         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10527         Add m->r,x alternatives.
10528         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10529         splitters using SWI48x mode iterator.
10530         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
10531         TARGET_64BIT.  Add m->x alternative.
10532         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10533         Add o->x alternative.  Enable for TARGET_SSE.
10534         (sse_storeq): Remove expander.
10535         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10536         with memory input operand.
10537         (*vec_extractv2di_1 splitter): New.
10538         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10539         * config/i386/i386.md (ssevecmodelower): New mode attribute.
10541 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10543         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
10544         (INT_LOWPART): Delete.
10545         (extract_MB): Adjust.
10546         (extract_ME): Adjust.
10547         (print_operand): Adjust.
10549 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10551         * config/rs6000/predicates.md (reg_or_add_cint_operand,
10552         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10553         (reg_or_logical_cint_operand, easy_fp_constant,
10554         logical_const_operand): Delete "CONST_DOUBLE" case.
10555         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10556         "HOST_BITS_PER_WIDE_INT == 64" test.
10557         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
10558         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10559         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10560         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10561         test.
10562         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10563         CONST_DOUBLE DImode/VOIDmode case.
10564         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10565         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10566         CONST_DOUBLE VOIDmode case.
10567         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10568         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10569         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10570         Delete CONST_DOUBLE case.
10571         (splitters for mov FMOVE64 const_double): Delete
10572         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10573         "HOST_BITS_PER_WIDE_INT >= 64" test.
10574         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10575         case.
10576         (mov DI const_double): Delete.
10578 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
10580         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10581         on op shows all bits zero in mode of a lowpart subreg, return zero.
10583 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
10585         PR target/57150
10586         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10587         to save TFmode registers and DImode to save TImode registers for
10588         caller save operations.
10589         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10590         mark being partially clobbered since they only use the first
10591         double word.
10593         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10594         and TDmode only use the upper 64-bits of each VSX register.
10596 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10598         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10599         (find_candidates_in_block): Disable slsr_process_phi.
10601 2013-05-03  Guozhi Wei  <carrot@google.com>
10603         * coverage.c (coverage_obj_init): Move the construction of gcov
10604         constructor to ...
10605         (build_init_ctor): ... here.
10607 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10609         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10610         (slsr_cand_d): Redefine def_phi.
10611         (stride_status, phi_adjust_status, count_phis_status): New enums.
10612         (find_phi_def): New.
10613         (find_basis_for_base_expr): New.
10614         (find_basis_for_candidate): Handle hidden bases.
10615         (alloc_cand_and_find_basis): Handle phi candidates.
10616         (slsr_process_phi): New.
10617         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10618         (create_mul_imm_cand): Likewise.
10619         (create_add_ssa_cand): Exclude phi base candidates.
10620         (create_add_imm_cand): Likewise.
10621         (slsr_process_cast): Likewise.
10622         (slsr_process_copy): Likewise.
10623         (find_candidates_in_block): Handle phi candidates.
10624         (dump_candidate): Likewise.
10625         (unconditional_cands): Delete.
10626         (unconditional_cands_with_known_stride_p): Delete.
10627         (phi_dependent_cand_p): New.
10628         (cand_increment): Handle phi-dependent candidates.
10629         (replace_dependent): Delete.
10630         (replace_mult_candidate): New.
10631         (replace_unconditional_candidate): New.
10632         (incr_vec_index): Move to avoid forward reference.
10633         (create_add_on_incoming_edge): New.
10634         (create_phi_basis): New.
10635         (replace_dependents): Delete.
10636         (replace_conditional_candidate): New.
10637         (phi_add_costs): New.
10638         (replace_uncond_cands_and_profitable_phis): New.
10639         (record_increment): Handle phi adjustments.
10640         (record_phi_increments): New.
10641         (record_increments): Handle phi adjustments.
10642         (phi_incr_cost): New.
10643         (lowest_cost_path): Handle phis.
10644         (total_savings): Likewise.
10645         (analyze_increments): Likewise.
10646         (ncd_with_phi): New.
10647         (ncd_of_cand_and_phis): New.
10648         (nearest_common_dominator_for_cands): Handle phi increments.
10649         (all_phi_incrs_profitable): New.
10650         (replace_profitable_candidates): Handle phi-dependent candidates.
10651         (analyze_candidates_and_replace): Likewise.
10653 2013-05-03  Teresa Johnson  <tejohnson@google.com>
10655         PR bootstrap/57154
10656         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10657         do not exceed REG_BR_PROB_BASE.
10659 2013-05-03  Jeff Law  <law@redhat.com>
10661         PR tree-optimization/57144
10662         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10663         operand of the condition will bit into the new type when eliminating
10664         a cast feeding a condition.
10666 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
10668         PR rtl-optimization/57130
10669         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10670         of COMPARE as in_code to the recursive call if needed.
10672 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
10674         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10675         (enabled): Handle new members.
10676         * config/i386/sse.md (*vec_concatv2si): Merge from
10677         *vec_concatv2si_sse2 and vec_concatv2si_sse.
10678         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10680 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
10682         PR tree-optimization/57027
10683         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10684         for fnms opportunity, check we got the prerequisite kind
10685         of tree / gimple before using accessor functions.
10687 2013-05-03  Richard Biener  <rguenther@suse.de>
10689         * double-int.h (lshift): New overload without precision
10690         and arith argument.
10691         (operator *=, operator +=, operator -=): Move ...
10692         * double-int.c (operator *=, operator +=, operator -=): ... here
10693         and implement more efficiently.
10694         (mul_double_with_sign): Remove.
10695         (lshift_double): Adjust to take unsinged shift argument, push
10696         dispatching code to callers.
10697         (mul_double_wide_with_sign): Add early out for callers that
10698         are not interested in high parts or overflow.
10699         (lshift): New function.
10700         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10701         dispatch code here.
10702         (lrotate, rrotate): Use logical shifts.
10703         * expr.c (get_inner_reference): Use lshift.
10704         * fixed-value.c (do_fixed_divide): Likewise.
10705         * tree-dfa.c (get_ref_base_and_extent): Likewise.
10706         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10707         (indirect_refs_may_alias_p): Likewise.
10708         (stmt_kills_ref_p_1): Likewise.
10710 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10712         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10714 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10716         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10717         scalar form of FABD instruction.
10719 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
10721         * lra-constraints.c (process_alt_operands): Add checking alt
10722         number to choose the best alternative.
10724 2013-05-02  Richard Biener  <rguenther@suse.de>
10726         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10727         bitmap and its handling.
10728         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10730 2013-05-02  Richard Biener  <rguenther@suse.de>
10732         PR middle-end/57140
10733         * tree-inline.c (copy_loops): Properly handle removed loops.
10734         (copy_cfg_body): Mark destination loops for fixup if source
10735         loops needed fixup.
10737 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
10739         PR target/56732
10740         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10741         generating simple_return for naked functions.
10743 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10745         PR middle-end/56988
10746         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10747         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10748         flags match.
10749         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10750         ipa_agg_replacement_value structures.
10751         (known_aggs_to_agg_replacement_list): Likewise.
10752         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10753         (read_agg_replacement_chain): Likewise.
10754         (ipcp_transform_function): Also check that by_ref flags match.
10756 2013-05-02  Richard Biener  <rguenther@suse.de>
10758         * graphds.h (struct graph): Add obstack member.
10759         * graphds.c (new_graph): Initialize obstack and allocate
10760         vertices from it.
10761         (add_edge): Allocate edge from the obstack.
10762         (free_graph): Free the obstack instead of all edges and vertices.
10764 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10766         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10767         divides.
10768         * cfg.c (update_bb_profile_for_threading): Ditto.
10769         * tree-inline.c (copy_bb): Ditto.
10770         (copy_edges_for_bb): Ditto.
10771         (initialize_cfun): Ditto.
10772         (copy_cfg_body): Ditto.
10773         (expand_call_inline): Ditto.
10774         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10775         (estimate_node_size_and_time): Ditto.
10776         (inline_merge_summary): Ditto.
10777         * cgraphclones.c (cgraph_clone_edge): Ditto.
10778         (cgraph_clone_node): Ditto.
10779         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10780         (compute_trg_info): Ditto.
10782 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10784         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10785         S reg when fp attribute set.
10786         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10788 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10790         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10791         New pattern.
10792         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10793         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10794         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10796 2013-05-02  Richard Biener  <rguenther@suse.de>
10798         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10799         (struct instantiate_cache_entry): New type.
10800         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10801         (struct instantiate_cache_type): New type.
10802         (set_instantiated_value, get_instantiated_value): Remove.
10803         (get_instantiated_value_entry): New function.
10804         (instantiate_scev_name): Use the new cache and adjust.
10805         (instantiate_scev_poly): Adjust.
10806         (instantiate_scev_binary): Likewise.
10807         (instantiate_array_ref): Likewise.
10808         (instantiate_scev_convert): Likewise.
10809         (instantiate_scev_not): Likewise.
10810         (instantiate_scev_3): Likewise.
10811         (instantiate_scev_2): Likewise.
10812         (instantiate_scev_r): Likewise.
10813         (instantiate_scev): Likewise.
10814         (resolve_mixers): Likewise.
10816 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10818         PR target/57091
10819         * lra-constraints.c (best_small_class_operands_num): Remove.
10820         (process_alt_operands): Remove small_class_operands_num.  Take
10821         small classes operands into losers and only if the operand is not
10822         matched.  Modify debugging output.
10823         (curr_insn_transform): Remove best_small_class_operands_num.
10824         Print insn name.
10826 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10828         * config/aarch64/aarch64-builtins.c
10829         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10830         * config/aarch64/aarch64-simd-builtins.def
10831         (reduc_splus_): Add new modes.
10832         (reduc_uplus_): New.
10833         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10834         (reduc_uplus_v4sf): Likewise.
10835         (reduc_splus_v4sf): Likewise.
10836         (aarch64_addv<mode>): Likewise.
10837         (reduc_uplus_<mode>): Likewise.
10838         (reduc_splus_<mode>): Likewise.
10839         (aarch64_addvv2di): Likewise.
10840         (reduc_uplus_v2di): Likewise.
10841         (reduc_splus_v2di): Likewise.
10842         (aarch64_addvv2si): Likewise.
10843         (reduc_uplus_v2si): Likewise.
10844         (reduc_splus_v2si): Likewise.
10845         (reduc_<sur>plus_<mode>): New.
10846         (reduc_<sur>plus_v2di): Likewise.
10847         (reduc_<sur>plus_v2si): Likewise.
10848         (reduc_<sur>plus_v4sf): Likewise.
10849         (aarch64_addpv4sf): Likewise.
10850         * config/aarch64/arm_neon.h
10851         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10852         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10853         add UNSPEC_SADDV, UNSPEC_UADDV.
10854         (SUADDV): New.
10855         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10857 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10859         * config/aarch64/arm_neon.h
10860         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10862 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10864         * config/aarch64/aarch64-builtins
10865         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10867 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10869         * config/aarch64/aarch64-simd-builtins.def
10870         (reduc_smax_): New.
10871         (reduc_smin_): Likewise.
10872         (reduc_umax_): Likewise.
10873         (reduc_umin_): Likewise.
10874         (reduc_smax_nan_): Likewise.
10875         (reduc_smin_nan_): Likewise.
10876         (fmax): Remove.
10877         (fmin): Likewise.
10878         (smax): Update for V2SF, V4SF and V2DF modes.
10879         (smin): Likewise.
10880         (smax_nan): New.
10881         (smin_nan): Likewise.
10882         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10883         (<su><maxmin><mode>3): ...This, refactor.
10884         (s<maxmin><mode>3): New.
10885         (<maxmin_uns><mode>3): Likewise.
10886         (reduc_<maxmin_uns>_<mode>): Refactor.
10887         (reduc_<maxmin_uns>_v4sf): Likewise.
10888         (reduc_<maxmin_uns>_v2si): Likewise.
10889         (aarch64_<fmaxmin><mode>: Remove.
10890         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10891         new builtin names.
10892         (vmin<q>_f<32,64>): Likewise.
10893         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10894         (FMAXMIN): New.
10895         (su): Add mappings for smax, smin, umax, umin.
10896         (maxmin): New.
10897         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10898         (FMAXMIN): Rename as...
10899         (FMAXMIN_UNS): ...This.
10900         (maxminv): Remove.
10901         (fmaxminv): Likewise.
10902         (fmaxmin): Likewise.
10903         (maxmin_uns): New.
10904         (maxmin_uns_op): Likewise.
10906 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10908         * config/aarch64/arm_neon.h
10909         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10910         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10911         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10913 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10915         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10916         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10918 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10920         * config/aarch64/aarch64-simd.md
10921         (vcond<mode>_internal): Handle special cases for constant masks.
10922         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10923         (vcondu<mode><mode>): Likewise.
10924         (vcond<v_cmp_result><mode>): New.
10926 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10928         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10929         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10930         * config/aarch64/aarch64-simd-builtins.def
10931         (cmeq): Update to BUILTIN_VALLDI.
10932         (cmgt): Likewise.
10933         (cmge): Likewise.
10934         (cmle): Likewise.
10935         (cmlt): Likewise.
10936         * config/aarch64/arm_neon.h
10937         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10938         to builtins or C as appropriate.
10940 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10942         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10943         (cmgeu): ...This.
10944         (cmhi): Rename to...
10945         (cmgtu): ...This.
10946         * config/aarch64/aarch64-simd.md
10947         (simd_mode): Add SF.
10948         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10949         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10950         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10951         (cstore<mode>_neg): ...This.
10952         * config/aarch64/iterators.md
10953         (VALLF): new.
10954         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10955         (COMPARISONS): New.
10956         (UCOMPARISONS): Likewise.
10957         (optab): Add missing comparisons.
10958         (n_optab): New.
10959         (cmp_1): Likewise.
10960         (cmp_2): Likewise.
10961         (CMP): Likewise.
10962         (cmp): Remove.
10963         (VCMP_S): Likewise.
10964         (VCMP_U): Likewise.
10965         (V_cmp_result): Add DF, SF modes.
10966         (v_cmp_result): Likewise.
10967         (v): Likewise.
10968         (vmtype): Likewise.
10969         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10971 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10973         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10974         define_insn to define_insn_and_split.
10975         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10976         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10977         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10978         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10979         (thumb2_negscc): Likewise.
10981 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10983         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10985 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10987         * config/arm/thumb2.md: Remove trailing whitespaces.
10989 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10991         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10992         Use gen_int_mode rather than GEN_INT.
10994 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10996         * value-prof.c (stream_in_histogram_value): Remove the strayed
10997         debug_gimple_stmt.
10999 2013-04-30  Richard Biener  <rguenther@suse.de>
11001         PR middle-end/57122
11002         * cfghooks.c (split_edge): Properly check for the loop latch edge.
11004 2013-04-30  Richard Biener  <rguenther@suse.de>
11006         PR middle-end/57107
11007         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11009 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
11011         PR rtl-optimization/56957
11012         PR rtl-optimization/57105
11013         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11014         variable.  Use just INSN_UID for determining whether an insn
11015         should be only disconnected from the insn stream.
11016         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11018 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
11020         PR tree-optimization/57104
11021         * tsan.c (instrument_expr): Don't instrument accesses to
11022         DECL_HARD_REGISTER VAR_DECLs.
11024 2013-04-30  Richard Biener  <rguenther@suse.de>
11026         * function.h (loops_for_fn): New inline function.
11027         (set_loops_for_fn): Likewise.
11028         * cfgloop.h (place_new_loop): Add struct function parameter.
11029         (get_loop): Likewise.
11030         (get_loops): Likewise.
11031         (number_of_loops): Likewise.
11032         (fel_next): Adjust.
11033         (fel_init): Likewise.
11034         * cfg.c (get_loop_copy): Adjust.
11035         * cfgloop.c (flow_loops_dump): Likewise.
11036         (record_loop_exits): Likewise.
11037         (verify_loop_structure): Likewise.
11038         * cfgloopanal.c (mark_irreducible_loops): Likewise.
11039         (estimate_reg_pressure_cost): Likewise.
11040         (mark_loop_exit_edges): Likewise.
11041         * cfgloopmanip.c (place_new_loop): Likewise.
11042         (add_loop): Likewise.
11043         (duplicate_loop): Likewise.
11044         * graph.c (draw_cfg_nodes): Likewise.
11045         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11046         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11047         (extract_affine_chrec): Likewise.
11048         (build_scop_iteration_domain): Likewise.
11049         * graphite.c (graphite_initialize): Likewise.
11050         * ira-build.c (create_loop_tree_nodes): Likewise.
11051         (more_one_region_p): Likewise.
11052         (rebuild_regno_allocno_maps): Likewise.
11053         (mark_loops_for_removal): Likewise.
11054         (mark_all_loops_for_removal): Likewise.
11055         (remove_unnecessary_regions): Likewise.
11056         (ira_build): Likewise.
11057         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11058         * loop-init.c (fix_loop_structure): Likewise.
11059         (gate_rtl_move_loop_invariants): Likewise.
11060         (gate_rtl_unswitch): Likewise.
11061         (gate_rtl_unroll_and_peel_loops): Likewise.
11062         (rtl_doloop): Likewise.
11063         * lto-streamer-in.c (input_cfg): Likewise.
11064         * lto-streamer-out.c (output_cfg): Likewise.
11065         * modulo-sched.c (sms_schedule): Likewise.
11066         * predict.c (tree_estimate_probability): Likewise.
11067         (tree_estimate_probability_driver): Likewise.
11068         (estimate_loops): Likewise.
11069         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11070         (move_sese_region_to_fn): Likewise.
11071         (debug_loop_num): Likewise.
11072         * tree-chrec.c (chrec_evaluate): Likewise.
11073         (hide_evolution_in_other_loops_than_loop): Likewise.
11074         (chrec_component_in_loop_num): Likewise.
11075         (reset_evolution_in_loop): Likewise.
11076         (evolution_function_is_invariant_rec_p): Likewise.
11077         * tree-if-conv.c (main_tree_if_conversion): Likewise.
11078         * tree-inline.c (copy_loops): Likewise.
11079         (copy_cfg_body): Likewise.
11080         (tree_function_versioning): Likewise.
11081         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11082         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11083         Likewise.
11084         (add_to_evolution_1): Likewise.
11085         (scev_const_prop): Likewise.
11086         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11087         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11088         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11089         (tree_ssa_lim_initialize): Likewise.
11090         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11091         (verify_loop_closed_ssa): Likewise.
11092         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
11093         (tree_ssa_loop_im): Likewise.
11094         (tree_ssa_loop_unswitch): Likewise.
11095         (tree_vectorize): Likewise.
11096         (check_data_deps): Likewise.
11097         (tree_ssa_loop_ivcanon): Likewise.
11098         (tree_ssa_loop_bounds): Likewise.
11099         (tree_complete_unroll): Likewise.
11100         (tree_complete_unroll_inner): Likewise.
11101         (tree_parallelize_loops): Likewise.
11102         (tree_ssa_loop_prefetch): Likewise.
11103         (tree_ssa_loop_ivopts): Likewise.
11104         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11105         * tree-vectorizer.c (vectorize_loops): Likewise.
11107 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
11109         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11110         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11111         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11112         with EABI_LINK_SPEC.
11114 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11116         PR target/44578
11117         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11118         alternative.
11120 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
11122         PR target/57097
11123         * lra-constraints.c (process_alt_operands): Discourage a bit more
11124         using memory for pseudos.  Print cost dump for alternatives.
11125         Modify cost values for conflicts with early clobbers.
11126         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11128 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11130         PR target/57098
11131         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11133 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11135         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11136         from/to S register.
11137         (movdi_aarch64): Support LDR/STR from/to D register.
11139 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11141         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11142         or higher by default.
11144 2013-04-29  Richard Biener  <rguenther@suse.de>
11146         PR middle-end/57075
11147         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11148         even if not adding abnormal edges for calls that can make
11149         abnormal gotos.
11151 2013-04-29  Richard Biener  <rguenther@suse.de>
11153         PR middle-end/57103
11154         * tree-cfg.c (move_stmt_op): Fix condition under which to update
11155         TREE_BLOCK.
11156         (move_stmt_r): Remove redundant checking.
11158 2013-04-29  Teresa Johnson  <tejohnson@google.com>
11160         PR bootstrap/57077
11161         * basic-block.h (apply_scale): New function.
11162         (apply_probability): Use apply_scale.
11163         * gimple-streamer-in.c (input_bb): Ditto.
11164         * lto-streamer-in.c (input_cfg): Ditto.
11165         * lto-cgraph.c (merge_profile_summaries): Ditto.
11166         * tree-optimize.c (execute_fixup_cfg): Ditto.
11167         * tree-inline.c (copy_bb): Update comment to use apply_scale.
11168         (copy_edges_for_bb): Ditto.
11169         (copy_cfg_body): Ditto.
11171 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11173         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11174         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11175         (tail_merge_optimize): Handle current_loops == NULL.
11177 2013-04-26  Jeff Law  <law@redhat.com>
11179         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11180         (simplify_cond_using_ranges): Generalize code to simplify
11181         COND_EXPRs where one argument is a constant and the other
11182         is an SSA_NAME created by an integral type conversion.
11184 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11186         * config/arm/arm.md (store_minmaxsi): Use only when
11187         optimize_insn_for_size_p.
11189 2013-04-29  Christian Bruel  <christian.bruel@st.com>
11191         PR target/57108
11192         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11194 2013-04-29  Richard Biener  <rguenther@suse.de>
11196         PR middle-end/57089
11197         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11198         loop tree make sure to schedule a fixup for the child as well.
11199         (expand_omp_for_generic): Properly add loops.
11200         (expand_omp_for_static_nochunk): Likewise.
11201         (expand_omp_for_static_chunk): Likewise.
11202         (expand_omp_for): For the degenerate case fixup loops.
11203         (expand_omp_sections): Fix default bb placement in loops.
11204         (expand_omp_atomic_pipeline): Properly add loops.
11206 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11208         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11210 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11212         * tree-ssa-tail-merge.c: Update header comment.
11214 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11216         * config/aarch64/arm_neon.h
11217         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11218         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11219         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11220         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11221         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11222         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11223         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11224         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11226 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11228         * config/aarch64/aarch64-simd.md
11229         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11230         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11231         fix_trunc, fixuns_trunc.
11232         (ftrunc<VDQF:mode>2): New.
11233         * config/aarch64/iterators.md (optab): Add fix, fixuns.
11234         (fix_trunc_optab): New.
11236 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11238         * config/aarch64/aarch64-builtins.c
11239         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11240         iceilf, lround, iroundf.
11242 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11244         PR target/54349
11245         * config/i386/i386.h (enum ix86_tune_indices)
11246         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11247         New, split from X86_TUNE_INTER_UNIT_MOVES.
11248         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11249         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11250         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11251         (TARGET_INTER_UNIT_MOVES): Remove.
11252         * config/i386/i386.c (initial_ix86_tune_features): Update.
11253         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11254         (ix86_expand_convert_uns_didf_sse): Use
11255         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11256         (ix86_expand_vector_init_one_nonzero): Ditto.
11257         (ix86_expand_vector_init_interleave): Ditto.
11258         (inline_secondary_memory_needed): Return true for moves from SSE class
11259         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11260         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11261         * config/i386/constraints.md (Yi, Ym): Depend on
11262         TARGET_INTER_UNIT_MOVES_TO_VEC.
11263         (Yj, Yn): New constraints.
11264         * config/i386/i386.md (*movdi_internal): Change constraints of
11265         operand 1 from Yi to Yj and from Ym to Yn.
11266         (*movsi_internal): Ditto.
11267         (*movdf_internal): Ditto.
11268         (*movsf_internal): Ditto.
11269         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11270         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11271         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11272         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11273         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11274         * config/i386/sse.md (movdi_to_sse): Ditto.
11275         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11276         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11277         TARGET_INTER_UNIT_MOVES.
11278         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11279         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11280         instead of TARGET_INTER_UNIT_MOVES.
11281         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11282         operand 1 from Yi to Yj and from Ym to Yn.
11284 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11286         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11287         (float_truncate_hi_): Likewise.
11288         (float_extend_lo_): Likewise.
11289         (float_truncate_lo_): Likewise.
11290         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11291         (aarch64_float_extend_lo_v2df): Likewise.
11292         (vec_unpacks_hi_v4sf): Likewise.
11293         (aarch64_float_truncate_lo_v2sf): Likewise.
11294         (aarch64_float_truncate_hi_v4sf): Likewise.
11295         (vec_pack_trunc_v2df): Likewise.
11296         (vec_pack_trunc_df): Likewise.
11298 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11300         * config/aarch64/aarch64-builtins.c
11301         (aarch64_fold_builtin): Fold float conversions.
11302         * config/aarch64/aarch64-simd-builtins.def
11303         (floatv2si, floatv4si, floatv2di): New.
11304         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11305         * config/aarch64/aarch64-simd.md
11306         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11307         * config/aarch64/iterators.md (FLOATUORS): New.
11308         (optab): Add float, floatuns.
11309         (su_optab): Likewise.
11311 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11313         * config/aarch64/aarch64-builtins.c
11314         (aarch64_builtin_vectorized_function): Use new names for
11315         fcvt builtins.
11316         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11317         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11318         (fcvtzu): Split as...
11319         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11320         (fcvtas): Split as...
11321         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11322         (fcvtau): Split as...
11323         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11324         (fcvtps): Split as...
11325         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11326         (fcvtpu): Split as...
11327         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11328         (fcvtms): Split as...
11329         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11330         (fcvtmu): Split as...
11331         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11332         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11333         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11334         (lfrintnusf, lfrintnudf): Likewise.
11335         * config/aarch64/aarch64-simd.md
11336         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11337         define_insn.
11338         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11339         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11340         (fcvt_pattern): Likewise.
11342 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11344         * config/aarch64/aarch64-simd.md
11345         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11346         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11348 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11350         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11351         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11352         (vrnd<a,m,n,p>_f32): Implement using builtins.
11353         (vrnd<i,x><q>_f<32, 64>): New.
11355 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11357         * config/aarch64/aarch64-builtins.c
11358         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11359         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11360         (frintz): Rename to...
11361         (btrunc): ...this.
11362         (frintp): Rename to...
11363         (ceil): ...this.
11364         (frintm): Rename to...
11365         (floor): ...this.
11366         (frinti): Rename to...
11367         (nearbyint): ...this.
11368         (frintx): Rename to...
11369         (rint): ...this.
11370         (frinta): Rename to...
11371         (round): ...this.
11372         * config/aarch64/aarch64-simd.md
11373         (aarch64_frint<frint_suffix><mode>): Delete.
11374         (<frint_pattern><mode>2): Convert to insn.
11375         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11376         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11377         (frint_pattern): Likewise.
11378         (frint_suffix): Likewise.
11380 2013-04-29  Richard Biener  <rguenther@suse.de>
11382         PR tree-optimization/57081
11383         * loop-init.c: Include tree-flow.h.
11384         (loop_optimizer_finalize): Free number of iteration estimates.
11385         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11387 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
11389         PR tree-optimization/57083
11390         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11391         non-singleton shift count range, zero extend low_bound for uns case.
11393         * config/i386/predicates.md (general_vector_operand): New predicate.
11394         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11395         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11396         if they aren't nonimmediate operands.  If their original values
11397         satisfy const_vector_equal_evenodd_p, don't shift them.
11398         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11399         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
11400         if not nonimmediate_operand.
11401         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11402         instead of register_operand.
11403         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11405 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
11407         * stor-layout.c (finalize_size_functions): Allocate a structure and
11408         reset cfun before dumping the functions.
11410 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
11412         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11414         PR target/56866
11415         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11416         use xop_pmacsdqh if uns_p.
11417         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11418         the immediate rotate count.
11420 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
11422         * rtl.h (struct rtx_def): Add comment for field jump.
11423         (LRA_SUBREG_P): New macro.
11424         * recog.c (register_operand): Check LRA_SUBREG_P.
11425         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11426         stack frame.
11427         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11428         (lra_final_code_change): Skip subreg change for operators.
11429         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11430         if there are no operand changes.
11431         * lra-constraints.c (curr_insn_set): New.
11432         (match_reload): Set LRA_SUBREG_P.
11433         (emit_spill_move): Ditto.
11434         (check_and_process_move): Use curr_insn_set. Process only single
11435         set insns.  Don't initialize sec_mem_p and change_p.
11436         (simplify_operand_subreg): Use LRA_SUBREG_P.
11437         (reg_in_class_p): New function.
11438         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11439         of #ifdef.  Add code to remove cycling.
11440         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11441         non-null disp.  Reload inner instead of disp when base and index
11442         are null.  Try to put lo_sum into register.
11443         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11444         (check_and_process_move): Move code for move cost check to
11445         simple_move_p.  Remove equiv_substitution.
11446         (simple_move_p): New function.
11447         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11448         curr_insn_set.  Call check_and_process_move only for single set
11449         insns.  Use the new function.  Move call of check_and_process_move
11450         after operand equiv substitution and address process.
11452 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11454         PR go/57045
11455         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11456         with nonlocal goto receivers or returns twice calls, ignore
11457         unininitialized values from abnormal edges to nl goto receiver
11458         or returns twice call.
11460 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11462         PR tree-optimization/57051
11463         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11464         and VEC_RSHIFT_EXPR if shift count is a multiple of element
11465         bitsize.
11467 2013-04-26  Richard Biener  <rguenther@suse.de>
11469         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11470         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
11471         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11472         (fixup_loop_arrays_after_move): New function.
11473         (move_sese_region_to_fn): Properly outline the loop tree parts
11474         of the SESE region.
11476 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
11478         * config/i386/i386.md (type, unit): Fix long lines.
11480 2013-04-26  Richard Biener  <rguenther@suse.de>
11482         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11483         (lto-streamer-out.o): Likewise.
11484         * cfgloop.c (init_loops_structure): Export, add struct function
11485         argument and adjust.
11486         (flow_loops_find): Adjust.
11487         * cfgloop.h (enum loop_estimation): Add EST_LAST.
11488         (init_loops_structure): Declare.
11489         * lto-streamer-in.c: Include cfgloop.h.
11490         (input_cfg): Input the loop tree.
11491         * lto-streamer-out.c: Include cfgloop.h.
11492         (output_cfg): Output the loop tree.
11493         (output_struct_function_base): Do not drop PROP_loops.
11495 2013-03-26  Richard Biener  <rguenther@suse.de>
11497         * tree-cfg.c (execute_build_cfg): Build the loop tree.
11498         (pass_build_cfg): Provide PROP_loops.
11499         (move_sese_region_to_fn): Remove loops that are outlined into fn
11500         for now.
11501         * tree-inline.c: Include cfgloop.h.
11502         (initialize_cfun): Do not drop PROP_loops.
11503         (copy_loops): New function.
11504         (copy_cfg_body): Copy loop structure.
11505         (tree_function_versioning): Initialize destination loop tree.
11506         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11507         (pass_parallelize_loops): Do IL verification.
11508         * loop-init.c (loop_optimizer_init): Fixup loops if required.
11509         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11510         the CFG make sure we fixup loops as well.
11511         * tree-ssa-tail-merge.c: Include cfgloop.h.
11512         (replace_block_by): When merging loop latches mark loops for fixup.
11513         * lto-streamer-out.c (output_struct_function_base): Drop
11514         PROP_loops for now.
11515         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11516         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11517         * ipa-split.c: Include cfgloop.h.
11518         (split_function): Add the new return block to the loop tree root.
11519         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11520         whether we have removed the forwarder block.
11521         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11522         * cfgloop.h (place_new_loop): Declare.
11523         * cfgloopmanip.c (place_new_loop): Export.
11524         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11525         (tree-switch-conversion.o): Likewise.
11526         (tree-complex.o): Likewise.
11527         (tree-inline.o): Likewise.
11528         (tree-ssa-tailmerge.o): Likewise.
11529         (ipa-split.o): Likewise.
11530         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11531         (tree-ssa-copy.o): Likewise.
11532         * tree-switch-conversion.c: Include cfgloop.h
11533         (process_switch): If we emit a bit-test cascade, schedule loops
11534         for fixup.
11535         * tree-complex.c: Include cfgloop.h.
11536         (expand_complex_div_wide): Properly add new basic-blocks to loops.
11537         * asan.c: Include cfgloop.h.
11538         (create_cond_insert_point): Properly add new basic-blocks to
11539         loops, schedule loop fixup.
11540         * cfgloop.c (verify_loop_structure): Check that looks are not
11541         marked for fixup.
11542         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11543         to loops.
11544         (expand_omp_for_generic): Likewise.
11545         (expand_omp_sections): Likewise.
11546         (expand_omp_atomic_pipeline): Schedule loops for fixup.
11547         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11548         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11549         is initialized, not when loops are present.
11550         * tree-parloops.c (parallelize_loops): Remove checking here.
11551         * passes.c (init_optimization_passes): Schedule a copy-propagation
11552         pass before complete unrolling of inner loops.
11554 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11556         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11557         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11558         (colorize_init): Add argument to _WIN32 version.
11559         * toplev.c: Include diagnostic-color.h.
11560         (process_options): Default to -fdiagnostics-color=auto if
11561         GCC_COLORS env var is in the environment.
11562         * common.opt (fdiagnostics-color=): Add Var and Init.
11563         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11564         env var is in the environment, the default is auto rather than never.
11566         * diagnostic.h (file_name_as_prefix): Add context argument.
11567         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
11568         the string as locus.
11569         * langhooks.c (lhd_print_error_function): Adjust caller.
11571 2013-04-25  Lawrence Crowl  <crowl@google.com>
11573         * var-tracking.c (shared_hash_def::htab):
11574         Change type to hash_table.  Update dependent calls and types.
11576 2013-04-25  Lawrence Crowl  <crowl@google.com>
11578         * Makefile.in: Update as needed below.
11580         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11581         Move declaration to after the type's method definitons.
11583         * attribs.c (htab_t scoped_attributes::attribute_hash):
11584         Change type to hash_table.  Update dependent calls and types.
11586         * bitmap.c (htab_t bitmap_desc_hash):
11587         Change type to hash_table.  Update dependent calls and types.
11589         * cselib.c (htab_t cselib_hash_table):
11590         Change type to hash_table.  Update dependent calls and types.
11592         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11593         (hash_string_slot_node): Move implementation into lto-streamer.h
11594         struct string_slot_hasher.
11595         (eq_string_slot_node): Likewise.
11597         * data-streamer-out.c: Update output_block::string_hash_table
11598         dependent calls and types.
11600         * dwarf2cfi.c (htab_t trace_index):
11601         Change type to hash_table.  Update dependent calls and types.
11603         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11604         Change type to hash_table.  Update dependent calls and types.
11605         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11606         (htab_t optimize_external_refs::map): Likewise.
11607         (htab_t output_comp_unit::extern_map): Likewise.
11608         (htab_t output_comdat_type_unit::extern_map): Likewise.
11609         (htab_t output_macinfo::macinfo_htab): Likewise.
11610         (htab_t optimize_location_lists::htab): Likewise.
11611         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11613         * except.c (htab_t ehspec_hash_type):
11614         Change type to hash_table.  Update dependent calls and types.
11615         (assign_filter_values::ttypes): Likewise.
