2013-09-04 Teresa Johnson <tejohnson@google.com>
[official-gcc.git] / gcc / ChangeLog
blobabf8182b09def41503471029071f1cf56e0d342f
1 2013-09-04  Teresa Johnson  <tejohnson@google.com>
3         * dumpfile.c (dump_finish): Don't close stderr/stdout.
5 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
9 2013-09-04  Jan Hubicka  <jh@suse.cz>
11         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
12         * ipa-devirt.c: Include diganostic.h
13         (odr_type_d): Add types and types_set.
14         (hash_type_name): Work for types with vtables during LTO.
15         (odr_hasher::remove): Fix comment; destroy types_set.
16         (add_type_duplicate): New function,
17         (get_odr_type): Use it.
18         (dump_type_inheritance_graph): Dump type duplicates.
19         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
20         graph.
21         * tree.c (types_same_for_odr): Give exact answers on types with
22         virtual tables.
24 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
26         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
27         explaining their differences.
29 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
31         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
33 2013-09-03  Jeff Law  <law@redhat.com>
35         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
36         when not threading through a joiner block.  Pass joiner/no joiner
37         state to register_jump_thread.
38         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
39         state from argument rather than implying on path length.
40         Dump the entire jump thread path into debugging dump.
41         * tree-flow.h (register_jump_thread): Update prototype.
43 2013-08-29  Xinliang David Li  <davidxl@google.com>
45         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
46         Remove a trivial gcc_assert.
48 2013-08-29  Xinliang David Li  <davidxl@google.com>
50         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
51         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
52         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
53         Delay base decl alignment adjustment.
54         * tree-vectorizer.c (vect_destroy_datarefs): New function.
55         * tree-vectorizer.h: New data structure.
56         (set_dr_misalignment): New function.
57         (dr_misalignment): Ditto.
58         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
59         (vectorizable_load): Ditto.
60         (ensure_base_align): New function.
61         (vectorize_loops): Add dbg_cnt support.
62         (execute_vect_slp): Ditto.
63         * dbgcnt.def: New debug counter.
64         * Makefile: New dependency.
66 2013-09-03  Meador Inge  <meadori@codesourcery.com>
68         Revert:
70         2013-08-30  Meador Inge  <meadori@codesourcery.com>
72         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
74 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
76         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
77         function descriptor.
79 2013-09-03  Richard Biener  <rguenther@suse.de>
81         * tree-affine.c (add_elt_to_tree): Fix association issue,
82         avoid useless converts and make sure to always return a
83         properly typed result.
85 2013-09-03  Richard Biener  <rguenther@suse.de>
87         PR middle-end/57656
88         * fold-const.c (negate_expr_p): Fix division case.
89         (negate_expr): Likewise.
91 2013-09-03  Richard Biener  <rguenther@suse.de>
93         PR lto/58285
94         * tree-streamer-out.c: Include tm.h.
95         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
97 2013-09-03  Jan Hubicka  <jh@suse.cz>
99         * tree-profile.c (tree_profiling): Cleanup CFG when done.
101 2013-09-03  Alan Modra  <amodra@gmail.com>
103         * config.gcc (powerpc*-*-linux*): Add support for little-endian
104         multilibs to big-endian target and vice versa.
105         * config/rs6000/t-linux64: Use := assignment on all vars.
106         (MULTILIB_EXTRA_OPTS): Remove fPIC.
107         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
108         * config/rs6000/t-linux64le: New file.
109         * config/rs6000/t-linux64bele: New file.
110         * config/rs6000/t-linux64lebe: New file.
112 2013-09-02  Jan Hubicka  <jh@suse.cz>
114         * ipa-inline-transform.c (inline_transform): Do not
115         optimize_inline_calls when not optimizing.
117 2013-09-02  Jan Hubicka  <jh@suse.cz>
119         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
120         duplicated nodes for assembler names.
121         * symtab.c (symtab_unregister_node): Do not attempt to unlink
122         hard registers from assembler name hash.
124 2013-09-02  Jan Hubicka  <jh@suse.cz>
126         * ipa-split.c (execute_split_functions): Split externally visible
127         functions called once.
129 2013-09-02  Martin Jambor  <mjambor@suse.cz>
131         PR ipa/58106
132         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
133         linked list.  When finding the correct duplicate, also consider also
134         the caller in additon to its inlined_to node.
136 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
138         * config/aarch64/aarch64-simd-builtins.def
139         (dup_lane_scalar): Remove.
140         * config/aarch64/aarch64-simd.md
141         (aarch64_simd_dup): Add 'w->w' alternative.
142         (aarch64_dup_lane<mode>): Allow for VALL.
143         (aarch64_dup_lane_scalar<mode>): Remove.
144         (aarch64_dup_lane_<vswap_width_name><mode>): New.
145         (aarch64_get_lane_signed<mode>): Add w->w altenative.
146         (aarch64_get_lane_unsigned<mode>): Likewise.
147         (aarch64_get_lane<mode>): Likewise.
148         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
149         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
150         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
151         (VCON): Change container of V2SF.
152         (vswap_width_name): Likewise.
153         * config/aarch64/arm_neon.h
154         (__aarch64_vdup_lane_any): New.
155         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
156         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
157         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
159 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
161         PR middle-end/56382
162         * expr.c (emit_move_complex): Do not move complex FP values as parts if
163         the source or the destination is a single hard register.
165 2013-09-02  Richard Biener  <rguenther@suse.de>
167         PR middle-end/57511
168         * tree-scalar-evolution.c (instantiate_scev_name): Allow
169         non-linear SCEVs.
171 2013-09-02  Richard Biener  <rguenther@suse.de>
173         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
174         arithmetic to sizetype.
176 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
178         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
179         Find auto-increment use both before and after candidate.
181 2013-09-02  Marek Polacek  <polacek@redhat.com>
183         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
185 2013-09-01  Jan Hubicka  <jh@suse.cz>
187         * Makefile.in: Add ipa-profile.o
188         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
189         * cgraph.c (struct cgraph_propagate_frequency_data,
190         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
191         ipa-profile.c; replace cgraph_ by ipa_ prefix.
192         * cgraph.h (cgraph_propagate_frequency): Remove.
193         * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
194         (inline_update_callee_summaries): Update.
195         * ipa-profile.c: New file.
196         * ipa-utils.h (ipa_propagate_frequency): Declare.
197         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
198         data-streamer.h, value-prof.h
199         (symtab_remove_unreachable_nodes): Update profile.
200         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
201         account_time_size, cmp_counts, dump_histogram,
202         ipa_profile_generate_summary, ipa_profile_write_summary,
203         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
204         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
205         Move to ipa-profile.c
207 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
209         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
211 2013-09-01  Jan Hubicka  <jh@suse.cz>
213         * common.opt (fdevirtualize-speculatively): New function.
214         * invoke.texi (fdevirtualize-speculatively): Document.
215         * ipa-devirt.c: Include ipa-inline.h
216         (likely_target_p): New function.
217         (ipa_devirt): New function.
218         (gate_ipa_devirt): New function.
219         (pass_data_ipa_devirt): New static var.
220         (pass_ipa_devirt): Likewise.
221         (make_pass_ipa_devirt): New function.
222         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
223         (common_handle_option): Disable devirtualization when
224         value range profiling is available.
225         * passes.def (pass_ipa_devirt): Add.
226         * timever.def (TV_IPA_DEVIRT): New timevar.
227         * tree-pass.h (make_pass_ipa_devirt): 
229 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
231         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
232         include sanitize(undefined).
234 2013-08-31  Diego Novillo  <dnovillo@google.com>
236         * Makefile.in (TREE_CORE_H): Define.
237         (TREE_H): Use.
238         (GTFILES): Add tree-core.h.
239         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
240         size the array.
241         * tree-core.h: New file.
242         Move all data structures, enum, typedefs, global
243         declarations and constants from ...
244         * tree.h: ... here.
246 2013-08-31  Jan Hubicka  <jh@suse.cz>
248         * bulitins.c (expand_builtin): Do not early exit for gcov
249         instrumented functions.
251 2013-08-31  Marek Polacek  <polacek@redhat.com>
253         * ubsan.c: Include tm_p.h.
255 2013-08-31  Jan Hubicka  <jh@suse.cz>
257         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
258         warning.
260         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
261         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
262         * tree-cfg.c (verify_gimple_label): ... here.
263         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
264         (ipa_merge_profiles): New function.
265         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
266         (lto_input_function_body): Likewise.
267         * ipa-utils.h (ipa_merge_profiles): Declare.
268         * lto-streamer.h (lto_input_function_body): Update prototype.
269         (emit_label_in_global_context_p): Remove.
270         * lto-symtab.c: Include ipa-utils.h
271         (lto_cgraph_replace_node): Use ipa_merge_profiles.
273 2013-08-31  Jan Hubicka  <jh@suse.cz>
275         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
277 2013-08-31  Jan Hubicka  <jh@suse.cz>
279         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
281 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
283         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
284         "cmp" RTX before signed_comparison_operator check to account
285         for "code" changes.
287 2013-08-30  Jan Hubicka  <jh@suse.cz>
289         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
290         (detect_type_change_1): Rename to ...
291         (detect_type_change): ... this one; early return on non-polymorphic
292         types.
293         (detect_type_change_ssa): Add comp_type parameter; update
294         use of detect_type_change.
295         (compute_complex_assign_jump_func): Add param_type parameter;
296         update use of detect_type_change_ssa.
297         (compute_complex_ancestor_jump_func): Likewise.
298         (ipa_get_callee_param_type): New function.
299         (ipa_compute_jump_functions_for_edge): Compute parameter type;
300         update calls to the jump function computation functions.
302 2013-08-30  Teresa Johnson  <tejohnson@google.com>
303             Steven Bosscher  <steven@gcc.gnu.org>
305         * cfgrtl.c (fixup_new_cold_bb): New routine.
306         (commit_edge_insertions): Invoke fixup_partitions.
307         (find_partition_fixes): New routine.
308         (fixup_partitions): Ditto.
309         (verify_hot_cold_block_grouping): Update comments.
310         (rtl_verify_edges): Invoke find_partition_fixes.
311         (rtl_verify_bb_pointers): Update comments.
312         (rtl_verify_bb_layout): Ditto.
313         * basic-block.h (probably_never_executed_edge_p): Declare.
314         (fixup_partitions): Ditto.
315         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
316         * bb-reorder.c (sanitize_hot_paths): New function.
317         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
318         sanitize_hot_paths.
319         * predict.c (probably_never_executed_edge_p): New routine.
320         * cfg.c (check_bb_profile): Add partition insanity warnings.
322 2013-08-30  Meador Inge  <meadori@codesourcery.com>
324         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
326 2013-08-30  Marek Polacek  <polacek@redhat.com>
328         * Makefile.in (ubsan.o): Add.
329         (c-family/c-ubsan.o): Add.
330         (builtins.o): Add ubsan.h dependency.
331         * ubsan.h: New file.
332         * ubsan.c: New file.
333         * common.opt: Add -fsanitize=undefined option.
334         (flag_sanitize): Add variable.
335         (fsanitize=): Add option.  Add Driver.
336         (fsanitize=thread): Remove option.
337         (fsanitize=address): Likewise.
338         (static-libubsan): New option.
339         * doc/invoke.texi: Document the new flag and -static-libubsan.
340         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
341         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
342         * builtin-attrs.def (ATTR_COLD): Define.
343         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
344         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
345         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
346         * flag-types.h (sanitize_code): New enum.
347         * opts.c (common_handle_option): Parse command line arguments
348         of -fsanitize=.  Add -fsanitize=unreachable option.
349         * varasm.c (get_variable_section): Adjust.
350         (assemble_noswitch_variable): Likewise.
351         (assemble_variable): Likewise.
352         (output_constant_def_contents): Likewise.
353         (categorize_decl_for_section): Likewise.
354         (place_block_symbol): Likewise.
355         (output_object_block): Likewise.
356         * builtins.def: Likewise.
357         * toplev.c (compile_file): Likewise.
358         (process_options): Likewise.
359         * cppbuiltin.c: Likewise.
360         * tsan.c (tsan_pass): Likewise.
361         (tsan_gate): Likewise.
362         (tsan_gate_O0): Likewise.
363         * cfgexpand.c (partition_stack_vars): Likewise.
364         (expand_stack_vars): Likewise.
365         (defer_stack_allocation): Likewise.
366         (expand_used_vars): Likewise.
367         * cfgcleanup.c (old_insns_match_p): Likewise.
368         * asan.c (asan_finish_file): Likewise.
369         (asan_instrument): Likewise.
370         (gate_asan): Likewise.
371         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
372         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
373         (asan_global_struct): Use pointer_sized_int_node instead
374         calling build_nonstandard_integer_type.
375         (initialize_sanitizer_builtins): Likewise.
376         (asan_finish_file): Likewise.
377         * gcc.c: Document %{%:function(args):X}.
378         (static_spec_functions): Add sanitize.
379         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
380         store funcval != NULL there.
381         (do_spec_1): Adjust handle_spec_function caller.
382         (handle_braces): Allow %:function(args) as condition.
383         (sanitize_spec_function): New function.
384         (ADD_STATIC_LIBUBSAN_LIBS): Define.
385         (LIBUBSAN_SPEC): Likewise.
386         (LIBUBSAN_EARLY_SPEC): Likewise.
387         (SANITIZER_SPEC): Handle libubsan.
388         (SANITIZER_EARLY_SPEC): Likewise.
389         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
390         instead of fsanitize=address.
391         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
392         instead of fsanitize=address*.
393         * builtins.c: Include ubsan.h.
394         (fold_builtin_0): Instrument __builtin_unreachable.
395         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
396         instead of flag_asan.
397         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
398         (pointer_sized_int_node): Define.
399         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
401 2013-08-30  Mike Stump  <mikestump@comcast.net>
403         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
404         with RE patterns.
406 2013-08-29  Jan Hubicka  <jh@suse.cz>
408         * cgraph.c (cgraph_function_body_availability): Handle weakref
409         correctly.
410         * passes.def: Remove pass_fixup_cfg.
411         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
412         track when we need to remove functions.
413         (gate_ipa_inline): Execute inlining always; add comment why.
414         (pass_data_ipa_inline): Remove TODO_remove_functions.
415         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
416         do not produce summaries.
417         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
418         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
419         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
420         constant pool and vtable.
422 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
424         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
425         New arm_neon.h's internal macros to specify 64-bit constants.
426         Avoid using stdint.h's macros.
428 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
430         * recog.c (verify_changes): Verify that changes[i].old is non-zero
431         before applying REG_P.
433 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
435         PR tree-optimization/58277
436         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
437         after seeing too many stmts with vdef in between dombb and current
438         bb, invalidate everything.
440 2013-08-30  Richard Biener  <rguenther@suse.de>
442         * fold-const.c (fold_single_bit_test): Fix overflow test.
444 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
446         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
447         and which can live in a register, always retrieve the value on entry.
448         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
449         passed by invisible reference specially.
450         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
451         (vt_add_function_parameter): Correctly deal with a parameter passed by
452         invisible reference.
454 2013-08-30  Jan Hubicka  <jh@suse.cz>
456         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
458 2013-08-30  Richard Biener  <rguenther@suse.de>
460         PR tree-optimization/58228
461         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
462         allow invariant loads in nested loop vectorization.
464 2013-08-30  Richard Biener  <rguenther@suse.de>
466         PR tree-optimization/58223
467         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
468         (has_anti_or_output_dependence): ... this and adjust to also
469         look for output dependences.
470         (mark_nodes_having_upstream_mem_writes): Adjust.
471         (rdg_flag_uses): Likewise.
473 2013-08-30  Richard Biener  <rguenther@suse.de>
475         PR tree-optimization/58010
476         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
477         assert that we have a loop-closed PHI.
479 2013-08-29  Jan Hubicka  <jh@suse.cz>
481         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
482         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
483         * lto-section-in.c (lto_free_function_in_decl_state): New function.
484         (lto_free_function_in_decl_state_for_node): New function.
486 2013-08-29  Xinliang David Li  <davidxl@google.com>
488         * loop-unroll.c (report_unroll_peel): Minor message change.
489         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
490         Emit alignment peeling message with default -fopt-info.
491         (vect_loop_versioning): Emit loop version info message.
492         * tree-vectorizer.c (vectorize_loops): Minor message change.
493         (execute_vect_slp): Ditto.
495 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
497         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
498         of the clone from the DECL_NAME of the original function.
500 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
502         * passes.c (register_pass): Add overload.
503         * tree-pass.h (register_pass): Forward declare it.  Add comment.
505 2013-08-29  Jan Hubicka  <jh@suse.cz>
507         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
508         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
509         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
510         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
511         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
512         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
513         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
514         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
516 2013-08-29  Teresa Johnson  <tejohnson@google.com>
518         * dumpfile.c (dump_loc): Output column number.
519         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
520         * doc/invoke.texi: Document optall -fopt-info flag.
521         * profile.c (read_profile_edge_counts): Use new dump framework.
522         (compute_branch_probabilities): Ditto.
523         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
524         when pass not in any opt group.
525         * pass_manager.h (pass_manager::get_pass_profile): New method.
526         * value-prof.c (check_counter): Use new dump framework.
527         (check_ic_target): Ditto.
528         * coverage.c (get_coverage_counts): Ditto.
529         (coverage_init): Setup new dump framework.
531 2013-08-29  Richard Biener  <rguenther@suse.de>
533         PR tree-optimization/58246
534         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
535         handle the dominance check inside a basic-block.
537 2013-08-29  Richard Biener  <rguenther@suse.de>
539         PR middle-end/57287
540         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
541         of default defs that appear in abnormal PHI nodes.
543 2013-08-29  Richard Biener  <rguenther@suse.de>
545         PR tree-optimization/57685
546         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
547         single-use operands to avoid exponential complexity.
549 2013-08-28  Dehao Chen  <dehao@google.com>
551         * ipa-inline.c (edge_badness): Fix integer underflow.
553 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
555         * gtm-builtins.def (_ITM_free): Declare leaf.
557 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
559         PR target/58067
560         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
561         (*tls_local_dynamic_base_64_largepic): Likewise.
562         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
563         Remove predicate from call operand.
564         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
565         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
567 2013-08-28  Jeff Law  <law@redhat.com>
569         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
570         checks for the number of predecessors and successors allowed.
571         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
572         which require copying a joiner block if there is a request which
573         is a subpath that requires no joiner block copying.
575 2013-08-28  Jan Hubicka  <jh@suse.cz>
577         * lto-streamer-out.c (DFS_write_tree_body): Drop
578         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
579         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
580         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
581         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
582         Do not read DECL_ERROR_ISSUED.
583         (unpack_ts_decl_with_vis_value_fields): Do not read
584         DECL_DEFER_OUTPUT.
585         (lto_input_ts_binfo_tree_pointers): Do not read
586         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
587         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
588         write DECL_ERROR_ISSUED..
589         (pack_ts_decl_with_vis_value_fields): Do not write
590         DECL_DEFER_OUTPUT.
591         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
592         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
593         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
594         * tree.h (tree_decl_common): Update comment.
595         (DECL_ERROR_ISSUED): Remove.
597 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
599         PR middle-end/58257
600         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
602 2013-08-28  Jan Hubicka  <jh@suse.cz>
604         * builtins.def (free): Declare leaf.
606 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
608         * gdbhooks.py: New.
609         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
610         * configure: Regenerate.
612 2013-08-27  Martin Jambor  <mjambor@suse.cz>
614         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
615         (ipa_ancestor_jf_data): Likewise.
616         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
617         (ipa_get_jf_pass_through_type_preserved): New function.
618         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
619         (ipa_get_jf_ancestor_type_preserved): New function.
620         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
621         (ipa_get_jf_ancestor_result): Likewise.
622         (propagate_vals_accross_pass_through): Use
623         ipa_get_jf_pass_through_result to do all the value mappings.
624         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
625         type_preserved flag.
626         (ipa_set_jf_cst_copy): New function.
627         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
628         (ipa_set_jf_arith_pass_through): Likewise.
629         (ipa_set_ancestor_jf): Likewise.
630         (compute_complex_assign_jump_func): Set type_preserved instead of
631         punting.
632         (ipa_compute_jump_functions_for_edge): Likewise.
633         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
634         (update_jump_functions_after_inlining): Update type_preserved.
635         Explicitely create jump functions when combining one with pass_through.
636         (ipa_write_jump_function): Stream the type_preserved flags.
637         (ipa_read_jump_function): Likewise.
639 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
640             Aldy Hernandez  <aldyh@redhat.com>
642         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
643         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
644         * function.h (struct function): Add has_force_vect_loops and
645         has_simduid_loops.
646         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
647         * gimple.c (gimple_build_omp_critical): Add KIND argument and
648         handle it.
649         * gimple.def: Update CLAUSES comments.
650         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
651         (gimple_build_omp_for): Add argument to prototype.
652         (gimple_omp_for_kind): New.
653         (gimple_omp_for_set_kind): New.
654         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
655         GOVD_DATA_SHARE_CLASS.
656         (enum omp_region_type): Add ORT_SIMD.
657         (gimple_add_tmp_var): Handle ORT_SIMD.
658         (gimplify_var_or_parm_decl): Same.
659         (is_gimple_stmt): Same.
660         (omp_firstprivatize_variable): Same.
661         (omp_add_variable): Only use splay_tree_insert if lookup failed.
662         (omp_notice_variable): Handle ORT_SIMD.
663         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
664         (omp_check_private): Handle ORT_SIMD.
665         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
666         OMP_CLAUSE_SAFELEN.
667         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
668         Handle OMP_CLAUSE_LASTPRIVATE.
669         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
670         OMP_CLAUSE_SAFELEN.
671         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
672         (gimplify_expr): Handle OMP_SIMD.
673         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
674         (expand_GOMP_SIMD_VF): New.
675         (expand_GOMP_SIMD_LAST_LANE): New.
676         * internal-fn.def (GOMP_SIMD_LANE): New.
677         (GOMP_SIMD_VF): New.
678         (GOMP_SIMD_LAST_LANE): New.
679         * omp-low.c: Include target.h.
680         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
681         OMP_CLAUSE_SAFELEN.
682         (check_omp_nesting_restrictions): Same.
683         (omp_max_vf): New.
684         (lower_rec_simd_input_clauses): New.
685         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
686         OMP_CLAUSE_LINEAR.
687         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
688         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
689         (expand_omp_build_assign): New.
690         (expand_omp_for_init_counts): New.
691         (expand_omp_for_init_vars): New.
692         (extract_omp_for_update_vars): New.
693         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
694         and rewrite accordingly.
695         (expand_omp_simd): New.
696         (expand_omp_for): Use expand_omp_simd.
697         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
698         (lower_omp_for): Do not lower the body.
699         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
700         in their own loops.
701         * tree-flow.h (find_omp_clause): Remove prototype.
702         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
703         forcing vectorization of the loop, or if flag_tree_vectorize.
704         (gate_tree_if_conversion): Similarly.
705         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
706         gimple_build_omp_for.
707         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
708         * tree-parloops (create_parallel_loop): Pass kind argument to
709         gimple_build_omp_for.
710         * tree-pretty-print.c (dump_omp_clause): Add cases for
711         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
712         OMP_CLAUSE__SIMDUID_.
713         (dump_generic_node): Handle OMP_SIMD.
714         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
715         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
716         unroll OMP_SIMD loops here.
717         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
718         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
719         loop->safelen.
720         (vect_analyze_data_refs): Handle simd loops.
721         * tree-vect-loop.c (vectorizable_live_operation): Handle
722         IFN_GOMP_SIMD*.
723         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
724         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
725         (vectorizable_load): Same.
726         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
727         (struct simduid_to_vf): New.
728         (simduid_to_vf::hash): New.
729         (simduid_to-vf::equal): New.
730         (struct simd_array_to_simduid): New.
731         (simd_array_to_simduid::hash): New.
732         (simd_array_to_simduid::equal): New.
733         (adjust_simduid_builtins): New.
734         (struct note_simd_array_uses_struct): New.
735         (note_simd_array_uses_cb): New.
736         (note_simd_array_uses): New.
737         (vectorize_loops): Handle simd hints and adjust simd builtins
738         accordingly.
739         * tree-vectorizer.h (struct _stmt_vec_info): Add
740         simd_lane_access_p field.
741         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
742         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
743         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
744         (omp_clause_code_name): Same.
745         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
746         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
747         * tree.def (OMP_SIMD): New entry.
748         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
749         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
750         (OMP_CLAUSE_DECL): Adjust range for new clauses.
751         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
752         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
753         (OMP_CLAUSE_LINEAR_STEP): New.
754         (OMP_CLAUSE_SAFELEN_EXPR): New.
755         (OMP_CLAUSE__SIMDUID__DECL): New.
756         (find_omp_clause): New prototype.
758 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
760         * config/i386/driver-i386.c (host_detect_local_cpu): Update
761         Haswell processor detection.
763 2013-08-27  Christian Widmer  <shadow@umbrox.de>
765         PR target/57927
766         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
767         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
768         AVX2 capable processors.
770 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
772         * config/aarch64/arm_neon.h: Replace all inline asm implementations
773         of vget_low_* with implementations in terms of other intrinsics.
775 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
777         PR middle-end/57219
778         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
779         values to -1, 0 and 1.
781 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
783         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
784         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
785         (<optab><mode>3_insn): Remove.
786         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
787         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
788         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
789         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
790         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
791         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
792         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
793         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
794         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
795         (ror<mode>3_insn): Likewise.
796         * config/aarch64/predicates.md (aarch64_simd_register): New.
798 2013-08-27  Richard Biener  <rguenther@suse.de>
800         PR tree-optimization/57521
801         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
802         one edge is non-critical.
803         (find_phi_replacement_condition): Make sure to use a non-critical
804         edge.  Cleanup and remove old bug workarounds.
805         (bb_postdominates_preds): Remove.
806         (if_convertible_loop_p_1): Do not compute post-dominators.
807         (combine_blocks): Do not free post-dominators.
808         (main_tree_if_conversion): Likewise.
809         (pass_data_if_conversion): Add TODO_verify_ssa.
811 2013-08-27  DJ Delorie  <dj@redhat.com>
813         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
815 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
817         * function.c (assign_parm_find_data_types): Set passed_mode and
818         nominal_mode to the TYPE_MODE of nominal_type for the built
819         pointer type in case of the struct-pass-by-reference.
821 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
823         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
824         (UINT16_TYPE): Change default to "unsigned int".
826         * config/avr/avr.opt (mfract-convert-truncate): New option.
827         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
828         is set, round negative fractional integers according to n1169
829         when converting to integer types.
831 2013-08-26  Jan Hubicka  <jh@suse.cz>
833         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
834         methods can not be called indirectly when their address is not taken.
836 2013-08-26  Jan Hubicka  <jh@suse.cz>
838         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
839         ctor_for_folding.
841 2013-08-26  Jan Hubicka  <jh@suse.cz>
843         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
844         can be unshared.
846 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
848         * reload.c (find_valid_class): Allow classes that do not include
849         FIRST_PSEUDO_REGISTER - 1.
851 2013-08-26  Jan Hubicka  <jh@suse.cz>
853         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
854         fix edge count/frequency when speculation failed; fix type check
855         for the direct call.
857 2013-08-26  Jan Hubicka  <jh@suse.cz>
859         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
861 2013-08-26  Jan Hubicka  <jh@suse.cz>
863         * ipa-inline-transform.c (inline_transform): Be ready for basic block
864         to be changed by edge redirection.
866 2013-08-26  Jan Hubicka  <jh@suse.cz>
868         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
869         formating; add sanity check.
870         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
871         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
872         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
873         (resolve_noninline_speculation): Update callee keys, too.
875 2013-08-26  Jan Hubicka  <jh@suse.cz>
877         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
878         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
880 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
882         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
883         into proper place.
885 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
887         * config/i386/i386.c (ix86_debug_options): Remove prototype.
888         (x86_64_elf_select_section): Ditto.
889         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
890         arguments.
891         (ix86_pass_by_reference): Ditto.
892         (output_set_got): Ditto.
893         (ix86_unary_operator_ok): Ditto.
894         (ix86_expand_builtin): Ditto.
896 2013-08-23  Jan Hubicka  <jh@suse.cz>
898         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
900 2013-08-23  Jan Hubicka  <jh@suse.cz>
902         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
903         (tree_decl_with_vis): Add FINAL field.
905 2013-08-23  Jeff Law  <law@redhat.com>
907         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
908         the debugging dump when the expression is fully redundant.
910 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
912         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
913         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
914         * pretty-print.c (pp_formatted_text_data): Likewise.
915         (pp_write_text_to_stream): Likewise.
916         (pp_write_text_as_dot_label_to_stream): Likewise.
917         (pp_append_r): Likewise.
918         (pp_format): Likewise.
919         (pp_flush): Likewise.
920         (pp_clear_output_area): Likewise.
921         (pp_append_text): Likewise.
922         (pp_formatted_text): Likewise.
923         (pp_remaining_character_count_for_line): Likewise.
924         (pp_newline): Likewise.
925         (pp_character): Likewise.
926         (output_buffer::~output_buffer): Define.
927         (pretty_printer::~pretty_printer): Destruct output buffer.
928         * pretty-print.h (output_buffer::~output_buffer): Declare.
929         (pretty_printer::~pretty_printer): Declare virtual.
931 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
933         PR other/57324
934         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
935         HOST_WIDE_INT_M1U): New macros.
936         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
937         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
938         unsigned -1 for lshift.
939         * cse.c (cse_insn): Likewise.
940         * double-int.c (rshift_double, lshift_double): Likewise.
941         * builtins.c (fold_builtin_bitop): Likewise.
942         * combine.c (force_to_mode): Likewise.
943         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
944         * simplify-rtx.c (simplify_const_unary_operation,
945         simplify_const_binary_operation): Likewise.
946         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
947         check_va_list_escapes): Likewise.
948         * rtlanal.c (nonzero_bits1): Likewise.
949         * expmed.c (expand_smod_pow2): Likewise.
950         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
952 2013-08-23  Jan Hubicka  <jh@suse.cz>
954         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
955         as having address taken.
957 2013-08-23  Jan Hubicka  <jh@suse.cz>
959         * ipa-utils.h (method_class_type): Declare.
960         * ipa-devirt.c (method_class_type): Export.
962         * cgraphunit.c (analyze_functions): Do basic devirtualization;
963         do not walk base classes of anonymous types.
965 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
967         PR rtl-optimization/58220
968         PR regression/58221
969         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
970         handle SEQUENCE insns properly.
972 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
974         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
975         definition.
976         (pp_newline_and_indent): Likewise.
977         (pp_separate_with): Likewise.
978         * pretty-print.c (pp_newline_and_flush): Define.
979         (pp_newline_and_indent): Likewise.
980         (pp_separate_with): Likewise.
982 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
984         PR target/58218
985         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
986         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
988 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
990         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
991         * gcc/config/i386/i386.md (*movti_internal): Use
992         predicate to determine if EVEX is needed.
993         (*movsi_internal): Ditto.
994         (*movdf_internal): Ditto.
995         (*movsf_internal): Ditto.
996         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
998 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
1000         PR tree-optimization/58209
1001         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
1002         (find_tail_calls): Give up for pointer result types if m is non-NULL.
1003         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
1004         emit POINTER_PLUS_EXPR.
1005         (create_tailcall_accumulator): For pointer result type accumulate in
1006         sizetype type.
1008 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
1010         * configure.ac: Add backslashes missing from the last change.
1011         * configure: Regenerate.
1013 2013-08-22  Jan Hubicka  <jh@susue.cz>
1015         * ipa.c (function_and_variable_visibility): First remember function
1016         was global and then make it local.
1018 2013-08-22  Julian Brown  <julian@codesourcery.com>
1020         * configure.ac: Add aarch64 to list of arches which use "nop" in
1021         debug_line test.
1022         * configure: Regenerate.
1024 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1026         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
1027         gnu_libc_has_function.
1028         * config/s390/tpf.h: Likewise.
1030 2013-08-22  Jan Hubicka  <jh@susue.cz>
1032         * timevar.c (validate_phases): Add cast.
1034 2013-08-22  Jan Hubicka  <jh@susue.cz>
1036         * timevar.c (validate_phases): Use size_t for memory.
1037         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
1039 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1041         * pretty-print.h (output_buffer::output_buffer): Declare.
1042         (pretty_printer::pretty_printer): Likewise.
1043         (pp_construct): Remove.
1044         * pretty-print.c (output_buffer::output_buffer): Define.
1045         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
1046         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
1047         (print_gimple_expr): Likewise.
1048         (print_gimple_seq): Likewise.
1049         (gimple_dump_bb): Likewise.
1050         * sched-vis.c (dump_value_slim): Likewise.
1051         (dump_insn_slim): Likewise.
1052         (dump_rtl_slim): Likewise.
1053         (str_pattern_slim): Likewise.
1054         * tree-mudflap.c (mf_varname_tree): Likewise.
1055         * graph.c (print_graph_cfg): Likewise.
1056         (start_graph_dump): Likewise.
1057         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
1058         placement-new.
1059         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
1060         pretty printer initialization.
1061         * coretypes.h (diagnostic_context): Remove superflous type alias
1062         declaration.
1063         (pretty_printer): Likewise.  Declare directly as a class.
1064         (pretty_print_info): Remove declaration as class.
1065         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
1066         and pp_clear_output_area.
1067         (asan_add_global): Likewise.
1069 2013-08-22  Jan Hubicka  <jh@suse.cz>
1071         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
1072         * ipa-utils.h (update_type_inheritance_graph): Declare.
1073         (possible_polymorphic_call_target_p): Declare.
1074         (possible_polymorphic_call_target_p): New.
1075         * ipa-devirt.c: Update toplevel comments.
1076         (cached_polymorphic_call_targets): Move up.
1077         (odr_type_d): Move ID down.
1078         (polymorphic_type_binfo_p): Update comment.
1079         (odr_hasher::remove): Likewise;
1080         (get_odr_type): Set anonymous_namespace.
1081         (dump_odr_type): Dump it.
1082         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
1083         (maybe_record_node): Record node in cached_polymorphic_call_targets.
1084         (record_binfo): Add comment.
1085         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
1086         built.
1087         (devirt_node_removal_hook): Do not iCE when cache is freed.
1088         (possible_polymorphic_call_target_p): New predicate.
1089         (update_type_inheritance_graph): New function.
1091 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1092             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1093             Sergey Lega  <sergey.s.lega@intel.com>
1094             Anna Tikhonova  <anna.tikhonova@intel.com>
1095             Ilya Tocar  <ilya.tocar@intel.com>
1096             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1097             Ilya Verbin  <ilya.verbin@intel.com>
1098             Kirill Yukhin  <kirill.yukhin@intel.com>
1099             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1101         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1102         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1103         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1104         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1105         (OPTION_MASK_ISA_AVX2_UNSET): Update.
1106         (OPTION_MASK_ISA_AVX512F_UNSET): New.
1107         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1108         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1109         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1110         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1111         OPT_mavx512pf, OPT_mavx512er cases.
1112         * config/i386/constraints.md (v): New constraint.
1113         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1114         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1115         (bit_AVX512CD): New.
1116         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1117         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1118         * config/i386/i386-c.c (ix86_target_macros_internal):
1119         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1120         __AVX512PF__.
1121         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1122         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1123         * config/i386/i386.c (regclass_map, dbx_register_map)
1124         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1125         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1126         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1127         -mavx512pf options.
1128         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1129         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
1130         -mavx512pf options.  Fix formatting.
1131         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1132         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
1133         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1134         -mavx512cd, -mavx512pf options.
1135         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1136         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1137         (ix86_preferred_output_reload_class): Replace SSE_REGS with
1138         ALL_SSE_REGS.
1139         (ix86_hard_regno_mode_ok): Support 512-bit registers.
1140         (ix86_set_reg_reg_cost): Ditto.
1141         (x86_order_regs_for_local_alloc): Ditto.
1142         (MAX_VECT_LEN): Extend to 64-byte.
1143         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1144         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1145         (TARGET_AVX512ER, TARGET_AVX512CD): New.
1146         (BIGGEST_ALIGNMENT): Extend to 512-bits.
1147         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1148         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1149         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1150         (SSE_REG_MODE_P): Support new modes.
1151         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1152         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1153         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1154         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1155         (REG_CLASS_CONTENTS): Add new registers.
1156         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1157         (EXT_REX_SSE_REGNO_P): New.
1158         (HI_REGISTER_NAMES): Add new registers.
1159         * config/i386/i386.md: Define constants for new registers.
1160         (mode): Add new 512-bit modes.
1161         (prefix): Support evex prefix.
1162         (isa): Support avx512f, noavx512f, fma_avx512f.
1163         (ssemodesuffix): Add new 512-bit modes.
1164         (movxi): New.
1165         (*movxi_internal_avx512f): Ditto.
1166         (*movdi_internal): Replace constraint "x" with the new constraint "v".
1167         Support MODE_XI.
1168         (*movsi_internal): Likewise.
1169         (*movdf_internal): Likewise.
1170         (*movsf_internal): Likewise.
1171         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1172         (<code><mode>3): Likewise.
1173         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1174         New.
1175         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1176         with the new constraint "v".
1177         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1178         modes.
1179         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1180         with the new constraint "v".
1181         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1182         (<sse2>_storedqu<avxsizesuffix>): Likewise.
1183         (*<plusminus_insn><mode>3): Likewise.
1184         (<sse>_vm<plusminus_insn><mode>3): Likewise.
1185         (*mul<mode>3): Likewise.
1186         (<sse>_vmmul<mode>3): Likewise.
1187         (<sse>_div<mode>3): Likewise.
1188         (<sse>_vmdiv<mode>3): Likewise.
1189         (<sse>_sqrt<mode>2): Likewise.
1190         (<sse>_vmsqrt<mode>2): Likewise.
1191         (*<code><mode>3_finite): Likewise.
1192         (*<code><mode>3) <smaxmin>: Likewise.
1193         (<sse>_vm<code><mode>3): Likewise.
1194         (*<code><mode>3) <any_logic>: Likewise.
1195         (*fma_fmadd_<mode>): Likewise.
1196         (*fma_fmsub_<mode>): Likewise.
1197         (*fma_fnmadd_<mode>): Likewise.
1198         (*fma_fnmsub_<mode>): Likewise.
1199         (*fma_fmaddsub_<mode>): Likewise.
1200         (*fma_fmsubadd_<mode>): Likewise.
1201         (*fmai_fmadd_<mode>): Likewise.
1202         (*fmai_fmsub_<mode>): Likewise.
1203         (*fmai_fnmadd_<mode>): Likewise.
1204         (*fmai_fnmsub_<mode>): Likewise.
1205         (sse_cvtsi2ss): Likewise.
1206         (sse_cvtsi2ssq): Likewise.
1207         (sse_cvtss2si): Likewise.
1208         (sse_cvtss2si_2): Likewise.
1209         (sse_cvtss2siq): Likewise.
1210         (sse_cvtss2siq_2): Likewise.
1211         (sse_cvttss2si): Likewise.
1212         (sse_cvtss2siq_2): Likewise.
1213         (float<sseintvecmodelower><mode>2): Likewise.
1214         (sse2_cvtsd2si_2): Likewise.
1215         (sse2_cvtsd2siq_2): Likewise.
1216         (*<plusminus_insn><mode>3): Likewise.
1217         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1218         (*<sse4_1_avx2>_mul<mode>3): Likewise.
1219         (ashr<mode>3): Likewise.
1220         (<shift_insn><mode>3): Likewise.
1221         (avx2_<code><mode>3): Likewise.
1222         (*avx2_<code><mode>3): Likewise.
1223         (*andnot<mode>3): Likewise.
1224         (*<code><mode>3) <any_logic>: Likewise.
1225         (abs<mode>2): Likewise.
1226         (avx2_permvar<mode>): Likewise.
1227         (avx2_perm<mode>_1): Likewise.
1228         (*avx_vpermilp<mode>): Likewise.
1229         (avx_vpermilvar<mode>3): Likewise.
1230         (avx2_ashrv<mode>): Likewise.
1231         (avx2_<shift_insn>v<mode>): Likewise.
1232         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1233         -mavx512cd.
1234         * doc/rtl.texi: Document XImode.
1236 2013-08-21  Jeff Law  <law@redhat.com>
1238         * tree-flow.h (register_jump_thread): Pass vector of edges
1239         instead of each important edge.
1240         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1241         thread path into a vector and pass that to register_jump_thread.
1242         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1243         passed in edge vector to the current 3-edge form.
1245         Revert:
1246         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1248         * dce.c (fini_dce): Call df_analyze again just in case
1249         delete_unmarked_insns removed anything.
1251 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
1253         * reload.h (struct reg_equivs): Rename to ..
1254         (struct reg_equivs_s): .. this.
1256 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
1258         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1260 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1262         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1264 2013-08-21  Jeff Law  <law@redhat.com>
1266         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1267         simplify assignments too.  If the RHS collapses to a singleton
1268         range, then return the value for the range.
1270 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
1272         * config/i386/sse.md (V16): Rename to...
1273         (VMOVE): this.
1274         (mov<mode>): Update iterator name.
1275         (*mov<mode>_internal): Ditto.
1276         (push<mode>1): Ditto.
1277         (movmisalign<mode>): Ditto.
1279 2013-08-20  Jan Hubicka  <jh@suse.cz>
1281         PR bootstrap/58186
1282         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1283         entry for direct edges.
1284         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1286 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1288         Revert my last two changes, r201865 and r201864:
1290         Revert r201865:
1291         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1293         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1294         instances can own GC refs.
1296         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1297         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1298         (gcc::context::gt_pch_nx): Likewise.
1299         (gcc::context::gt_pch_nx):  Likewise.
1300         * ggc.h (gt_ggc_mx <T>): New.
1301         (gt_pch_nx_with_op <T>): New.
1302         (gt_pch_nx <T>): New.
1303         * passes.c (opt_pass::gt_ggc_mx): New.
1304         (opt_pass::gt_pch_nx): New.
1305         (opt_pass::gt_pch_nx_with_op): New.
1306         (pass_manager::gt_ggc_mx): New.
1307         (pass_manager::gt_pch_nx): New.
1308         (pass_manager::gt_pch_nx_with_op): New.
1309         (pass_manager::operator new): Use
1310         ggc_internal_cleared_alloc_stat rather than xcalloc.
1311         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1312         (pass_manager::gt_ggc_mx): New.
1313         (pass_manager::gt_pch_nx): New.
1314         (pass_manager::gt_pch_nx_with_op): New.
1315         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1316         (opt_pass::operator new): New.
1317         (opt_pass::gt_ggc_mx): New.
1318         (opt_pass::gt_pch_nx): New.
1319         (opt_pass::gt_pch_nx_with_op): New.
1321         Revert r201864:
1322         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1324         * Makefile.in (GTFILES): Add context.h.
1325         * context.c (gcc::context::operator new): New.
1326         (gcc::context::gt_ggc_mx): New.
1327         (gcc::context::gt_pch_nx): New.
1328         (gcc::context::gt_pch_nx): New.
1329         * context.h (gcc::context): Add GTY((user)) marking.
1330         (gcc::context::operator new): New.
1331         (gcc::context::gt_ggc_mx): New.
1332         (gcc::context::gt_pch_nx): New.
1333         (gcc::context::gt_pch_nx): New.
1334         (g): Add GTY marking.
1335         (gt_ggc_mx (gcc::context *)): New.
1336         (gt_pch_nx (gcc::context *)): New.
1337         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1338         void *cookie)): New.
1339         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1341 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1343         * dce.c (fini_dce): Call df_analyze again just in case
1344         delete_unmarked_insns removed anything.
1346 2013-08-20  Teresa Johnson  <tejohnson@google.com>
1348         PR rtl-optimizations/57451
1349         * final.c (reemit_insn_block_notes): Prevent lexical blocks
1350         from crossing split section boundaries.
1352 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1354         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1355         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1356         with MULTLIB_DEFAULTS.
1358 2013-08-20  Nick Clifton  <nickc@redhat.com>
1360         * target.def (narrow_volatile_bitfield): Note that the default
1361         value is false, not !TARGET_STRICT_ALIGN.
1362         * doc/tm.texi: Regenerate.
1364 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
1366         Fix LIB_SPEC for systems without libpthread.
1368         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1369         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1370         for Android.
1371         * config/i386/linux-common.h: Likewise.
1372         * config/mips/linux-common.h: Likewise.
1374 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
1376         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1377         checks.
1379 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1381         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1382         instances can own GC refs.
1384         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1385         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1386         (gcc::context::gt_pch_nx): Likewise.
1387         (gcc::context::gt_pch_nx):  Likewise.
1388         * ggc.h (gt_ggc_mx <T>): New.
1389         (gt_pch_nx_with_op <T>): New.
1390         (gt_pch_nx <T>): New.
1391         * passes.c (opt_pass::gt_ggc_mx): New.
1392         (opt_pass::gt_pch_nx): New.
1393         (opt_pass::gt_pch_nx_with_op): New.
1394         (pass_manager::gt_ggc_mx): New.
1395         (pass_manager::gt_pch_nx): New.
1396         (pass_manager::gt_pch_nx_with_op): New.
1397         (pass_manager::operator new): Use
1398         ggc_internal_cleared_alloc_stat rather than xcalloc.
1399         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1400         (pass_manager::gt_ggc_mx): New.
1401         (pass_manager::gt_pch_nx): New.
1402         (pass_manager::gt_pch_nx_with_op): New.
1403         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1404         (opt_pass::operator new): New.
1405         (opt_pass::gt_ggc_mx): New.
1406         (opt_pass::gt_pch_nx): New.
1407         (opt_pass::gt_pch_nx_with_op): New.
1409 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
1411         * Makefile.in (GTFILES): Add context.h.
1412         * context.c (gcc::context::operator new): New.
1413         (gcc::context::gt_ggc_mx): New.
1414         (gcc::context::gt_pch_nx): New.
1415         (gcc::context::gt_pch_nx): New.
1416         * context.h (gcc::context): Add GTY((user)) marking.
1417         (gcc::context::operator new): New.
1418         (gcc::context::gt_ggc_mx): New.
1419         (gcc::context::gt_pch_nx): New.
1420         (gcc::context::gt_pch_nx): New.
1421         (g): Add GTY marking.
1422         (gt_ggc_mx (gcc::context *)): New.
1423         (gt_pch_nx (gcc::context *)): New.
1424         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1425         void *cookie)): New.
1426         * gengtype.c (open_base_files) <ifiles>: Add context.h.
1428 2013-08-20  Alan Modra  <amodra@gmail.com>
1430         PR target/57865
1431         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1432         (rs6000_emit_epilogue): Likewise.
1434 2013-08-19  Dehao Chen  <dehao@google.com>
1436         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1438 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
1439             Jakub Jelinek  <jakub@redhat.com>
1441         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1442         (BUILT_IN_FABSD64): Likewise.
1443         (BUILT_IN_FABSD128): Likewise.
1444         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1445         (fold_builtin_1): Likewise.
1446         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1447         destination and source operands.
1448         (*abstd2_fpr): Likewise.
1449         (*nabstd2_fpr): Likewise.
1451 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
1453         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1454         JUMP_P and INSN_P.
1456 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
1458         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1459         incomplete.
1461 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1463         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1464         * builtins.c (default_libc_has_function): New.
1465         (gnu_libc_has_function): Ditto.
1466         (no_c99_libc_has_function): Ditto.
1467         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1468         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1469         (fold_builtin_sincos): Likewise.
1470         (fold_builtin_cexp): Likewise.
1471         * builtins.def (DEF_C94_BUILTIN): Likewise.
1472         (DEF_C99_BUILTIN): Likewise.
1473         (DEF_C99_C90RES_BUILTIN): Likewise.
1474         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1475         definitions to using this define.
1476         * config/darwin-protos.h (darwin_libc_has_function): New.
1477         * config/darwin.c (darwin_libc_has_function): Ditto.
1478         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1479         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1480         * config/darwin.h: Ditto.
1481         * config/elfos.h: Ditto.
1482         * config/freebsd.h: Ditto.
1483         * config/i386/cygming.h: Ditto.
1484         * config/i386/djgpp.h: Ditto.
1485         * config/i386/i386-interix.h: Ditto.
1486         * config/microblaze/microblaze.h: Ditto.
1487         * config/mmix/mmix.h: Ditto.
1488         * config/gnu-user.h: Ditto.
1489         * config/ia64/hpux.h: Ditto.
1490         * config/pa/pa-hpux.h: Ditto.
1491         * config/pdp11/pdp11.h: Ditto.
1492         * config/picochip/picochip.h: Ditto.
1493         * config/linux.h: Ditto.
1494         * config/netbsd.h: Ditto.
1495         * config/openbsd.h: Ditto.
1496         * config/rs6000/aix43.h: Ditto.
1497         * config/rs6000/aix51.h: Ditto.
1498         * config/rs6000/aix52.h: Ditto.
1499         * config/rs6000/aix53.h: Ditto.
1500         * config/rs6000/aix61.h: Ditto.
1501         * config/rs6000/darwin.h: Ditto.
1502         * config/rs6000/linux.h: Ditto.
1503         * config/rs6000/linux64.h: Ditto.
1504         * config/s390/tpf.h: Ditto.
1505         * config/sol2-10.h: Ditto.