11616         (assign_filter_values::ehspec): Likewise.
11617         (sjlj_assign_call_site_values::ar_hash): Likewise.
11618         (convert_to_eh_region_ranges::ar_hash): Likewise.
11620         * gcse.c (htab_t pre_ldst_table):
11621         Change type to hash_table.  Update dependent calls and types.
11623         * ggc-common.c (htab_t saving_htab):
11624         Change type to hash_table.  Update dependent calls and types.
11625         (htab_t loc_hash): Likewise.
11626         (htab_t ptr_hash): Likewise.
11627         (call_count): Rename ggc_call_count.
11628         (call_alloc): Rename ggc_call_alloc.
11629         (loc_descriptor): Rename make_loc_descriptor.
11630         (add_statistics): Rename ggc_add_statistics.
11632         * ggc-common.c (saving_htab):
11633         Change type to hash_table.  Update dependent calls and types.
11635         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11636         (push_gimplify_context): Likewise.
11637         (pop_gimplify_context): Likewise.
11638         (struct gimple_temp_hash_elt): Added.
11639         (struct gimplify_hasher): Likewise.
11640         (struct gimplify_ctx.temp_htab):
11641         Change type to hash_table.  Update dependent calls and types.
11643         * gimple-fold.c: Include gimplify-ctx.h.
11645         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11646         Change type to hash_table.  Update dependent calls and types.
11647         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11648         avoid potential global name collision.
11650         * gimplify.c: Include gimplify-ctx.h.
11651         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11652         (htab_t gimplify_ctx::temp_htab):
11653         Update dependent calls and types for new type hash_table.
11654         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11655         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11657         * gimplify-ctx.h: New.
11658         (struct gimple_temp_hash_elt): Move from gimplify.c.
11659         (class gimplify_hasher): New.
11660         (struct gimplify_ctx): Move from gimple.h.
11661         (htab_t gimplify_ctx::temp_htab):
11662         Change type to hash_table.  Update dependent calls and types.
11664         * graphite-clast-to-gimple.c: Include graphite-htab.h.
11665         (htab_t ivs_params::newivs_index):
11666         Change type to hash_table.  Update dependent calls and types.
11667         (htab_t ivs_params::params_index): Likewise.
11668         (htab_t print_generated_program::params_index): Likewise.
11669         (htab_t gloog::newivs_index): Likewise.
11670         (htab_t gloog::params_index): Likewise.
11672         * graphite.c: Include graphite-htab.h.
11673         4htab_t graphite_transform_loops::bb_pbb_mapping):
11674         Change type to hash_table.  Update dependent calls and types.
11676         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11677         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11678         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11680         * graphite-dependences.c: Include graphite-htab.h.
11681         (loop_is_parallel_p): Change hash table type of parameter.
11683         * graphite-htab.h: New.
11684         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11685         (extern find_pbb_via_hash): Move from graphite-poly.h.
11686         (extern loop_is_parallel_p): Move from graphite-poly.h.
11687         (extern get_loop_body_pbbs): Move from graphite-poly.h.
11689         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11690         (extern loop_is_parallel_p): Move to graphite-htab.h.
11691         (extern get_loop_body_pbbs): Move to graphite-htab.h.
11693         * haifa-sched.c (htab_t delay_htab):
11694         Change type to hash_table.  Update dependent calls and types.
11695         (htab_t delay_htab_i2): Likewise.
11697         * ira-color.c (htab_t allocno_hard_regs_htab):
11698         Change type to hash_table.  Update dependent calls and types.
11700         * ira-costs.c (htab_t cost_classes_htab):
11701         Change type to hash_table.  Update dependent calls and types.
11703         * loop-invariant.c (htab_t merge_identical_invariants::eq):
11704         Change type to hash_table.  Update dependent calls and types.
11706         * loop-iv.c (htab_t bivs):
11707         Change type to hash_table.  Update dependent calls and types.
11709         * loop-unroll.c (htab_t opt_info::insns_to_split):
11710         Change type to hash_table.  Update dependent calls and types.
11711         (htab_t opt_info::insns_with_var_to_expand): Likewise.
11713         * lto-streamer.h (struct string_slot): Move from data-streamer.h
11714         (struct string_slot_hasher): New.
11715         (htab_t output_block::string_hash_table):
11716         Change type to hash_table.  Update dependent calls and types.
11718         * lto-streamer-in.c (freeing_string_slot_hasher): New.
11719         (htab_t file_name_hash_table):
11720         Change type to hash_table.  Update dependent calls and types.
11722         * lto-streamer-out.c: Update output_block::string_hash_table dependent
11723         calls and types.
11725         * lto-streamer.c (htab_t tree_htab):
11726         Change type to hash_table.  Update dependent calls and types.
11728         * omp-low.c: Include gimplify-ctx.h.
11730         * passes.c (htab_t name_to_pass_map):
11731         Change type to hash_table.  Update dependent calls and types.
11732         (pass_traverse): Rename to passes_pass_traverse.
11734         * plugin.c (htab_t event_tab):
11735         Change type to hash_table.  Update dependent calls and types.
11737         * postreload-gcse.c (htab_t expr_table):
11738         Change type to hash_table.  Update dependent calls and types.
11739         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11741         * sese.c (debug_rename_map_1): Make extern.
11742         (htab_t copy_bb_and_scalar_dependences::rename_map):
11743         Change type to hash_table.  Update dependent calls and types.
11745         * sese.h (extern debug_rename_map): Move to .c file.
11747         * store-motion.c (htab_t store_motion_mems_table):
11748         Change type to hash_table.  Update dependent calls and types.
11750         * trans-mem.c (htab_t tm_new_mem_hash):
11751         Change type to hash_table.  Update dependent calls and types.
11753         * tree-browser.c (htab_t TB_up_ht):
11754         Change type to hash_table.  Update dependent calls and types.
11756         * tree-cfg.c (htab_t discriminator_per_locus):
11757         Change type to hash_table.  Update dependent calls and types.
11759         * tree-complex.c: Include tree-hasher.h
11760         (htab_t complex_variable_components):
11761         Change type to hash_table.  Update dependent calls and types.
11763         * tree-eh.c (htab_t finally_tree):
11764         Change type to hash_table.  Update dependent calls and types.
11766         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11767         struct int_tree_hasher.
11768         (extern int_tree_map_eq): Likewise.
11769         (uid_decl_map_hash): Removed.
11770         (extern decl_tree_map_eq): Likewise.
11772         * tree-hasher.h: New.
11773         (struct int_tree_hasher): New.
11774         (typedef int_tree_htab_type): New.
11776         * tree-inline.c: Include gimplify-ctx.h.
11778         * tree-mudflap.c: Include gimplify-ctx.h.
11780         * tree-parloops.c: Include tree-hasher.h.
11781         (htab_t eliminate_local_variables_stmt::decl_address):
11782         Change type to hash_table.  Update dependent calls and types.
11783         (htab_t separate_decls_in_region::decl_copies): Likewise.
11785         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11786         Change type to hash_table.  Update dependent calls and types.
11788         * tree-sra.c (candidates):
11789         Change type to hash_table.  Update dependent calls and types.
11791         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11792         in tree-flow.h.
11793         (int_tree_map_hash): Likewise.
11795         * tree-ssa-dom.c (htab_t avail_exprs):
11796         Change type to hash_table.  Update dependent calls and types.
11798         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11799         Change type to hash_table.  Update dependent calls and types.
11801         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11802         Change type to hash_table.  Update dependent calls and types.
11804         * tree-ssa-phiopt.c (seen_ssa_names):
11805         Change type to hash_table.  Update dependent calls and types.
11807         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11808         Change type to hash_table.  Update dependent calls and types.
11810         * tree-ssa-uncprop.c (equiv):
11811         Change type to hash_table.  Update dependent calls and types.
11813 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11815         PR rtl-optimization/57003
11816         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11817         call note_stores with kill_clobbered_value callback again after
11818         killing regs_invalidated_by_call.
11820 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11822         * config/aarch64/aarch64-simd.md
11823         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11824         (aarch64_simd_bsl<mode>): Likewise.
11825         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11827 2013-04-25  Marek Polacek  <polacek@redhat.com>
11829         PR tree-optimization/57066
11830         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11832 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11834         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11836 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11838         * config/aarch64/aarch64-builtins.c
11839         (aarch64_fold_builtin): New.
11840         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11841         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11842         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11843         * config/aarch64/arm_neon.h
11844         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11846 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11847             Tejas Belagod  <tejas.belagod@arm.com>
11849         * config/aarch64/aarch64-builtins.c
11850         (aarch64_gimple_fold_builtin): New.
11851         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11852         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11853         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11854         (addvv4sf): Update.
11855         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11857 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11859         * config/aarch64/aarch64.md
11860         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11862 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11864         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11865         (*ngcsi_uxtw): New pattern.
11867 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11868             Julian Brown  <julian@codesourcery.com>
11870         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11871         (TB_DREG): Add T_V4HF.
11872         (v4hf_UP): New macro.
11873         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11874         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11875         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11876         built-ins.
11877         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11878         (arm_vector_mode_supported_p): Handle V4HF.
11879         (arm_mangle_map): Handle V4HFmode.
11880         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11881         * config/arm/arm_neon_builtins.def: Add entries for
11882         vcvtv4hfv4sf, vcvtv4sfv4hf.
11883         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11884         (neon_vcvtv4hfv4sf): Likewise.
11885         * config/arm/neon-gen.ml: Handle half-precision floating point
11886         features.
11887         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11888         * config/arm/arm_neon.h: Regenerate.
11889         * config/arm/neon.ml (type elts): Add F16.
11890         (type vectype): Add T_float16x4, T_floatHF.
11891         (type vecmode): Add V4HF.
11892         (type features): Add Requires_FP_bit feature.
11893         (elt_width): Handle F16.
11894         (elt_class): Likewise.
11895         (elt_of_class_width): Likewise.
11896         (mode_of_elt): Refactor.
11897         (type_for_elt): Handle F16, fix error messages.
11898         (vectype_size): Handle T_float16x4.
11899         (vcvt_sh): New function.
11900         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11901         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11902         (string_of_mode): Handle V4HF.
11903         * doc/arm-neon-intrinsics.texi: Regenerate.
11905 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11907         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11908         format specifier in 'X' case.
11910 2013-04-25  Alan Modra  <amodra@gmail.com>
11912         PR target/57052
11913         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11914         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11915         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11916         Repeat for many other rotate/shift and mask patterns using subregs.
11917         Name lshiftrt insns.
11918         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11919         on WORDS_BIG_ENDIAN.
11921 2013-04-25  Alan Modra  <amodra@gmail.com>
11923         * config.gcc: Support little-endian powerpc-linux targets.
11924         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11925         (LINK_OS_LINUX_SPEC): Define.
11926         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11927         Preserve MASK_LITTLE_ENDIAN.
11928         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11929         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11930         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11931         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11932         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11933         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11934         Correct fp word order for little-endian.  Don't shift toc entries
11935         smaller than a word for little-endian.
11936         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11937         (bswapdi2 splits): Correct low-part subreg for little-endian.
11938         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11939         low/high where such is correct only for be.
11940         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11941         little-endian for -mcall-aixdesc.
11943 2013-04-25  Alan Modra  <amodra@gmail.com>
11945         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11946         replace_equiv_address_nv.
11948 2013-04-25  Alan Modra  <amodra@gmail.com>
11950         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11952 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11954         Revert:
11955         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11956         * rtl.h (struct rtx_def): ...
11958 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11960         PR rtl-optimizations/57046
11961         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11962         for multi-reg splits.
11964 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11966         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11968 2013-04-24  Sterling Augustine  <saugustine@google.com>
11970         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11971         (comp_dir_string, debug_str_dwo_section): New.
11972         (DEBUG_STR_DWO_SECTION): Rename to ...
11973         (DEBUG_DWO_STR_SECTION): ... this.
11974         (DEBUG_NORM_STR_SECTION): Delete.
11975         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11976         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11977         (find_AT_string): Move most logic to ...
11978         (find_AT_string_in_table): ... here.  New.
11979         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11980         add_skeleton_AT_string.  Delete logic.
11981         (output_skeleton_debug_sections): Remove call to
11982         add_top_level_skeleton_die_attrs.
11983         (add_comp_dir_attribute): Move logic to comp_dir_string.
11984         (dwarf2out_init): Initialize debug_str_dwo_section.
11985         (output_indirect_string): Call find_string_form.
11986         (output_indirect_strings): Rewrite.
11987         (prune_unused_types): Empty skeleton_debug_str_hash.
11988         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11989         (dwarf2out_finish):  Call output_indirect_strings.
11991 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11993         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11995 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11997         * rtl.h (struct rtx_def): Add comment for field jump.
11998         (LRA_SUBREG_P): New macro.
11999         * recog.c (register_operand): Check LRA_SUBREG_P.
12000         * lra.c (lra): Add note at the end of RTL code. Align non-empty
12001         stack frame.
12002         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12003         (lra_final_code_change): Skip subreg change for operators.
12004         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12005         if there are no operand changes.
12006         * lra-constraints.c (curr_insn_set): New.
12007         (match_reload): Set LRA_SUBREG_P.
12008         (emit_spill_move): Ditto.
12009         (check_and_process_move): Use curr_insn_set. Process only single
12010         set insns.  Don't initialize sec_mem_p and change_p.
12011         (simplify_operand_subreg): Use LRA_SUBREG_P.
12012         (reg_in_class_p): New function.
12013         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
12014         of #ifdef.  Add code to remove cycling.
12015         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12016         non-null disp.  Reload inner instead of disp when base and index
12017         are null.  Try to put lo_sum into register.
12018         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12019         (check_and_process_move): Move code for move cost check to
12020         simple_move_p.  Remove equiv_substitution.
12021         (simple_move_p): New function.
12022         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
12023         curr_insn_set.  Call check_and_process_move only for single set
12024         insns.  Use the new function.  Move call of check_and_process_move
12025         after operand equiv substitution and address process.
12027 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
12029         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12030         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12031         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12033 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12035         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12037 2013-04-24  Marek Polacek  <polacek@redhat.com>
12039         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12040         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12041         (select_loops_exit_conditions): Likewise.
12042         (number_of_iterations_for_all_loops): Likewise.
12043         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12044         (scev_analysis): Likewise.
12046 2013-04-02  Catherine Moore  <clm@codesourcery.com>
12047             Chao-ying Fu  <fu@mips.com>
12049         * config/mips/micromips.md (jraddiusp): New pattern.
12050         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12051         instruction if possible.
12053 2013-04-24  Alan Modra  <amodra@gmail.com>
12055         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12057 2013-04-24  Julian Brown  <julian@codesourcery.com>
12058             Chung-Lin Tang  <cltang@codesourcery.com>
12060         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12061         dependency behavior in enumeration type DIE generation. Add TODO note
12062         to comments about future DW_FORM_sdata/udata re-work of related code.
12064 2013-04-23  Lawrence Crowl  <crowl@google.com>
12066         * Makefile.in: Update as needed below.
12068         * hash-table.h (class hash_table):
12069         Correct many methods with parameter types compare_type to the correct
12070         value_type.  (Correct code was unlikely to notice the change.)
12071         (hash_table::elements_with_deleted) New.
12072         (class hashtable::iterator): New.
12073         (hashtable::begin()): New.
12074         (hashtable::end()): New.
12075         (FOR_EACH_HASH_TABLE_ELEMENT): New.
12077         * statistics.c (statistics_hashes):
12078         Change type to hash_table.  Update dependent calls and types.
12080         * tree-into-ssa.c (var_infos):
12081         Change type to hash_table.  Update dependent calls and types.
12083         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12084         Change type to hash_table.  Update dependent calls and types.
12086         * tree-ssa-loop-im.c (struct mem_ref.refs):
12087         Change type to hash_table.  Update dependent calls and types.
12089         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12090         Change type to hash_table.  Update dependent calls and types.
12092         * tree-ssa-sccvn.c (vn_tables_s::nary):
12093         Change type to hash_table.  Update dependent calls and types.
12094         (vn_tables_s::phis): Likewise.
12095         (vn_tables_s::references): Likewise.
12097         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12098         (vn_reference_eq): Update parameter and return types.
12100         * tree-ssa-structalias.c (pointer_equiv_class_table):
12101         Change type to hash_table.  Update dependent calls and types.
12102         (location_equiv_class_table): Likewise.
12104         * tree-vect-data-refs.c: Consequential changes for making
12105         peeling a hash_table.
12107         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12108         (destroy_loop_vec_info): Dependent hash_table update.
12110         * tree-vectorizer.h (peeling_htab):
12111         Change type to hash_table.  Update dependent calls and types.
12113 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
12115         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12116         to check the register content is equal or not.
12117         * lra-constraints.c (match_reload): Use lra_assign_reg_val
12118         to assign register content record.
12119         * lra-eliminations.c (update_reg_eliminate): Use
12120         lra_update_reg_val_offset to update register content offset.
12121         * lra-int.h (struct lra_reg): Add offset member.
12122         (lra_reg_val_equal_p): New static inline function.
12123         (lra_update_reg_val_offset): New static inline function.
12124         (lra_assign_reg_val): New static inline function.
12125         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12126         to assign register content record.
12127         (initialize_lra_reg_info_element): Initial offset to zero.
12129 2013-04-23  Catherine Moore  <clm@codesourcery.com>
12131         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12132         operands.  Record compression.
12134 2013-04-23  Xinliang David Li  <davidxl@google.com>
12136         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12138 2013-04-23  Richard Biener  <rguenther@suse.de>
12140         PR middle-end/57036
12141         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12142         parameter, only add abnormal goto edges from the copied body
12143         if the call could perform abnormal gotos.
12144         (copy_cfg_body): Adjust.
12146 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
12148         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12150 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
12152         * coretypes.h (gimple_stmt_iterator): Add struct to make
12153         compatible with C.
12155 2013-04-23  Richard Biener  <rguenther@suse.de>
12157         PR tree-optimization/57026
12158         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12159         from SSA names occuring in abnormal PHI nodes.
12161 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12163         * lto/lto.c (print_lto_report_1): Fix LTO report names.
12165 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12167         * lto/lto.c (print_lto_report_1): Declare early.
12168         (read_cgraph_and_symbols): Call print_lto_report_1 early.
12170 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12172         * common.opt (-flto-report-wpa): Add.
12173         * doc/invoke.texi (-flto-report-wpa): Add.
12174         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12175         (lto_main): dito.
12177 2013-04-22  Xinliang David Li  <davidxl@google.com>
12179         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12180         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12181         * Makefile.in: New dependency
12183             David Daney <ddaney.cavm@gmail.com>
12185         * configure.ac (gcc_cv_as_micromips_support):  Use the
12186         --fatal-warnings option.
12187         * configure: Regenerate.
12189 2013-04-22  Marek Polacek  <polacek@redhat.com>
12191         PR sanitizer/56990
12192         * tsan.c (instrument_expr): Don't instrument expression
12193         in case its size is zero.
12195 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
12197         PR target/57032
12198         Revert:
12199         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
12201         * config/alpha/alpha.c (TARGET_LRA_P): New define.
12203 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12205         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12206         (gimple_stmt_iterator): New typedef.
12207         * gimple.h (gimple_stmt_iterator): Rename to...
12208         (gimple_stmt_iterator_d): ... This.
12209         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12210         trees be valid for GIMPLE and GENERIC.
12211         (TARGET_GIMPLE_FOLD_BUILTIN): New.
12212         * gimple-fold.c (gimple_fold_call): Call target hook
12213         gimple_fold_builtin.
12214         * hooks.c (hook_bool_gsiptr_false): New.
12215         * hooks.h (hook_bool_gsiptr_false): New.
12216         * target.def (fold_stmt): New.
12217         * doc/tm.texi: Regenerate.
12219 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
12221         PR target/57018
12222         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12223         a set sp if no stack realignment.
12225 2013-04-22  Nick Clifton  <nickc@redhat.com>
12227         * config.gcc (tilegx-linux): Extend extra_objs rather than
12228         overwriting it.
12229         (tilepro-linux): Likewise.
12231 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12233         * config/aarch64/aarch64-builtins.c
12234         (CF): Remove.
12235         (CF0, CF1, CF2, CF3, CF4, CF10): New.
12236         (VAR<1-12>): Add MAP parameter.
12237         (BUILTIN_*): Likewise.
12238         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12239         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12240         (aarch64_ushl_n<mode>): Likewise.
12241         (aarch64_sshr_n<mode>): Likewise.
12242         (aarch64_ushr_n<mode>): Likewise.
12243         (aarch64_<maxmin><mode>): Likewise.
12244         (aarch64_sqrt<mode>): Likewise.
12245         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12246         (vshr<q>_n_*): Likewise.
12248 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12250         * config/aarch64/aarch64-builtins.c
12251         (aarch64_simd_builtin_type_mode): Handle SF types.
12252         (sf_UP): Define.
12253         (BUILTIN_GPF): Define.
12254         (aarch64_init_simd_builtins): Handle SF types.
12255         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12256         (frecps): Likewise.
12257         (frecpx): Likewise.
12258         * config/aarch64/aarch64-simd.md
12259         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12260         (aarch64_frecpe<mode>): New.
12261         (aarch64_frecps<mode>): Likewise.
12262         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12263         (v8type): Add frecp<esx>.
12264         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12265         (aarch64_frecps<mode>): Likewise.
12266         * config/aarch64/iterators.md (FRECP): New.
12267         (frecp_suffix): Likewise.
12268         * config/aarch64/arm_neon.h
12269         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12271 2013-04-22  Christian Bruel  <christian.bruel@st.com>
12273         PR target/56995
12274         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12275         (REG_CLASS_NAMES): Idem.
12276         (REG_CLASS_CONTENTS): Idem.
12277         (REGCLASS_HAS_FP_REG): Idem.
12278         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12279         (sh_conditional_register_usage): Idem.
12281 2013-04-21  Jeff Law  <law@redhat.com>
12283         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12284         (ssa_forward_propagate_and_combine): Use it.
12286 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12288         * lra.c: Update the flow chart diagram.
12290 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12292         PR rtl-optimization/56847
12293         * lra-constraints.c (process_alt_operands): Discourage alternative
12294         with non-matche doffsettable memory constraint fro memory with
12295         known offset.
12297 2013-04-19  Richard Biener  <rguenther@suse.de>
12299         PR tree-optimization/56982
12300         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12301         function.
12302         * gimplify.c (gimplify_call_expr): Notice special calls.
12303         (gimplify_modify_expr): Likewise.
12304         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12305         abnormal control flow receivers.
12306         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12307         in the same way as cfun->has_nonlocal_labels.
12308         (gimple_purge_dead_abnormal_call_edges): Likewise.
12309         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12310         receivers start a basic-block.
12312 2013-04-19  Richard Biener  <rguenther@suse.de>
12314         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12315         member ...
12316         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
12317         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12318         (SLP_TREE_LOAD_PERMUTATION): Add.
12319         (vect_transform_slp_perm_load): Adjust prototype.
12320         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12321         (vect_free_slp_instance): Likewise.
12322         (vect_create_new_slp_node): Likewise.
12323         (vect_supported_slp_permutation_p): Remove.
12324         (vect_slp_rearrange_stmts): Adjust.
12325         (vect_supported_load_permutation_p): Likewise.  Inline
12326         vect_supported_slp_permutation_p here.
12327         (vect_analyze_slp_instance): Compute load permutations per
12328         slp node instead of per instance.
12329         (vect_get_slp_defs): Adjust.
12330         (vect_transform_slp_perm_load): Likewise.
12331         (vect_schedule_slp_instance): Remove redundant code.
12332         (vect_schedule_slp): Remove hack for PR56270, add it ...
12333         * tree-vect-stmts.c (vectorizable_load): ... here, do not
12334         CSE loads for SLP.  Adjust.
12336 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12338         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12339         spelling in two comments.
12341 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12343         PR target/56797
12344         * config/arm/arm.c (load_multiple_sequence): Require SP
12345         as base register for loads if SP is in the register list.
12347 2013-04-19  Martin Jambor  <mjambor@suse.cz>
12349         PR tree-optimization/56718
12350         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12351         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12352         and made public.  Adjusted all callers.
12353         (ipa_intraprocedural_devirtualization): New function.
12354         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12355         (ipa_intraprocedural_devirtualization): Likewise.
12356         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12358 2013-04-19  Richard Biener  <rguenther@suse.de>
12360         PR tree-optimization/57000
12361         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12363 2013-04-19  Terry Guo  <terry.guo@arm.com>
12365         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12366         Replace with ...
12367         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
12368         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12369         (cortex_m4_fmacs): Use new reservations.
12370         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12372 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
12374         PR rtl-optimization/56999
12375         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12376         related code.
12377         (lra_coalesce): Remove split_origin_bitmap and related code.
12378         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
12379         ranges if necessary.
12381 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
12383         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12384         New array.
12385         (ix86_expand_call): Remove clobbered_registers array and use
12386         x86_64_ms_sysv_extra_clobbered_registers instead.
12387         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12388         Declare here.
12389         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12390         predicate.
12391         * config/i386/i386.md (*call_rex64_ms_sysv): Use
12392         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
12393         (*call_value_rex64_ms_sysv): Ditto.
12395 2013-04-18  Cary Coutant  <ccoutant@google.com>
12397         * dwarf2out.c (output_pubnames): Check die_perennial_p of
12398         parent instead of die_mark.
12400 2013-04-18  Diego Novillo  <dnovillo@google.com>
12402         * gimple.c (create_gimple_tmp): New.
12403         (get_expr_type): New.
12404         (build_assign): New.
12405         (build_type_cast): New.
12406         * gimple.h (enum ssa_mode): Define.
12407         (gimple_seq_set_location): New.
12408         * asan.c (build_check_stmt): Change some gimple_build_* calls
12409         to use build_assign and build_type_cast.
12411 2013-04-18  Richard Biener  <rguenther@suse.de>
12413         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12414         handle negative step.  Remove redundant checks.
12415         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12416         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12417         for negative step and grouped loads fail to vectorize.
12419 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12421         * emit-rtl.c (reset_insn_used_flags): New function.
12422         (reset_all_used_flags): Use it.
12423         (verify_insn_sharing): New function.
12424         (verify_rtl_sharing): Fix verification for SEQUENCEs.
12426 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
12428         PR tree-optimization/56984
12429         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12430         and (x >> M) >= N don't register any assertion if N << M is the
12431         minimum value.
12433 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12435         * lower-subreg.c (resolve_simple_move): If called self-recursive,
12436         do not delete_insn insns that have not yet been emitted, only
12437         unlink them with remove_insn.
12438         * df-scan.c (df_insn_delete): Revert r197492.
12440 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
12442         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12443         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12445 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12447         * config/arm/arm.md (movsicc_insn): Convert define_insn into
12448         define_insn_and_split.
12449         (and_scc,ior_scc,negscc): Likewise.
12450         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12452 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12454         * config/arm/arm.c (use_return_insn): Return 0 for targets that
12455         can benefit from using a sequence of LDRD instructions in epilogue
12456         instead of a single LDM instruction.
12458 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12460         PR 45688
12461         * doc/extend.texi: Fix typo.
12463 2013-04-17  Richard Biener  <rguenther@suse.de>
12465         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12466         (vect_build_slp_tree): ... here.
12467         (vect_build_slp_tree_1): Compute which stmts of the SLP group
12468         match.  Remove special-casing of mismatched complex loads.
12469         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12470         re-try the match with swapped commutative operands.
12471         (vect_supported_load_permutation_p): Remove special-casing of
12472         mismatched complex loads.
12473         (vect_analyze_slp_instance): Adjust.
12475 2013-04-17  Richard Biener  <rguenther@suse.de>
12477         PR rtl-optimization/56921
12478         * cfgloop.h (struct loop): Add simple_loop_desc member.
12479         (struct niter_desc): Mark with GTY(()).
12480         (simple_loop_desc): Do not use aux field but simple_loop_desc.
12481         * loop-iv.c (get_simple_loop_desc): Likewise.
12482         (free_simple_loop_desc): Likewise.
12484         Revert
12485         2013-04-16  Richard Biener  <rguenther@suse.de>
12487         PR rtl-optimization/56921
12488         * loop-init.c (pass_rtl_move_loop_invariants): Add
12489         TODO_do_not_ggc_collect to todo_flags_finish.
12490         (pass_rtl_unswitch): Same.
12491         (pass_rtl_unroll_and_peel_loops): Same.
12492         (pass_rtl_doloop): Same.
12494 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
12496         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12497         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12498         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12499         references.
12500         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12501         * tree-streamer.c (record_common_node): Adjust reference in comment.
12503 2013-04-17  Terry Guo  <terry.guo@arm.com>
12505         * config/arm/cortex-m4.md: Add a new bypass.
12507 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12509         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12510         New pattern.
12511         (*subs_<optab><mode>_multp2): New pattern.
12512         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12513         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12515 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12517         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12518         (*subs_mul_imm_<mode>): New pattern.
12520 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
12522         PR target/56948
12523         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12524         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
12525         (vsx_movti_32bit): Change j->wa to O->wa.
12527 2013-04-16  Richard Biener  <rguenther@suse.de>
12529         PR rtl-optimization/56921
12530         * loop-init.c (pass_rtl_move_loop_invariants): Add
12531         TODO_do_not_ggc_collect to todo_flags_finish.
12532         (pass_rtl_unswitch): Same.
12533         (pass_rtl_unroll_and_peel_loops): Same.
12534         (pass_rtl_doloop): Same.
12536 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12538         * config/arm/arm.c (emit_multi_reg_push): New declaration
12539         for an existing function.
12540         (arm_emit_strd_push): New function.
12541         (arm_expand_prologue): Used here.
12542         (arm_emit_ldrd_pop): New function.
12543         (arm_expand_epilogue): Used here.
12544         (arm_get_frame_offsets): Update condition.
12545         (arm_emit_multi_reg_pop): Add a special case for load of a single
12546         register with writeback.
12548 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
12550         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12551         description.
12553 2013-04-16  Richard Biener  <rguenther@suse.de>
12555         PR tree-optimization/56756
12556         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12557         (first_mem_ref_loc): New.
12558         (execute_sm): Place the load temporarily before a previous
12559         access instead of in the latch edge to ensure its SSA dependencies
12560         are defined at points dominating the load.
12562 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
12564         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12565         correct fix by moving header and footer insn to the footer of
12566         the merged basic block.  Clear BB_END of the merged-away block.
12568         PR middle-end/43631
12569         * emit-rtl.c (make_note_raw): New function.
12570         (link_insn_into_chain): New static inline function.
12571         (add_insn): Use it.
12572         (add_insn_before, add_insn_after): Factor insn chain linking code...
12573         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12574         using link_insn_into_chain.
12575         (note_outside_basic_block_p): New helper function for emit_note_after
12576         and emit_note_before.
12577         (emit_note_after): Use nobb variant of add_insn_after if the note
12578         should not be contained in a basic block.
12579         (emit_note_before): Use nobb variant of add_insn_before if the note
12580         should not be contained in a basic block.
12581         (emit_note_copy): Use make_note_raw.
12582         (emit_note): Likewise.
12583         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12584         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12585         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12586         the moved barrier the tail of the basic block it follows.
12587         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12589 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
12591         PR tree-optimization/56962
12592         * gimple-ssa-strength-reduction.c (record_increment): Only set
12593         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12594         either rhs1 or rhs2 is equal to c->base_expr.
12596 2013-04-15  Richard Biener  <rguenther@suse.de>
12598         PR tree-optimization/56933
12599         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12600         member.
12601         (GROUP_READ_WRITE_DEPENDENCE): Remove.
12602         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12603         * tree-vect-data-refs.c (vect_analyze_group_access): Move
12604         dependence check ...
12605         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12606         ... here.
12607         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12608         GROUP_READ_WRITE_DEPENDENCE.
12610 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12612         * emit-rtl.c (reset_all_used_flags): New function.
12613         (verify_rtl_sharing): Call reset_all_used_flags before and after
12614         performing the checks.
12616 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12618         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12619         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12620         * config/arm/constraints.md (De): New constraint.
12621         * config/arm/neon.md (anddi3_neon): Delete.
12622         (neon_vand<mode>): Expand to standard anddi3 pattern.
12623         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12624         Move earlier in the file.
12625         (neon_inv_logic_op2): Likewise.
12626         (arm_anddi_operand_neon): New predicate.
12628 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12630         * configure.ac (gcc_cv_ld_as_needed): Set
12631         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12632         Use -z ignore, -z record on *-*-solaris2*.
12633         (HAVE_LD_AS_NEEDED): Update comment.
12634         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12635         * configure: Regenerate.
12636         * config.in: Regenerate.
12637         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12638         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12639         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12640         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12641         equivalents.  Fix markup.
12642         * doc/tm.texi: Regenerate.
12644 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
12646         * config/i386/i386.opt: New option mstack-protector-guard=.
12647         * config/i386/i386-opts.h: Add enum stack_protector_guard.
12648         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12649         TARGET_SSP_TLS_GUARD.
12650         * config/i386/i386.c (ix86_option_override_internal): Set
12651         ix86_stack_protector_guard.
12652         * config/i386/i386.md (stack_protect_set): Enable for
12653         TARGET_SSP_TLS_GUARD only.
12654         (stack_protect_set_<mode>): Ditto.
12655         (stack_protect_test): Ditto.
12656         (stack_protect_test_<mode>): Ditto.
12657         * doc/invoke.texi (i386 Option): Document.
12659 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
12661         PR target/56890
12662         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12663         (S_MODES): Set H_MODE bit.
12664         (SF_MODES): Set only S_MODE and SF_MODE bits.
12665         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12666         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12667         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12668         <MODE_FLOAT>: Likewise.
12670 2013-04-15  Joey Ye  <joey.ye@arm.com>
12672         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12674 2013-04-15  Joey Ye  <joey.ye@arm.com>
12676         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12677         for real far jump.
12678         (thumb_far_jump_used_p): Count instruction size and set
12679         far_jump_used.
12681 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
12683         * reorg.c (fill_simple_delay_slots): Reindent block of code.
12684         * resource.c (mark_target_live_regs): Reformat conditional block.
12686 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12688         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12689         notes, they are emitted only just before final.
12690         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12692 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12694         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12695         * cfgrtl.c (delete_insn): Call it here instead.
12696         * lra-spills.c (lra_final_code_change): Use delete_insn.
12697         * haifa-sched.c (sched_remove_insn): Likewise.
12698         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12699         returning to the nop pool.
12700         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12701         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
12703 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12705         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12706         * doc/tm.texi: Regenerated.
12708 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
12710         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12711         QImode checks.
12713 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12715         * df-core.c (df_find_def): Compare register numbers.
12716         (df_find_use): Likewise.
12718 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
12720         PR target/56903
12721         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12722         lra_in_progress for return.
12724 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12726         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12727         define_insn into define_insn_and_split and emit movsicc patterns.
12729 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12731         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12733 2013-04-12  Richard Biener  <rguenther@suse.de>
12735         * tree-pass.h (TODO_do_not_ggc_collect): New.
12736         * passes.c (execute_one_ipa_transform_pass): Honor
12737         TODO_do_not_ggc_collect.
12738         (execute_one_pass): Likewise.