1506         * config/sol2.h: Ditto.
1507         * config/vms/vms.h: Ditto.
1508         * config/vxworks.h: Ditto.
1509         * config/linux-android.c (linux_android_libc_has_function):
1510         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1511         * config/linux-protos.h (linux_android_libc_has_function):
1512         New declaration.
1513         * config/i386/i386.c (ix86_libc_has_function): New.
1514         * config/i386/i386-protos.h
1515         (ix86_libc_has_function): New declaration.
1516         * config/i386/i386.md
1517         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1518         ("isinf<mode>2): Likewise.
1519         * convert.c (convert_to_integer): Using new target hook
1520         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1521         TARGET_C99_FUNCTIONS.
1522         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1523         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1524         * coretypes.h (function_class): New enum for different
1525         classes of functions.
1526         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1527         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1528         (TARGET_HAS_SINCOS): Likewise.
1529         (TARGET_LIBC_HAS_FUNCTION): New.
1530         * doc/tm.texi: Regenerated.
1531         * targhooks.h (default_libc_has_function): New declaration.
1532         (no_c99_libc_has_function): Ditto.
1533         (gnu_libc_has_function): Ditto.
1534         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1535         and TARGET_HAS_SINCOS.
1537 2013-08-18  Jan Hubicka  <jh@suse.cz>
1539         * Makeifle-in (ipa-devirt.o): New.
1540         (GTFILES): Add ipa-utils.h and ipa-devirt.c
1541         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1542         (analyze_functions): Look into possible targets of polymorphic call.
1543         * dumpfile.c (dump_files): Add type-inheritance dump.
1544         * dumpfile.h (TDI_inheritance): New.
1545         * ipa-devirt.c: New file.
1546         * ipa-utils.h (odr_type_d): Forward declare.
1547         (odr_type): New type.
1548         (build_type_inheritance_graph): Declare.
1549         (possible_polymorphic_call_targets): Declare and introduce inline
1550         variant when only edge is pased.
1551         (dump_possible_polymorphic_call_targets): Likewise.
1552         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1553         * tree.c (type_in_anonymous_namespace_p): Break out from ...
1554         (types_same_for_odr): ... here.
1555         * tree.h (type_in_anonymous_namespace_p): Declare.
1557 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
1559         PR tree-optimization/58006
1560         * tree-parloops.c (take_address_of): Don't ICE if get_name
1561         returns NULL.
1562         (eliminate_local_variables_stmt): Remove clobber stmts.
1564 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
1566         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1567         error message is issued for an alias to undefined symbol.
1569 2013-08-18  Jan Hubicka  <jh@suse.cz>
1571         * cgraph.c (cgraph_create_indirect_edge): Discover
1572         polymorphic calls and record basic info into indirect_info.
1573         * gimple-fold.c (gimple_fold_call): When doing BINFO based
1574         devirtualization, ignore objc function calls.
1575         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1576         call with no parm index info.
1577         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1578         * tree.c (virtual_method_call_p): New function.
1579         * tree.h (virtual_method_call_p): Declare.
1581 2013-08-16  Jan Hubicka  <jh@suse.cz>
1583         PR middle-end/58179
1584         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1586 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
1588         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1589         attribute.
1591 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1593         * gengtype.c (type_for_name): Add special-case support for
1594         locating types within the "gcc::" namespace.
1595         (open_base_files): Emit a "using namespace gcc" directive.
1597 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1599         PR target/58160
1600         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1601         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1603         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1604         array instead of each individual operand as a separate argument.
1605         (emit_fusion_gpr_load): Likewise.
1606         (expand_fusion_gpr_load): Add new function declaration.
1608         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1609         signature to have the operands passed as an array, instead of as
1610         separate arguments.  Allow ZERO_EXTEND to be in the memory
1611         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
1612         depend on the register live/dead flags when peepholes are run.
1613         (expand_fusion_gpr_load): New function to be called from the
1614         peephole2 pass, to change the register that addis sets to be the
1615         target register.
1616         (emit_fusion_gpr_load): Change the calling signature to have the
1617         operands passed as an array, instead of as separate arguments.
1618         Allow ZERO_EXTEND to be in the memory address, and also
1619         SIGN_EXTEND if -mpower8-fusion-sign.
1621         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1622         unspec enumeration.
1623         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1624         adjust the register addis loads up in the peephole2 pass.  Do not
1625         depend on the register live/dead state when the peephole pass is done.
1627 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1629         * gengtype.c (create_user_defined_type): Ensure that the kind
1630         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1631         declaration is seen before the GTY((user)) marking.
1633 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1635         PR target/58105
1636         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1638 2013-08-16  Jan Hubicka  <jh@suse.cz>
1640         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1641         arugment expected_type.
1642         (gimple_fold_call): Use it.
1643         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1644         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1645         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1646         (try_make_edge_direct_virtual_call): Likewise.
1647         * tree.c (obj_type_ref_class): New.
1648         * tree.h (obj_type_ref_class): Use it.
1650 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1652         * sched-vis.c (rtl_slim_pp_initialized): Remove.
1653         (rtl_slim_pp): Likewise.
1654         (init_rtl_slim_pretty_print): Likewise.
1655         (dump_value_slim):  Don't call it.  Use local pretty printer.
1656         (dump_insn_slim): Likewise.
1657         (dump_rtl_slim): Likewise.
1658         (str_pattern_slim): Likewise.
1659         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1660         Simplify.
1662 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
1664         PR tree-optimization/58164
1665         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1666         walk gimple_goto_dest of GIMPLE_GOTO.
1668         PR tree-optimization/58165
1669         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1670         bi_call must be the last stmt in a bb, don't split_block, instead
1671         use fallthru edge from it and give up if there is none.
1672         Release conds vector when returning early.
1674 2013-08-14  Xinliang David Li  <davidxl@google.com>
1676         * config/i386/i386.c (ix86_option_override_internal):
1677         Remove unused variable and field.
1679 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1681         PR target/57949
1682         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1683         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1684         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1685         and Linux, correct BLKmode alignment when 128-bit alignment is
1686         required and compatibility flag is not set.
1687         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1688         for zero-size arguments when compatibility flag is not set.
1690 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1692         PR tree-optimization/58145
1693         * tree-sra.c (build_ref_for_offset): If prev_base has
1694         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1696 2013-08-14  Xinliang David Li  <davidxl@google.com>
1698         * config/i386/i386.c (ix86_option_override_internal):
1699         Fix uninitialized variable error.
1701 2013-08-14  Xinliang David Li  <davidxl@google.com>
1703         * config/i386/i386.opt: Define two new options.
1704         * config/i386/x86-tune.def: Add arch selector field in macros.
1705         * config/i386/i386.h: Adjust macro definition.
1706         * config/i386/i386.c (ix86_option_override_internal):
1707         Refactor the code.
1708         (parse_mtune_ctrl_str): New function.
1709         (set_ix86_tune_features): New function.
1710         (ix86_function_specific_restore): Call the new helper function.
1712 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
1714         PR rtl-optimization/57662
1715         * sel-sched.c (code_motion_process_successors): When the current insn
1716         is removed after the recursive traversal, break from the loop.
1717         Add comments and debug printouts.
1719 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1720             Alexandre Oliva  <aoliva@redhat.com>
1722         PR target/58067
1723         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1724         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1725         there also UNSPEC_PLTOFF.
1727 2013-08-14  Marek Polacek  <polacek@redhat.com>
1729         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1730         NUM_CONDITIONS bit positions.
1732 2013-08-13  Cary Coutant  <ccoutant@google.com>
1734         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1735         (attr_checksum): Hash vector contents instead of pointer.
1736         (attr_checksum_ordered): Likewise.
1738 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
1740         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1741         when Pmode != word_mode.  Add length_address attribute.
1742         (sse3_monitor_<mode>): Merge from sse3_monitor and
1743         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
1744         Pmode != word_mode.  Update insn length attribute.
1745         * config/i386/i386.c (ix86_option_override_internal): Update
1746         ix86_gen_monitor selection for merged sse3_monitor insn.
1748 2013-08-13  Julian Brown  <julian@codesourcery.com>
1750         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1751         perform invalid legitimization on greater-than-word-size modes for
1752         TARGET_E500_DOUBLE.
1754 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
1756         * ira.c (setup_class_translate_array): Use aclass instead of cl
1757         for classes not fully covered by allocno classes.
1759 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
1761         PR tree-optimization/57661
1762         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1763         * tree-inline.c (tree_function_versioning): Initialize it.
1764         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1765         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1766         that is not being copied.
1768         PR sanitizer/56417
1769         * asan.c (instrument_strlen_call): Fix typo in comment.
1770         Use char * type even for the lhs of POINTER_PLUS_EXPR.
1772 2013-08-13  Steve Ellcey  <sellcey@mips.com>
1774         * config/mips/mips.md (prefetch): Use lw instead of ld on
1775         loongson in 32bit mode.
1777 2013-08-13  Nick Clifton  <nickc@redhat.com>
1779         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1781 2013-08-13  Jan Hubicka  <jh@suse.cz>
1783         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1784         introduced edge; fix typo in sanity check.
1785         (cgraph_resolve_speculation): Export; improve diagnostic.
1786         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1787         speculation at type mismatch.
1788         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1789         (cgraph_resolve_speculation): Declare.
1790         (symtab_can_be_discarded): New function.
1791         * value-prof.c (gimple_ic_transform): Remove actual transform code.
1792         * ipa-inline-transform.c (speculation_removed): New global var.
1793         (clone_inlined_nodes): See if speculation can be removed.
1794         (inline_call): If speculations was removed, we growths may not match.
1795         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1796         (speculation_useful_p): New function.
1797         (resolve_noninline_speculation): New function.
1798         (inline_small_functions): Resolve useless speculations.
1799         * ipa-inline.h (speculation_useful_p): Declare
1800         * ipa.c (can_replace_by_local_alias): Simplify.
1801         (ipa_profile): Produce speculative calls in non-lto, too;
1802         add simple cost model; produce local aliases.
1804 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
1806         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1807         PASS_MANAGER_H.
1809 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
1811         * config/i386/i386.c (ix86_function_versions): Use error + inform.
1813 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
1815         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1816         iterator instead of X87MODEF.
1818 2013-08-12  Perez Read  <netfirewall@gmail.com>
1820         PR target/58132
1821         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
1822         operand 0 for intel asm alternative.
1823         (*movabs<mode>_2): Ditto for operand 1.
1825 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
1827         * config/aarch64/arm_none.h
1828         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1830 2013-08-12  Nick Clifton  <nickc@redhat.com>
1832         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1834 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1836         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1837         expand for QI/HImode operand to produce more effictive code for
1838         unsigned char(short) --> float(double) conversion.
1840 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1842         * doc/invoke.texi: Mention that -ftls-model does not force the final
1843         model.
1845 2013-08-12  Marek Polacek  <polacek@redhat.com>
1846             Marc Glisse  <marc.glisse@inria.fr>
1848         PR tree-optimization/57980
1849         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1850         when creating -1 constant.
1852 2013-08-10  Jan Hubicka  <jh@suse.cz>
1854         Workaround binutils PR14342.
1855         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1856         (gimple_init_edge_profiler): Likewise.
1857         (gimple_gen_ic_func_profiler): Likewise.
1859 2013-08-09  Jan Hubicka  <jh@suse.cz>
1861         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1863 2013-08-09  Xinliang David Li  <davidxl@google.com>
1865         * config/i386/stringop.def: New file.
1866         * config/i386/stringop.opt: New file.
1867         * config/i386/i386-opts.h: Include stringopt.def.
1868         * config/i386/i386.opt: Include stringopt.opt.
1869         * config/i386/i386.c (ix86_option_override_internal):
1870         Override default size based stringop inline strategies with options.
1871         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1872         New function.
1874 2013-08-09  Jan Hubicka  <jh@suse.cz>
1876         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1878 2013-08-09  Jan Hubicka  <jh@suse.cz>
1880         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1881         CGRAPH_FREQ_MAX.
1882         (dump_cgraph_node): Dump profile-id.
1883         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1884         and common_target_probability.
1885         * lto-cgraph.c (lto_output_edge): Stream common targets.
1886         (lto_output_node): Stream profile ids.
1887         (input_node): Stream profile ids.
1888         (input_edge): Stream common targets.
1889         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1890         * ipa.c: Include value-prof.h
1891         (ipa_profile_generate_summary): Turn indirect call statement histograms
1892         into common targets.
1893         (ipa_profile): Turn common targets into speculative edges.
1895 2013-08-09  Jan Hubicka  <jh@suse.cz>
1897         * cgraph.h (cgraph_node): Add profile_id.
1898         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1899         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1900         (del_node_map): Update.
1901         (find_func_by_funcdef_no): Replace by ...
1902         (find_func_by_profile_id): ... this one.
1903         (gimple_ic_transform): Do not remove useful histograms when
1904         speculation is not done; dump info when indirect call removal
1905         can happen at LTO.
1906         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1907         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1908         (__gcov_indirect_call_profiler_v2): .. this one.
1909         * profile.h (init_node_map): Update.
1910         * coverage.c (coverage_compute_profile_id): New function.
1911         * coverage.h (coverage_compute_profile_id): Declare.
1912         * tree-profile.c (init_ic_make_global_vars): Make
1913         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1914         (gimple_init_edge_profiler): Update prototype of
1915         __gcov_indirect_call_profiler.
1916         (gimple_gen_ic_func_profiler): Simplify.
1917         (tree_profiling): Use init_node_map
1919 2013-08-09  Jan Hubicka  <jh@suse.cz>
1921         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1922         non-speculative refs.
1923         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1924         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1925         (cgraph_set_call_stmt): Likewise.
1926         (cgraph_create_edge_1): Fix release checking compilatoin;
1927         clear lto_stmt_uid.
1928         (cgraph_free_edge): Free indirect info.
1929         (cgraph_turn_edge_to_speculative): New function.
1930         (cgraph_speculative_call_info): New function.
1931         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1932         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1933         (dump_cgraph_node): Dump speculation.
1934         (verify_edge_count_and_frequency): Accept speculative edges.
1935         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1936         (verify_cgraph_node): Handle speculation.
1937         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1938         (cgraph_set_call_stmt): Update prototype.
1939         (cgraph_make_edge_direct): Update prototype.
1940         (cgraph_speculative_call_info): Declare.
1941         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1942         to change; update call of ipa_find_references.
1943         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1944         lto_stmt_uid and speculative flags.
1945         (ipa_dump_references): Dump speculation.
1946         (ipa_clone_references): Clone speculative flag.
1947         (ipa_clone_referring): Likewise.
1948         (ipa_clone_ref): New function.
1949         (ipa_find_reference): Look into lto_stmt_uids
1950         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1951         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1952         (ipa_find_reference): Update declaration.
1953         (ipa_clone_ref): Declare.
1954         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1955         stream speculative flag.
1956         (lto_output_ref): Stream statements uids and speculation.
1957         (input_ref): Likewise.
1958         (input_edge): Stream speuclation.
1959         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1960         (cgraph_set_call_stmt_including_clones): Handle speculation.
1961         * ipa-inline.c (heap_edge_removal_hook): New function.
1962         (inline_small_functions): Register it.
1963         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1964         also initialize refs.
1965         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1966         edge to change.
1967         (try_make_edge_direct_simple_call): Likewise.
1968         (try_make_edge_direct_simple_call): Likewise.
1969         (update_indirect_edges_after_inlining): Likewise.
1970         (remove_described_reference): Look proper lto_stmt_uid.
1971         (propagate_controlled_uses): Likewise.
1972         (propagate_controlled_uses): Liekwise.
1973         * tree-inline.c (copy_bb): Copy speculative edges.
1974         (redirect_all_calls): New function.
1975         (copy_cfg_body): Do redirection after loop info is updated.
1976         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1978 2013-08-09  Jan Hubicka  <jh@suse.cz>
1980         * lto-streamer-out.c (output_function): Renumber PHIs.
1981         * lto-streamer-in.c (input_function): Likewise.
1983 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1985         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1986         (get_lane_unsigned): Likewise.
1987         (dup_lane_scalar): Likewise.
1988         (get_lane): enable for VALL.
1989         * config/aarch64/aarch64-simd.md
1990         (aarch64_dup_lane_scalar<mode>): Remove.
1991         (aarch64_get_lane_signed<mode>): Likewise.
1992         (aarch64_get_lane_unsigned<mode>): Likewise.
1993         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1994         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1995         (aarch64_get_lane<mode>): Enable for all vector modes.
1996         (aarch64_get_lanedi): Remove misleading constraints.
1997         * config/aarch64/arm_neon.h
1998         (__aarch64_vget_lane_any): Define.
1999         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
2000         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
2001         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
2002         * config/aarch64/iterators.md (VDQQH): New.
2003         (VDQQHS): Likewise.
2004         (vwcore): Likewise.
2006 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
2008         * configure.ac: Add GAS check for LEON instructions on SPARC.
2009         * configure: Regenerate.
2010         * config.in: Likewise.
2011         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
2012         sparc*-*-* block.
2013         * config/sparc/sparc.opt (LEON, LEON3): New masks.
2014         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
2015         for LEON or LEON3.
2016         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
2017         (AS_LEON_FLAG): New macro.
2018         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
2019         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
2020         Deal with LEON and LEON3 for the memory model.
2021         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
2022         (atomic_compare_and_swap<mode>_1): Likewise.
2023         (*atomic_compare_and_swap<mode>_1): Likewise.
2025 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2027         * config/arm/neon.md (vcond): Fix floating-point vector
2028         comparisons against 0.
2030 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
2032         * lra-constraints.c (emit_spill_move): Remove assert.
2033         (process_alt_operands): Add more debugging
2034         output.  Increase reject for spilling into memory.  Decrease
2035         reject for reloading scratch.
2036         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
2038 2013-08-08  Steve Ellcey  <sellcey@mips.com>
2040         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
2041         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
2042         micromips incompatible.  Add nan2008.
2043         (MULTILIB_DIRNAMES): Add nan2008.
2044         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2045         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
2046         and micromips incompatible.  Add nan2008.
2047         (MULTILIB_DIRNAMES): Add nan2008.
2048         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2050 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
2052         PR rtl-optimization/58079
2053         * combine.c (combine_simplify_rtx): Avoid using SUBST if
2054         simplify_comparison has widened a comparison with an integer.
2056 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2058         * config/arm/neon.md (movmisalign<mode>): Disable when we
2059         don't allow unaligned accesses.
2060         (*movmisalign<mode>_neon_store): Likewise.
2061         (*movmisalign<mode>_neon_load): Likewise.
2062         (*movmisalign<mode>_neon_store): Likewise.
2063         (*movmisalign<mode>_neon_load): Likewise.
2065 2013-08-08  Jan Hubicka  <jh@suse.cz>
2067         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
2068         (make_pass_rebuild_cgraph_edges): Also clear references.
2069         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
2070         * ipa-inline-transform.c (inline_transform): Remove all references
2071         after inlining.
2072         * cgraphunit.c (expand_function): Remove all references after
2073         expansion.
2074         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
2075         (ipa_find_reference): Rewrite to iterator.
2076         (remove_stmt_references): Likewise.
2077         (ipa_clear_stmts_in_references): New function.
2078         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
2079         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
2080         clear references.
2081         * ipa-split.c (split_function): Remove references in split function.
2083 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2085         PR target/57431
2086         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
2087         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
2089 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2091         PR target/56979
2092         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2093         suggested mode for the assignment isn't compatible with the
2094         registers required.
2096 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2098         PR target/58065
2099         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2101 2013-08-07  Xinliang David Li  <davidxl@google.com>
2103         * config/i386/i386.opt: New option -mtune-ctrl=.
2104         * config/i386/x86-tune.def: New file.
2105         * config/i386/i386.h: include x86-tune.def.
2106         * config/i386/i386.c (ix86_option_override_internal):
2107         Parsing -mtune-ctrl= option and set tune features.
2109 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
2111         PR other/12081
2112         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2113         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2114         to gen_2arg_fn_t.
2116 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2118         * rtl.h (update_alignments): Declare.
2119         * final.c (grow_label_align): New function extracted from...
2120         (shorten_branches): ...here.  Call it.
2121         (update_alignments): New function.
2122         * reorg.c (sibling_labels): New variable.
2123         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
2124         the new label along with it onto the sibling_labels vector.
2125         (fill_simple_delay_slots): Adjust call to get_label_before.
2126         (fill_slots_from_thread): Likewise.
2127         (relax_delay_slots): Likewise.
2128         (make_return_insns): Likewise.
2129         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2131 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2133         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2134         document its semantics.
2135         (diagnostic_report_diagnostic): Adjust accordingly.
2137 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2139         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2140         (sparc_option_override): ...and port to new C++ pass API.
2141         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2143 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
2145         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2147 2013-08-06  Caroline Tice  <cmtice@google.com>
2149         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2150         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2151         * tree-pass.h: Add pass_vtable_verify.
2152         * varasm.c (assemble_variable): Add code to properly set the comdat
2153         section and name for the .vtable_map_vars section.
2154         (assemble_vtyv_preinit_initializer): New function.
2155         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
2156         LINK_ONCE flag.
2157         * output.h: Add function decl for assemble_vtv_preinit_initializer.
2158         * vtable-verify.c: New file.
2159         * vtable-verify.h: New file.
2160         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2161         initialiation levels.
2162         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2163         * passes.def: Insert pass_vtable_verify.
2164         * aclocal.m4: Reorder includes.
2165         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
2166         -fvtv-counts options.
2167         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2168         as appropriate, if -fvtable-verify=... is used.
2169         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2170         -fvtable-verify=... is used.
2171         * Makefile.in (OBJS):  Add vtable-verify.o to list.
2172         (vtable-verify.o): Add new build rule.
2173         (GTFILES): Add vtable-verify.c to list.
2174         * common.opt (fvtable-verify=): New flag.
2175         (vtv_priority): Values for fvtable-verify= flag.
2176         (fvtv-counts): New flag.
2177         (fvtv-debug): New flag.
2178         * tree.h (save_vtable_map_decl): New extern function decl.
2180 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2182         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2183         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2184         (pass_data_rl78_devirt): ...new pass_data instance and...
2185         (make_pass_rl78_devirt): ...new function.
2186         (rl78_asm_file_start): Port pass registration to new C++ API.
2188 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2190         * coretypes.h (rtl_opt_pass): Add.
2191         (gcc::context): Add.
2192         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2193         (epiphany_init): Port to new C++ pass API.
2194         (epiphany_optimize_mode_switching): Likewise.
2195         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2196         (pass_manager::get_pass_mode_switching): New.
2197         (pass_manager::get_pass_peephole2): New.
2198         * mode-switching.c (pass_mode_switching): Add clone method.
2199         * recog.c (pass_peephole2): Add clone method.
2200         (pass_split_all_insns): Add clone method.
2202 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
2204         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2205         (mips_option_override): ...here, porting to new C++ API for passes.
2207 2013-08-06  Jan Hubicka  <jh@suse.cz>
2209         * cgraph.c (cgraph_get_body): New function based on lto.c
2210         implementation.
2211         * cgraph.h (cgraph_get_body): Declare.
2212         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2213         LTO paths.
2214         * cgraphunit.c (expand_function): Get body prior expanding.
2215         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2216         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2217         really need.
2218         * passes.c (do_per_function_toporder): Get body.
2219         * tree-inline.c (expand_call_inline): Get body prior inlining it.
2220         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2222 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2224         PR fortran/57987
2225         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2226         is not re-finalized.  Rename second parameter to no_collect.
2228 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2230         PR middle-end/58041
2231         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2232         MEM_REF has proper alignment information.
2234 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
2236         PR other/12081
2237         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2238         class insn_gen_fn.
2239         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2240         rtx (*) (rtx, ...) with insn_gen_fn.
2241         * genoutput.c (output_insn_data): Cast gen_? function pointers to
2242         insn_gen_fn::stored_funcptr.  Add initializer braces.
2244 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2246         Rewrite how instances of passes are cloned to remove assumptions
2247         about their sizes (thus allowing pass subclasses to have
2248         additional data fields, albeit non-GC-managed ones at this point).
2250         * passes.c (make_pass_instance): Now that passes have clone
2251         methods, rewrite this function to eliminate XNEW and memcpy
2252         calls that used hardcoded sizes.  Since this function no longer
2253         creates pass instances, rename it to...
2254         (add_pass_instance): ...this.  Document the old way that passes were
2255         numbered and flagged, and rework this function to continue using it.
2256         (next_pass_1): Add an initial_pass argument for use by
2257         add_pass_instance.
2258         (position_pass): When adding multiple instances of a pass, use
2259         the pass's clone method, rather than relying on the XNEW/memcpy
2260         within the former make_pass_instance (now add_pass_instance).
2261         (pass_manager::pass_manager): When invoking next_pass_1, also supply
2262         the initial instance of the current pass within the pass manager.
2264 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2266         This is the automated part of the conversion of passes from C
2267         structs to C++ classes.
2269         Patch autogenerated by refactor_passes.py from
2270         https://github.com/davidmalcolm/gcc-refactoring-scripts
2271         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2273         * asan.c (pass_asan): Convert from a global struct to a subclass of
2274         gimple_opt_pass along with...
2275         (pass_data_asan): ...new pass_data instance and...
2276         (make_pass_asan): ...new function.
2277         (pass_asan_O0): Convert from a global struct to a subclass of
2278         gimple_opt_pass along with...
2279         (pass_data_asan_O0): ...new pass_data instance and...
2280         (make_pass_asan_O0): ...new function.
2281         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2282         subclass of rtl_opt_pass along with...
2283         (pass_data_inc_dec): ...new pass_data instance and...
2284         (make_pass_inc_dec): ...new function.
2285         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2286         a subclass of rtl_opt_pass along with...
2287         (pass_data_reorder_blocks): ...new pass_data instance and...
2288         (make_pass_reorder_blocks): ...new function.
2289         (pass_duplicate_computed_gotos): Convert from a global struct to a
2290         subclass of rtl_opt_pass along with...
2291         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2292         (make_pass_duplicate_computed_gotos): ...new function.
2293         (pass_partition_blocks): Convert from a global struct to a subclass of
2294         rtl_opt_pass along with...
2295         (pass_data_partition_blocks): ...new pass_data instance and...
2296         (make_pass_partition_blocks): ...new function.
2297         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2298         struct to a subclass of rtl_opt_pass along with...
2299         (pass_data_branch_target_load_optimize1): ...new pass_data instance
2300         and...
2301         (make_pass_branch_target_load_optimize1): ...new function.
2302         (pass_branch_target_load_optimize2): Convert from a global struct to a
2303         subclass of rtl_opt_pass along with...
2304         (pass_data_branch_target_load_optimize2): ...new pass_data instance
2305         and...
2306         (make_pass_branch_target_load_optimize2): ...new function.
2307         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2308         of rtl_opt_pass along with...
2309         (pass_data_jump): ...new pass_data instance and...
2310         (make_pass_jump): ...new function.
2311         (pass_jump2): Convert from a global struct to a subclass of
2312         rtl_opt_pass along with...
2313         (pass_data_jump2): ...new pass_data instance and...
2314         (make_pass_jump2): ...new function.
2315         * cfgexpand.c (pass_expand): Convert from a global struct to a
2316         subclass of rtl_opt_pass along with...
2317         (pass_data_expand): ...new pass_data instance and...
2318         (make_pass_expand): ...new function.
2319         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2320         of rtl_opt_pass along with...
2321         (pass_data_free_cfg): ...new pass_data instance and...
2322         (make_pass_free_cfg): ...new function.
2323         (pass_into_cfg_layout_mode): Convert from a global struct to a
2324         subclass of rtl_opt_pass along with...
2325         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2326         (make_pass_into_cfg_layout_mode): ...new function.
2327         (pass_outof_cfg_layout_mode): Convert from a global struct to a
2328         subclass of rtl_opt_pass along with...
2329         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2330         (make_pass_outof_cfg_layout_mode): ...new function.
2331         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2332         struct to a subclass of gimple_opt_pass along with...
2333         (pass_data_build_cgraph_edges): ...new pass_data instance and...
2334         (make_pass_build_cgraph_edges): ...new function.
2335         (pass_rebuild_cgraph_edges): Convert from a global struct to a
2336         subclass of gimple_opt_pass along with...
2337         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2338         (make_pass_rebuild_cgraph_edges): ...new function.
2339         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2340         subclass of gimple_opt_pass along with...
2341         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2342         and...
2343         (make_pass_remove_cgraph_callee_edges): ...new function.
2344         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2345         struct to a subclass of rtl_opt_pass along with...
2346         (pass_data_stack_adjustments): ...new pass_data instance and...
2347         (make_pass_stack_adjustments): ...new function.
2348         * combine.c (pass_combine): Convert from a global struct to a subclass
2349         of rtl_opt_pass along with...
2350         (pass_data_combine): ...new pass_data instance and...
2351         (make_pass_combine): ...new function.
2352         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2353         global struct to a subclass of rtl_opt_pass along with...
2354         (pass_data_compare_elim_after_reload): ...new pass_data instance
2355         and...
2356         (make_pass_compare_elim_after_reload): ...new function.
2357         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2358         of rtl_opt_pass along with...
2359         (pass_data_rtl_cprop): ...new pass_data instance and...
2360         (make_pass_rtl_cprop): ...new function.
2361         * cse.c (pass_cse): Convert from a global struct to a subclass of
2362         rtl_opt_pass along with...
2363         (pass_data_cse): ...new pass_data instance and...
2364         (make_pass_cse): ...new function.
2365         (pass_cse2): Convert from a global struct to a subclass of
2366         rtl_opt_pass along with...
2367         (pass_data_cse2): ...new pass_data instance and...
2368         (make_pass_cse2): ...new function.
2369         (pass_cse_after_global_opts): Convert from a global struct to a
2370         subclass of rtl_opt_pass along with...
2371         (pass_data_cse_after_global_opts): ...new pass_data instance and...
2372         (make_pass_cse_after_global_opts): ...new function.
2373         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2374         of rtl_opt_pass along with...
2375         (pass_data_ud_rtl_dce): ...new pass_data instance and...
2376         (make_pass_ud_rtl_dce): ...new function.
2377         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2378         rtl_opt_pass along with...
2379         (pass_data_fast_rtl_dce): ...new pass_data instance and...
2380         (make_pass_fast_rtl_dce): ...new function.
2381         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2382         a subclass of rtl_opt_pass along with...
2383         (pass_data_df_initialize_opt): ...new pass_data instance and...
2384         (make_pass_df_initialize_opt): ...new function.
2385         (pass_df_initialize_no_opt): Convert from a global struct to a
2386         subclass of rtl_opt_pass along with...
2387         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2388         (make_pass_df_initialize_no_opt): ...new function.
2389         (pass_df_finish): Convert from a global struct to a subclass of
2390         rtl_opt_pass along with...
2391         (pass_data_df_finish): ...new pass_data instance and...
2392         (make_pass_df_finish): ...new function.
2393         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2394         rtl_opt_pass along with...
2395         (pass_data_rtl_dse1): ...new pass_data instance and...
2396         (make_pass_rtl_dse1): ...new function.
2397         (pass_rtl_dse2): Convert from a global struct to a subclass of
2398         rtl_opt_pass along with...
2399         (pass_data_rtl_dse2): ...new pass_data instance and...
2400         (make_pass_rtl_dse2): ...new function.
2401         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2402         subclass of rtl_opt_pass along with...
2403         (pass_data_dwarf2_frame): ...new pass_data instance and...
2404         (make_pass_dwarf2_frame): ...new function.
2405         * except.c (pass_set_nothrow_function_flags): Convert from a global
2406         struct to a subclass of rtl_opt_pass along with...
2407         (pass_data_set_nothrow_function_flags): ...new pass_data instance
2408         and...
2409         (make_pass_set_nothrow_function_flags): ...new function.
2410         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2411         subclass of rtl_opt_pass along with...
2412         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2413         and...
2414         (make_pass_convert_to_eh_region_ranges): ...new function.
2415         * final.c (pass_compute_alignments): Convert from a global struct to a
2416         subclass of rtl_opt_pass along with...
2417         (pass_data_compute_alignments): ...new pass_data instance and...
2418         (make_pass_compute_alignments): ...new function.
2419         (pass_final): Convert from a global struct to a subclass of
2420         rtl_opt_pass along with...
2421         (pass_data_final): ...new pass_data instance and...
2422         (make_pass_final): ...new function.
2423         (pass_shorten_branches): Convert from a global struct to a subclass of
2424         rtl_opt_pass along with...
2425         (pass_data_shorten_branches): ...new pass_data instance and...
2426         (make_pass_shorten_branches): ...new function.
2427         (pass_clean_state): Convert from a global struct to a subclass of
2428         rtl_opt_pass along with...
2429         (pass_data_clean_state): ...new pass_data instance and...
2430         (make_pass_clean_state): ...new function.
2431         * function.c (pass_instantiate_virtual_regs): Convert from a global
2432         struct to a subclass of rtl_opt_pass along with...
2433         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2434         (make_pass_instantiate_virtual_regs): ...new function.
2435         (pass_leaf_regs): Convert from a global struct to a subclass of
2436         rtl_opt_pass along with...
2437         (pass_data_leaf_regs): ...new pass_data instance and...
2438         (make_pass_leaf_regs): ...new function.
2439         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2440         subclass of rtl_opt_pass along with...
2441         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2442         and...
2443         (make_pass_thread_prologue_and_epilogue): ...new function.
2444         (pass_match_asm_constraints): Convert from a global struct to a
2445         subclass of rtl_opt_pass along with...
2446         (pass_data_match_asm_constraints): ...new pass_data instance and...
2447         (make_pass_match_asm_constraints): ...new function.
2448         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2449         subclass of rtl_opt_pass along with...
2450         (pass_data_rtl_fwprop): ...new pass_data instance and...
2451         (make_pass_rtl_fwprop): ...new function.
2452         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2453         rtl_opt_pass along with...
2454         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2455         (make_pass_rtl_fwprop_addr): ...new function.
2456         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2457         rtl_opt_pass along with...
2458         (pass_data_rtl_pre): ...new pass_data instance and...
2459         (make_pass_rtl_pre): ...new function.
2460         (pass_rtl_hoist): Convert from a global struct to a subclass of
2461         rtl_opt_pass along with...
2462         (pass_data_rtl_hoist): ...new pass_data instance and...
2463         (make_pass_rtl_hoist): ...new function.
2464         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2465         subclass of gimple_opt_pass along with...
2466         (pass_data_lower_cf): ...new pass_data instance and...
2467         (make_pass_lower_cf): ...new function.
2468         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2469         from a global struct to a subclass of gimple_opt_pass along with...
2470         (pass_data_strength_reduction): ...new pass_data instance and...
2471         (make_pass_strength_reduction): ...new function.
2472         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2473         of rtl_opt_pass along with...
2474         (pass_data_rtl_ifcvt): ...new pass_data instance and...
2475         (make_pass_rtl_ifcvt): ...new function.
2476         (pass_if_after_combine): Convert from a global struct to a subclass of
2477         rtl_opt_pass along with...
2478         (pass_data_if_after_combine): ...new pass_data instance and...
2479         (make_pass_if_after_combine): ...new function.
2480         (pass_if_after_reload): Convert from a global struct to a subclass of
2481         rtl_opt_pass along with...
2482         (pass_data_if_after_reload): ...new pass_data instance and...
2483         (make_pass_if_after_reload): ...new function.
2484         * init-regs.c (pass_initialize_regs): Convert from a global struct to
2485         a subclass of rtl_opt_pass along with...
2486         (pass_data_initialize_regs): ...new pass_data instance and...
2487         (make_pass_initialize_regs): ...new function.
2488         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2489         of ipa_opt_pass_d along with...
2490         (pass_data_ipa_cp): ...new pass_data instance and...
2491         (make_pass_ipa_cp): ...new function.
2492         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2493         global struct to a subclass of gimple_opt_pass along with...
2494         (pass_data_inline_parameters): ...new pass_data instance and...
2495         (make_pass_inline_parameters): ...new function.
2496         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2497         subclass of gimple_opt_pass along with...
2498         (pass_data_early_inline): ...new pass_data instance and...
2499         (make_pass_early_inline): ...new function.
2500         (pass_ipa_inline): Convert from a global struct to a subclass of
2501         ipa_opt_pass_d along with...
2502         (pass_data_ipa_inline): ...new pass_data instance and...
2503         (make_pass_ipa_inline): ...new function.
2504         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2505         struct to a subclass of gimple_opt_pass along with...
2506         (pass_data_local_pure_const): ...new pass_data instance and...
2507         (make_pass_local_pure_const): ...new function.
2508         (pass_ipa_pure_const): Convert from a global struct to a subclass of
2509         ipa_opt_pass_d along with...
2510         (pass_data_ipa_pure_const): ...new pass_data instance and...
2511         (make_pass_ipa_pure_const): ...new function.
2512         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2513         to a subclass of ipa_opt_pass_d along with...
2514         (pass_data_ipa_reference): ...new pass_data instance and...
2515         (make_pass_ipa_reference): ...new function.
2516         * ipa-split.c (pass_split_functions): Convert from a global struct to
2517         a subclass of gimple_opt_pass along with...
2518         (pass_data_split_functions): ...new pass_data instance and...
2519         (make_pass_split_functions): ...new function.
2520         (pass_feedback_split_functions): Convert from a global struct to a
2521         subclass of gimple_opt_pass along with...
2522         (pass_data_feedback_split_functions): ...new pass_data instance and...
2523         (make_pass_feedback_split_functions): ...new function.
2524         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2525         global struct to a subclass of simple_ipa_opt_pass along with...
2526         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2527         instance and...
2528         (make_pass_ipa_function_and_variable_visibility): ...new function.
2529         (pass_ipa_free_inline_summary): Convert from a global struct to a
2530         subclass of simple_ipa_opt_pass along with...
2531         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2532         (make_pass_ipa_free_inline_summary): ...new function.
2533         (pass_ipa_whole_program_visibility): Convert from a global struct to a
2534         subclass of ipa_opt_pass_d along with...
2535         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2536         and...
2537         (make_pass_ipa_whole_program_visibility): ...new function.
2538         (pass_ipa_profile): Convert from a global struct to a subclass of
2539         ipa_opt_pass_d along with...
2540         (pass_data_ipa_profile): ...new pass_data instance and...
2541         (make_pass_ipa_profile): ...new function.
2542         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2543         ipa_opt_pass_d along with...
2544         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2545         (make_pass_ipa_cdtor_merge): ...new function.
2546         * ira.c (pass_ira): Convert from a global struct to a subclass of
2547         rtl_opt_pass along with...
2548         (pass_data_ira): ...new pass_data instance and...
2549         (make_pass_ira): ...new function.
2550         (pass_reload): Convert from a global struct to a subclass of
2551         rtl_opt_pass along with...
2552         (pass_data_reload): ...new pass_data instance and...
2553         (make_pass_reload): ...new function.
2554         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2555         subclass of rtl_opt_pass along with...
2556         (pass_data_cleanup_barriers): ...new pass_data instance and...
2557         (make_pass_cleanup_barriers): ...new function.
2558         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2559         of rtl_opt_pass along with...
2560         (pass_data_loop2): ...new pass_data instance and...
2561         (make_pass_loop2): ...new function.
2562         (pass_rtl_loop_init): Convert from a global struct to a subclass of
2563         rtl_opt_pass along with...
2564         (pass_data_rtl_loop_init): ...new pass_data instance and...
2565         (make_pass_rtl_loop_init): ...new function.
2566         (pass_rtl_loop_done): Convert from a global struct to a subclass of
2567         rtl_opt_pass along with...
2568         (pass_data_rtl_loop_done): ...new pass_data instance and...
2569         (make_pass_rtl_loop_done): ...new function.
2570         (pass_rtl_move_loop_invariants): Convert from a global struct to a
2571         subclass of rtl_opt_pass along with...
2572         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2573         (make_pass_rtl_move_loop_invariants): ...new function.
2574         (pass_rtl_unswitch): Convert from a global struct to a subclass of
2575         rtl_opt_pass along with...
2576         (pass_data_rtl_unswitch): ...new pass_data instance and...
2577         (make_pass_rtl_unswitch): ...new function.
2578         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2579         subclass of rtl_opt_pass along with...
2580         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2581         and...
2582         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2583         (pass_rtl_doloop): Convert from a global struct to a subclass of
2584         rtl_opt_pass along with...
2585         (pass_data_rtl_doloop): ...new pass_data instance and...
2586         (make_pass_rtl_doloop): ...new function.
2587         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2588         a subclass of rtl_opt_pass along with...
2589         (pass_data_lower_subreg): ...new pass_data instance and...
2590         (make_pass_lower_subreg): ...new function.
2591         (pass_lower_subreg2): Convert from a global struct to a subclass of
2592         rtl_opt_pass along with...
2593         (pass_data_lower_subreg2): ...new pass_data instance and...
2594         (make_pass_lower_subreg2): ...new function.
2595         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2596         struct to a subclass of ipa_opt_pass_d along with...
2597         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2598         (make_pass_ipa_lto_gimple_out): ...new function.
2599         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2600         of ipa_opt_pass_d along with...
2601         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2602         (make_pass_ipa_lto_finish_out): ...new function.
2603         * mode-switching.c (pass_mode_switching): Convert from a global struct
2604         to a subclass of rtl_opt_pass along with...
2605         (pass_data_mode_switching): ...new pass_data instance and...
2606         (make_pass_mode_switching): ...new function.
2607         * modulo-sched.c (pass_sms): Convert from a global struct to a
2608         subclass of rtl_opt_pass along with...
2609         (pass_data_sms): ...new pass_data instance and...
2610         (make_pass_sms): ...new function.
2611         * omp-low.c (pass_expand_omp): Convert from a global struct to a
2612         subclass of gimple_opt_pass along with...
2613         (pass_data_expand_omp): ...new pass_data instance and...
2614         (make_pass_expand_omp): ...new function.
2615         (pass_lower_omp): Convert from a global struct to a subclass of
2616         gimple_opt_pass along with...
2617         (pass_data_lower_omp): ...new pass_data instance and...
2618         (make_pass_lower_omp): ...new function.
2619         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2620         of gimple_opt_pass along with...
2621         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2622         (make_pass_diagnose_omp_blocks): ...new function.
2623         * passes.c (pass_early_local_passes): Convert from a global struct to
2624         a subclass of simple_ipa_opt_pass along with...
2625         (pass_data_early_local_passes): ...new pass_data instance and...
2626         (make_pass_early_local_passes): ...new function.
2627         (pass_all_early_optimizations): Convert from a global struct to a
2628         subclass of gimple_opt_pass along with...
2629         (pass_data_all_early_optimizations): ...new pass_data instance and...
2630         (make_pass_all_early_optimizations): ...new function.
2631         (pass_all_optimizations): Convert from a global struct to a subclass
2632         of gimple_opt_pass along with...
2633         (pass_data_all_optimizations): ...new pass_data instance and...
2634         (make_pass_all_optimizations): ...new function.
2635         (pass_all_optimizations_g): Convert from a global struct to a subclass
2636         of gimple_opt_pass along with...
2637         (pass_data_all_optimizations_g): ...new pass_data instance and...
2638         (make_pass_all_optimizations_g): ...new function.
2639         (pass_rest_of_compilation): Convert from a global struct to a subclass
2640         of rtl_opt_pass along with...
2641         (pass_data_rest_of_compilation): ...new pass_data instance and...
2642         (make_pass_rest_of_compilation): ...new function.
2643         (pass_postreload): Convert from a global struct to a subclass of
2644         rtl_opt_pass along with...
2645         (pass_data_postreload): ...new pass_data instance and...
2646         (make_pass_postreload): ...new function.
2647         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2648         subclass of rtl_opt_pass along with...
2649         (pass_data_gcse2): ...new pass_data instance and...
2650         (make_pass_gcse2): ...new function.
2651         * postreload.c (pass_postreload_cse): Convert from a global struct to
2652         a subclass of rtl_opt_pass along with...
2653         (pass_data_postreload_cse): ...new pass_data instance and...
2654         (make_pass_postreload_cse): ...new function.
2655         * predict.c (pass_profile): Convert from a global struct to a subclass
2656         of gimple_opt_pass along with...
2657         (pass_data_profile): ...new pass_data instance and...
2658         (make_pass_profile): ...new function.
2659         (pass_strip_predict_hints): Convert from a global struct to a subclass
2660         of gimple_opt_pass along with...
2661         (pass_data_strip_predict_hints): ...new pass_data instance and...
2662         (make_pass_strip_predict_hints): ...new function.
2663         * recog.c (pass_peephole2): Convert from a global struct to a subclass
2664         of rtl_opt_pass along with...
2665         (pass_data_peephole2): ...new pass_data instance and...
2666         (make_pass_peephole2): ...new function.
2667         (pass_split_all_insns): Convert from a global struct to a subclass of
2668         rtl_opt_pass along with...
2669         (pass_data_split_all_insns): ...new pass_data instance and...
2670         (make_pass_split_all_insns): ...new function.
2671         (pass_split_after_reload): Convert from a global struct to a subclass
2672         of rtl_opt_pass along with...
2673         (pass_data_split_after_reload): ...new pass_data instance and...
2674         (make_pass_split_after_reload): ...new function.
2675         (pass_split_before_regstack): Convert from a global struct to a
2676         subclass of rtl_opt_pass along with...
2677         (pass_data_split_before_regstack): ...new pass_data instance and...
2678         (make_pass_split_before_regstack): ...new function.
2679         (pass_split_before_sched2): Convert from a global struct to a subclass
2680         of rtl_opt_pass along with...
2681         (pass_data_split_before_sched2): ...new pass_data instance and...
2682         (make_pass_split_before_sched2): ...new function.
2683         (pass_split_for_shorten_branches): Convert from a global struct to a
2684         subclass of rtl_opt_pass along with...
2685         (pass_data_split_for_shorten_branches): ...new pass_data instance
2686         and...
2687         (make_pass_split_for_shorten_branches): ...new function.
2688         * ree.c (pass_ree): Convert from a global struct to a subclass of
2689         rtl_opt_pass along with...
2690         (pass_data_ree): ...new pass_data instance and...
2691         (make_pass_ree): ...new function.
2692         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2693         subclass of rtl_opt_pass along with...
2694         (pass_data_stack_regs): ...new pass_data instance and...
2695         (make_pass_stack_regs): ...new function.
2696         (pass_stack_regs_run): Convert from a global struct to a subclass of
2697         rtl_opt_pass along with...
2698         (pass_data_stack_regs_run): ...new pass_data instance and...
2699         (make_pass_stack_regs_run): ...new function.
2700         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2701         subclass of rtl_opt_pass along with...
2702         (pass_data_cprop_hardreg): ...new pass_data instance and...
2703         (make_pass_cprop_hardreg): ...new function.
2704         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2705         subclass of rtl_opt_pass along with...
2706         (pass_data_reginfo_init): ...new pass_data instance and...
2707         (make_pass_reginfo_init): ...new function.
2708         * regmove.c (pass_regmove): Convert from a global struct to a subclass
2709         of rtl_opt_pass along with...
2710         (pass_data_regmove): ...new pass_data instance and...
2711         (make_pass_regmove): ...new function.
2712         * regrename.c (pass_regrename): Convert from a global struct to a
2713         subclass of rtl_opt_pass along with...
2714         (pass_data_regrename): ...new pass_data instance and...
2715         (make_pass_regrename): ...new function.
2716         * reorg.c (pass_delay_slots): Convert from a global struct to a
2717         subclass of rtl_opt_pass along with...
2718         (pass_data_delay_slots): ...new pass_data instance and...
2719         (make_pass_delay_slots): ...new function.
2720         (pass_machine_reorg): Convert from a global struct to a subclass of
2721         rtl_opt_pass along with...
2722         (pass_data_machine_reorg): ...new pass_data instance and...
2723         (make_pass_machine_reorg): ...new function.
2724         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2725         of rtl_opt_pass along with...
2726         (pass_data_sched): ...new pass_data instance and...
2727         (make_pass_sched): ...new function.
2728         (pass_sched2): Convert from a global struct to a subclass of
2729         rtl_opt_pass along with...
2730         (pass_data_sched2): ...new pass_data instance and...
2731         (make_pass_sched2): ...new function.
2732         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2733         to a subclass of rtl_opt_pass along with...
2734         (pass_data_stack_ptr_mod): ...new pass_data instance and...
2735         (make_pass_stack_ptr_mod): ...new function.
2736         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2737         to a subclass of rtl_opt_pass along with...
2738         (pass_data_rtl_store_motion): ...new pass_data instance and...
2739         (make_pass_rtl_store_motion): ...new function.
2740         * tracer.c (pass_tracer): Convert from a global struct to a subclass
2741         of gimple_opt_pass along with...
2742         (pass_data_tracer): ...new pass_data instance and...
2743         (make_pass_tracer): ...new function.
2744         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2745         to a subclass of gimple_opt_pass along with...
2746         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2747         (make_pass_diagnose_tm_blocks): ...new function.
2748         (pass_lower_tm): Convert from a global struct to a subclass of
2749         gimple_opt_pass along with...
2750         (pass_data_lower_tm): ...new pass_data instance and...
2751         (make_pass_lower_tm): ...new function.
2752         (pass_tm_init): Convert from a global struct to a subclass of
2753         gimple_opt_pass along with...