12740         Revert
12741         2013-04-10  Richard Biener  <rguenther@suse.de>
12743         * passes.c (init_optimization_passes): Remove reload pass.
12744         * ira.c (do_reload): Merge into ...
12745         (ira): ... this.
12746         (rest_of_handle_reload): Remove.
12747         (pass_reload): Likewise.
12748         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12749         of reload for vzeroupper pass placement.
12751 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12753         PR tree-optimization/56918
12754         PR tree-optimization/56920
12755         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12756         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12757         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12758         use wide_mul_with_sign method.
12760 2013-04-12  Richard Biener  <rguenther@suse.de>
12762         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12763         not be considered a gimple constant.
12765 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12767         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12768         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12769         operations.
12771 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12772             Jakub Jelinek  <jakub@redhat.com>
12774         * opts.c: Include diagnostic-color.h.
12775         (common_handle_option): Handle OPT_fdiagnostics_color_.
12776         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12777         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12778         (diagnostic-color.o): New.
12779         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12780         (diagnostic_color_rule): New enum.
12781         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12782         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12783         the location string.
12784         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12785         either NULL, or color kind.
12786         * diagnostic-color.c: New file.
12787         * diagnostic-color.h: New file.
12788         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12789         arguments.
12790         * doc/invoke.texi (-fdiagnostics-color): Document.
12791         * pretty-print.h (pp_show_color): Define.
12792         (struct pretty_print_info): Add show_color field.
12793         * diagnostic.c: Include diagnostic-color.h.
12794         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12795         macros.  Colorize error:, warning: etc. strings and also the location
12796         string.
12797         (diagnostic_show_locus): Colorize the caret line.
12798         * pretty-print.c: Include diagnostic-color.h.
12799         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12800         inside of %< %> quotes or quoted through q format modifier.
12802 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12804         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12806 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12808         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12809         code in CC_NZ mode.
12810         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12811         pattern.
12813 2013-04-11  Marek Polacek  <polacek@redhat.com>
12815         PR tree-optimization/48184
12816         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12818 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12820         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12821         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12822         (skip_simple_arithmetic): Tidy up.
12823         * tree.h (skip_simple_constant_arithmetic): Declare.
12825 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12827         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12829 2013-04-11  Richard Biener  <rguenther@suse.de>
12831         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12832         generate vector constants.
12834 2013-04-11  Richard Biener  <rguenther@suse.de>
12836         PR tree-optimization/56878
12837         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12838         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12839         New function.
12840         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12841         Prefer to align the DR with the most invariant base address.
12843 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12845         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12846         comment.
12848 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12850         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12851         floating-point vector comparisons against 0.
12853 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12855         PR tree-optimization/56899
12856         * fold-const.c (extract_muldiv_1): Apply distributive law
12857         only if TYPE_OVERFLOW_WRAPS (ctype).
12859 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12861         PR target/56124
12862         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12863         loading has side effect.
12865 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12867         * config/sparc/sparc.c: Include tree-pass.h.
12868         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12869         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12870         head of file.  Change return type.  Split off gate function.
12871         (sparc_gate_work_around_errata): New function.
12872         (pass_work_around_errata): New pass definition.
12873         (insert_pass_work_around_errata) New pass insert definition to
12874         insert pass_work_around_errata just after delayed-branch scheduling.
12875         (sparc_option_override): Insert the pass.
12876         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12878 2013-04-10  David S. Miller  <davem@davemloft.net>
12880         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12881         or -mcpu=hypersparc.
12883         * target.def (cstore_mode): New hook.
12884         * target.h: Include insn-codes.h
12885         * targhooks.c: Likewise.
12886         (default_cstore_mode): New function.
12887         * targhooks.h: Declare it.
12888         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12889         * doc/tm.texi: Rebuild.
12890         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12891         target hook, rather than inspecting the insn_data.
12892         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12893         (TARGET_CSTORE_MODE): Redefine.
12894         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12895         result patterns.
12896         * config/sparc/predicates.md (cstore_result_operand): New special
12897         predicate.
12898         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12899         Use it for operand 0.
12900         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12901         (*snesi_special): Likewise.
12902         (*snesi_zero): Likewise.
12903         (*seqsi_zero): Likewise.
12904         (*sltu_insn): Likewise.
12905         (*sgeu_insn): Likewise.
12906         (*seqdi_special): Make operand 0 and comparison operation be of
12907         DImode.
12908         (*snedi_special): Likewise.
12909         (*snedi_special_vis3): Likewise.
12910         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12911         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12912         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12913         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12914         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12915         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12916         (*sltu_extend_sp64): Likewise.
12917         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12918         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12919         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12920         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12921         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12923 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12925         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12926         (aarch64_start_file): Use the new function.
12928 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12929             Jason Merrill  <jason@redhat.com>
12931         * common.opt: Add -gdwarf.
12932         * opts.c (common_handle_option): Handle it.
12933         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12935 2013-04-10  Richard Biener  <rguenther@suse.de>
12937         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12938         (execute_one_ipa_transform_pass): But unconditionally here.
12939         (execute_one_pass): And here.
12940         (init_optimization_passes): Remove reload pass.
12941         * tree-pass.h (TODO_ggc_collect): Remove.
12942         (pass_reload): Likewise.
12943         * ira.c (do_reload): Merge into ...
12944         (ira): ... this.
12945         (rest_of_handle_reload): Remove.
12946         (pass_reload): Likewise.
12947         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12948         of reload for vzeroupper pass placement.
12949         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12950         and todo_flags_finish of all passes.
12952 2013-04-10  Richard Biener  <rguenther@suse.de>
12954         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12955         first_const_oprnd field, rename first_def_type to first_op_type.
12956         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12957         (vect_get_and_check_slp_defs): Always use the type of the
12958         operand.  Allow mixed vect_external_def, vect_constant_def types.
12959         (vect_get_constant_vectors): Handle mixed vect_external_def,
12960         vect_constant_def types.
12962 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
12964         PR tree-optimization/55524
12965         * tree-ssa-math-opts.c
12966         (convert_mult_to_fma): Don't use an fms construct
12967         when we don't have an fms operation, but fnma, and it looks
12968         likely that we'll be able to use the latter.
12970 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
12972         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12973         function.
12974         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12975         inline fail caused by overwritable functions.
12977 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12979         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12980         unnecessary bits in the constant power of two case.
12982 2013-04-10  Richard Biener  <rguenther@suse.de>
12984         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12985         broken code swapping operands.
12986         (vect_build_slp_tree): Do not compute load permutations here.
12987         (vect_analyze_slp_instance): Compute load permutations here,
12988         after building the SLP tree.
12990 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12992         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12993         of next/prev_real_insn.
12995 2013-04-09  Jan Hubicka  <jh@suse.cz>
12997         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12998         Drop aliased parameter.
12999         (function_and_variable_visibility): Do not handle alias pairs.
13000         * cgraph.c (varpool_externally_visible_p): Update prototype.
13001         * varpool.c (varpool_add_new_variable): Update.
13003 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13005         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13007 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
13009         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13011         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13013 2013-04-09  Marek Polacek  <polacek@redhat.com>
13015         PR tree-optimization/48762
13016         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13018 2013-04-09  Richard Biener  <rguenther@suse.de>
13020         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13021         dealing with cost.
13022         (vect_build_slp_tree): Likewise.
13023         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13024         calculating the cost of a SLP instance.
13025         (vect_analyze_slp_instance): Use it from here, after building
13026         the SLP tree.
13028 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13030         PR middle-end/56883
13031         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13032         expand_omp_for_static_chunk): Use simple_p = true in
13033         force_gimple_operand_gsi calls when assigning to addressable decls.
13035 2013-04-09  Jeff Law  <law@redhat.com>
13037         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13038         when the boolean was created by converting a wider object which
13039         had a boolean range.
13041 2013-04-09  Richard Biener  <rguenther@suse.de>
13043         * tree-vectorizer.h (slp_void_p): Remove.
13044         (slp_tree): Typedef before _slp_tree declaration.
13045         (struct _slp_tree): Use a vector of slp_tree as children.
13046         (vect_get_place_in_interleaving_chain): Remove.
13047         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13048         Move ...
13049         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13050         and make static.
13051         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13052         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13053         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13054         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13055         Use slp_node instead of slp_void_p and adjust.
13057 2013-04-09  Richard Biener  <rguenther@suse.de>
13059         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13060         work that is not necessary.
13062 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13064         PR tree-optimization/56854
13065         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13066         forward into clobber stmts if it would change MEM_REF lhs into
13067         non-MEM_REF.
13069 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
13071         * tree.c (type_hash_lookup, type_hash_add): Make static.
13072         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13074 2013-04-09  Richard Biener  <rguenther@suse.de>
13076         * tree.h (unsave_expr_now): Remove.
13077         * tree-inline.c (mark_local_for_remap_r): Remove.
13078         (unsave_expr_1): Likewise.
13079         (unsave_r): Likewise.
13080         (unsave_expr_now): Likewise.
13081         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13082         (propagate_tree_value): Likewise.
13084 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
13086         * doc/rtl.texi (sequence): Rewrite documentation to match the
13087         current use of SEQUENCE rtl objects.
13088         * rtl.def (SEQUENCE): Likewise.
13090         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13091         Update documentation.
13092         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13093         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13095         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13097 2013-04-08  Teresa Johnson  <tejohnson@google.com>
13099         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13100         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13101         methods.
13102         (estimate_edge_size_and_time): Add comment to suggest using rounding
13103         methods.
13104         (estimate_node_size_and_time): Ditto.
13105         (remap_edge_change_prob): Use helper rounding divide methods.
13106         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13107         (gimple_mod_pow2_value_transform): Ditto.
13108         (gimple_mod_subtract_transform): Ditto.
13109         (gimple_ic_transform): Ditto.
13110         (gimple_stringops_transform): Ditto.
13111         * stmt.c (conditional_probability): Ditto.
13112         (emit_case_dispatch_table): Ditto.
13113         * lto-cgraph.c (merge_profile_summaries): Ditto.
13114         * tree-optimize.c (execute_fixup_cfg): Ditto.
13115         * cfgcleanup.c (try_forward_edges): Ditto.
13116         * cfgloopmanip.c (scale_loop_profile): Ditto.
13117         (loopify): Ditto.
13118         (duplicate_loop_to_header_edge): Ditto.
13119         (lv_adjust_loop_entry_edge): Ditto.
13120         * tree-vect-loop.c (vect_transform_loop): Ditto.
13121         * profile.c (compute_branch_probabilities): Ditto.
13122         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13123         * lto-streamer-in.c (input_cfg): Ditto.
13124         * gimple-streamer-in.c (input_bb): Ditto.
13125         * ipa-cp.c (update_profiling_info): Ditto.
13126         (update_specialized_profile): Ditto.
13127         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13128         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13129         rounding methods.
13130         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13131         (compute_trg_info): Ditto.
13132         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13133         (purge_dead_edges): Ditto.
13134         * loop-unswitch.c (unswitch_loop): Ditto.
13135         * cgraphclones.c (cgraph_clone_edge): Ditto.
13136         (cgraph_clone_node): Ditto.
13137         * tree-inline.c (copy_bb): Ditto.
13138         (copy_edges_for_bb): Ditto.
13139         (initialize_cfun): Ditto.
13140         (copy_cfg_body): Ditto.
13141         (expand_call_inline): Ditto.
13143 2013-04-08  Kai Tietz  <ktietz@redhat.com>
13145         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13146         TARGET_CYGWIN64 by TARGET_64BIT.
13148 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13150         * config/epiphany/epiphany.md (GPR_1): New constant.
13151         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13152         * config/epiphany/epiphany.c (gen_compare_reg):
13153         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13154         is already in place.
13155         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13156         Don't require being called during rtl expansion; If y operlaps r0,
13157         return 0.
13158         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13159         (epiphany_expand_epilogue): Likewise.
13161         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13162         Don't use CC_FPmode for ORDERED / UNORDERED.
13163         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13165         * config/epiphany/constraints.md (CnL): New constraint.
13166         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13167         * config/epiphany/predicates.md (add_operand): Allow 1024.
13169         * config/epiphany/epiphany.md (logical_op): New code iterator.
13170         (op_mnc): New code attribute.
13171         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13172         (mov_f+1, mov_f+2): New peephole2 patterns.
13174         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13175         (cstoresi4): Also allow re-use of zero result when doing a NE
13176         comparison to a non-zero operand.
13177         Use (clobber (scratch)) for first insn if the gpr output is not needed.
13179         * config/epiphany/epiphany.md (<insn_opname>v2si3):
13180         Use gen_addsi3_i / gen_subsi3_i.
13182 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13184         PR c++/34949
13185         PR c++/50243
13186         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13187         contain anything but clobbers, at most one __builtin_stack_restore,
13188         optionally debug stmts and final resx, and if it has at least one
13189         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
13190         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13191         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13192         which isn't defaut definition, remove them.
13193         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13194         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13195         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13196         with MEM_REF LHS with SSA_NAME address.
13198 2013-04-08  Jeff Law  <law@redhat.com>
13200         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13202 2013-04-08  Richard Biener  <rguenther@suse.de>
13204         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13205         extra newline.
13206         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13207         determined vector type.
13208         (vect_analyze_data_refs): Likewise.
13209         (vect_get_new_vect_var): Adjust.
13210         (vect_create_destination_var): Preserve SSA name versions.
13211         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13212         not dump anything here.
13214 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13216         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13217         Add member lr_slot_known.
13218         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13219         if necessary.
13220         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13221         Remove code that sets lr_slot_offset according to what a previous
13222         version of epiphany_emit_save_restore used to do.
13223         (epiphany_emit_save_restore): When doing an lr save or restore,
13224         set/verify lr_slot_known and lr_slot_offset.
13226 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
13228         PR target/54338
13229         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13230         in ALL_REGS.
13232 2013-04-08  Richard Biener  <rguenther@suse.de>
13234         * alias.c (find_base_term): Fix thinko in previous change.
13236 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13238         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13239         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13240         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
13241         if possible to compute val.
13242         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13243         For QImode integers don't require anything about precision.  Use
13244         const_with_all_bytes_same to find out if the constant doesn't have
13245         repeated bytes in it.
13247 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13249         * config/s390/s390.c (s390_expand_insv): Only accept insertions
13250         within mode size.
13252 2013-04-08  Marek Polacek  <polacek@redhat.com>
13254         PR rtl-optimization/48182
13255         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13256         value to 1.
13258 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13260         PR target/55487
13261         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13262         nuses, make sure we have a label.
13264 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13266         PR target/56843
13267         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13268         (rs6000_emit_swdiv_low_precision): Remove.
13269         (rs6000_emit_swdiv): Rewrite to handle between one and four
13270         iterations of Newton-Raphson generally; modify required number of
13271         iterations for some cases.
13272         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13274 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
13276         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13277         set-but-unused variable.
13279         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13280         basic blocks of released function bodies garbage-collectable.
13282         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13283         (struct rtl_opt_pass): Add TODO_df_finish.
13285         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13287 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13289         * config/arm/constraints.md (q): New constraint.
13290         * config/arm/ldrdstrd.md: New file.
13291         * config/arm/arm.md (ldrdstrd.md) New include.
13292         (arm_movdi): Use "q" instead of "r" constraint
13293         for double-word memory access.
13294         (movdf_soft_insn): Likewise.
13295         * config/arm/vfp.md (movdi_vfp): Likewise.
13296         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13297         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13298         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13299         (mem_ok_for_ldrd_strd): Likewise.
13300         (output_move_double): Update assertion.
13302 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13304         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13306 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13308         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13309         define_insn_and_split.
13310         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13312 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13314         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13315         define_insn_and_split.
13316         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13317         (shiftsi3_compare): New pattern.
13318         (rrx): New pattern.
13319         * config/arm/unspecs.md (UNSPEC_RRX): New.
13321 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13323         * config/arm/arm.md (negdi_extendsidi): New pattern.
13324         (negdi_zero_extendsidi): Likewise.
13326 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13328         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13329         define_insn_and_split.
13330         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13331         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13333 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13335         * config/arm/arm.md (arm_subdi3): Convert define_insn into
13336         define_insn_and_split.
13337         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13338         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13340 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13342         * config/arm/arm.md (subsi3_carryin): New pattern.
13343         (subsi3_carryin_const): Likewise.
13344         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13345         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13347 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13349         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13351 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13353         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13354         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13356 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13358         * config/arm/arm.c (arm_expand_builtin): Change fcode
13359         type to unsigned int.
13361 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13363         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13365 2013-04-04  Ian Lance Taylor  <iant@google.com>
13367         * doc/standards.texi (Standards): The Go frontend supports the Go 1
13368         language standard.
13370 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
13372         PR middle-end/56729
13373         * df-scan.c (df_insn_delete): Disable failing assert.
13375 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13377         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13378         New function prototype.
13379         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13380         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13381         (arm_builtin_vectorized_function): New function.
13383 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13385         * config/arm/arm_neon_builtins.def: New file.
13386         * config/arm/arm.c (neon_builtin_data): Move contents to
13387         arm_neon_builtins.def.
13388         (enum arm_builtins): Include neon builtin definitions.
13389         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13390         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13392 2013-04-04  Marek Polacek  <polacek@redhat.com>
13394         PR tree-optimization/48186
13395         * predict.c (maybe_hot_frequency_p): Return false if
13396         HOT_BB_FREQUENCY_FRACTION is 0.
13397         (cgraph_maybe_hot_edge_p): Likewise.
13399 2013-04-04  Richard Biener  <rguenther@suse.de>
13401         PR tree-optimization/56826
13402         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13403         more accurately.
13405 2013-04-04  Richard Biener  <rguenther@suse.de>
13407         PR tree-optimization/56213
13408         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13409         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13411 2013-04-04  Richard Biener  <rguenther@suse.de>
13413         PR tree-optimization/56837
13414         * tree-loop-distribution.c (classify_partition): For non-zero
13415         values require that the value has the same precision as its
13416         mode to be useful as memset value.
13418 2013-04-03  Nick Clifton  <nickc@redhat.com>
13420         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13421         (fmssf4): Use fmsf.s on E3V5 architectures.
13422         (fnmasf4): Use fnmaf.s on E3V5 architectures.
13423         (fnmssf4): Use fnmsf.s on E3V5 architectures.
13425 2013-04-03  Jeff Law  <law@redhat.com>
13427         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13428         (lra-eliminations.o): Likewise.
13430 2013-04-03  Teresa Johnson  <tejohnson@google.com>
13432         * gcov-io.c (compute_working_sets): Moved most of body of old
13433         compute_working_sets here from profile.c.
13434         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13435         (gcov_working_set_t): Moved typedef here from basic-block.h
13436         (compute_working_set): Declare.
13437         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13438         (get_working_sets): Renamed from compute_working_set,
13439         replace most of body with call to new compute_working_sets.
13440         (get_exec_counts): Replace call to compute_working_sets
13441         to get_working_sets.
13442         * profile.h (get_working_sets): Renamed from compute_working_set.
13443         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13444         to get_working_sets.
13445         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13446         * gcov-dump.c (dump_working_sets): New function.
13448 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
13450         * hwint.c (sext_hwi, zext_hwi): New functions.
13451         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13452         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13453         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13454         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13455         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13456         (sext_hwi, zext_hwi): New functions.
13458 2013-04-03  Jeff Law  <law@redhat.com>
13460         PR tree-optimization/56799
13461         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13462         back test for widening conversion erroneously dropped in prior change.
13464 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13466         PR target/56809
13467         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13468         instead of next_real_insn.
13470 2013-04-03  Marek Polacek  <polacek@redhat.com>
13472         PR sanitizer/55702
13473         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13475 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13477         PR target/56809
13478         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13479         next_real_insn.
13480         (thumb1_output_casesi): Likewise.
13481         (thumb2_output_casesi): Likewise.
13483 2013-04-03  Richard Biener  <rguenther@suse.de>
13485         PR tree-optimization/56817
13486         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13487         Split out ...
13488         (tree_unroll_loops_completely_1): ... new function to manually
13489         walk the loop tree, properly defering outer loops of unrolled
13490         loops to later iterations.
13492 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13494         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13495         (vectorizable_load): Likewise.
13496         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13497         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13499 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13501         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13502         BIT_FIELD_REF.
13504 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13506         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13508 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
13510         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13512 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13514         PR tree-optimization/56790
13515         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13516         folding.
13518 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13520         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13521         Handle VEC_MERGE.
13522         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13523         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13524         equal arguments.
13526 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
13528         PR c/19449
13529         * tree.h (force_folding_builtin_constant_p): New decl.
13530         * builtins.c (force_folding_builtin_constant_p): New variable.
13531         (fold_builtin_constant_p): Fold immediately also if
13532         force_folding_builtin_constant_p.
13534 2013-04-03  Richard Biener  <rguenther@suse.de>
13536         PR tree-optimization/56812
13537         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13538         DRs of the same interleaving chain are independent.
13540 2013-04-02  Jason Merrill  <jason@redhat.com>
13542         * gdbinit.in (pbb): Use debug fn.
13544 2013-04-02  Lawrence Crowl  <crowl@google.com>
13546         * sese.h (struct ivtype_map_elt_s): Remove unused.
13547         (extern debug_ivtype_map): Remove unused.
13548         (extern eq_ivtype_map_elts): Remove unused.
13549         * sese.c (debug_ivtype_map): Removed unused.
13550         (debug_ivtype_map_1): Removed unused.
13551         (debug_ivtype_elt): Remove unused.
13552         (eq_ivtype_map_elts): Remove unused.
13555 2013-04-02  Kai Tietz  <ktietz@redhat.com>
13557         PR target/52790
13558         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13559         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13560         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13561         function.
13562         (legitimize_pe_coff_symbol): Likewise.
13563         (is_imported_p): New helper-function.
13564         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13565         for Windows x64 targets.
13566         (ix86_expand_prologue): Optimize for pe-coff targets.
13567         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13568         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13569         medium/large code-model.
13570         (legitimize_pic_address): Likewise.
13571         (legitimize_tls_address): Likewise.
13572         (ix86_expand_call): Likewise.
13573         (x86_output_mi_thunk): Likewise.
13574         (get_dllimport_decl): Add new beimport argument.
13575         (construct_plt_address): Don't assert for x64 pe-coff targets.
13576         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13577         targets.
13578         (SYMBOL_FLAG_STUBVAR): New macro.
13579         (SYMBOL_REF_STUBVAR_P): Likewise.
13580         * config/i386/winnt.c (stub_list): New structure.
13581         (stub_head): New local variable.
13582         (i386_pe_record_stub): New function.
13583         (i386_pe_file_end): Emit refptr-stubs.
13585 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
13587         PR rtl-optimization/56745
13588         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13589         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13591         PR c++/34949
13592         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13593         and both of them are MEM_REFs, just compare first argument for
13594         equality and attempt to deal even with differing offsets.
13596         PR c++/34949
13597         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13598         of gimple_clobber_p to be MEM_REF.
13599         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13600         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
13601         after gimplification.
13602         * asan.c (get_mem_ref_of_assignment): Don't instrument
13603         gimple_clobber_p stmts.
13604         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13605         gimple_clobber_p stmt if they have MEM_REF lhs and
13606         are dead because of another gimple_clobber_p stmt.
13607         * tree-ssa-live.c (clear_unused_block_pointer): Treat
13608         gimple_clobber_p stmts like debug stmts.
13609         (remove_unused_locals): Remove clobbers with MEM_REF lhs
13610         that refer to unused VAR_DECLs or uninitialized values.
13611         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13612         gimple_clobber_p stmts if they refer to removed parameters.
13613         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13614         formatting.
13616 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
13618         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13619         using SWI48 mode attribute.
13621 2013-04-02  Wei Mi  <wmi@google.com>
13623         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13624         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13625         *<rotate_insn><mode>3_mask in i386.md.
13627 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13629         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13631 2013-04-02  Richard Biener  <rguenther@suse.de>
13633         PR tree-optimization/56778
13634         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13635         Runtime alias tests are not supported for gather loads.
13636         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13637         stmts referenced from SSA operands before updating SSA form.
13639 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
13640             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13642         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13643         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13644         * config/arm/cortex-a53.md: New file.
13645         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13646         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13647         * config/arm/arm.c (arm_issue_rate): Likewise.
13648         * config/arm/arm-tune.md: Regenerate
13649         * config/arm/arm-tables.opt: Regenerate.
13650         * config/arm/arm-cores.def: Add cortex-a53.
13652 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
13654         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13655         non-static link.
13657 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13659         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13660         scalar load/store operations using B/H registers.
13661         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13663 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13665         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13666         scalar move.
13667         * config/aarch64/aarch64.c
13668         (aarch64_simd_scalar_immediate_valid_for_move): New.
13669         * config/aarch64/aarch64-protos.h
13670         (aarch64_simd_scalar_immediate_valid_for_move): New.
13671         * config/aarch64/constraints.md (Dh, Dq): New.
13672         * config/aarch64/iterators.md (hq): New.
13674 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
13676         * reorg.c (get_branch_condition): Deal with conditional returns.
13677         (fill_simple_delay_slots): Remove dead code dealing with jumps.
13679 2013-04-01  Wei Mi  <wmi@google.com>
13681         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13682         Truncate operand 2 using %b asm operand modifier.
13683         (*<shift_insn><mode>3_mask): Ditto.
13684         (*<rotate_insn><mode>3_mask): Ditto.
13686 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
13688         PR middle-end/56798
13689         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13691 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
13693         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13694         of next_real_insn.
13695         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13697 2013-03-30  Lawrence Crowl  <crowl@google.com>
13699         * dse.c (clear_alias_sets): Remove never set.
13700         (disqualified_clear_alias_sets): Remove never set.
13701         (clear_alias_mode_pool): Remove never set.
13702         (dse_step0): Remove condition that is never true.
13703         (canon_address): Remove condition that is never true.
13704         (dse_step7): Remove condition that is never true.
13705         (rest_of_handle_dse): Remove condition that is never true.
13706         (rest_of_handle_dse::did_global): Remove never read from above.
13707         (dse_step2_spill): Remove never called from above.
13708         (dse_step5_spill): Remove never called from above.
13710 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
13712         * doc/md.texi (Standard Names) <casesi>: Update documentation for
13713         JUMP_TABLE_DATA changes.
13714         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13715         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13716         (Insns) <jump_table_data>: New entry.
13717         * doc/tm.texi: Regenerate.
13719         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13721         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13722         for table jump at the end of a basic block using tablejump_p.
13723         * targhooks.c (default_invalid_within_doloop): Likewise.
13724         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13725         target hook implementation that is identical to the default hook.
13726         (rs6000_invalid_within_doloop): Remove.
13728         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13729         unused variable from tablejump_p call.
13731         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13732         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13733         (INSN_DELETED_P): Likewise.
13734         (emit_jump_table_data): New prototype.
13735         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13736         after 4th as unused.
13737         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13738         * sched-vis.c (print_insn): Likewise.
13739         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13740         insn for compatibility with back ends that use next_active_insn to
13741         identify jump table data.
13742         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13743         (remove_insn): Likewise.
13744         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13745         to be emitted.
13746         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13747         (emit_jump_table_data): New function.
13749         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13750         basic block, a JUMP_TABLE_DATA never is.
13751         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13752         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13753         off from code handling real insns.
13754         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13755         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13756         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13757         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13758         is not a NONDEBUG_INSN_P.
13759         * ira-costs.c (scan_one_insn): Likewise.
13760         * jump.c (mark_all_labels): Likewise.
13761         (mark_jump_label_1): Likewise.
13762         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13763         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13764         a basic block.
13765         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13766         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13767         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13768         JUMP_TABLE_DATA_P insns.
13769         (calculate_elim_costs_all_insns): Likewise.
13770         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13771         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13772         (delete_output_reload): Code style fixups.
13773         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13774         insn flags on this non-insn.
13775         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13776         as scheduling barriers, for pre-change compatibility.
13777         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13778         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13780         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13781         redundant JUMP_TABLE_DATA_P test.
13782         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13783         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13784         (frv_for_each_packet): Likewise.
13785         * config/i386/i386.c (min_insn_size): Likewise.
13786         (ix86_avoid_jump_mispredicts): Likewise.
13787         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13788         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13789         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13790         (mips16_insn_length): Robustify.
13791         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13792         (mips16_split_long_branches): Likewise.
13793         * config/pa/pa.c (pa_combine_instructions): Likewise.
13794         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13795         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13796         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13797         as contributing to pool range lengths.
13798         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13799         Remove redundant JUMP_TABLE_DATA_P test.
13800         (sh_loop_align): Likewise.
13801         (split_branches): Likewise.
13802         (sh_insn_length_adjustment): Likewise.
13803         * config/spu/spu.c (get_branch_target): Likewise.
13805 2013-03-29  Jan Hubicka  <jh@suse.cz>
13807         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13808         gcov streaming; stream hot bb threshold to ltrans.
13809         * predict.c (get_hot_bb_threshold): Break out from ....
13810         (maybe_hot_count_p): ... here.
13811         (set_hot_bb_threshold): New function.
13812         * lto-section-in.c (lto_section_name): Add profile.
13813         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13814         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13815         and data-streamer.h
13816         (histogram_entry): New structure.
13817         (histogram, histogram_pool): New global vars.
13818         (histogram_hash): New structure.
13819         (histogram_hash::hash): New method.
13820         (histogram_hash::equal): Likewise.
13821         (account_time_size): New function.
13822         (cmp_counts): New function.
13823         (dump_histogram): New function.
13824         (ipa_profile_generate_summary): New function.
13825         (ipa_profile_write_summary): New function.
13826         (ipa_profile_read_summary): New function.
13827         (ipa_profile): Decide on threshold.
13828         (pass_ipa_profile): Add ipa_profile_write_summary and
13829         ipa_profile_read_summary.
13830         * Makefile.in (ipa.o): Update dependencies.
13831         * lto-streamer.h (LTO_section_ipa_profile): New section.
13833 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13835         * tree.h (VAR_P): New.
13837 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13839         PR lto/56777
13840         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13842 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13844         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13845         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13846         (control_flow_insn_p): Likewise.
13847         * cfgrtl.c (duplicate_insn_chain): Likewise.
13848         * final.c (get_attr_length_1): Likewise.
13849         (shorten_branches): Likewise.
13850         (final_scan_insn): Likewise.
13851         * function.c (instantiate_virtual_regs): Likewise.
13852         * gcse.c (insert_insn_end_basic_block): Likewise.
13853         * ira-costs.c (scan_one_insn): Likewise.
13854         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13855         * lra.c (check_rtl): Likewise.
13856         * reload1.c (elimination_costs_in_insn): Likewise.
13857         * reorg.c (follow_jumps): Likewise.
13859         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13860         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13861         (thumb_far_jump_used_p): Likewise.
13862         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13863         (workaround_speculation): Likewise.
13864         (add_sched_insns_for_speculation): Likewise.
13865         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13866         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13867         (frv_for_each_packet): Likewise.
13868         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13869         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13870         (final_emit_insn_group_barriers): Likewise.
13871         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13872         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13873         (mips16_insn_length): Likewise.
13874         * config/pa/pa.c (pa_reorg): Likewise.
13875         (pa_combine_instructions): Likewise.
13876         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13877         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13878         (sh_reorg): Likewise.
13879         (split_branches): Likewise.
13880         * config/spu/spu.c (get_branch_target): Likewise.
13882         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13883         JUMP_TABLE_DATA_P.
13885 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13887         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13888         Fix declaration name.
13890 2013-03-28  Lawrence Crowl  <crowl@google.com>
13892         * graphds.h (struct graph.indicies): Remove unused.
13893         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13894         (SCOP_ORIGINAL_PDDRS): Remove unused.
13895         * sese.h (extern insert_loop_close_phis): Removed unused.
13896         (extern insert_guard_phis): Removed unused.
13897         (extern ivtype_map_elt_info): Removed unused.
13898         (new_ivtype_map_elt): Removed unused.
13899         * sese.c (ivtype_map_elt_info): Removed unused.
13901 2013-03-28  Lawrence Crowl  <crowl@google.com>
13903         * Makefile.in: Add several missing include dependences.
13904         (DUMPFILE_H): New.
13905         (test-dump.o): New.  This object is not added to any executable,
13906         but is present for ad-hoc testing.
13907         * bitmap.c
13908         (debug (const bitmap_head_def &)): New.
13909         (debug (const bitmap_head_def *)): New.
13910         * bitmap.h
13911         (extern debug (const bitmap_head_def &)): New.
13912         (extern debug (const bitmap_head_def *)): New.
13913         * cfg.c
13914         (debug (edge_def &)): New.
13915         (debug (edge_def *)): New.
13916         * cfghooks.c
13917         (debug (basic_block_def &)): New.
13918         (debug (basic_block_def *)): New.
13919         * dumpfile.h
13920         (dump_node (const_tree, int, FILE *)): Correct source file.
13921         * dwarf2out.c
13922         (debug (die_struct &)): New.
13923         (debug (die_struct *)): New.
13924         * dwarf2out.h
13925         (extern debug (die_struct &)): New.
13926         (extern debug (die_struct *)): New.
13927         * gimple-pretty-print.c
13928         (debug (gimple_statement_d &)): New.
13929         (debug (gimple_statement_d *)): New.
13930         * gimple-pretty-print.h
13931         (extern debug (gimple_statement_d &)): New.
13932         (extern debug (gimple_statement_d *)): New.
13933         * ira-build.c
13934         (debug (ira_allocno_copy &)): New.
13935         (debug (ira_allocno_copy *)): New.
13936         (debug (ira_allocno &)): New.
13937         (debug (ira_allocno *)): New.
13938         * ira-int.h
13939         (extern debug (ira_allocno_copy &)): New.
13940         (extern debug (ira_allocno_copy *)): New.
13941         (extern debug (ira_allocno &)): New.
13942         (extern debug (ira_allocno *)): New.
13943         * ira-lives.c
13944         (debug (live_range &)): New.
13945         (debug (live_range *)): New.
13946         * lra-int.h
13947         (debug (lra_live_range &)): New.
13948         (debug (lra_live_range *)): New.
13949         * lra-lives.c
13950         (debug (lra_live_range &)): New.
13951         (debug (lra_live_range *)): New.
13952         * omega.c
13953         (debug (omega_pb_d &)): New.
13954         (debug (omega_pb_d *)): New.
13955         * omega.h
13956         (extern debug (omega_pb_d &)): New.
13957         (extern debug (omega_pb_d *)): New.
13958         * print-rtl.c
13959         (debug (const rtx_def &)): New.
13960         (debug (const rtx_def *)): New.
13961         * print-tree.c
13962         (debug_tree (tree): Move within file.
13963         (debug_raw (const tree_node &)): New.
13964         (debug_raw (const tree_node *)): New.
13965         (dump_tree_via_hooks (const tree_node *, int)): New.
13966         (debug (const tree_node &)): New.
13967         (debug (const tree_node *)): New.
13968         (debug_verbose (const tree_node &)): New.
13969         (debug_verbose (const tree_node *)): New.
13970         (debug_head (const tree_node &)): New.
13971         (debug_head (const tree_node *)): New.