2754         (pass_data_tm_init): ...new pass_data instance and...
2755         (make_pass_tm_init): ...new function.
2756         (pass_tm_mark): Convert from a global struct to a subclass of
2757         gimple_opt_pass along with...
2758         (pass_data_tm_mark): ...new pass_data instance and...
2759         (make_pass_tm_mark): ...new function.
2760         (pass_tm_edges): Convert from a global struct to a subclass of
2761         gimple_opt_pass along with...
2762         (pass_data_tm_edges): ...new pass_data instance and...
2763         (make_pass_tm_edges): ...new function.
2764         (pass_tm_memopt): Convert from a global struct to a subclass of
2765         gimple_opt_pass along with...
2766         (pass_data_tm_memopt): ...new pass_data instance and...
2767         (make_pass_tm_memopt): ...new function.
2768         (pass_ipa_tm): Convert from a global struct to a subclass of
2769         simple_ipa_opt_pass along with...
2770         (pass_data_ipa_tm): ...new pass_data instance and...
2771         (make_pass_ipa_tm): ...new function.
2772         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2773         subclass of gimple_opt_pass along with...
2774         (pass_data_call_cdce): ...new pass_data instance and...
2775         (make_pass_call_cdce): ...new function.
2776         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2777         subclass of gimple_opt_pass along with...
2778         (pass_data_build_cfg): ...new pass_data instance and...
2779         (make_pass_build_cfg): ...new function.
2780         (pass_split_crit_edges): Convert from a global struct to a subclass of
2781         gimple_opt_pass along with...
2782         (pass_data_split_crit_edges): ...new pass_data instance and...
2783         (make_pass_split_crit_edges): ...new function.
2784         (pass_warn_function_return): Convert from a global struct to a
2785         subclass of gimple_opt_pass along with...
2786         (pass_data_warn_function_return): ...new pass_data instance and...
2787         (make_pass_warn_function_return): ...new function.
2788         (pass_warn_function_noreturn): Convert from a global struct to a
2789         subclass of gimple_opt_pass along with...
2790         (pass_data_warn_function_noreturn): ...new pass_data instance and...
2791         (make_pass_warn_function_noreturn): ...new function.
2792         (pass_warn_unused_result): Convert from a global struct to a subclass
2793         of gimple_opt_pass along with...
2794         (pass_data_warn_unused_result): ...new pass_data instance and...
2795         (make_pass_warn_unused_result): ...new function.
2796         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2797         a subclass of gimple_opt_pass along with...
2798         (pass_data_merge_phi): ...new pass_data instance and...
2799         (make_pass_merge_phi): ...new function.
2800         * tree-complex.c (pass_lower_complex): Convert from a global struct to
2801         a subclass of gimple_opt_pass along with...
2802         (pass_data_lower_complex): ...new pass_data instance and...
2803         (make_pass_lower_complex): ...new function.
2804         (pass_lower_complex_O0): Convert from a global struct to a subclass of
2805         gimple_opt_pass along with...
2806         (pass_data_lower_complex_O0): ...new pass_data instance and...
2807         (make_pass_lower_complex_O0): ...new function.
2808         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2809         subclass of gimple_opt_pass along with...
2810         (pass_data_lower_eh): ...new pass_data instance and...
2811         (make_pass_lower_eh): ...new function.
2812         (pass_refactor_eh): Convert from a global struct to a subclass of
2813         gimple_opt_pass along with...
2814         (pass_data_refactor_eh): ...new pass_data instance and...
2815         (make_pass_refactor_eh): ...new function.
2816         (pass_lower_resx): Convert from a global struct to a subclass of
2817         gimple_opt_pass along with...
2818         (pass_data_lower_resx): ...new pass_data instance and...
2819         (make_pass_lower_resx): ...new function.
2820         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2821         of gimple_opt_pass along with...
2822         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2823         (make_pass_lower_eh_dispatch): ...new function.
2824         (pass_cleanup_eh): Convert from a global struct to a subclass of
2825         gimple_opt_pass along with...
2826         (pass_data_cleanup_eh): ...new pass_data instance and...
2827         (make_pass_cleanup_eh): ...new function.
2828         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2829         to a subclass of simple_ipa_opt_pass along with...
2830         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2831         (make_pass_ipa_lower_emutls): ...new function.
2832         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2833         a subclass of gimple_opt_pass along with...
2834         (pass_data_if_conversion): ...new pass_data instance and...
2835         (make_pass_if_conversion): ...new function.
2836         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2837         subclass of gimple_opt_pass along with...
2838         (pass_data_build_ssa): ...new pass_data instance and...
2839         (make_pass_build_ssa): ...new function.
2840         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2841         global struct to a subclass of gimple_opt_pass along with...
2842         (pass_data_loop_distribution): ...new pass_data instance and...
2843         (make_pass_loop_distribution): ...new function.
2844         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2845         subclass of gimple_opt_pass along with...
2846         (pass_data_mudflap_1): ...new pass_data instance and...
2847         (make_pass_mudflap_1): ...new function.
2848         (pass_mudflap_2): Convert from a global struct to a subclass of
2849         gimple_opt_pass along with...
2850         (pass_data_mudflap_2): ...new pass_data instance and...
2851         (make_pass_mudflap_2): ...new function.
2852         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2853         subclass of gimple_opt_pass along with...
2854         (pass_data_mudflap_1): ...new pass_data instance and...
2855         (make_pass_mudflap_1): ...new function.
2856         (pass_mudflap_2): Convert from a global struct to a subclass of
2857         gimple_opt_pass along with...
2858         (pass_data_mudflap_2): ...new pass_data instance and...
2859         (make_pass_mudflap_2): ...new function.
2860         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2861         gimple_opt_pass along with...
2862         (pass_data_nrv): ...new pass_data instance and...
2863         (make_pass_nrv): ...new function.
2864         (pass_return_slot): Convert from a global struct to a subclass of
2865         gimple_opt_pass along with...
2866         (pass_data_return_slot): ...new pass_data instance and...
2867         (make_pass_return_slot): ...new function.
2868         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2869         to a subclass of gimple_opt_pass along with...
2870         (pass_data_object_sizes): ...new pass_data instance and...
2871         (make_pass_object_sizes): ...new function.
2872         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2873         global struct to a subclass of gimple_opt_pass along with...
2874         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2875         and...
2876         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2877         (pass_fixup_cfg): Convert from a global struct to a subclass of
2878         gimple_opt_pass along with...
2879         (pass_data_fixup_cfg): ...new pass_data instance and...
2880         (make_pass_fixup_cfg): ...new function.
2881         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2882         (make_pass_mudflap_1): ...new function.
2883         (pass_mudflap_2): Replace declaration with that of...
2884         (make_pass_mudflap_2): ...new function.
2885         (pass_asan): Replace declaration with that of...
2886         (make_pass_asan): ...new function.
2887         (pass_asan_O0): Replace declaration with that of...
2888         (make_pass_asan_O0): ...new function.
2889         (pass_tsan): Replace declaration with that of...
2890         (make_pass_tsan): ...new function.
2891         (pass_tsan_O0): Replace declaration with that of...
2892         (make_pass_tsan_O0): ...new function.
2893         (pass_lower_cf): Replace declaration with that of...
2894         (make_pass_lower_cf): ...new function.
2895         (pass_refactor_eh): Replace declaration with that of...
2896         (make_pass_refactor_eh): ...new function.
2897         (pass_lower_eh): Replace declaration with that of...
2898         (make_pass_lower_eh): ...new function.
2899         (pass_lower_eh_dispatch): Replace declaration with that of...
2900         (make_pass_lower_eh_dispatch): ...new function.
2901         (pass_lower_resx): Replace declaration with that of...
2902         (make_pass_lower_resx): ...new function.
2903         (pass_build_cfg): Replace declaration with that of...
2904         (make_pass_build_cfg): ...new function.
2905         (pass_early_tree_profile): Replace declaration with that of...
2906         (make_pass_early_tree_profile): ...new function.
2907         (pass_cleanup_eh): Replace declaration with that of...
2908         (make_pass_cleanup_eh): ...new function.
2909         (pass_sra): Replace declaration with that of...
2910         (make_pass_sra): ...new function.
2911         (pass_sra_early): Replace declaration with that of...
2912         (make_pass_sra_early): ...new function.
2913         (pass_early_ipa_sra): Replace declaration with that of...
2914         (make_pass_early_ipa_sra): ...new function.
2915         (pass_tail_recursion): Replace declaration with that of...
2916         (make_pass_tail_recursion): ...new function.
2917         (pass_tail_calls): Replace declaration with that of...
2918         (make_pass_tail_calls): ...new function.
2919         (pass_tree_loop): Replace declaration with that of...
2920         (make_pass_tree_loop): ...new function.
2921         (pass_tree_loop_init): Replace declaration with that of...
2922         (make_pass_tree_loop_init): ...new function.
2923         (pass_lim): Replace declaration with that of...
2924         (make_pass_lim): ...new function.
2925         (pass_tree_unswitch): Replace declaration with that of...
2926         (make_pass_tree_unswitch): ...new function.
2927         (pass_predcom): Replace declaration with that of...
2928         (make_pass_predcom): ...new function.
2929         (pass_iv_canon): Replace declaration with that of...
2930         (make_pass_iv_canon): ...new function.
2931         (pass_scev_cprop): Replace declaration with that of...
2932         (make_pass_scev_cprop): ...new function.
2933         (pass_empty_loop): Replace declaration with that of...
2934         (make_pass_empty_loop): ...new function.
2935         (pass_record_bounds): Replace declaration with that of...
2936         (make_pass_record_bounds): ...new function.
2937         (pass_graphite): Replace declaration with that of...
2938         (make_pass_graphite): ...new function.
2939         (pass_graphite_transforms): Replace declaration with that of...
2940         (make_pass_graphite_transforms): ...new function.
2941         (pass_if_conversion): Replace declaration with that of...
2942         (make_pass_if_conversion): ...new function.
2943         (pass_loop_distribution): Replace declaration with that of...
2944         (make_pass_loop_distribution): ...new function.
2945         (pass_vectorize): Replace declaration with that of...
2946         (make_pass_vectorize): ...new function.
2947         (pass_slp_vectorize): Replace declaration with that of...
2948         (make_pass_slp_vectorize): ...new function.
2949         (pass_complete_unroll): Replace declaration with that of...
2950         (make_pass_complete_unroll): ...new function.
2951         (pass_complete_unrolli): Replace declaration with that of...
2952         (make_pass_complete_unrolli): ...new function.
2953         (pass_parallelize_loops): Replace declaration with that of...
2954         (make_pass_parallelize_loops): ...new function.
2955         (pass_loop_prefetch): Replace declaration with that of...
2956         (make_pass_loop_prefetch): ...new function.
2957         (pass_iv_optimize): Replace declaration with that of...
2958         (make_pass_iv_optimize): ...new function.
2959         (pass_tree_loop_done): Replace declaration with that of...
2960         (make_pass_tree_loop_done): ...new function.
2961         (pass_ch): Replace declaration with that of...
2962         (make_pass_ch): ...new function.
2963         (pass_ccp): Replace declaration with that of...
2964         (make_pass_ccp): ...new function.
2965         (pass_phi_only_cprop): Replace declaration with that of...
2966         (make_pass_phi_only_cprop): ...new function.
2967         (pass_build_ssa): Replace declaration with that of...
2968         (make_pass_build_ssa): ...new function.
2969         (pass_build_alias): Replace declaration with that of...
2970         (make_pass_build_alias): ...new function.
2971         (pass_build_ealias): Replace declaration with that of...
2972         (make_pass_build_ealias): ...new function.
2973         (pass_dominator): Replace declaration with that of...
2974         (make_pass_dominator): ...new function.
2975         (pass_dce): Replace declaration with that of...
2976         (make_pass_dce): ...new function.
2977         (pass_dce_loop): Replace declaration with that of...
2978         (make_pass_dce_loop): ...new function.
2979         (pass_cd_dce): Replace declaration with that of...
2980         (make_pass_cd_dce): ...new function.
2981         (pass_call_cdce): Replace declaration with that of...
2982         (make_pass_call_cdce): ...new function.
2983         (pass_merge_phi): Replace declaration with that of...
2984         (make_pass_merge_phi): ...new function.
2985         (pass_split_crit_edges): Replace declaration with that of...
2986         (make_pass_split_crit_edges): ...new function.
2987         (pass_pre): Replace declaration with that of...
2988         (make_pass_pre): ...new function.
2989         (pass_profile): Replace declaration with that of...
2990         (make_pass_profile): ...new function.
2991         (pass_strip_predict_hints): Replace declaration with that of...
2992         (make_pass_strip_predict_hints): ...new function.
2993         (pass_lower_complex_O0): Replace declaration with that of...
2994         (make_pass_lower_complex_O0): ...new function.
2995         (pass_lower_complex): Replace declaration with that of...
2996         (make_pass_lower_complex): ...new function.
2997         (pass_lower_vector): Replace declaration with that of...
2998         (make_pass_lower_vector): ...new function.
2999         (pass_lower_vector_ssa): Replace declaration with that of...
3000         (make_pass_lower_vector_ssa): ...new function.
3001         (pass_lower_omp): Replace declaration with that of...
3002         (make_pass_lower_omp): ...new function.
3003         (pass_diagnose_omp_blocks): Replace declaration with that of...
3004         (make_pass_diagnose_omp_blocks): ...new function.
3005         (pass_expand_omp): Replace declaration with that of...
3006         (make_pass_expand_omp): ...new function.
3007         (pass_expand_omp_ssa): Replace declaration with that of...
3008         (make_pass_expand_omp_ssa): ...new function.
3009         (pass_object_sizes): Replace declaration with that of...
3010         (make_pass_object_sizes): ...new function.
3011         (pass_strlen): Replace declaration with that of...
3012         (make_pass_strlen): ...new function.
3013         (pass_fold_builtins): Replace declaration with that of...
3014         (make_pass_fold_builtins): ...new function.
3015         (pass_stdarg): Replace declaration with that of...
3016         (make_pass_stdarg): ...new function.
3017         (pass_early_warn_uninitialized): Replace declaration with that of...
3018         (make_pass_early_warn_uninitialized): ...new function.
3019         (pass_late_warn_uninitialized): Replace declaration with that of...
3020         (make_pass_late_warn_uninitialized): ...new function.
3021         (pass_cse_reciprocals): Replace declaration with that of...
3022         (make_pass_cse_reciprocals): ...new function.
3023         (pass_cse_sincos): Replace declaration with that of...
3024         (make_pass_cse_sincos): ...new function.
3025         (pass_optimize_bswap): Replace declaration with that of...
3026         (make_pass_optimize_bswap): ...new function.
3027         (pass_optimize_widening_mul): Replace declaration with that of...
3028         (make_pass_optimize_widening_mul): ...new function.
3029         (pass_warn_function_return): Replace declaration with that of...
3030         (make_pass_warn_function_return): ...new function.
3031         (pass_warn_function_noreturn): Replace declaration with that of...
3032         (make_pass_warn_function_noreturn): ...new function.
3033         (pass_cselim): Replace declaration with that of...
3034         (make_pass_cselim): ...new function.
3035         (pass_phiopt): Replace declaration with that of...
3036         (make_pass_phiopt): ...new function.
3037         (pass_forwprop): Replace declaration with that of...
3038         (make_pass_forwprop): ...new function.
3039         (pass_phiprop): Replace declaration with that of...
3040         (make_pass_phiprop): ...new function.
3041         (pass_tree_ifcombine): Replace declaration with that of...
3042         (make_pass_tree_ifcombine): ...new function.
3043         (pass_dse): Replace declaration with that of...
3044         (make_pass_dse): ...new function.
3045         (pass_nrv): Replace declaration with that of...
3046         (make_pass_nrv): ...new function.
3047         (pass_rename_ssa_copies): Replace declaration with that of...
3048         (make_pass_rename_ssa_copies): ...new function.
3049         (pass_sink_code): Replace declaration with that of...
3050         (make_pass_sink_code): ...new function.
3051         (pass_fre): Replace declaration with that of...
3052         (make_pass_fre): ...new function.
3053         (pass_check_data_deps): Replace declaration with that of...
3054         (make_pass_check_data_deps): ...new function.
3055         (pass_copy_prop): Replace declaration with that of...
3056         (make_pass_copy_prop): ...new function.
3057         (pass_vrp): Replace declaration with that of...
3058         (make_pass_vrp): ...new function.
3059         (pass_uncprop): Replace declaration with that of...
3060         (make_pass_uncprop): ...new function.
3061         (pass_return_slot): Replace declaration with that of...
3062         (make_pass_return_slot): ...new function.
3063         (pass_reassoc): Replace declaration with that of...
3064         (make_pass_reassoc): ...new function.
3065         (pass_rebuild_cgraph_edges): Replace declaration with that of...
3066         (make_pass_rebuild_cgraph_edges): ...new function.
3067         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
3068         (make_pass_remove_cgraph_callee_edges): ...new function.
3069         (pass_build_cgraph_edges): Replace declaration with that of...
3070         (make_pass_build_cgraph_edges): ...new function.
3071         (pass_local_pure_const): Replace declaration with that of...
3072         (make_pass_local_pure_const): ...new function.
3073         (pass_tracer): Replace declaration with that of...
3074         (make_pass_tracer): ...new function.
3075         (pass_warn_unused_result): Replace declaration with that of...
3076         (make_pass_warn_unused_result): ...new function.
3077         (pass_diagnose_tm_blocks): Replace declaration with that of...
3078         (make_pass_diagnose_tm_blocks): ...new function.
3079         (pass_lower_tm): Replace declaration with that of...
3080         (make_pass_lower_tm): ...new function.
3081         (pass_tm_init): Replace declaration with that of...
3082         (make_pass_tm_init): ...new function.
3083         (pass_tm_mark): Replace declaration with that of...
3084         (make_pass_tm_mark): ...new function.
3085         (pass_tm_memopt): Replace declaration with that of...
3086         (make_pass_tm_memopt): ...new function.
3087         (pass_tm_edges): Replace declaration with that of...
3088         (make_pass_tm_edges): ...new function.
3089         (pass_split_functions): Replace declaration with that of...
3090         (make_pass_split_functions): ...new function.
3091         (pass_feedback_split_functions): Replace declaration with that of...
3092         (make_pass_feedback_split_functions): ...new function.
3093         (pass_strength_reduction): Replace declaration with that of...
3094         (make_pass_strength_reduction): ...new function.
3095         (pass_ipa_lower_emutls): Replace declaration with that of...
3096         (make_pass_ipa_lower_emutls): ...new function.
3097         (pass_ipa_function_and_variable_visibility): Replace declaration with
3098         that of...
3099         (make_pass_ipa_function_and_variable_visibility): ...new function.
3100         (pass_ipa_tree_profile): Replace declaration with that of...
3101         (make_pass_ipa_tree_profile): ...new function.
3102         (pass_early_local_passes): Replace declaration with that of...
3103         (make_pass_early_local_passes): ...new function.
3104         (pass_ipa_whole_program_visibility): Replace declaration with that
3105         of...
3106         (make_pass_ipa_whole_program_visibility): ...new function.
3107         (pass_ipa_lto_gimple_out): Replace declaration with that of...
3108         (make_pass_ipa_lto_gimple_out): ...new function.
3109         (pass_ipa_increase_alignment): Replace declaration with that of...
3110         (make_pass_ipa_increase_alignment): ...new function.
3111         (pass_ipa_inline): Replace declaration with that of...
3112         (make_pass_ipa_inline): ...new function.
3113         (pass_ipa_free_lang_data): Replace declaration with that of...
3114         (make_pass_ipa_free_lang_data): ...new function.
3115         (pass_ipa_free_inline_summary): Replace declaration with that of...
3116         (make_pass_ipa_free_inline_summary): ...new function.
3117         (pass_ipa_cp): Replace declaration with that of...
3118         (make_pass_ipa_cp): ...new function.
3119         (pass_ipa_reference): Replace declaration with that of...
3120         (make_pass_ipa_reference): ...new function.
3121         (pass_ipa_pure_const): Replace declaration with that of...
3122         (make_pass_ipa_pure_const): ...new function.
3123         (pass_ipa_pta): Replace declaration with that of...
3124         (make_pass_ipa_pta): ...new function.
3125         (pass_ipa_lto_finish_out): Replace declaration with that of...
3126         (make_pass_ipa_lto_finish_out): ...new function.
3127         (pass_ipa_tm): Replace declaration with that of...
3128         (make_pass_ipa_tm): ...new function.
3129         (pass_ipa_profile): Replace declaration with that of...
3130         (make_pass_ipa_profile): ...new function.
3131         (pass_ipa_cdtor_merge): Replace declaration with that of...
3132         (make_pass_ipa_cdtor_merge): ...new function.
3133         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3134         of...
3135         (make_pass_cleanup_cfg_post_optimizing): ...new function.
3136         (pass_init_datastructures): Replace declaration with that of...
3137         (make_pass_init_datastructures): ...new function.
3138         (pass_fixup_cfg): Replace declaration with that of...
3139         (make_pass_fixup_cfg): ...new function.
3140         (pass_expand): Replace declaration with that of...
3141         (make_pass_expand): ...new function.
3142         (pass_instantiate_virtual_regs): Replace declaration with that of...
3143         (make_pass_instantiate_virtual_regs): ...new function.
3144         (pass_rtl_fwprop): Replace declaration with that of...
3145         (make_pass_rtl_fwprop): ...new function.
3146         (pass_rtl_fwprop_addr): Replace declaration with that of...
3147         (make_pass_rtl_fwprop_addr): ...new function.
3148         (pass_jump): Replace declaration with that of...
3149         (make_pass_jump): ...new function.
3150         (pass_jump2): Replace declaration with that of...
3151         (make_pass_jump2): ...new function.
3152         (pass_lower_subreg): Replace declaration with that of...
3153         (make_pass_lower_subreg): ...new function.
3154         (pass_cse): Replace declaration with that of...
3155         (make_pass_cse): ...new function.
3156         (pass_fast_rtl_dce): Replace declaration with that of...
3157         (make_pass_fast_rtl_dce): ...new function.
3158         (pass_ud_rtl_dce): Replace declaration with that of...
3159         (make_pass_ud_rtl_dce): ...new function.
3160         (pass_rtl_dce): Replace declaration with that of...
3161         (make_pass_rtl_dce): ...new function.
3162         (pass_rtl_dse1): Replace declaration with that of...
3163         (make_pass_rtl_dse1): ...new function.
3164         (pass_rtl_dse2): Replace declaration with that of...
3165         (make_pass_rtl_dse2): ...new function.
3166         (pass_rtl_dse3): Replace declaration with that of...
3167         (make_pass_rtl_dse3): ...new function.
3168         (pass_rtl_cprop): Replace declaration with that of...
3169         (make_pass_rtl_cprop): ...new function.
3170         (pass_rtl_pre): Replace declaration with that of...
3171         (make_pass_rtl_pre): ...new function.
3172         (pass_rtl_hoist): Replace declaration with that of...
3173         (make_pass_rtl_hoist): ...new function.
3174         (pass_rtl_store_motion): Replace declaration with that of...
3175         (make_pass_rtl_store_motion): ...new function.
3176         (pass_cse_after_global_opts): Replace declaration with that of...
3177         (make_pass_cse_after_global_opts): ...new function.
3178         (pass_rtl_ifcvt): Replace declaration with that of...
3179         (make_pass_rtl_ifcvt): ...new function.
3180         (pass_into_cfg_layout_mode): Replace declaration with that of...
3181         (make_pass_into_cfg_layout_mode): ...new function.
3182         (pass_outof_cfg_layout_mode): Replace declaration with that of...
3183         (make_pass_outof_cfg_layout_mode): ...new function.
3184         (pass_loop2): Replace declaration with that of...
3185         (make_pass_loop2): ...new function.
3186         (pass_rtl_loop_init): Replace declaration with that of...
3187         (make_pass_rtl_loop_init): ...new function.
3188         (pass_rtl_move_loop_invariants): Replace declaration with that of...
3189         (make_pass_rtl_move_loop_invariants): ...new function.
3190         (pass_rtl_unswitch): Replace declaration with that of...
3191         (make_pass_rtl_unswitch): ...new function.
3192         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3193         (make_pass_rtl_unroll_and_peel_loops): ...new function.
3194         (pass_rtl_doloop): Replace declaration with that of...
3195         (make_pass_rtl_doloop): ...new function.
3196         (pass_rtl_loop_done): Replace declaration with that of...
3197         (make_pass_rtl_loop_done): ...new function.
3198         (pass_web): Replace declaration with that of...
3199         (make_pass_web): ...new function.
3200         (pass_cse2): Replace declaration with that of...
3201         (make_pass_cse2): ...new function.
3202         (pass_df_initialize_opt): Replace declaration with that of...
3203         (make_pass_df_initialize_opt): ...new function.
3204         (pass_df_initialize_no_opt): Replace declaration with that of...
3205         (make_pass_df_initialize_no_opt): ...new function.
3206         (pass_reginfo_init): Replace declaration with that of...
3207         (make_pass_reginfo_init): ...new function.
3208         (pass_inc_dec): Replace declaration with that of...
3209         (make_pass_inc_dec): ...new function.
3210         (pass_stack_ptr_mod): Replace declaration with that of...
3211         (make_pass_stack_ptr_mod): ...new function.
3212         (pass_initialize_regs): Replace declaration with that of...
3213         (make_pass_initialize_regs): ...new function.
3214         (pass_combine): Replace declaration with that of...
3215         (make_pass_combine): ...new function.
3216         (pass_if_after_combine): Replace declaration with that of...
3217         (make_pass_if_after_combine): ...new function.
3218         (pass_ree): Replace declaration with that of...
3219         (make_pass_ree): ...new function.
3220         (pass_partition_blocks): Replace declaration with that of...
3221         (make_pass_partition_blocks): ...new function.
3222         (pass_match_asm_constraints): Replace declaration with that of...
3223         (make_pass_match_asm_constraints): ...new function.
3224         (pass_regmove): Replace declaration with that of...
3225         (make_pass_regmove): ...new function.
3226         (pass_split_all_insns): Replace declaration with that of...
3227         (make_pass_split_all_insns): ...new function.
3228         (pass_fast_rtl_byte_dce): Replace declaration with that of...
3229         (make_pass_fast_rtl_byte_dce): ...new function.
3230         (pass_lower_subreg2): Replace declaration with that of...
3231         (make_pass_lower_subreg2): ...new function.
3232         (pass_mode_switching): Replace declaration with that of...
3233         (make_pass_mode_switching): ...new function.
3234         (pass_sms): Replace declaration with that of...
3235         (make_pass_sms): ...new function.
3236         (pass_sched): Replace declaration with that of...
3237         (make_pass_sched): ...new function.
3238         (pass_ira): Replace declaration with that of...
3239         (make_pass_ira): ...new function.
3240         (pass_reload): Replace declaration with that of...
3241         (make_pass_reload): ...new function.
3242         (pass_clean_state): Replace declaration with that of...
3243         (make_pass_clean_state): ...new function.
3244         (pass_branch_prob): Replace declaration with that of...
3245         (make_pass_branch_prob): ...new function.
3246         (pass_value_profile_transformations): Replace declaration with that
3247         of...
3248         (make_pass_value_profile_transformations): ...new function.
3249         (pass_postreload_cse): Replace declaration with that of...
3250         (make_pass_postreload_cse): ...new function.
3251         (pass_gcse2): Replace declaration with that of...
3252         (make_pass_gcse2): ...new function.
3253         (pass_split_after_reload): Replace declaration with that of...
3254         (make_pass_split_after_reload): ...new function.
3255         (pass_branch_target_load_optimize1): Replace declaration with that
3256         of...
3257         (make_pass_branch_target_load_optimize1): ...new function.
3258         (pass_thread_prologue_and_epilogue): Replace declaration with that
3259         of...
3260         (make_pass_thread_prologue_and_epilogue): ...new function.
3261         (pass_stack_adjustments): Replace declaration with that of...
3262         (make_pass_stack_adjustments): ...new function.
3263         (pass_peephole2): Replace declaration with that of...
3264         (make_pass_peephole2): ...new function.
3265         (pass_if_after_reload): Replace declaration with that of...
3266         (make_pass_if_after_reload): ...new function.
3267         (pass_regrename): Replace declaration with that of...
3268         (make_pass_regrename): ...new function.
3269         (pass_cprop_hardreg): Replace declaration with that of...
3270         (make_pass_cprop_hardreg): ...new function.
3271         (pass_reorder_blocks): Replace declaration with that of...
3272         (make_pass_reorder_blocks): ...new function.
3273         (pass_branch_target_load_optimize2): Replace declaration with that
3274         of...
3275         (make_pass_branch_target_load_optimize2): ...new function.
3276         (pass_leaf_regs): Replace declaration with that of...
3277         (make_pass_leaf_regs): ...new function.
3278         (pass_split_before_sched2): Replace declaration with that of...
3279         (make_pass_split_before_sched2): ...new function.
3280         (pass_compare_elim_after_reload): Replace declaration with that of...
3281         (make_pass_compare_elim_after_reload): ...new function.
3282         (pass_sched2): Replace declaration with that of...
3283         (make_pass_sched2): ...new function.
3284         (pass_stack_regs): Replace declaration with that of...
3285         (make_pass_stack_regs): ...new function.
3286         (pass_stack_regs_run): Replace declaration with that of...
3287         (make_pass_stack_regs_run): ...new function.
3288         (pass_df_finish): Replace declaration with that of...
3289         (make_pass_df_finish): ...new function.
3290         (pass_compute_alignments): Replace declaration with that of...
3291         (make_pass_compute_alignments): ...new function.
3292         (pass_duplicate_computed_gotos): Replace declaration with that of...
3293         (make_pass_duplicate_computed_gotos): ...new function.
3294         (pass_variable_tracking): Replace declaration with that of...
3295         (make_pass_variable_tracking): ...new function.
3296         (pass_free_cfg): Replace declaration with that of...
3297         (make_pass_free_cfg): ...new function.
3298         (pass_machine_reorg): Replace declaration with that of...
3299         (make_pass_machine_reorg): ...new function.
3300         (pass_cleanup_barriers): Replace declaration with that of...
3301         (make_pass_cleanup_barriers): ...new function.
3302         (pass_delay_slots): Replace declaration with that of...
3303         (make_pass_delay_slots): ...new function.
3304         (pass_split_for_shorten_branches): Replace declaration with that of...
3305         (make_pass_split_for_shorten_branches): ...new function.
3306         (pass_split_before_regstack): Replace declaration with that of...
3307         (make_pass_split_before_regstack): ...new function.
3308         (pass_convert_to_eh_region_ranges): Replace declaration with that
3309         of...
3310         (make_pass_convert_to_eh_region_ranges): ...new function.
3311         (pass_shorten_branches): Replace declaration with that of...
3312         (make_pass_shorten_branches): ...new function.
3313         (pass_set_nothrow_function_flags): Replace declaration with that of...
3314         (make_pass_set_nothrow_function_flags): ...new function.
3315         (pass_dwarf2_frame): Replace declaration with that of...
3316         (make_pass_dwarf2_frame): ...new function.
3317         (pass_final): Replace declaration with that of...
3318         (make_pass_final): ...new function.
3319         (pass_rtl_seqabstr): Replace declaration with that of...
3320         (make_pass_rtl_seqabstr): ...new function.
3321         (pass_release_ssa_names): Replace declaration with that of...
3322         (make_pass_release_ssa_names): ...new function.
3323         (pass_early_inline): Replace declaration with that of...
3324         (make_pass_early_inline): ...new function.
3325         (pass_inline_parameters): Replace declaration with that of...
3326         (make_pass_inline_parameters): ...new function.
3327         (pass_update_address_taken): Replace declaration with that of...
3328         (make_pass_update_address_taken): ...new function.
3329         (pass_convert_switch): Replace declaration with that of...
3330         (make_pass_convert_switch): ...new function.
3331         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3332         to a subclass of simple_ipa_opt_pass along with...
3333         (pass_data_ipa_tree_profile): ...new pass_data instance and...
3334         (make_pass_ipa_tree_profile): ...new function.
3335         * tree-sra.c (pass_sra_early): Convert from a global struct to a
3336         subclass of gimple_opt_pass along with...
3337         (pass_data_sra_early): ...new pass_data instance and...
3338         (make_pass_sra_early): ...new function.
3339         (pass_sra): Convert from a global struct to a subclass of
3340         gimple_opt_pass along with...
3341         (pass_data_sra): ...new pass_data instance and...
3342         (make_pass_sra): ...new function.
3343         (pass_early_ipa_sra): Convert from a global struct to a subclass of
3344         gimple_opt_pass along with...
3345         (pass_data_early_ipa_sra): ...new pass_data instance and...
3346         (make_pass_early_ipa_sra): ...new function.
3347         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3348         subclass of gimple_opt_pass along with...
3349         (pass_data_ccp): ...new pass_data instance and...
3350         (make_pass_ccp): ...new function.
3351         (pass_fold_builtins): Convert from a global struct to a subclass of
3352         gimple_opt_pass along with...
3353         (pass_data_fold_builtins): ...new pass_data instance and...
3354         (make_pass_fold_builtins): ...new function.
3355         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3356         subclass of gimple_opt_pass along with...
3357         (pass_data_copy_prop): ...new pass_data instance and...
3358         (make_pass_copy_prop): ...new function.
3359         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3360         global struct to a subclass of gimple_opt_pass along with...
3361         (pass_data_rename_ssa_copies): ...new pass_data instance and...
3362         (make_pass_rename_ssa_copies): ...new function.
3363         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3364         subclass of gimple_opt_pass along with...
3365         (pass_data_dce): ...new pass_data instance and...
3366         (make_pass_dce): ...new function.
3367         (pass_dce_loop): Convert from a global struct to a subclass of
3368         gimple_opt_pass along with...
3369         (pass_data_dce_loop): ...new pass_data instance and...
3370         (make_pass_dce_loop): ...new function.
3371         (pass_cd_dce): Convert from a global struct to a subclass of
3372         gimple_opt_pass along with...
3373         (pass_data_cd_dce): ...new pass_data instance and...
3374         (make_pass_cd_dce): ...new function.
3375         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3376         subclass of gimple_opt_pass along with...
3377         (pass_data_dominator): ...new pass_data instance and...
3378         (make_pass_dominator): ...new function.
3379         (pass_phi_only_cprop): Convert from a global struct to a subclass of
3380         gimple_opt_pass along with...
3381         (pass_data_phi_only_cprop): ...new pass_data instance and...
3382         (make_pass_phi_only_cprop): ...new function.
3383         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3384         subclass of gimple_opt_pass along with...
3385         (pass_data_dse): ...new pass_data instance and...
3386         (make_pass_dse): ...new function.
3387         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3388         a subclass of gimple_opt_pass along with...
3389         (pass_data_forwprop): ...new pass_data instance and...
3390         (make_pass_forwprop): ...new function.
3391         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3392         struct to a subclass of gimple_opt_pass along with...
3393         (pass_data_tree_ifcombine): ...new pass_data instance and...
3394         (make_pass_tree_ifcombine): ...new function.
3395         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3396         subclass of gimple_opt_pass along with...
3397         (pass_data_ch): ...new pass_data instance and...
3398         (make_pass_ch): ...new function.
3399         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3400         subclass of gimple_opt_pass along with...
3401         (pass_data_tree_loop): ...new pass_data instance and...
3402         (make_pass_tree_loop): ...new function.
3403         (pass_tree_loop_init): Convert from a global struct to a subclass of
3404         gimple_opt_pass along with...
3405         (pass_data_tree_loop_init): ...new pass_data instance and...
3406         (make_pass_tree_loop_init): ...new function.
3407         (pass_lim): Convert from a global struct to a subclass of
3408         gimple_opt_pass along with...
3409         (pass_data_lim): ...new pass_data instance and...
3410         (make_pass_lim): ...new function.
3411         (pass_tree_unswitch): Convert from a global struct to a subclass of
3412         gimple_opt_pass along with...
3413         (pass_data_tree_unswitch): ...new pass_data instance and...
3414         (make_pass_tree_unswitch): ...new function.
3415         (pass_predcom): Convert from a global struct to a subclass of
3416         gimple_opt_pass along with...
3417         (pass_data_predcom): ...new pass_data instance and...
3418         (make_pass_predcom): ...new function.
3419         (pass_vectorize): Convert from a global struct to a subclass of
3420         gimple_opt_pass along with...
3421         (pass_data_vectorize): ...new pass_data instance and...
3422         (make_pass_vectorize): ...new function.
3423         (pass_graphite): Convert from a global struct to a subclass of
3424         gimple_opt_pass along with...
3425         (pass_data_graphite): ...new pass_data instance and...
3426         (make_pass_graphite): ...new function.
3427         (pass_graphite_transforms): Convert from a global struct to a subclass
3428         of gimple_opt_pass along with...
3429         (pass_data_graphite_transforms): ...new pass_data instance and...
3430         (make_pass_graphite_transforms): ...new function.
3431         (pass_check_data_deps): Convert from a global struct to a subclass of
3432         gimple_opt_pass along with...
3433         (pass_data_check_data_deps): ...new pass_data instance and...
3434         (make_pass_check_data_deps): ...new function.
3435         (pass_iv_canon): Convert from a global struct to a subclass of
3436         gimple_opt_pass along with...
3437         (pass_data_iv_canon): ...new pass_data instance and...
3438         (make_pass_iv_canon): ...new function.
3439         (pass_scev_cprop): Convert from a global struct to a subclass of
3440         gimple_opt_pass along with...
3441         (pass_data_scev_cprop): ...new pass_data instance and...
3442         (make_pass_scev_cprop): ...new function.
3443         (pass_record_bounds): Convert from a global struct to a subclass of
3444         gimple_opt_pass along with...
3445         (pass_data_record_bounds): ...new pass_data instance and...
3446         (make_pass_record_bounds): ...new function.
3447         (pass_complete_unroll): Convert from a global struct to a subclass of
3448         gimple_opt_pass along with...
3449         (pass_data_complete_unroll): ...new pass_data instance and...
3450         (make_pass_complete_unroll): ...new function.
3451         (pass_complete_unrolli): Convert from a global struct to a subclass of
3452         gimple_opt_pass along with...
3453         (pass_data_complete_unrolli): ...new pass_data instance and...
3454         (make_pass_complete_unrolli): ...new function.
3455         (pass_parallelize_loops): Convert from a global struct to a subclass
3456         of gimple_opt_pass along with...
3457         (pass_data_parallelize_loops): ...new pass_data instance and...
3458         (make_pass_parallelize_loops): ...new function.
3459         (pass_loop_prefetch): Convert from a global struct to a subclass of
3460         gimple_opt_pass along with...
3461         (pass_data_loop_prefetch): ...new pass_data instance and...
3462         (make_pass_loop_prefetch): ...new function.
3463         (pass_iv_optimize): Convert from a global struct to a subclass of
3464         gimple_opt_pass along with...
3465         (pass_data_iv_optimize): ...new pass_data instance and...
3466         (make_pass_iv_optimize): ...new function.
3467         (pass_tree_loop_done): Convert from a global struct to a subclass of
3468         gimple_opt_pass along with...
3469         (pass_data_tree_loop_done): ...new pass_data instance and...
3470         (make_pass_tree_loop_done): ...new function.
3471         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3472         struct to a subclass of gimple_opt_pass along with...
3473         (pass_data_cse_reciprocals): ...new pass_data instance and...
3474         (make_pass_cse_reciprocals): ...new function.
3475         (pass_cse_sincos): Convert from a global struct to a subclass of
3476         gimple_opt_pass along with...
3477         (pass_data_cse_sincos): ...new pass_data instance and...
3478         (make_pass_cse_sincos): ...new function.
3479         (pass_optimize_bswap): Convert from a global struct to a subclass of
3480         gimple_opt_pass along with...
3481         (pass_data_optimize_bswap): ...new pass_data instance and...
3482         (make_pass_optimize_bswap): ...new function.
3483         (pass_optimize_widening_mul): Convert from a global struct to a
3484         subclass of gimple_opt_pass along with...
3485         (pass_data_optimize_widening_mul): ...new pass_data instance and...
3486         (make_pass_optimize_widening_mul): ...new function.
3487         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3488         subclass of gimple_opt_pass along with...
3489         (pass_data_phiopt): ...new pass_data instance and...
3490         (make_pass_phiopt): ...new function.
3491         (pass_cselim): Convert from a global struct to a subclass of
3492         gimple_opt_pass along with...
3493         (pass_data_cselim): ...new pass_data instance and...
3494         (make_pass_cselim): ...new function.
3495         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3496         subclass of gimple_opt_pass along with...
3497         (pass_data_phiprop): ...new pass_data instance and...
3498         (make_pass_phiprop): ...new function.
3499         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3500         subclass of gimple_opt_pass along with...
3501         (pass_data_pre): ...new pass_data instance and...
3502         (make_pass_pre): ...new function.
3503         (pass_fre): Convert from a global struct to a subclass of
3504         gimple_opt_pass along with...
3505         (pass_data_fre): ...new pass_data instance and...
3506         (make_pass_fre): ...new function.
3507         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3508         subclass of gimple_opt_pass along with...
3509         (pass_data_reassoc): ...new pass_data instance and...
3510         (make_pass_reassoc): ...new function.
3511         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3512         subclass of gimple_opt_pass along with...
3513         (pass_data_sink_code): ...new pass_data instance and...
3514         (make_pass_sink_code): ...new function.
3515         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3516         subclass of gimple_opt_pass along with...
3517         (pass_data_strlen): ...new pass_data instance and...
3518         (make_pass_strlen): ...new function.
3519         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3520         struct to a subclass of gimple_opt_pass along with...
3521         (pass_data_build_alias): ...new pass_data instance and...
3522         (make_pass_build_alias): ...new function.
3523         (pass_build_ealias): Convert from a global struct to a subclass of
3524         gimple_opt_pass along with...
3525         (pass_data_build_ealias): ...new pass_data instance and...
3526         (make_pass_build_ealias): ...new function.
3527         (pass_ipa_pta): Convert from a global struct to a subclass of
3528         simple_ipa_opt_pass along with...
3529         (pass_data_ipa_pta): ...new pass_data instance and...
3530         (make_pass_ipa_pta): ...new function.
3531         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3532         subclass of gimple_opt_pass along with...
3533         (pass_data_uncprop): ...new pass_data instance and...
3534         (make_pass_uncprop): ...new function.
3535         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3536         global struct to a subclass of gimple_opt_pass along with...
3537         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3538         (make_pass_late_warn_uninitialized): ...new function.
3539         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3540         to a subclass of gimple_opt_pass along with...
3541         (pass_data_init_datastructures): ...new pass_data instance and...
3542         (make_pass_init_datastructures): ...new function.
3543         (pass_early_warn_uninitialized): Convert from a global struct to a
3544         subclass of gimple_opt_pass along with...
3545         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3546         (make_pass_early_warn_uninitialized): ...new function.
3547         (pass_update_address_taken): Convert from a global struct to a
3548         subclass of gimple_opt_pass along with...
3549         (pass_data_update_address_taken): ...new pass_data instance and...
3550         (make_pass_update_address_taken): ...new function.
3551         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3552         struct to a subclass of gimple_opt_pass along with...
3553         (pass_data_release_ssa_names): ...new pass_data instance and...
3554         (make_pass_release_ssa_names): ...new function.
3555         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3556         subclass of gimple_opt_pass along with...
3557         (pass_data_stdarg): ...new pass_data instance and...
3558         (make_pass_stdarg): ...new function.
3559         * tree-switch-conversion.c (pass_convert_switch): Convert from a
3560         global struct to a subclass of gimple_opt_pass along with...
3561         (pass_data_convert_switch): ...new pass_data instance and...
3562         (make_pass_convert_switch): ...new function.
3563         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3564         to a subclass of gimple_opt_pass along with...
3565         (pass_data_tail_recursion): ...new pass_data instance and...
3566         (make_pass_tail_recursion): ...new function.
3567         (pass_tail_calls): Convert from a global struct to a subclass of
3568         gimple_opt_pass along with...
3569         (pass_data_tail_calls): ...new pass_data instance and...
3570         (make_pass_tail_calls): ...new function.
3571         * tree-vect-generic.c (pass_lower_vector): Convert from a global
3572         struct to a subclass of gimple_opt_pass along with...
3573         (pass_data_lower_vector): ...new pass_data instance and...
3574         (make_pass_lower_vector): ...new function.
3575         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3576         gimple_opt_pass along with...
3577         (pass_data_lower_vector_ssa): ...new pass_data instance and...
3578         (make_pass_lower_vector_ssa): ...new function.
3579         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3580         to a subclass of gimple_opt_pass along with...
3581         (pass_data_slp_vectorize): ...new pass_data instance and...
3582         (make_pass_slp_vectorize): ...new function.
3583         (pass_ipa_increase_alignment): Convert from a global struct to a
3584         subclass of simple_ipa_opt_pass along with...
3585         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3586         (make_pass_ipa_increase_alignment): ...new function.
3587         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3588         gimple_opt_pass along with...
3589         (pass_data_vrp): ...new pass_data instance and...
3590         (make_pass_vrp): ...new function.
3591         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3592         subclass of simple_ipa_opt_pass along with...
3593         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3594         (make_pass_ipa_free_lang_data): ...new function.
3595         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3596         gimple_opt_pass along with...
3597         (pass_data_tsan): ...new pass_data instance and...
3598         (make_pass_tsan): ...new function.
3599         (pass_tsan_O0): Convert from a global struct to a subclass of
3600         gimple_opt_pass along with...
3601         (pass_data_tsan_O0): ...new pass_data instance and...
3602         (make_pass_tsan_O0): ...new function.
3603         * var-tracking.c (pass_variable_tracking): Convert from a global
3604         struct to a subclass of rtl_opt_pass along with...
3605         (pass_data_variable_tracking): ...new pass_data instance and...
3606         (make_pass_variable_tracking): ...new function.
3607         * web.c (pass_web): Convert from a global struct to a subclass of
3608         rtl_opt_pass along with...
3609         (pass_data_web): ...new pass_data instance and...
3610         (make_pass_web): ...new function.
3611         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3612         declaration with that of...
3613         (make_pass_mode_switch_use): ...new function.
3614         (pass_resolve_sw_modes): Replace declaration with that of...
3615         (make_pass_resolve_sw_modes): ...new function.
3616         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3617         from a global struct to a subclass of rtl_opt_pass along with...
3618         (pass_data_mode_switch_use): ...new pass_data instance and...
3619         (make_pass_mode_switch_use): ...new function.
3620         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3621         from a global struct to a subclass of rtl_opt_pass along with...
3622         (pass_data_resolve_sw_modes): ...new pass_data instance and...
3623         (make_pass_resolve_sw_modes): ...new function.
3624         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3625         struct to a subclass of rtl_opt_pass along with...
3626         (pass_data_insert_vzeroupper): ...new pass_data instance and...
3627         (make_pass_insert_vzeroupper): ...new function.
3628         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3629         global struct to a subclass of rtl_opt_pass along with...
3630         (pass_data_work_around_errata): ...new pass_data instance and...
3631         (make_pass_work_around_errata): ...new function.
3632         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3633         struct to a subclass of rtl_opt_pass along with...
3634         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3635         (make_pass_mips_machine_reorg2): ...new function.
3637 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3639         * passes.c (pass_manager::operator new): New.
3641 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3643         Handwritten part of conversion of passes to C++ classes.
3645         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3646         (toplev.o): Add dep on PASS_MANAGER_H.
3647         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3648         of early local pases to reflect this moving from a global to a
3649         member of gcc::pass_manager.
3650         (cgraph_add_new_function): Likewise.
3651         * lto-cgraph.c (lto_output_node): Update for conversion of
3652         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3653         * passes.c (opt_pass::clone): New.
3654         (opt_pass::gate): New.
3655         (opt_pass::execute): New.
3656         (opt_pass::opt_pass): New.
3657         (pass_manager::execute_early_local_passes): New.
3658         (pass_manager::execute_pass_mode_switching): new.
3659         (finish_optimization_passes): Convert to...
3660         (pass_manager::finish_optimization_passes): ...this.
3661         (finish_optimization_passes): Update for conversion of passes to
3662         C++ classes.
3663         (register_dump_files_1): Use has_gate since we cannot portably
3664         check a vtable entry against NULL.
3665         (dump_one_pass): Likewise.
3666         (ipa_write_summaries_2): Likewise.
3667         (ipa_write_optimization_summaries_1): Likewise.
3668         (ipa_read_summaries_1): Likewise.
3669         (ipa_read_optimization_summaries_1): Likewise.
3670         (execute_ipa_stmt_fixups): Likewise.
3671         (pass_manager::pass_manager): Rewrite pass-creation, invoking
3672         pass-creation functions rather than wiring up globals, and
3673         storing the results in fields of pass_manager generated using
3674         pass-instances.def.
3675         (pass_manager::dump_profile_report): Update for conversion of
3676         passes to C++ classes.
3677         (pass_manager::execute_ipa_summary_passes): Likewise.
3678         (execute_one_ipa_transform_pass): Likewise.
3679         (execute_one_pass): Use has_gate and has_execute since we cannot
3680         portably check a vtable entry against NULL.