13972         (debug_body (const tree_node &)): New.
13973         (debug_body (const tree_node *)): New.
13974         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13975         (debug (vec<tree, va_gc> &)): New.
13976         (debug (vec<tree, va_gc> *)): New.
13977         * rtl.h
13978         (extern debug (const rtx_def &)): New.
13979         (extern debug (const rtx_def *)): New.
13980         * sbitmap.c
13981         (debug_raw (simple_bitmap_def &)): New.
13982         (debug_raw (simple_bitmap_def *)): New.
13983         (debug (simple_bitmap_def &)): New.
13984         (debug (simple_bitmap_def *)): New.
13985         * sbitmap.h
13986         (extern debug (simple_bitmap_def &)): New.
13987         (extern debug (simple_bitmap_def *)): New.
13988         (extern debug_raw (simple_bitmap_def &)): New.
13989         (extern debug_raw (simple_bitmap_def *)): New.
13990         * sel-sched-dump.c
13991         (debug (vinsn_def &)): New.
13992         (debug (vinsn_def *)): New.
13993         (debug_verbose (vinsn_def &)): New.
13994         (debug_verbose (vinsn_def *)): New.
13995         (debug (expr_def &)): New.
13996         (debug (expr_def *)): New.
13997         (debug_verbose (expr_def &)): New.
13998         (debug_verbose (expr_def *)): New.
13999         (debug (vec<rtx> &)): New.
14000         (debug (vec<rtx> *)): New.
14001         * sel-sched-dump.h
14002         (extern debug (vinsn_def &)): New.
14003         (extern debug (vinsn_def *)): New.
14004         (extern debug_verbose (vinsn_def &)): New.
14005         (extern debug_verbose (vinsn_def *)): New.
14006         (extern debug (expr_def &)): New.
14007         (extern debug (expr_def *)): New.
14008         (extern debug_verbose (expr_def &)): New.
14009         (extern debug_verbose (expr_def *)): New.
14010         (extern debug (vec<rtx> &)): New.
14011         (extern debug (vec<rtx> *)): New.
14012         * sel-sched-ir.h
14013         (_list_iter_cond_expr): Make inline instead of static.
14014         * sreal.c
14015         (debug (sreal &)): New.
14016         (debug (sreal *)): New.
14017         * sreal.h
14018         (extern debug (sreal &)): New.
14019         (extern debug (sreal *)): New.
14020         * tree.h
14021         (extern debug_raw (const tree_node &)): New.
14022         (extern debug_raw (const tree_node *)): New.
14023         (extern debug (const tree_node &)): New.
14024         (extern debug (const tree_node *)): New.
14025         (extern debug_verbose (const tree_node &)): New.
14026         (extern debug_verbose (const tree_node *)): New.
14027         (extern debug_head (const tree_node &)): New.
14028         (extern debug_head (const tree_node *)): New.
14029         (extern debug_body (const tree_node &)): New.
14030         (extern debug_body (const tree_node *)): New.
14031         (extern debug (vec<tree, va_gc> &)): New.
14032         (extern debug (vec<tree, va_gc> *)): New.
14033         * tree-cfg.c
14034         (debug (struct loop &)): New.
14035         (debug (struct loop *)): New.
14036         (debug_verbose (struct loop &)): New.
14037         (debug_verbose (struct loop *)): New.
14038         * tree-dump.c: Add header dependence.
14039         * tree-flow.h
14040         (extern debug (struct loop &)): New.
14041         (extern debug (struct loop *)): New.
14042         (extern debug_verbose (struct loop &)): New.
14043         (extern debug_verbose (struct loop *)): New.
14044         * tree-data-ref.c
14045         (debug (data_reference &)): New.
14046         (debug (data_reference *)): New.
14047         (debug (vec<data_reference_p> &)): New.
14048         (debug (vec<data_reference_p> *)): New.
14049         (debug (vec<ddr_p> &)): New.
14050         (debug (vec<ddr_p> *)): New.
14051         * tree-data-ref.h
14052         (extern debug (data_reference &)): New.
14053         (extern debug (data_reference *)): New.
14054         (extern debug (vec<data_reference_p> &)): New.
14055         (extern debug (vec<data_reference_p> *)): New.
14056         (extern debug (vec<ddr_p> &)): New.
14057         (extern debug (vec<ddr_p> *)): New.
14058         * tree-ssa-alias.c
14059         (debug (pt_solution &)): New.
14060         (debug (pt_solution *)): New.
14061         * tree-ssa-alias.h
14062         (extern debug (pt_solution &)): New.
14063         (extern debug (pt_solution *)): New.
14064         * tree-ssa-alias.c
14065         (debug (_var_map &)): New.
14066         (debug (_var_map *)): New.
14067         (debug (tree_live_info_d &)): New.
14068         (debug (tree_live_info_d *)): New.
14069         * tree-ssa-alias.h
14070         (extern debug (_var_map &)): New.
14071         (extern debug (_var_map *)): New.
14072         (extern debug (tree_live_info_d &)): New.
14073         (extern debug (tree_live_info_d *)): New.
14075 2013-03-28  Jan Hubicka  <jh@suse.cz>
14077         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14079 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
14081         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14082         record only when desired or required.
14084 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
14086         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14087         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
14089 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14091         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14092         (*andsi3_compare0_uxtw): New pattern.
14093         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14094         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14096 2013-03-28  Jan Hubicka  <jh@suse.cz>
14098         * data-streamer-in.c (streamer_read_gcov_count): New function.
14099         * gimple-streamer-out.c: Include value-prof.h.
14100         (output_gimple_stmt): Output histogram.
14101         (output_bb): Use streamer_write_gcov_count.
14102         * value-prof.c: Include data-streamer.h
14103         (dump_histogram_value): Add HIST_TYPE_MAX.
14104         (stream_out_histogram_value): New function.
14105         (stream_in_histogram_value): New function.
14106         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14107         (stream_out_histogram_value, stream_in_histogram_value): Declare.
14108         * data-streamer-out.c (streamer_write_gcov_count): New function.
14109         (streamer_write_gcov_count_stream): New function.
14110         * lto-cgraph.c (lto_output_edge): Update counter streaming.
14111         (lto_output_node): Likewise.
14112         (input_node, input_edge): Likewise.
14113         * lto-streamer-out.c (output_cfg): Update streaming.
14114         * lto-streamer-in.c (input_cfg): Likewise.
14115         * data-streamer.h (streamer_write_gcov_count,
14116         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14117         * gimple-streamer-in.c: Include value-prof.h
14118         (input_gimple_stmt): Input histograms.
14119         (input_bb): Update profile streaming.
14121 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14123         * genmodes.c (emit_max_int): New function.
14124         (emit_insn_modes_h): Added call to emit_max_function.
14125         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14126         Added doc.
14127         * machmode.def: Fixed comment.
14129 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14131         * combine.c (try_combine): Removed useless assert.
14132         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14134 2013-03-28  Marek Polacek  <polacek@redhat.com>
14135             Richard Biener  <rguenther@suse.de>
14137         PR tree-optimization/56695
14138         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14139         build signed result of a vector comparison.
14140         * tree-cfg.c (verify_gimple_comparison): Check that a result
14141         of a vector comparison has signed type.
14143 2013-03-28  Richard Biener  <rguenther@suse.de>
14145         PR tree-optimization/37021
14146         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14147         do not restrict gaps between groups.
14148         * tree-vect-stmts.c (vectorizable_load): Properly account for
14149         a gap between groups.
14151 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
14153         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14154         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14155         is not enabled.
14157 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
14159         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14160         * doc/extend.texi (Named Address Spaces): Ditto.
14161         (Variable Attributes): Ditto.
14163 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14165         * config.build: Add support for cygwin x64 target.
14166         * config.gcc: Likewise.
14167         * config.host: Likewise.
14168         * configure.ac: Likewise
14169         * configure: Regenerated.
14171 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14173         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14174         * config/i386/t-cygwin-w64: New file.
14175         * config/i386/cygwin-w64.h: New file.
14176         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14177         and add support for x64-cygwin target.
14178         (CPP_SPEC): Likewise.
14179         (CXX_WRAP_SPEC_LIST): Undefine before define.
14180         (LIBGCJ_SONAME): Use 15 as version.
14182 2013-03-27  Richard Biener  <rguenther@suse.de>
14184         PR tree-optimization/56716
14185         * tree-ssa-structalias.c (perform_var_substitution): Adjust
14186         dumping for ref nodes.
14188 2013-03-27  Martin Jambor  <mjambor@suse.cz>
14190         PR tree-optimization/55334
14191         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14192         restricted pointers to arrays.
14194 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14196         * Makefile.in (.SUFFIXES): Add .cc.
14197         (.c.o): Apply same recipe for implicit rule .cc.o.
14199 2013-03-27  Richard Biener  <rguenther@suse.de>
14201         PR tree-optimization/37021
14202         * tree-vect-data-refs.c (vect_check_strided_load): Allow
14203         REALPART/IMAGPART_EXPRs around the supported refs.
14204         * tree-ssa-structalias.c (find_func_aliases): Assume that
14205         floating-point values are not used to transfer pointers.
14207 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
14209         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14210         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14211         * gcc/doc/tm.texi: Regenerate.
14212         * gcc/targhooks.h (default_has_ifunc_p): New.
14213         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14214         * gcc/config/linux-protos.h: New file.
14215         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14216         this hook for linux which disables support of indirect functions in
14217         android.
14218         * gcc/config/linux-android.c: New file.
14219         * gcc/config/t-linux-android.c: Ditto.
14220         * gcc/config.gcc: Added new object file linux-android.o.
14221         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14222         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14223         * gcc/varasm.c (do_assemble_alias): Likewise.
14224         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14225         doesn't support indirect functions.
14226         * configure: Regenerate.
14228 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
14230         PR target/56102
14231         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14232         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14233         mult-word mode.
14235 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14237         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14239 2013-03-27  Terry Guo  <terry.guo@arm.com>
14241         * config/arm/arm-cores.def: Added core cortex-r7.
14242         * config/arm/arm-tune.md: Regenerated.
14243         * config/arm/arm-tables.opt: Regenerated.
14244         * doc/invoke.texi: Added entry for core cortex-r7.
14246 2013-03-27  Walter Lee  <walt@tilera.com>
14248         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14249         double-decrement of next_scratch_regno.
14251 2013-03-27  Walter Lee  <walt@tilera.com>
14253         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14254         input operands.
14255         (insn_v1mulus): Ditto.
14256         (insn_v2muls): Ditto.
14258 2013-03-27  Walter Lee  <walt@tilera.com>
14260         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14261         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14263 2013-03-27  Walter Lee  <walt@tilera.com>
14265         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14266         (*sibcall_value): Ditto.
14268 2013-03-27  Walter Lee  <walt@tilera.com>
14270         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14271         (insn_mnz_v8qi): ... this ...
14272         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
14273         vector equivalent.
14274         (insn_v<n>mnz): Replaced by ...
14275         (insn_v1mnz): ... this ...
14276         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
14277         equivalent.
14278         (insn_mz_<mode>): Replaced by ...
14279         (insn_mz_v8qi): ... this ...
14280         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
14281         vector equivalent.
14282         (insn_v<n>mz): Replaced by ...
14283         (insn_v1mz): ... this ...
14284         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
14285         equivalent.
14287 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
14289         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14291 2013-03-26  Roland McGrath  <mcgrathr@google.com>
14293         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14294         than fprintf with a non-constant, non-format string.
14296 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14298         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14299         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
14300         operand 0 predicate.
14301         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14302         attribute.  Use general_x64nomem_operand as operand 1 predicate.
14303         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14304         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14305         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14306         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14307         (mov<mode>_insv_1): Remove expander.  Merge insn with
14308         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14309         Use general_x64nomem_operand as operand 1 predicate.
14310         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14311         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14312         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14313         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14314         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
14315         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14316         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14317         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14318         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
14319         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14320         (general_x64nomem_operand): Ditto.
14322 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14324         * config/rtems.opt: Add -pthread option.
14326 2013-03-26  Richard Biener  <rguenther@suse.de>
14328         * alias.c (find_base_term): Avoid redundant and not used recursion.
14329         (base_alias_check): Get the initial base term from the caller.
14330         (true_dependence_1): Compute and pass base terms to base_alias_check.
14331         (write_dependence_p): Likewise.
14332         (may_alias_p): Likewise.
14334 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
14336         * config/aarch64/aarch64.c (aarch64_classify_address): Support
14337         PC-relative load in SI modes and above only.
14339 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
14341         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14342         * config/arm/iwmmxt.md (WCGR0): Update.
14343         (WCGR1, WCGR2, WCGR3): Likewise.
14345 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14347         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14348         Use x64 and nox64 isa attributes.
14350 2013-03-26  Richard Biener  <rguenther@suse.de>
14352         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14353         alignment computations and rely on get_object_alignment_1
14354         for the !TYPE_P case.
14355         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14357 2013-03-26  Walter Lee  <walt@tilera.com>
14359         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14360         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14362 2013-03-25  Jeff Law  <law@redhat.com>
14364         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14365         check for INTEGRAL_TYPE_P that was missing due to checking in
14366         wrong version of prior patch.
14368 2013-03-25  Walter Lee  <walt@tilera.com>
14370         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14371         TILEGX_INSN_SHUFFLEBYTES1.
14372         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14373         shufflebytes1.
14374         (tilegx_builtins): Ditto.
14375         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14377 2013-03-25  Walter Lee  <walt@tilera.com>
14379         * config/tilegx/tilegx.md (floatsisf2): New pattern.
14380         (floatunssisf2): New pattern.
14381         (floatsidf2): New pattern.
14382         (floatunssidf2): New pattern.
14384 2013-03-25  Walter Lee  <walt@tilera.com>
14386         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14387         tests for constraint J, K, N, P.
14389 2013-03-25  Walter Lee  <walt@tilera.com>
14391         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14392         Use indirect/pcrel encoding.
14393         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14394         Ditto.
14396 2013-03-25  Steve Ellcey  <sellcey@mips.com>
14398         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14399         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14400         * config/mips/mips.c (mips_option_override): Set IMADD default.
14401         * config/mips/mips.h (PTF_AVOID_IMADD): New.
14402         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14403         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14404         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14405         * doc/invoke.texi (-mimadd/-mno-imadd): New.
14407 2013-03-25  Jeff Law  <law@redhat.com>
14409         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14410         slightly to avoid creating and folding useless trees.  Simplify
14411         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14413 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
14415         * config/i386/i386.md (*zero_extendsidi2): Merge with
14416         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
14417         * config/i386/predicates.md (x86_64_zext_operand): Rename from
14418         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
14419         targets.  Clarify comment.
14421 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14423         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14424         pass-through jump functions differently.
14425         (ipa_read_jump_function): Likewise.  Also use setter functions to set
14426         up jump functions.
14428 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14430         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14431         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14432         process it.
14433         (ipa_get_indirect_edge_target): New function.
14434         (devirtualization_time_bonus): New parameter known_aggs, pass it to
14435         ipa_get_indirect_edge_target.  Update all callers.
14436         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
14437         ipa_get_indirect_edge_target_1 instead of calling
14438         ipa_get_indirect_edge_target.
14439         (create_specialized_node): Pass aggvlas to
14440         ipcp_discover_new_direct_edges.
14442 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14444         * config/arm/arm.md (f_sels, f_seld): New types.
14445         (*cmov<mode>): New pattern.
14446         * config/arm/predicates.md (arm_vsel_comparison_operator): New
14447         predicate.
14449 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14451         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14452         POSIX-printf for mingw-hosted  builds.
14454 2013-03-25  Richard Biener  <rguenther@suse.de>
14456         PR middle-end/56694
14457         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14458         must-not-throw stmt location.
14460 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14462         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14463         Emit load-acquire versions when acq is true.
14464         (arm_emit_store_exclusive): Add rel parameter.
14465         Emit store-release versions when rel is true.
14466         (arm_split_compare_and_swap): Use acquire-release instructions
14467         instead.
14468         of barriers when appropriate.
14469         (arm_split_atomic_op): Likewise.
14470         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14471         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14472         (VUNSPEC_SLX): Likewise.
14473         (VUNSPEC_LDA): Likewise.
14474         (VUNSPEC_STL): Likewise.
14475         * config/arm/sync.md (atomic_load<mode>): New pattern.
14476         (atomic_store<mode>): Likewise.
14477         (arm_load_acquire_exclusive<mode>): Likewise.
14478         (arm_load_acquire_exclusivesi): Likewise.
14479         (arm_load_acquire_exclusivedi): Likewise.
14480         (arm_store_release_exclusive<mode>): Likewise.
14482 2013-03-25  Catherine Moore  <clm@codesourcery.com>
14484         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14485         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14486         * config/mip/predicates.md (lwsp_swsp_operand,
14487         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14488         sb16_operand, db4_operand, db7_operand, ib3_operand,
14489         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14490         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14491         andi16_operand): New predicates.
14492         * config/mips/mips.md (compression): New attribute.
14493         (enabled): New attribute.
14494         (length): Consider compression in computing length.
14495         (shift_compression): New code attribute.
14496         (*add<mode>3): New operands. Record compression.
14497         (sub<mode>3): Likewise.
14498         (one_cmpl<mode>2): Likewise.
14499         (*and<mode>3): Likewise.
14500         (*ior<mode>3): Likewise.
14501         (unnamed pattern for xor): Likewise.
14502         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14503         (*<optab><mode>3): Likewise.
14504         (*mov<mode>_internal: Likewise.
14505         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14506         (mips_unsigned_immediate_p): New.
14507         (umips_lwsp_swsp_address_p): New.
14508         (m16_based_address_p): New.
14509         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14510         (mips_unsigned_immediate_p): New prototype.
14511         (lwsp_swsp_address_p): New prototype.
14512         (m16_based_address_p): New prototype.
14513         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14514         (mips_signed_immediate_p): New function.
14515         (m16_based_address_p): New function.
14516         (lwsp_swsp_address_p): New function.
14517         (mips_print_operand_punctuation): Recognize short delay slot insns
14518         for microMIPS.add<mode>3"
14520 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14522         PR target/56720
14523         * config/arm/iterators.md (v_cmp_result): New mode attribute.
14524         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14526 2013-03-25  Richard Biener  <rguenther@suse.de>
14528         PR tree-optimization/56689
14529         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14530         any edge.
14532 2013-03-25  Richard Biener  <rguenther@suse.de>
14534         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14535         of bitmap.
14536         (memory_references): Likewise.
14537         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14538         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14539         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14540         (gather_mem_refs_in_loops): Fold into ...
14541         (analyze_memory_references): ... this.  Move initialization
14542         to tree_ssa_lim_initialize.
14543         (fill_always_executed_in): Rename to ...
14544         (fill_always_executed_in_1): ... this.
14545         (fill_always_executed_in): Move contains_call computation to
14546         this new function from ...
14547         (tree_ssa_lim_initialize): ... here.
14548         (tree_ssa_lim): Call fill_always_executed_in.
14550 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14552         * postreload.c (reload_combine): Fix code detecting returns.
14554 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14556         * function.c (emit_use_return_register_into_block): On cc0 targets,
14557         do not emit the sequence between cc0 setter and user.
14559 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14561         * config/i386/predicates.md (local_symbolic_operand): Interpret
14562         dll-imported symbols as none-local.
14564 2013-03-25  Richard Biener  <rguenther@suse.de>
14566         * tree-ssa-loop-im.c (struct depend): Remove.
14567         (struct lim_aux_data): Make depends a vec of gimples.
14568         (free_lim_aux_data): Adjust.
14569         (add_dependency): Likewise.
14570         (set_level): Likewise.
14572 2013-03-25  Richard Biener  <rguenther@suse.de>
14574         PR middle-end/56434
14575         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14576         the pointer returned by calls with ECF_MALLOC set.
14578 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14580         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14582 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14584         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14585         using MMXMODE mode iterator.
14586         (*move<mode>_internal): Merge with *movv2sf_internal and
14587         *movv2sf_internal_rex64 using MMXMODE mode iterator.
14589 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14591         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14592         (record_last_mem_set_info): Likewise.
14594         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14595         of XNEWVEC followed by memset.
14596         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14598 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14600         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14601         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14602         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14603         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14604         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14605         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14606         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14607         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14608         BARRIER_P instead of GET_CODE.
14610 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14612         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14613         inaccuracy in the probing code.
14615         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14616         (ctrapdi4): Likewise.
14618 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14620         * calls.c (expand_call): Add missing guard to code handling return
14621         of non-BLKmode structures in MSB.
14622         * function.c (expand_function_end): Likewise.
14624 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14626         * combine.c (try_combine): Adjust comment.  Do not add the set of
14627         insn #0 if the destination indirectly is set or dies in insn #2.
14628         Tidy up code to distribute a new note.
14630 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14632         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14633         also for alternatives 16 and 17.
14635 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14637         * config/i386/sse.md (*mov<mode>_internal): Merge with
14638         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
14639         Emit insn template depending on type attribute.  Use
14640         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14641         movd instead of movq mnemonic for interunit moves.  Rewrite mode
14642         attribute calculation.  Remove unit attribute calculation.
14643         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14644         Set prefix_data16 attribute for DImode ssemov types.
14645         Use Ym instead of y for SSE-MMX conversion alternatives.
14646         Reorder operand constraints.
14648 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
14650         * df.h (df_insn_delete): Adjust prototype.
14651         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14652         and let it decide whether mark the basic block dirty.
14653         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14654         * df-scan.c (df_insn_info_delete): New helper function, split
14655         off from df_insn_delete.
14656         (df_scan_free_bb_info): Use it.
14657         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14658         Likewise.
14659         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
14660         that the insn is actually an insn and it has a non-NULL basic block.
14661         Do not mark basic block dirty if only deleting a DEBUG_INSN.
14663 2013-03-22  Richard Biener  <rguenther@suse.de>
14665         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14666         dep_ref members.
14667         (mem_ref_alloc): Do not allocate them.
14668         (refs_independent_p): Do not query or maintain a cache.
14670 2013-03-22  Richard Biener  <rguenther@suse.de>
14672         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14673         (gather_mem_refs_in_loops): Do not compute it.
14674         (analyze_memory_references): Do not allocate it.
14675         (tree_ssa_lim_finalize): Do not free it.
14676         (for_all_locs_in_loop): Do not query all_refs_in_loop.
14678 2013-03-22  Richard Biener  <rguenther@suse.de>
14680         * is-a.h (as_a): Use gcc_checking_assert.
14682 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
14684         * config/aarch64/aarch64.c (aarch64_print_operand): New
14685         format specifier for printing a constant in hex.
14686         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14687         format specifier for printing second operand.
14689 2013-03-22  Richard Biener  <rguenther@suse.de>
14691         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14692         bitmaps.
14693         (gather_mem_refs_in_loops): Perform store accumulation here.
14694         (create_vop_ref_mapping_loop): Remove.
14695         (create_vop_ref_mapping): Likewise.
14696         (analyze_memory_references): Initialize refs_stored_in_loop.
14697         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14698         (record_indep_loop): Remove.
14699         (record_dep_loop): New function.
14700         (ref_indep_loop_p_1): Adjust to only walk over references
14701         in the loop, not its subloops.
14702         (ref_indep_loop_p): Rename to ...
14703         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14704         maintaining a more fine-grained cache.
14705         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14706         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14708 2013-03-22  Richard Biener  <rguenther@suse.de>
14710         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14711         (struct mem_ref): Make accesses_in_loop a vec of a vec of
14712         aggregate mem_ref_loc.
14713         (free_mem_ref_locs): Inline into ...
14714         (memref_free): ... this and adjust.
14715         (mem_ref_alloc): Adjust.
14716         (mem_ref_locs_alloc): Remove.
14717         (record_mem_ref_loc): Adjust.
14718         (get_all_locs_in_loop): Rewrite into ...
14719         (for_all_locs_in_loop): ... this iterator.
14720         (rewrite_mem_ref_loc): New functor.
14721         (rewrite_mem_refs): Use for_all_locs_in_loop.
14722         (sm_set_flag_if_changed): New functor.
14723         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14724         (ref_always_accessed): New functor.
14725         (ref_always_accessed_p): Use for_all_locs_in_loop.
14727 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14729         * tree-pass.h (PROP_gimple_lvec): New.
14730         * passes.c (dump_properties): Handle PROP_gimple_lvec.
14731         (init_optimization_passes): Move pass_lower_vector.
14732         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14733         PROP_gimple_lvec.
14734         (pass_lower_vector): Provide PROP_gimple_lvec.
14735         (pass_lower_vector_ssa): Likewise.
14736         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14738 2013-03-21  Mark Wielaard  <mjw@redhat.com>
14740         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14742 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14744         * config/i386/i386.md (*movdi_internal): Disparage slightly
14745         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14746         conversion alternatives.
14748 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14750         PR middle-end/48087
14751         * diagnostic.def (DK_WERROR): New kind.
14752         * diagnostic.h (werrorcount): Define.
14753         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14754         promoted to DK_ERROR, increment DK_WERROR counter instead of
14755         DK_ERROR counter.
14756         * toplev.c (toplev_main): Call print_ignored_options even if
14757         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14758         even if just werrorcount is non-zero.
14760         PR debug/55608
14761         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14762         on failure.
14763         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14764         (string_cst_pool_decl): New function.
14765         (optimize_one_addr_into_implicit_ptr): New function.
14766         (resolve_addr_in_expr): Optimize DWARF location expression
14767         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14768         which doesn't live in memory, but has DW_AT_location or
14769         DW_AT_const_value, or refers to a string literal, into
14770         DW_OP_GNU_implicit_pointer.
14771         (optimize_location_into_implicit_ptr): New function.
14772         (resolve_addr): If removing DW_AT_location of a variable because
14773         it was DW_OP_addr of address of the variable, but the variable doesn't
14774         live in memory, try to emit const value attribute for the initializer.
14776 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14778         * tree.h (VECTOR_TYPE_P): New macro.
14779         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14780         TYPE_MODE): Use it.
14781         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14782         VEC_COND_EXPR cannot be lvalues.
14783         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14785 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14787         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14788         Restrict the transformation to equal modes.
14790 2013-03-21  Richard Biener  <rguenther@suse.de>
14792         PR tree-optimization/39326
14793         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14794         (MEM_ANALYZABLE): Adjust.
14795         (record_mem_ref_loc): Move bitmap ops ...
14796         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14797         unanalyzable refs, do not record locations for it.
14798         (analyze_memory_references): Allocate ref zero as shared
14799         unanalyzable ref.
14800         (refs_independent_p): Do not test for unanalyzed mems here.
14801         (ref_indep_loop_p_1): Special-case disambiguation against
14802         the unanalyzed ref.
14803         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14805 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14807         * config/arm/arm-protos.h (tune_params): Add
14808         prefer_neon_for_64bits field.
14809         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14810         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14811         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14812         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14813         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14814         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14815         (arm_option_override): Handle -mneon-for-64bits new option.
14816         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14817         (prefer_neon_for_64bits): Declare new variable.
14818         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14819         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14820         (arch_enabled): Handle new arch types. Remove support for onlya8
14821         and nota8.
14822         (one_cmpldi2): Use new arch names.
14823         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14824         * config/arm/arm.opt (mneon-for-64bits): Add option.
14825         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14826         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14827         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14828         of onlya8.
14829         * doc/invoke.texi (-mneon-for-64bits): Document.
14831 2013-03-21  Richard Biener  <rguenther@suse.de>
14833         PR tree-optimization/39326
14834         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14835         (sort_bbs_in_loop_postorder_cmp): New function.
14836         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14838 2013-03-21  Richard Biener  <rguenther@suse.de>
14840         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14841         (vect_insert_into_interleaving_chain): Likewise.
14842         (vect_drs_dependent_in_basic_block): Inline ...
14843         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14844         split out from ...
14845         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14846         (vect_check_interleaving): Simplify.
14847         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14848         (vect_slp_analyze_data_ref_dependences): ... this new function.
14849         (dr_group_sort_cmp): New function.
14850         (vect_analyze_data_ref_accesses): Compute data-reference groups
14851         here instead of in vect_analyze_data_ref_dependence.  Use
14852         a more efficient algorithm.
14853         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14854         vect_slp_analyze_data_ref_dependences.  Call
14855         vect_analyze_data_ref_accesses earlier.
14856         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14857         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14858         (vect_slp_analyze_data_ref_dependences): New prototype.
14860 2013-03-21  Richard Biener  <rguenther@suse.de>
14862         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14863         ref is stored in the loop.
14864         (find_refs_for_sm): Walk only over all stores.
14865         (store_motion_loop): Allocate from lim_bitmap_obstack.
14866         (store_motion): Likewise.
14868 2013-03-21  Richard Biener  <rguenther@suse.de>
14870         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14871         Update virtual SSA form.
14873 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14875         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14876         * configure: Regenerate.
14877         * config.in: Regenerate.
14878         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14879         if !HAVE_LD_EH_FRAME_CIEV3.
14881 2013-03-21  Richard Biener  <rguenther@suse.de>
14883         * tree-cfg.c (verify_expr_no_block): New function.
14884         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14885         nor DECL_VALUE_EXPR have locations with associated blocks.
14886         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14887         (clear_unused_block_pointer): Remove code dealing with
14888         blocks in DECL_DEBUG_EXPR locations.
14890 2013-03-21  Richard Biener  <rguenther@suse.de>
14892         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14893         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14894         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14895         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14896         instead of DECL_DEBUG_EXPR_IS_FROM.
14897         * gimplify.c (gimplify_modify_expr): Likewise.
14898         * tree-cfg.c (verify_expr_location_1): Likewise.
14899         * tree-complex.c (create_one_component_var): Likewise.
14900         * tree-sra.c (create_access_replacement): Likewise.
14901         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14902         (clear_unused_block_pointer): Likewise.
14903         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14904         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14905         * var-tracking.c (var_debug_decl): Likewise.
14906         (track_expr_p): Likewise.
14907         * tree-inline.c (add_local_variables): Likewise.  Set
14908         DECL_HAS_DEBUG_EXPR_P after copying it.
14909         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14910         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14912 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14914         PR bootstrap/56656
14915         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14916         * configure: Regenerate.
14917         * config.in: Regenerate.
14918         * config/i386/i386.md (*movdf_internal): Use
14919         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14920         movd instead of movq mnemonic for interunit moves.
14921         (*movdi_internal): Ditto.
14923 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14925         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14926         (abd<mode>_3): New pattern.
14927         (aba<mode>_3): New pattern.
14928         (fabd<mode>_3): New pattern.
14930 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14932         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14933         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14934         occurrence of REGISTER_PREFIX as its empty string.
14936 2013-03-20  Jeff Law  <law@redhat.com>
14938         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14939         addititional equivalences for equality comparisons between an SSA_NAME
14940         and a constant where the SSA_NAME was set from a widening conversion.
14942 2013-03-20  Walter Lee  <walt@tilera.com>
14944         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14946 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14948         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14949         depending on type attribute.
14950         (*movti_internal): Ditto.
14951         (*movtf_internal): Ditto.
14952         (*movxf_internal): Ditto.
14953         (*movdf_internal): Ditto.
14954         (*movsf_internal): Ditto.
14956 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14958         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14959         maybe_vex for sselog1 and ssemov types.
14960         (*movdi_internal): Reorder operand constraints.
14961         (*movsi_internal): Ditto.  Set prefix attribute to
14962         maybe_vex for sselog1 and ssemov types.
14963         (*movtf_internal): Set prefix attribute to maybe_vex
14964         for sselog1 and ssemov types.
14965         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14966         DImode ssemov types.  Reorder operand constraints.
14967         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14968         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14969         attribute for SImode ssemov types.  Reorder operand constraints.
14971 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14973         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14974         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14976 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
14978         * config/rs6000/predicates.md (indexed_address, update_address_mem
14979         update_indexed_address_mem): New predicates.
14980         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14981         attribute for load/store instructions.
14982         * config/rs6000/dfp.md (movsd_store): Likewise.
14983         (movsd_load): Likewise.
14984         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14985         (unnamed HI->DI extend define_insn): Likewise.
14986         (unnamed SI->DI extend define_insn): Likewise.
14987         (unnamed QI->SI extend define_insn): Likewise.
14988         (unnamed QI->HI extend define_insn): Likewise.
14989         (unnamed HI->SI extend define_insn): Likewise.
14990         (unnamed HI->SI extend define_insn): Likewise.
14991         (extendsfdf2_fpr): Likewise.
14992         (movsi_internal1): Likewise.
14993         (movsi_internal1_single): Likewise.
14994         (movhi_internal): Likewise.
14995         (movqi_internal): Likewise.
14996         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14997         attribute for load/store instructions.
14998         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14999         instructions.
15000         (mov<mode>_softfloat): Likewise.
15001         (mov<mode>_hardfloat32): Likewise.
15002         (mov<mode>_hardfloat64): Likewise.
15003         (mov<mode>_softfloat64): Likewise.
15004         (movdi_internal32): Likewise.
15005         (movdi_internal64): Likewise.
15006         (probe_stack_<mode>): Likewise.
15008 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
15010         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15011         floating point, and decimal floating point to reload iterator.
15013         * config/rs6000/constraints.md (wl constraint): New constraints to
15014         return FLOAT_REGS if certain options are used to reduce the number
15015         of separate patterns that exist in the file.
15016         (wx constraint): Likewise.
15017         (wz constraint): Likewise.
15019         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15020         -mdebug=reg, print wg, wl, wx, and wz constraints.
15021         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15022         Initialize the reload functions for 64-bit binary/decimal floating
15023         point types.
15024         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15025         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15026         create the buffer on the stack to overcome not having a 32-bit
15027         load and store.
15028         (rs6000_emit_move): Likewise.
15029         (rs6000_secondary_memory_needed_rtx): Likewise.
15030         (rs6000_alloc_sdmode_stack_slot): Likewise.
15031         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15032         via xxlxor, just like DFmode 0.0.
15034         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15035         define as 1 if we are running on a power7 or newer.
15036         (enum r6000_reg_class_enum): Add new constraints.
15038         * config/rs6000/dfp.md (movsd): Delete, combine with binary
15039         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15040         with other moves by using conditional constraits (wg).  Use LFIWZX
15041         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
15042         (movsd splitter): Likewise.
15043         (movsd_hardfloat): Likewise.
15044         (movsd_softfloat): Likewise.
15046         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15047         binary and decimal floating point moves.
15048         (fmove_ok): New attributes to combine binary and decimal floating
15049         point moves, and to combine power6x (mfpgpr) moves along normal
15050         floating moves.
15051         (real_value_to_target): Likewise.
15052         (f32_lr): Likewise.
15053         (f32_lm): Likewise.
15054         (f32_li): Likewise.
15055         (f32_sr): Likewise.
15056         (f32_sm): Likewise.
15057         (f32_si): Likewise.
15058         (movsf): Combine binary and decimal floating point moves.  Combine
15059         power6x (mfpgpr) moves with other moves by using conditional
15060         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
15061         (mov<mode> for SFmode/SDmode); Likewise.
15062         (SFmode/SDmode splitters): Likewise.
15063         (movsf_hardfloat): Likewise.
15064         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15065         (movsf_softfloat): Likewise.