3681         * pass_manager.h (pass_manager::finish_optimization_passes): New.
3682         (pass_manager): Use pass-instances.def to add fields for the
3683         various pass instances.
3684         * toplev.c (finalize): Update for move of
3685         finish_optimization_passes to a method of gcc::pass_manager.
3686         * toplev.h (finish_optimization_passes): Move to method of class
3687         pass_manager.
3688         * tree-pass.h (struct pass_data): New.
3689         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3690         (opt_pass::gate): Convert to virtual function.
3691         (opt_pass::~opt_pass): New.
3692         (opt_pass::clone): New.
3693         (opt_pass::execute): Convert to virtual function.
3694         (opt_pass::opt_pass): New.
3695         (opt_pass::ctxt_): new.
3696         (gimple_opt_pass): Convert to subclass of opt_pass.
3697         (gimple_opt_pass::gimple_opt_pass): New.
3698         (rtl_opt_pass): Convert to subclass of opt_pass.
3699         (rtl_opt_pass::rtl_opt_pass): New.
3700         (ipa_opt_pass_d): Convert to subclass of opt_pass.
3701         (ipa_opt_pass_d::ipa_opt_pass_d): New.
3702         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3703         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3704         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3705         invocation of pass_mode_switching to reflect this moving from a
3706         global to a member of gcc::pass_manager.
3707         (ix86_option_override): Rework how pass_insert_vzeroupper is
3708         added to the pass_manager to reflect autogenerated changes.
3709         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3711 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
3713         PR rtl-optimization/57708
3714         * recog.c (peep2_find_free_register): Validate all regs in a
3715         multi-reg mode.
3717 2013-08-05  Jan Hubicka  <jh@suse.cz>
3719         PR lto/57602
3720         * cgraph.c (verify_cgraph_node): Accept local flags from other
3721         partitions.
3722         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3723         (function_and_variable_visibility): Likewise.
3724         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3726 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3728         * graph.c (init_graph_slim_pretty_print): Remove.
3729         (print_graph_cfg): Do not call it.  Use local pretty printer.
3730         (start_graph_dump): Likewise.
3732 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3734         * gimple-pretty-print.c (buffer): Remove.
3735         (initialized): Likewise.
3736         (maybe_init_pretty_print): Likewise.
3737         (print_gimple_stmt): Do not call it.  Use non-static local
3738         pretty_printer variable.
3739         (print_gimple_expr): Likewise.
3740         (print_gimple_seq): Likewise.
3741         (gimple_dump_bb): Likewise.
3743 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3745         * asan.c (asan_pp): Remove.
3746         (asan_pp_initialized): Likewise.
3747         (asan_pp_initialize): Likewise.
3748         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
3749         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
3750         (asan_add_global): Likewise.
3752 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3754         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
3755         * diagnostic.h (diagnostic_flush_buffer): Adjust.
3756         * pretty-print.c (pp_formatted_text_data): Likewise.
3757         (pp_indent): Rename from pp_base_indent.
3758         (pp_format): Rename from pp_base_format.
3759         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3760         (pp_format_verbatim): Rename from pp_base_format_verbatim.
3761         (pp_flush): Rename from pp_base_flush.
3762         (pp_set_line_maximum_length): Rename from
3763         pp_base_set_line_maximum_length.
3764         (pp_clear_output_area): Rename from pp_base_clear_output_area.
3765         (pp_set_prefix): Rename from pp_base_set_prefix.
3766         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3767         (pp_emit_prefix): Rename from pp_base_emit_prefix.
3768         (pp_append_text): Rename from pp_base_append_text.
3769         (pp_formatted_text): Rename from pp_base_formatted_text.
3770         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3771         (pp_remaining_character_count_for_line): Rename from
3772         pp_base_remaining_character_count_for_line.
3773         (pp_newline): Rename from pp_base_newline.
3774         (pp_character): Rename from pp_base_character.
3775         (pp_string): Rename from pp_base_string.
3776         (pp_maybe_space): Rename from pp_base_maybe_space.
3777         * asan.c (asan_pp_string): Adjust.
3778         (asan_emit_stack_protection): Likewise.
3779         (asan_add_global): Likewise.
3780         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3781         * tree-mudflap.c (mf_varname_tree): Likewise.
3782         * tree-pretty-print.c (pp_tree_identifier): Rename from
3783         pp_base_tree_identifier.
3784         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3785         Declare as function.
3787 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3789         * pretty-print.h (pp_bar_bar): New.
3790         (pp_ampersand_ampersand): Likewise.
3791         (pp_less_equal): Likewise.
3792         (pp_greater_equal): Likewise.
3793         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3794         printer functions instead of pp_string or operators and punctuators.
3795         (dump_gimple_call): Likewise.
3796         (dump_gimple_omp_for): Likewise.
3797         (dump_gimple_transaction): Likewise.
3798         (dump_gimple_phi): Likewise.
3799         (pp_gimple_stmt_1): Likewise.
3800         * sched-vis.c (print_insn): Likewise.
3801         * tree-mudflap.c (mf_varname_tree): Likewise.
3802         * tree-pretty-print.c (dump_block_node): Likewise.
3803         (dump_generic_node): Likewise.
3805 2013-08-02  Jan Hubicka  <jh@suse.cz>
3807         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3808         boundaries.
3809         * lto-streamer-out.c (tree_is_indexable): Results decls and
3810         parm decls are not indexable.
3811         (DFS_write_tree_body): Do not follow args and results.
3812         (hash_tree): Likewise.
3813         (output_functions): Rearrange so struct function is needed
3814         only when real body is output; be able to also ouptut abstract
3815         functions; output DECL_ARGUMENTS and DECL_RESULT.
3816         (lto_output): When not in WPA, ale store abstract functions.
3817         (write_symbol): Do not care about RESULT_DECL.
3818         (output_symbol_p): Handle correctly sbtract decls.
3819         * lto-streamer-in.c (input_function): Rearrange so struct
3820         function can be NULL at entry; allow streaming of
3821         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3822         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3823         sanity check during LTO.
3824         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3825         RESULT_DECl and DECL_ARGUMENTS.
3826         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3827         Likewise.
3829 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3831         * pretty-print.h (pp_underscore): New.
3832         (pp_comma): Tidy.
3833         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3834         printer functions instead of pp_character.
3835         (dump_binary_rhs): Likewise.
3836         (dump_ternary_rhs): Likewise.
3837         (dump_gimple_call_args): Likewise.
3838         (pp_points_to_solution): Likewise.
3839         (dump_gimple_call): Likewise.
3840         (dump_gimple_switch): Likewise.
3841         (dump_gimple_cond): Likewise.
3842         (dump_gimple_bind): Likewise.
3843         (dump_gimple_try): Likewise.
3844         (dump_gimple_omp_for): Likewise.
3845         (dump_gimple_omp_continue): Likewise.
3846         (dump_gimple_omp_single): Likewise.
3847         (dump_gimple_omp_sections): Likewise.
3848         (dump_gimple_omp_block): Likewise.
3849         (dump_gimple_omp_critical): Likewise.
3850         (dump_gimple_transaction): Likewise.
3851         (dump_gimple_asm): Likewise.
3852         (dump_gimple_phi): Likewise.
3853         (dump_gimple_omp_parallel): Likewise.
3854         (dump_gimple_omp_task): Likewise.
3855         (dump_gimple_omp_atomic_load): Likewise.
3856         (dump_gimple_omp_atomic_store): Likewise.
3857         (dump_gimple_mem_ops): Likewise.
3858         (pp_gimple_stmt_1): Likewise.
3859         (pp_cfg_jump): Likewise.
3860         (dump_implicit_edges): Likewise.
3861         (gimple_dump_bb_for_graph): Likewise.
3862         * graph.c (draw_cfg_node): Likewise.
3863         * langhooks.c (lhd_print_error_function): Likewise.
3864         * sched-vis.c (print_exp): Likewise.
3865         (print_value): Likewise.
3866         (print_pattern): Likewise.
3867         (print_insn): Likewise.
3868         (rtl_dump_bb_for_graph): Likewise.
3869         * tree-pretty-print.c (dump_function_declaration): Likewise.
3870         (dump_array_domain): Likewise.
3871         (dump_omp_clause): Likewise.
3872         (dump_location): Likewise.
3873         (dump_generic_node): Likewise.
3874         (print_struct_decl): Likewise.
3875         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3877 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3879         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3880         candidate table when replacing a candidate statement.
3881         (replace_rhs_if_not_dup): Likewise.
3882         (replace_one_candidate): Likewise.
3884 2013-08-02  Jan Hubicka  <jh@suse.cz>
3885             Martin Liska  <marxin.liska@gmail.com>
3887         * cgraphunit.c (add_new_function): Fix logic when adding from
3888         late IPA pass.
3889         (assemble_thunk): Rename to ...
3890         (expand_thunk); .. this one; export; get it working with
3891         general functions; make produced gimple valid.
3892         * cgraph.h (expand_thunk): Declare.
3894 2013-08-02  Jan Hubicka  <jh@suse.cz>
3896         * ipa-cp.c (gather_context_independent_values): Use
3897         ipa_get_param_move_cost.
3898         (get_replacement_map): Remove PARAM; move parameter folding
3899         into tree-inline.c
3900         (create_specialized_node): Update.
3901         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3902         assert that we have gimple body; update move_cost.
3903         (count_formal_params): Assert that we have gimple body.
3904         (ipa_dump_param): New function.
3905         (ipa_alloc_node_params): Break out from ...
3906         (ipa_initialize_node_params): ... here.
3907         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3908         (ipa_write_node_info): Stream move costs.
3909         (ipa_read_node_info): Read move costs.
3910         (ipa_update_after_lto_read): Do not recompute node params.
3911         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3912         (ipa_get_param): Check we are not in WPA.
3913         (ipa_get_param_move_cost): New.
3914         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3915         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3916         parm numbers to be present.
3918 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3920         PR rtl-optimization/58048
3921         * lra-constraints.c (process_alt_operands): Don't check asm
3922         operand on register.
3924 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3926         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3927         the implied StoreLoad barrier for atomic operations if before.
3929 2013-08-02  Jan Hubicka  <jh@suse.cz>
3930             Martin Liska  <marxin.liska@gmail.com>
3932         * cgraph.c (cgraph_function_body_availability): Do not check
3933         cgraph flags.
3934         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3935         symtab_node_availability): Declare.
3936         * ipa.c (can_replace_by_local_alias): New.
3937         (function_and_variable_visibility): Use it.
3938         * symtab.c (symtab_for_node_and_aliases,
3939         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3941 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3943         PR rtl-optimization/57963
3944         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3945         (lra_constraints): Use them.
3947 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3949         * config/arm/types.md (define_attr "type"): Add "load_acq"
3950         and "store_rel".
3951         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3952         changes.
3953         (cortex_a53_store1): Likewise.
3955 2013-08-01  Jan Hubicka  <jh@suse.cz>
3957         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3958         partitions are not needed.
3960 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3962         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3963         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3964         MAYBE_NON_Q_CLASS_P where appropriate.
3966 2013-08-01  Jan Hubicka  <jh@suse.cz>
3968         * cgraph.h (release_function_body): Declare.
3969         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3970         of unused delcarations.
3972 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3974         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3975         RTL form when subtracting a constant.
3977 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3979         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3980         Generate canonical plus rtx with negated immediate instead of minus
3981         where appropriate.
3982         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3984 2013-08-01  Jan Hubicka  <jh@suse.cz>
3986         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3987         (cgraph_release_function_body): Likewise.
3988         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3989         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3990         to used_as_abstract_origin.
3991         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3992         symbols used as abstract origins.
3993         * cgraphunit.c (analyze_functions): Update.
3994         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3995         used_as_abstract_origin.
3996         * tree-inline.c (tree_function_versioning): Update
3997         used_as_abstract_origin; be ready for DECL_RESULT and
3998         DECL_ARGUMENTS to be NULL.
4000         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
4001         for abstract functions.
4002         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
4003         real symbols.
4005 2013-08-01  Jan Hubicka  <jh@suse.cz>
4007         * profile.c (compute_value_histograms): Fix thinko.
4009 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
4011         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
4012         aarch-common-protos.h to extra_headers.
4013         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4014         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
4015         * config/aarch64/t-aarch64 (aarch-common.o): Define.
4017 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
4019         * config/aarch64/aarch64.md (define_attr "type"): Delete.
4020         Include "../arm/types.md".  Define "type" attribute for all patterns.
4021         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
4022         attribute changes.
4024 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
4026         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
4027         to support power8 load fusion.
4028         (fusion_gpr_mem_load): Likewise.
4030         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
4032         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
4033         declarations for power8 load fusion.
4034         (emit_fusion_gpr_load): Likewise.
4036         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
4037         tuning for power8, turn on fusion mode by default.  Turn on sign
4038         extending fusion mode if normal fusion mode is on, and we are at
4039         -O2 or -O3.
4040         (fusion_gpr_load_p): New function, return true if we can fuse an
4041         addis instruction with a dependent load to a GPR.
4042         (emit_fusion_gpr_load): Emit the instructions for power8 load
4043         fusion to GPRs.
4045         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
4046         (VSX load fusion peepholes): New peepholes to fuse together an
4047         addi instruction with a VSX load instruction.
4049         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
4050         peepholes to fuse an addis instruction with a load to a GPR base
4051         register.  If we are supporting sign extending fusions, convert
4052         sign extending loads to zero extending loads and add an explicit
4053         sign extension.
4055 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
4057         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
4058         aarch-common-protos.h to extra_headers.
4059         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4060         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
4061         (arm_early_store_addr_dep): Likewise.
4062         (arm_no_early_alu_shift_dep): Likewise.
4063         (arm_no_early_alu_shift_value_dep): Likewise.
4064         (arm_no_early_mul_dep): Likewise.
4065         (arm_no_early_store_addr_dep): Likewise.
4066         (arm_mac_accumulator_is_mul_result): Likewise.
4067         (arm_mac_accumulator_is_result): Likewise.
4068         * config/arm/aarch-common.c: ... here.  New file.
4069         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
4070         here to ...
4071         (arm_early_store_addr_dep): Likewise.
4072         (arm_no_early_alu_shift_dep): Likewise.
4073         (arm_no_early_alu_shift_value_dep): Likewise.
4074         (arm_no_early_mul_dep): Likewise.
4075         (arm_no_early_store_addr_dep): Likewise.
4076         (arm_mac_accumulator_is_mul_result): Likewise.
4077         (arm_mac_accumulator_is_result): Likewise.
4078         * config/arm/aarch-common-protos.h: ... here.  New file.
4079         * config/arm/t-arm (aarch-common.o): Define.
4081 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
4083         * config/arm/arm.md: Include new file "types.md".
4084         (define_attr "type"): Move from here to ...
4085         (define_attr "mul32"): Likewise.
4086         (define_attr "mul64"): Likewise.
4087         * config/arm/types.md: ... here.  New file.
4089 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4091         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4092         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4094 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4096         * gen-pass-instances.awk: Fix offset of substr().
4098 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
4100         * Makefile.in (pass-instances.def): New.
4101         (passes.o): Replace dependency on passes.def with one on
4102         pass-instances.def
4104         * gen-pass-instances.awk: New.
4106         * passes.c (pass_manager::pass_manager): Use pass-instances.def
4107         rather than passes.def, updating local definition of NEXT_PASS
4108         macro to add an extra NUM parameter (currently unused).
4110 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
4112         * Makefile.in (PASS_MANAGER_H): New.
4113         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4114         (passes.o): Likewise.
4115         (statistics.o): Likewise.
4116         (cgraphunit.o): Likewise.
4117         (context.o): Depend on PASS_MANAGER_H.
4119         * pass_manager.h: New.
4121         * cgraphunit.c (cgraph_add_new_function): Update for moves
4122         of globals to fields of pass_manager.
4123         (analyze_function): Likewise.
4124         (expand_function): Likewise.
4125         (ipa_passes): Likewise.
4126         (compile): Likewise.
4128         * context.c (context::context): New.
4129         * context.h  (context::context): New.
4130         (context::get_passes): New.
4131         (context::passes_): New.
4133         * lto-cgraph.c (input_node): Update for moves of globals to
4134         fields of pass_manager.
4136         * passes.c (all_passes): Remove, in favor of a field of the
4137         same name within the new class pass_manager.
4138         (all_small_ipa_passes): Likewise.
4139         (all_lowering_passes): Likewise.
4140         (all_regular_ipa_passes): Likewise.
4141         (all_late_ipa_passes): Likewise.
4142         (all_lto_gen_passes): Likewise.
4143         (passes_by_id): Likewise.
4144         (passes_by_id_size): Likewise.
4145         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4146         the new class pass_manager.
4147         (set_pass_for_id): Convert to...
4148         (pass_manager::set_pass_for_id): ...method.
4149         (get_pass_for_id): Convert to...
4150         (pass_manager::get_pass_for_id): ...method.
4151         (register_one_dump_file): Move body of implementation into...
4152         (pass_manager::register_one_dump_file): ...here.
4153         (register_dump_files_1): Convert to...
4154         (pass_manager::register_dump_files_1): ...method.
4155         (register_dump_files): Convert to...
4156         (pass_manager::register_dump_files): ...method.
4157         (create_pass_tab): Update for moves of globals to fields of
4158         pass_manager.
4159         (dump_passes): Move body of implementation into...
4160         (pass_manager::dump_passes): ...here.
4161         (register_pass): Move body of implementation into...
4162         (pass_manager::register_pass): ...here.
4163         (init_optimization_passes): Convert into...
4164         (pass_manager::pass_manager): ...constructor for new
4165         pass_manager class, and initialize the pass_lists array.
4166         (check_profile_consistency): Update for moves of globals to
4167         fields of pass_manager.
4168         (dump_profile_report): Move body of implementation into...
4169         (pass_manager::dump_profile_report): ...here.
4170         (ipa_write_summaries_1): Update for moves of pass lists from
4171         being globals to fields of pass_manager.
4172         (ipa_write_optimization_summaries): Likewise.
4173         (ipa_read_summaries):  Likewise.
4174         (ipa_read_optimization_summaries): Likewise.
4175         (execute_all_ipa_stmt_fixups): Likewise.
4177         * statistics.c (statistics_fini): Update for moves of globals to
4178         fields of pass_manager.
4180         * toplev.c (general_init): Replace call to
4181         init_optimization_passes with construction of the pass_manager
4182         instance.
4184         * tree-pass.h (all_passes): Remove, in favor of a field of the
4185         same name within the new class pass_manager.
4186         (all_small_ipa_passes): Likewise.
4187         (all_lowering_passes): Likewise.
4188         (all_regular_ipa_passes): Likewise.
4189         (all_lto_gen_passes): Likewise.
4190         (all_late_ipa_passes): Likewise.
4191         (passes_by_id): Likewise.
4192         (passes_by_id_size): Likewise.
4193         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4194         the new class pass_manager.
4195         (get_pass_for_id): Remove.
4197 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4199         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4200         configs.
4202 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4204         * arm.md (mulhi3): New expand pattern.
4206 2013-07-30  Jan Hubicka  <jh@suse.cz>
4207             Martin Liska  <marxin.liska@gmail.com>
4209         * profile.c (compute_value_histograms): Do not ICE when
4210         there is mismatch only on some counters.
4212 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4214         PR rtl-optimization/57637
4215         * function.c (move_insn_for_shrink_wrap): Also check the
4216         GEN set of the LIVE problem for the liveness analysis
4217         if it exists, otherwise give up.
4219 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4221         PR tree-optimization/57993
4222         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4223         replaced statement in the candidate table.
4224         (phi_add_costs): Return infinite cost when the hidden basis does
4225         not dominate all phis on which the candidate is dependent.
4226         (replace_one_candidate): Record replaced statement in the
4227         candidate table.
4229 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
4231         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4232         (ashlv2si3): New expander.
4233         (*ashlv2si3_i): New define_insn_and_split.
4234         * predicates.md (float_operation): Allow patterns with three
4235         basic sub-patterns.
4237         PR rtl-optimization/58021
4238         * mode-switching.c (create_pre_exit): Always split off preceding
4239         insns if we are not at the basic block head.
4241 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
4243         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4244         (UCLIBC_DYNAMIC_LINKER): New macro.
4245         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4246         `-mnan=2008'.
4247         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4248         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
4249         `-mnan=2008'.
4250         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4251         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4252         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4253         for SF and DF modes.  Use ieee_quad_format for TF mode.
4254         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4255         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4256         (mips_option_override): Handle `-mnan=legacy'.
4257         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4258         `-mabs=2008' and `-mnan=2008'.
4259         (OPTION_DEFAULT_SPECS): Add "nan" default.
4260         (ASM_SPEC): Handle `-mnan='.
4261         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4262         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4263         comment accordingly.
4264         (neg<mode>2): Likewise.
4265         * config/mips/mips.opt (mabs, mnan): New options.
4266         * doc/install.texi (Configuration): Document `--with-nan=' option.
4267         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4268         `-mnan=' options.
4269         (MIPS Options): Document them.
4270         * config.gcc <mips*-*-*>: Handle `--with-nan='.
4271         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4272         * configure: Regenerate.
4273         * config.in: Regenerate.
4275 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4277         * config/i386/i386.md (float post-reload splitters): Do not check
4278         for subregs of SSE registers.
4280 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
4281             H.J. Lu  <hongjiu.lu@intel.com>
4283         PR target/57954
4284         PR target/57988
4285         * config/i386/i386.md (post-reload splitter
4286         to avoid partial SSE reg dependency stalls): New pattern.
4288 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4290         * config/s390/s390.md ("movcc"): Swap load and store instructions.
4292 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
4294         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4295         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4297 2013-07-26  Cary Coutant  <ccoutant@google.com>
4299         * dwarf2out.c (die_checksum_ordered): Don't include template
4300         instantiations in signature.
4301         (is_template_parameter): New function.
4302         (is_template_instantiation): New function.
4303         (generate_skeleton_bottom_up): Don't include template instantiations
4304         in type unit DIE.
4305         (generate_skeleton): Likewise.
4306         (break_out_comdat_types): Move recursive call to break out nested
4307         types earlier.
4308         (prune_unused_types_mark_generic_parms_dies): Call
4309         is_template_parameter.
4311 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
4313         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4314         uses vector registers.
4315         * config/aarch64/iterators.md: Add attributes rtn and vas.
4317 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4318             Richard Earnshaw  <richard.earnshaw@arm.com>
4320         * combine.c (simplify_comparison): Re-canonicalize operands
4321         where appropriate.
4322         * config/arm/arm.md (movcond_addsi): New splitter.
4324 2013-07-25  Sterling Augustine  <saugustine@google.com>
4326         * dwarf2out.c (size_of_pubnames): Move code to...
4327         (include_pubname_in_output): ...here.  New.
4328         (want_pubnames): Rearrange.
4329         (output_pubnames): Call include_pubname_in_output.  Move assertion.
4331 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4333         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4335 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
4337         PR target/38836
4338         * doc/extend.texi: Remove obsolete builtins. Fix
4339         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4341 2013-07-25  Jan Hubicka  <jh@suse.cz>
4343         * cgraph.c (release_function_body): Break out from ...
4344         (cgraph_release_function_body): ... this one; also release DECL_RESULT
4345         and DECL_ARGUMENTS.
4346         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4347         old_tree in the map.
4348         (create_specialized_node): Update.
4349         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4350         into index.
4351         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4352         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4353         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4354         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4355         DECL_RESULT.
4357 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4359         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4360         addsi3_carryin_alt2_<optab>): Correct output template.
4362 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4364         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4365         Adjust for arm_restrict_it.
4366         Remove trailing whitespace.
4368 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
4370         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4371         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4373         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4375 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
4377         PR rtl-optimization/57960
4378         * lra-constraints.c (process_alt_operands): Use the right mode
4379         when checking strict_low.
4381 2013-07-25  Jan Hubicka  <jh@suse.cz>
4383         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4384         * cgraph.c (cgraph_remove_node): Do not release function body
4385         when in cgraph streaming.
4386         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4387         in other partitions are not considered reachable; fix handling of
4388         clones.
4390 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4392         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4394 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4396         PR target/19599
4397         PR target/57731
4398         PR target/57837
4399         * config/arm/arm.md ("*sibcall_insn): Replace use of
4400         Ss with US. Adjust output for v5 and v4t.
4401         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4403         * config/arm/constraints.md ("Ss"): Rename to US.
4405 2013-07-25  Terry Guo  <terry.guo@arm.com>
4407         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4408         shift_add/shift_sub0/shift_sub1 RTXs.
4410 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
4411             Anton Blanchard  <anton@au1.ibm.com>
4413         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4414         (altivec_vpks<VI_char>ss): Likewise.
4415         (altivec_vpks<VI_char>us): Likewise.
4416         (altivec_vpku<VI_char>us): Likewise.
4417         (altivec_vpku<VI_char>um): Likewise.
4419 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
4421         Introduce context class.
4423         * Makefile.in (CONTEXT_H): New.
4424         (OBJS): Add context.o.
4425         (toplev.o): Add CONTEXT_H to dependencies.
4426         (context.o): New.
4428         * toplev.c (general_init): Create the singleton gcc::context instance.
4430         * context.c: New.
4432         * context.h: New.
4434 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
4436         PR rtl-optimization/57968
4437         * mode-switching.c (create_pre_exit): Allow instructions that
4438         don't set a return register to need a non-exit mode.
4440 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4441             Anton Blanchard  <anton@au1.ibm.com>
4443         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4444         operands to vperm for little endian.
4445         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4446         of lvsl to create the control mask for a vperm for little endian.
4448 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4449             Anton Blanchard  <anton@au1.ibm.com>
4451         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4452         two operands for little-endian.
4454 2013-07-23  Steve Ellcey  <sellcey@mips.com>
4456         * config/mips/mips.c (mips_case_values_threshold): New.
4457         (TARGET_CASE_VALUES_THRESHOLD): Define.
4459 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4460             Anton Blanchard  <anton@au1.ibm.com>
4462         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4463         selection of field for vector splat in little endian mode.
4465 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4467         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4468         expanders to rs6000.md.
4469         (ior<mode>3): Likewise.
4470         (and<mode>3): Likewise.
4471         (one_cmpl<mode>2): Likewise.
4472         (nor<mode>3): Likewise.
4473         (andc<mode>3): Likewise.
4474         (eqv<mode>3): Likewise.
4475         (nand<mode>3): Likewise.
4476         (orc<mode>3): Likewise.
4478         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4479         declaration.
4481         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4482         to split multi-word logical operations.
4483         (rs6000_split_logical_di): Likewise.
4484         (rs6000_split_logical): Likewise.
4486         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4487         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4488         and allow TImode operations in 32-bit.
4489         (vsx_and<mode>3_64bit): Likewise.
4490         (vsx_ior<mode>3_32bit): Likewise.
4491         (vsx_ior<mode>3_64bit): Likewise.
4492         (vsx_xor<mode>3_32bit): Likewise.
4493         (vsx_xor<mode>3_64bit): Likewise.
4494         (vsx_one_cmpl<mode>2_32bit): Likewise.
4495         (vsx_one_cmpl<mode>2_64bit): Likewise.
4496         (vsx_nor<mode>3_32bit): Likewise.
4497         (vsx_nor<mode>3_64bit): Likewise.
4498         (vsx_andc<mode>3_32bit): Likewise.
4499         (vsx_andc<mode>3_64bit): Likewise.
4500         (vsx_eqv<mode>3_32bit): Likewise.
4501         (vsx_eqv<mode>3_64bit): Likewise.
4502         (vsx_nand<mode>3_32bit): Likewise.
4503         (vsx_nand<mode>3_64bit): Likewise.
4504         (vsx_orc<mode>3_32bit): Likewise.
4505         (vsx_orc<mode>3_64bit): Likewise.
4507         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4508         logical types in GPRs.
4510         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4511         logical insns to rs6000.md, and allow TImode operations in
4512         32-bit.
4513         (altivec_ior<mode>3): Likewise.
4514         (altivec_xor<mode>3): Likewise.
4515         (altivec_one_cmpl<mode>2): Likewise.
4516         (altivec_nor<mode>3): Likewise.
4517         (altivec_andc<mode>3): Likewise.
4519         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4520         attributes for moving the 128-bit logical operations into
4521         rs6000.md.
4522         (BOOL_REGS_OUTPUT): Likewise.
4523         (BOOL_REGS_OP1): Likewise.
4524         (BOOL_REGS_OP2): Likewise.
4525         (BOOL_REGS_UNARY): Likewise.
4526         (BOOL_REGS_AND_CR0): Likewise.
4527         (one_cmpl<mode>2): Add support for DI logical operations on
4528         32-bit, splitting the operations to 32-bit.
4529         (anddi3): Likewise.
4530         (iordi3): Likewise.
4531         (xordi3): Likewise.
4532         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4533         changes to combine the 32/64-bit code, allow logical operations on
4534         TI mode in 32-bit, and to use similar match_operator patterns like
4535         scalar mode uses.  Combine the Altivec and VSX code for logical
4536         operations, and move it here.
4537         (ior<mode>3, 128-bit types): Likewise.
4538         (xor<mode>3, 128-bit types): Likewise.
4539         (one_cmpl<mode>3, 128-bit types): Likewise.
4540         (nor<mode>3, 128-bit types): Likewise.
4541         (andc<mode>3, 128-bit types): Likewise.
4542         (eqv<mode>3, 128-bit types): Likewise.
4543         (nand<mode>3, 128-bit types): Likewise.
4544         (orc<mode>3, 128-bit types): Likewise.
4545         (and<mode>3_internal): Likewise.
4546         (bool<mode>3_internal): Likewise.
4547         (boolc<mode>3_internal1): Likewise.
4548         (boolc<mode>3_internal2): Likewise.
4549         (boolcc<mode>3_internal1): Likewise.
4550         (boolcc<mode>3_internal2): Likewise.
4551         (eqv<mode>3_internal1): Likewise.
4552         (eqv<mode>3_internal2): Likewise.
4553         (one_cmpl1<mode>3_internal): Likewise.
4555 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4557         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4558         Rename flag_stack_usage to flag_stack_usage_info.
4560 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4562         * config/microblaze/sync.md: New file.
4563         * config/microblaze/microblaze.md: Include sync.md
4564         * config/microblaze/microblaze.c: Add print_operand 'y'.
4565         * config/microblaze/constraints.md: Add memory_contraint
4566         'Q' which is a single register.
4568 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
4570         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4572 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4574         * reload.c (find_reloads): Exit loop once we find this operand
4575         cannot be reloaded somehow for this alternative.
4577         * reload.c (find_reloads): Exit loop once we find a hard register.
4579         * rtlanal.c (computed_jump_p): Exit loop once we find label
4580         reference is used.
4582         * i386.c (ix86_pad_returns): Exit loop after setting replace.
4584         * cfgloopmanip.c (remove_path): Exit loop after setting
4585         irred_invalidated.
4587         * gensupport.c (subst_dup): Avoid loop if code is not
4588         MATCH_DUP nor MATCH_OP_DUP.
4590 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
4592         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4594 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4596         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4597         true for SP_REGNUM if mode == ptr_mode.
4598         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4599         with value R0_REGNUM + 31.
4601 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4603         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4604         pad pointer-typed argument downward.
4606 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4608         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4609         and __ILP32__ when the ILP32 model is in use.
4611 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4613         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4614         (aarch64_load_symref_appropriately): In the case of
4615         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4616         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4617         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4618         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4619         if 'mode' doesn't equal to 'ptr_mode'.
4620         (aarch64_output_mi_thunk): Add an assertion on the alignment of
4621         'vcall_offset'; change to call aarch64_emit_move differently depending
4622         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4623         to calculate the upper bound of 'vcall_offset'.
4624         (aarch64_cannot_force_const_mem): Change to also return true if
4625         mode != ptr_mode.
4626         (aarch64_legitimize_reload_address): In the case of large
4627         displacements, add new local variable 'xmode' and an assertion
4628         based on it; change to use 'xmode' to generate the new rtx and
4629         reload.
4630         (aarch64_asm_trampoline_template): Change to generate the template
4631         differently depending on TARGET_ILP32 or not; change to use
4632         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4633         (aarch64_trampoline_size): Removed.
4634         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4635         and replace immediate literals with it.  Change to use 'ptr_mode'
4636         instead of 'DImode' and call convert_memory_address if the mode
4637         of 'fnaddr' doesn't equal to 'ptr_mode'.
4638         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4639         to output symbol.
4640         (aarch64_elf_asm_destructor): Likewise.
4641         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4642         on TARGET_ILP32 instead of aarch64_trampoline_size.
4643         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4644         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4645         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4646         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
4647         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4648         (storewb_pair<GPI:mode>_<P:mode>): ... this.
4649         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4650         depending on the value of 'mode'.
4651         (add_losym_<mode>): New.
4652         (ldr_got_small_<mode>): New, based on ldr_got_small.
4653         (ldr_got_small): Remove.
4654         (ldr_got_small_sidi): New.
4655         * config/aarch64/iterators.md (P): New.
4656         (PTR): Change to 'ptr_mode' in the condition.
4658 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4660         * config.gcc (aarch64*-*-*): Support --with-abi.
4661         (aarch64*-*-elf): Support --with-multilib-list.
4662         (aarch64*-*-linux*): Likewise.
4663         (supported_defaults): Add abi to aarch64*-*-*.
4664         * configure.ac: Mention AArch64 for --with-multilib-list.
4665         * configure: Re-generated.
4666         * config/aarch64/biarchilp32.h: New file.
4667         * config/aarch64/biarchlp64.h: New file.
4668         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4669         (ABI_SPEC): Ditto.
4670         (MULTILIB_DEFAULTS): Ditto.
4671         (DRIVER_SELF_SPECS): Ditto.
4672         (ASM_SPEC): Update to also substitute -mabi.
4673         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4674         file whose name depends on -mabi= and -mbig-endian.
4675         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4676         TARGET_ILP32.
4677         (POINTER_SIZE): New define.
4678         (POINTERS_EXTEND_UNSIGNED): Ditto.
4679         (enum aarch64_abi_type): New enumeration tag.
4680         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4681         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4682         (TARGET_ILP32): New define.
4683         * config/aarch64/aarch64.opt (mabi): New.
4684         (aarch64_abi): New.
4685         (ilp32, lp64): New values for -mabi.
4686         * config/aarch64/t-aarch64 (comma): New define.
4687         (MULTILIB_OPTIONS): Ditto.
4688         (MULTILIB_DIRNAMES): Ditto.
4689         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4690         * doc/invoke.texi: Document -mabi for AArch64.
4692 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
4694         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4696 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4697             Anton Blanchard  <anton@au1.ibm.com>
4699         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4700         endianness when selecting field to splat.
4702 2013-07-22  Eric Christopher  <echristo@gmail.com>
4704         * dwarf2out.c (die_odr_checksum): New function to use
4705         CHECKSUM_ macros and ULEB128 for DIE tag.
4706         (generate_type_signature): Use.
4708 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
4710         * config.gcc (sparc*-*-*): Accept leon3 processor.
4711         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4712         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4713         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4714         * config/sparc/sparc.opt (enum processor_type): Add leon3.
4715         (mfix-ut699): Adjust comment.
4716         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4717         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4718         (CPP_CPU_SPEC): Likewise.
4719         (ASM_CPU_SPEC): Likewise.
4720         * config/sparc/sparc.c (leon3_cost): New constant.
4721         (sparc_option_override): Add leon3 support.
4722         (mem_ref): New function.
4723         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4724         (sparc_do_work_around_errata): Look into the instruction in the delay
4725         slot and adjust accordingly.  Add fix for the data cache nullify issues
4726         of the UT699.  Change insertion position for the NOP.
4727         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4728         (leon3_load): New reservation.
4729         (leon_store): Bump latency to 2.
4730         (grfpu): New automaton.
4731         (grfpu_alu): New unit.
4732         (grfpu_ds): Likewise.
4733         (leon_fp_alu): Adjust.
4734         (leon_fp_mult): Delete.
4735         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4736         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4737         * config/sparc/sparc.md (cpu): Add leon3.
4738         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4739         (swapsi): Likewise.
4740         (atomic_test_and_set): Likewise.
4741         (ldstub): Likewise.
4743 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
4745         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
4746         default for R5900 targets.
4747         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4748         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4749         * config/mips/mips.c (mips_option_override): Report an error for
4750         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
4751         for -march=r5900 -mhard-float.
4753 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4755         * df-problems.c (can_move_insns_across): Exit loop once we
4756         find a non-fixed, non-global register.
4758         * ipa-pure-const.c (propagate_nothrow): Exit loop after
4759         setting can_throw.
4761         * omega.c (omega_eliminate_red): Break after setting red_found.
4762         (omega_problem_has_red_equations): Similarly after setting found.
4763         (omega_query_variable): Similarly after setting coupled.
4765 2013-07-22  Marek Polacek  <polacek@redhat.com>
4767         * gimplify.c: Don't include gimple.h twice.
4769 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4771         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4772         instead of TARGET_THUMB1.
4773         (Pz): New constraint.
4774         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4775         encodings.
4776         (compare_negsi_si): Likewise.
4777         (compare_addsi2_op0): Likewise.
4778         (compare_addsi2_op1): Likewise.
4779         (addsi3_carryin_<optab>): Likewise.
4780         (addsi3_carryin_alt2_<optab>): Likewise.
4781         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4782         for arm_restrict_it.
4783         (subsi3_carryin): Likewise.
4784         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4785         (minmax_arithsi): Disable for arm_restrict_it.
4786         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4787         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4788         (satsi_<SAT:code>_shift): Likewise.
4789         (arm_shiftsi3): Add alternative for 16-bit encoding.
4790         (arm32_movhf): Disable for arm_restrict_it.
4791         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4792         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4794 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
4796         * config/arm/arm.md (attribute "insn"): Delete.
4797         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4798         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4799         (not_shiftsi): Update for attribute change.
4800         (not_shiftsi_compare0): Likewise.
4801         (not_shiftsi_compare0_scratch): Likewise.
4802         (arm_one_cmplsi2): Likewise.
4803         (thumb1_one_cmplsi2): Likewise.
4804         (notsi_compare0): Likewise.
4805         (notsi_compare0_scratch): Likewise.
4806         (thumb1_movdi_insn): Likewise.
4807         (arm_movsi_insn): Likewise.
4808         (movhi_insn_arch4): Likewise.
4809         (movhi_bytes): Likewise.
4810         (arm_movqi_insn): Likewise.
4811         (thumb1_movqi_insn): Likewise.
4812         (arm32_movhf): Likewise.
4813         (thumb1_movhf): Likewise.
4814         (arm_movsf_soft_insn): Likewise.
4815         (thumb1_movsf_insn): Likewise.
4816         (thumb_movdf_insn): Likewise.
4817         (movsicc_insn): Likewise.
4818         (movsfcc_soft_insn): Likewise.
4819         (and_scc): Likewise.
4820         (cond_move): Likewise.
4821         (if_move_not): Likewise.
4822         (if_not_move): Likewise.
4823         (if_shift_move): Likewise.
4824         (if_move_shift): Likewise.
4825         (if_shift_shift): Likewise.
4826         (if_not_arith): Likewise.
4827         (if_arith_not): Likewise.
4828         (cond_move_not): Likewise.
4829         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4830         (neon_mov<mode>): Likewise.
4831         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4832         (thumb2_movsi_vfp): Likewise.
4833         (movsf_vfp): Likewise.
4834         (thumb2_movsf_vfp): Likewise.
4835         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4836         change.
4837         (cortexa7_older_only): Likewise.
4838         (cortexa7_younger): Likewise.
4839         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4840         (1020alu_shift_op): Likewise.
4841         (1020alu_shift_reg_op): Likewise.
4842         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4843         (alu_shift_op): Likewise.
4844         (alu_shift_reg_op): Likewise.
4845         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4846         (11_alu_shift_op): Likewise.
4847         (11_alu_shift_reg_op): Likewise.
4848         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4849         (9_alu_shift_reg_op): Likewise.
4850         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4851         change.
4852         (cortex_a15_alu_shift): Likewise.
4853         (cortex_a15_alu_shift_reg): Likewise.
4854         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4855         (cortex_a5_alu_shift): Likewise.
4856         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4857         change.
4858         (cortex_a53_alu_shift): Likewise.
4859         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4860         change.
4861         (cortex_a7_alu_reg): Likewise.
4862         (cortex_a7_alu_shift): Likewise.
4863         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4864         (cortex_a8_alu_shift): Likewise.
4865         (cortex_a8_alu_shift_reg): Likewise.
4866         (cortex_a8_mov): Likewise.
4867         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4868         (cortex_a9_dp_shift): Likewise.
4869         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4870         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4871         (cortex_r4_mov): Likewise.
4872         (cortex_r4_alu_shift): Likewise.
4873         (cortex_r4_alu_shift_reg): Likewise.
4874         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4875         (526_alu_shift_op): Likewise.
4876         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4877         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4878         (626te_alu_shift_op): Likewise.
4879         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4880         (726te_alu_op): Likewise.
4881         (726te_alu_shift_op): Likewise.
4882         (726te_alu_shift_reg_op): Likewise.
4883         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4884         (mp626_alu_shift_op): Likewise.
4885         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4886         (pj4_alu_e1_conds): Likewise.
4887         (pj4_alu): Likewise.
4888         (pj4_alu_conds): Likewise.
4889         (pj4_shift): Likewise.
4890         (pj4_shift_conds): Likewise.
4891         (pj4_alu_shift): Likewise.
4892         (pj4_alu_shift_conds): Likewise.
4894 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4896         * config/arm/predicates.md (shiftable_operator_strict_it):
4897         New predicate.
4898         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4899         Disable cond_exec version for arm_restrict_it.
4900         (thumb2_smaxsi3): Convert to generate cond_exec.
4901         (thumb2_sminsi3): Likewise.
4902         (thumb32_umaxsi3): Likewise.
4903         (thumb2_uminsi3): Likewise.
4904         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4905         (thumb2_neg_abssi2): Likewise.
4906         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4907         (thumb2_movsicc_insn): Adjust alternatives.
4908         (thumb2_mov_negscc): Disable for arm_restrict_it.
4909         (thumb2_mov_negscc_strict_it): New pattern.
4910         (thumb2_mov_notscc_strict_it): New pattern.
4911         (thumb2_mov_notscc): Disable for arm_restrict_it.
4912         (thumb2_ior_scc): Likewise.
4913         (thumb2_ior_scc_strict_it): New pattern.
4914         (thumb2_cond_move): Adjust for arm_restrict_it.
4915         (thumb2_cond_arith): Disable for arm_restrict_it.
4916         (thumb2_cond_arith_strict_it): New pattern.
4917         (thumb2_cond_sub): Adjust for arm_restrict_it.
4918         (thumb2_movcond): Likewise.
4919         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4920         (thumb2_zero_extendhisi2_v6): Likewise.
4921         (thumb2_zero_extendqisi2_v6): Likewise.
4922         (orsi_notsi_si): Likewise.
4923         (orsi_not_shiftsi_si): Likewise.
4925 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4927         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4928         instruction sequence is 1 byte shorter.
4930 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4932         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4933         it is not needed after split.
4935 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4937         PR target/51784
4938         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4939         second label for nonlocal goto receivers. Don't output pic base labels
4940         unless we're producing PIC; mark that action unreachable().
4941         (ix86_save_reg): If the function contains a nonlocal label, save the
4942         PIC base reg.
4943         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4944         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4945         (update_pic_label_number_if_needed): New.
4946         (machopic_output_function_base_name): Adjust for nonlocal receiver
4947         case.
4948         (machopic_should_output_picbase_label): New.
4949         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4950         (nonlocal_goto_receiver): New insn and split.
4952 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4954         * config/aarch64/aarch64-builtins.c
4955         (aarch64_fold_builtin): Fold abs in all modes.
4956         * config/aarch64/aarch64-simd-builtins.def
4957         (abs): Enable for all modes.
4958         * config/aarch64/arm_neon.h
4959         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4960         (vabs_f64): Add missing intrinsic.
4962 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4964         * config/aarch64/arm_neon.h (vabs_s64): New function
4966 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4968         PR target/57516
4969         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4970         * config/avr/avr.md (adjust_len): Add `round'.
4971         * config/avr/avr-protos.h (avr_out_round): New prototype.
4972         (avr_out_plus): Add `out_label' argument.
4973         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4974         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4975         Handle the case where `insn' is just a pattern.
4976         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4977         (avr_out_round): New function.
4978         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4980 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
4982         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4983         Add check for flag_stack_usage to handle -fstack-usage support
4985 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
4987         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4988         interaction for new Power8 flags and VSX.
4990 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4992         PR middle-end/57698
4993         * tree-inline.c (expand_call_inline): Emit errors during
4994         early_inlining only if optimization is not turned on.
4996 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4998         * passes.def: New.
5000         * passes.c (init_optimization_passes): Move the construction of
5001         the pass hierarchy into a new passes.def file.
5003         * Makefile.in (passes.o): Add dependency on passes.def.
5005 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5007         * passes.c (init_optimization_passes): Introduce macros for
5008         constructing the tree of passes (INSERT_PASSES_AFTER,
5009         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
5010         TERMINATE_PASS_LIST).
5012 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
5013             Wei Mi  <wmi@google.com>
5015         PR rtl-optimization/57878
5016         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
5017         top.
5018         (reload_pseudo_compare_func): Check nregs first for reload
5019         pseudos.
5021 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5023         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
5025 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
5027         * read-rtl.c (validate_const_int): Once an invalid character is
5028         seen, quit the loop.
5030         * gengtype.c (write_roots): Similarly once we find the "deletable"
5031         or "if_marked" option.
5033 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
5035         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
5036         "xtab" and "sat".  Move value "clz" from here to ...
5037         (attriubte "type"): ... here.
5038         (satsi_<SAT:code>): Delete "insn" attribute.
5039         (satsi_<SAT:code>_shift): Likewise.
5040         (arm_zero_extendqisi2addsi): Likewise.
5041         (arm_extendqisi2addsi): Likewise.
5042         (clzsi2): Update for attribute changes.
5043         (rbitsi2): Likewise.
5044         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
5045         attribute.
5046         (arm_usatsihi): Likewise.
5047         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
5049 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
5051         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
5052         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
5053         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
5054         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
5055         in alphabetical order.
5056         (attribute "core_cycles"): Update for attribute changes.
5057         (arm_addsi3): Likewise.
5058         (addsi3_compare0): Likewise.
5059         (addsi3_compare0_scratch): Likewise.
5060         (addsi3_compare_op1): Likewise.
5061         (addsi3_compare_op2): Likewise.
5062         (compare_addsi2_op0): Likewise.
5063         (compare_addsi2_op1): Likewise.
5064         (addsi3_carryin_shift_<optab>): Likewise.
5065         (subsi3_carryin_shift): Likewise.
5066         (rsbsi3_carryin_shift): Likewise.
5067         (arm_subsi3_insn): Likewise.
5068         (subsi3_compare0): Likewise.
5069         (subsi3_compare): Likewise.
5070         (arm_andsi3_insn): Likewise.
5071         (thumb1_andsi3_insn): Likewise.
5072         (andsi3_compare0): Likewise.
5073         (andsi3_compare0_scratch): Likewise.
5074         (zeroextractsi_compare0_scratch
5075         (andsi_not_shiftsi_si): Likewise.
5076         (iorsi3_insn): Likewise.
5077         (iorsi3_compare0): Likewise.
5078         (iorsi3_compare0_scratch): Likewise.
5079         (arm_xorsi3): Likewise.
5080         (thumb1_xorsi3_insn): Likewise.
5081         (xorsi3_compare0): Likewise.
5082         (xorsi3_compare0_scratch): Likewise.
5083         (satsi_<SAT:code>_shift): Likewise.
5084         (rrx): Likewise.
5085         (arm_shiftsi3): Likewise.
5086         (shiftsi3_compare0): Likewise.
5087         (not_shiftsi): Likewise.
5088         (not_shiftsi_compare0): Likewise.
5089         (not_shiftsi_compare0_scratch): Likewise.
5090         (arm_one_cmplsi2): Likewise.
5091         (thumb_one_complsi2): Likewise.
5092         (notsi_compare0): Likewise.
5093         (notsi_compare0_scratch): Likewise.
5094         (thumb1_zero_extendhisi2): Likewise.
5095         (arm_zero_extendhisi2): Likewise.
5096         (arm_zero_extendhisi2_v6): Likewise.
5097         (arm_zero_extendhisi2addsi): Likewise.
5098         (thumb1_zero_extendqisi2): Likewise.
5099         (thumb1_zero_extendqisi2_v6): Likewise.
5100         (arm_zero_extendqisi2): Likewise.
5101         (arm_zero_extendqisi2_v6): Likewise.
5102         (arm_zero_extendqisi2addsi): Likewise.
5103         (thumb1_extendhisi2): Likewise.
5104         (arm_extendhisi2): Likewise.
5105         (arm_extendhisi2_v6): Likewise.
5106         (arm_extendqisi): Likewise.
5107         (arm_extendqisi_v6): Likewise.
5108         (arm_extendqisi2addsi): Likewise.
5109         (thumb1_extendqisi2): Likewise.
5110         (thumb1_movdi_insn): Likewise.
5111         (arm_movsi_insn): Likewise.
5112         (movsi_compare0): Likewise.