15066         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15068         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15069         wx and wz constraints.
15071         * config/rs6000/constraints.md (wg constraint): New constraint to
15072         return FLOAT_REGS if -mmfpgpr (power6x) was used.
15074         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15075         constraint.
15077         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15078         -mdebug=reg, print wg, wl, wx, and wz constraints.
15079         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15080         Initialize the reload functions for 64-bit binary/decimal floating
15081         point types.
15082         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15083         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15084         create the buffer on the stack to overcome not having a 32-bit
15085         load and store.
15086         (rs6000_emit_move): Likewise.
15087         (rs6000_secondary_memory_needed_rtx): Likewise.
15088         (rs6000_alloc_sdmode_stack_slot): Likewise.
15089         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15090         via xxlxor, just like DFmode 0.0.
15092         * config/rs6000/dfp.md (movdd): Delete, combine with binary
15093         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15094         with other moves by using conditional constraits (wg).  Use LFIWZX
15095         and STFIWX for loading SDmode on power7.
15096         (movdd splitters): Likewise.
15097         (movdd_hardfloat32): Likewise.
15098         (movdd_softfloat32): Likewise.
15099         (movdd_hardfloat64_mfpgpr): Likewise.
15100         (movdd_hardfloat64): Likewise.
15101         (movdd_softfloat64): Likewise.
15103         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15104         64-bit binary and decimal floating point moves.
15105         (FMOVE64X): Likewise.
15106         (movdf): Combine 64-bit binary and decimal floating point moves.
15107         Combine power6x (mfpgpr) moves with other moves by using
15108         conditional constraits (wg).
15109         (mov<mode> for DFmode/DDmode): Likewise.
15110         (DFmode/DDmode splitters): Likewise.
15111         (movdf_hardfloat32): Likewise.
15112         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15113         (movdf_softfloat32): Likewise.
15114         (movdf_hardfloat64_mfpgpr): Likewise.
15115         (movdf_hardfloat64): Likewise.
15116         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15117         (movdf_softfloat64): Likewise.
15118         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15119         (reload_<mode>_load): Move to later in the file so they aren't in
15120         the middle of the floating point move insns.
15121         (reload_<mode>_store): Likewise.
15123         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15124         constraint.
15126         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15127         constraint if -mdebug=reg.
15128         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15129         Enable using dd reload support if needed.
15131         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15132         binary and decimal floating point moves in rs6000.md.
15133         (movtd_internal): Likewise.
15135         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15136         decimal floating point moves.
15137         (movtf): Likewise.
15138         (movtf_internal): Likewise.
15139         (mov<mode>_internal, TDmode/TFmode): Likewise.
15140         (movtf_softfloat): Likewise.
15141         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15143         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15144         movdi_internal64, using wg constraint for move direct operations.
15145         (movdi_internal64): Likewise.
15147         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15148         MODES_TIEABLE_P for selected modes.  Print the numerical value of
15149         the various virtual registers. Use GPR/FPR first/last values,
15150         instead of hard coding the register numbers.  Print which modes
15151         have reload functions registered.
15152         (rs6000_option_override_internal): If -mdebug=reg, trace the options
15153         settings before/after setting cpu, target and subtarget settings.
15154         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15155         and for secondary reload failures in rs6000_secondary_reload_inner.
15156         (rs6000_secondary_reload_fail): Likewise.
15157         (rs6000_secondary_reload_inner): Likewise.
15159         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15160         macros for first/last GPR and FPR registers.
15161         (LAST_GPR_REGNO): Likewise.
15162         (FIRST_FPR_REGNO): Likewise.
15163         (LAST_FPR_REGNO): Likewise.
15165         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15166         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15167         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15168         (vcond<mode><mode>): Likewise.
15169         (vcondu<mode><mode>): Likewise.
15170         (vector_gtu<mode>): Likewise.
15171         (vector_gte<mode>): Likewise.
15172         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15173         to prevent the compiler from converting DImode operations to TImode.
15174         (ior<mode>3): Likewise.
15175         (and<mode>3): Likewise.
15176         (one_cmpl<mode>2): Likewise.
15177         (nor<mode>3): Likewise.
15178         (andc<mode>3): Likewise.
15180         * config/rs6000/constraints.md (wt constraint): New constraint
15181         that returns VSX_REGS if TImode is allowed in VSX registers.
15183         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15184         constant under VSX.
15186         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15187         similar to TImode, but it is restricted to being in the GPRs.
15189         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15190         TImode to occupy a single VSX register.
15192         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15193         -mvsx-timode for power7/power8.
15194         (power7 cpu): Likewise.
15195         (power8 cpu): Likewise.
15197         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15198         sure that TFmode/TDmode take up two registers if they are ever
15199         allowed in the upper VSX registers.
15200         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15201         registers.
15202         (rs6000_init_hard_regno_mode_ok): Likewise.
15203         (rs6000_debug_reg_global): Add debugging for PTImode and wt
15204         constraint.  Print if LRA is turned on.
15205         (rs6000_option_override_internal): Give an error if -mvsx-timode
15206         and VSX is not enabled.
15207         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
15208         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15209         to reg+offset addressing.  Use PTImode when checking offset
15210         addresses for validity.
15211         (reg_offset_addressing_ok_p): Likewise.
15212         (rs6000_legitimate_offset_address_p): Likewise.
15213         (rs6000_legitimize_address): Likewise.
15214         (rs6000_legitimize_reload_address): Likewise.
15215         (rs6000_legitimate_address_p): Likewise.
15216         (rs6000_eliminate_indexed_memrefs): Likewise.
15217         (rs6000_emit_move): Likewise.
15218         (rs6000_secondary_reload): Likewise.
15219         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
15220         reloads to fpr registers to continue to use reg+offset addressing,
15221         but 64-bit reloads to altivec registers need reg+reg addressing.
15222         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15223         it.  Treat LO_SUM like a PLUS operation.
15224         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15225         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15226         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15227         registers to share a register with a smaller sized type, since VSX
15228         puts scalars in the upper 64-bits.
15229         (print_operand): Add support for PTImode.
15230         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15231         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15232         registers, but don't have arithmetic support.
15233         (rs6000_memory_move_cost): Add test for VSX.
15234         (rs6000_opt_masks): Add -mvsx-timode.
15236         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15237         for TImode.
15238         (VSs): Likewise.
15239         (VSr): Use wt constraint for TImode.
15240         (VSv): Drop TImode support.
15241         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15242         (vsx_movti_64bit): Likewise.
15243         (vsx_movti_32bit): Likewise.
15244         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15245         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15246         one '?' on the appropriate output constraint.  Do not allow TImode
15247         logical operations on 32-bit systems.
15248         (vsx_ior<mode>3): Likewise.
15249         (vsx_xor<mode>3): Likewise.
15250         (vsx_one_cmpl<mode>2): Likewise.
15251         (vsx_nor<mode>3): Likewise.
15252         (vsx_andc<mode>3): Likewise.
15253         (vsx_concat_<mode>): Likewise.
15254         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15256         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15257         OPTION_MASK_VSX_TIMODE.
15258         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15259         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15261         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15262         (TI2 iterator): New iterator for TImode, PTImode.
15263         (wd mode attribute): Add values for vector types.
15264         (movti_string): Replace TI move operations with operations for TImode
15265         and PTImode.  Add support for TImode being allowed in VSX registers.
15266         (mov<mode>_string, TImode/PTImode): Likewise.
15267         (movti_ppc64): Likewise.
15268         (mov<mode>_ppc64, TImode/PTImode): Likewise.
15269         (TI mode splitters): Likewise.
15271         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15272         constraint.
15274 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
15276         PR tree-optimization/56355
15277         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15278         Also handle integers with undefined overflow.
15280 2013-03-20  Catherine Moore  <clm@codesourcery.com>
15281             Maciej W. Rozycki  <macro@codesourcery.com>
15282             Tom de Vries  <tom@codesourcery.com>
15283             Nathan Sidwell  <nathan@codesourcery.com>
15284             Iain Sandoe  <iain@codesourcery.com>
15285             Nathan Froyd  <froydnj@codesourcery.com>
15286             Chao-ying Fu  <fu@mips.com>
15288         * doc/extend.texi: (micromips, nomicromips, nocompression):
15289         Document new function attributes.
15290         * doc/invoke.texi (minterlink-compressed, mmicromips,
15291         m14k, m14ke, m14kec): Document new options.
15292         (minterlink-mips16): Update documentation.
15293         * doc/md.texi (ZC, ZD): Document new constraints.
15294         * configure.ac (gcc_cv_as_micromips): Check if linker
15295         supports the .set micromips directive.
15296         * configure: Regenerate.
15297         * config.in: Regenerate.
15298         * config/mips/mips-tables.opt: Regenerate.
15299         * config/mips/micromips.md: New file.
15300         * constraints.md (ZC, ZD): New constraints.
15301         * config/mips/predicates.md (movep_src_register): New predicate.
15302         (movep_src_operand): New predicate.
15303         (non_volatile_mem_operand): New predicate.
15304         * config/mips/mips.md (multimem): New type.
15305         (length): Differentiate between 17-bit and 18-bit branch offsets.
15306         (MOVEP1, MOVEP2): New mode iterator.
15307         (mov_<load>l): Use ZC constraint.
15308         (mov_<load>r): Likewise.
15309         (mov_<store>l): Likewise.
15310         (mov_<store>r): Likewise.
15311         (*branch_equality<mode>_inverted): Add microMIPS support.
15312         (*branch_equality<mode>): Likewise.
15313         (*jump_absolute): Likewise.
15314         (indirect_jump_<mode>): Likewise.
15315         (tablejump_<mode>): Likewise.
15316         (<optab>_internal): Likewise.
15317         (sibcall_internal): Likewise.
15318         (sibcall_value_internal): Likewise.
15319         (prefetch): Use constraint ZD.
15320         * config/mips/mips.opt (minterlink-compressed): New option.
15321         (minterlink-mips16): Now an alias for minterlink-compressed.
15322         (mmicromips): New option.
15323         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15324         (compare_and_swap_12): Likewise.
15325         (sync_add<mode>): Likewise.
15326         (sync_<optab>_12): Likewise.
15327         (sync_old_<optab>_12): Likewise.
15328         (sync_new_<optab>_12): Likewise.
15329         (sync_nand_12): Likewise.
15330         (sync_old_nand_12): Likewise.
15331         (sync_new_nand_12): Likewise.
15332         (sync_sub<mode>): Likewise.
15333         (sync_old_add<mode>): Likewise.
15334         (sync_old_sub<mode>): Likewise.
15335         (sync_new_add<mode>): Likewise.
15336         (sync_new_sub<mode>): Likewise.
15337         (sync_<optab><mode>): Likewise.
15338         (sync_old_<optab><mode>): Likewise.
15339         (sync_new_<optab><mode>): Likewise.
15340         (sync_nand<mode>): Likewise.
15341         (sync_old_nand<mode>): Likewise.
15342         (sync_new_nand<mode>): Likewise.
15343         (sync_lock_test_and_set<mode>): Likewise.
15344         (test_and_set_12): Likewise.
15345         (atomic_compare_and_swap<mode>): Likewise.
15346         (atomic_exchange<mode>_llsc): Likewise.
15347         (atomic_fetch_add<mode>_llsc): Likewise.
15348         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15349         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15350         (umips_save_restore_pattern_p): Likewise.
15351         (umips_load_store_pair_p): Likewise.
15352         (umips_output_load_store_pair): Likewise.
15353         (umips_movep_target_p): Likewise.
15354         (umips_12bit_offset_address_p): Likewise.
15355         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15356         (mips_base_mips16): Rename this...
15357         (mips_base_compression_flags): ...to this. Update all uses.
15358         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15359         (mips_mips16_decl_p): Delete.
15360         (mips_nomips16_decl_p): Delete.
15361         (mips_get_compress_on_flags): New function.
15362         (mips_get_compress_off_flags): New function.
15363         (mips_get_compress_mode): New function.
15364         (mips_get_compress_on_name): New function.
15365         (mips_get_compress_off_name): New function.
15366         (mips_insert_attributes): Support multiple compression types.
15367         (mips_merge_decl_attributes): Likewise.
15368         (umips_12bit_offset_address_p): New function.
15369         (mips_start_function_definition): Emit .set micromips directive.
15370         (mips_call_may_need_jalx_p): New function.
15371         (mips_function_ok_for_sibcall): Add microMIPS support.
15372         (mips_print_operand_punctuation): Support short delay slots and
15373         compact jumps.
15374         (umips_swm_mask, umips_swm_encoding): New.
15375         (umips_build_save_restore): New function.
15376         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15377         (was_mips16_p): Remove.
15378         (old_compression_mode): New.
15379         (mips_set_compression_mode): New function.
15380         (mips_set_current_function): Add microMIPS support.
15381         (mips_option_override): Likewise.
15382         (umips_save_restore_pattern_p): New function.
15383         (umips_output_save_restore): New function.
15384         (umips_load_store_pair_p_1): New function.
15385         (umips_load_store_pair_p): New function.
15386         (umips_output_load_store_pair_1): New function.
15387         (umips_output_load_store_pair): New function.
15388         (umips_movep_target_p) New function.
15389         (mips_prepare_pch_save): Add microMIPS support.
15390         * config/mips/mips.h (TARGET_COMPRESSION): New.
15391         (TARGET_CPU_CPP_BUILTINS): Update macro
15392         to use new compression flags and to support microMIPS.
15393         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15394         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15395         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15396         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15397         (ASM_SPEC): Support mmicromips and mno-micromips.
15398         (M16STORE_REG_P): New macro.
15399         (MIPS_CALL): Support TARGET_MICROMIPS.
15400         (MICROMIPS_J): New macro.
15401         (mips_base_mips16): Rename this...
15402         (mips_base_compression_flags): ...to this.
15403         (UMIPS_12BIT_OFFSET_P): New macro.
15404         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15405         (MULTILIB_DIRNAMES): Likewise.
15406 2013-03-20  Richard Biener  <rguenther@suse.de>
15408         PR tree-optimization/56661
15409         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15410         the result does not have to be distinct.
15412 2013-03-20  Richard Biener  <rguenther@suse.de>
15414         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15415         remap_gimple_op_r.
15417 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15418             Steven Bosscher  <steven@gcc.gnu.org>
15420         PR rtl-optimization/56605
15421         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15423 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15425         PR bootstrap/56656
15426         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15427         that require movd instead of movq.
15429 2013-03-20  Richard Biener  <rguenther@suse.de>
15431         * tree-ssa-structalias.c (struct variable_info): Add pointer
15432         to the first field of an aggregate with sub-vars.  Make
15433         this and the pointer to the next subfield its ID.
15434         (vi_next): New function.
15435         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15436         storedanything_id, integer_id): Increment by one.
15437         (new_var_info, get_call_vi, lookup_call_clobber_vi,
15438         get_call_clobber_vi): Adjust.
15439         (solution_set_expand): Simplify and speedup.
15440         (solution_set_add): Inline into ...
15441         (set_union_with_increment): ... this.  Adjust accordingly.
15442         (do_sd_constraint): Likewise.
15443         (do_ds_constraint): Likewise.
15444         (do_complex_constraint): Simplify.
15445         (build_pred_graph): Adjust.
15446         (solve_graph): Likewise.  Simplify and speedup.
15447         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15448         get_constraint_for_component_ref, get_constraint_for_1,
15449         first_vi_for_offset, first_or_preceding_vi_for_offset,
15450         create_function_info_for, create_variable_info_for_1,
15451         create_variable_info_for, intra_create_variable_infos): Adjust.
15452         (init_base_vars): Push NULL for ID zero.
15453         (compute_points_to_sets): Adjust.
15455 2013-03-20  Richard Biener  <rguenther@suse.de>
15457         * cfgloop.c (verify_loop_structure): Streamline and avoid
15458         ICEing on corrupt loop tree.
15459         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15460         loop tree.
15462 2013-03-20  Richard Biener  <rguenther@suse.de>
15464         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15465         check whether an SSA update is needed.
15467 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
15469         * config/mips/constraints.md (T): Rename to...
15470         (Yf): ...this.
15471         (U): Rename to...
15472         (Yd): ...this.
15473         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15474         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15476 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15478         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15479         (*subsi3_carryin_uxtw): Likewise.
15481 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15483         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15484         (*rorsi3_insn_uxtw): Likewise.
15486 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15488         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15489         (*extrsi5_insn_uxtw): Likewise.
15491 2013-03-19  Richard Biener  <rguenther@suse.de>
15493         PR tree-optimization/56273
15494         * passes.c (init_optimization_passes): Move second VRP after DOM.
15496 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
15498         * config/i386/i386.md (*movti_internal): Merge from
15499         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
15500         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
15501         nox64 isa attributes.
15503 2013-03-18  Richard Biener  <rguenther@suse.de>
15505         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15506         (unite): Likewise.
15507         (merge_node_constraints): Likewise.
15508         (build_succ_graph): Likewise.
15509         (valid_graph_edge): Inline into single caller.
15510         (unify_nodes): Likewise.  Use bitmap_set_bit return value
15511         and cache varinfo.
15512         (scc_visit): Fix formatting and variable use.
15513         (do_sd_constraint): Use gcc_checking_assert.
15514         (do_ds_constraint): Likewise.
15515         (do_complex_constraint): Likewise.
15516         (condense_visit): Likewise.  Cleanup.
15517         (dump_pred_graph): New function.
15518         (perform_var_substitution): Dump the pred-graph before
15519         variable substitution.
15520         (find_equivalent_node): Use gcc_checking_assert.
15521         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15523 2013-03-18  Richard Biener  <rguenther@suse.de>
15525         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15526         Remove cond_expr_stmt_list argument and do not gimplify the
15527         built expression.
15528         (vect_loop_versioning): Adjust.
15529         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15530         Cleanup to use less temporaries.
15531         (vect_create_data_ref_ptr): Cleanup.
15533 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15535         PR tree-optimization/56635
15536         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15537         require types_compatible_p types.
15539 2013-03-18  Nick Clifton  <nickc@redhat.com>
15541         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15542         spurious backslash.
15544         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15545         Add missing line to comment describing function.
15547 2013-03-18  Richard Biener  <rguenther@suse.de>
15549         PR tree-optimization/56210
15550         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15551         Handle string / character search functions.
15552         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15554 2013-03-18  Richard Biener  <rguenther@suse.de>
15556         PR middle-end/56483
15557         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15558         and implement properly.
15559         * gimple.h (gimple_cond_single_var_p): Remove.
15561 2013-03-18  Richard Biener  <rguenther@suse.de>
15563         * tree-data-ref.h (find_data_references_in_loop): Declare.
15564         * tree-data-ref.c (get_references_in_stmt): Use a stack
15565         vector pre-allocated in the callers.
15566         (find_data_references_in_stmt): Adjust.
15567         (graphite_find_data_references_in_stmt): Likewise.
15568         (create_rdg_vertices): Likewise.
15569         (find_data_references_in_loop): Export.
15570         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15571         Compute dependences here...
15572         (vect_analyze_data_refs): ...not here.  When we encounter
15573         a non-vectorizable data reference in basic-block vectorization
15574         truncate the data reference vector.  Do not bother to
15575         fixup data-dependence information for gather loads.
15576         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15577         of data references, as reported.
15579 2013-03-18  Richard Biener  <rguenther@suse.de>
15581         PR tree-optimization/3713
15582         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
15583         has_constants and expr.
15584         (stmt_has_constants): Properly valueize SSA names when deciding
15585         whether the stmt has constants.
15587 2013-03-18  Richard Biener  <rguenther@suse.de>
15589         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15590         whole function when there is nothing to do.
15591         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15592         * tree-vectorizer.c (vectorize_loops): Update virtual and
15593         loop-closed SSA once.
15594         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15596 2013-03-18  Richard Biener  <rguenther@suse.de>
15598         PR middle-end/56113
15599         * domwalk.c (bb_postorder): New global static.
15600         (cmp_bb_postorder): New function.
15601         (walk_dominator_tree): Replace scheme imposing an order for
15602         visiting dominator sons by one sorting them at the time they
15603         are pushed on the stack.
15605 2013-03-18  Richard Biener  <rguenther@suse.de>
15607         PR tree-optimization/39326
15608         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15609         (struct mem_ref): Replace mem member with ao_ref typed member.
15610         (MEM_ANALYZABLE): Adjust.
15611         (memref_eq): Likewise.
15612         (mem_ref_alloc): Likewise.
15613         (gather_mem_refs_stmt): Likewise.
15614         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15615         (execute_sm_if_changed_flag_set): Adjust.
15616         (execute_sm): Likewise.
15617         (ref_always_accessed_p): Likewise.
15618         (refs_independent_p): Likewise.
15619         (can_sm_ref_p): Likewise.
15621 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15623         PR c/56566
15624         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15625         return 1 even for !unsignedp.
15627 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15629         * config/i386/i386.md (isa): Add x64 and nox64.
15630         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15631         (*pushtf): Enable *roF alternative for x64 isa only.
15632         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
15633         mode attribute of integer alternatives to DImode for TARGET_64BIT.
15634         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
15635         (*movtf_internal): Merge from *movtf_internal_rex64 and
15636         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
15637         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
15638         nox64 isa attributes.
15639         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
15640         nox64 isa attributes.
15641         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15643 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15645         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15647 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
15649         PR target/56640
15650         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15651         class names.  Remove trailing comma after "ALL_REGS".
15653 2013-03-16  Jan Hubicka  <jh@suse.cz>
15655         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15656         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15657         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15658         of cgraph_get_create_node.
15659         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15661 2013-03-16  Jason Merrill  <jason@redhat.com>
15663         PR debug/49090
15664         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15665         with DW_AT_default_value.
15667 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
15669         * BASE-VER: Set to 4.9.0.
15671 2013-03-14  Andi Kleen  <ak@linux.intel.com>
15673         PR target/56619
15674         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15675         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15676         Document _x* TSX intrinsics.
15678 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
15679             David Holsgrove  <david.holsgrove@xilinx.com>
15681         * configure.ac: Add MicroBlaze TLS support detection.
15682         * configure: Regenerate.
15683         * config/microblaze/microblaze-protos.h
15684         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15685         symbol_mentioned_p, label_mentioned_p): Add prototypes.
15686         * config/microblaze/microblaze.c (microblaze_address_type): Add
15687         ADDRESS_TLS and tls_reloc address types.
15688         (microblaze_address_info): Add tls_reloc.
15689         (TARGET_HAVE_TLS): Define.
15690         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15691          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15692          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15693         load_tls_operand,  microblaze_call_tls_get_addr,
15694         microblaze_legitimize_tls_address): New functions.
15695         (microblaze_classify_unspec): Handle UNSPEC_TLS.
15696         (get_base_reg): Use microblaze_tls_symbol_p.
15697         (microblaze_classify_address): Handle TLS.
15698         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15699         label_mentioned_p and microblaze_tls_referenced_p.
15700         (microblaze_legitimize_address): Handle TLS.
15701         (microblaze_address_insns): Handle ADDRESS_TLS.
15702         (pic_address_needs_scratch): Handle TLS.
15703         (print_operand_address): Handle TLS.
15704         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15705         (microblaze_expand_move): Handle TLS.
15706         (microblaze_legitimate_constant_p): Check
15707         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15708         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15709         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15710         (PIC_OFFSET_TABLE_REGNUM): Set.
15711         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15712         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15713         (addsi3, movsi_internal2, movdf_internal): Update constraints
15714         * config/microblaze/predicates.md (arith_plus_operand): Define
15715         (move_operand): Redefine as move_src_operand,
15716         check microblaze_tls_referenced_p.
15718 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15720         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15721         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15723 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15725         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15726         CC mode for AND.
15728 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
15730         PR tree-optimization/53265
15731         * common.opt (Waggressive-loop-optimizations): New option.
15732         * tree-ssa-loop-niter.c: Include tree-pass.h.
15733         (do_warn_aggressive_loop_optimizations): New function.
15734         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
15735         if number_of_latch_executions returned constant.
15736         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15737         early.  If number_of_latch_executions returned constant, set
15738         nb_iterations_upper_bound back to it.
15739         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15740         field.
15741         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15742         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15744         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15745         (MULTILIB_OSDIRNAMES): Set.
15746         * genmultilib: If defaultosdirname doesn't start with :: , set
15747         defaultosdirname2 instead, clear it and emit two . multilib_raw
15748         entries instead of just one.
15750 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15752         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15753         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15754         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15755         (SUBTARGET_OVERRIDE_OPTIONS): New.
15757 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15759         PR target/49880
15760         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15761         (musermode): Convert to Var(TARGET_USERMODE).
15762         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15763         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15764         * config/sh/sh.c (sh_option_override): Use
15765         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15766         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15767         condition.
15768         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15769         TARGET_SH4.
15770         (udivsi3_i4_single, divsi3_i4_single): Use
15771         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15773 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15775         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15776         default setting.
15778 2013-03-13  Richard Biener  <rguenther@suse.de>
15780         PR tree-optimization/56608
15781         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15782         calls when vectorizing basic-blocks.
15784 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15786         PR plugins/45078
15787         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15788         tm_file.
15790 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15792         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15794 2013-03-11  Jan Hubicka  <jh@suse.cz>
15796         PR lto/56557
15797         * lto-streamer-out.c (output_symbol_p): Skip references from
15798         constructors of external variables.
15800 2013-03-11  Jan Hubicka  <jh@suse.cz>
15802         PR middle-end/56571
15803         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15804         from pseudos.
15805         * emit-rtl.c (verify_rtx_sharing): Likewise.
15806         (copy_insn_1): Likewise.
15807         * rtl.c (copy_rtx): Likewise.
15809 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15811         PR target/56591
15812         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15813         output_operand_lossage message.
15815 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15817         PR target/56470
15818         * arm.c (shift_op): Validate RTL pattern on the fly.
15819         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15820         the RTL.
15822 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15824         PR target/56347
15825         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15826         new call patterns that clobber %fr12.
15827         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15828         split and postreload patterns.
15829         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15830         registers %fr12 and %fr12R as call used.
15832 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15834         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15835         (canon_address, record_store, replace_read, check_mem_read_rtx,
15836         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15837         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15838         rest_of_handle_dse): Likewise.
15840 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15842         PR middle-end/56524
15843         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15844         Add base_optabs.
15845         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15846         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15847         (save_optabs_if_changed): Replace with...
15848         (init_tree_optimization_optabs): ...this.
15849         * optabs.c (save_optabs_if_changed): Rename to...
15850         (init_tree_optimization_optabs): ...this.  Take the optimization node
15851         as argument.  Do nothing if the base optabs are already correct.
15852         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15853         to recompute optabs.
15854         * function.h (function): Remove optabs field.
15855         * function.c (invoke_set_current_function_hook): Call
15856         init_tree_optimization_optabs.  Use the result to initialize
15857         this_fn_optabs.
15859 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15861         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15862         if GTMA_HAS_NO_INSTRUMENTATION.
15863         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15864         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15865         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15866         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15867         GTMA_HAS_NO_INSTRUMENTATION.
15869 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15871         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15872         libasan_preinit.o.
15874 2013-03-08  Marek Polacek  <polacek@redhat.com>
15875             Jakub Jelinek  <jakub@redhat.com>
15877         PR tree-optimization/56478
15878         * predict.c (is_comparison_with_loop_invariant_p): Change the
15879         type of loop_step to tree.
15880         (predict_loops): Adjust.
15881         (predict_iv_comparison): Perform the computations on double_ints.
15883 2013-03-08  Richard Biener  <rguenther@suse.de>
15885         PR tree-optimization/56570
15886         * tree-cfg.c (verify_expr_location_1): Verify locations for
15887         DECL_DEBUG_EXPR.
15888         * tree-sra.c (create_access_replacement): Strip locations
15889         from DECL_DEBUG_EXPRs.
15891 2013-03-08  Richard Biener  <rguenther@suse.de>
15893         * tree-inline.c (expand_call_inline): Do not associate
15894         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15895         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15897 2013-03-08  Richard Biener  <rguenther@suse.de>
15899         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15900         or block changes with -Og.  Fix for location / block encoding
15901         changes and PHI arguments with locations.
15903 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15905         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15906         for all counters.
15907         (struct output_info): Likewise.
15908         (register_overhead): Remove bad gcc_assert.
15909         (bitmap_find_bit): If there is only a single bitmap element, do not
15910         count a miss as a search.
15911         (print_statistics): Update for counter type changes.
15912         (dump_bitmap_statistics): Likewise.  Print headers such that they
15913         are properly lined up with the printed counters.
15915 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15917         PR tree-optimization/56559
15918         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15919         check that it has only a single use.
15921 2013-03-07  Richard Biener  <rguenther@suse.de>
15923         * doc/invoke.texi (fwhole-program): Discourage use in combination
15924         with -flto.
15926 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15928         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15930         PR tree-optimization/56539
15931         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15932         instead of GSI_CONTINUE_LINKING as last argument to
15933         force_gimple_operand_gsi.  Adjust function comment.
15935         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15936         aarch64-cores.def.
15938         PR middle-end/56548
15939         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15940         promoted mode, convert the result back to the original mode.
15942 2013-03-06  Richard Biener  <rguenther@suse.de>
15944         PR middle-end/56294
15945         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15946         (insert_updated_phi_nodes_compare_uids): New function.
15947         (update_ssa): Sort symbols_to_rename after UID before
15948         traversing it to insert PHI nodes.
15950 2013-03-06  Richard Biener  <rguenther@suse.de>
15952         PR middle-end/50494
15953         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15954         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15956         Revert
15957         2013-02-13  Richard Biener  <rguenther@suse.de>
15959         PR lto/50494
15960         * varasm.c (output_constant_def_1): Get the decl representing
15961         the constant as argument.
15962         (output_constant_def): Wrap output_constant_def_1.
15963         (make_decl_rtl): Use output_constant_def_1 with the decl
15964         representing the constant.
15965         (build_constant_desc): Optionally re-use a decl already
15966         representing the constant.
15967         (tree_output_constant_def): Adjust.
15969 2013-03-06  Joey Ye  <joey.ye@arm.com>
15971         PR lto/50293
15972         * gcc.c (convert_white_space): New function.
15973         (main): Handles white space in function name.
15975 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15977         PR target/56529
15978         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15979         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15980         to SH_DIV_CALL_TABLE for TARGET_SH2.
15981         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15982         list.
15983         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15984         call-table options.
15986 2013-03-05  Sterling Augustine  <saugustine@google.com>
15987             Cary Coutant  <ccoutant@google.com>
15989         PR debug/55364
15990         * dwarf2out.c (resolve_addr): Don't call
15991         remove_loc_list_addr_table_entries a second time for the same
15992         expression.
15994 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15996         PR debug/56510
15997         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15998         (avoid_complex_debug_insns): New function.
15999         (expand_debug_locations): Call it.
16001         PR rtl-optimization/56484
16002         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16003         lifetimes of hard registers on small register class machines.
16005 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16007         * config/microblaze/microblaze-protos.h: Rename
16008         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16009         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16010         fast_interrupt.
16011         (microblaze_fast_interrupt_function_p): New function.
16012         (microblaze_is_interrupt_handler): Rename to
16013         microblaze_is_interrupt_variant and add fast_interrupt check.
16014         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16015         (save_restore_insns): Likewise.
16016         (compute_frame_size): Likewise.
16017         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16018         (microblaze_globalize_label): Likewise.
16019         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16020         * config/microblaze/microblaze.md: Use wrapper
16021         microblaze_is_interrupt_variant.
16023 2013-03-05  Kai Tietz  <ktietz@redhat.com>
16025         * sdbout.c (sdbout_one_type): Switch to current function's section
16026         supporting cold/hot.
16028 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16030         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16031         -mxl-reorder.
16033 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16035         PR middle-end/56461
16036         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16037         if VALGRIND_GET_VBITS is defined, temporarily make object
16038         memory all defined, and restore previous valgrind addressability
16039         and definability afterwards.  Free this_object at the end.
16041         PR middle-end/56461
16042         * lra.c (lra): Call lra_clear_live_ranges if live_p,
16043         right before calling lra_create_live_ranges, also call it
16044         when clearing live_p.  Only call lra_clear_live_ranges
16045         at the end if live_p.
16047         PR middle-end/56461
16048         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16050 2013-03-05  Richard Biener  <rguenther@suse.de>
16052         PR tree-optimization/56521
16053         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16054         value-id.
16056 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
16058         PR c++/55135
16059         * except.h (remove_unreachable_eh_regions): New prototype.
16060         * except.c (remove_eh_handler_splicer): New function, split out
16061         of remove_eh_handler.
16062         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
16063         warning about running it on many EH regions one at a time.
16064         (remove_unreachable_eh_regions_worker): New function, walk the
16065         EH tree in depth-first order and remove non-marked regions.
16066         (remove_unreachable_eh_regions): New function.
16067         * tree-eh.c (mark_reachable_handlers): New function, split out
16068         from remove_unreachable_handlers.
16069         (remove_unreachable_handlers): Use mark_reachable_handlers and
16070         remove_unreachable_eh_regions.
16071         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16072         and remove_unreachable_eh_regions.
16074 2013-03-05  Richard Biener  <rguenther@suse.de>
16076         PR middle-end/56525
16077         * loop-init.c (fix_loop_structure): Remove loops in two stages,
16078         not freeing them until the end.
16080 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16082         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16084 2013-03-05  Richard Biener  <rguenther@suse.de>
16086         PR tree-optimization/56270
16087         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16088         of loads after scheduling an SLP instance.
16090 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16092         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16093         tic6x.exp.
16094         (check_gcc_parallelize): Run guality.exp as a separate job from
16095         vect.exp with unsorted.exp and $(dg_target_exps) separately from
16096         struct-layout-1.exp with stackalign.exp.
16098         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16100         PR middle-end/56461
16101         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16102         load_index sbitmap even if some bit in it isn't set.
16104         PR middle-end/56461
16105         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16106         (discover_iteration_bound_by_body_walk): Change queues to
16107         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
16108         spelling in comment.  Call safe_push on queues[bound_index] directly.
16109         Release queues[queue_index] in every iteration unconditionally.
16110         Release bounds vector.
16112         PR middle-end/56461
16113         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16114         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16115         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16116         inner_phis vector.
16118 2013-03-05  Richard Biener  <rguenther@suse.de>
16120         PR lto/56515
16121         * tree-inline.c (remap_blocks_to_null): New function.
16122         (expand_call_inline): When expanding a call stmt without
16123         an associated block inline remap all callee blocks to NULL.
16125 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16127         PR rtl-optimization/56494
16128         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16129         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16130         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16132         PR middle-end/56461
16133         * sel-sched-ir.c (free_sched_pools): Release
16134         succs_info_pool.stack[succs_info_pool.max_top] vectors too
16135         if succs_info_pool.max_top isn't -1.
16137         PR bootstrap/56509
16138         * opts.c (opts_obstack, opts_concat): Moved to...
16139         * opts-common.c (opts_obstack, opts_concat): ... here.
16141 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16143         PR middle-end/56461
16144         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16146 2013-03-04  Martin Jambor  <mjambor@suse.cz>
16148         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16149         all appropriate places.
16151 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
16153         PR tree-optimization/56424
16154         * ipa-split.c (split_function): Do not set the RSO flag if result is
16155         not by reference and its type is a register type.
16157 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
16159         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16160         (microblaze_legitimate_pic_operand): Likewise
16161         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16162         new function microblaze_legitimate_pic_operand
16163         * config/microblaze/microblaze-protos.h
16164         (microblaze_legitimate_pic_operand): Declare.
16166 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16168         * config/microblaze/predicates.md (call_insn_simple_operand):
16169         New predicate for supported rtx code types.
16170         * config/microblaze/microblaze.md (call_internal1): Use
16171         call_insn_simple_operand predicate.
16173 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16175         PR middle-end/56461
16176         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16177         partitions.ordered_remove.
16179         PR middle-end/56461
16180         * tree-vect-stmts.c (vectorizable_conversion): Don't call
16181         vec_oprnds0.create (1) for modifier == NONE.
16183         PR middle-end/56461
16184         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16185         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16186         vec_oprnds1 right before pushing anything to it for
16187         scalar_shift_arg.
16189         PR middle-end/56461
16190         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16191         set nbbs to 0 instead of having separate code path.
16192         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16193         instead of false as last argument if returning NULL.
16195 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
16197         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16198         the attribute is now called "target" instead of "option".
16199         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16200         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
16201         attribute/pragma name for TARGET_OPTION_VALID_P and
16202         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
16203         * doc/tm.texi: Regenerated.
16205 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
16207         * config/microblaze/microblaze.c:
16208         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16209         * config/microblaze/microblaze.h: Add -mxl-reorder to
16210         DRIVER_SELF_SPECS.
16211         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16212         instructions emitted if TARGET_REORDER.
16213         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16214         or 0 for -m/-mno case, but initialises as 2 to detect default use case
16215         separately.
16217 2013-03-01  Xinliang David Li  <davidxl@google.com>
16219         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16220         walk length.
16222 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
16224         PR middle-end/56461
16225         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16226         vector even when returning true.  Fix up function comment formatting.
16228         PR middle-end/56461
16229         * ira-build.c (ira_loop_nodes_count): New variable.
16230         (create_loop_tree_nodes): Initialize it.
16231         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16233         PR middle-end/56461
16234         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16235         method on dr_chain and result_chain.
16236         * tree-vect-stmts.c (vectorizable_store): Only call
16237         result_chain.create if j == 0.
16239         PR middle-end/56461
16240         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16241         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16242         before overwriting it.
16244 2013-03-01  Tobias Burnus  <burnus@net-b.de>
16246         * doc/extended.texi (C Extensions): Change order in @menu
16247         to match @node.
16248         (Other MIPS Built-in Functions): Move last MIPS entry before
16249         "picoChip Built-in Functions".
16250         (SH Built-in Functions): Move after RX Built-in Functions.
16251         * doc/gcc.texi (Introduction): Change order in @menu
16252         to match @node.
16253         * doc/md.texi (Constraints): Ditto.
16254         * gty.texi (Type Information): Ditto.
16255         (User-provided marking routines for template types): Make
16256         subsection.
16257         * doc/invoke.texi (AArch64 Options): Move before
16258         "Adapteva Epiphany Options".
16260 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
16261             Jakub Jelinek  <jakub@redhat.com>
16263         PR sanitizer/56454
16264         * asan.c (gate_asan): Lookup no_sanitize_address instead of
16265         no_address_safety_analysis attribute.
16266         * doc/extend.texi (no_address_safety_attribute): Rename to
16267         no_sanitize_address attribute, mention no_address_safety_analysis
16268         attribute as deprecated alias.
16270 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
16272         PR middle-end/56461
16273         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16274         type to vec<vec<tree> > *.
16275         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
16276         to be vec<tree> instead of vec<tree> *, set vec_defs
16277         to vNULL and call vec_defs.create (number_of_vects), adjust other
16278         uses of vec_defs.
16279         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16280         vectorizable_condition): Adjust vect_get_slp_defs callers.
16282 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16284         * config/aarch64/aarch64.c
16285         (aarch64_float_const_representable): Remove unused variable.
16287 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16289         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16291 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16293         * config/aarch64/aarch64-builtins.c
16294         (aarch64_init_simd_builtins): Make static.
16296 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16298         * config/aarch64/aarch64.c
16299         (aarch64_simd_make_constant): Make static.
16301 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16303         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16304         with no initialization to the RHS of debug statements.
16306 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16308         PR tree-optimization/56294
16309         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16310         Adjust dumping.
16311         (get_access_replacement): Do not call create_access_replacement.
16312         Assert a replacement exists.
16313         (get_repl_default_def_ssa_name): Create the replacement declaration
16314         itself.
16316 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16318         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16319         final_end_function.
16321 2013-02-28  Marek Polacek  <polacek@redhat.com>
16323         PR rtl-optimization/56466
16324         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16325         if we're changing a loop.
16326         (peel_loops_completely): Likewise.
16328 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
16330         PR c++/55813
16331         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16333 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16335         PR target/56445
16336         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16337         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16338         INTX_FTYPE_FX, FX_FTYPE_INTX.
16339         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16341 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16343         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16344         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16345         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16346         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16347         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16348         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16349         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16350         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16351         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16352         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16353         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16354         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16355         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16356         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16357         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16358         (avrxmega6): Increase max flash segments from 5 to 6.
16359         * config/avr/t-multilib: Regenerate.
16360         * config/avr/avr-tables.opt: Regenerate.
16361         * doc/avr-mmcu.texi: Regenerate.
16363 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16365         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16366         (avr_device_to_arch): Rename to avr_device_to_ld.
16367         (avr_device_to_as): New prototype.
16368         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16369         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16370         * config/avr/driver-avr.c (avr_device_to_as): New.
16371         (avr_device_to_arch): Rename to avr_device_to_ld.
16373 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16375         PR middle-end/56461
16376         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16377         method on dr_chain and result_chain.
16379         PR middle-end/56461
16380         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16381         pointer_set_destroy on not_executed_last_iteration.
16383         PR middle-end/56461
16384         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16386         PR middle-end/56461
16387         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16388         FOR_EACH_DEFINED_FUNCTION when freeing state.
16390         PR middle-end/56461
16391         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16392         pool_free.
16393         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16394         overwriting it.
16396         PR middle-end/56461
16397         * ipa-cp.c (decide_whether_version_node): Call vec_free on
16398         known_aggs[i].items and release known_aggs vector.
16400         PR middle-end/56461
16401         * ipa-reference.c (propagate): Free node_info even for alias nodes.
16403 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16405         * config/microblaze/microblaze.c (microblaze_emit_compare):
16406         Use xor for EQ/NE comparisions.
16407         * config/microblaze/microblaze.md (cstoresf4): Add constraints
16408         (cbranchsf4): Adjust operator to comparison_operator.
16410 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16412         PR middle-end/56461
16413         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16414         vector.
16415         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16416         vec_safe_push, always update *slot.
16417         (redirect_edge_var_map_clear): Use vec_free.
16418         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16419         (free_var_map_entry): Use vec_free.
16420         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16421         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16423 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
16425         PR middle-end/45472
16426         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16427         when the may_trap_p bit of the exprs being merged differs.
16428         Reorder tests for speculativeness in the logical and operator.
16430 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16432         * incpath.c (add_standard_paths): Use reconcat instead of concat
16433         where appropriate and avoid leaking memory.
16435         * opts.h: Include obstack.h.
16436         (opts_concat): New prototype.
16437         (opts_obstack): New declaration.
16438         * opts.c (opts_concat): New function.
16439         (opts_obstack): New variable.
16440         (init_options_struct): Call gcc_init_obstack on opts_obstack.
16441         (finish_options): Use opts_concat instead of concat
16442         and XOBNEWVEC instead of XNEWVEC.
16443         * opts-common.c (generate_canonical_option, decode_cmdline_option,
16444         generate_option): Likewise.
16445         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16446         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16448         PR target/56455
16449         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16450         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16452 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16454         PR middle-end/56461
16455         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16457 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16459         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16460         (arm_block_move_unaligned_straight): Likewise.
16461         (arm_adjust_block_mem): Likewise.
16463 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16465         PR target/48901
16466         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16467         temp, cond and label.
16468         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16470         PR target/52500
16471         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16472         * config/c6x/c6x.h (dbx_register_map): Update declaration.
16474         PR target/52501
16475         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16476         of prologue/epilogue functions.
16478         PR target/52550
16479         * config/tilegx/tilegx.c (tilegx_expand_prologue):
16480         Remove unused variable cfa_offset.
16481         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16483         PR target/54639
16484         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16485         type promotion to unsigned.
16487         PR target/54640
16488         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16489         for HOST_WIDE_INT of 32 bit / same size as int.
16490         (arm_block_move_unaligned_straight): Likewise.
16491         (arm_adjust_block_mem): Likewise.
16493         PR target/54662
16494         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16495         ALL_CFLAGS.
16497 2013-02-26  Marek Polacek  <polacek@redhat.com>
16499         PR tree-optimization/56426
16500         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16502 2013-02-26  Richard Biener  <rguenther@suse.de>
16504         PR target/56444
16505         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16506         unused variable loops.
16508 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16510         PR tree-optimization/56448
16511         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16512         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16513         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16514         later operands of the references, or even first operand for
16515         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16517         PR tree-optimization/56443
16518         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16519         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16520         to type_for_mode langhook.
16522 2013-02-25  Matt Turner  <mattst88@gmail.com>
16524         * doc/invoke.texi: Document r4700.
16526 2013-02-25  Richard Biener  <rguenther@suse.de>
16528         PR tree-optimization/56175
16529         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16530         split out from ...
16531         (simplify_bitwise_binary): ... here.  Also guard the conversion
16532         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16534 2013-02-25  Catherine Moore  <clm@codesourcery.com>
16536         Revert:
16537         2013-02-24  Catherine Moore  <clm@codesourcery.com>
16538             Maciej W. Rozycki  <macro@codesourcery.com>
16539             Tom de Vries  <tom@codesourcery.com>
16540             Nathan Sidwell  <nathan@codesourcery.com>
16541             Iain Sandoe  <iain@codesourcery.com>
16542             Nathan Froyd  <froydnj@codesourcery.com>
16543             Chao-ying Fu  <fu@mips.com>
16545         * doc/extend.texi: (micromips, nomicromips, nocompression):
16546         Document new function attributes.
16547         * doc/invoke.texi (minterlink-compressed, mmicromips,
16548         m14k, m14ke, m14kec): Document new options.
16549         (minterlink-mips16): Update documentation.
16550         * doc/md.texi (ZC, ZD): Document new constraints.
16551         * configure.ac (gcc_cv_as_micromips): Check if linker
16552         supports the .set micromips directive.
16553         * configure: Regenerate.
16554         * config.in: Regenerate.
16555         * config/mips/mips-tables.opt: Regenerate.
16556         * config/mips/micromips.md: New file.
16557         * constraints.md (ZC, AD): New constraints.
16558         * config/mips/predicates.md (movep_src_register): New predicate.
16559         (movep_src_operand): New predicate.
16560         (non_volatile_mem_operand): New predicate.
16561         * config/mips/mips.md (multimem): New type.
16562         (length): Differentiate between 17-bit and 18-bit branch offsets.
16563         (MOVEP1, MOVEP2): New mode iterator.
16564         (mov_<load>l): Use ZC constraint.
16565         (mov_<load>r): Likewise.
16566         (mov_<store>l): Likewise.
16567         (mov_<store>r): Likewise.
16568         (*branch_equality<mode>_inverted): Add microMIPS support.
16569         (*branch_equality<mode>): Likewise.
16570         (*jump_absolute): Likewise.
16571         (indirect_jump_<mode>): Likewise.
16572         (tablejump_<mode>): Likewise.
16573         (<optab>_internal): Likewise.
16574         (sibcall_internal): Likewise.
16575         (sibcall_value_internal): Likewise.
16576         (prefetch): Use constraint ZD.
16577         * config/mips/mips.opt (minterlink-compressed): New option.
16578         (minterlink-mips16): Now an alias for minterlink-compressed.
16579         (mmicromips): New option.
16580         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16581         (compare_and_swap_12): Likewise.
16582         (sync_add<mode>): Likewise.
16583         (sync_<optab>_12): Likewise.
16584         (sync_old_<optab>_12): Likewise.
16585         (sync_new_<optab>_12): Likewise.
16586         (sync_nand_12): Likewise.
16587         (sync_old_nand_12): Likewise.
16588         (sync_new_nand_12): Likewise.
16589         (sync_sub<mode>): Likewise.
16590         (sync_old_add<mode>): Likewise.
16591         (sync_old_sub<mode>): Likewise.
16592         (sync_new_add<mode>): Likewise.
16593         (sync_new_sub<mode>): Likewise.
16594         (sync_<optab><mode>): Likewise.
16595         (sync_old_<optab><mode>): Likewise.
16596         (sync_new_<optab><mode>): Likewise.
16597         (sync_nand<mode>): Likewise.
16598         (sync_old_nand<mode>): Likewise.
16599         (sync_new_nand<mode>): Likewise.
16600         (sync_lock_test_and_set<mode>): Likewise.
16601         (test_and_set_12): Likewise.
16602         (atomic_compare_and_swap<mode>): Likewise.
16603         (atomic_exchange<mode>_llsc): Likewise.
16604         (atomic_fetch_add<mode>_llsc): Likewise.
16605         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16606         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16607         (umips_save_restore_pattern_p): Likewise.
16608         (umips_load_store_pair_p): Likewise.
16609         (umips_output_load_store_pair): Likewise.
16610         (umips_movep_target_p): Likewise.
16611         (umips_12bit_offset_address_p): Likewise.
16612         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16613         (mips_base_mips16): Rename this...
16614         (mips_base_compression_flags): ...to this. Update all uses.
16615         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16616         (mips_mips16_decl_p): Delete.
16617         (mips_nomips16_decl_p): Delete.
16618         (mips_get_compress_on_flags): New function.
16619         (mips_get_compress_off_flags): New function.
16620         (mips_get_compress_mode): New function.
16621         (mips_get_compress_on_name): New function.
16622         (mips_get_compress_off_name): New function.
16623         (mips_insert_attributes): Support multiple compression types.
16624         (mips_merge_decl_attributes): Likewise.
16625         (umips_12bit_offset_address_p): New function.
16626         (mips_start_function_definition): Emit .set micromips directive.
16627         (mips_call_may_need_jalx_p): New function.
16628         (mips_function_ok_for_sibcall): Add microMIPS support.
16629         (mips_print_operand_punctuation): Support short delay slots and
16630         compact jumps.
16631         (umips_swm_mask, umips_swm_encoding): New.
16632         (umips_build_save_restore): New function.
16633         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16634         (was_mips16_p): Remove.
16635         (old_compression_mode): New.
16636         (mips_set_compression_mode): New function.
16637         (mips_set_current_function): Add microMIPS support.
16638         (mips_option_override): Likewise.
16639         (umips_save_restore_pattern_p): New function.
16640         (umips_output_save_restore): New function.
16641         (umips_load_store_pair_p_1): New function.
16642         (umips_load_store_pair_p): New function.
16643         (umips_output_load_store_pair_1): New function.
16644         (umips_output_load_store_pair): New function.
16645         (umips_movep_target_p) New function.
16646         (mips_prepare_pch_save): Add microMIPS support.
16647         * config/mips/mips.h (TARGET_COMPRESSION): New.
16648         (TARGET_CPU_CPP_BUILTINS): Update macro
16649         to use new compression flags and to support microMIPS.
16650         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16651         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16652         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16653         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16654         (ASM_SPEC): Support mmicromips and mno-micromips.
16655         (M16STORE_REG_P): New macro.
16656         (MIPS_CALL): Support TARGET_MICROMIPS.
16657         (MICROMIPS_J): New macro.
16658         (mips_base_mips16): Rename this...
16659         (mips_base_compression_flags): ...to this.
16660         (UMIPS_12BIT_OFFSET_P): New macro.
16661         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16662         (MULTILIB_DIRNAMES): Likewise.
16664 2013-02-25  Tom de Vries  <tom@codesourcery.com>
16666         PR rtl-optimization/56131
16667         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16668         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16669         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16671 2013-02-25  Tobias Burnus  <burnus@net-b.de>
16673         * doc/invoke.texi (-fsanitize=): Move from optimization
16674         to debugging options.
16676 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16678         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16680 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16681             Alexander Monakov  <amonakov@ispras.ru>
16683         PR middle-end/56077
16684         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16685         flush pending lists also on non-jumps.  Adjust comment.
16687 2013-02-24  Catherine Moore  <clm@codesourcery.com>
16688             Maciej W. Rozycki  <macro@codesourcery.com>
16689             Tom de Vries  <tom@codesourcery.com>
16690             Nathan Sidwell  <nathan@codesourcery.com>
16691             Iain Sandoe  <iain@codesourcery.com>
16692             Nathan Froyd  <froydnj@codesourcery.com>
16693             Chao-ying Fu  <fu@mips.com>
16695         * doc/extend.texi: (micromips, nomicromips, nocompression):
16696         Document new function attributes.
16697         * doc/invoke.texi (minterlink-compressed, mmicromips,
16698         m14k, m14ke, m14kec): Document new options.
16699         (minterlink-mips16): Update documentation.
16700         * doc/md.texi (ZC, ZD): Document new constraints.
16701         * configure.ac (gcc_cv_as_micromips): Check if linker
16702         supports the .set micromips directive.
16703         * configure: Regenerate.
16704         * config.in: Regenerate.
16705         * config/mips/mips-tables.opt: Regenerate.
16706         * config/mips/micromips.md: New file.
16707         * constraints.md (ZC, AD): New constraints.
16708         * config/mips/predicates.md (movep_src_register): New predicate.
16709         (movep_src_operand): New predicate.
16710         (non_volatile_mem_operand): New predicate.
16711         * config/mips/mips.md (multimem): New type.
16712         (length): Differentiate between 17-bit and 18-bit branch offsets.
16713         (MOVEP1, MOVEP2): New mode iterator.
16714         (mov_<load>l): Use ZC constraint.
16715         (mov_<load>r): Likewise.
16716         (mov_<store>l): Likewise.
16717         (mov_<store>r): Likewise.
16718         (*branch_equality<mode>_inverted): Add microMIPS support.
16719         (*branch_equality<mode>): Likewise.
16720         (*jump_absolute): Likewise.
16721         (indirect_jump_<mode>): Likewise.
16722         (tablejump_<mode>): Likewise.
16723         (<optab>_internal): Likewise.
16724         (sibcall_internal): Likewise.
16725         (sibcall_value_internal): Likewise.
16726         (prefetch): Use constraint ZD.
16727         * config/mips/mips.opt (minterlink-compressed): New option.
16728         (minterlink-mips16): Now an alias for minterlink-compressed.
16729         (mmicromips): New option.
16730         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16731         (compare_and_swap_12): Likewise.
16732         (sync_add<mode>): Likewise.
16733         (sync_<optab>_12): Likewise.
16734         (sync_old_<optab>_12): Likewise.
16735         (sync_new_<optab>_12): Likewise.
16736         (sync_nand_12): Likewise.
16737         (sync_old_nand_12): Likewise.
16738         (sync_new_nand_12): Likewise.
16739         (sync_sub<mode>): Likewise.
16740         (sync_old_add<mode>): Likewise.
16741         (sync_old_sub<mode>): Likewise.
16742         (sync_new_add<mode>): Likewise.
16743         (sync_new_sub<mode>): Likewise.
16744         (sync_<optab><mode>): Likewise.
16745         (sync_old_<optab><mode>): Likewise.
16746         (sync_new_<optab><mode>): Likewise.
16747         (sync_nand<mode>): Likewise.
16748         (sync_old_nand<mode>): Likewise.
16749         (sync_new_nand<mode>): Likewise.
16750         (sync_lock_test_and_set<mode>): Likewise.
16751         (test_and_set_12): Likewise.
16752         (atomic_compare_and_swap<mode>): Likewise.
16753         (atomic_exchange<mode>_llsc): Likewise.
16754         (atomic_fetch_add<mode>_llsc): Likewise.
16755         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16756         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16757         (umips_save_restore_pattern_p): Likewise.
16758         (umips_load_store_pair_p): Likewise.
16759         (umips_output_load_store_pair): Likewise.
16760         (umips_movep_target_p): Likewise.
16761         (umips_12bit_offset_address_p): Likewise.
16762         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16763         (mips_base_mips16): Rename this...
16764         (mips_base_compression_flags): ...to this. Update all uses.
16765         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16766         (mips_mips16_decl_p): Delete.
16767         (mips_nomips16_decl_p): Delete.
16768         (mips_get_compress_on_flags): New function.
16769         (mips_get_compress_off_flags): New function.
16770         (mips_get_compress_mode): New function.
16771         (mips_get_compress_on_name): New function.
16772         (mips_get_compress_off_name): New function.
16773         (mips_insert_attributes): Support multiple compression types.
16774         (mips_merge_decl_attributes): Likewise.
16775         (umips_12bit_offset_address_p): New function.
16776         (mips_start_function_definition): Emit .set micromips directive.
16777         (mips_call_may_need_jalx_p): New function.
16778         (mips_function_ok_for_sibcall): Add microMIPS support.
16779         (mips_print_operand_punctuation): Support short delay slots and
16780         compact jumps.
16781         (umips_swm_mask, umips_swm_encoding): New.
16782         (umips_build_save_restore): New function.
16783         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16784         (was_mips16_p): Remove.
16785         (old_compression_mode): New.
16786         (mips_set_compression_mode): New function.
16787         (mips_set_current_function): Add microMIPS support.
16788         (mips_option_override): Likewise.
16789         (umips_save_restore_pattern_p): New function.
16790         (umips_output_save_restore): New function.
16791         (umips_load_store_pair_p_1): New function.
16792         (umips_load_store_pair_p): New function.
16793         (umips_output_load_store_pair_1): New function.
16794         (umips_output_load_store_pair): New function.
16795         (umips_movep_target_p) New function.
16796         (mips_prepare_pch_save): Add microMIPS support.
16797         * config/mips/mips.h (TARGET_COMPRESSION): New.
16798         (TARGET_CPU_CPP_BUILTINS): Update macro
16799         to use new compression flags and to support microMIPS.
16800         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16801         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16802         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16803         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16804         (ASM_SPEC): Support mmicromips and mno-micromips.
16805         (M16STORE_REG_P): New macro.
16806         (MIPS_CALL): Support TARGET_MICROMIPS.
16807         (MICROMIPS_J): New macro.
16808         (mips_base_mips16): Rename this...
16809         (mips_base_compression_flags): ...to this.
16810         (UMIPS_12BIT_OFFSET_P): New macro.
16811         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16812         (MULTILIB_DIRNAMES): Likewise.
16814 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16816         PR target/52555
16817         * target-globals.c (save_target_globals): For init_reg_sets and
16818         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16820 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16822         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16823         * config/aarch64/t-aarch64
16824         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16826 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16828         PR inline-asm/56148
16829         * lra-constraints.c (process_alt_operands): Reload operand
16830         conflicting with earlier clobber only if no more other conflicting
16831         operands.
16833 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16835         PR sanitizer/56393
16836         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16837         if not linking a shared library.
16839 2013-02-22  Seth LaForge  <sethml@google.com>
16841         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16843 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16845         * config/arm/arm.md (split for extendsidi): Update condition.
16846         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16847         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16848         (qhs_zextenddi_cstr): Likewise.
16850 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16852         PR middle-end/56420
16853         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16854         avoid signed wrapping.
16855         (expand_mult): Handle properly multiplication by
16856         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16857         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16858         in the compiler if coeff is HOST_WIDE_INT_MIN.
16859         (expand_divmod): Don't make ext_op1 static, change it's type to
16860         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16862         PR rtl-optimization/50339
16863         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16864         field.
16865         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16866         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16867         into splitting_ashiftrt field.
16868         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16869         ASHIFTRT.
16870         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16871         choices.
16873 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16875         PR middle-end/56108
16876         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16877         are sure to go irrevocable.
16879 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16881         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16882         scalars are valid operands.
16884 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16886         PR tree-optimization/56310
16887         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16888         only matching indices and non-negative final offsets.
16889         (intersect_aggregates_with_edge): Pass src_idx to
16890         agg_replacements_to_vector.  Pass src_idx insstead of index to
16891         intersect_with_agg_replacements.
16893 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16895         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16896         instead of hard-wired defaults.
16898 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16900         * doc/invoke.texi (MIPS Options): Update documentation of the
16901         floating-point multiply-accumulate instruction restrictions.
16903 2013-02-21  Kostya Serebryany  <kcc@google.com>
16905         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16906         asan_shadow_offset on x86_64 linux.
16908 2013-02-21  Richard Biener  <rguenther@suse.de>
16910         PR tree-optimization/56415
16911         Revert
16912         2013-02-11  Richard Biener  <rguenther@suse.de>
16914         PR tree-optimization/56273
16915         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16916         first VRP run.
16918 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16920         PR bootstrap/56258
16921         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16922         instead of @itemx.
16924         PR inline-asm/56405
16925         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16926         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16928 2013-02-20  Jan Hubicka  <jh@suse.cz>
16930         PR tree-optimization/56265
16931         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16932         when target is referenced for first time.
16934 2013-02-20  Richard Biener  <rguenther@suse.de>
16936         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16937         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16938         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16939         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16940         not return anything.
16941         (rename_ssa_copies): Do not remove unused locals.
16942         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16943         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16944         * passes.c (execute_function_todo): Do not schedule unused locals
16945         removal if cleanup_tree_cfg did something.
16946         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16947         about the number of removed locals.
16949 2013-02-20  Richard Biener  <rguenther@suse.de>
16951         PR tree-optimization/56398
16952         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16954 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16956         PR tree-optimization/55334
16957         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16958         restricted pointers to arrays.
16960 2013-02-20  Richard Biener  <rguenther@suse.de>
16961             Jakub Jelinek  <jakub@redhat.com>
16963         PR tree-optimization/56396
16964         * tree-ssa-ccp.c (n_const_val): New static variable.
16965         (get_value): Return NULL for SSA names we don't have a lattice
16966         entry for.
16967         (ccp_initialize): Initialize n_const_val.
16968         * tree-ssa-copy.c (n_copy_of): New static variable.
16969         (init_copy_prop): Initialize n_copy_of.
16970         (get_value): Return NULL_TREE for SSA names we don't have a
16971         lattice entry for.
16973 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16975         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16977 2013-02-20  Richard Biener  <rguenther@suse.de>
16979         * genpreds.c (write_lookup_constraint): Do not compare first
16980         letter of the constraint again.
16982 2013-02-20  Richard Biener  <rguenther@suse.de>
16984         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16985         and ceil_log2.
16986         (get_use_iv_cost): Terminate hashtable walk when coming across
16987         an empty entry.
16989 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16991         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16992         reassociation for avx2 targets.
16994 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16996         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16997         Add version check for v8.10.a to enable microblaze_has_clz
16998         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16999         version and TARGET_PATTERN_COMPARE check
17000         * config/microblaze/microblaze.md: New clzsi2 instruction
17002 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17004         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17005         function before branching.
17007 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17009         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17010         DUMP_INSN_RTX_UID.
17011         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17013 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17015         PR middle-end/55889
17016         * sel-sched.c: Include ira.h.
17017         (implicit_clobber_conflict_p): New function.
17018         (moveup_expr): Use it.
17019         * Makefile.in (sel-sched.o): Depend on ira.h.
17021 2013-02-19  Richard Biener  <rguenther@suse.de>
17023         PR tree-optimization/56384
17024         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17025         (vn_hash_type): Split out from ...
17026         (vn_hash_constant_with_type): ... here.
17027         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17028         (vn_phi_eq): Compare types from vn_phi_s structure.
17029         (vn_phi_lookup): Populate vn_phi_s type.
17030         (vn_phi_insert): Likewise.
17032 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
17034         PR tree-optimization/56350
17035         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17036         if haven't found reduction or nested cycle operand, rather than
17037         asserting we must find it.
17039         PR tree-optimization/56381
17040         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17041         to fold_build3.
17043 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
17044             Jakub Jelinek  <jakub@redhat.com>
17046         PR target/52555
17047         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17048         (swap_optab_enable): Same.
17049         (init_all_optabs): Use argument instead of global.
17050         * tree.h (struct tree_optimization_option): New field target_optabs.
17051         * expr.h (init_all_optabs): Add argument to prototype.
17052         (TREE_OPTIMIZATION_OPTABS): New.
17053         (save_optabs_if_changed): Protoize.
17054         * optabs.h: Declare this_fn_optabs.
17055         * optabs.c (save_optabs_if_changed): New.
17056         Declare this_fn_optabs.
17057         (init_optabs): Add argument to init_all_optabs() call.
17058         * function.c (invoke_set_current_function_hook): Handle per
17059         function optabs.
17060         * function.h (struct function): New field optabs.
17061         * config/mips/mips.c (mips_set_mips16_mode): Handle when
17062         optimization_current_node has changed.
17063         * target-globals.h (save_target_globals_default_opts): Protoize.
17064         * target-globals.c (save_target_globals_default_opts): New.
17066 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17068         PR target/56347
17069         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17070         registers %fr12 and %fr12R as call used.
17072         PR target/56214
17073         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17074         and HImode, require all displacements to be an integer multiple of
17075         their mode size.
17076         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17077         only allow QImode and HImode when reload is in progress and strict is
17078         true.  Likewise for symbolic addresses.  Use base14_operand to check
17079         displacements in REG+BASE addresses.
17081 2013-02-18  Richard Biener  <rguenther@suse.de>
17083         PR tree-optimization/56366
17084         * tree-vect-loop.c (get_initial_def_for_induction): Properly
17085         handle sign-conversion of outer-loop initial induction value.
17087 2013-02-18  Richard Biener  <rguenther@suse.de>
17089         PR middle-end/56349
17090         * cfghooks.c (merge_blocks): If we merge a latch into another
17091         block adjust references to it.
17092         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17093         (verify_loop_structure): Verify that a recorded latch is in fact
17094         a latch.
17096 2013-02-18  Richard Biener  <rguenther@suse.de>
17098         PR tree-optimization/56321
17099         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17100         order SSA name release and virtual operand unlinking.
17102 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17104         * config/microblaze/microblaze.md (save_stack_block): Define.
17105         (restore_stack_block): Likewise.
17107 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17109         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17110         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17111         * config/microblaze/microblaze.c (microblaze_option_override):
17112         Bail out early for PIC modes when target does not support PIC.
17114 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17116         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17117         Replace with a microblaze version.
17118         (microblaze_trampoline_init): Adapt for microblaze.
17119         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17120         microblaze.
17122 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
17123             Dodji Seketeli  <dodji@redhat.com>
17125         PR asan/56330
17126         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17127         (instrument_mem_region_access): Do not forget to always put
17128         instrumentation of the of 'base' and 'base + len' in a "if (len !=
17129         0) statement, even for cases where either 'base' or 'base + len'
17130         are not instrumented -- because they have been previously
17131         instrumented.  Simplify the logic by putting all the statements
17132         instrument 'base + len' inside a sequence, and then insert that
17133         sequence right before the current insertion point.  Then, to
17134         instrument 'base + len', just get an iterator on that statement.
17135         And do not forget to update the pointer to iterator the function
17136         received as argument.
17138 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
17140         PR rtl-optimization/56348
17141         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17143 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
17145         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17146         (clean_graph_dump_file): Pass base to start_graph_dump.
17148 2013-02-14  Richard Henderson  <rth@redhat.com>
17150         PR target/55941
17151         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17153 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
17155         * collect2-aix.h: Define F_LOADONLY.
17157 2013-02-14  Richard Biener  <rguenther@suse.de>
17159         PR lto/50494
17160         * varasm.c (output_constant_def_1): Get the decl representing
17161         the constant as argument.
17162         (output_constant_def): Wrap output_constant_def_1.
17163         (make_decl_rtl): Use output_constant_def_1 with the decl
17164         representing the constant.
17165         (build_constant_desc): Optionally re-use a decl already
17166         representing the constant.
17167         (tree_output_constant_def): Adjust.
17169 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
17171         Fix an asan crash
17172         * asan.c (instrument_builtin_call):  Really put the length of the
17173         second source argument into src1_len.
17175 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17177         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17178         argument.  If it is false, don't create edge from then_bb to
17179         fallthru_bb.
17180         (insert_if_then_before_iter): Pass true to it.
17181         (build_check_stmt): Pass false to it.
17182         (transform_statements): Flush hash table only on extended basic
17183         block boundaries, rather than at the beginning of every bb.
17184         Don't flush hash table on nonfreeing_call_p calls.
17185         * tree-flow.h (nonfreeing_call_p): New prototype.
17186         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17188 2013-02-13  David S. Miller  <davem@davemloft.net>
17190         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17192 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
17194         PR target/56184
17195         * ira.c (max_regno_before_ira): Move from ...
17196         (ira): ... here.
17197         (fix_reg_equiv_init): Use max_regno_before_ira instead of
17198         vec_safe_length.
17200 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17202         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17204 2013-02-13  Richard Biener  <rguenther@suse.de>
17206         PR lto/56295
17207         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17208         globals in MEM_REFs.
17210 2013-02-13  Richard Biener  <rguenther@suse.de>
17212         * loop-init.c (loop_optimizer_init): Clear loop state when
17213         re-initializing preserved loops.
17214         * loop-unswitch.c (unswitch_single_loop): Return whether
17215         we unswitched the loop.  Do not verify loop state here.
17216         (unswitch_loops): When we unswitched a loop discover new loops.
17218 2013-02-13  Kostya Serebryany  <kcc@google.com>
17220         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17221         on x86_64 linux.
17222         * sanitizer.def: Rename __asan_init to __asan_init_v1.
17224 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
17226         Avoid instrumenting duplicated memory access in the same basic block
17227         * Makefile.in (asan.o): Add new dependency on hash-table.h
17228         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17229         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17230         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17231         (free_mem_ref_resources, has_mem_ref_been_instrumented)
17232         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17233         (get_mem_ref_of_assignment): New functions.
17234         (get_mem_refs_of_builtin_call): Extract from
17235         instrument_builtin_call and tweak a little bit to make it fit with
17236         the new signature.
17237         (instrument_builtin_call): Use the new
17238         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
17239         of is_gimple_builtin_call.
17240         (instrument_derefs, instrument_mem_region_access): Insert the
17241         instrumented memory reference into the hash table.
17242         (maybe_instrument_assignment): Renamed instrument_assignment into
17243         this, and change it to advance the iterator when instrumentation
17244         actually happened and return true in that case.  This makes it
17245         homogeneous with maybe_instrument_assignment, and thus give a
17246         chance to callers to be more 'regular'.
17247         (transform_statements): Clear the memory reference hash table
17248         whenever we enter a new BB, when we cross a function call, or when
17249         we are done transforming statements.  Use
17250         maybe_instrument_assignment instead of instrumentation.  No more
17251         need to special case maybe_instrument_assignment and advance the
17252         iterator after calling it; it's now handled just like
17253         maybe_instrument_call.  Update comment.
17255 2013-02-13  Richard Biener  <rguenther@suse.de>
17257         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17258         Fix loop discovery code.