5113         (movhi_insn_arch4): Likewise.
5114         (movhi_bytes): Likewise.
5115         (arm_movqi_insn): Likewise.
5116         (thumb1_movqi_insn): Likewise.
5117         (arm32_movhf): Likewise.
5118         (thumb1_movhf): Likewise.
5119         (arm_movsf_soft_insn): Likewise.
5120         (thumb1_movsf_insn): Likewise.
5121         (movdf_soft_insn): Likewise.
5122         (thumb_movdf_insn): Likewise.
5123         (arm_cmpsi_insn): Likewise.
5124         (cmpsi_shiftsi): Likewise.
5125         (cmpsi_shiftsi_swp): Likewise.
5126         (arm_cmpsi_negshiftsi_si): Likewise.
5127         (movsicc_insn): Likewise.
5128         (movsfcc_soft_insn): Likewise.
5129         (arith_shiftsi): Likewise.
5130         (arith_shiftsi_compare0
5131         (arith_shiftsi_compare0_scratch
5132         (sub_shiftsi): Likewise.
5133         (sub_shiftsi_compare0
5134         (sub_shiftsi_compare0_scratch
5135         (and_scc): Likewise.
5136         (cond_move): Likewise.
5137         (if_plus_move): Likewise.
5138         (if_move_plus): Likewise.
5139         (if_move_not): Likewise.
5140         (if_not_move): Likewise.
5141         (if_shift_move): Likewise.
5142         (if_move_shift): Likewise.
5143         (if_shift_shift): Likewise.
5144         (if_not_arith): Likewise.
5145         (if_arith_not): Likewise.
5146         (cond_move_not): Likewise.
5147         (thumb1_ashlsi3): Set type attribute.
5148         (thumb1_ashrsi3): Likewise.
5149         (thumb1_lshrsi3): Likewise.
5150         (thumb1_rotrsi3): Likewise.
5151         (shiftsi3_compare0_scratch): Likewise.
5152         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5153         (neon_mov<mode>): Likewise.
5154         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5155         attribute changes.
5156         (thumb2_movsi_insn): Likewise.
5157         (thumb2_cmpsi_neg_shiftsi): Likewise.
5158         (thumb2_extendqisi_v6): Likewise.
5159         (thumb2_zero_extendhisi2_v6): Likewise.
5160         (thumb2_zero_extendqisi2_v6): Likewise.
5161         (thumb2_shiftsi3_short): Likewise.
5162         (thumb2_addsi3_compare0_scratch): Likewise.
5163         (orsi_not_shiftsi_si): Likewise.
5164         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5165         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5166         changes.
5167         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5168         (1020alu_shift_op): Likewise.
5169         (1020alu_shift_reg_op): Likewise.
5170         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5171         (alu_shift_op): Likewise.
5172         (alu_shift_reg_op): Likewise.
5173         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5174         (11_alu_shift_op): Likewise.
5175         (11_alu_shift_reg_op): Likewise.
5176         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5177         (9_alu_shift_reg_op): Likewise.
5178         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5179         changes.
5180         (cortex_a15_alu_shift): Likewise.
5181         (cortex_a15_alu_shift_reg): Likewise.
5182         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5183         changes.
5184         (cortex_a5_alu_shift): Likewise.
5185         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5186         changes.
5187         (cortex_a53_alu_shift): Likewise.
5188         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5189         changes.
5190         (cortex_a7_alu_reg): Likewise.
5191         (cortex_a7_alu_shift): Likewise.
5192         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5193         changes.
5194         (cortex_a8_alu_shift): Likewise.
5195         (cortex_a8_alu_shift_reg): Likewise.
5196         (cortex_a8_mov): Likewise.
5197         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5198         (cortex_a9_dp_shift): Likewise.
5199         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5200         changes.
5201         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5202         changes.
5203         (cortex_r4_mov): Likewise.
5204         (cortex_r4_alu_shift): Likewise.
5205         (cortex_r4_alu_shift_reg): Likewise.
5206         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5207         (526_alu_shift_op): Likewise.
5208         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5209         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5210         (626te_alu_shift_op): Likewise.
5211         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5212         (726te_alu_op): Likewise.
5213         (726te_alu_shift_op): Likewise.
5214         (726te_alu_shift_reg_op): Likewise.
5215         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5216         (mp626_alu_shift_op): Likewise.
5217         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5218         (pj4_alu_e1_conds): Likewise.
5219         (pj4_alu): Likewise.
5220         (pj4_alu_conds): Likewise.
5221         (pj4_shift): Likewise.
5222         (pj4_shift_conds): Likewise.
5223         (pj4_alu_shift): Likewise.
5224         (pj4_alu_shift_conds): Likewise.
5225         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5226         changes.
5227         (cortexa7_older_only): Likewise.
5228         (cortexa7_younger): Likewise.
5230 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5232         * ipa-pure-const.c (generate_summary): Rename to...
5233         (pure_const_generate_summary): ... this.
5235 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
5237         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5239 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
5241         PR target/57909
5242         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5243         usage in HI mode.
5245 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5247         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5248         enabled without -march=zEC12.
5249         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5250         flags to be set.
5252 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5254         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5255         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5256         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5257         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5258         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5259         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5260         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5261         ISA_HAS_FP4.
5262         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5263         and ISA_HAS_NMADD3_NMSUB3.
5264         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5265         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5266         (nmsub4<mode>, nmsub3<mode>): Likewise.
5267         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5269 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5271         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5272         TARGET_MIPS5400 checking.
5274 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
5275             Peter Bergner  <bergner@vnet.ibm.com>
5277         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5278         registers in the comment.
5279         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5280         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5281         rather than FIRST_PSEUDO_REGISTERS.
5283 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
5285         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5286         enable extra ISA flags with TARGET_HTM.
5288 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
5290         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5291         Fix comment typos.
5293 2013-07-15  Cong Hou  <congh@google.com>
5295         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5296         in compare function for sorting.
5298 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
5300         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5301         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5302         * config/rs6000/rs6000.opt: Add -mhtm option.
5303         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5304         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5305         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5306         __HTM__ if the HTM instructions are available.
5307         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5308         htm_spr_reg_operand): New define_predicates.
5309         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5310         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5311         Include htm.md.
5312         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5313         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5314         HTM builtin functions.
5315         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5316         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5317         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5318         (rs6000_builtin_mask_calculate): Likewise.
5319         (rs6000_option_override_internal): Likewise.
5320         (bdesc_htm): Add new HTM builtin support.
5321         (htm_spr_num): New function.
5322         (htm_spr_regno): Likewise.
5323         (rs6000_htm_spr_icode): Likewise.
5324         (htm_expand_builtin): Likewise.
5325         (htm_init_builtins): Likewise.
5326         (rs6000_expand_builtin): Add support for HTM builtin functions.
5327         (rs6000_init_builtins): Likewise.
5328         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5329         option.
5330         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5331         (TARGET_HTM, MASK_HTM): Define macros.
5332         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5333         (FIXED_REGISTERS): Likewise.
5334         (CALL_USED_REGISTERS): Likewise.
5335         (CALL_REALLY_USED_REGISTERS): Likewise.
5336         (REG_ALLOC_ORDER): Likewise.
5337         (enum reg_class): Likewise.
5338         (REG_CLASS_NAMES): Likewise.
5339         (REG_CLASS_CONTENTS): Likewise.
5340         (REGISTER_NAMES): Likewise.
5341         (ADDITIONAL_REGISTER_NAMES): Likewise.
5342         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5343         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5344         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5345         * config/rs6000/htm.md: New file.
5346         * config/rs6000/htmintrin.h: New file.
5347         * config/rs6000/htmxlintrin.h: New file.
5349 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5351         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5352         Define SYMBOL_TINY_GOT, update comment.
5353         * config/aarch64/aarch64.c
5354         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5355         (aarch64_expand_mov_immediate): Likewise.
5356         (aarch64_print_operand): Likewise.
5357         (aarch64_classify_symbol): Likewise.
5358         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5359         (ldr_got_tiny): Define.
5361 2013-07-13  Tobias Grosser  <tobias@grosser.es>
5363         PR tree-optimization/54094
5364         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5365           scheduling dimension for the parallelism check from the polyhedral
5366           information in the AST.
5367         * graphite-dependences.c (carries_deps): Do not assume the schedule is
5368           in 2D + 1 form.
5370 2013-07-13  Jason Merrill  <jason@redhat.com>
5372         * print-tree.c (debug_vec_tree): Use debug_raw.
5373         (debug_raw (vec<tree, va_gc> &)): New.
5374         (debug_raw (vec<tree, va_gc> *)): New.
5375         * tree.h: Declare them.
5377 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
5379         * ifcvt.c (ifcvt_after_combine): New static variable.
5380         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5381         for size.
5382         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
5383         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5384         rest_of_handle_if_after_reload): Pass new argument for if_convert.
5386 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
5388         * config/mips/mips.c (mips_expand_call): Remove empty statement.
5390 2013-07-12  Michael Matz  <matz@suse.de>
5392         PR middle-end/55771
5393         * convert.c (convert_to_real): Reject non-float inner types.
5395 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
5397         * config/aarch64/aarch64-protos.h
5398         (aarch64_simd_immediate_valid_for_move): Remove.
5399         * config/aarch64/aarch64.c (simd_immediate_info): New member.
5400         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5401         cases.
5402         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5404 2013-07-11  Steve Ellcey  <sellcey@mips.com>
5406         * config/mips/mips.c (mips_conditional_register_usage): Do not
5407         use t[0-7] registers in MIPS16 mode when optimizing for size.
5409 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
5411         * config/i386/i386.c (dispatch_function_versions): Fix array
5412         indexing of function_version_info to match actual_versions.
5414 2013-07-11  Teresa Johnson  <tejohnson@google.com>
5416         * vec.h (struct va_gc): Move release out-of-line.
5417         (va_gc::release): Call ggc_free on released vec.
5419 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5421         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5422         Require GOT register as additional operand in UNSPEC.
5423         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5424         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5425         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5426         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5427         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5428         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5429         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5431 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
5433         PR target/57631
5434         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5435         name seen by assembler/linker rather if available.
5437 2013-07-11  Andreas Schwab  <schwab@suse.de>
5439         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5441 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
5443         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5445 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
5447         * doc/tm.texi.in: Move hook documentation to ....
5448         * target.def: ... here.
5450         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5451         text on @hook line.
5452         * doc/tm.texi: Regenerate.
5454 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
5456         PR c++/57869
5457         * doc/invoke.texi: Document -Wconditionally-supported.
5459 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5461         PR target/57844
5462         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5463         of my_fp.
5465 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5467         PR target/57506
5468         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5469         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5470         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5471         Remove duplicate devices.
5472         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5473         * config/avr/t-multilib: Regenerate.
5474         * config/avr/avr-tables.opt: Regenerate.
5475         * doc/avr-mmcu.texi: Regenerate.
5477 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5479         PR target/56987
5480         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5482 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
5484         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5485         the cost of MULT when optimizing for size.
5487 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5489         * config/cr16/cr16-protos.h: Don't include target.h.
5491 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
5493         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5494         adjust register size for TDmode and TFmode for VSX registers.
5496 2013-07-08  Kai Tietz  <ktietz@redhat.com>
5498         PR target/56892
5499         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5500         hook_bool_const_tree_true.
5502 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5504         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5505         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5506         * config/s390/s390.md: Define FPR*_REGNUM constants.
5507         Fix FPR2_REGNUM constant (18 -> 17).
5508         ("*trunc<BFP:mode><DFP_ALL:mode>2")
5509         ("*trunc<DFP_ALL:mode><BFP:mode>2")
5510         ("trunc<BFP:mode><DFP_ALL:mode>2")
5511         ("trunc<DFP_ALL:mode><BFP:mode>2")
5512         ("*extend<BFP:mode><DFP_ALL:mode>2")
5513         ("*extend<DFP_ALL:mode><BFP:mode>2")
5514         ("extend<BFP:mode><DFP_ALL:mode>2")
5515         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5516         FPR4_REGNUM.
5518 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
5520         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5522 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5524         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5525         and cfun_fpr_bit_p to cfun_fpr_save_p.
5526         (s390_frame_area, s390_register_info, s390_frame_info)
5527         (s390_emit_prologue, s390_emit_epilogue)
5528         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5529         register numbers.
5530         * config/s390/s390.h: Define *_REGNUM macros for floating point
5531         register numbers.
5533 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
5535         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5537 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
5539         PR rtl-optimization/57786
5540         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5541         and break out of the loop when it is set to false.
5543 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
5545         PR target/57819
5546         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5547         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5548         (const_int 63)) 0)).
5549         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5550         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5551         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5553         PR rtl-optimization/57829
5554         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5555         mask bits outside of mode are just sign-extension from mode to HWI.
5557 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5559         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5560         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5561         adjust_address instead of change_address to keep info about alignment.
5562         (emit_strmov): Remove.
5563         (emit_memmov): New function.
5564         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5565         (expand_movmem_epilogue): Likewise and return updated rtx for
5566         destination.
5567         (expand_constant_movmem_prologue): Likewise and return updated rtx for
5568         destination and source.
5569         (decide_alignment): Refactor, handle vector_loop.
5570         (ix86_expand_movmem): Likewise.
5571         (ix86_expand_setmem): Likewise.
5572         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5574 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
5576         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5577         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5579 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5581         * config/i386/sse.md (sse_movlhps): Change alternative 3
5582         of operand 2 to "m".
5584 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5586         PR target/57807
5587         * config/i386/sse.md (iptr): New mode attribute.
5588         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5589         (<sse>_vm<plusminus_insn><mode>3): Ditto.
5590         (<sse>_vmmul<mode>3): Ditto.
5591         (<sse>_vmdiv<mode>3): Ditto.
5592         (sse_vmrcpv4sf2): Ditto.
5593         (<sse>_vmsqrt<mode>2): Ditto.
5594         (sse_vmrsqrtv4sf2): Ditto.
5595         (<sse>_vm<code><mode>3): Ditto.
5596         (avx_vmcmp<mode>3): Ditto.
5597         (<sse>_vmmaskcmp<mode>3): Ditto.
5598         (<sse>_comi): Ditto.
5599         (<sse>_ucomi): Ditto.
5600         (*xop_vmfrcz_<mode>): Ditto.
5601         (*fmai_fmadd_<mode>): Ditto.
5602         (*fmai_fmsub_<mode>): Ditto.
5603         (*fmai_fnmadd_<mode>): Ditto.
5604         (*fmai_fnmsub_<mode>): Ditto.
5605         (*fma4i_vmfmadd_<mode>): Ditto.
5606         (*fma4i_vmfmsub_<mode>): Ditto.
5607         (*fma4i_vmfnmadd_<mode>): Ditto.
5608         (*fma4i_vmfnmsub_<mode>): Ditto.
5609         (*xop_vmfrcz_<mode>): Ditto.
5610         (sse_cvtps2pi): Ditto.
5611         (sse_cvttps2pi): Ditto.
5612         (sse_cvtss2si): Ditto.
5613         (sse_cvtss2si_2): Ditto.
5614         (sse_cvtss2siq_2): Ditto.
5615         (sse_cvttss2si): Ditto.
5616         (sse_cvttss2siq): Ditto.
5617         (sse_cvtsd2si): Ditto.
5618         (sse_cvtsd2si_2): Ditto.
5619         (sse_cvtsd2siq_2): Ditto.
5620         (sse_cvttsd2si): Ditto.
5621         (sse_cvttsd2siq): Ditto.
5622         (sse_cvtsd2ss): Ditto.
5623         (sse_cvtss2sd): Ditto.
5624         (avx2_pbroadcast<mode>): Ditto.
5625         (avx2_pbroadcast<mode>_1): Ditto.
5626         (*avx_vperm_broadcast_v4sf): Ditto.
5628         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5629         (sse_movlhps): Ditto.
5630         (sse_storehps): Ditto.
5631         (sse_loadhps): Ditto.
5632         (sse_storelps): Ditto.
5633         (sse_loadlps): Ditto.
5634         (*vec_concatv4sf): Ditto.
5635         (*vec_interleave_highv2df): Ditto.
5636         (*vec_interleave_lowv2df): Ditto.
5637         (*vec_extractv2df_1_sse): Ditto.
5638         (*vec_extractv2df_0_sse): Ditto.
5639         (sse2_storelpd): Ditto.
5640         (sse2_loadlpd): Ditto.
5641         (sse2_movsd): Ditto.
5642         (*vec_concatv4si): Ditto.
5643         (vec_concatv2di): Ditto.
5645         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5646         for Intel asm dialect.
5647         (mmx_punpcklwd): Ditto.
5648         (mmx_punpckldq): Ditto.
5650         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5651         for intel assembler dialect.
5653 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
5655         PR target/29776
5656         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5657         for BUILT_IN_C{LZ,LRSB}*.
5658         * tree.h (CASE_INT_FN): Add FN##IMAX case.
5659         * tree-vrp.c (extract_range_basic): Handle
5660         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
5661         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5662         fall thru to code calling set_value*.
5663         * builtins.c (expand_builtin): Remove *IMAX cases.
5664         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5665         if width is bigger than 2*HWI.
5667 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
5669         PR rtl-optimization/55342
5670         * lra-int.h (lra_subreg_reload_pseudos): New.
5671         * lra.c: Add undoing optional reloads to the block diagram.
5672         (lra_subreg_reload_pseudos): New.
5673         (lra_optional_reload_pseudos): Change comments.
5674         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
5675         lra_optional_reload_pseudos after undo transformations.
5676         * lra-assigns.c (pseudo_prefix_title): New.
5677         (lra_setup_reg_renumber): Use it.
5678         (spill_for): Ditto.  Check subreg reload pseudos too.
5679         (assign_by_spills): Consider subreg reload pseudos too.
5680         * lra-constraints.c (simplify_operand_subreg): Use
5681         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5682         (curr_insn_transform): Recognize and do optional reloads.
5683         (undo_optional_reloads): New.
5684         (lra_undo_inheritance): Call undo_optional_reloads.
5686 2013-07-05  Thomas Quinot  <quinot@adacore.com>
5688         * tree-complex.c (expand_complex_operations_1): Fix typo.
5690 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
5692         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5693         (tune_params): New member 'const vec_costs'.
5694         * config/aarch64/aarch64.c (generic_vector_cost): New.
5695         (generic_tunings): New member 'generic_vector_cost'.
5696         (aarch64_builtin_vectorization_cost): New.
5697         (aarch64_add_stmt_cost): New.
5698         (TARGET_VECTORIZE_ADD_STMT_COST): New.
5699         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5701 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
5703         PR target/57777
5704         * config/i386/predicates.md (vsib_address_operand): Disallow
5705         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5707 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
5709         PR middle-end/55030
5710         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5711         expand_builtin_setjmp_receiver.
5712         (expand_label): Adjust, call expand_builtin_setjmp_receiver
5713         with NULL for the label parameter.
5714         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5715         the frame-pointer.  Adjust comments.
5716         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5717         only if LABEL is non-NULL.
5719 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
5721         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5722         (ARM_ABI_AAPCS64): Ditto.
5723         (arm_abi): Ditto.
5724         (ARM_DEFAULT_ABI): Ditto.
5726 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
5728         * config/aarch64/aarch64-builtins.c
5729         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5730         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5731         (st1): Likewise.
5732         * config/aarch64/aarch64-simd.md
5733         (aarch64_ld1<VALL:mode>): New.
5734         (aarch64_st1<VALL:mode>): Likewise.
5735         * config/aarch64/arm_neon.h
5736         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5738 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
5740         * config/i386/i386.c (gate_insert_vzeroupper): Check if
5741         target ISA is AVX.
5742         (ix86_option_override_internal):Turn on all -mavx target flags by
5743         default as they are dependent on AVX anyway.
5745 2013-07-02  Cary Coutant  <ccoutant@google.com>
5747         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5748         deterministic hash.
5749         (loc_checksum_ordered): Likewise.
5750         (hash_loc_operands): Remove inline keyword.
5752 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
5754         PR tree-optimization/57741
5755         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5756         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5757         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5758         Allow REAL_CST step_exprs if flag_associative_math.
5759         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5761 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5763         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5765 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5767         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5769 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5771         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5772         encoding.
5773         (iorsi3_insn): Likewise.
5774         (arm_xorsi3): Likewise.
5776 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
5778         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
5779         to ...
5780         (attribute "type"): ... here, and prefix with "wmmx_".
5781         (attribute "core_cycles"): Update for attribute changes.
5782         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5783         (tbcstv4hi): Likewise.
5784         (tbcstv2si): Likewise.
5785         (iwmmxt_iordi3): Likewise.
5786         (iwmmxt_xordi3): Likewise.
5787         (iwmmxt_anddi3): Likewise.
5788         (iwmmxt_nanddi3): Likewise.
5789         (iwmmxt_arm_movdi): Likewise.
5790         (iwmmxt_movsi_insn): Likewise.
5791         (mov<mode>_internal): Likewise.
5792         (and<mode>3_iwmmxt): Likewise.
5793         (ior<mode>3_iwmmxt): Likewise.
5794         (xor<mode>3_iwmmxt): Likewise.
5795         (add<mode>3_iwmmxt): Likewise.
5796         (ssaddv8qi3): Likewise.
5797         (ssaddv4hi3): Likewise.
5798         (ssaddv2si3): Likewise.
5799         (usaddv8qi3): Likewise.
5800         (usaddv4hi3): Likewise.
5801         (usaddv2si3): Likewise.
5802         (sub<mode>3_iwmmxt): Likewise.
5803         (sssubv8qi3): Likewise.
5804         (sssubv4hi3): Likewise.
5805         (sssubv2si3): Likewise.
5806         (ussubv8qi3): Likewise.
5807         (ussubv4hi3): Likewise.
5808         (ussubv2si3): Likewise.
5809         (mulv4hi3_iwmmxt): Likewise.
5810         (smulv4hi3_highpart): Likewise.
5811         (umulv4hi3_highpart): Likewise.
5812         (iwmmxt_wmacs): Likewise.
5813         (iwmmxt_wmacsz): Likewise.
5814         (iwmmxt_wmacu): Likewise.
5815         (iwmmxt_wmacuz): Likewise.
5816         (iwmmxt_clrdi): Likewise.
5817         (iwmmxt_clrv8qi): Likewise.
5818         (iwmmxt_clr4hi): Likewise.
5819         (iwmmxt_clr2si): Likewise.
5820         (iwmmxt_uavgrndv8qi3): Likewise.
5821         (iwmmxt_uavgrndv4hi3): Likewise.
5822         (iwmmxt_uavgv8qi3): Likewise.
5823         (iwmmxt_uavgv4hi3): Likewise.
5824         (iwmmxt_tinsrb): Likewise.
5825         (iwmmxt_tinsrh): Likewise.
5826         (iwmmxt_tinsrw): Likewise.
5827         (iwmmxt_textrmub): Likewise.
5828         (iwmmxt_textrmsb): Likewise.
5829         (iwmmxt_textrmuh): Likewise.
5830         (iwmmxt_textrmsh): Likewise.
5831         (iwmmxt_textrmw): Likewise.
5832         (iwmxxt_wshufh): Likewise.
5833         (eqv8qi3): Likewise.
5834         (eqv4hi3): Likewise.
5835         (eqv2si3): Likewise.
5836         (gtuv8qi3): Likewise.
5837         (gtuv4hi3): Likewise.
5838         (gtuv2si3): Likewise.
5839         (gtv8qi3): Likewise.
5840         (gtv4hi3): Likewise.
5841         (gtv2si3): Likewise.
5842         (smax<mode>3_iwmmxt): Likewise.
5843         (umax<mode>3_iwmmxt): Likewise.
5844         (smin<mode>3_iwmmxt): Likewise.
5845         (umin<mode>3_iwmmxt): Likewise.
5846         (iwmmxt_wpackhss): Likewise.
5847         (iwmmxt_wpackwss): Likewise.
5848         (iwmmxt_wpackdss): Likewise.
5849         (iwmmxt_wpackhus): Likewise.
5850         (iwmmxt_wpackwus): Likewise.
5851         (iwmmxt_wpackdus): Likewise.
5852         (iwmmxt_wunpckihb): Likewise.
5853         (iwmmxt_wunpckihh): Likewise.
5854         (iwmmxt_wunpckihw): Likewise.
5855         (iwmmxt_wunpckilb): Likewise.
5856         (iwmmxt_wunpckilh): Likewise.
5857         (iwmmxt_wunpckilw): Likewise.
5858         (iwmmxt_wunpckehub): Likewise.
5859         (iwmmxt_wunpckehuh): Likewise.
5860         (iwmmxt_wunpckehuw): Likewise.
5861         (iwmmxt_wunpckehsb): Likewise.
5862         (iwmmxt_wunpckehsh): Likewise.
5863         (iwmmxt_wunpckehsw): Likewise.
5864         (iwmmxt_wunpckelub): Likewise.
5865         (iwmmxt_wunpckeluh): Likewise.
5866         (iwmmxt_wunpckeluw): Likewise.
5867         (iwmmxt_wunpckelsb): Likewise.
5868         (iwmmxt_wunpckelsh): Likewise.
5869         (iwmmxt_wunpckelsw): Likewise.
5870         (ror<mode>3): Likewise.
5871         (ashr<mode>3_iwmmxt): Likewise.
5872         (lshr<mode>3_iwmmxt): Likewise.
5873         (ashl<mode>3_iwmmxt): Likewise.
5874         (ror<mode>3_di): Likewise.
5875         (ashr<mode>3_di): Likewise.
5876         (lshr<mode>3_di): Likewise.
5877         (ashl<mode>3_di): Likewise.
5878         (iwmmxt_wmadds): Likewise.
5879         (iwmmxt_wmaddu): Likewise.
5880         (iwmmxt_tmia): Likewise.
5881         (iwmmxt_tmiaph): Likewise.
5882         (iwmmxt_tmiabb): Likewise.
5883         (iwmmxt_tmiatb): Likewise.
5884         (iwmmxt_tmiabt): Likewise.
5885         (iwmmxt_tmiatt): Likewise.
5886         (iwmmxt_tmovmskb): Likewise.
5887         (iwmmxt_tmovmskh): Likewise.
5888         (iwmmxt_tmovmskw): Likewise.
5889         (iwmmxt_waccb): Likewise.
5890         (iwmmxt_wacch): Likewise.
5891         (iwmmxt_waccw): Likewise.
5892         (iwmmxt_waligni): Likewise.
5893         (iwmmxt_walignr): Likewise.
5894         (iwmmxt_walignr0): Likewise.
5895         (iwmmxt_walignr1): Likewise.
5896         (iwmmxt_walignr2): Likewise.
5897         (iwmmxt_walignr3): Likewise.
5898         (iwmmxt_wsadb): Likewise.
5899         (iwmmxt_wsadh): Likewise.
5900         (iwmmxt_wsadbz): Likewise.
5901         (iwmmxt_wsadhz): Likewise.
5902         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5903         (iwmmxt_wabsdiffb): Likewise.
5904         (iwmmxt_wabsdiffh): Likewise.
5905         (iwmmxt_wabsdiffw): Likewise.
5906         (iwmmxt_waddsubhx): Likewise
5907         (iwmmxt_wsubaddhx): Likewise.
5908         (addc<mode>3): Likewise.
5909         (iwmmxt_avg4): Likewise.
5910         (iwmmxt_avg4r): Likewise.
5911         (iwmmxt_wmaddsx): Likewise.
5912         (iwmmxt_wmaddux): Likewise.
5913         (iwmmxt_wmaddsn): Likewise.
5914         (iwmmxt_wmaddun): Likewise.
5915         (iwmmxt_wmulwsm): Likewise.
5916         (iwmmxt_wmulwum): Likewise.
5917         (iwmmxt_wmulsmr): Likewise.
5918         (iwmmxt_wmulumr): Likewise.
5919         (iwmmxt_wmulwsmr): Likewise.
5920         (iwmmxt_wmulwumr): Likewise.
5921         (iwmmxt_wmulwl): Likewise.
5922         (iwmmxt_wqmulm): Likewise.
5923         (iwmmxt_wqmulwm): Likewise.
5924         (iwmmxt_wqmulmr): Likewise.
5925         (iwmmxt_wqmulwmr): Likewise.
5926         (iwmmxt_waddbhusm): Likewise.
5927         (iwmmxt_waddbhusl): Likewise.
5928         (iwmmxt_wqmiabb): Likewise.
5929         (iwmmxt_wqmiabt): Likewise.
5930         (iwmmxt_wqmiatb): Likewise.
5931         (iwmmxt_wqmiatt): Likewise.
5932         (iwmmxt_wqmiabbn): Likewise.
5933         (iwmmxt_wqmiabtn): Likewise.
5934         (iwmmxt_wqmiatbn): Likewise.
5935         (iwmmxt_wqmiattn): Likewise.
5936         (iwmmxt_wmiabb): Likewise.
5937         (iwmmxt_wmiabt): Likewise.
5938         (iwmmxt_wmiatb): Likewise.
5939         (iwmmxt_wmiatt): Likewise.
5940         (iwmmxt_wmiabbn): Likewise.
5941         (iwmmxt_wmiabtn): Likewise.
5942         (iwmmxt_wmiatbn): Likewise.
5943         (iwmmxt_wmiattn): Likewise.
5944         (iwmmxt_wmiawbb): Likewise.
5945         (iwmmxt_wmiawbt): Likewise.
5946         (iwmmxt_wmiawtb): Likewise.
5947         (iwmmxt_wmiawtt): Likewise.
5948         (iwmmxt_wmiawbbn): Likewise.
5949         (iwmmxt_wmiawbtn): Likewise.
5950         (iwmmxt_wmiawtbn): Likewise.
5951         (iwmmxt_wmiawttn): Likewise.
5952         (iwmmxt_wmerge): Likewise.
5953         (iwmmxt_tandc<mode>3): Likewise.
5954         (iwmmxt_torc<mode>3): Likewise.
5955         (iwmmxt_torvsc<mode>3): Likewise.
5956         (iwmmxt_textrc<mode>3): Likewise.
5957         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5958         (wmmxt_pack): Likewise.
5959         (wmmxt_mult_c1): Likewise.
5960         (wmmxt_mult_c2): Likewise.
5961         (wmmxt_alu_c1): Likewise.
5962         (wmmxt_alu_c2): Likewise.
5963         (wmmxt_alu_c3): Likewise.
5964         (wmmxt_transfer_c1): Likewise.
5965         (wmmxt_transfer_c2): Likewise.
5966         (wmmxt_transfer_c3): Likewise.
5967         (marvell_f_iwmmxt_wstr): Likewise.
5968         (marvell_f_iwmmxt_wldr): Likewise.
5970 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5972         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5974 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5976         Revert:
5977         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5978         * lra-constraints.c (need_for_split_p): Check call used hard regs
5979         living through calls.
5981         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5982         call used regs for call insn.
5984 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5986         PR target/57736
5987         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5988         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5990 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5992         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5994 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5996         * lra-constraints.c (need_for_split_p): Check call used hard regs
5997         living through calls.
5999 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
6001         PR target/57744
6002         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
6003         to tie with any other modes.  Eliminate Altivec vector mode tests,
6004         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
6005         code, to return 0 if testing MODE2 for a condition, if we've
6006         already tested MODE1 for the same condition.
6008 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6010         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
6011         layout.
6013 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6015         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
6016         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
6018 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6020         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
6021         Define.
6022         (aarch64_symbolic_constant_p): Remove.
6023         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
6024         static.  Fix line length and white space.
6025         (aarch64_symbolic_constant_p): Remove.
6026         * config/aarch64/predicates.md (aarch64_valid_symref):
6027         Use aarch64_classify_symbol_expression.
6029 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6031         * config/arm/constraints.md (Ts): New constraint.
6032         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
6033         16-bit encodings.
6034         (compare_scc): Use "Ts" constraint for operand 0.
6035         (ior_scc_scc): Likewise.
6036         (and_scc_scc): Likewise.
6037         (and_scc_scc_nodom): Likewise.
6038         (ior_scc_scc_cmp): Likewise for operand 7.
6039         (and_scc_scc_cmp): Likewise.
6040         * config/arm/thumb2.md (thumb2_movsi_insn):
6041         Add alternatives for 16-bit encodings.
6042         (thumb2_movhi_insn): Likewise.
6043         (thumb2_movsicc_insn): Likewise.
6044         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
6045         (thumb2_negscc): Use "Ts" constraint.
6046         Move mvn instruction outside cond_exec block.
6047         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
6048         for 16-bit encodings.
6050 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6052         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
6053         encoding.
6054         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
6055         (mulsi3subsi): Likewise.
6056         (mulsidi3adddi): Likewise.
6057         (mulsidi3_v6): Likewise.
6058         (umulsidi3_v6): Likewise.
6059         (umulsidi3adddi_v6): Likewise.
6060         (smulsi3_highpart_v6): Likewise.
6061         (umulsi3_highpart_v6): Likewise.
6062         (mulhisi3tb): Likewise.
6063         (mulhisi3bt): Likewise.
6064         (mulhisi3tt): Likewise.
6065         (maddhisi4): Likewise.
6066         (maddhisi4tb): Likewise.
6067         (maddhisi4tt): Likewise.
6068         (maddhidi4): Likewise.
6069         (maddhidi4tb): Likewise.
6070         (maddhidi4tt): Likewise.
6071         (zeroextractsi_compare0_scratch): Likewise.
6072         (insv_zero): Likewise.
6073         (insv_t2): Likewise.
6074         (anddi_notzesidi_di): Likewise.
6075         (anddi_notsesidi_di): Likewise.
6076         (andsi_notsi_si): Likewise.
6077         (iordi_zesidi_di): Likewise.
6078         (xordi_zesidi_di): Likewise.
6079         (andsi_iorsi3_notsi): Likewise.
6080         (smax_0): Likewise.
6081         (smax_m1): Likewise.
6082         (smin_0): Likewise.
6083         (not_shiftsi): Likewise.
6084         (unaligned_loadsi): Likewise.
6085         (unaligned_loadhis): Likewise.
6086         (unaligned_loadhiu): Likewise.
6087         (unaligned_storesi): Likewise.
6088         (unaligned_storehi): Likewise.
6089         (extv_reg): Likewise.
6090         (extzv_t2): Likewise.
6091         (divsi3): Likewise.
6092         (udivsi3): Likewise.
6093         (arm_zero_extendhisi2addsi): Likewise.
6094         (arm_zero_extendqisi2addsi): Likewise.
6095         (compareqi_eq0): Likewise.
6096         (arm_extendhisi2_v6): Likewise.
6097         (arm_extendqisi2addsi): Likewise.
6098         (arm_movt): Likewise.
6099         (thumb2_ldrd): Likewise.
6100         (thumb2_ldrd_base): Likewise.
6101         (thumb2_ldrd_base_neg): Likewise.
6102         (thumb2_strd): Likewise.
6103         (thumb2_strd_base): Likewise.
6104         (thumb2_strd_base_neg): Likewise.
6105         (arm_negsi2): Add alternative for 16-bit encoding.
6106         (arm_one_cmplsi2): Likewise.
6108 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6110         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6111         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6112         (movdfcc): Likewise.
6113         * config/arm/vfp.md (*thumb2_movsf_vfp):
6114         Disable predication for arm_restrict_it.
6115         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6116         (*thumb2_movdfcc_vfp): Likewise.
6117         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6118         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6119         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6120         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6121         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6122         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6123         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6124         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6125         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6126         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6127         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6128         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6129         Disable predication for arm_restrict_it.
6131 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
6133         * config/i386/bmiintrin.h (_bextr_u32): New.
6134         (_bextr_u64): Ditto.
6136 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
6138         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6139         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6140         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6141         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6142         * config/mips/n32-elf.h: ...this new file.
6144 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
6146         PR target/57224
6147         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6148         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6150 2013-06-27  Catherine Moore  <clm@codesourcery.com>
6152         * config/mips/mips-tables.opt: Regenerate.
6153         * config/mips/mips-cpus.def: Add m14ke and m14kec.
6154         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6155         * doc/invoke.texi: Add -m14kc.
6157 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6159         PR target/57623
6160         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6161         constraints of operand 1 and 2.
6163         PR target/57623
6164         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6165         to match RTL canonicalization.  Swap predicates and
6166         constraints of operand 1 and 2.
6168 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
6170         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6171         Process OP_INOUT regs for splitting too.
6173 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6175         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6176         decl before the loop, initialize to NULL.
6177         (vectorizable_load): Initialize ptr_incr to NULL.
6179 2013-06-27  Martin Jambor  <mjambor@suse.cz>
6181         PR lto/57208
6182         * ipa-ref.h (ipa_maybe_record_reference): Declare.
6183         * ipa-ref.c (ipa_maybe_record_reference): New function.
6184         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6185         * ipa-cp.c (create_specialized_node): Record potential references from
6186         aggvals.
6187         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6189 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6191         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6192         parameter 'mode' of type 'enum machine_mode mode'; change to pass
6193         'mode' to force_reg.
6194         (aarch64_add_offset): Update calls to aarch64_force_temporary.
6195         (aarch64_expand_mov_immediate): Likewise.
6197 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6199         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6200         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6202 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6204         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6205         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6206         (struct machine_function): Add tbegin_p.
6207         (s390_canonicalize_comparison): Fold CC mode compares to
6208         conditional jump if possible.
6209         (s390_emit_jump): Return the emitted jump.
6210         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6211         Handle CCRAWmode compares.
6212         (s390_option_override): Default to -mhtm if available.
6213         (s390_reg_clobbered_rtx): Handle floating point regs as well.
6214         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6215         FPRs instead of df_regs_ever_live_p.
6216         (s390_optimize_nonescaping_tx): New function.
6217         (s390_init_frame_layout): Extend clobbered_regs array to cover
6218         FPRs as well.
6219         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6220         (s390_expand_tbegin): New function.
6221         (enum s390_builtin): New enum definition.
6222         (code_for_builtin): New array definition.
6223         (s390_init_builtins): New function.
6224         (s390_expand_builtin): New function.
6225         (TARGET_INIT_BUILTINS): Define.
6226         (TARGET_EXPAND_BUILTIN): Define.
6227         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6228         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6229         (s390_alc_comparison): Likewise.
6230         * config/s390/s390-modes.def: Add CCRAWmode.
6231         * config/s390/s390.h (processor_flags): Add PF_TX.
6232         (TARGET_CPU_HTM): Define macro.
6233         (TARGET_HTM): Define macro.
6234         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6235         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6236         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6237         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6238         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6239         (TBEGIN_MASK, TBEGINC_MASK): New constants.
6240         ("*cc_to_int"): Move up.
6241         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6242         constants other than 0.
6243         ("*ccraw_to_int"): New insn and splitter definition.
6244         ("tbegin", "tbegin_nofloat", "tbegin_retry")
6245         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6246         ("tx_assist"): New expander.
6247         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6248         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6249         * config/s390/s390.opt: Add -mhtm option.
6250         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6251         * config/s390/htmxlintrin.h: New file.
6252         * config/s390/htmintrin.h: New file.
6253         * config/s390/s390intrin.h: New file.
6254         * doc/extend.texi: Document htm builtins.
6255         * config.gcc: Add the new header files to extra_headers.
6257 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
6259         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6260         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6262 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
6263             Pat Haugen  <pthaugen@us.ibm.com>
6264             Peter Bergner  <bergner@vnet.ibm.com>
6266         * config/rs6000/power8.md: New.
6267         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6268         setting for power8 entry.
6269         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6270         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6271         test for Power4/Power5 only.
6272         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6273         support.
6274         (force_new_group): Adjust comment.
6275         * config/rs6000/rs6000.md: Include power8.md.
6277 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
6279         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6280         * config/arm/arm-protos.h (arm_max_conditional_execute): New
6281         declaration.
6282         (tune_params): Update comment.
6283         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6284         (arm_max_conditional_execute): New function.
6285         (thumb2_final_prescan_insn): Use max_insn_skipped and
6286         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6288 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
6290         PR tree-optimization/57705
6291         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6292         SSA_NAME step, provided that it is not defined inside the loop.
6293         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6294         (get_initial_def_for_induction): Handle SSA_NAME IV step.
6296 2013-06-25  Martin Jambor  <mjambor@suse.cz>
6298         PR middle-end/57670
6299         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6300         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6301         calls in the dump.
6302         (ipa_note_param_call): Initialize member_ptr flag.
6303         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6304         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6305         (ipa_write_indirect_edge_info): Stream member_ptr flag.
6306         (ipa_read_indirect_edge_info): Likewise.
6308 2013-06-25  Richard Biener  <rguenther@suse.de>
6310         PR middle-end/56977
6311         * passes.c (init_optimization_passes): Move pass_fold_builtins
6312         and pass_dce earlier with -Og.
6314 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
6316         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6317         <BIT_FIELD_REF>: Remove trailing TAB.
6318         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6319         remove blank line.
6321 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6323         PR tree-optimization/57358
6324         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6325         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6326         (ipa_analyze_params_uses): Generate pessimistic info when true.
6328 2013-06-24  Martin Jambor  <mjambor@suse.cz>
6330         PR tree-optimization/57539
6331         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6332         global.inlined_to of the new node to it.  All callers changed.
6333         * ipa-inline-transform.c (clone_inlined_nodes): New variable
6334         inlining_into, pass it to cgraph_clone_node.
6335         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6336         ipa_free_edge_args_substructures.
6337         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6338         rdesc linked list.  Do not assert rdesc edges have inlined caller.
6339         Assert we have found an rdesc in the rdesc list.
6341 2013-06-24  Richard Biener  <rguenther@suse.de>
6343         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6344         (pointer_set_lookup): Declare.
6345         (class pointer_map): New template class implementing a
6346         generic pointer to T map.
6347         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6348         pointer_map<T>::contains, pointer_map<T>::insert,
6349         pointer_map<T>::traverse): New functions.
6350         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6351         (pointer_set_lookup): New function.
6352         (pointer_set_contains): Use pointer_set_lookup.
6353         (pointer_set_insert): Likewise.
6354         (insert_aux): Remove.
6355         (struct pointer_map_t): Embed a pointer_set_t.
6356         (pointer_map_create): Adjust.
6357         (pointer_map_destroy): Likewise.
6358         (pointer_map_contains): Likewise.
6359         (pointer_map_insert): Likewise.
6360         (pointer_map_traverse): Likewise.
6361         * tree-streamer.h (struct streamer_tree_cache_d): Use a
6362         pointer_map<unsigned> instead of a pointer_map_t.
6363         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6364         (streamer_tree_cache_lookup): Likewise.
6365         (streamer_tree_cache_create): Likewise.
6366         (streamer_tree_cache_delete): Likewise.
6367         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6368         pointer_map<unsigned> instead of a pointer_map_t.
6369         (lto_init_tree_ref_encoder): Adjust.
6370         (lto_destroy_tree_ref_encoder): Likewise.
6371         * lto-section-out.c (lto_output_decl_index): Likewise.
6372         (lto_record_function_out_decl_state): Likewise.
6373         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6375 2013-06-24  Richard Biener  <rguenther@suse.de>
6377         PR tree-optimization/57488
6378         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6380 2013-06-24  Alan Modra  <amodra@gmail.com>
6382         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6383         (gen_easy_altivec_constant): Likewise.
6384         * config/rs6000/predicates.md (easy_vector_constant_add_self,
6385         easy_vector_constant_msb): Likewise.
6387 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
6389         PR target/57688
6390         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6391         add missing return true.
6393 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
6395         PR target/52483
6396         * config/sh/predicates.md (general_extend_operand): Invoke
6397         general_movsrc_operand for memory operands.
6398         (general_movsrc_operand): Allow reg+reg addressing, do not use
6399         general_operand for memory operands.
6401 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
6403         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
6404         when current target options does not apply.
6405         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6406         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6407         * config/i386/bmiintrin.h: Pass appropriate target
6408         attributes to header.
6409         * config/i386/mmintrin.h: Ditto.
6410         * config/i386/nmmintrin.h: Ditto.
6411         * config/i386/avx2intrin.h: Ditto.
6412         * config/i386/fxsrintrin.h: Ditto.
6413         * config/i386/tbmintrin.h: Ditto.
6414         * config/i386/xsaveintrin.h: Ditto.
6415         * config/i386/f16cintrin.h: Ditto.
6416         * config/i386/xtestintrin.h: Ditto.
6417         * config/i386/xsaveoptintrin.h: Ditto.
6418         * config/i386/bmi2intrin.h: Ditto.
6419         * config/i386/lzcntintrin.h: Ditto.
6420         * config/i386/smmintrin.h: Ditto.
6421         * config/i386/wmmintrin.h: Ditto.
6422         * config/i386/x86intrin.h: Remove all header include guards.
6423         * config/i386/prfchwintrin.h: Ditto.
6424         * config/i386/pmmintrin.h: Ditto.
6425         * config/i386/tmmintrin.h: Ditto.
6426         * config/i386/xmmintrin.h: Ditto.
6427         * config/i386/popcntintrin.h: Ditto.
6428         * config/i386/rdseedintrin.h: Ditto.
6429         * config/i386/ammintrin.h: Ditto.
6430         * config/i386/emmintrin.h: Ditto.
6431         * config/i386/immintrin.h: Remove all header include guards.
6432         * config/i386/fma4intrin.h: Ditto.
6433         * config/i386/lwpintrin.h: Ditto.
6434         * config/i386/xopintrin.h: Ditto.
6435         * config/i386/ia32intrin.h: Ditto.
6436         * config/i386/avxintrin.h: Ditto.
6437         * config/i386/rtmintrin.h: Ditto.
6438         * config/i386/fmaintrin.h: Ditto.
6439         * config/i386/mm3dnow.h: Ditto.
6441 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
6443         * common/config/i386/i386-common.c: Handle LZCNT.
6445 2013-06-22  Andi Kleen  <ak@linux.intel.com>
6447         * doc/extend.texi: Use __atomic_store_n instead of
6448         __atomic_store in HLE example.
6450 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
6452         * config/sh/sh.c: Remove <cstdlib> workaround.
6454 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6456         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
6458 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6460         * doc/extend.texi: Document that __atomic_clear and
6461           __atomic_test_and_set should only be used with bool.
6463 2013-06-20  Jan Hubicka  <jh@suse.cz>
6465         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6466         types_same_for_odr.
6467         * tree.c (decls_same_for_odr): New function.
6468         (same_for_edr): New function.
6469         (types_same_for_odr): New function.
6470         (get_binfo_at_offset): Use it.
6471         * tree.h (types_same_for_odr): Declare.
6473 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
6474             Jason Merrill  <jason@redhat.com>
6476         * system.h: Include <cstdlib> as well as <stdlib.h>.
6478 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
6480         PR target/57655
6481         * config/i386/i386.c (construct_container): Report error if
6482         long double is used with disabled x87 float returns.
6484 2013-06-20  Jan Hubicka  <jh@suse.cz>
6486         * lto-cgraph.c (input_symtab): Do not set cgraph state.
6488 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
6490         PR rtl-optimization/57425
6491         PR rtl-optimization/57569
6492         * alias.c (write_dependence_p): Remove parameters mem_mode and
6493         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
6494         Changed all callers.
6495         (canon_anti_dependence): Get comments and semantics in sync.
6496         Add parameter mem_canonicalized.  Changed all callers.
6497         * rtl.h (canon_anti_dependence): Update prototype.
6499 2013-06-20  Richard Biener  <rguenther@suse.de>
6501         * data-streamer-in.c (streamer_read_uhwi): Optimize single
6502         byte case, inline streamer_read_uchar and defer section
6503         overrun check.
6505 2013-06-20  Richard Biener  <rguenther@suse.de>
6507         PR tree-optimization/57584
6508         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6509         SSA names into the expanded expression that take part in
6510         abnormal coalescing.
6512 2013-06-19  Sharad Singhai  <singhai@google.com>
6514         * gcov.c (print_usage): Handle new option.
6515         (process_args): Ditto.
6516         (get_gcov_intermediate_filename): New function.
6517         (output_intermediate_file): New function.
6518         (output_gcov_file): New function
6519         (generate_results): Handle new option.
6520         (release_function): Relase demangled name.
6521         (read_graph_file): Handle demangled name.
6522         (output_lines): Ditto.
6523         * doc/gcov.texi: Document gcov intermediate format.
6525 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
6527         PR bootstrap/57604
6528         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6529         (lra_emit_add): Use the functions.  Add comment about Y as an
6530         address segment.
6532 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
6534         PR driver/57652
6535         * collect2.c (collect_atexit): New.
6536         (collect_exit): Delete.
6537         (main): Register collect_atexit with atexit.
6538         (collect_wait): Change collect_exit to exit.
6539         (do_wait): Same.
6540         * collect2.h (collect_exit): Delete.
6541         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6543 2013-06-19  Wei Mi  <wmi@google.com>
6545         PR rtl-optimization/57518
6546         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6547         if regno is used in paradoxical subreg.
6548         (update_equiv_regs): Check pdx_subregs[regno] before
6549         set a reg to be equivalent with a mem.
6551 2013-06-19  Matthias Klose  <doko@ubuntu.com>
6553         PR driver/57651
6554         * file-find.h (find_a_file): Add a mode parameter.
6555         * file-find.c (find_a_file): Likewise.
6556         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6557         with X_OK for the executables.
6558         * collect2.c (main): Call find_a_file with X_OK.