17260 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
17262         PR inline-asm/56148
17263         * lra-constraints.c (process_alt_operands): Match early clobber
17264         operand with itself.  Check conflicts with earlyclobber only if
17265         the operand is not reloaded.  Prefer to reload conflicting operand
17266         if earlyclobber and matching operands are the same.
17268 2013-02-12  Richard Biener  <rguenther@suse.de>
17270         PR lto/56297
17271         * lto-streamer-out.c (write_symbol): Do not output symbols
17272         for hard register variables.
17274 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
17276         PR target/54222
17277         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17278         (umulsidi3_insn, mulsidi3_insn): New insns.
17280 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
17282         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17283         (struct tune_params): Add vec_costs field.
17284         * config/arm/arm.c (arm_builtin_vectorization_cost)
17285         (arm_add_stmt_cost): New functions.
17286         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17287         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17288         (arm_default_vec_cost): New struct of type cpu_vec_costs.
17289         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17290         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17291         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17292         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17294 2013-02-12  Richard Biener  <rguenther@suse.de>
17296         PR lto/56295
17297         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17298         decls again if possible.
17300 2013-02-12  Richard Biener  <rguenther@suse.de>
17302         PR middle-end/56288
17303         * tree-ssa.c (verify_ssa_name): Fix check, move
17304         SSA_NAME_IN_FREE_LIST check up.
17306 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
17307             Steven Bosscher   <steven@gcc.gnu.org>
17309         PR rtl-optimization/56151
17310         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17311         equal to op0 or op1, and last_insn pattern is CODE operation
17312         with MEM dest and one of the operands matches that MEM.
17314 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
17316         * doc/extend.texi: Document Function Multiversioning and "default"
17317         parameter string to target attribute.
17318         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17319         target attribute parameter is "default".
17320         (ix86_compare_version_priority): Remove checks for target attribute.
17321         (ix86_mangle_function_version_assembler_name): Change error to sorry.
17322         Remove check for target attribute equal to NULL. Add assert.
17323         (ix86_generate_version_dispatcher_body): Change error to sorry.
17325 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
17326             Jack Howarth  <howarth@bromo.med.uc.edu>
17327             Patrick Marlier  <patrick.marlier@gmail.com>
17329         PR libitm/55693
17330         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17331         define ENDFILE_SPEC as TM_DESTRUCTOR.
17332         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17334 2013-02-11  Alexander Potapenko  <glider@google.com>
17335             Jack Howarth  <howarth@bromo.med.uc.edu>
17336             Jakub Jelinek  <jakub@redhat.com>
17338         PR sanitizer/55617
17339         * config/darwin.c (cdtor_record): Rename ctor_record.
17340         (sort_cdtor_records): Rename sort_ctor_records.
17341         (finalize_dtors): New routine to sort destructors by
17342         priority before use in assemble_integer.
17343         (machopic_asm_out_destructor): Use finalize_dtors if needed.
17345 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
17347         PR rtl-optimization/56275
17348         * simplify-rtx.c (avoid_constant_pool_reference): Check that
17349         offset is non-negative and less than cmode size before
17350         calling simplify_subreg.
17352 2013-02-11  Richard Biener  <rguenther@suse.de>
17354         PR tree-optimization/56264
17355         * cfgloop.h (fix_loop_structure): Adjust prototype.
17356         * loop-init.c (fix_loop_structure): Return the number of
17357         newly discovered loops.
17358         * tree-cfgcleanup.c (repair_loop_structures): When new loops
17359         are discovered, do a full loop-closed SSA rewrite.
17361 2013-02-11  Richard Biener  <rguenther@suse.de>
17363         PR tree-optimization/56273
17364         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17365         first VRP run.
17366         (check_array_ref): Fix missing newline in dumps.
17367         (search_for_addr_array): Likewise.
17369 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
17371         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17373 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
17375         PR target/56256
17376         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17378 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
17380         PR rtl-optimization/56246
17381         * lra-constraints.c (simplify_operand_subreg): Try to reuse
17382         reload pseudo.
17383         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17384         constraints are satisfied.
17386 2013-02-08  Jeff Law  <law@redhat.com>
17388         PR debug/53948
17389         * emit-rtl.c (reg_is_parm_p): New function.
17390         * regs.h (reg_is_parm_p): New prototype.
17391         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17392         callee-clobbered registers.
17394 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
17396         PR target/56043
17397         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17398         If there is no implicit builtin declaration, just return NULL.
17400 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17402         * config/i386/sse.md (FMAMODEM): New mode iterator.
17403         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17404         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17406 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17408         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17409         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17410         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17412 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17414         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17415         (microblaze*-*-elf): Likewise.
17416         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17417         LINK_SPEC.
17418         * config/microblaze/microblaze-c.c: Add builtin defines for
17419         _LITTLE_ENDIAN and _BIG_ENDIAN.
17420         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17421         add to TARGET_DEFAULT flags.
17422         Expand ASM_SPEC and LINK_SPEC.
17423         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17424         * config/microblaze/microblaze.md: Update extendsidi2 and
17425         movdi_internal instructions to use low-order / high-order reg
17426         print_operands.
17427         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17428         options and inversemask / mask of LITTLE_ENDIAN.
17429         * config/microblaze/t-microblaze: Expand multilib options to
17430         include mlittle-endian (le) and update exceptions patterns.
17432 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
17434         PR rtl-optimization/56195
17435         * lra-constraints.c (get_reload_reg): Don't reuse regs
17436         if they have smaller mode than requested, if they have
17437         wider mode than requested, try to return a SUBREG.
17439         PR tree-optimization/56250
17440         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17441         if type is unsigned and code isn't MULT_EXPR.
17443 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17445         PR tree-optimization/56064
17446         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17447         bits according to mode.
17448         * fixed-value.h (fixed_from_double_int)
17449         (const_fixed_from_double_int): Adjust comments.
17451 2013-02-08  Richard Biener  <rguenther@suse.de>
17453         PR lto/56231
17454         * lto-streamer.h (struct data_in): Remove current_file, current_line
17455         and current_col members.
17456         * lto-streamer-out.c (lto_output_location): Stream changed bits
17457         en-block for efficiency.
17458         * lto-streamer-in.c (clear_line_info): Remove.
17459         (lto_input_location): Cache current file, line and column
17460         globally via local statics.  Read changed bits en-block.
17461         (input_function): Do not call clear_line_info.
17462         (lto_read_body): Likewise.
17463         (lto_input_toplevel_asms): Likewise.
17465 2013-02-08  Michael Matz  <matz@suse.de>
17467         PR tree-optimization/52448
17468         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17469         (nt_call_phase): New static.
17470         (add_or_mark_expr): Only mark accesses with newer phase than any
17471         call seen.
17472         (nonfreeing_call_p): New.
17473         (nt_init_block): Update nt_call_phase, mark blocks as visited.
17474         (nt_fini_block): Keep blocks marked as visited.
17475         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17477 2013-02-08  Richard Biener  <rguenther@suse.de>
17479         * ira.c (ira): Free broken dominator information.
17481 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17483         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17485 2013-02-08  Marek Polacek  <polacek@redhat.com>
17487         * cfgloop.c (verify_loop_structure): Add more checking of headers.
17489 2013-02-08  Richard Biener  <rguenther@suse.de>
17491         PR middle-end/56181
17492         * cfgloop.h (flow_loops_find): Adjust.
17493         (bb_loop_header_p): Declare.
17494         * cfgloop.c (bb_loop_header_p): New function split out from ...
17495         (flow_loops_find): ... here.  Adjust function signature,
17496         support incremental loop structure update.
17497         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
17498         * cfgloopmanip.c (fix_loop_structure): Move ...
17499         * loop-init.c (fix_loop_structure): ... here.
17500         (apply_loop_flags): Split out from ...
17501         (loop_optimizer_init): ... here.
17502         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
17503         in incremental mode, only remove dead loops here.
17505 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17507         PR target/54222
17508         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17509         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17510         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17511         (*round<mode>3.libgcc): New insns for fixed-modes.
17512         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17513         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17514         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17515         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17516         implementations.  Define to __builtin_avr_absFX,
17517         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17518         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17519         __builtin_avr_countlsFX, respectively.
17520         * config/avr/avr-c.c (target.h): Include it.
17521         (enum avr_builtin_id): New enum.
17522         (avr_resolve_overloaded_builtin): New static function.
17523         (avr_register_target_pragmas): Use it to set
17524         targetm.resolve_overloaded_builtin.
17525         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17526         tree nodes used by DEF_BUILTIN.
17527         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17528         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17529         <AVR_BUILTIN_xxBITS>: Same.
17531 2013-02-08  Richard Biener  <rguenther@suse.de>
17533         * cfgloop.c (verify_loop_structure): Properly handle
17534         a loop exiting to another loop header.
17535         * ira-int.h (ira_loops): Remove.
17536         * ira.c (ira_loops): Remove.
17537         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17538         (do_reload): Use loop_optimizer_finalize.
17539         * ira-build.c (create_loop_tree_nodes): Use get_loops and
17540         number_of_loops to access the loop tree.
17541         (more_one_region_p): Likewise.
17542         (finish_loop_tree_nodes): Likewise.
17543         (rebuild_regno_allocno_maps): Likewise.
17544         (mark_loops_for_removal): Likewise.
17545         (mark_all_loops_for_removal): Likewise.
17546         (remove_unnecessary_regions): Likewise.
17547         (ira_build): Likewise.
17548         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17550 2013-02-08  Richard Biener  <rguenther@suse.de>
17552         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17553         * ipa-pure-const.c (analyze_function): Avoid calling
17554         mark_irreducible_loops twice.
17555         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17557 2013-02-07  David S. Miller  <davem@davemloft.net>
17559         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17560         on 'reg'.
17561         * var-tracking.c (vt_add_function_parameter): Test the presence of
17562         HAVE_window_save properly and do not remap argument registers when
17563         we have a leaf function.
17565 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
17567         PR bootstrap/56227
17568         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17569         instead of "ll".
17570         * config/i386/i386.c (ix86_print_operand): Ditto.
17572 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17574         * lra-constraints.c (process_alt_operands): Fix recently added comment.
17576 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17578         PR rtl-optimization/56225
17579         * lra-constraints.c (process_alt_operands): Check that reload hard
17580         reg can hold value for strict_low_part.
17582 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17584         PR debug/56154
17585         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17586         dwarf2out_end_function.
17587         (in_first_function_p, maybe_at_text_label_p,
17588         first_loclabel_num_not_at_text_label): New variables.
17589         (dwarf2out_var_location): In the first function find out
17590         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17591         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17592         functions.
17594 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
17596         PR rtl-optimization/56178
17597         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17598         SUBREG of a register.  Tidy up related block of code.
17599         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17600         note if the source is a register or a SUBREG of a register.
17602 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17604         PR target/56228
17605         * config/rs6000/rs6000.md (ptrm): New mode attr.
17606         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17607         call_value_indirect_aix<pttrsize>,
17608         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17609         m in constraints.
17611 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17613         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17614         if -bnortl. Convert to strcmp and strncmp.
17616 2013-02-07  Alan Modra  <amodra@gmail.com>
17618         PR target/54009
17619         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17620         addresses won't wrap when offsetting.
17621         (rs6000_secondary_reload): Provide secondary reloads needed for
17622         wrapping LO_SUM addresses.
17624 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
17626         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17627         MACH, just __MACH__.
17629 2013-02-06  Richard Biener  <rguenther@suse.de>
17631         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17632         instead of calling fix_loop_structure.
17634 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
17636         PR middle-end/56217
17637         * omp-low.c (use_pointer_for_field): Return false if
17638         lower_send_shared_vars doesn't generate any copy-out code.
17640 2013-02-06  Tom de Vries  <tom@codesourcery.com>
17642         PR rtl-optimization/56131
17643         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17644         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
17645         of the label is NULL.  Add comment.
17647 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
17649         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17651         PR sanitizer/55374
17652         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17653         (STATIC_LIBTSAN_LIBS): Likewise.
17654         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17655         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17656         is defined, don't add anything else beyond that.
17657         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17658         (LINK_COMMAND_SPEC): Use them.
17660         PR tree-optimization/56205
17661         * tree-stdarg.c (check_all_va_list_escapes): Return true if
17662         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17663         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17665 2013-02-05  Richard Biener  <rguenther@suse.de>
17667         PR tree-optimization/53342
17668         PR tree-optimization/53185
17669         * tree-vectorizer.h (vect_check_strided_load): Remove.
17670         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17671         not disallow peeling for vectorized strided loads.
17672         (vect_check_strided_load): Make static and simplify.
17673         (vect_analyze_data_refs): Adjust.
17674         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17675         correctly when vectorizing strided loads.
17677 2013-02-05  Richard Biener  <rguenther@suse.de>
17679         * doc/install.texi: Refer to ISL, not PPL.
17681 2013-02-05  Jan Hubicka  <jh@suse.cz>
17683         PR tree-optimization/55789
17684         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17686 2013-02-05  Jan Hubicka  <jh@suse.cz>
17688         PR tree-optimization/55789
17689         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17690         the dead call anyway.
17692 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
17694         PR sanitizer/55374
17695         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17697 2013-02-04  Alexander Potapenko  <glider@google.com>
17698             Jack Howarth  <howarth@bromo.med.uc.edu>
17699             Jakub Jelinek  <jakub@redhat.com>
17701         PR sanitizer/55617
17702         * config/darwin.c (sort_ctor_records): Stabilized qsort
17703         on constructor priority by using original position.
17704         (finalize_ctors): New routine to sort constructors by
17705         priority before use in assemble_integer.
17706         (machopic_asm_out_constructor): Use finalize_ctors if needed.
17708 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
17710         PR libstdc++/54314
17711         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17712         about visibility on artificial decls.
17713         * config/sol2.c (solaris_assemble_visibility): Likewise.
17715 2013-02-04  Kai Tietz  <ktietz@redhat.com>
17717         PR target/56186
17718         * config/i386/i386.c (function_value_ms_64): Add additional valtype
17719         argument and improve checking of return-argument types for 16-byte
17720         modes.
17721         (ix86_function_value_1): Add additional valtype argument on call
17722         of function_value_64.
17723         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17724         handling infunction_value_64 function.
17726 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
17728         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17730 2013-02-04  Richard Biener  <rguenther@suse.de>
17732         PR tree-optimization/56188
17733         * tree-ssa-structalias.c (label_visit): Consider case with
17734         initially non-empty points-to set.
17735         (perform_var_substitution): Dump node mapping and clean up.
17737 2013-02-04  Richard Guenther  <rguenther@suse.de>
17739         PR lto/56168
17740         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17741         node prevail as last resort.
17742         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17743         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17745 2013-02-04  Richard Biener  <rguenther@suse.de>
17747         PR tree-optimization/56113
17748         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17749         Merge into ...
17750         (equiv_class_lookup_or_add): ... this.
17751         (label_visit): Adjust and fix error in previous patch.
17752         (perform_var_substitution): Adjust.
17754 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17756         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17757         * config/sh/sh4-300.md: Likewise.
17758         * config/sh/sh4a.md: Likewise.
17759         * config/sh/constraints.md: Likewise.
17760         * config/sh/sh.md: Likewise.
17761         * config/sh/netbsd-elf.h: Likewise.
17762         * config/sh/predicates.md: Likewise.
17763         * config/sh/sh-protos.h: Likewise.
17764         * config/sh/ushmedia.h: Likewise.
17765         * config/sh/linux.h: Likewise.
17766         * config/sh/sh.c: Likewise.
17767         * config/sh/superh.h: Likewise.
17768         * config/sh/elf.h: Likewise.
17769         * config/sh/sh4.md: Likewise.
17770         * config/sh/sh.h: Likewise.
17772 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17774         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17775         constraint to match_test floating_point_store_memory_operand().
17776         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17777         (base14_operand): New.
17778         (floating_point_store_memory_operand): New.
17779         (integer_store_memory_operand): Revise to use base14_operand and
17780         reg_plus_base_memory_operand.
17781         (move_dest_operand): Allow symbolic_memory_operands.
17782         (symbolic_memory_operand): Check for LO_SOM.
17783         (symbolic_operand): Change default case to break.
17784         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17785         CONST_DOUBLE values to be reloaded by putting them into memory when
17786         the destination is a floating point register.
17787         (movdf): Remove code to handle CONST_DOUBLE.
17788         (movsf): Likewise.
17789         (reload_indf_r1): New.
17790         (reload_insf_r1): New.
17791         Consistently use "Q" and "T" constraints with integer and floating
17792         point move instructions, respectively.
17793         (movdi): Remove FAIL.
17794         Change predicate for source operand unamed DImode move from
17795         general_operand to move_src_operand.
17796         (umulsidi3): Change predicate for destination operand to
17797         register_operand.
17798         Likewise for similar unamed patterns.
17799         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17800         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17801         (hppa_legitimize_address): Simplify mask calculation.
17802         (pa_emit_move_sequence): Revised handling of secondary reloads from
17803         REG+D addresses for floating point loads and stores.  Directly handle
17804         loading CONST0_RTX (mode) to a floating point register.
17805         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17806         to floating point registers.  Don't restrict secondary reloads to
17807         floating point registers to integer modes.  Revise some comments and
17808         cleanup some code.
17809         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17810         (pa_legitimate_address_p): New.
17811         (pa_legitimize_reload_address): New.
17812         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17813         (STRICT_REG_OK_FOR_BASE_P): New.
17814         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17815         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17817 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17818             Andrew Dixie  <andrewd@gentrack.com>
17820         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17821         flag set.
17823 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17825         * expmed.c (extract_bit_field_1): Pass the full width of the
17826         structure to get_best_reg_extraction_insn.
17828 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17830         PR target/54601
17831         * configure.ac (use_cxa_atexit): Add AIX.
17832         * configure: Regenerate.
17834         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17836 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17838         PR debug/54793
17839         * final.c (need_profile_function): New variable.
17840         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17841         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17842         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17843         notes, targetm.asm_out.function_prologue doesn't emit anything,
17844         HAVE_prologue and profiler should be emitted before prologue,
17845         set need_profile_function instead of emitting it.
17846         (final_scan_insn): If need_profile_function, emit
17847         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17848         NOTE_INSN_FUNCTION_BEG note.
17850 2013-02-01  Richard Henderson  <rth@redhat.com>
17852         * config/rs6000/rs6000.md (smulditi3): New.
17853         (umulditi3): New.
17855         * config/alpha/alpha.md (umulditi3): New.
17857 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17859         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17860         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17862 2013-02-01  Richard Biener  <rguenther@suse.de>
17864         PR tree-optimization/56113
17865         * tree-ssa-structalias.c (label_visit): Reduce work for
17866         single-predecessor nodes.
17868 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17870         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17871         range isn't testing for zero.
17873 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17875         PR middle-end/56113
17876         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17878 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17879             Nick Clifton  <nickc@redhat.com>
17881         * config/v850/constraints.md (Q): Define as a memory constraint.
17882         * config/v850/predicates.md (label_ref_operand): New predicate.
17883         (e3v5_shift_operand): New predicate.
17884         (ior_operator): New predicate.
17885         * config/v850/t-v850: Add e3v5 multilib.
17886         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17887         (v850_gen_movdi): Prototype.
17888         * config/v850/v850.c: Add support for e3v5 architecture.
17889         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17890         TARGET_V850E_UP.
17891         (construct_save_jarl): Add e3v5 long JARL support.
17892         (v850_adjust_insn_length): New function.  Adjust length of call
17893         insns when using e3v5 instructions.
17894         (v850_gen_movdi): New function: Generate instructions to move a
17895         DImode value.
17896         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17897         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17898         (TARGET_USE_FPU): Enable for e3v5.
17899         (CONST_OK_FOR_W): New macro.
17900         (ADJUST_INSN_LENGTH): Define.
17901         * config/v850/v850.md (UNSPEC_LOOP): Define.
17902         (attr cpu): Add v850e3v5.
17903         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17904         (movdi): New pattern.
17905         (movdi_internal): New pattern.
17906         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17907         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17908         (cstoresf4): Likewise.
17909         (cstoredf4): Likewise.
17910         (insv): New pattern.
17911         (rotlso3_a): New pattern.
17912         (rotlsi3_b): New pattern
17913         (rotlsi3_v850e3v5): New pattern.
17914         (doloop_begin): New pattern.
17915         (fix_loop_counter): New pattern.
17916         (doloop_end): New pattern.
17917         (branch_normal): Add e3v5 long branch support.
17918         (branch_invert): Likewise.
17919         (branch_z_normal): Likewise.
17920         (branch_z_invert): Likewise.
17921         (branch_nz_normal): Likewise.
17922         (branch_nz_invert): Likewise.
17923         (call_internal_short): Add e3v5 register-indirect JARL support.
17924         (call_internal_long): Likewise.
17925         (call_value_internal_short): Likewise.
17926         (call_value_internal_long): Likewise.
17927         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17928         (mloop): New option.
17929         * config.gcc: Add support for configuring v840e3v5 target.
17930         * doc/invoke.texi: Document new v850 specific command line options.
17932 2013-01-31  Paul Koning  <ni1d@arrl.net>
17934         PR debug/55059
17935         PR debug/54508
17936         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17937         children if parent is a class.
17938         (prune_unused_types_prune): Don't add DW_AT_declaration.
17940 2013-01-31  Richard Biener  <rguenther@suse.de>
17942         PR tree-optimization/56157
17943         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17944         match up operand with SLP child.
17946 2013-01-31  Jason Merrill  <jason@redhat.com>
17948         PR debug/54410
17949         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17950         parameters the first time.
17951         (gen_scheduled_generic_parms_dies): Check completeness here.
17953 2013-01-31  Richard Biener  <rguenther@suse.de>
17955         PR middle-end/53073
17956         * common.opt (faggressive-loop-optimizations): New flag,
17957         enabled by default.
17958         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17959         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17960         infer_loop_bounds_from_undefined by it.
17962 2013-01-31  Richard Biener  <rguenther@suse.de>
17964         PR tree-optimization/56150
17965         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17966         visit virtual operands.
17967         (find_uses_to_rename_bb): Likewise.
17969 2013-01-31  Richard Biener  <rguenther@suse.de>
17971         PR tree-optimization/56150
17972         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17973         mixed store non-store stmts.
17975 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17977         PR sanitizer/55374
17978         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17979         LIBASAN_EARLY_SPEC is defined.
17980         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17981         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17982         before %o.
17983         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17985         PR c++/55742
17986         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17987         invalid args instead of ICEing on it.
17988         (ix86_valid_target_attribute_tree): Return error_mark_node if
17989         ix86_valid_target_attribute_inner_p failed.
17990         (ix86_valid_target_attribute_p): Return false only if
17991         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17992         target("default") attribute.
17993         (sorted_attr_string): Change argument from const char * to tree,
17994         merge in all target attribute arguments rather than just one.
17995         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17996         instead of free.  Avoid using strcat.
17997         (ix86_mangle_function_version_assembler_name): Mangle
17998         target("default") as if no target attribute is present.  Adjust
17999         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
18000         instead of xmalloc and XDELETEVEC instead of free.
18001         (ix86_function_versions): Don't return true if one of the decls
18002         doesn't have target attribute.  If they don't and one of the decls
18003         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
18004         sorted_attr_string caller.  Use XDELETEVEC instead of free.
18005         (ix86_supports_function_versions): Remove.
18006         (make_name): Fix up formatting.
18007         (make_dispatcher_decl): Remove resolver_name and its initialization.
18008         Avoid leaking memory.
18009         (is_function_default_version): Return true if there is
18010         target("default") attribute rather than no target attribute at all.
18011         (make_resolver_func): Avoid leaking memory.
18012         (ix86_generate_version_dispatcher_body): Likewise.
18013         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18014         * target.def (supports_function_versions): Remove.
18015         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18016         * doc/tm.texi: Regenerated.
18018 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
18020         PR rtl-optimization/56144
18021         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18022         for values with side effects.
18024 2013-01-30  Richard Biener  <rguenther@suse.de>
18026         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18027         (sparseset_pop): Likewise.
18028         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
18029         to be able to use quick_push in the worker loop.
18031 2013-01-30  Marek Polacek  <polacek@redhat.com>
18033         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18035 2013-01-30  Richard Biener  <rguenther@suse.de>
18037         PR lto/56147
18038         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18040 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
18042         PR tree-optimization/56064
18043         * fixed-value.c (fixed_from_double_int): New function.
18044         * fixed-value.h (fixed_from_double_int): New prototype.
18045         (const_fixed_from_double_int): New static inline function.
18046         * fold-const.c (native_interpret_fixed): New static function.
18047         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18048         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18049         (native_encode_fixed): New static function.
18050         (native_encode_expr) <FIXED_CST>: Use it.
18051         (native_interpret_int): Move double_int worker code to...
18052         * double-int.c (double_int::from_buffer): ...this new static method.
18053         * double-int.h (double_int::from_buffer): Prototype it.
18055 2013-01-30  Richard Biener  <rguenther@suse.de>
18057         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18058         New pointer-map and obstack.
18059         (init_alias_vars): Allocate pointer-map and obstack.
18060         (delete_points_to_sets): Free them.
18061         (find_what_var_points_to): Cache result.
18062         (find_what_p_points_to): Adjust for changed interface of
18063         find_what_var_points_to.
18064         (compute_points_to_sets): Likewise.
18065         (ipa_pta_execute): Likewise.
18067 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18069         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18070         * configure: Regenerate.
18071         * config.in: Regenerate.
18072         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18073         #nobits/#progbits if supported.
18075 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
18077         PR target/56121
18078         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18079         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18080         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18082 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18084         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18085         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18087 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18089         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18090         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18092 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18094         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18095         declaration.
18096         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18097         * config/arm/cortex-a7.md: New bypasses using
18098         arm_mac_accumulator_is_result.
18100 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18102         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
18103         (cortex_a7_neon_mla): Likewise.
18104         (cortex_a7_fpfmad): New reservation.
18105         (cortex_a7_fpmacs): Use ffmas and update required units.
18106         (cortex_a7_fpmuld): Update required units and latency.
18107         (cortex_a7_fpmacd): Likewise.
18108         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18109         (cortex_a7_neon). Likewise.
18110         (bypass) Update participating units.
18112 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18114         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18115         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18116         from fmac to ffma.
18117         * config/arm/vfp11.md (vfp_farith): Use ffmas.
18118         (vfp_fmul): Use ffmad.
18119         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18120         (cortex_r4_fmacd): Use ffmad.
18121         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18122         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
18123         (cortex_a9_fmacd): Use ffmad.
18124         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18125         (cortex_a8_vfp_macd): Use ffmad.
18126         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18127         (cortex_a5_fpmacd): Use ffmad.
18128         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18129         (cortex_a15_vfp_macd): Use ffmad.
18130         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18132 2013-01-29  Jason Merrill  <jason@redhat.com>
18134         PR libstdc++/54314
18135         * varasm.c (default_assemble_visibility): Don't warn about
18136         visibility on artificial decls.
18138 2013-01-29  Richard Biener  <rguenther@suse.de>
18140         PR tree-optimization/56113
18141         * tree-ssa-structalias.c (equiv_class_lookup): Also return
18142         the bitmap leader.
18143         (label_visit): Free duplicate bitmaps and record the leader instead.
18144         (perform_var_substitution): Adjust.
18146 2013-01-29  Richard Biener  <rguenther@suse.de>
18148         PR tree-optimization/55270
18149         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18150         the CFG, schedule loops for fixup.
18152 2013-01-29  Nick Clifton  <nickc@redhat.com>
18154         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18155         SP_REG.
18157 2013-01-28  Leif Ekblad  <leif@rdos.net>
18159         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18160         * config/i386/i386.h (TARGET_RDOS): New macro.
18161         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18162         * config/i386/i386.c (ix86_option_override_internal): For 64bit
18163         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18164         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18165         DEFAULT_LARGE_SECTION_THRESHOLD.
18166         * config/i386/i386.md (R14_REG, R15_REG): New constants.
18167         * config/i386/rdos.h: New file.
18168         * config/i386/rdos64.h: New file.
18170 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
18172         PR other/54814
18173         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18174         TEST_HARD_REG_BIT.
18176 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18178         PR rtl-optimization/56117
18179         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18180         call cselib_lookup_from_insn on the MEM before calling
18181         add_insn_mem_dependence.
18183 2013-01-28  Richard Biener  <rguenther@suse.de>
18185         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18186         to a stmt that didn't have one.
18187         (copy_phis_for_bb): Likewise for PHI arguments.
18188         (copy_debug_stmt): Likewise for debug stmts.
18190 2013-01-28  Richard Biener  <rguenther@suse.de>
18192         PR tree-optimization/56034
18193         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18194         (partition_builtin_p): Adjust.
18195         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
18196         it is the last partition.
18197         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18198         up the vertex for the definition.
18199         (classify_partition): Classify whether a partition is a
18200         PKIND_REDUCTION, thus has uses outside of the loop.
18201         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18202         Merge all PKIND_REDUCTION partitions into the last partition.
18203         (tree_loop_distribution): Seed partitions from reductions as well.
18205 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18207         PR tree-optimization/56125
18208         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18209         pow(x,c) into sqrt(x) * powi(x, n/2) or
18210         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18211         optimizing for size.
18212         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18213         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18214         integer.
18216         PR tree-optimization/56094
18217         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18218         to UNKNOWN_LOCATION while gimplifying expr.
18220 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
18222         PR target/56114
18223         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18224         operand 0 in movabs insn template for -masm=intel asm alternative.
18225         (*movabs<mode>_2): Ditto for operand 1.
18227 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
18229         PR target/54663
18230         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18231         of microblaze-c.o
18233 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18235         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18236         tm_file.
18238 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18240         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18241         Undef to avoid warning.
18243 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
18245         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18246         * configure: Regenerate.
18248 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
18250         PR tree-optimization/56098
18251         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18252         for stmts with volatile ops.
18253         (cond_store_replacement): Don't optimize if assign has volatile ops.
18254         (cond_if_else_store_replacement_1): Don't optimize if either
18255         then_assign or else_assign have volatile ops.
18256         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18257         volatile ops.
18259 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18261         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18263 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18265         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18266         missing ':' in asm example.
18268 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
18270         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18271         entries into lane and laneq entries.
18272         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18273         Remove AdvSIMD scalar modes.
18274         (aarch64_sq<r>dmulh_laneq<mode>): New.
18275         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18276         modes.
18277         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18278         builtin implementations to relfect changes in RTL in aarch64-simd.md.
18279         * config/aarch64/iterators.md (VCOND): New.
18280         (VCONQ): New.
18282 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18284         PR target/54222
18285         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18286         Add NULL LIBNAME argument to existing definitions.
18287         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18288         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18289         * config/avr/avr.c (DEF_BUILTIN): Same.
18290         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18291         (avr_expand_builtin): Expand to a vanilla call if a libgcc
18292         implementation is available (DECL_ASSEMBLER_NAME is set).
18293         (avr_fold_absfx): New static function.
18294         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18295         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18296         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18297         AVR_BUILTIN_ABSLLK.
18298         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18299         (abshk, absk, abslk, absllk): Provide as static inline functions.
18301 2013-01-25  Marek Polacek  <polacek@redhat.com>
18303         PR tree-optimization/56035
18304         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18306 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
18308         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18309         (*movtf_internal_rex64): Add (!o,C) alternative
18310         (*movxf_internal_rex64): Ditto.
18311         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18313 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
18315         * doc/invoke.texi: fix typo.
18316         * doc/objc.texi: fix typo.
18318 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
18320         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18321         for the first two alternatives.
18323 2013-01-24  Diego Novillo  <dnovillo@google.com>
18325         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
18326         (ggc-zone.o): Remove.
18327         * configure.ac: Remove option --with-gc.
18328         * configure: Re-generate.
18329         * doc/install.texi: Remove documentation for --with-gc.
18330         * gengtype.c (write_enum_defn): Remove.  Update all users.
18331         (write_Types_process_field): Remove generation of gt_e_* argument.
18332         (output_type_enum): Remove.  Update all users.
18333         (write_enum_defn): Remove.  Update all users.
18334         (enum alloc_zone): Remove.  Update all users.
18335         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18336         * ggc-common.c (ggc_splay_alloc): Remove first argument.
18337         Update all callers.
18338         (struct ptr_data): Remove field TYPE.  Update all users.
18339         (gt_pch_note_object): Remove argument TYPE.  Update all users.
18340         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18341         Update all users.
18342         * ggc-none.c (ggc_alloc_typed_stat): Remove.
18343         (struct alloc_zone): Remove.
18344         (ggc_internal_alloc_zone_stat): Remove.
18345         (ggc_internal_cleared_alloc_zone_stat): Remove.
18346         * ggc-page.c (ggc_alloc_typed_stat): Remove.
18347         (ggc_pch_count_object): Remove last argument.  Update all users.
18348         (ggc_pch_alloc_object): Remove last argument.  Update all users.
18349         (struct alloc_zone): Remove.
18350         * ggc-zone.c: Remove.
18351         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
18352         (struct alloc_zone): Remove.
18353         (ggc_alloc_typed_stat): Remove.
18354         (ggc_alloc_typed): Remove.
18355         (ggc_splay_alloc): Remove first argument.
18356         (rtl_zone): Remove.  Update all users.
18357         (tree_zone): Remove.  Update all users.
18358         (tree_id_zone): Remove.  Update all users.
18359         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
18360         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
18361         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
18362         * tree-ssanames.c: Remove references to zone allocator in comments.
18364 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
18366         * config/avr/avr.c (avr_out_fract): Make register numbers that
18367         might be outside of source operand signed.
18369 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
18371         * config/i386/constraints.md (Yf): New constraint.
18372         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18373         of f constraint to conditionaly disable x87 register preferences.
18374         (*movdf_internal): Ditto.
18375         (*movsf_internal): Ditto.
18377 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
18379         PR inline-asm/55934
18380         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18381         that have operands with impossible constraints.
18382         Add a FIXME for a speed-up opportunity.
18383         * lra-constraints.c (process_alt_operands): Verify that a class
18384         selected from constraints on asms is valid for the operand mode.
18385         (curr_insn_transform): Remove incorrect comment.
18387 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
18389         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18390         TOC operand is a valid symbol ref in the constant pool.
18392 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18394         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18396 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
18398         PR target/54222
18399         * config/avr/stdfix.h: New file.
18400         * t-avr (stdfix-gcc.h): New rule to build it.
18401         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18403 2013-01-23  Kostya Serebryany  <kcc@google.com>
18405         * config/darwin.h: remove dependency on
18406         CoreFoundation (asan on Mac OS).
18408 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
18410         PR target/49069
18411         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18412         instead of cmpdi_operand for first comparison operand.
18413         Don't assert that comparison operands aren't both constants.
18415 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18417         * doc/install.texi (Downloading the Source): Update references to
18418         downloading separate components.
18420 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18422         * doc/extend.texi (__int128): Improve grammar.
18424 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
18426         PR target/56028
18427         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18428         alternative to (o,r).
18429         (*movdi_internal_rex64): Remove (!o,n) alternative.
18430         (DImode immediate->memory splitter): Remove.
18431         (DImode immediate->memory peephole2): Remove.