6560 2013-06-19  Steve Ellcey  <sellcey@mips.com>
6562         PR target/56942
6563         * config/mips/mips.md (casesi_internal_mips16_<mode>):
6564         Use NEXT_INSN instead of next_real_insn.
6566 2013-06-19  Jan Hubicka  <jh@suse.cz>
6568         * cgraph.h (const_value_known_p): Replace by ...
6569         (ctor_for_folding): .. this one.
6570         * cgraphunit.c (process_function_and_variable_attributes): Use it.
6571         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6572         * expr.c (expand_expr_real_1): Likewise.
6573         (string_constant): Likewise.
6574         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6575         * ipa.c (process_references): Likewise.
6576         (symtab_remove_unreachable_nodes): Likewise.
6577         * ipa-inline-analysis.c (param_change_prob): Likewise.
6578         * gimple-fold.c (canonicalize_constructor_val): Likewise.
6579         (get_base_constructor): Likwise.
6580         * varpool.c (varpool_remove_node): Likewise.
6581         (varpool_remove_initializer): LIkewise.
6582         (dump_varpool_node): LIkwise.
6583         (const_value_known_p): Rewrite to ...
6584         (ctor_for_folding): ... this one.
6586 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
6588         PR driver/57651
6589         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6590         PERSONALITY in $PATH derived prefixes.
6592 2013-06-19  Jeff Law  <law@redhat.com>
6594         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6595         in comment.
6597         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6598         (simplify_bitwise_binary): Use it to simpify certain binary ops on
6599         booleans.
6601 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
6603         * config/arm/vfp.md: Move VFP instruction classification documentation
6604         to ...
6605         * config/arm/arm.md: ... here.  Update instruction classification
6606         documentation.
6608 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
6610         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6611         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6612         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
6614 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
6616         PR target/57609
6617         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6618         with NEXT_INSN.  Use tablejump_p to check for jump table data
6619         insns.
6621 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
6623         PR c++/56544
6624         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6625         that now in C++ the value is correct per the C++ standards.
6627 2013-06-19  Richard Biener  <rguenther@suse.de>
6629         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6630         for global context.
6632 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6634         Revert:
6635         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6637         PR target/57609
6638         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6639         with next_active_insn.
6641 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
6643         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6644         functions are inlined during failures to flag an error.
6645         * tree-inline.c (expand_call_inline): Allow the error to be flagged
6646         in early inline pass.
6648 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
6650         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6651         in comments.
6653 2013-06-18  Julian Brown  <julian@codesourcery.com>
6655         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6656         Permit virtual register pre-reload if !strict.
6657         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6658         change.
6659         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6660         prototype.
6661         * config/arm/neon.md (movmisalign<mode>): Use
6662         neon_perm_struct_or_reg_operand instead of
6663         neon_struct_or_register_operand.
6664         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6665         neon_permissive_struct_operand instead of neon_struct_operand.
6666         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6667         neon_vector_mem_operand.
6668         * config/arm/predicates.md (neon_struct_operand): Adjust call to
6669         neon_vector_mem_operand.
6670         (neon_permissive_struct_operand): New.
6671         (neon_struct_or_register_operand): Rename to...
6672         (neon_perm_struct_or_reg_operand): This. Adjust call to
6673         neon_vector_mem_operand.
6675 2013-06-18  Richard Biener  <rguenther@suse.de>
6677         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6678         * lto-streamer.h: Include pointer-set.h.
6679         (struct lto_decl_slot): Remove.
6680         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6681         Remove next_index entry.
6682         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6683         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6684         (lto_init_tree_ref_encoder): Adjust.
6685         (lto_destroy_tree_ref_encoder): Likewise.
6686         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6687         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6688         (lto_output_decl_index): Adjust.
6689         (lto_new_out_decl_state): Likewise.
6690         (lto_record_function_out_decl_state): Likewise.
6691         * lto-streamer-out.c (copy_function): Likewise.
6693 2013-06-18  Richard Biener  <rguenther@suse.de>
6695         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6696         * cgraphunit.c: Include cfgloop.h.
6697         (init_lowered_empty_function): Initialize the loop tree.
6698         (assemble_thunk): Insert new BBs into loops.
6700 2013-06-18  Richard Biener  <rguenther@suse.de>
6702         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6703         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6704         the map from cache entry to cache index optional.
6705         (streamer_tree_cache_replace_tree): Adjust accordingly.
6706         (streamer_tree_cache_append): Likewise.
6707         (streamer_tree_cache_delete): Likewise.
6708         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6709         streamer cache map from cache entry to cache index.
6710         * lto-streamer-out.c (create_output_block): Adjust.
6712 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
6714         * config/arm/arm.md (attribute "insn"): Move multiplication and
6715         division attributes to...
6716         (attribute "type"): ... here.  Remove mult.
6717         (attribute "mul32"): New attribute.
6718         (attribute "mul64"): Add umaal.
6719         (*arm_mulsi3): Update attributes.
6720         (*arm_mulsi3_v6): Likewise.
6721         (*thumb_mulsi3): Likewise.
6722         (*thumb_mulsi3_v6): Likewise.
6723         (*mulsi3_compare0): Likewise.
6724         (*mulsi3_compare0_v6): Likewise.
6725         (*mulsi_compare0_scratch): Likewise.
6726         (*mulsi_compare0_scratch_v6): Likewise.
6727         (*mulsi3addsi): Likewise.
6728         (*mulsi3addsi_v6): Likewise.
6729         (*mulsi3addsi_compare0): Likewise.
6730         (*mulsi3addsi_compare0_v6): Likewise.
6731         (*mulsi3addsi_compare0_scratch): Likewise.
6732         (*mulsi3addsi_compare0_scratch_v6): Likewise.
6733         (*mulsi3subsi): Likewise.
6734         (*mulsidi3adddi): Likewise.
6735         (*mulsi3addsi_v6): Likewise.
6736         (*mulsidi3adddi_v6): Likewise.
6737         (*mulsidi3_nov6): Likewise.
6738         (*mulsidi3_v6): Likewise.
6739         (*umulsidi3_nov6): Likewise.
6740         (*umulsidi3_v6): Likewise.
6741         (*umulsidi3adddi): Likewise.
6742         (*umulsidi3adddi_v6): Likewise.
6743         (*smulsi3_highpart_nov6): Likewise.
6744         (*smulsi3_highpart_v6): Likewise.
6745         (*umulsi3_highpart_nov6): Likewise.
6746         (*umulsi3_highpart_v6): Likewise.
6747         (mulhisi3): Likewise.
6748         (*mulhisi3tb): Likewise.
6749         (*mulhisi3bt): Likewise.
6750         (*mulhisi3tt): Likewise.
6751         (maddhisi4): Likewise.
6752         (*maddhisi4tb): Likewise.
6753         (*maddhisi4tt): Likewise.
6754         (maddhidi4): Likewise.
6755         (*maddhidi4tb): Likewise.
6756         (*maddhidi4tt): Likewise.
6757         (divsi3): Likewise.
6758         (udivsi3): Likewise.
6759         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6760         (thumb2_mulsi_short_compare0): Likewise.
6761         (thumb2_mulsi_short_compare0_scratch): Likewise.
6762         * config/arm/arm1020e.md (1020mult1): Update attribute change.
6763         (1020mult2): Likewise.
6764         (1020mult3): Likewise.
6765         (1020mult4): Likewise.
6766         (1020mult5): Likewise.
6767         (1020mult6): Likewise.
6768         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6769         change.
6770         (cortex_a15_mult64): Likewise.
6771         (cortex_a15_sdiv): Likewise.
6772         (cortex_a15_udiv): Likewise.
6773         * config/arm/arm1026ejs.md (mult1): Update attribute change.
6774         (mult2): Likewise.
6775         (mult3): Likewise.
6776         (mult4): Likewise.
6777         (mult5): Likewise.
6778         (mult6): Likewise.
6779         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6780         (pj4_ir_div): Likewise.
6781         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6782         (11_mult2): Likewise.
6783         (11_mult3): Likewise.
6784         (11_mult4): Likewise.
6785         (11_mult5): Likewise.
6786         (11_mult6): Likewise.
6787         (11_mult7): Likewise.
6788         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6789         (cortex_a8_mla): Likewise.
6790         (cortex_a8_mull): Likewise.
6791         (cortex_a8_smulwy): Likewise.
6792         (cortex_a8_smlald): Likewise.
6793         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6794         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6795         (cortex_r4_mul_3): Likewise.
6796         (cortex_r4_mla_4): Likewise.
6797         (cortex_r4_mla_3): Likewise.
6798         (cortex_r4_smlald): Likewise.
6799         (cortex_r4_mull): Likewise.
6800         (cortex_r4_sdiv): Likewise.
6801         (cortex_r4_udiv): Likewise.
6802         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6803         (cortex_a7_idiv): Likewise.
6804         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6805         (9_mult2): Likewise.
6806         (9_mult3): Likewise.
6807         (9_mult4): Likewise.
6808         (9_mult5): Likewise.
6809         (9_mult6): Likewise.
6810         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6811         (cortex_a53_sdiv): Likewise.
6812         (cortex_a53_udiv): Likewise.
6813         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6814         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6815         (mp626_mult2): Likewise.
6816         (mp626_mult3): Likewise.
6817         (mp626_mult4): Likewise.
6818         * config/arm/fa526.md (526_mult1): Update attribute change.
6819         (526_mult2): Likewise.
6820         * config/arm/arm-generic.md (mult): Update attribute change.
6821         (mult_ldsched_strongarm): Likewise.
6822         (mult_ldsched): Likewise.
6823         (multi_cycle): Likewise.
6824         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6825         * config/arm/fa606te.md (606te_mult1): Update attribute change.
6826         (606te_mult2): Likewise.
6827         (606te_mult3): Likewise.
6828         (606te_mult4): Likewise.
6829         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6830         (cortex_a9_mac16): Likewise.
6831         (cortex_a9_multiply): Likewise.
6832         (cortex_a9_mac): Likewise.
6833         (cortex_a9_multiply_long): Likewise.
6834         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6835         (626te_mult2): Likewise.
6836         (626te_mult3): Likewise.
6837         (626te_mult4): Likewise.
6839 2013-06-18  Richard Biener  <rguenther@suse.de>
6841         PR lto/57334
6842         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6844 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6846         PR target/57609
6847         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6848         with next_active_insn.
6850 2013-06-18  Alan Modra  <amodra@gmail.com>
6852         * config/rs6000/rs6000.h (enum data_align): New.
6853         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6854         (DATA_ABI_ALIGNMENT): Define.
6855         (CONSTANT_ALIGNMENT): Correct comment.
6856         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6857         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6859 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6861         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6862         ATTRIBUTE_UNUSED marking.
6864 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6866         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6867         alternative and update.
6868         (aarch64_dup_lanedi): Delete.
6869         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6870         * config/aarch64/aarch64-simd-builtins.def: Update.
6872 2013-06-17  Richard Biener  <rguenther@suse.de>
6874         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6875         (lto_input_scc): Declare.
6876         (lto_input_tree_1): Likewise.
6877         (struct lto_stats_d): Add num_tree_bodies_output and
6878         num_pickle_refs_output.
6879         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6880         (lto_read_tree_1): Split out from ...
6881         (lto_read_tree): ... this.
6882         (lto_input_scc): New function.
6883         (lto_input_tree_1): Split out from ...
6884         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6885         (lto_data_in_create): Create the streamer cache without hashes.
6886         * lto-streamer-out.c (create_output_block): Create the streamer
6887         cache with hashes when not doing WPA.
6888         (lto_write_tree_1): Split out from ...
6889         (lto_write_tree): ... this.
6890         (get_symbol_initial_value): New function.
6891         (lto_output_tree_1): Split out from ...
6892         (lto_output_tree): ... this.  Write trees as series of SCCs
6893         using a DFS walk via DFS_write_tree.
6894         (struct sccs, struct scc_entry): New types.
6895         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6896         (DFS_write_tree_body): New function.
6897         (DFS_write_tree): Likewise.
6898         (hash_tree): Likewise.
6899         (scc_entry_compare): Likewise.
6900         (hash_scc): Likewise.
6901         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6902         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6903         TREE_CHAIN as regular reference.
6904         (streamer_read_integer_cst): Remove.
6905         (streamer_get_pickled_tree): Adjust.
6906         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6907         of DECL_EXTERNALs in BLOCK_VARS for now.
6908         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6909         reference.
6910         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6911         Add hash value argument and record that if hashes are recorded
6912         in the cache.
6913         (streamer_tree_cache_insert_1): Adjust.
6914         (streamer_tree_cache_insert): Likewise.
6915         (streamer_tree_cache_insert_at): Rename to ...
6916         (streamer_tree_cache_replace_tree): ... this and adjust.
6917         (streamer_tree_cache_append): Adjust.
6918         (record_common_node): Likewise.
6919         (streamer_tree_cache_create): Add argument whether to
6920         record hash values together with trees.
6921         (streamer_tree_cache_delete): Adjust.
6922         * tree-streamer.h (struct streamer_tree_cache_d): Add
6923         vector of hashes.
6924         (streamer_read_integer_cst): Remove.
6925         (streamer_tree_cache_insert): Adjust.
6926         (streamer_tree_cache_append): Likewise.
6927         (streamer_tree_cache_insert_at): Rename to ...
6928         (streamer_tree_cache_replace_tree): ... this and adjust.
6929         (streamer_tree_cache_create): Add argument whether to record hashes.
6930         (streamer_tree_cache_get): Rename to ...
6931         (streamer_tree_cache_get_tree): ... this.
6932         (streamer_tree_cache_get_hash): New function.
6933         * tree.c (cache_integer_cst): New function.
6934         * tree.h (cache_integer_cst): Declare.
6935         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6936         * lto-symtab.c (lto_varpool_replace_node): Only release
6937         DECL_INITIAL of non-prevailing decls.
6938         * varpool.c (varpool_remove_initializer): Do not release
6939         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6941 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6943         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6944         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6945         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6946         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6947         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6948         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6949         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6950         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6951         instead of TARGET_64BIT.
6952         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6953         Require ISA_HAS_<D>DIV.
6955 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6957         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6958         (mips*-*-linux*): Move default with_llsc setting to where other
6959         defaults are set.
6960         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6961         with_arch block.
6962         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6963         Likewise.  Remove default with_tune setting.  Move default float
6964         setting to its own block.  Handle with_llsc in the same block as above.
6966 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
6968         PR rtl-optimization/57425
6969         PR rtl-optimization/57569
6970         * alias.c (write_dependence_p): Add new parameters mem_mode,
6971         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6972         Changed all callers.
6973         (canon_anti_dependence): New function.
6974         * cse.c (check_dependence): Use canon_anti_dependence.
6975         * cselib.c (cselib_invalidate_mem): Likewise.
6976         * rtl.h (canon_anti_dependence): Declare.
6978 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6980         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6981         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6982         ".set mips3" for 64-bit targets.
6984 2013-06-15  Dehao Chen  <dehao@google.com>
6986         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6987         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6988         (gimple_check_call_args): Likewise.
6989         * value-prof.c (check_ic_target): Likewise.
6990         * ipa-inline.c (early_inliner): Likewise.
6991         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6992         * cgraph.c (cgraph_create_edge_1): Likewise.
6993         (cgraph_make_edge_direct): Likewise.
6995 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6997         PR target/57615
6998         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6999         rs6000_output_move_128bit to handle emitting quad memory
7000         operations.  Set attribute length to 8 bytes.
7002 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
7004         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
7005         New pattern.
7006         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
7007         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
7008         (aarch64_<su>mlsl<mode>): Likewise.
7010 2013-06-14  Mike Stump  <mikestump@comcast.net>
7012         * Makefile.in (TARGET_H): Add insn-codes.h.
7014 2013-06-14  Alan Modra  <amodra@gmail.com>
7016         PR middle-end/57134
7017         PR middle-end/57586
7018         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
7019         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
7020         bitfield expansion when EXPAND_MEMORY.
7021         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
7023 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
7025         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
7026         test for clearing quad memory on 32-bit later.
7028 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
7030         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
7031         (fold_negate_expr): Likewise.
7032         (fold_real_zero_addition_p): Handle vectors.
7033         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
7035 2013-06-14  Alan Modra  <amodra@gmail.com>
7037         * varasm.c (force_const_mem): Revert 2013-06-07 change.
7039 2013-06-13  Jan Hubicka  <jh@suse.cz>
7041         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
7042         Local comdats are not externally visible.
7043         * symtab.c (dump_symtab_base): Dump externally visible.
7044         (verify_symtab_base): Verify back links in the symtab hash.
7046 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
7048         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
7049         CONVERT_EXPR as equal nodes.
7051 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
7053         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
7055 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
7057         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
7058         Generalize to complex and vector.
7059         * tree.c (build_all_ones_cst): New function.
7060         * tree.h (build_all_ones_cst): Declare it.
7062 2013-06-13  Alan Modra  <amodra@gmail.com>
7064         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
7065         * config/rs6000/rs6000.md (signbittf2): New insn.
7066         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
7067         (abstf2_internal, cmptf_internal2): Likewise.
7068         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
7070 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
7071             Pat Haugen  <pthaugen@us.ibm.com>
7072             Peter Bergner  <bergner@vnet.ibm.com>
7074         * config/rs6000/rs6000.c (emit_load_locked): Add support for
7075         power8 byte, half-word, and quad-word atomic instructions.
7076         (emit_store_conditional): Likewise.
7077         (rs6000_expand_atomic_compare_and_swap): Likewise.
7078         (rs6000_expand_atomic_op): Likewise.
7080         * config/rs6000/sync.md (larx): Add new modes for power8.
7081         (stcx): Likewise.
7082         (AINT): New mode iterator to include TImode as well as normal
7083         integer modes on power8.
7084         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
7085         that VSX registers are not considered.  Use AINT mode iterator
7086         instead of INT1 to allow inclusion of quad word atomic operations
7087         on power8.
7088         (load_locked<mode>): Likewise.
7089         (store_conditional<mode>): Likewise.
7090         (atomic_compare_and_swap<mode>): Likewise.
7091         (atomic_exchange<mode>): Likewise.
7092         (atomic_nand<mode>): Likewise.
7093         (atomic_fetch_<fetchop_name><mode>): Likewise.
7094         (atomic_nand_fetch<mode>): Likewise.
7095         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7096         each type.
7097         (ATOMIC): On power8, add QImode, HImode modes.
7098         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7099         modes that promote to SImode.
7100         (load_lockedti): Convert TImode arguments to PTImode, so that we
7101         get a guaranteed even/odd register pair.
7102         (load_lockedpti): Likewise.
7103         (store_conditionalti): Likewise.
7104         (store_conditionalpti): Likewise.
7106         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7107         atomic load/store instructions.
7108         (HSI): Likewise.
7110 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
7112         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7113         loads.
7114         (insn_count): New attribute, with most cases extracted from...
7115         (length): ...here.  Redefine most cases in terms of insn_count.
7116         (single_insn): Delete.
7117         (can_delay): Use insn_count to check for single instructions.
7118         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7119         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7120         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7121         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7122         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7123         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7124         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7125         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7126         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7127         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7128         rather than "length".
7129         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
7130         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7131         Use "insn_count" rather than "length".
7132         * config/mips/mips-dsp.md
7133         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7134         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7135         length attributes.
7137 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
7139         PR tree-optimization/57361
7140         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7142 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
7144         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7145         to split.
7146         (aarch64_simd_combine<mode>): New instruction expansion.
7147         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7148         function prototype.
7149         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7150         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7152 2013-06-12  Jan Hubicka  <jh@suse.cz>
7154         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7155         decl has when in streaming stage.
7156         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7157         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7159 2013-06-12  Roland Stigge  <stigge@antcom.de>
7161         PR target/57578
7162         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7164 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7166         PR tree-optimization/57537
7167         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7168         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7170 2013-06-12  Richard Biener  <rguenther@suse.de>
7172         * data-streamer.h (streamer_write_char_stream): CSE
7173         obs->current_pointer.
7174         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7175         streamer_write_char_stream manually and optimize the resulting loop.
7176         (streamer_write_hwi_stream): Likewise.
7178 2013-06-12  Jan Hubicka  <jh@suse.cz>
7180         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7181         * cgraph.h (varpool_create_empty_node): Declare.
7182         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7183         duplicated nodes.
7184         * symtab.c (symtab_unregister_node): Be lax about missin entries
7185         in node hash.
7186         (symtab_get_node): Update comment.
7187         * varpool.c (varpool_create_empty_node): Break out from ...
7188         (varpool_node_for_decl): ... here.
7189         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7191 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
7193         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7194         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7195         part.  Use straight-line flow at the end.
7196         <COMPONENT_REF>: Remove superfluous else.
7197         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7199 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7201         PR target/56564
7202         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7203         target hook even for !TREE_PUBLIC decls.  If no resolution info
7204         is available, return false for common and external decls.
7206 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7208         * config/rl78/constraints.md (U): New constraint.
7209         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7210         valloc attribute.
7212 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7214         PR target/57589
7215         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7216         to allow returning address to AT_PLATFORM name.
7218 2013-06-11  Jan Hubicka  <jh@suse.cz>
7220         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7221         * cgraph.h (symtab_node_base): Add weakref flag.
7222         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7223         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7224         (output_weakrefs): Use weakref flag.
7225         * fold-const.c (simple_operand_p): Handle WEAK.
7226         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7227         * ipa.c (varpool_externally_visible_p): Drop weakref.
7228         (function_and_variable_visibility): Update comment; fix weakref
7229         sanity checks; do not clear DECL_WEAK on them.
7230         * lto-cgraph.c (lto_output_node): update.
7231         (lto_output_varpool_node): Update.
7232         (input_overwrite_node): Update.
7233         (input_node): Update.
7234         (input_varpool_node): Update.
7235         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7236         (lto_symtab_merge_symbols): Add sanity check.
7237         (lto_symtab_prevailing_decl): Do not special case weakrefs.
7238         * passes.c (rest_of_decl_compilation): Set static flag, too.
7239         * symtab.c (dump_symtab_base): Dump weakref.
7240         (verify_symtab_base): Sanity check weakrefs.
7241         (symtab_make_decl_local): Remove duplicated code.
7242         (symtab_alias_ultimate_target): Simplify.
7243         * varpool.c (varpool_create_variable_alias): Set weakref flag.
7245 2013-06-11  DJ Delorie  <dj@redhat.com>
7247         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7248         (rl78_unwind_word_mode): New.
7250 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7252         * final.c (debug_prefix_maps): Make static.
7254 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7256         * function.c (initial_trampoline): Remove stray copy.
7258 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
7260         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7262 2013-06-11  Martin Jambor  <mjambor@suse.cz>
7264         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7265         within bounds at the beginning of the function.
7267 2013-06-11  Alan Modra  <amodra@gmail.com>
7269         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
7270         reporting.
7271         (get_named_section): Don't NULL !DECL_P decl.
7273 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
7275         * doc/invoke.texi (core-avx2): Document.
7276         (slm): Likewise.
7277         (atom): Updated with MOVBE.
7279 2013-06-11  Richard Biener  <rguenther@suse.de>
7281         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7283 2013-06-11  Anton Blanchard  <anton@samba.org>
7285         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7286         correct shift value in little-endian mode.
7288 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
7290         PR target/56564
7291         * varasm.c (get_variable_align): Move #endif to the right place.
7293 2013-06-10  Cary Coutant  <ccoutant@google.com>
7295         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7296         for hash so that hash table traversal order is deterministic.
7298 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
7299             Pat Haugen  <pthaugen@us.ibm.com>
7300             Peter Bergner  <bergner@vnet.ibm.com>
7302         * config/rs6000/vector.md (GPR move splitter): Do not split moves
7303         of vectors in GPRS if they are direct moves or quad word load or
7304         store moves.
7306         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7307         declaration.
7308         (direct_move_p): Likewise.
7309         (quad_load_store_p): Likewise.
7311         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7312         classes into bins based on the physical register type.
7313         (reg_class_to_reg_type): Likewise.
7314         (IS_STD_REG_TYPE): Likewise.
7315         (IS_FP_VECT_REG_TYPE): Likewise.
7316         (reload_fpr_gpr): Arrays to determine what insn to use if we can
7317         use direct move instructions.
7318         (reload_gpr_vsx): Likewise.
7319         (reload_vsx_gpr): Likewise.
7320         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7321         information that is a simplification of register classes.  Also
7322         precalculate direct move reload helpers.
7323         (direct_move_p): New function to return true if the operation can
7324         be done as a direct move instruciton.
7325         (quad_load_store_p): New function to return true if the operation
7326         is a quad memory operation.
7327         (rs6000_legitimize_address): If quad memory, only allow register
7328         indirect for TImode addresses.
7329         (rs6000_legitimate_address_p): Likewise.
7330         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7331         (rs6000_reload_register_type): Likewise.
7332         (register_to_reg_type): Return register type.
7333         (rs6000_secondary_reload_simple_move): New helper function for
7334         secondary reload and secondary memory needed to identify anything
7335         that is a simple move, and does not need reloading.
7336         (rs6000_secondary_reload_direct_move): New helper function for
7337         secondary reload to identify cases that can be done with several
7338         instructions via the direct move instructions.
7339         (rs6000_secondary_reload_move): New helper function for secondary
7340         reload to identify moves between register types that can be done.
7341         (rs6000_secondary_reload): Add support for quad memory operations
7342         and for direct move.
7343         (rs6000_secondary_memory_needed): Likewise.
7344         (rs6000_debug_secondary_memory_needed): Change argument names.
7345         (rs6000_output_move_128bit): New function to return the move to
7346         use for 128-bit moves, including knowing about the various
7347         limitations of quad memory operations.
7349         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7350         memory operations.  call rs6000_output_move_128bit for the actual
7351         instruciton(s) to generate.
7352         (vsx_movti_64bit): Likewise.
7354         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7355         (UNSPEC_P8V_MTVSRWZ): Likewise.
7356         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7357         (UNSPEC_P8V_MTVSRD): Likewise.
7358         (UNSPEC_P8V_XXPERMDI): Likewise.
7359         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7360         (UNSPEC_FUSION_GPR): Likewise.
7361         (FMOVE128_GPR): New iterator for direct move.
7362         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7363         (f32_sv): Likewise.
7364         (f32_dm): Likewise.
7365         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7366         loads and direct move instructions.
7367         (zero_extendsidi2_lfiwzx): Likewise.
7368         (extendsidi2_lfiwax): Likewise.
7369         (extendsidi2_nocell): Likewise.
7370         (floatsi<mode>2_lfiwax): Likewise.
7371         (lfiwax): Likewise.
7372         (floatunssi<mode>2_lfiwzx): Likewise.
7373         (lfiwzx): Likewise.
7374         (fix_trunc<mode>_stfiwx): Likewise.
7375         (fixuns_trunc<mode>_stfiwx): Likewise.
7376         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7377         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7378         (parity<mode>2_cmpb): Set length/type attr.
7379         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7380         for 'mr.' to fast_compare.
7381         (bpermd_<mode>): Change type attr to popcnt.
7382         (p8_fmrgow_<mode>): New insns for power8 direct move support.
7383         (p8_mtvsrwz_1): Likewise.
7384         (p8_mtvsrwz_2): Likewise.
7385         (reload_fpr_from_gpr<mode>): Likewise.
7386         (p8_mtvsrd_1): Likewise.
7387         (p8_mtvsrd_2): Likewise.
7388         (p8_xxpermdi_<mode>): Likewise.
7389         (reload_vsx_from_gpr<mode>): Likewise.
7390         (reload_vsx_from_gprsf): Likewise.
7391         (p8_mfvsrd_3_<mode>): LIkewise.
7392         (reload_gpr_from_vsx<mode>): Likewise.
7393         (reload_gpr_from_vsxsf): Likewise.
7394         (p8_mfvsrd_4_disf): Likewise.
7395         (multi-word GPR splits): Do not split direct moves or quad memory
7396         operations.
7398 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
7400         * tree-into-ssa.c (interesting_blocks): Make static.
7402 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
7404         PR target/56564
7405         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7406         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7407         Use DATA_ABI_ALIGNMENT for that case instead if defined.
7408         (get_variable_align): New function.
7409         (get_variable_section, emit_bss, emit_common,
7410         assemble_variable_contents, place_block_symbol): Use
7411         get_variable_align instead of DECL_ALIGN.
7412         (assemble_noswitch_variable): Add align argument, use it
7413         instead of DECL_ALIGN.
7414         (assemble_variable): Adjust caller.  Use get_variable_align
7415         instead of DECL_ALIGN.
7416         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7417         caller.
7418         (DATA_ABI_ALIGNMENT): Define.
7419         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7420         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
7421         opt is false, only return the psABI mandated alignment increase.
7422         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7423         (DATA_ABI_ALIGNMENT): ... this.
7424         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7425         (DATA_ABI_ALIGNMENT): ... this.
7426         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7427         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7428         (DATA_ABI_ALIGNMENT): ... this.
7429         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7430         * doc/tm.texi: Regenerated.
7432 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
7434         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7435         cmp_code to construct REG_EQUAL note.
7437 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
7439         PR target/57568
7440         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7441         that operands[2] doesn't overlap with operands[0].
7443 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
7444             Jan Hubicka  <jh@suse.cz>
7446         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7447         hack to mark symbols as used.
7449 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
7451         PR rtl-optimization/57559
7452         * lra-constraints.c (process_alt_operands): Don't discourage
7453         memory with known offset for offsetable memory constraint.
7454         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7456 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
7458         * varasm.c (struct oc_local_state): Reorder fields.
7459         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7460         and adjust accordingly.
7461         (output_constructor): Reorder initialization code and adjust call to
7462         output_constructor_bitfield.
7464 2013-06-07  Jan Hubicka  <jh@suse.cz>
7466         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7468 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
7470         * tree-object-size.c (unknown): Make const.
7472 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7474         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7475         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7476         for last alternative in the cpu_facility attribute.
7478 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7480         PR target/56315
7481         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7482         (xordi3): Change operand 2 constraint to arm_xordi_operand.
7483         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7484         * config/arm/constraints.md (Dg): New constraint.
7485         * config/arm/neon.md (xordi3_neon): Remove.
7486         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7487         * config/arm/predicates.md (arm_xordi_operand): New predicate.
7489 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7491         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7492         Clean up alternatives.
7494 2013-06-07  Alan Modra  <amodra@gmail.com>
7496         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7497         va_list_gpr_size.
7499 2013-06-07  Alan Modra  <amodra@gmail.com>
7501         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7503 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7505         * config/arm/constraints.md (Df): New constraint.
7506         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7507         Correct length attribute for last two alternatives.
7509 2013-06-07  Alan Modra  <amodra@gmail.com>
7511         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7512         override user -mfp-in-toc.
7513         (offsettable_ok_by_alignment): Consider just the current access
7514         rather than the whole object, unless BLKmode.  Handle
7515         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7516         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7517         for -mcmodel=medium.
7518         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7519         override user -mfp-in-toc or -msum-in-toc.  Default to
7520         -mno-fp-in-toc for -mcmodel=medium.
7522 2013-06-06  DJ Delorie  <dj@redhat.com>
7524         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7525         TARGET_VALID_POINTER_MODE.
7527 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
7528             Pat Haugen  <pthaugen@us.ibm.com>
7529             Peter Bergner  <bergner@vnet.ibm.com>
7531         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7532         Document new power8 builtins.
7534         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7535         condition code register, to allow 128-bit logical operations to be
7536         done in the VSX or GPR registers.
7537         (nor<mode>3): Use the canonical form for nor.
7538         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7539         vclz*, and vpopcnt* vector instructions.
7540         (nand<mode>3): Likewise.
7541         (orc<mode>3): Likewise.
7542         (clz<mode>2): LIkewise.
7543         (popcount<mode>2): Likewise.
7545         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7546         that only the GPRs are recognized.
7548         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7549         support for new power8 builtins.
7551         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7552         builtin functions.
7553         (xscvdpspn): Likewise.
7554         (vclz): Likewise.
7555         (vclzb): Likewise.
7556         (vclzh): Likewise.
7557         (vclzw): Likewise.
7558         (vclzd): Likewise.
7559         (vpopcnt): Likewise.
7560         (vpopcntb): Likewise.
7561         (vpopcnth): Likewise.
7562         (vpopcntw): Likewise.
7563         (vpopcntd): Likewise.
7564         (vgbbd): Likewise.
7565         (vmrgew): Likewise.
7566         (vmrgow): Likewise.
7567         (eqv): Likewise.
7568         (eqv_v16qi3): Likewise.
7569         (eqv_v8hi3): Likewise.
7570         (eqv_v4si3): Likewise.
7571         (eqv_v2di3): Likewise.
7572         (eqv_v4sf3): Likewise.
7573         (eqv_v2df3): Likewise.
7574         (nand): Likewise.
7575         (nand_v16qi3): Likewise.
7576         (nand_v8hi3): Likewise.
7577         (nand_v4si3): Likewise.
7578         (nand_v2di3): Likewise.
7579         (nand_v4sf3): Likewise.
7580         (nand_v2df3): Likewise.
7581         (orc): Likewise.
7582         (orc_v16qi3): Likewise.
7583         (orc_v8hi3): Likewise.
7584         (orc_v4si3): Likewise.
7585         (orc_v2di3): Likewise.
7586         (orc_v4sf3): Likewise.
7587         (orc_v2df3): Likewise.
7589         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7590         allow power8 quad mode in 64-bit.
7591         (rs6000_builtin_vectorized_function): Add support to vectorize
7592         ISA 2.07 count leading zeros, population count builtins.
7593         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7594         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7595         (builtin_function_type): Add vgbbd builtin function which takes an
7596         unsigned argument.
7597         (altivec_expand_vec_perm_const): Add support for new power8 merge
7598         instructions.
7600         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7601         that does not include TImdoe for use with 32-bit.
7602         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7603         instructions.
7604         (UNSPEC_VSX_CVDPSPN): Likewise.
7605         (vsx_xscvdpspn): Likewise.
7606         (vsx_xscvspdpn): Likewise.
7607         (vsx_xscvdpspn_scalar): Likewise.
7608         (vsx_xscvspdpn_directmove): Likewise.
7609         (vsx_and<mode>3): Split logical operations into 32-bit and
7610         64-bit. Add support to do logical operations on TImode as well as
7611         VSX vector types.  Allow logical operations to be done in either
7612         VSX registers or in general purpose registers in 64-bit mode.  Add
7613         splitters if GPRs were used. For AND, add clobber of CCmode to
7614         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
7615         encoding.
7616         (vsx_and<mode>3_32bit): Likewise.
7617         (vsx_and<mode>3_64bit): Likewise.
7618         (vsx_ior<mode>3): Likewise.
7619         (vsx_ior<mode>3_32bit): Likewise.
7620         (vsx_ior<mode>3_64bit): Likewise.
7621         (vsx_xor<mode>3): Likewise.
7622         (vsx_xor<mode>3_32bit): Likewise.
7623         (vsx_xor<mode>3_64bit): Likewise.
7624         (vsx_one_cmpl<mode>2): Likewise.
7625         (vsx_one_cmpl<mode>2_32bit): Likewise.
7626         (vsx_one_cmpl<mode>2_64bit): Likewise.
7627         (vsx_nor<mode>3): Likewise.
7628         (vsx_nor<mode>3_32bit): Likewise.
7629         (vsx_nor<mode>3_64bit): Likewise.
7630         (vsx_andc<mode>3): Likewise.
7631         (vsx_andc<mode>3_32bit): Likewise.
7632         (vsx_andc<mode>3_64bit): Likewise.
7633         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7634         and xxlorc instructions.
7635         (vsx_eqv<mode>3_64bit): Likewise.
7636         (vsx_nand<mode>3_32bit): Likewise.
7637         (vsx_nand<mode>3_64bit): Likewise.
7638         (vsx_orc<mode>3_32bit): Likewise.
7639         (vsx_orc<mode>3_64bit): Likewise.
7641         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7643         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7644         instruction.
7645         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7646         (p8_vmrgow): Likewise.
7647         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7648         GPRs to be split under VSX.
7649         (p8v_clz<mode>2): Add power8 count leading zero support.
7650         (p8v_popcount<mode>2): Add power8 population count support.
7651         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7652         support.
7654         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7655         instruction.
7657         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7658         builtin functions.
7659         (vec_nand): Likewise.
7660         (vec_vclz): Likewise.
7661         (vec_vclzb): Likewise.
7662         (vec_vclzd): Likewise.
7663         (vec_vclzh): Likewise.
7664         (vec_vclzw): Likewise.
7665         (vec_vgbbd): Likewise.
7666         (vec_vmrgew): Likewise.
7667         (vec_vmrgow): Likewise.
7668         (vec_vpopcnt): Likewise.
7669         (vec_vpopcntb): Likewise.
7670         (vec_vpopcntd): Likewise.
7671         (vec_vpopcnth): Likewise.
7672         (vec_vpopcntw): Likewise.
7674 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7676         PR rtl-optimization/57468
7677         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7678         spilled pseudos.
7680 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7682         PR rtl-optimization/57459
7683         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7684         type when setting live regs.
7686 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7688         * config/s390/s390.opt (mlra): New option.
7689         * config/s390/s390.c (s390_decompose_address): Check displacement
7690         for all registers for LRA.
7691         (s390_secondary_reload): Don't used secondary reloads for LRA.
7692         (s390_lra_p): New function.
7693         (TARGET_LRA_P): Define.
7694         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7695         of attribute cpu_facility to zarch for the last alternative.
7696         (*cmpmem_short): Ditto.
7698 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
7700         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7701         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
7702         (arm_expand_prologue): Likewise.
7704 2013-06-06  Teresa Johnson  <tejohnson@google.com>
7706         PR c++/53743
7707         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7708         as this is now done by redirect_edge_and_branch_force.
7709         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7710         barriers, and fix interaction with splitting.
7711         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7712         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7713         reflect changes made in the routine.
7714         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7715         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7716         since this is called in cfglayout mode, and replace partition fixup
7717         with assert as that is now done by force_nonfallthru_and_redirect.
7718         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7719         already be marked with region crossing note.
7720         (insert_section_boundary_note): Make non-static, gate on flag
7721         has_bb_partition, rewrite to also check for multiple partitions.
7722         (rest_of_handle_reorder_blocks): Remove call to
7723         insert_section_boundary_note, now done later during free_cfg.
7724         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7725         * bb-reorder.h (insert_section_boundary_note): Declare.
7726         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7727         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7728         invoke insert_section_boundary_note.
7729         (try_redirect_by_replacing_jump): Remove unnecessary
7730         check for region crossing note.
7731         (fixup_partition_crossing): New function.
7732         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7733         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7734         in non-cfglayout mode.
7735         (force_nonfallthru_and_redirect): Fixup partition boundaries,
7736         remove old code that tried to do this. Emit barrier correctly
7737         when we are in cfglayout mode.
7738         (last_bb_in_partition): New function.
7739         (rtl_split_edge): Correctly fixup partition boundaries.
7740         (commit_one_edge_insertion): Remove old code that tried to
7741         fixup region crossing edge since this is now handled in
7742         split_block, and set up insertion point correctly since
7743         block may now end in a jump.
7744         (verify_hot_cold_block_grouping): Guard against checking when not in
7745         linearized RTL mode.
7746         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7747         notes.
7748         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7749         rtl_verify_flow_info, so not called in cfglayout mode.
7750         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7751         (fixup_reorder_chain): Remove old code that attempted to fixup region
7752         crossing note as this is now handled in force_nonfallthru_and_redirect.
7753         (duplicate_insn_chain): Don't duplicate switch section notes.
7754         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7755         note.
7756         * basic-block.h (emit_barrier_after_bb): Declare.
7758 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7760         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7761         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7762         arm_usatsihi): Adjust alternatives for arm_restrict_it.
7764 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7766         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7767         where appropriate.
7768         * config/arm/ldmstm.md: Regenerate.
7770 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7772         * config/arm/sync.md (atomic_loaddi_1):
7773         Disable predication for arm_restrict_it.
7774         (arm_load_exclusive<mode>): Likewise.
7775         (arm_load_exclusivesi): Likewise.
7776         (arm_load_exclusivedi): Likewise.
7777         (arm_load_acquire_exclusive<mode>): Likewise.
7778         (arm_load_acquire_exclusivesi): Likewise.
7779         (arm_load_acquire_exclusivedi): Likewise.
7780         (arm_store_exclusive<mode>): Likewise.
7781         (arm_store_exclusive<mode>): Likewise.
7782         (arm_store_release_exclusivedi): Likewise.
7783         (arm_store_release_exclusive<mode>): Likewise.
7785 2013-06-06  Richard Biener  <rguenther@suse.de>
7787         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7788         after LTO_null.
7789         (lto_tag_is_tree_code_p): Adjust.
7790         (lto_tag_is_gimple_code_p): Likewise.
7791         (lto_gimple_code_to_tag): Likewise.
7792         (lto_tag_to_gimple_code): Likewise.
7793         (lto_tree_code_to_tag): Likewise.
7794         (lto_tag_to_tree_code): Likewise.
7795         * data-streamer.h (streamer_write_hwi_in_range): Use
7796         uhwi streaming to stream the normalized range.
7797         (streamer_read_hwi_in_range): Likewise.
7799 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7801         * config/arm/arm.md (enabled_for_depr_it): New attribute.
7802         (predicable_short_it): Likewise.
7803         (predicated): Likewise.
7804         (enabled): Handle above.
7805         (define_cond_exec): Set predicated attribute to yes.
7807 2013-06-05  Mike Stump  <mikestump@comcast.net>
7809         * gdbinit.in (__FUNCTION__): Add.
7811 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
7813         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7814         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7816 2013-06-05  Jan Hubicka  <jh@suse.cz>
7818         * varasm.c (mark_decl_referenced): Revert the removal until targets
7819         are fixed.
7821 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
7823         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7824         instead of mark_decl_referenced.
7826 2013-06-05  Jan Hubicka  <jh@suse.cz>
7828         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7829         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7830         and symtab_used_from_object_file_p.
7831         (cgraph_make_node_local_1): Clear forced_by_abi.
7832         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7833         * cgraph.h (symtab_node_base): Add forced_by_abi.
7834         (decide_is_variable_needed): Remove.
7835         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7836         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7837         (decide_is_symbol_needed): ... this one; handle symbols in general;
7838         always analyze virtuals; honnor forced_by_abi.
7839         (cgraph_finalize_function): Update.
7840         (varpool_finalize_decl): Update.
7841         (symbol_defined_and_needed): Remove.
7842         (analyze_functions): Update.
7843         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7844         output_refs, input_overwrite_node): Handle forced_by_abi.
7845         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7846         (address_taken_from_non_vtable_p): ... this one.
7847         (comdat_can_be_unshared_p_1): New function.
7848         (cgraph_comdat_can_be_unshared_p): Rename to ...
7849         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7850         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7851         (function_and_variable_visibility): Clear forced_by_abi as needed.
7852         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7853         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7854         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7855         * varpool.c (decide_is_variable_needed): Remove.
7857 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7859         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7860         (arm_option_override): Override arm_restrict_it where appropriate.
7861         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7862         * config/arm/arm.opt (mrestrict-it): New command-line option.
7863         * doc/invoke.texi: Document -mrestrict-it.
7865 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7867         * tsan.c (tsan_atomic_table): Make const.
7869 2013-06-05  Richard Biener  <rguenther@suse.de>
7871         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7872         index associated with the tree we are supposed to replace.
7873         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7874         TREE_ASM_WRITTEN as zero for everything but SSA names.
7876 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7878         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7880 2013-06-04  Jan Hubicka  <jh@suse.cz>
7882         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7883         (input_node, input_varpool_node): Handle correctly external same
7884         body aliases.
7885         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7886         nodes at ltrans stage.
7888 2013-06-04  Jan Hubicka  <jh@suse.cz>
7890         * ipa-inline.c (update_caller_keys): Fix availability test.
7891         (update_callee_keys): Likewise.
7892         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7893         to follow ELF standard.
7895 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7897         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7898         (mips64r5900el-*-elf*): New configurations.
7899         * config/mips/mips-cpus.def (r5900): New processor.
7900         * config/mips/mips-tables.opt: Regenerate.
7901         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7902         (mips_issue_rate): Handle PROCESSOR_R5900.
7903         (mips_reorg_process_insns): Force reorder mode for the R5900.
7904         * config/mips/mips.h (TARGET_MIPS5900): Define.
7905         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7906         TARGET_MIPS5900.
7907         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7908         TARGET_MIPS5900.
7909         * config/mips/mips.md (processor): Add r5900.
7910         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7912 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7914         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7915         into function to generate MOVI instruction.
7916         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7917         (aarch64_preferred_simd_mode): Turn into wrapper.
7918         (aarch64_output_scalar_simd_mov_immediate): New function.
7919         * config/aarch64/aarch64-protos.h: Add prototype for above.
7921 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7923         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7924         element_char member.
7925         (sizetochar): Return signed char.
7926         (aarch64_simd_valid_immediate): Remove elchar and other
7927         unnecessary variables.
7928         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7929         Calculate element_char as required.
7930         * config/aarch64/aarch64-protos.h: Update and move prototype
7931         for aarch64_output_simd_mov_immediate.
7932         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7933         Update arguments.
7935 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7937         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7938         information completed by aarch64_simd_valid_immediate.
7939         (aarch64_legitimate_constant_p): Update arguments.
7940         (aarch64_simd_valid_immediate): Work with struct rather than many
7941         pointers.
7942         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7943         (aarch64_simd_make_constant): Update arguments.
7944         (aarch64_output_simd_mov_immediate): Work with struct rather than
7945         many pointers.  Output immediate directly rather than as operand.
7946         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7947         Update prototype.
7948         * config/aarch64/constraints.md (Dn): Update arguments.
7950 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7952         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7953         longer static.
7954         (aarch64_simd_immediate_valid_for_move): Remove.
7955         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7956         (aarch64_simd_make_constant): Update call.
7957         (aarch64_output_simd_mov_immediate): Update call.
7958         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7959         Add prototype.
7960         * config/aarch64/constraints.md (Dn): Update call.
7962 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7964         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7965         return type to bool for prototype.
7966         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7967         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7968         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7970 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7972         * config/mips/mips.opt (meva): New.
7973         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7974         (ASM_SPEC): Handle -meva.
7975         * doc/invoke.texi (meva):  Document.
7977 2013-06-04  Alan Modra  <amodra@gmail.com>
7979         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7980         constant output.
7982 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7984         * rtl.def: Add extra fourth optional field to define_cond_exec.
7985         * gensupport.c (process_one_cond_exec): Process attributes from
7986         define_cond_exec.
7987         * doc/md.texi: Document fourth field in define_cond_exec.
7989 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7991         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7992         out the processing order as in store_bit_field_1.
7994 2013-06-04  Jan Hubicka  <jh@suse.cz>
7996         PR middle-end/57500
7997         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7998         non-VAR_DECL node if it does not exist yet.
8000 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
8002         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
8003         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
8004         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
8005         target_cpu_default setting.
8007 2013-06-03  Teresa Johnson  <tejohnson@google.com>
8009         * dumpfile.c (opt_info_switch_p): Change -fopt-info
8010         default to -fopt-info=optimized instead of all.
8011         * doc/invoke.texi: Ditto.
8012         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
8013         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
8014         (execute_vect_slp): Emit BB vectorization success under
8015         MSG_OPTIMIZED_LOCATIONS.
8016         * tree-vect-slp.c (vect_slp_transform_bb): Change
8017         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
8018         * tree-vect-loop.c (vect_transform_loop): Ditto.
8020 2013-06-03  Jason Merrill  <jason@redhat.com>
8022         PR c++/57415
8023         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8024         Use TARGET_EXPR for C++.
8026 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
8028         PR rtl-optimization/57268
8029         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
8030         if DEBUG_INSN_P (insn).
8032         Reapply
8033         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8035         PR rtl-optimization/57268
8036         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8037         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8039 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8041         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
8042         (ix86_avoid_lea_for_addr): Likewise.
8043         (exact_dependency_1): Likewise.
8044         (ix86_adjust_cost): Likewise.
8045         (swap_top_of_ready_list): Fix formatting and !reload_completed check
8046         removed.
8047         (do_reorder_for_imul): Fix typo, formatting and
8048         !reload_completed check removed.
8049         (ix86_sched_reorder): Fix typo and formatting.
8050         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
8051         list.
8053 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
8055         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
8057 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
8059         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
8060         <STRING_CST>: Likewise.
8061         <VECTOR_CST>: Likewise.
8063 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
8064             Mikael Morin  <mikael@gcc.gnu.org>
8066         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
8067         * config.in: Regenerated.
8068         * configure: Regenerated.
8070 2013-06-01  Jan Hubicka  <jh@suse.cz>
8072         PR middle-end/57366
8073         * cgraphunit.c (compile): When weakref is not supported,
8074         set up transparent aliases before final output pass.
8075         * varasm.c (assemble_alias): Do not try to do it here.
8077 2013-06-01  Jan Hubicka  <jh@suse.cz>
8079         PR middle-end/57467
8080         * passes.c (for_per_function): Skip unanalyzed functions.
8082 2013-06-01  Jan Hubicka  <jh@suse.cz>
8084         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
8085         (lto_symtab_merge_symbols_1): ... this one.
8086         (lto_symtab_merge_cgraph_nodes): Rename to ...
8087         (lto_symtab_merge_symbols): ... this one; simplify.
8088         * cgraph.c (same_body_aliases_done): Rename to ...
8089         (cpp_implicit_aliases_done): ... this one.
8090         (cgraph_create_function_alias): Update.
8091         (cgraph_same_body_alias): Update.
8092         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8093         (verify_edge_corresponds_to_fndecl): Simplify.
8094         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8095         (cgraph_node): Remove same_body_alias.
8096         (varpool_node): Remove alias_of and extra_name_alias.
8097         (same_body_aliases_done): Rename to ..
8098         (cpp_implicit_aliases_done): ... this one.
8099         (symtab_alias_ultimate_target): Add default parameter.
8100         (symtab_resolve_alias): New function.
8101         (fixup_same_cpp_alias_visibility): Declare.
8102         (cgraph_function_node): Add default parameter.
8103         (cgraph_node_asm_name): Likewise.
8104         (cgraph_function_or_thunk_node): Add default parameter; do
8105         not ICE when it is NULL.
8106         (varpool_variable_node): Likewise.
8107         * tree-emutls.c (create_emultls_var): Update.
8108         (ipa_lower_emutls): Update.
8109         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8110         (cgraph_reset_node): Reset alias info.
8111         (cgraph_finalize_function): Update.
8112         (fixup_same_cpp_alias_visibility): Move to symtab.c.
8113         (analyze_function): Simplify.
8114         (cgraph_process_same_body_aliases): Simplify.
8115         (analyze_functions): Fixup same body aliases.
8116         (handle_alias_pairs): Simplify.
8117         (assemble_thunk): Update.
8118         (assemble_thunks_and_aliases): Update.
8119         (output_weakrefs): Rewrite.
8120         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8121         (lto_output_varpool_node): Likewise.
8122         (compute_ltrans_boundary): Remve assert.
8123         (get_alias_symbol): New functoin.
8124         (input_node): Rewrite alias handling.
8125         (input_varpool_node): Likewise.
8126         * ipa-pure-const.c (propagate_pure_const): Fix formating.
8127         * ipa.c (process_references): Handle weakrefs correctly.
8128         (symtab_remove_unreachable_nodes): Likewise.
8129         * trans-mem.c (get_cg_data): Update.
8130         (ipa_tm_create_version_alias): Update.
8131         (ipa_tm_execute): Update.
8132         * symtab.c (dump_symtab_base): Dump aliases.
8133         (verify_symtab_base): Verify aliases.
8134         (symtab_node_availability): New function.
8135         (symtab_alias_ultimate_target): Simplify.
8136         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8137         handle all the fixup cases.
8138         (symtab_resolve_alias): New function.
8139         * passes.c (ipa_write_summaries): Handle weakrefs.
8140         * varpool.c (varpool_analyze_node): Simplify.
8141         (assemble_aliases): Update.
8142         (varpool_create_variable_alias): Simplify.
8143         (varpool_extra_name_alias): Simplify.
8144         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8145         (lto_symtab_merge_symbols): ... this one.
8147 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
8149         Revert
8150         PR rtl-optimization/57268
8151         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8152         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8154 2013-06-01  Tobias Burnus  <burnus@net-b.de>
8156         Partially reverted:
8157         2013-05-31  Tobias Burnus  <burnus@net-b.de>
8159         PR middle-end/57073
8160         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8161         further up.
8163 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8165         PR rtl-optimization/57268
8166         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8167         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8169 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8171         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8172         unordered comparison operators when -fno-trapping-math is in effect
8173         on the e500.
8174         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8175         and implement unordered comparison operators properly on the e500.
8177 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8179         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8180         for constant scalar integers.
8181         (simplify_relational_operation_1): Likewise.
8183 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
8185         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8186         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
8187         Fix comment.
8189 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8190             Igor Zamyatin  <igor.zamyatin@intel.com>
8192         Silvermont (SLM) architecture performance tuning.
8193         * config/i386/i386.h (enum ix86_tune_indices): Add
8194         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8195         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8197         * config/i386/i386.c (initial_ix86_tune_features)
8198         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8199         (ix86_lea_outperforms): Handle Silvermont tuning.
8200         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8201         call.
8202         (ix86_use_lea_for_mov): Likewise.
8203         (ix86_avoid_lea_for_addr): Likewise.
8204         (ix86_lea_for_add_ok): Likewise.
8205         (exact_dependency_1): New function.
8206         (exact_store_load_dependency): Likewise.
8207         (ix86_adjust_cost): Handle Silvermont tuning.
8208         (do_reoder_for_imul): Likewise.
8209         (swap_top_of_ready_list): New function.
8210         (ix86_sched_reorder): Changed to handle Silvermont tuning.
8212         * config/i386/i386.md (peepholes that split memory operand in fp
8213         converts): New.
8215 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8217         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8218         Remove un-necessary braces.
8220 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8222         * config/aarch64/aarch64.c (aarch64_classify_symbol):
8223         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8225 2013-05-31  Tobias Burnus  <burnus@net-b.de>
8227         PR middle-end/57073
8228         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8230 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8232         PR target/56315
8233         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8234         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8235         * config/arm/neon.md (iordi3_neon): Remove.
8236         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8237         * config/arm/predicates.md (imm_for_neon_logic_operand):
8238         Move to earlier in the file.
8239         (neon_logic_op2): Likewise.
8240         (arm_iordi_operand_neon): New predicate.
8242 2013-05-31  Richard Biener  <rguenther@suse.de>
8244         PR tree-optimization/57478
8245         PR tree-optimization/57453
8246         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8247         are life as well.
8249 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8251         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8252         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8254 2013-05-30  Tobias Burnus  <burnus@net-b.de>
8255             Thomas Koenig  <tkoenig@gcc.gnu.org>
8257         PR middle-end/57073
8258         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8259         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8261 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
8263         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8265 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
8267         * target.def (register_usage_leveling_p): New hook.
8268         * targhooks.c (default_register_usage_leveling_p): New.
8269         * targhooks.h (default_register_usage_leveling_p): New prototype.
8270         * lra-assigns.c (register_usage_leveling_p): Use the hook.
8271         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8272         * doc/tm.texi: Update.
8273         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8275 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
8277         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8278         (*insv_reg<mode>): New define_insn.
8280 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
8282         PR rtl-optimization/57439
8283         * postreload.c (move2add_valid_value_p): Check that we have
8284         a zero subreg_regno_offset when accessing the register in
8285         the requested mode.
8287 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8288             Igor Zamyatin  <igor.zamyatin@intel.com>
8290         Silvermont (SLM) architecture pipeline model, tuning and
8291         insn selection.
8292         * config.gcc: Add slm config options and target.
8294         * config/i386/slm.md: New.
8296         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8298         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
8299         PROCESSOR_SLM.
8300         (ix86_target_macros_internal): Likewise.
8302         * gcc/config/i386/i386.c (slm_cost): New cost.
8303         (m_SLM): New macro flag.
8304         (initial_ix86_tune_features): Set m_SLM.
8305         (x86_accumulate_outgoing_args): Likewise.
8306         (x86_arch_always_fancy_math_387): Likewise.
8307         (processor_target_table): Add slm cost.
8308         (cpu_names): Add slm cpu name.
8309         (x86_option_override_internal): Set SLM ISA.
8310         (ix86_issue_rate): New case PROCESSOR_SLM.
8311         (ia32_multipass_dfa_lookahead): Likewise.
8312         (fold_builtin_cpu): Add slm.
8314         * config/i386/i386.h (TARGET_SLM): New target macro.
8315         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8316         (processor_type): Add PROCESSOR_SLM.
8318         * config/i386/i386.md (cpu): Add new value "slm".
8319         (slm.md): Include slm.md.
8321 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
8322             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8324         * config/arm/arm-protos.h: Add and update function protos.
8325         * config/arm/arm.c (use_simple_return_p): New added.
8326         (thumb2_expand_return): Check simple_return flag.
8327         * config/arm/arm.md: Add simple_return and conditional simple_return.
8328         * config/arm/iterators.md: Add iterator for return and simple_return.
8330 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
8332         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8333         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8334         (arm_emit_vfp_multi_reg_pop): Likewise.
8335         (thumb2_emit_ldrd_pop): Likewise.
8336         (arm_expand_epilogue): Add misc REG_CFA notes.
8337         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8339 2013-05-29  Lawrence Crowl  <crowl@google.com>
8341         * config/arm/t-arm: Update for below.
8343         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8344         Change type to hash_table.  Update dependent calls and types.
8346         * config/i386/t-cygming: Update for below.
8348         * config/i386/t-interix: Update for below.
8350         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8351         Change type to hash_table.  Update dependent calls and types.
8352         (i386_find_on_wrapper_list::wrappers): Likewise.
8354         * config/ia64/t-ia64: Update for below.
8356         * config/ia64/ia64.c (bundle_state_table):
8357         Change type to hash_table.  Update dependent calls and types.
8359         * config/mips/mips.c (mips_reorg_process_insns::htab):
8360         Change type to hash_table.  Update dependent calls and types.
8362         * config/sol2.c (solaris_comdat_htab):
8363         Change type to hash_table.  Update dependent calls and types.
8365         * config/t-sol2: Update for above.
8367 2013-05-29  Teresa Johnson  <tejohnson@google.com>
8369         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8370         functions are not yet marked as defined.
8372 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
8373             Pat Haugen  <pthaugen@us.ibm.com>
8374             Peter Bergner  <bergner@vnet.ibm.com>
8376         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8377         instructions.
8378         (VEC_A): Likewise.
8379         (VEC_C): Likewise.
8380         (vrotl<mode>3): Likewise.
8381         (vashl<mode>3): Likewise.
8382         (vlshr<mode>3): Likewise.
8383         (vashr<mode>3): Likewise.
8385         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8386         support for power8 V2DI builtins.
8388         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8389         power8 V2DI builtins.
8390         (vupkhsw): Likewise.
8391         (vupklsw): Likewise.
8392         (vaddudm): Likewise.
8393         (vminsd): Likewise.
8394         (vmaxsd): Likewise.
8395         (vminud): Likewise.
8396         (vmaxud): Likewise.
8397         (vpkudum): Likewise.
8398         (vpksdss): Likewise.
8399         (vpkudus): Likewise.
8400         (vpksdus): Likewise.
8401         (vrld): Likewise.
8402         (vsld): Likewise.
8403         (vsrd): Likewise.
8404         (vsrad): Likewise.
8405         (vsubudm): Likewise.
8406         (vcmpequd): Likewise.
8407         (vcmpgtsd): Likewise.
8408         (vcmpgtud): Likewise.
8409         (vcmpequd_p): Likewise.
8410         (vcmpgtsd_p): Likewise.
8411         (vcmpgtud_p): Likewise.
8412         (vupkhsw): Likewise.
8413         (vupklsw): Likewise.
8414         (vaddudm): Likewise.
8415         (vmaxsd): Likewise.
8416         (vmaxud): Likewise.
8417         (vminsd): Likewise.
8418         (vminud): Likewise.
8419         (vpksdss): Likewise.
8420         (vpksdus): Likewise.
8421         (vpkudum): Likewise.
8422         (vpkudus): Likewise.
8423         (vrld): Likewise.
8424         (vsld): Likewise.
8425         (vsrad): Likewise.
8426         (vsrd): Likewise.
8427         (vsubudm): Likewise.
8429         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8430         support for power8 V2DI instructions.
8432         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8433         power8 V2DI instructions.  Combine pack and unpack insns to use an
8434         iterator for each mode.  Check whether a particular mode supports
8435         Altivec instructions instead of just checking TARGET_ALTIVEC.
8436         (UNSPEC_VPKUWUM): Likewise.
8437         (UNSPEC_VPKSHSS): Likewise.
8438         (UNSPEC_VPKSWSS): Likewise.
8439         (UNSPEC_VPKUHUS): Likewise.
8440         (UNSPEC_VPKSHUS): Likewise.
8441         (UNSPEC_VPKUWUS): Likewise.
8442         (UNSPEC_VPKSWUS): Likewise.
8443         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8444         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8445         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8446         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8447         (UNSPEC_VUPKHSB): Likewise.
8448         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8449         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8450         (UNSPEC_VUPKHSH): Likewise.
8451         (UNSPEC_VUPKLSB): Likewise.
8452         (UNSPEC_VUPKLSH): Likewise.
8453         (VI2): Likewise.
8454         (VI_char): Likewise.
8455         (VI_scalar): Likewise.
8456         (VI_unit): Likewise.
8457         (VP): Likewise.
8458         (VP_small): Likewise.
8459         (VP_small_lc): Likewise.
8460         (VU_char): Likewise.
8461         (add<mode>3): Likewise.
8462         (altivec_vaddcuw): Likewise.
8463         (altivec_vaddu<VI_char>s): Likewise.
8464         (altivec_vadds<VI_char>s): Likewise.
8465         (sub<mode>3): Likewise.
8466         (altivec_vsubcuw): Likewise.
8467         (altivec_vsubu<VI_char>s): Likewise.
8468         (altivec_vsubs<VI_char>s): Likewise.
8469         (altivec_vavgs<VI_char>): Likewise.
8470         (altivec_vcmpbfp): Likewise.
8471         (altivec_eq<mode>): Likewise.
8472         (altivec_gt<mode>): Likewise.
8473         (altivec_gtu<mode>): Likewise.
8474         (umax<mode>3): Likewise.
8475         (smax<mode>3): Likewise.
8476         (umin<mode>3): Likewise.
8477         (smin<mode>3): Likewise.
8478         (altivec_vpkuhum): Likewise.
8479         (altivec_vpkuwum): Likewise.
8480         (altivec_vpkshss): Likewise.
8481         (altivec_vpkswss): Likewise.
8482         (altivec_vpkuhus): Likewise.
8483         (altivec_vpkshus): Likewise.
8484         (altivec_vpkuwus): Likewise.
8485         (altivec_vpkswus): Likewise.
8486         (altivec_vpks<VI_char>ss): Likewise.
8487         (altivec_vpks<VI_char>us): Likewise.
8488         (altivec_vpku<VI_char>us): Likewise.
8489         (altivec_vpku<VI_char>um): Likewise.
8490         (altivec_vrl<VI_char>): Likewise.
8491         (altivec_vsl<VI_char>): Likewise.
8492         (altivec_vsr<VI_char>): Likewise.
8493         (altivec_vsra<VI_char>): Likewise.
8494         (altivec_vsldoi_<mode>): Likewise.
8495         (altivec_vupkhsb): Likewise.
8496         (altivec_vupkhs<VU_char>): Likewise.
8497         (altivec_vupkls<VU_char>): Likewise.
8498         (altivec_vupkhsh): Likewise.
8499         (altivec_vupklsb): Likewise.
8500         (altivec_vupklsh): Likewise.
8501         (altivec_vcmpequ<VI_char>_p): Likewise.
8502         (altivec_vcmpgts<VI_char>_p): Likewise.
8503         (altivec_vcmpgtu<VI_char>_p): Likewise.
8504         (abs<mode>2): Likewise.
8505         (vec_unpacks_hi_v16qi): Likewise.
8506         (vec_unpacks_hi_v8hi): Likewise.
8507         (vec_unpacks_lo_v16qi): Likewise.
8508         (vec_unpacks_hi_<VP_small_lc>): Likewise.
8509         (vec_unpacks_lo_v8hi): Likewise.
8510         (vec_unpacks_lo_<VP_small_lc>): Likewise.
8511         (vec_pack_trunc_v8h): Likewise.
8512         (vec_pack_trunc_v4si): Likewise.
8513         (vec_pack_trunc_<mode>): Likewise.
8515         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8516         V2DI builtins.
8517         (vec_vmaxsd): Likewise.
8518         (vec_vmaxud): Likewise.
8519         (vec_vminsd): Likewise.
8520         (vec_vminud): Likewise.
8521         (vec_vpksdss): Likewise.
8522         (vec_vpksdus): Likewise.
8523         (vec_vpkudum): Likewise.
8524         (vec_vpkudus): Likewise.
8525         (vec_vrld): Likewise.
8526         (vec_vsld): Likewise.
8527         (vec_vsrad): Likewise.
8528         (vec_vsrd): Likewise.
8529         (vec_vsubudm): Likewise.
8530         (vec_vupkhsw): Likewise.
8531         (vec_vupklsw): Likewise.
8533 2013-05-29  Jan Hubicka  <jh@suse.cz>
8535         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8536         flags; reorder rest of fields in more consistent way.
8537         (varpool_node): Remove analyzed, finalized and alias.
8538         (cgraph_ndoe): Likewise.
8539         (symtab_alias_ultimate_target): New function.
8540         (cgraph_function_node): Move offline.
8541         (cgraph_reset_node): Declare.
8542         (cgraph_comdat_can_be_unshared_p): Remove.
8543         (varpool_remove_initializer): Declare.
8544         (varpool_first_defined_variable, varpool_next_defined_variable
8545         cgraph_first_defined_function, cgraph_next_defined_function): Update.
8546         (cgraph_function_with_gimple_body_p): Update.
8547         (varpool_all_refs_explicit_p): Update.
8548         (symtab_alias_target): New function.
8549         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8550         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8551         (cgraph_function_or_thunk_node): Simplify using
8552         symtab_alias_ultimate_target.
8553         (varpool_variable_node): Likewise.
8554         * cgraph.c (cgraph_create_function_alias): Update.
8555         (cgraph_add_thunk): Update.
8556         (cgraph_remove_node): Update.
8557         (dump_cgraph_node): Do not dump removed flags.
8558         (cgraph_function_body_availability): Update.
8559         (cgraph_propagate_frequency): Update.
8560         (verify_cgraph_node): Check sanity of local flag.
8561         (cgraph_function_node): Move here from cgraph.h; revamp for
8562         cgraph_function_or_thunk_node.
8563         * lto-symtab.c (lto_varpool_replace_node): Update.
8564         (lto_symtab_resolve_can_prevail_p): Update.
8565         (lto_symtab_merge_cgraph_nodes): Update.
8566         * ipa-cp.c (determine_versionability, initialize_node_lattices,
8567         propagate_constants_accross_call, devirtualization_time_bonus,
8568         ipcp_propagate_stage): Update.
8569         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8570         * ipa-inline-transform.c (clone_inlined_nodes,
8571         preserve_function_body_p): Update.
8572         * ipa-reference.c (propagate): Update.
8573         (write_node_summary_p): Update.
8574         * toplev.c (wrapup_global_declaration_2): Update.
8575         * cgraphunit.c (cgraph_analyze_function): Rename to ...
8576         (analyze_function) ... this one.
8577         (cgraph_process_new_functions): Update.
8578         (cgraph_reset_node): Export.
8579         (cgraph_finalize_function): Update.
8580         (cgraph_add_new_function): Update.
8581         (process_function_and_variable_attributes): Update.
8582         (varpool_finalize_decl): Update.
8583         (symbol_finalized): Remove.
8584         (symbol_finalized_and_needed): Rename to ...
8585         (symbol_defined_and_needed): ... update.
8586         (cgraph_analyze_functions): Update.
8587         (handle_alias_pairs): Update.
8588         (mark_functions_to_output): Update.
8589         (assemble_thunk): Update.
8590         (output_in_order): Update.
8591         (output_weakrefs): Update.
8592         (finalize_compilation_unit): Update.
8593         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8594         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8595         input_node, input_varpool_node): Update.
8596         * dbxout.c (dbxout_expand_expr): Update.
8597         * cgraphclones.c (cgraph_clone_node): Update.
8598         (cgraph_copy_node_for_versioning): Update.
8599         (cgraph_materialize_clone): Update.
8600         (cgraph_materialize_all_clones): Update.
8601         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8602         propagate_pure_const, propagate_nothrow): Update.
8603         * lto-streamer-out.c (lto_output, write_symbol): Update.
8604         * ipa-utils.c (ipa_reverse_postorder): Update.
8605         * ipa-inline.c (can_inline_edge_p): Update.
8606         (update_caller_keys, ipa_inline): Update.
8607         * dwarf2out.c (reference_to_unused,
8608         premark_types_used_by_global_vars_helper): Update.
8609         * tree-eh.c (tree_could_trap_p): Update.
8610         * ipa-split.c (consider_split, execute_split_functions): Update.
8611         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8612         has_addr_references_p): Update; move ahead in file for better
8613         readability.
8614         (process_references): Simplify.
8615         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8616         bodies are removed.
8617         (cgraph_comdat_can_be_unshared_p): Make static.
8618         (cgraph_externally_visible_p): Update.
8619         (varpool_externally_visible_p): Update.
8620         (function_and_variable_visibility): Update.
8621         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8622         ipa_tm_mark_force_output_node): Update.
8623         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8624         estimate_edge_devirt_benefit, inline_generate_summary,
8625         inline_write_summary): Update.
8626         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8627         * ipa-prop.c (ipa_compute_jump_functions): Update.
8628         (ipa_print_node_params, ipa_prop_read_section,
8629         ipa_update_after_lto_read, read_replacements_section): Update.
8630         * varasm.c (mark_decl_referenced): Update.
8631         (assemble_alias, dump_tm_clone_pairs): Update.
8632         * tree-inline.c (copy_bb): Update.
8633         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8634         Update.
8635         * symtab.c (dump_symtab_base): Print new flags.
8636         (verify_symtab_base): Verify new flags.
8637         (symtab_alias_ultimate_target): New function.
8638         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8639         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8640         Update.
8641         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8642         Update.
8643         * i386.c (ix86_get_function_versions_dispatcher,
8644         ix86_generate_version_dispatcher_body): Update.
8645         (fold_builtin_cpu): Use varpool_add_new_variable.
8646         * varpool.c (varpool_remove_initializer): Break out from ...
8647         (varpool_remove_node): ... this one.
8648         (dump_varpool_node, varpool_node_for_asm,
8649         cgraph_variable_initializer_availability, varpool_analyze_node,
8650         varpool_assemble_decl, varpool_remove_unreferenced_decls,
8651         varpool_finalize_named_section_flags, varpool_create_variable_alias):
8652         Update.
8654 2013-05-29  Jan Hubicka  <jh@suse.cz>
8656         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8658 2013-05-29  Easwaran Raman  <eraman@google.com>
8660         PR tree-optimization/57442
8661         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8662         when control exits the main loop.
8664 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
8666         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8667         and RX600.
8668         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8669         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8670         * rx/t-rx: Add rx100 under multi library matches option for nofpu
8671         option.
8673 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8675         PR tree-optimization/57441
8676         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8677         Don't limit size of incr_vec to number of candidates.
8679 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
8681         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8682         and mips16 directories.
8683         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8684         (MULTILIB_DIRNAMES): Ditto.
8685         (MULTILIB_EXCEPTIONS): Add new exceptions.
8686         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8687         (MULTILIB_DIRNAMES): Ditto.
8688         (MULTILIB_EXCEPTIONS): Add new exceptions.
8690 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8691             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8693         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8694         SYMBOL_TINY_ABSOLUTE.
8695         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8696         SYMBOL_TINY_ABSOLUTE.
8697         (aarch64_expand_mov_immediate): Likewise.
8698         (aarch64_classify_symbol): Likewise.
8699         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8700         Permit SYMBOL_TINY_ABSOLUTE.
8701         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8703 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8704             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8706         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8707         Refactor if/switch.  Replace gcc_assert with if.
8709 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8711         * config/i386/i386.c (initial_ix86_tune_features): Enable
8712         FP Reassociation for AMD bdver1 and bdver2.
8714 2013-05-29  Martin Jambor  <mjambor@suse.cz>
8716         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8717         and IMAGPART_EXPR do not occur within other handled_components.
8719 2013-05-29  Richard Biener  <rguenther@suse.de>
8721         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8722         access on whether the use is in the BB we currently try to
8723         vectorize.
8724         (vect_bb_vectorization_profitable_p): Pass the BB we currently
8725         vectorize to vect_bb_slp_scalar_cost.
8727 2013-05-29  Richard Biener  <rguenther@suse.de>
8729         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8730         computing scalar cost offsetted by stmts that are kept live
8731         by scalar uses.
8732         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8733         for computation of scalar cost.
8735 2013-05-28  Steve Ellcey  <sellcey@mips.com>
8737         * config/mips/mips-cpus.def (mips32r2): Change processor type.
8739 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8741         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8742         array notation built-in reduction functions.
8743         * doc/passes.texi (Passes): Added documentation about changes done
8744         for Cilk Plus.
8745         * doc/invoke.texi (C Dialect Options): Added documentation about
8746         the -fcilkplus flag.
8747         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8748         (BUILTINS_DEF): Depend on cilkplus.def.
8749         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
8750         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8751         * cilkplus.def: New file.
8753 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
8755         PR rtl-optimization/57439
8756         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8758 2013-05-28  Easwaran Raman  <eraman@google.com>
8760         PR tree-optimization/57337
8761         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8762         (find_insert_point): Correctly identify the insertion point
8763         when two statements with the same UID is compared.
8765 2013-05-28  Richard Biener  <rguenther@suse.de>
8767         PR tree-optimization/56787
8768         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8769         from the list of data references.
8770         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8771         clobbers.
8772         (vect_analyze_loop_operations): Likewise.
8773         (vect_transform_loop): Remove clobbers.
8775 2013-05-28  Martin Jambor  <mjambor@suse.cz>
8777         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8778         and REALPART_EXPRs have scalar type.
8780 2013-05-28  Richard Biener  <rguenther@suse.de>
8782         PR tree-optimization/57411
8783         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8784         virtual operands.
8785         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8786         virtual operand propagation.
8788 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8790         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8791         destination register for bmasksi_vis.
8792         (vector_init_bshuffle): Likewise.
8793         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8795 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8797         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8798         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8799         mode if the instruction isn't available in the original mode.
8800         * config/sparc/sparc.opt (mfix-ut699): New option.
8801         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8802         (divdf3): Turn into expander.
8803         (divdf3_nofix): New insn.
8804         (divdf3_fix): Likewise.
8805         (divsf3): Disable if -mfix-ut699.
8806         (sqrtdf2): Turn into expander.
8807         (sqrtdf2_nofix): New insn.
8808         (sqrtdf2_fix): Likewise.
8809         (sqrtsf2): Disable if -mfix-ut699.
8811 2013-05-27  Richard Biener  <rguenther@suse.de>
8813         PR middle-end/57412
8814         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8815         block for the new loop.
8817 2013-05-27  Richard Biener  <rguenther@suse.de>
8819         PR tree-optimization/57343
8820         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8821         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8822         (number_of_iterations_cond): Do not build the folded tree.
8824 2013-05-27  Richard Biener  <rguenther@suse.de>
8826         Revert
8827         PR middle-end/57381
8828         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8829         OEP_CONSTANT_ADDRESS_OF retained.
8831         PR tree-optimization/57417
8832         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8833         for unchanged base.
8834         (set_ssa_val_to): Compare addresses using
8835         get_addr_base_and_unit_offset.
8837 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
8839         PR rtl-optimization/56833
8840         * postreload.c (move2add_record_mode): New function.
8841         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8842         (move2add_use_add2_insn): Use move2add_record_sym_value.
8843         (move2add_use_add3_insn): Likewise.
8844         (reload_cse_move2add): Use move2add_valid_value_p and
8845         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8846         affected regs by setting reg_mode to VOIDmode.
8847         (move2add_note_store): Don't pretend the inside of a SUBREG is
8848         the actual destination.  Invalidate single/leading registers by
8849         setting reg_mode to VOIDmode.
8850         Use move2add_record_sym_value, move2add_valid_value_p and
8851         move2add_record_mode.
8853 2013-05-27  Richard Biener  <rguenther@suse.de>
8855         PR tree-optimization/57396
8856         * tree-affine.c (double_int_constant_multiple_p): Properly
8857         return false for val == 0 and div != 0.
8859 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8861         * config/mips/mips.h: Use #elif in preprocessor conditions.
8863 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8865         PR target/53916
8866         * config/mips/constraints.md (kl): New constraint.
8867         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8868         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8869         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8870         from LO for MIPS16.
8871         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8873 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8875         PR target/55777
8876         * config/mips/mips.c (mips_can_inline_p): New function.
8877         (TARGET_CAN_INLINE_P): Define.
8879 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8881         * sched-int.h (ds_t, dw_t): Make unsigned int.
8882         Fix documentation that describes how all the ds_t bits are used.
8883         Reserve the last bit for delayed-branch scheduling.
8884         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8885         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8886         (gen_dep_weak_1): Remove prototype.
8887         * sched-deps.c (get_dep_weak_1): Make static.
8888         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8889         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8890         * doc/tm.texi: Regenerate.
8891         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8893 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8895         PR debug/56950
8896         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8898 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8899             Sandra Loosemore  <sandra@codesourcery.com>
8901         * config.gcc (powerpc-*): Allow native for with-cpu.
8903 2013-05-24  Jeff Law  <law@redhat.com>
8905         PR tree-optimization/57124
8906         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8907         conversion feeding a condition if the range has an overflow
8908         if -fstrict-overflow.  Add warnings for when we do make the
8909         transformation.
8911 2013-05-24  Dehao Chen  <dehao@google.com>
8913         * tree-cfg.c (locus_discrim_map): Fix the typo.
8914         (locus_discrim_hasher): Likewise.
8915         (locus_discrim_hasher::hash): Likewise.
8916         (locus_discrim_hasher::equal): Likewise.
8918 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8920         PR tree-optimization/57294
8921         * cgraph.h (ipa_record_stmt_references): Declare.
8922         * cgraphbuild.c (ipa_record_stmt_references): New function.
8923         (build_cgraph_edges): Use ipa_record_stmt_references.
8924         (rebuild_cgraph_edges): Likewise.
8925         (cgraph_rebuild_references): Likewise.
8926         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8927         associated with the old statement and build references from the
8928         newly built statements.
8929         * ipa-ref.c (ipa_remove_stmt_references): New function.
8930         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8932 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8934         * lra-constraints.c (emit_spill_move): Use smaller mode for
8935         mem-mem moves.
8936         (check_and_process_move): Consider mem-reg moves for secondary
8937         too.
8938         (curr_insn_transform): Don't lose insns emitted before for
8939         secondary memory moves.
8940         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8941         reg set up in the current insn.
8943 2013-05-24  Dehao Chen  <dehao@google.com>
8945         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8946         hash function.
8947         (locus_descrim_hasher::equal): Likewise.
8948         (build_gimple_cfg): New discriminator assignment algorithm.
8949         (make_edges): Likewise.
8950         (next_discriminator_for_locus): Likewise.
8951         (same_line_p): Likewise.
8952         (assign_discriminators): Likewise.
8953         (make_cond_expr_edges): Likewise.
8954         (make_gimple_switch_edges): Likewise.
8955         (make_goto_expr_edges): Likewise.
8956         (make_gimple_asm_edges): Likewise.
8958 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8960         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8961         X format specifier to only display bottom 16 bits.
8962         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8963         immediate to match for operand 2, since it will be masked.
8965 2013-05-24  Richard Biener  <rguenther@suse.de>
8967         PR tree-optimization/57287
8968         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8969         all SSA names that occur in abnormal PHIs.
8971 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8973         PR tree-ssa/57385
8974         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8975         that index is not negative.
8977 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8979         PR rtl-optimization/55177
8980         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8981         (simplify_byte_swapping_operation): New.
8982         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8983         (simplify_relational_operation_1): Deal with BSWAP.
8985 2013-05-23  Richard Henderson  <rth@redhat.com>
8987         PR target/56742
8988         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8989         (ix86_reorg): Call it.
8991 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8993         PR target/57379
8994         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8995         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8996         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8998 2013-05-23  Christian Bruel  <christian.bruel@st.com>
9000         PR debug/57351
9001         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
9003 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9004             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9006         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
9007         * config/aarch64/constraints.md (Usa): Remove.
9008         * doc/md.texi (AArch64 Usa): Remove.
9010 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9011             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9013         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
9014         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
9015         * config/aarch64/predicates.md (aarch64_const_address): Remove.
9016         (aarch64_mov_operand): Use aarch64_mov_operand_p.
9018 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
9020         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
9021         instruction (AdvSIMD).
9022         * config/aarch64/aarch64-builtins.c
9023         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
9024         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
9026 2013-05-23  Martin Jambor  <mjambor@suse.cz>
9028         PR middle-end/57347
9029         * tree.h (contains_bitfld_component_ref_p): Declare.
9030         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
9031         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
9032         caller.
9033         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
9034         not access a bit-field.  Assert all final offsets are byte-aligned.
9036 2013-05-23  Richard Biener  <rguenther@suse.de>
9038         PR tree-optimization/57380
9039         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
9040         least one invariant or re-used load.
9041         * passes.c (init_optimization_passes): Move pass_phiprop before
9042         pass_forwprop.
9044 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
9046         * config/aarch64/aarch64-simd.md
9047         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
9049 2013-05-23  Richard Biener  <rguenther@suse.de>
9051         PR middle-end/57381
9052         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
9053         OEP_CONSTANT_ADDRESS_OF retained.
9055 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
9057         PR middle-end/57344
9058         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
9059         don't lower unit.  Handle unit not being always BITS_PER_WORD.
9061 2013-05-23  Richard Biener  <rguenther@suse.de>
9063         PR rtl-optimization/57341
9064         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
9065         instead of true_dependence.
9067 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
9069         * bb-reorder.c (branch_threshold): Make const.
9070         (exec_threshold): Ditto.
9072 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9073             Pat Haugen  <pthaugen@us.ibm.com>
9074             Peter Bergner  <bergner@vnet.ibm.com>
9076         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
9077         documentation for the power8 crypto builtins.
9079         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
9081         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
9082         macros for defining power8 builtin functions.
9083         (BU_P8V_AV_2): Likewise.
9084         (BU_P8V_AV_P): Likewise.
9085         (BU_P8V_VSX_1): Likewise.
9086         (BU_P8V_OVERLOAD_1): Likewise.
9087         (BU_P8V_OVERLOAD_2): Likewise.
9088         (BU_CRYPTO_1): Likewise.
9089         (BU_CRYPTO_2): Likewise.
9090         (BU_CRYPTO_3): Likewise.
9091         (BU_CRYPTO_OVERLOAD_1): Likewise.
9092         (BU_CRYPTO_OVERLOAD_2): Likewise.
9093         (XSCVSPDP): Fix typo, point to the correct instruction.
9094         (VCIPHER): Add power8 crypto builtins.
9095         (VCIPHERLAST): Likewise.
9096         (VNCIPHER): Likewise.
9097         (VNCIPHERLAST): Likewise.
9098         (VPMSUMB): Likewise.
9099         (VPMSUMH): Likewise.
9100         (VPMSUMW): Likewise.
9101         (VPERMXOR_V2DI): Likewise.
9102         (VPERMXOR_V4SI: Likewise.
9103         (VPERMXOR_V8HI: Likewise.
9104         (VPERMXOR_V16QI: Likewise.
9105         (VSHASIGMAW): Likewise.
9106         (VSHASIGMAD): Likewise.
9107         (VPMSUM): Likewise.
9108         (VPERMXOR): Likewise.
9109         (VSHASIGMA): Likewise.
9111         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9112         __CRYPTO__ if the crypto instructions are available.
9113         (altivec_overloaded_builtins): Add support for overloaded power8
9114         builtins.
9116         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9117         support for power8 crypto builtins.
9118         (builtin_function_type): Likewise.
9119         (altivec_init_builtins): Add support for builtins that take vector
9120         long long (V2DI) arguments.
9122         * config/rs6000/crypto.md: New file, define power8 crypto
9123         instructions.
9125 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9126             Pat Haugen  <pthaugen@us.ibm.com>
9127             Peter Bergner  <bergner@vnet.ibm.com>
9129         * doc/invoke.texi (Option Summary): Add power8 options.
9130         (RS/6000 and PowerPC Options): Likewise.
9132         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9133         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
9134         wm, wn, wr documentation.
9136         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9137         registers if direct move instructions are enabled.
9138         (wn): New constraint for no registers.
9139         (wq): New constraint for quad word even GPR registers.
9140         (wr): New constraint if 64-bit instructions are enabled.
9141         (wv): New constraint if power8 vector instructions are enabled.
9142         (wQ): New constraint for quad word memory locations.
9144         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9145         constraint for 0..15 for crypto instructions.
9146         (gpc_reg_operand): If VSX allow registers in VSX registers as well
9147         as GPR and floating point registers.
9148         (int_reg_operand): New predicate to match only GPR registers.
9149         (base_reg_operand): New predicate to match base registers.
9150         (quad_int_reg_operand): New predicate to match even GPR registers
9151         for quad memory operations.
9152         (vsx_reg_or_cint_operand): New predicate to allow vector logical
9153         operations in both GPR and VSX registers.
9154         (quad_memory_operand): New predicate for quad memory operations.
9155         (reg_or_indexed_operand): New predicate for direct move support.
9157         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9158         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9159         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9160         (POWERPC_MASKS): Add power8 options.
9161         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9162         various options.
9164         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9165         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9167         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9168         (-mpower8-fusion): New power8 options.
9169         (-mpower8-fusion-sign): Likewise.
9170         (-mpower8-vector): Likewise.
9171         (-mcrypto): Likewise.
9172         (-mdirect-move): Likewise.
9173         (-mquad-memory): Likewise.
9175         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9176         power8.
9177         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9178         registers.
9179         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9180         (rs6000_debug_vector_unit): Add p8_vector.
9181         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9182         definitions.  Also print fusion state.
9183         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9184         (rs6000_builtin_mask_calculate): Add power8 builtin support.
9185         (rs6000_option_override_internal): Add support for power8.
9186         (rs6000_common_init_builtins): Add debugging for skipped builtins
9187         if -mdebug=builtin.
9188         (rs6000_adjust_cost): Add power8 support.
9189         (rs6000_issue_rate): Likewise.
9190         (insn_must_be_first_in_group): Likewise.
9191         (insn_must_be_last_in_group): Likewise.
9192         (force_new_group): Likewise.
9193         (rs6000_register_move_cost): Likewise.
9194         (rs6000_opt_masks): Likewise.
9196         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9197         power8 capable assembler, default to power7 options.
9198         (TARGET_DIRECT_MOVE): Likewise.
9199         (TARGET_CRYPTO): Likewise.
9200         (TARGET_P8_VECTOR): Likewise.
9201         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9202         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9203         (VECTOR_MEM_P8_VECTOR_P): Likewise.
9204         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9205         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9206         (TARGET_XSCVDPSPN): Likewise.
9207         (TARGET_XSCVSPDPN): Likewsie.
9208         (TARGET_SYNC_HI_QI): Likewise.
9209         (TARGET_SYNC_TI): Likewise.
9210         (MASK_CRYPTO): Likewise.
9211         (MASK_DIRECT_MOVE): Likewise.
9212         (MASK_P8_FUSION): Likewise.
9213         (MASK_P8_VECTOR): Likewise.
9214         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9215         temporary used by some of the direct move instructions to get two FP
9216         temporary registers does not force creation of a stack frame.
9217         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9218         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9219         that any VSX registers are tieable, even if they are also an
9220         Altivec vector mode.
9221         (r6000_reg_class_enum): Add wm, wr, wv constraints.
9222         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9223         (RS6000_BTM_CRYPTO): Likewise.
9224         (RS6000_BTM_COMMON): Likewise.
9226         * config/rs6000/rs6000.md (cpu attribute): Add power8.
9227         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9228         (enum rs6000_vector): Add power8 vector support.
9230 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9232         PR target/19599
9233         PR target/57340
9234         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9235         (any_sibcall_could_use_r3): this and handle indirect calls.
9236         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9238 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9240         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9242 2013-05-22  Richard Biener  <rguenther@suse.de>
9244         PR middle-end/57349
9245         * profile.c (branch_prob): Do not split blocks that are
9246         abnormally receiving from ECF_RETURNS_TWICE functions.
9248 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9250         * recog.c (offsettable_address_addr_space_p): Fix calculation of
9251         address mode.  Move pointer mode initialization to the same place.
9253 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
9255         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9256         while it has any effect.
9258 2013-05-21  Easwaran Raman  <eraman@google.com>
9260         PR tree-optimization/57322
9261         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9262         UID of the statement added to the BB to be 1.
9264 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
9266         PR tree-optimization/57331
9267         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9268         of conversion from pointer type to integral type with integer.
9270 2013-05-21  Martin Jambor  <mjambor@suse.cz>
9272         PR lto/57289
9273         * ipa-prop.c (ipa_read_node_info): Process param_used and
9274         controlled_uses in the same order as when writing.
9276 2013-05-21  Magnus Granberg  <baldrick@free.fr>
9278         PR plugins/56754
9279         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9281 2013-05-21  Richard Biener  <rguenther@suse.de>
9283         PR tree-optimization/57318
9284         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9285         estimate stmts with side-effects as likely eliminated.
9287 2013-05-21  Richard Biener  <rguenther@suse.de>
9289         PR tree-optimization/57330
9290         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9291         preserve the call stmts fntype.
9293 2013-05-21  Richard Biener  <rguenther@suse.de>
9295         PR tree-optimization/57303
9296         * tree-ssa-sink.c (statement_sink_location): Improve killing
9297         stmt detection and properly handle self-assignments.
9299 2013-05-21  Christian Bruel  <christian.bruel@st.com>
9301         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9302         spanning registers. LEAF_REG_REMAP is supported only for contiguous
9303         registers. Set register size out of the PARALLEL loop.
9305 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
9307         PR target/56547
9308         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9309         (*fmasf4, *fmasf4_media): New insns.
9311 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9313         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9314         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9315         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9316         (mips_idiv_insns): Update the comments to say that the returned
9317         instruction counts are in units of BASE_INSN_LENGTH.
9318         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9319         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
9320         using 2 rather than 4 as the length of indirect MIPS16 and
9321         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
9322         length of a NOP.  Don't divide MIPS16 lengths by 2.
9323         (mips16_split_long_branches): Assume a branch is long if the
9324         length is greater than 4 rather than 8.
9325         * config/mips/mips.md (length): Give MIPS16 lengths directly,
9326         rather than multiplying them by 2.  Multiply instruction counts
9327         by BASE_INSN_LENGTH rather than 4.
9328         (*jump_mips16, tls_get_tp_mips16_<mode>)
9329         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9331 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9333         * config/mips/mips.md (extended_mips16): Remove branch case.
9334         (length): Remove duplicated extended_mips16 test.
9336 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
9338         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9340 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
9342         * recog.h (Recog_data): Rename to...
9343         (recog_data_d): ...this.
9344         (recog_data): Update accordingly.
9345         * recog.c (recog_data): Likewise.
9346         * reload.c (save_recog_data): Likewise.
9347         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9348         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9350 2013-05-17  Julian Brown  <julian@codesourcery.com>
9352         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9353         found in a REG_EQUAL note, invalidate it.
9355 2013-05-17  Easwaran Raman  <eraman@google.com>
9357         * tree-ssa-reassoc.c (find_insert_point): New function.
9358         (insert_stmt_after): Likewise.
9359         (get_def_stmt): Likewise.
9360         (ensure_ops_are_available): Likewise.
9361         (not_dominated_by): Likewise.
9362         (rewrite_expr_tree): Do not move statements beyond what is
9363         necessary. Remove call to swap_ops_for_binary_stmt...
9364         (reassociate_bb): ... and move it here.
9365         (build_and_add_sum): Assign UIDs for new statements.
9366         (linearize_expr): Likewise.
9367         (do_reassoc): Renumber gimple statement UIDs.
9369 2013-05-17  Jan Hubicka  <jh@suse.cz>
9371         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9372         weakrefs.
9373         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9374         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9375         weakrefs.
9376         (output_weakrefs): Update.
9378 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
9379             Martin Jambor  <mjambor@suse.cz>
9381         PR middle-end/57276
9382         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9383         value that corresponds to the given aggval is found in values vector.
9385 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
9387         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9388         sse, sse2, sse3, ssse3 and sse4a flags to options.
9390 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
9392         * gengtype-state.c: (s_expr_writer): New class, to handle
9393         prettifying of output layout of s-expressions.
9394         (state_writer): New class, to write out gtype.state.
9395         (state_written_type_count): Move this variable into member data of
9396         state_writer.
9397         (s_expr_writer::s_expr_writer): New code: constructor for new class
9398         (state_writer::state_writer(): ditto
9399         (s_expr_writer::write_new_line): New function
9400         (s_expr_writer::write_any_indent): ditto
9401         (s_expr_writer::begin_s_expr): ditto
9402         (s_expr_writer::end_s_expr): ditto
9403         (write_state_fileloc): convert to method of state_writer...
9404         (state_writer:: write_state_fileloc): ...and use methods of
9405         s_expr_writer to write indentation into the gtype.state output file
9406         to visually represent the hierarchical structure of the list
9407         structures
9408         (write_state_fields): ditto, renaming to...
9409         (state_writer::write_state_fields)
9410         (write_state_a_string): ditto, renaming to...
9411         (state_writer::write_state_a_string)
9412         (write_state_string_option): ditto, renaming to...
9413         (state_writer::write_state_string_option)
9414         (write_state_type_option): ditto, renaming to...
9415         (state_writer::write_state_type_option)
9416         (write_state_nested_option): ditto, renaming to...
9417         (state_writer::write_state_nested_option)
9418         (write_state_option): ditto, renaming to...
9419         (state_writer::write_state_option)
9420         (write_state_options): ditto, renaming to...
9421         (state_writer::write_state_options)
9422         (write_state_lang_bitmap): ditto, renaming to...
9423         (state_writer::write_state_lang_bitmap)
9424         (write_state_version): ditto, renaming to...
9425         (state_writer::write_state_version)
9426         (write_state_scalar_type): ditto, renaming to...
9427         (state_writer::write_state_scalar_type)
9428         (write_state_string_type): ditto, renaming to...
9429         (state_writer::write_state_string_type)
9430         (write_state_undefined_type): ditto, renaming to...
9431         (state_writer::write_state_undefined_type)
9432         (write_state_struct_union_type): ditto, renaming to...
9433         (state_writer::write_state_struct_union_type)
9434         (write_state_struct_type): ditto, renaming to...
9435         (state_writer::write_state_struct_type)
9436         (write_state_user_struct_type): ditto, renaming to...
9437         (state_writer::write_state_user_struct_type)
9438         (write_state_lang_struct_type): ditto, renaming to...
9439         (state_writer::write_state_lang_struct_type)
9440         (write_state_param_struct_type): ditto, renaming to...
9441         (state_writer::write_state_param_struct_type)
9442         (write_state_pointer_type): ditto, renaming to...
9443         (state_writer::write_state_pointer_type)
9444         (write_state_array_type): ditto, renaming to...
9445         (state_writer::write_state_array_type)
9446         (write_state_gc_used): ditto, renaming to...
9447         (state_writer::write_state_gc_used)
9448         (write_state_common_type_content): ditto, renaming to...
9449         (state_writer::write_state_common_type_content)
9450         (write_state_type): ditto, renaming to...
9451         (state_writer::write_state_type)
9452         (write_state_pair_list): ditto, renaming to...
9453         (state_writer::write_state_pair_list)
9454         (write_state_pair): ditto, renaming to...
9455         (state_writer::write_state_pair)
9456         (write_state_typedefs): ditto, renaming to...
9457         (state_writer::write_state_typedefs)
9458         (write_state_structures): ditto, renaming to...
9459         (state_writer::write_state_structures)
9460         (write_state_param_structs): ditto, renaming to...
9461         (state_writer::write_state_param_structs)
9462         (write_state_variables): ditto, renaming to...
9463         (state_writer::write_state_variables)
9464         (write_state_srcdir): ditto, renaming to...
9465         (state_writer::write_state_srcdir)
9466         (write_state_files_list): ditto, renaming to...
9467         (state_writer::write_state_files_list)
9468         (write_state_languages): ditto, renaming to...
9469         (state_writer::write_state_languages)
9470         (write_state): create a state_writer instance and use it when
9471         writing out the state file
9473 2013-05-17  Mike Stump  <mikestump@comcast.net>
9475         PR rtl-optimization/57304
9476         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9477         accessing DF_REF_REAL_LOC.
9479 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9481         PR rtl-optimization/57281
9482         PR rtl-optimization/57300
9483         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9484         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9485         what the other splitter did if the registers are dead.
9487 2013-05-17  Richard Biener  <rguenther@suse.de>
9489         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9490         MEM_REF offsets.
9492 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9494         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9495         linking.
9497 2013-05-17  Marek Polacek  <polacek@redhat.com>
9499         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9500         length when doing non-zero store of storing '\0' to '\0'.
9502 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9504         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9505         vect_external_def oprnd1 with loop_vinfo, try to emit
9506         optional cast, negation and and stmts on the loop preheader
9507         edge instead of into the pattern def seq.
9509         PR tree-optimization/57051
9510         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9511         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9513 2013-05-16  Nick Clifton  <nickc@redhat.com>
9515         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9516         (rl78_is_naked_func): New function.
9517         (rl78_expand_prologue): Skip prologue generation for naked functions.
9518         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9519         * doc/extend.texi (naked): Add RL78 to the list of processors
9520         that supports this attribute.