18432         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18433         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18434         alternative to (!o,*r).
18435         (*movtf_internal_sse): New pattern.
18436         (*movxf_internal_rex64): New pattern.
18437         (*movxf_internal): Disable for TARGET_64BIT.
18438         (*movdf_internal_rex64): Remove (!o,F) alternative.
18440 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
18442         PR middle-end/56074
18443         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18444         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18445         * tree-vect-loop-manip.c (find_loop_location): Also ignore
18446         stmt locations where LOCATION_LOCUS of the stmt location is
18447         UNKNOWN_LOCATION or BUILTINS_LOCATION.
18449         PR target/55686
18450         * config/i386/i386.md (UNSPEC_STOS): New.
18451         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18452         *strsetqi_1): Add UNSPEC_STOS.
18454 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
18456         PR c++/56067
18457         * doc/invoke.texi: Remove left over -Wsynth example.
18459 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18461         PR tree-optimization/56051
18462         * fold-const.c (fold_binary_loc): Don't fold
18463         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18464         a narrowing conversion, or widening conversion from signed
18465         to unsigned.
18467 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18469         PR rtl-optimization/56023
18470         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18471         dependent on debug instruction.
18473 2013-01-21  Martin Jambor  <mjambor@suse.cz>
18475         PR middle-end/56022
18476         * function.c (allocate_struct_function): Call
18477         invoke_set_current_function_hook earlier.
18479 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18481         * reload1.c (init_reload): Only initialize reload_obstack
18482         during the first call.
18484 2013-01-21  Marek Polacek  <polacek@redhat.com>
18486         * cfgloop.c (verify_loop_structure): Fix up grammar.
18488 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
18490         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18491         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18493 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18495         PR target/56058
18496         * config/arm/marvell-pj4.md: Update copyright year.
18497         Fix up use of alu to alu_reg and simple_alu_imm.
18499 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18501         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18503 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
18505         PR target/55433
18506         * lra-constraints.c (curr_insn_transform): Don't reuse original
18507         insn for secondary memory move when memory mode should be different.
18509 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18511         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18512         atomic_storedi_1): New patterns.
18514 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18516         btver2 pipeline descriptions.
18517         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18518         descriptions.
18519         * config/i386/i386.md (btver2_decode): New type attributes.
18520         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18521         type attributes.
18522         * config/i386/btver2.md: New file describing btver2 pipelines.
18524 2013-01-19  Andrew Pinski  <apinski@cavium.com>
18526         PR tree-optimization/52631
18527         * tree-ssa-sccvn (visit_use): Before looking up the original
18528         statement, try looking up the simplified expression.
18530 2013-01-19  Anthony Green  <green@moxielogic.com>
18532         * config/moxie/moxie.c (moxie_expand_prologue): Set
18533         current_function_static_stack_size.
18535 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18537         PR tree-optimization/56029
18538         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18539         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18541 2013-01-18  Sharad Singhai  <singhai@google.com>
18543         PR tree-optimization/55995
18544         * dumpfile.c (dump_loc): Print location only if available.
18545         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18547 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
18549         PR target/55433
18550         * lra-constraints.c (curr_insn_transform): Reuse original insn for
18551         secondary memory move.
18552         (inherit_reload_reg): Use rclass instead of cl for
18553         check_secondary_memory_needed_p.
18555 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18557         PR middle-end/56015
18558         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18559         the case where writing real complex part of target modifies op1.
18561 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
18563         * config/aarch64/aarch64-simd.md
18564         (aarch64_vcond_internal<mode>): Handle unordered cases.
18565         * config/aarch64/iterators.md (v_cmp_result): New.
18567 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
18568             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18570         * config/arm/marvell-pj4.md: New file.
18571         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18572         * config/arm/arm.md (generic_sched): Add marvell_pj4.
18573         (generic_vfp): Likewise.
18574         * config/arm/arm-cores.def: Add marvell-pj4.
18575         * config/arm/arm-tune.md: Regenerate.
18576         * config/arm/arm-tables.opt: Regenerate.
18577         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18578         * doc/invoke.texi: Document marvell-pj4.
18580 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
18582         * config/aarch64/arm_neon.h: Map scalar types to standard types.
18584 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18586         PR debug/54114
18587         PR debug/54402
18588         PR debug/49888
18589         * var-tracking.c (negative_power_of_two_p): New.
18590         (global_get_addr_cache, local_get_addr_cache): New.
18591         (get_addr_from_global_cache, get_addr_from_local_cache): New.
18592         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
18593         heading comment.
18594         (vt_stack_offset_p): Remove.
18595         (vt_canon_true_dep): Always canonicalize loc's address.
18596         (clobber_overlapping_mems): Make sure we have a MEM.
18597         (local_get_addr_clear_given_value): New.
18598         (val_reset): Clear local cached entries.
18599         (compute_bb_dataflow): Create and release the local cache.
18600         Disable duplicate MEMs clobbering.
18601         (emit_notes_in_bb): Clobber MEMs likewise.
18602         (vt_emit_notes): Create and release the local cache.
18603         (vt_initialize, vt_finalize): Create and release the global
18604         cache, respectively.
18605         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18607 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18609         PR libmudflap/53359
18610         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18611         not found in the symtab.
18613 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18615         PR debug/56006
18616         PR rtl-optimization/55547
18617         PR rtl-optimization/53827
18618         PR debug/53671
18619         PR debug/49888
18620         * alias.c (offset_overlap_p): New, factored out of...
18621         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
18622         the conservative special case for symbolic constants.  Don't
18623         adjust zero sizes on alignment.
18625 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
18627         PR rtl-optimization/52573
18628         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18629         REG_UNUSED for the same register.
18631 2013-01-17  Richard Biener  <rguenther@suse.de>
18632             Marek Polacek  <polacek@redhat.com>
18634         PR rtl-optimization/55833
18635         * loop-unswitch.c (unswitch_loops): Move loop verification...
18636         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
18637         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18638         Set it to true when we're removing a loop from hierarchy tree in
18639         an irreducible region.
18640         (fix_bb_placements): Adjust caller.
18641         (fix_loop_placements): Likewise.
18643 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
18645         * config/avr/builtins.def (DEF_BUILTIN): Factor out
18646         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18647         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18648         Remove ID.  Adjust comments.
18649         * config/avr/avr-c.c (avr_builtin_name): Remove.
18650         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18651         * config/avr/avr.c (avr_tolower): New static function.
18652         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
18653         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18654         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18655         default expansion.
18657 2013-01-17  Jan Hubicka  <jh@suse.cz>
18659         PR tree-optimization/55273
18660         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18662 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
18664         PR target/55981
18665         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18666         store through atomic_store<mode>_1.
18667         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18669 2013-01-17  Martin Jambor  <mjambor@suse.cz>
18671         PR tree-optimizations/55264
18672         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18673         for virtual methods.
18674         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18675         virtual methods before inlining is over.
18676         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18677         virtual functions.
18678         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18679         non-virtual.
18681 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
18683         PR rtl-optimization/56005
18684         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18685         pending reads for prefetch.
18687 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
18689         * config/aarch64/aarch64.md
18690         (*cstoresi_neg_uxtw): New pattern.
18691         (*cmovsi_insn_uxtw): New pattern.
18692         (*<optab>si3_uxtw): New pattern.
18693         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18694         (*<optab>si3_insn_uxtw): New pattern.
18695         (*bswapsi2_uxtw): New pattern.
18697 2013-01-16  Richard Biener  <rguenther@suse.de>
18699         * tree-inline.c (tree_function_versioning): Remove set but
18700         never used variable.
18702 2013-01-16  Richard Biener  <rguenther@suse.de>
18704         PR tree-optimization/55964
18705         * tree-flow.h (rename_variables_in_loop): Remove.
18706         (rename_variables_in_bb): Likewise.
18707         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18708         (copy_loop_before): Adjust and delete update-ssa status.
18709         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18710         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
18711         (rename_variables_in_loop): Remove.
18712         (slpeel_update_phis_for_duplicate_loop): Likewise.
18713         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18714         use available cfg machinery instead of duplicating it.
18715         Update PHI nodes and perform poor-mans SSA update here.
18716         (slpeel_tree_peel_loop_to_edge): Adjust.
18718 2013-01-16  Richard Biener  <rguenther@suse.de>
18720         PR tree-optimization/54767
18721         PR tree-optimization/53465
18722         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18723         (vrp_visit_phi_node): For PHI arguments coming via backedges
18724         drop all symbolical range information.
18725         (execute_vrp): Compute backedges.
18727 2013-01-16  Richard Biener  <rguenther@suse.de>
18729         * doc/install.texi: Update CLooG and ISL requirements to
18730         0.18.0 and 0.11.1.
18732 2013-01-16  Christian Bruel  <christian.bruel@st.com>
18734         PR target/55301
18735         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18736         (broken_move): Handle UNSPECV_SP_SWITCH_B.
18737         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18739 2013-01-16  DJ Delorie  <dj@redhat.com>
18741         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18742         (UNSPECV_SP_SWITCH_E): New.
18743         (sp_switch_1): Change to an unspec.
18744         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18745         replace $r15.
18747 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18749         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18750         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18751         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18752         (expand_mem_thread_fence): Ditto.
18753         (expand_mem_signal_fence): Ditto.
18754         (expand_atomic_load): Ditto.
18755         (expand_atomic_store): Ditto.
18757 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18759         PR rtl-optimization/55547
18760         PR rtl-optimization/53827
18761         PR debug/53671
18762         PR debug/49888
18763         * alias.c (memrefs_conflict_p): Set sizes to negative after
18764         AND adjustments.
18766 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18768         PR target/55940
18769         * function.c (thread_prologue_and_epilogue_insns): Always
18770         add crtl->drap_reg to set_up_by_prologue.set, even if
18771         stack_realign_drap is false.
18773 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18775         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18776         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18777         *call): Fix indention.
18779 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18781         PR target/55876
18782         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18783         Update comment.
18785 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18787         PR rtl-optimization/55153
18788         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18790 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18792         PR tree-optimization/55920
18793         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18794         accesses as grp_to_be_debug_replaced.
18796 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18798         PR tree-optimization/55920
18799         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18800         there is non-useless type conversion needed from debug rhs to lhs,
18801         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18803 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18804             Mikael Pettersson  <mikpe@it.uu.se>
18806         PR target/43961
18807         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18808         Thumb.
18809         (ASM_OUTPUT_CASE_LABEL): Remove.
18810         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18811         * final.c (shorten_branches): Update alignment of labels before
18812         jump tables if CASE_VECTOR_SHORTEN_MODE.
18814 2013-01-15  Richard Biener  <rguenther@suse.de>
18816         PR bootstrap/55961
18817         * system.h: Do not include gmp.h for building host tools.
18819 2013-01-15  Richard Biener  <rguenther@suse.de>
18821         PR middle-end/55882
18822         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18823         account for bitpos when computing alignment.
18825 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18827         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18828         (ix86_target_macros_internal): Likewise.
18830         * config/i386/i386.c (m_CORE2I7): Removed.
18831         (m_CORE_HASWELL): New macro.
18832         (m_CORE_ALL): Likewise.
18833         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18834         (initial_ix86_arch_features): Likewise.
18835         (processor_target_table): Initializations for Core avx2.
18836         (cpu_names): New names "core-avx2".
18837         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18838         PROCESSOR_CORE_HASWELL.
18839         (ix86_issue_rate): New case.
18840         (ia32_multipass_dfa_lookahead): Likewise.
18841         (ix86_sched_init_global): Likewise.
18843         * config/i386/i386.h (TARGET_HASWELL): New macro.
18844         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18845         (processor_type): New PROCESSOR_HASWELL.
18847 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18849         PR tree-optimization/55955
18850         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18851         *SHIFT_EXPR and *ROTATE_EXPR codes.
18853         PR tree-optimization/48766
18854         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18855         -ftrapv disable -fwrapv.
18857 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18859         PR target/55974
18860         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18861         etc. to 1 and not to __flash.
18862         Use LL suffix for __INT24_MAX__ with -mint8.
18863         Use ULL suffix for __UINT24_MAX__ with -mint8.
18865 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18867         * config/avr/avr-arch.h
18868         (struct base_arch_s): Use typedef avr_arch_t instead.
18869         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18870         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18871         * config/avr/avr.c: Same.
18872         * config/avr/avr-devices.c: Same.
18873         * config/avr/driver-avr.c: Same.
18874         * config/avr/gen-avr-mmcu-texi.c: Same.
18875         * config/avr/avr-mcus.def: Adjust comment.
18877 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18879         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18880         * config/aarch64/iterators.md (VALLDI): New.
18882 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18883             Andi Kleen  <ak@linux.intel.com>
18885         PR target/55948
18886         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18887         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18888         memmodel flag.
18890 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18892         * config/avr/avr-stdint.h: Remove trailing blanks.
18893         * config/avr/avr-log.h: Same.
18894         * config/avr/avr-arch.h: Same.
18895         * config/avr/avr-devices.c: Same.
18896         * config/avr/avr-dimode.md: Same.
18897         * config/avr/predicates.md: Same.
18898         * config/avr/avr-c.c: Same.  And fix typo.
18900         * config/avr/avr-protos.h: Same.  And:
18901         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18902         (init_cumulative_args): Rename to avr_init_cumulative_args.
18903         (expand_prologue): Rename to avr_expand_prologue.
18904         (expand_epilogue): Rename to avr_expand_epilogue.
18905         (adjust_insn_length): Rename to avr_adjust_insn_length.
18906         (notice_update_cc): Rename to avr_notice_update_cc.
18907         (final_prescan_insn): Rename to avr_final_prescan_insn.
18908         * config/avr/avr.c: Same.
18909         * config/avr/avr.h: Same.
18910         * config/avr/avr.md: Remove trailing blanks.
18911         (prologue): Use avr_expand_prologue.
18912         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18914 2013-01-14  Richard Biener  <rguenther@suse.de>
18916         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18917         verify_location, collect_subblocks): New functions.
18918         (verify_gimple_in_cfg): Verify that locations only reference
18919         BLOCKs in the functions BLOCK tree.
18921 2013-01-14  Richard Biener  <rguenther@suse.de>
18923         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18924         PHI argument.
18925         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18926         unshare reference.
18927         (insert_out_of_ssa_copy_on_edge): Likewise.
18928         (rewrite_close_phi_out_of_ssa): Likewise.
18929         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18930         debug expressions.
18931         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18932         propagated constants.
18933         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18934         can not be shared.
18936 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18938         * config/avr/avr-modes.def: Add GPL copyright notice.
18940 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18942         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18943         MEMMODEL_MASK to determine memory model.
18944         (atomic_store<mode>): Ditto from operands[2].
18945         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18947 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18949         PR fortran/55935
18950         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18951         (fold_gimple_assign): Don't call unshare_expr here.
18952         (fold_ctor_reference): Call unshare_expr.
18954 2013-01-13  Terry Guo  <terry.guo@arm.com>
18956         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18957         * doc/fragments.texi: Document MULTILIB_REUSE.
18958         * gcc.c (multilib_reuse): New internal spec.
18959         (set_multilib_dir): Also search multilib from multilib_reuse.
18960         * genmultilib (tmpmultilib3): Refactor code.
18961         (tmpmultilib4): Ditto.
18962         (multilib_reuse): New multilib argument.
18964 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18966         * Makefile.in: Update copyright.
18968 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18970         PR middle-end/55890
18971         * calls.c (expand_call): Check if arg_nr is valid.
18973 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18975         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18976         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18977         documentation.  Add missing '__' in front of
18978         __builtin_ia32_packssdw256.
18980 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18982         PR target/55719
18983         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18984         NO_REGS for larl operands.
18985         (s390_reload_larl_operand): Use s390_load_address instead of
18986         emit_move_insn.
18988 2013-01-11  Richard Biener  <rguenther@suse.de>
18990         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18991         (verify_node_sharing): ... here.
18992         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18994 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18996         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18997         Substitute TREECHECKING.
18998         * configure: Regenerate.
18999         * Makefile.in (TREECHECKING): New.
19001 2013-01-11  Richard Guenther  <rguenther@suse.de>
19003         PR tree-optimization/44061
19004         * tree-vrp.c (extract_range_basic): Compute zero as
19005         value-range for __builtin_constant_p of function parameters.
19007 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
19009         Update copyright years.
19011 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
19013         PR rtl-optimization/55672
19014         * lra-eliminations.c (mark_not_eliminable): Permit addition with
19015         const to be eliminable.
19017 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
19019         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19020         * configure: Regenerate.
19022 2013-01-10  Richard Biener  <rguenther@suse.de>
19024         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19026 2013-01-10  Richard Biener  <rguenther@suse.de>
19028         PR bootstrap/55792
19029         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19030         locations for virtual PHI arguments.
19031         (rewrite_update_phi_arguments): Likewise.
19033 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
19035         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19036         on to assembler.
19038 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
19040         PR tree-optimization/55921
19041         * tree-complex.c (expand_complex_asm): New function.
19042         (expand_complex_operations_1): Call it for GIMPLE_ASM.
19044 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19046         PR target/55718
19047         * config/s390/s390.c (s390_symref_operand_p)
19048         (s390_loadrelative_operand_p): Merge the two functions.
19049         (s390_check_qrst_address, print_operand_address): Add parameters
19050         to s390_loadrelative_operand_p invokation.
19051         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19052         (s390_reload_larl_operand, s390_secondary_reload): Use
19053         s390_loadrelative_operand_p instead of s390_symref_operand_p.
19054         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19056 2013-01-09  Mike Stump  <mikestump@comcast.net>
19058         * dse.c (record_store): Remove unnecessary assert.
19060 2013-01-09  Jan Hubicka  <jh@suse.cz>
19062         PR tree-optimization/55569
19063         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19064         * cfgloop.h (scale_loop_profile): Likewise.
19066 2013-01-09  Jan Hubicka  <jh@suse.cz>
19068         PR lto/45375
19069         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19070         functions.
19071         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19073 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
19075         PR middle-end/55114
19076         * expr.h (maybe_emit_group_store): Declare.
19077         * expr.c (maybe_emit_group_store): New function.
19078         * builtins.c (expand_builtin_int_roundingfn): Call it.
19079         (expand_builtin_int_roundingfn_2): Likewise.
19081 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
19083         PR rtl-optimization/55829
19084         * lra-constraints.c (match_reload): Add code for absent output.
19085         (curr_insn_transform): Add code for reloads of matched inputs
19086         without output.
19088 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
19090         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19091         attribute of movddup insn to DF.
19092         (*vec_interleave_lowv2df): Ditto.
19093         (vec_dupv2df): Ditto.
19095 2013-01-09  Jan Hubicka  <jh@suse.cz>
19097         PR tree-optimiation/55875
19098         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19099         EVERY_ITERATION parameter.
19100         (number_of_iterations_exit): Check if exit is executed every iteration.
19101         (idx_infer_loop_bounds): Similarly here.
19102         (n_of_executions_at_most): Simplify
19103         to only test for cases where statement is dominated by the
19104         particular bound; handle correctly the "postdominance" test.
19105         (scev_probably_wraps_p): Use max loop iterations info
19106         as a global bound first.
19108 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
19109             Nick Clifton  <nickc@redhat.com>
19111         * config/v850/v850.md (cbranchsf4): New pattern.
19112         (cstoresf4): New pattern.
19113         (cbranchdf4): New pattern.
19114         (cstoredf4): New pattern.
19115         (movsicc): Disallow floating point comparisons.
19116         (cmpsf_le_insn): Fix order of operators.
19117         (cmpsf_lt_insn): Likewise.
19118         (cmpsf_eq_insn): Likewise.
19119         (cmpdf_le_insn): Likewise.
19120         (cmpdf_lt_insn): Likewise.
19121         (cmpdf_eq_insn): Likewise.
19122         (cmpsf_ge_insn): Use LE comparison.
19123         (cmpdf_ge_insn): Likewise.
19124         (cmpsf_gt_insn): Use LT comparison.
19125         (cmpdf_gt_insn): Likewise.
19126         (cmpsf_ne_insn): Delete pattern.
19127         (cmpdf_ne_insn): Delete pattern.
19128         * config/v850/v850.c (v850_gen_float_compare): Use
19129         gen_cmpdf_eq_insn for NE comparison.
19130         (v850_float_z_comparison_operator)
19131         (v850_float_nz_comparison_operator): Move from here ...
19132         * config/v850/predicates.md: ... to here.  Move GT and GE
19133         comparisons into v850_float_z_comparison_operator.
19134         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19135         Delete prototype.
19136         (v850_float_nz_comparison_operator): Likewise.
19138 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19140         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19141         with calls to gen_insvsi/gen_insvdi.
19143 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19145         * config/i386/i386.c (initial_ix86_tune_features): Set up
19146         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19148 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
19149             Jakub Jelinek  <jakub@redhat.com>
19151         PR tree-optimization/48189
19152         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19153         If nitercst is 0, don't predict the exit edge.
19155 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19157         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19158         in asm_fprintf with reg_names.
19159         (aarch64_print_operand_address): Likewise.
19160         (aarch64_return_addr): Likewise.
19161         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19163 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19165         * config/pa/pa.h (VAL_U6_BITS_P): Define.
19166         (INT_U6_BITS): Likewise.
19167         * config/pa/predicates.md (uint6_operand): New predicate.
19168         (shift5_operand, shift6_operand): Likewise.
19169         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19170         arith32_operand.
19171         (lshrdi3): Use shift6_operand.
19172         (shrpsi4, shrpdi4): New insn patterns.
19173         (extzv): Delete expander.
19174         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
19175         predicates in unamed zero extract patterns.  Tighten common constraint.
19176         (extv): Delete expander.
19177         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
19178         predicates in unamed sign extract patterns.  Tighten common constraint.
19179         (insv): Delete expander.
19180         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
19181         predicates in unamed insert patterns.  Tighten common constraint.
19182         Change uint32_operand predicate to uint6_operand predicate in unamed
19183         DImode pattern to insert constant values of type 1...1xxxx.
19185 2013-01-04  Jan Hubicka  <jh@suse.cz>
19187         PR tree-optimization/55823
19188         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19189         issue.
19191 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19192             Uros Bizjak  <ubizjak@gmail.com>
19194         PR rtl-optimization/55845
19195         * df-problems.c (can_move_insns_across): Stop scanning at
19196         volatile_insn_p source instruction or give up if
19197         across_from .. across_to range contains any volatile_insn_p
19198         instructions.
19200 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19202         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19203         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19204         Declare.
19205         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19206         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19208 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19210         PR fortran/55341
19211         * asan.c (asan_clear_shadow): New function.
19212         (asan_emit_stack_protection): Use it.
19214 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19216         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19217         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19218         with tab instead of space.
19220 2013-01-08  Nick Clifton  <nickc@redhat.com>
19222         * config/rl78/rl78.c (rl78_expand_prologue): Always select
19223         register bank 0 at the start of an interrupt handler.
19224         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19225         MDBH registers.
19227 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19229         * config/aarch64/aarch64-simd.md
19230         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19231         (aarch64_simd_bsl): Likewise.
19232         (aarch64_vcond_internal<mode>): Likewise.
19233         (vcond<mode><mode>): Likewise.
19234         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19235         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19237 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19239         * config/aarch64/aarch64-builtins.c
19240         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19242 2013-01-08  Martin Jambor  <mjambor@suse.cz>
19244         PR debug/55579
19245         * tree-sra.c (analyze_access_subtree): Return true also after
19246         potentially creating a debug-only replacement.
19248 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19250         PR middle-end/55890
19251         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19253         PR tree-optimization/54120
19254         * tree-vrp.c (range_fits_type_p): Don't allow
19255         src_precision < precision from signed vr to unsigned_p
19256         if vr->min or vr->max is negative.
19257         (simplify_float_conversion_using_ranges): Test can_float_p
19258         against CODE_FOR_nothing.
19260 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19261             Richard Biener  <rguenther@suse.de>
19263         PR middle-end/55851
19264         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19265         types instead of just INTEGER_TYPE types.
19267 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
19269         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19270         TF_SIZE): Define.
19272 2013-01-07  Steve Ellcey  <sellcey@mips.com>
19274         PR target/42661
19275         * config/mips/mips.opt: Change mad to mmad to match documentation.
19277 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19279         PR target/55897
19280         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19281         .progmemx.data now.
19283 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19285         PR target/55897
19286         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19287         (avr_addrspace_t): Add .section_name field.
19288         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19289         array size.
19290         (avr_addrspace): Same.  Initialize .section_name.  Remove last
19291         NULL entry.  Put __memx into .progmemx.data.
19292         (progmem_section_prefix): Remove.
19293         (avr_asm_init_sections): No need to initialize progmem_section.
19294         (avr_asm_named_section): Use avr_addrspace[].section_name to get
19295         section name prefix.
19296         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
19297         retrieve the progmem section.
19298         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19299         boundary to run over avr_addrspace[].
19300         (avr_register_target_pragmas): Ditto.
19302 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
19304         * varasm.c (output_constant_def_contents): For asan_protect_global
19305         protected strings, adjust DECL_ALIGN if needed, before testing for
19306         anchored symbols.
19307         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19308         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
19309         normal decls.
19310         (output_object_block): For asan protected decls, emit asan padding
19311         after their contents.
19312         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19313         (asan_finish_file): Test it here instead.
19315 2013-01-07  Nick Clifton  <nickc@redhat.com>
19316             Matthias Klose  <doko@debian.org>
19317             Doug Kwan  <dougkwan@google.com>
19318             H.J. Lu  <hongjiu.lu@intel.com>
19320         PR driver/55470
19321         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19323         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19325         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19327         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19329         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19331 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19333         PR target/54461
19334         * doc/install.texi (Cross-Compiler-Specific Options): Document
19335         --with-avrlibc.
19337 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
19339         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19340         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19341         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19342         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19343         vqmovun_high_s64): Fix source operand number and update copyright.
19345 2013-01-07  Richard Biener  <rguenther@suse.de>
19347         PR middle-end/55890
19348         * gimple.h (gimple_call_builtin_p): New overload.
19349         * gimple.c (validate_call): New function.
19350         (gimple_call_builtin_p): Likewise.
19351         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19352         Use gimple_call_builtin_p.
19353         (find_func_clobbers): Likewise.
19354         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19355         (strlen_optimize_stmt): Likewise.
19357 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19359         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19360         (vld1q_dup_*): Likewise.
19361         (vld1_*): Likewise.
19362         (vld1q_*): Likewise.
19363         (vld1_lane_*): Likewise.
19364         (vld1q_lane_*): Likewise.
19366 2013-01-07  Richard Biener  <rguenther@suse.de>
19368         * lto-streamer.h (LTO_minor_version): Bump to 2.
19370 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19372         * config/aarch64/aarch64-protos.h
19373         (aarch64_const_double_zero_rtx_p): Rename to...
19374         (aarch64_float_const_zero_rtx_p): ...this.
19375         (aarch64_float_const_representable_p): New.
19376         (aarch64_output_simd_mov_immediate): Likewise.
19377         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19378         move immediate case.
19379         * config/aarch64/aarch64.c
19380         (aarch64_const_double_zero_rtx_p): Rename to...
19381         (aarch64_float_const_zero_rtx_p): ...this.
19382         (aarch64_print_operand): Allow printing of new constants.
19383         (aarch64_valid_floating_const): New.
19384         (aarch64_legitimate_constant_p): Check for valid floating-point
19385         constants.
19386         (aarch64_simd_valid_immediate): Likewise.
19387         (aarch64_vect_float_const_representable_p): New.
19388         (aarch64_float_const_representable_p): Likewise.
19389         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19390         (aarch64_output_simd_mov_immediate): New.
19391         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19392         (*movdf_aarch64): Likewise.
19393         * config/aarch64/constraints.md (Ufc): New.
19394         (Y): call aarch64_float_const_zero_rtx.
19395         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19397 2013-01-07  Richard Biener  <rguenther@suse.de>
19399         PR tree-optimization/55888
19400         PR tree-optimization/55862
19401         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19402         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19403         not if it is contained therein.
19405 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19407         * config/avr/t-avr: Typo.
19409 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19411         PR55243
19412         * config/avr/t-avr: Don't automatically rebuild
19413         $(srcdir)/config/avr/t-multilib
19414         $(srcdir)/config/avr/avr-tables.opt
19415         $(srcdir)/doc/avr-mmcu.texi
19416         (avr-mcus): New phony target to build them on request.
19417         (s-avr-mlib, s-avr-mmcu-texi): Remove.
19418         * avr/avr-mcus.def: Adjust comments.
19420 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
19422         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19424 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19426         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19428 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19430         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19432 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
19434         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19435         to generate profiling.
19436         * config/rs6000/aix64.h (LIB_SPEC): Same.
19438 2013-01-04  Andrew Pinski  <apinski@cavium.com>
19440         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19441         New function.
19442         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19444 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
19446         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19447         unconditionally.
19448         (ix86_expand_move): Ditto.
19449         (ix86_zero_extend_to_Pmode): Ditto.
19450         (ix86_expand_call): Ditto.
19451         (ix86_expand_special_args_builtin): Ditto.
19452         (ix86_expand_builtin): Ditto.
19454 2013-01-04  Richard Biener  <rguenther@suse.de>
19456         PR tree-optimization/55862
19457         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19458         translating them through PHI nodes.
19460 2013-01-04  Martin Jambor  <mjambor@suse.cz>
19462         PR tree-optimization/55755
19463         * tree-sra.c (sra_modify_assign): Do not check that an access has no
19464         children when trying to avoid producing a VIEW_CONVERT_EXPR.
19466 2013-01-04  Marek Polacek  <polacek@redhat.com>
19468         PR middle-end/55859
19469         * opts.c (default_options_optimization): Clarify error message.
19471 2013-01-04  Richard Biener  <rguenther@suse.de>
19473         PR middle-end/55863
19474         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19475         reassociation.
19477 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19479         PR target/53789
19480         * config/pa/pa.md (movsi): Revert previous change.
19481         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19482         references.
19484 2013-01-03  Richard Henderson  <rth@redhat.com>
19486         * config/i386/i386.c (ix86_expand_move): Always assign to op1
19487         after eliminating TLS symbols.
19489 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
19491         PR bootstrap/50167
19492         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19493         * graphite-poly.c (debug_gmp_value): Likewise.
19495 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
19497         PR target/55712
19498         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19499         selected code model, define __code_mode_small__, __code_model_medium__,
19500         __code_model_large__, __code_model_32__ or __code_model_kernel__.
19501         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19502         xchg temporary register with %k.  Declare temporary register as
19503         early clobbered.
19504         [__x86_64__]: For medium and large code models, preserve %rbx register.
19506 2013-01-03  Richard Biener  <rguenther@suse.de>
19508         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19509         (dump_subscript): Adjust.
19510         (finalize_ddr_dependent): Do not dump redundant info.
19511         (analyze_siv_subscript): Adjust.
19512         (subscript_dependence_tester): Likewise.
19513         (compute_affine_dependence): Likewise.
19515 2013-01-03  Richard Biener  <rguenther@suse.de>
19517         Revert
19518         2013-01-03  Richard Biener  <rguenther@suse.de>
19520         PR tree-optimization/55857
19521         * tree-vect-stmts.c (vectorizable_load): Do not setup
19522         re-alignment for invariant loads.
19524         2013-01-02  Richard Biener  <rguenther@suse.de>
19526         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19527         invariant load do not generate a vector load from the scalar location.
19529 2013-01-03  Richard Biener  <rguenther@suse.de>
19531         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19532         for not vectorizing.
19533         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19534         not build INDIRECT_REFs, call get_name once only.
19535         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
19536         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19538 2013-01-03  Richard Biener  <rguenther@suse.de>
19540         PR tree-optimization/55857
19541         * tree-vect-stmts.c (vectorizable_load): Do not setup
19542         re-alignment for invariant loads.
19544 2013-01-03  Richard Biener  <rguenther@suse.de>
19546         PR lto/55848
19547         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19548         prefer a built-in decl.
19550 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19552         * gcc.c (process_command): Update copyright notice dates.
19553         * gcov.c (print_version): Likewise.
19554         * gcov-dump.c (print_version): Likewise.
19556         PR rtl-optimization/55838
19557         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19558         iv0.step, iv1.step and step.
19560 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19561             Marc Glisse  <marc.glisse@inria.fr>
19563         PR tree-optimization/55832
19564         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19565         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19566         integer_{one,zero}_node.
19568 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19570         PR debug/54402
19571         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19572         * var-tracking.c (reverse_op): Don't add reverse ops to
19573         VALUEs that have already
19574         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19576 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
19578         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19580 2013-01-02  Teresa Johnson  <tejohnson@google.com>
19582         * dumpfile.c (dump_loc): Print filename with location.
19583         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19584         new location_t parameter to emit complete unroll message with
19585         new dump framework.
19586         (canonicalize_loop_induction_variables): Compute loops location
19587         and pass to try_unroll_loop_completely.
19588         * loop-unroll.c (report_unroll_peel): New function.
19589         (peel_loops_completely): Use new dump format with location
19590         for main dumpfile message, and invoke report_unroll_peel on success.
19591         (decide_unrolling_and_peeling): Ditto.
19592         (decide_peel_once_rolling): Remove old dumpfile message subsumed
19593         by report_unroll_peel.
19594         (decide_peel_completely): Ditto.
19595         (decide_unroll_constant_iterations): Ditto.
19596         (decide_unroll_runtime_iterations): Ditto.
19597         (decide_peel_simple): Ditto.
19598         (decide_unroll_stupid): Ditto.
19599         * cfgloop.c (get_loop_location): New function.
19600         * cfgloop.h (get_loop_location): Declare.
19602 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19604         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19605         NULL.
19607 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19609         PR middle-end/55198
19610         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19611         BLKmode objects when EXPAND_MEMORY is specified.
19613 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19615         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19616         in loop predicate.
19617         (fold_builtin_cpu): Do not share cpu model decls across statements.
19619 2013-01-02  Jason Merrill  <jason@redhat.com>
19621         PR c++/55804
19622         * tree.c (build_array_type_1): Revert earlier change.
19624 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
19626         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19627         "cortex-a57".
19628         * config/aarch64/aarch64-tune.md: Re-generate.
19630 2013-01-02  Richard Biener  <rguenther@suse.de>
19632         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19633         invariant load do not generate a vector load from the scalar location.
19635 2013-01-02  Richard Biener  <rguenther@suse.de>
19637         PR bootstrap/55784
19638         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19639         * configure: Regenerate.
19641 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19643         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19644         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19645         (expand_builtin_int_roundingfn_2): Keep the original target around
19646         for the fallback case.
19648 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19650         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19651         to be clear for sign changes.
19653 2013-01-01  Jan Hubicka  <jh@suse.cz>
19655         * ipa-inline-analysis.c: Fix formatting.
19657 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
19659         PR tree-optimization/55831
19660         * tree-vect-loop.c (get_initial_def_for_induction): Use
19661         gsi_after_labels instead of gsi_start_bb.
19663 Copyright (C) 2013 Free Software Foundation, Inc.
19665 Copying and distribution of this file, with or without modification,
19666 are permitted in any medium without royalty provided the copyright
19667 notice and this notice are preserved.