9522 2013-05-16  Jeff Law  <law@redhat.com>
9524         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9526 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9528         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9529         cache parameters using detect_caches_amd also for CYRIX,
9530         NSC and TM2 signatures.
9532 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9533             Dzianis Kahanovich  <mahatma@eu.by>
9535         PR target/45359
9536         PR target/46396
9537         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9538         VIA/Centaur processors and determine their cache parameters
9539         using detect_caches_amd.
9541 2013-05-16  Teresa Johnson  <tejohnson@google.com>
9543         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9544         (rtl_verify_edges): New function.
9545         (rtl_verify_bb_insns): Ditto.
9546         (rtl_verify_bb_pointers): Ditto.
9547         (rtl_verify_bb_insn_chain): Ditto.
9548         (rtl_verify_fallthru): Ditto.
9549         (rtl_verify_bb_layout): Ditto.
9550         (rtl_verify_flow_info_1): Outline checks into new functions.
9551         (rtl_verify_flow_info): Ditto.
9553 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
9555         * cfghooks.c (copy_bbs): Add update_dominance argument.
9556         * cfghooks.h (copy_bbs): Update prototype.
9557         * tree-cfg.c (gimple_duplicate_sese_region):
9558         Add update_dominance argument.
9559         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9560         * tree-ssa-loop-ch.c (copy_loop_headers): Update
9561         gimple_duplicate_sese_region call.
9562         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9563         Update copy_bbs call.
9564         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9565         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9567 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9569         * tree-vectorizer.h (NUM_PATTERNS): Increment.
9570         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9571         vect_recog_rotate_pattern.
9572         (vect_recog_rotate_pattern): New function.
9574 2013-05-16  Jason Merrill  <jason@redhat.com>
9576         * Makefile.in (LLINKER): New variable.
9577         (mostlyclean): Remove link mutex.
9578         * configure.ac: Handle --enable-link-mutex.
9579         * lock-and-run.sh: New script.
9581 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9583         PR target/19599
9584         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9585         for NULL decl.
9587 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9589         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9591 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
9593         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9594         * config/arm/arm.c (next_consecutive_mem): New function.
9595         (gen_movmem_ldrd_strd): Likewise.
9596         * config/arm/arm.md (movmemqi): Update condition and code.
9597         (unaligned_loaddi, unaligned_storedi): New patterns.
9599 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9601         * config.gcc: Obsolete *-*-solaris2.9*.
9602         * doc/install.texi (Specific, *-*-solaris2*): Document it.
9604 2013-05-16  Richard Biener  <rguenther@suse.de>
9606         * passes.c (init_optimization_passes): Move pass_parallelize_loops
9607         earlier, after GRAPHITE transforms and IV canonicalization.
9609 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9611         * omp-low.c (extract_omp_for_data): For collapsed loops,
9612         if at least one of the loops is known at compile time to
9613         iterate zero times, set count to 0.
9614         (expand_omp_regimplify_p): New function.
9615         (expand_omp_for_generic): For collapsed loops, if at least
9616         one of the loops isn't known to iterate at least once,
9617         add runtime check with setting count to 0.
9618         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9619         For unsigned types if it isn't known at compile time that
9620         the loop will iterate at least once, add runtime check to bypass
9621         the whole loop if initial condition isn't true.
9623 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
9625         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9627 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
9629         PR middle-end/57286
9630         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9631         transformations to avoid an infinite loop.
9633 2013-05-16  Marek Polacek  <polacek@redhat.com>
9635         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9637 2013-05-15  Leif Ekblad  <leif@rdos.net>
9639         * config/i386/i386.c (ix86_decompose_address): Use
9640         DEFAULT_TLS_SEG_REG to access TLS segment register.
9641         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9642         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9643         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9645 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
9647         PR target/57260
9648         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9649         sibling calls to functions that would normally be lazily bound,
9650         unless $gp is call-clobbered.
9652 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
9654         * config/i386/i386.c (ix86_option_override_internal): Update
9655         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
9656         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9657         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9658         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9659         of TARGET_3DNOW.
9660         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9662 2013-05-15  Andreas Schwab  <schwab@suse.de>
9664         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9665         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
9666         third operand.
9668 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9670         * loop-unroll.c (report_unroll_peel): Check decision before
9671         emitting unroll/peel message.
9673 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9675         * function.h (has_bb_partition): New rtl_data flag.
9676         (bb_reorder_complete): Ditto.
9677         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9678         instead of flag_reorder_blocks_and_partition.
9679         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9680         with some enhancements.
9681         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9682         * bb-reorder.c (connect_traces): Check for has_bb_partition
9683         instead of flag_reorder_blocks_and_partition.
9684         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9685         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9686         verify_hot_cold_block_grouping.
9687         (partition_hot_cold_basic_blocks): Set has_bb_partition.
9689 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9691         PR target/19599
9692         * config/arm/predicates.md (call_insn_operand): New predicate.
9693         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
9694         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9695         if insn is not a tail call.
9696         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9697         registers.
9698         * config/arm/arm.h (enum reg_class): New caller save register class.
9699         (REG_CLASS_NAMES): Likewise.
9700         (REG_CLASS_CONTENTS): Likewise.
9701         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9702         without decls.
9704 2013-05-15  Richard Biener  <rguenther@suse.de>
9706         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9707         of MSG_OPTIMIZED_LOCATIONS.
9708         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9709         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9710         message.
9711         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9712         of MSG_OPTIMIZED_LOCATIONS.
9713         (execute_vect_slp): Likewise.
9714         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9715         (vect_create_cond_for_alias_checks): Likewise.
9716         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9717         (vect_recog_widen_mult_pattern): Likewise.
9718         (vect_recog_widen_sum_pattern): Likewise.
9719         (vect_recog_over_widening_pattern): Likewise.
9720         (vect_recog_widen_shift_pattern): Likewise.
9721         (vect_recog_vector_vector_shift_pattern): Likewise.
9722         (vect_recog_divmod_pattern): Likewise.
9723         (vect_recog_mixed_size_cond_pattern): Likewise.
9724         (vect_recog_bool_pattern): Likewise.
9725         (vect_pattern_recog_1): Likewise.
9727 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9729         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9730         non-functions to builtin_unreachable.
9731         * ipa-inline-transform.c (inline_call): Do not assert estimates were
9732         correct when new direct edges were discovered.
9734 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9736         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9737         header, print symbol order instead of node uid, print more information
9738         about indirect edge targets.
9739         (ipa_make_edge_direct_to_target): Print symbol order instead of node
9740         uids.
9741         (ipa_make_edge_direct_to_target): Likewise.
9742         (remove_described_reference): Likewise.
9743         (propagate_controlled_uses): Likewise.
9744         (ipa_print_node_params): Also print symbol order.
9745         (ipcp_transform_function): Print symbol order instead of node uids.
9746         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9747         (cgraph_get_create_real_symbol_node): Likewise.
9748         * ipa-cp.c (print_lattice): Likewise.
9749         (print_all_lattices): Likewise.
9750         (determine_versionability): Likewise.
9751         (initialize_node_lattices): Likewise.
9752         (estimate_local_effects): Likewise.
9753         (update_profiling_info): Likewise.
9754         (create_specialized_node): Likewise.
9755         (perhaps_add_new_callers): Likewise.
9756         (decide_about_value): Likewise.
9757         (decide_whether_version_node): Likewise.
9758         (identify_dead_nodes): Likewise.
9759         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9760         (dump_inline_summary): Likewise.
9761         (estimate_node_size_and_time): Likewise.
9762         (inline_analyze_function): Likewise.
9763         * ipa-inline.c (report_inline_failed_reason): Likewise.
9764         (want_early_inline_function_p): Likewise.
9765         (edge_badness): Likewise.
9766         (update_edge_key): Likewise.
9767         (inline_small_functions): Likewise.  Add dumping of order to two other
9768         dumps.
9769         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9770         instead of node uids.
9771         (propagate_pure_const): Likewise.
9772         (propagate_pure_const): Likewise.
9773         * ipa-utils.c (dump_cgraph_node_set): Likewise.
9774         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9775         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9776         of node uids.
9777         * tree-pretty-print.c (dump_function_header): Likewise.
9778         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9779         Print symbol order instead of node uids.
9781 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9783         * config/s390/s390.c (s390_register_move_cost): Don't impose the
9784         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9786 2013-05-15  Richard Biener  <rguenther@suse.de>
9788         PR tree-optimization/57275
9789         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9790         return value for fail to do runtime alias checks for gather loads.
9792 2013-05-15  Jan Hubicka  <jh@suse.cz>
9794         PR lto/57038
9795         PR lto/47375
9796         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9797         weakrefs are not external.
9798         (lto_symtab_merge_decls): Fix thinko when dealing with
9799         non-lto_symtab decls.
9800         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9801         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9802         * varpool.c (dump_varpool_node): Dump more flags.
9804 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9806         * config/i386/i386.c (processor_alias_table): Add instruction
9807         FSGSBASE for AMD bdver3 architecture.
9809 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9811         * tree.c (warn_deprecated_use): Print file:line using locus color.
9812         * diagnostic.c (diagnostic_report_current_module): Print file:line
9813         and file:line:column using locus color.
9815 2013-05-14  Mike Stump  <mikestump@comcast.net>
9817         * gdbinit.in: Add __null.
9819 2013-05-14  Mike Stump  <mikestump@comcast.net>
9821         * recog.h: Rename struct recog_data to Recog_data.
9822         * recog.c: Likewise.
9823         * reload.c (can_reload_into): Likewise.
9824         * config/picochip/picochip.c: Likewise.
9826 2013-05-14  Mike Stump  <mikestump@comcast.net>
9828         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9830 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9832         * resource.h (struct resources): Remove unch_memory member.
9833         (CLEAR_RESOURCE): Don't clear unch_memory.
9834         * resource.c (mark_referenced_resources): Don't set it.
9835         (mark_set_resources): Likewise.
9836         (mark_target_live_regs): Don't clear it.
9837         (init_resource_info): Likewise.
9838         * reorg.c (resource_conflicts_p): Don't compare it.
9839         (redundant_insn): Don't set it.
9841         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9842         Remove prototypes.
9843         * emit-rtl.c (next_label): Remove unused function.
9844         (skip_consecutive_labels, link_cc0_insns): Move to ...
9845         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9846         only place where these functions are used, and make them static.
9848 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9850         * fold-const.c (fold_negate_expr): Handle vectors.
9851         (fold_truth_not_expr): Make it static.
9852         (fold_invert_truthvalue): New static function.
9853         (invert_truthvalue_loc): Handle vectors. Do not call
9854         fold_truth_not_expr directly.
9855         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9856         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9857         (fold_comparison): Handle vector constants.
9858         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9859         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9860         * tree.h (fold_truth_not_expr): Remove declaration.
9862 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9864         * config/aarch64/aarch64-simd.md
9865         (aarch64_vcond_internal<mode>): Rename to...
9866         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9867         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9868         float modes. Clarify all iterator modes.
9869         (vcond<mode><mode>): Use new name for vcond expanders.
9870         (vcond<v_cmp_result><mode>): Likewise.
9871         (vcondu<mode><mode>: Likewise.
9872         * config/aarch64/iterators.md (VDQF_COND): New.
9874 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9876         PR bootstrap/57266
9877         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9878         variable for the shift amount. Check that we shift by non-negative
9879         amounts.
9881 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9883         PR target/42017
9884         * config/arm/arm.h (EPILOGUE_USES): Only return true
9885         for LR_REGNUM after epilogue_completed.
9887 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
9889         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9890         is error_mark_node.
9892 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9894         PR target/57261
9895         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9896         and Solaris 11+/x86 with gld.
9897         * configure: Regenerate.
9899 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9901         * expmed.c (expand_shift_1): Canonicalize rotates by
9902         constant bitsize / 2 to bitsize - 1.
9903         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9904         case ROTATERT>: Likewise.
9906         Revert:
9907         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9909         * config/i386/i386.md (rotateinv): New code attr.
9910         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9911         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9912         roll $31, %eax, etc.
9914 2013-05-14  Richard Biener  <rguenther@suse.de>
9916         PR middle-end/57235
9917         * tree-eh.c (sink_clobbers): Give up for successors with
9918         multiple predecessors and no virtual uses.
9920 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9922         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9923         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9925 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9927         PR middle-end/57251
9928         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9929         the case when both op0 and op1 have VOIDmode.
9931 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9933         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9934         in multiply-accumulate mode.
9936 2013-05-13  Guozhi Wei  <carrot@google.com>
9938         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9940 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9942         PR target/56975
9943         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9944         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9945         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9946         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9947         (ix86_expand_prologue): Likewise.
9948         (ix86_expand_split_stack_prologue): Likewise.
9949         (legitimate_pic_address_disp_p): Likewise.
9950         (legitimize_pic_address): Likewise.
9951         (legitimize_tls_address): Likewise.
9952         (legitimize_pe_coff_symbol): Likewise.
9953         (output_pic_addr_const): Likewise.
9954         (construct_plt_address): Likewise.
9955         (ix86_expand_call): Likewise.
9956         (x86_output_mi_thunk): Likewise.
9957         (x86_function_profiler): Likewise.
9959 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9961         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9962         similar switch cases.
9963         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9964         (aarch64_simd_mov_to_<mode>low): Delete.
9965         (aarch64_simd_mov_to_<mode>high): Delete.
9966         (move_lo_quad_<mode>): Add w<-r alternative.
9967         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9968         (aarch64_simd_mov_from_*): Update type attribute.
9969         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9970         statement.
9972 2013-05-13  Jan Hubicka  <jh@suse.cz>
9974         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9975         * config/i386/i386.c (ix86_compute_frame_layout,
9976         ix86_expand_epilogue, emit_i387_cw_initialization,
9977         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9978         ix86_local_alignment): Fix use of size/speed predicates.
9980 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9982         PR tree-optimization/45216
9983         PR tree-optimization/57157
9984         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9985         the (-Y) & (B - 1) variant if OP is |.
9986         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9987         return shifted.  Use (-op1) & (prec - 1) as other_amount
9988         instead of prec - op1.
9990 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9992         PR middle-end/42371
9993         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9994         (ipa_constant_data): New type.
9995         (ipa_jump_func): Use ipa_constant_data to hold information about
9996         constant jump functions.
9997         (ipa_get_jf_constant): Adjust to jump function type changes.
9998         (ipa_get_jf_constant_rdesc): New function.
9999         (ipa_param_descriptor): New field controlled_uses.
10000         (ipa_get_controlled_uses): New function.
10001         (ipa_set_controlled_uses): Likewise.
10002         * ipa-ref.h (ipa_find_reference): Declare.
10003         * ipa-prop.c (ipa_cst_ref_desc): New type.
10004         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
10005         changes.
10006         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
10007         New parameter cs.  Adjust all callers.
10008         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
10009         (remove_described_reference): New function.
10010         (jfunc_rdesc_usable): Likewise.
10011         (try_make_edge_direct_simple_call): Decrement controlled use count,
10012         attempt to remove reference if it hits zero.
10013         (combine_controlled_uses_counters): New function.
10014         (propagate_controlled_uses): Likewise.
10015         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
10016         (ipa_edge_duplication_hook): Duplicate reference descriptions.
10017         (ipa_print_node_params): Print described use counter.
10018         (ipa_write_jump_function): Adjust to jump function type changes.
10019         (ipa_read_jump_function): New parameter CS, pass it to
10020         ipa_set_jf_constant.  Adjust caller.
10021         (ipa_write_node_info): Stream controlled use count
10022         (ipa_read_node_info): Likewise.
10023         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
10024         asserting.
10025         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
10026         count.  Remove cloning-added reference if it reaches zero.
10027         * ipa-ref.c (ipa_find_reference): New function.
10029 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10031         * config/i386/i386.c (processor_target_table): Modified default
10032         alignment values for AMD BD and BT architectures.
10034 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
10036         * tree-vect-generic.c (uniform_vector_p): Move ...
10037         * tree.c (uniform_vector_p): ... here.
10038         * tree.h (uniform_vector_p): Declare it.
10039         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
10040         into a scalar.
10042 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
10044         PR tree-optimization/57230
10045         * tree-ssa-strlen.c (handle_char_store): Record length for
10046         array store from STRING_CST.
10048         PR tree-optimization/57230
10049         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
10050         check.
10052 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
10054         * config/epiphany/epiphany.c (epiphany_init): Check size of
10055         NUM_MODES_FOR_MODE_SWITCHING.
10056         (epiphany_expand_prologue):
10057         Remove CONFIG_REGNUM initial value handling code.
10058         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
10059         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
10060         (emit_set_fp_mode, epiphany_mode_after): Likewise.
10061         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
10062         Don't return 1 for FP_MODE_NONE.
10063         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
10064         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
10065         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
10066         * config/epiphany/epiphany.md (save_config): New pattern.
10068 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
10070         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
10072 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
10074         * config/i386/i386.md (memory): Handle sseishft1.
10075         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
10076         (*vec_extractv2di_1): Ditto.
10078 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
10080         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
10081         saved registers.
10083 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
10085         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
10086         Add mthumb/march=armv7-a multilib.
10087         Add mthumb/march=armv7-r multilib.
10088         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10090 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
10092         * config/v850/t-rtems: Add more multilibs.
10094 2013-05-10  Richard Biener  <rguenther@suse.de>
10096         PR tree-optimization/57214
10097         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10098         not propagate from SSA names that occur in abnormal PHI nodes.
10100 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
10102         * stor-layout.c (element_precision): New function.
10103         * machmode.h (element_precision): Declare it.
10104         * tree.c (build_minus_one_cst): New function.
10105         (element_precision): Likewise.
10106         * tree.h (build_minus_one_cst): Declare new function.
10107         (element_precision): Likewise.
10108         * fold-const.c (operand_equal_p): Use element_precision.
10109         (fold_binary_loc): Handle vector types.
10110         * convert.c (convert_to_integer): Use element_precision.
10111         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10112         separately.
10114 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
10116         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10117         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10118         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10119         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10120         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10121         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10122         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10123         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10124         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10125         (Uuw8): New constraints.
10126         (Usb4): Move into alphabetical order.
10127         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10128         (sd8_operand, ub8_operand, uw8_operand): New predicates.
10129         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10130         previously unnamed patterns.
10131         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10132         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10133         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10134         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10135         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
10136         of specific lengths.
10138 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
10140         * config/i386/i386.md (rotateinv): New code attr.
10141         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10142         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10143         roll $31, %eax, etc.
10145         PR tree-optimization/45216
10146         PR tree-optimization/57157
10147         * tree-ssa-forwprop.c (simplify_rotate): New function.
10148         (ssa_forward_propagate_and_combine): Call it.
10150 2013-05-10  Richard Biener  <rguenther@suse.de>
10152         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10153         disable peeling when we version for aliasing.
10154         (vector_alignment_reachable_p): Honor explicit user alignment.
10155         (vect_supportable_dr_alignment): Likewise.
10156         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10157         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10158         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10159         then peeling to arrange for the cost-model check to come first.
10161 2013-05-10  Alan Modra  <amodra@gmail.com>
10163         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10164         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10165         * configure: Regenerate.
10167 2013-05-10  Alan Modra  <amodra@gmail.com>
10169         PR target/55033
10170         * varasm.c (default_elf_select_section): Move !DECL_P check..
10171         (get_named_section): ..to here before calling get_section_name.
10172         Adjust assertion.
10173         (default_section_type_flags): Add DECL_P check.
10174         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10175         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10177 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
10179         * config/epiphany/epiphany.c (epiphany_expand_prologue):
10180         When using gen_stack_adjust_str with a register offset, add a
10181         REG_FRAME_RELATED_EXPR note.
10183 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
10185         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10186         (*vec_extractv4si_zext_mem): Ditto.
10187         (*vec_extractv2di): Add 0->x and x->x alternatives.
10188         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10189         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10191 2013-05-09  Jason Merrill  <jason@redhat.com>
10193         N3639 C++1y VLA support
10194         * gimplify.c (gimplify_vla_decl): Don't touch an existing
10195         DECL_VALUE_EXPR.
10197         * tree.c (build_constructor_va): New.
10198         * tree.h: Declare it.
10200 2013-05-09  Martin Jambor  <mjambor@suse.cz>
10202         PR lto/57084
10203         * gimple-fold.c (canonicalize_constructor_val): Call
10204         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10206 2013-05-09  Jan Hubicka  <jh@suse.cz>
10207             Richard Biener  <rguenther@suse.de>
10209         PR lto/54095
10210         * symtab.c (symtab_make_decl_local): Do not add private names.
10212 2013-05-09  Jan Hubicka  <jh@suse.cz>
10214         PR lto/54095
10215         * symtab.c (insert_to_assembler_name_hash): Handle clones.
10216         (unlink_from_assembler_name_hash): Likewise.
10217         (symtab_prevail_in_asm_name_hash, symtab_register_node,
10218         symtab_unregister_node, symtab_initialize_asm_name_hash,
10219         change_decl_assembler_name): Update.
10221 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
10223         * config/aarch64/aarch64.md: New movtf split.
10224         (*movtf_aarch64): Update.
10225         (aarch64_movdi_tilow): Handle TF modes and rename to
10226         aarch64_movdi_<mode>low.
10227         (aarch64_movdi_tihigh): Handle TF modes and rename to
10228         aarch64_movdi_<mode>high
10229         (aarch64_movtihigh_di): Handle TF modes and rename to
10230         aarch64_mov<mode>high_di
10231         (aarch64_movtilow_di): Handle TF modes and rename to
10232         aarch64_mov<mode>low_di
10233         (aarch64_movtilow_tilow): Remove spurious whitespace.
10234         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10235         splits.
10236         (aarch64_print_operand): Update.
10238 2013-05-09  Alan Modra  <amodra@gmail.com>
10240         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10241         powerpc64le.
10242         * configure: Regenerate.
10244 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10246         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10247         splitter preparation statements.
10248         * config/i386/sse.md (*vec_extract* splitters): Ditto.
10249         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10250         adjust_address_nv.
10252 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10254         * gimple-ssa-strength-reduction.c (count_candidates): Change
10255         return value to int.
10256         (analyze_candidates_and_replace): Change type of length to int.
10258 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10260         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10261         (*vec_extract<mode>): Use VI12_128 mode iterator.
10262         (*vec_extract<mode>_mem): Ditto.
10263         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10264         attribute.
10266 2013-05-08  Diego Novillo  <dnovillo@google.com>
10268         PR bootstrap/54659
10270         Revert:
10271         2012-08-17  Diego Novillo  <dnovillo@google.com>
10273         PR bootstrap/54281
10274         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10275         * config.in: Regenerate.
10276         * configure: Regenerate.
10277         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10279 2013-05-08  Jan Hubicka  <jh@suse.cz>
10281         PR lto/54095
10282         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10283         * cgraph.h (symtab_node_base): Add unique_name.
10284         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10285         input_overwrite_node, input_varpool_node): Stream unique_name.
10286         * cgraphclones.c (cgraph_create_virtual_clone,
10287         cgraph_function_versioning): Set unique_name.
10288         * ipa.c (function_and_variable_visibility): Set unique_name.
10290 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10292         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10293         (alloc_cand_and_find_basis): Restrict conditional candidate
10294         processing to CAND_MULTs.
10296 2013-05-08  Jan Hubicka  <jh@suse.cz>
10298         PR lto/54095
10299         lto-symtab.c (lto_symtab_symbol_p): New function.
10300         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10301         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10302         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10303         Skip static symbols.
10305 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
10307         PR tree-optimization/57200
10308         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10309         Only call inform if the preceding warning_at returns true.
10311 2013-05-07  Han Shen  <shenhan@google.com>
10313         * cfgexpand.c (record_or_union_type_has_array_p): New function.
10314         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10315         * common.opt (fstack-protector-strong): New option.
10316         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10317         * doc/invoke.texi (Optimization Options): Document
10318         "-fstack-protector-strong".
10319         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10321 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10323         * config/mips/mips.c (mips_machine_reorg2): Return 0.
10325 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
10327         * ira.c (update_equiv_regs): Add insn having equiv memory even if
10328         it is not lhs of the insn.
10329         (setup_reg_equiv): Remove insn having equiv memory which it is not
10330         lhs of the insn.
10331         * lra-constraints.c (process_address): Try to improve generation
10332         code for address base + disp.
10333         (lra_constraints): Make correct the code for checking insn setting
10334         up backward equivalence.  Remove insn only if it is in the init
10335         insn list.
10336         * lra-eliminations.c (update_reg_eliminate): Change return value.
10337         (lra_eliminate): Use the result.
10339 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
10341         * config/i386/sse.md (ssescalarnummask): New mode attribute.
10342         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10343         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10344         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10345         register target operands.
10346         (*vec_extractv8hi_sse2): New pattern.
10347         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10348         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10349         (*vec_extract<mode>_mem): New insn and split pattern.
10351 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
10353         * config/arm/arm.c (arm_asan_shadow_offset): New function.
10354         (TARGET_ASAN_SHADOW_OFFSET): Define.
10355         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10356         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10358 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10360         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10361         (incr_vec_index): Return -1 if increment not found.
10362         (create_add_on_incoming_edge): Assert if increment not found.
10363         (record_increment): Limit number of increments recorded.
10364         (all_phi_incrs_profitable): Return false if an increment not found.
10365         (replace_profitable_candidates): Don't process increments that were
10366         not recorded.
10367         (analyze_candidates_and_replace): Limit size of incr_vec.
10369 2013-05-07  Richard Biener  <rguenther@suse.de>
10371         * calls.c (special_function_p): setjmp-like functions are leaf.
10372         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10373         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10375 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
10377         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10378         (aarch64_simd_mov<mode>): New expander.
10379         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10380         (aarch64_simd_mov_to_<mode>high): Likewise.
10381         (aarch64_simd_mov_from_<mode>low): Likewise.
10382         (aarch64_simd_mov_from_<mode>high): Likewise.
10383         (aarch64_dup_lane<mode>): Update.
10384         (aarch64_dup_lanedi): New instruction pattern.
10385         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10386         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10388 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10390         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10391         (replace_mult_candidate): Remove unnecessary argument; remove
10392         unnecessary parameter from call to introduce_cast_before_cand.
10393         (replace_unconditional_candidate): Remove unnecessary parameter
10394         from call to replace_mult_candidate.
10395         (replace_conditional_candidate): Likewise.
10396         (insert_initializers): Use make_temp_ssa_name.
10397         (introduce_cast_before_cand): Remove unnecessary argument; use
10398         make_temp_ssa_name.
10399         (replace_one_candidate): Remove unnecessary argument; remove
10400         unnecessary parameter from calls to introduce_cast_before_cand.
10401         (replace_profitable_candidates): Remove unnecessary parameters
10402         from calls to replace_one_candidate.
10404 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10406         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10407         phi def as possibly hiding a basis for a CAND_ADD whose operands
10408         have been commuted in the analysis.
10409         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10411 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10413         * config/aarch64/aarch64.md
10414         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10415         shift value between 0-4.
10417 2013-05-07  Richard Biener  <rguenther@suse.de>
10419         * double-int.h (rshift): New overload.
10420         * double-int.c (rshift): New function.
10421         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10422         (create_reference_ops_from_ref): Remove.
10423         (vn_reference_insert): Use shared ops for constructing the
10424         reference and copy it.
10426 2013-05-07  Richard Biener  <rguenther@suse.de>
10428         PR middle-end/57190
10429         * tree-eh.c (sink_clobbers): Properly propagate
10430         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10432 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
10434         PR tree-optimization/57149
10435         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10436         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10437         collect_phi_def_edges, execute_late_warn_uninitialized): Use
10438         uninit_undefined_value_p instead of ssa_undefined_value_p.
10440         PR debug/57184
10441         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10442         for modifier == EXPAND_INITIALIZER.
10444 2013-05-07  Anton Blanchard  <anton@samba.org>
10446         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10447         for powerpc64 little endian.
10448         * configure: Regenerate.
10450 2013-05-06  Graham Stott  <grahams@btinternet.com>
10452         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10453         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10454         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10455         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10457 2013-05-06  Graham Stott  <grahams@btinternet.com>
10459         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10460         codes which allow non-lvalues.
10462 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
10464         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10465         components are all 1s.
10466         (integer_minus_onep): New function.
10467         * tree.h (integer_minus_onep): Declare it.
10468         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10469         integer_minus_onep instead of integer_all_onesp.
10471 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10473         PR target/52933
10474         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10475         variations of these patterns.
10477 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10479         * config/i386/i386.md (isa): Add x64_sse4 member.
10480         (enabled): Handle x64_sse4.
10481         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10482         instruction for 64bit SSE4_1 targets.  Update insn attributes.
10483         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10484         instruction for SSE4_1 targets.  Update insn attributes.
10485         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10486         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10487         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10488         const_1 selector.
10489         (*vec_extractv4si): Rename from *sse4_1_pextrd.
10490         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10491         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10493 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10495         PR target/57108
10496         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10498 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
10500         * final.c (do_assembler_dialects): Don't handle curly braces and
10501         vertical bar escaped by % as dialect delimiters.
10502         (output_asm_insn): Print curly braces and vertical bar if escaped
10503         by % and ASSEMBLER_DIALECT defined.
10504         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10505         * doc/tm.texi: Regenerated.
10507 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10509         * config/mips/mips.c: Include tree-pass.h.
10510         (mips_reorg): Split in pre- and post-dbr_schedule parts.
10511         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10512         (pass_mips_machine_reorg2): New machine specific pass.
10513         (insert_pass_mips_machine_reorg2): New pass plugin definition.
10514         (mips_option_override): Register the new pass.
10515         * rtl.h (cleanup_barriers): Remove prototype.
10516         (dbr_schedule): Likewise.
10517         * jump.c (cleanup_barriers): Make static.
10518         * reorg.c (dbr_schedule): Likewise.
10520 2013-05-06  Richard Biener  <rguenther@suse.de>
10522         PR tree-optimization/57185
10523         * tree-parloops.c (add_field_for_reduction): Handle anonymous
10524         SSA names properly.
10526 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10528         PR target/57106
10529         * config/i386/i386.c (add_parameter_dependencies): Add dependence
10530         between "first_arg" and "insn", not "last" and "insn".
10532 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10534         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10535         (find_candidates_in_block): Re-enable slsr_process_phi.
10536         (create_phi_basis): Fix double counting of candidate adjustment.
10538 2013-05-06  Richard Biener  <rguenther@suse.de>
10540         PR middle-end/57147
10541         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10542         the edge is also fallthru, preserve it and just clear the
10543         abnormal flag.
10544         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10545         also complex, preserve that and just clear the fallthru flag.
10546         * tree-inline.c (update_ssa_across_abnormal_edges): Also
10547         update virtual operands.
10549 2013-05-06  Alan Modra  <amodra@gmail.com>
10551         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10552         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10553         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10554         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10555         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10556         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10558 2013-05-06  Alan Modra  <amodra@gmail.com>
10560         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10561         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
10562         (DEFAULT_ASM_ENDIAN): Define.
10563         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10564         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10565         Update -K PIC clause from sysv4.h.
10566         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10567         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10569 2013-05-06  Alan Modra  <amodra@gmail.com>
10571         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10572         twice for little-endian.
10573         (ashrdi3_no_power, ashrdi3): Support little-endian.
10575 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10577         PR target/55303
10578         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10579         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10580         related expanders.
10581         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10582         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10583         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10584         New predicates.
10586 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
10587             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
10589         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10590         * config/pa/pa.opt: Make mbig-switch a no-op.
10591         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10592         (CASE_VECTOR_MODE): Always return SImode.
10593         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10594         for the !TARGET_BIG_SWITCH case.
10595         * config/pa/pa-linux.h: Likewise.
10596         * config/pa/pa-openbsd.h: Likewise.
10597         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10598         * config/pa/pa.md (short_jump): Remove define_insn.
10599         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10600         (casesi0): Remove define_insn.
10601         (type): Remove btable_branch.
10602         (pa_combine_type): Likewise.
10603         (in_nullified_branch_delay): Likewise.
10604         (in_call_delay): Likewise.
10605         (define_delay): Likewise.
10606         (define_insn_reservation "Z3"): Likewise.
10607         (define_insn_reservation "Z4"): Likewise.
10608         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10609         (pa_adjust_insn_length): Remove adjustment for btable branches.
10610         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10611         and mno-big-switch
10613 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
10615         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10616         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10617         Add m->r,x alternatives.
10618         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10619         splitters using SWI48x mode iterator.
10620         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
10621         TARGET_64BIT.  Add m->x alternative.
10622         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10623         Add o->x alternative.  Enable for TARGET_SSE.
10624         (sse_storeq): Remove expander.
10625         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10626         with memory input operand.
10627         (*vec_extractv2di_1 splitter): New.
10628         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10629         * config/i386/i386.md (ssevecmodelower): New mode attribute.
10631 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10633         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
10634         (INT_LOWPART): Delete.
10635         (extract_MB): Adjust.
10636         (extract_ME): Adjust.
10637         (print_operand): Adjust.
10639 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10641         * config/rs6000/predicates.md (reg_or_add_cint_operand,
10642         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10643         (reg_or_logical_cint_operand, easy_fp_constant,
10644         logical_const_operand): Delete "CONST_DOUBLE" case.
10645         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10646         "HOST_BITS_PER_WIDE_INT == 64" test.
10647         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
10648         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10649         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10650         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10651         test.
10652         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10653         CONST_DOUBLE DImode/VOIDmode case.
10654         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10655         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10656         CONST_DOUBLE VOIDmode case.
10657         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10658         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10659         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10660         Delete CONST_DOUBLE case.
10661         (splitters for mov FMOVE64 const_double): Delete
10662         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10663         "HOST_BITS_PER_WIDE_INT >= 64" test.
10664         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10665         case.
10666         (mov DI const_double): Delete.
10668 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
10670         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10671         on op shows all bits zero in mode of a lowpart subreg, return zero.
10673 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
10675         PR target/57150
10676         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10677         to save TFmode registers and DImode to save TImode registers for
10678         caller save operations.
10679         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10680         mark being partially clobbered since they only use the first
10681         double word.
10683         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10684         and TDmode only use the upper 64-bits of each VSX register.
10686 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10688         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10689         (find_candidates_in_block): Disable slsr_process_phi.
10691 2013-05-03  Guozhi Wei  <carrot@google.com>
10693         * coverage.c (coverage_obj_init): Move the construction of gcov
10694         constructor to ...
10695         (build_init_ctor): ... here.
10697 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10699         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10700         (slsr_cand_d): Redefine def_phi.
10701         (stride_status, phi_adjust_status, count_phis_status): New enums.
10702         (find_phi_def): New.
10703         (find_basis_for_base_expr): New.
10704         (find_basis_for_candidate): Handle hidden bases.
10705         (alloc_cand_and_find_basis): Handle phi candidates.
10706         (slsr_process_phi): New.
10707         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10708         (create_mul_imm_cand): Likewise.
10709         (create_add_ssa_cand): Exclude phi base candidates.
10710         (create_add_imm_cand): Likewise.
10711         (slsr_process_cast): Likewise.
10712         (slsr_process_copy): Likewise.
10713         (find_candidates_in_block): Handle phi candidates.
10714         (dump_candidate): Likewise.
10715         (unconditional_cands): Delete.
10716         (unconditional_cands_with_known_stride_p): Delete.
10717         (phi_dependent_cand_p): New.
10718         (cand_increment): Handle phi-dependent candidates.
10719         (replace_dependent): Delete.
10720         (replace_mult_candidate): New.
10721         (replace_unconditional_candidate): New.
10722         (incr_vec_index): Move to avoid forward reference.
10723         (create_add_on_incoming_edge): New.
10724         (create_phi_basis): New.
10725         (replace_dependents): Delete.
10726         (replace_conditional_candidate): New.
10727         (phi_add_costs): New.
10728         (replace_uncond_cands_and_profitable_phis): New.
10729         (record_increment): Handle phi adjustments.
10730         (record_phi_increments): New.
10731         (record_increments): Handle phi adjustments.
10732         (phi_incr_cost): New.
10733         (lowest_cost_path): Handle phis.
10734         (total_savings): Likewise.
10735         (analyze_increments): Likewise.
10736         (ncd_with_phi): New.
10737         (ncd_of_cand_and_phis): New.
10738         (nearest_common_dominator_for_cands): Handle phi increments.
10739         (all_phi_incrs_profitable): New.
10740         (replace_profitable_candidates): Handle phi-dependent candidates.
10741         (analyze_candidates_and_replace): Likewise.
10743 2013-05-03  Teresa Johnson  <tejohnson@google.com>
10745         PR bootstrap/57154
10746         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10747         do not exceed REG_BR_PROB_BASE.
10749 2013-05-03  Jeff Law  <law@redhat.com>
10751         PR tree-optimization/57144
10752         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10753         operand of the condition will bit into the new type when eliminating
10754         a cast feeding a condition.
10756 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
10758         PR rtl-optimization/57130
10759         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10760         of COMPARE as in_code to the recursive call if needed.
10762 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
10764         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10765         (enabled): Handle new members.
10766         * config/i386/sse.md (*vec_concatv2si): Merge from
10767         *vec_concatv2si_sse2 and vec_concatv2si_sse.
10768         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10770 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
10772         PR tree-optimization/57027
10773         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10774         for fnms opportunity, check we got the prerequisite kind
10775         of tree / gimple before using accessor functions.
10777 2013-05-03  Richard Biener  <rguenther@suse.de>
10779         * double-int.h (lshift): New overload without precision
10780         and arith argument.
10781         (operator *=, operator +=, operator -=): Move ...
10782         * double-int.c (operator *=, operator +=, operator -=): ... here
10783         and implement more efficiently.
10784         (mul_double_with_sign): Remove.
10785         (lshift_double): Adjust to take unsinged shift argument, push
10786         dispatching code to callers.
10787         (mul_double_wide_with_sign): Add early out for callers that
10788         are not interested in high parts or overflow.
10789         (lshift): New function.
10790         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10791         dispatch code here.
10792         (lrotate, rrotate): Use logical shifts.
10793         * expr.c (get_inner_reference): Use lshift.
10794         * fixed-value.c (do_fixed_divide): Likewise.
10795         * tree-dfa.c (get_ref_base_and_extent): Likewise.
10796         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10797         (indirect_refs_may_alias_p): Likewise.
10798         (stmt_kills_ref_p_1): Likewise.
10800 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10802         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10804 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10806         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10807         scalar form of FABD instruction.
10809 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
10811         * lra-constraints.c (process_alt_operands): Add checking alt
10812         number to choose the best alternative.
10814 2013-05-02  Richard Biener  <rguenther@suse.de>
10816         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10817         bitmap and its handling.
10818         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10820 2013-05-02  Richard Biener  <rguenther@suse.de>
10822         PR middle-end/57140
10823         * tree-inline.c (copy_loops): Properly handle removed loops.
10824         (copy_cfg_body): Mark destination loops for fixup if source
10825         loops needed fixup.
10827 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
10829         PR target/56732
10830         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10831         generating simple_return for naked functions.
10833 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10835         PR middle-end/56988
10836         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10837         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10838         flags match.
10839         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10840         ipa_agg_replacement_value structures.
10841         (known_aggs_to_agg_replacement_list): Likewise.
10842         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10843         (read_agg_replacement_chain): Likewise.
10844         (ipcp_transform_function): Also check that by_ref flags match.
10846 2013-05-02  Richard Biener  <rguenther@suse.de>
10848         * graphds.h (struct graph): Add obstack member.
10849         * graphds.c (new_graph): Initialize obstack and allocate
10850         vertices from it.
10851         (add_edge): Allocate edge from the obstack.
10852         (free_graph): Free the obstack instead of all edges and vertices.
10854 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10856         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10857         divides.
10858         * cfg.c (update_bb_profile_for_threading): Ditto.
10859         * tree-inline.c (copy_bb): Ditto.
10860         (copy_edges_for_bb): Ditto.
10861         (initialize_cfun): Ditto.
10862         (copy_cfg_body): Ditto.
10863         (expand_call_inline): Ditto.
10864         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10865         (estimate_node_size_and_time): Ditto.
10866         (inline_merge_summary): Ditto.
10867         * cgraphclones.c (cgraph_clone_edge): Ditto.
10868         (cgraph_clone_node): Ditto.
10869         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10870         (compute_trg_info): Ditto.
10872 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10874         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10875         S reg when fp attribute set.
10876         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10878 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10880         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10881         New pattern.
10882         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10883         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10884         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10886 2013-05-02  Richard Biener  <rguenther@suse.de>
10888         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10889         (struct instantiate_cache_entry): New type.
10890         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10891         (struct instantiate_cache_type): New type.
10892         (set_instantiated_value, get_instantiated_value): Remove.
10893         (get_instantiated_value_entry): New function.
10894         (instantiate_scev_name): Use the new cache and adjust.
10895         (instantiate_scev_poly): Adjust.
10896         (instantiate_scev_binary): Likewise.
10897         (instantiate_array_ref): Likewise.
10898         (instantiate_scev_convert): Likewise.
10899         (instantiate_scev_not): Likewise.
10900         (instantiate_scev_3): Likewise.
10901         (instantiate_scev_2): Likewise.
10902         (instantiate_scev_r): Likewise.
10903         (instantiate_scev): Likewise.
10904         (resolve_mixers): Likewise.
10906 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10908         PR target/57091
10909         * lra-constraints.c (best_small_class_operands_num): Remove.
10910         (process_alt_operands): Remove small_class_operands_num.  Take
10911         small classes operands into losers and only if the operand is not
10912         matched.  Modify debugging output.
10913         (curr_insn_transform): Remove best_small_class_operands_num.
10914         Print insn name.
10916 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10918         * config/aarch64/aarch64-builtins.c
10919         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10920         * config/aarch64/aarch64-simd-builtins.def
10921         (reduc_splus_): Add new modes.
10922         (reduc_uplus_): New.
10923         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10924         (reduc_uplus_v4sf): Likewise.
10925         (reduc_splus_v4sf): Likewise.
10926         (aarch64_addv<mode>): Likewise.
10927         (reduc_uplus_<mode>): Likewise.
10928         (reduc_splus_<mode>): Likewise.
10929         (aarch64_addvv2di): Likewise.
10930         (reduc_uplus_v2di): Likewise.
10931         (reduc_splus_v2di): Likewise.
10932         (aarch64_addvv2si): Likewise.
10933         (reduc_uplus_v2si): Likewise.
10934         (reduc_splus_v2si): Likewise.
10935         (reduc_<sur>plus_<mode>): New.
10936         (reduc_<sur>plus_v2di): Likewise.
10937         (reduc_<sur>plus_v2si): Likewise.
10938         (reduc_<sur>plus_v4sf): Likewise.
10939         (aarch64_addpv4sf): Likewise.
10940         * config/aarch64/arm_neon.h
10941         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10942         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10943         add UNSPEC_SADDV, UNSPEC_UADDV.
10944         (SUADDV): New.
10945         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10947 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10949         * config/aarch64/arm_neon.h
10950         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10952 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10954         * config/aarch64/aarch64-builtins
10955         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10957 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10959         * config/aarch64/aarch64-simd-builtins.def
10960         (reduc_smax_): New.
10961         (reduc_smin_): Likewise.
10962         (reduc_umax_): Likewise.
10963         (reduc_umin_): Likewise.
10964         (reduc_smax_nan_): Likewise.
10965         (reduc_smin_nan_): Likewise.
10966         (fmax): Remove.
10967         (fmin): Likewise.
10968         (smax): Update for V2SF, V4SF and V2DF modes.
10969         (smin): Likewise.
10970         (smax_nan): New.
10971         (smin_nan): Likewise.
10972         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10973         (<su><maxmin><mode>3): ...This, refactor.
10974         (s<maxmin><mode>3): New.
10975         (<maxmin_uns><mode>3): Likewise.
10976         (reduc_<maxmin_uns>_<mode>): Refactor.
10977         (reduc_<maxmin_uns>_v4sf): Likewise.
10978         (reduc_<maxmin_uns>_v2si): Likewise.
10979         (aarch64_<fmaxmin><mode>: Remove.
10980         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10981         new builtin names.
10982         (vmin<q>_f<32,64>): Likewise.
10983         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10984         (FMAXMIN): New.
10985         (su): Add mappings for smax, smin, umax, umin.
10986         (maxmin): New.
10987         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10988         (FMAXMIN): Rename as...
10989         (FMAXMIN_UNS): ...This.
10990         (maxminv): Remove.
10991         (fmaxminv): Likewise.
10992         (fmaxmin): Likewise.
10993         (maxmin_uns): New.
10994         (maxmin_uns_op): Likewise.
10996 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10998         * config/aarch64/arm_neon.h
10999         (vac<ge, gt><sd>_f<32, 64>): Rename to...
11000         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
11001         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
11003 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11005         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
11006         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
11008 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11010         * config/aarch64/aarch64-simd.md
11011         (vcond<mode>_internal): Handle special cases for constant masks.
11012         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
11013         (vcondu<mode><mode>): Likewise.
11014         (vcond<v_cmp_result><mode>): New.
11016 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11018         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
11019         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
11020         * config/aarch64/aarch64-simd-builtins.def
11021         (cmeq): Update to BUILTIN_VALLDI.
11022         (cmgt): Likewise.
11023         (cmge): Likewise.
11024         (cmle): Likewise.
11025         (cmlt): Likewise.
11026         * config/aarch64/arm_neon.h
11027         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
11028         to builtins or C as appropriate.
11030 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11032         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
11033         (cmgeu): ...This.
11034         (cmhi): Rename to...
11035         (cmgtu): ...This.
11036         * config/aarch64/aarch64-simd.md
11037         (simd_mode): Add SF.
11038         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
11039         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
11040         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
11041         (cstore<mode>_neg): ...This.
11042         * config/aarch64/iterators.md
11043         (VALLF): new.
11044         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
11045         (COMPARISONS): New.
11046         (UCOMPARISONS): Likewise.
11047         (optab): Add missing comparisons.
11048         (n_optab): New.
11049         (cmp_1): Likewise.
11050         (cmp_2): Likewise.
11051         (CMP): Likewise.
11052         (cmp): Remove.
11053         (VCMP_S): Likewise.
11054         (VCMP_U): Likewise.
11055         (V_cmp_result): Add DF, SF modes.
11056         (v_cmp_result): Likewise.
11057         (v): Likewise.
11058         (vmtype): Likewise.
11059         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
11061 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
11063         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
11064         define_insn to define_insn_and_split.
11065         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
11066         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
11067         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
11068         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
11069         (thumb2_negscc): Likewise.
11071 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
11073         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
11075 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
11077         * config/arm/thumb2.md: Remove trailing whitespaces.
11079 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11081         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
11082         Use gen_int_mode rather than GEN_INT.
11084 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
11086         * value-prof.c (stream_in_histogram_value): Remove the strayed
11087         debug_gimple_stmt.
11089 2013-04-30  Richard Biener  <rguenther@suse.de>
11091         PR middle-end/57122
11092         * cfghooks.c (split_edge): Properly check for the loop latch edge.
11094 2013-04-30  Richard Biener  <rguenther@suse.de>
11096         PR middle-end/57107
11097         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11099 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
11101         PR rtl-optimization/56957
11102         PR rtl-optimization/57105
11103         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11104         variable.  Use just INSN_UID for determining whether an insn
11105         should be only disconnected from the insn stream.
11106         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11108 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
11110         PR tree-optimization/57104
11111         * tsan.c (instrument_expr): Don't instrument accesses to
11112         DECL_HARD_REGISTER VAR_DECLs.
11114 2013-04-30  Richard Biener  <rguenther@suse.de>
11116         * function.h (loops_for_fn): New inline function.
11117         (set_loops_for_fn): Likewise.
11118         * cfgloop.h (place_new_loop): Add struct function parameter.
11119         (get_loop): Likewise.
11120         (get_loops): Likewise.
11121         (number_of_loops): Likewise.
11122         (fel_next): Adjust.
11123         (fel_init): Likewise.
11124         * cfg.c (get_loop_copy): Adjust.
11125         * cfgloop.c (flow_loops_dump): Likewise.
11126         (record_loop_exits): Likewise.
11127         (verify_loop_structure): Likewise.
11128         * cfgloopanal.c (mark_irreducible_loops): Likewise.
11129         (estimate_reg_pressure_cost): Likewise.
11130         (mark_loop_exit_edges): Likewise.
11131         * cfgloopmanip.c (place_new_loop): Likewise.
11132         (add_loop): Likewise.
11133         (duplicate_loop): Likewise.
11134         * graph.c (draw_cfg_nodes): Likewise.
11135         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11136         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11137         (extract_affine_chrec): Likewise.
11138         (build_scop_iteration_domain): Likewise.
11139         * graphite.c (graphite_initialize): Likewise.
11140         * ira-build.c (create_loop_tree_nodes): Likewise.
11141         (more_one_region_p): Likewise.
11142         (rebuild_regno_allocno_maps): Likewise.
11143         (mark_loops_for_removal): Likewise.
11144         (mark_all_loops_for_removal): Likewise.
11145         (remove_unnecessary_regions): Likewise.
11146         (ira_build): Likewise.
11147         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11148         * loop-init.c (fix_loop_structure): Likewise.
11149         (gate_rtl_move_loop_invariants): Likewise.
11150         (gate_rtl_unswitch): Likewise.
11151         (gate_rtl_unroll_and_peel_loops): Likewise.
11152         (rtl_doloop): Likewise.
11153         * lto-streamer-in.c (input_cfg): Likewise.
11154         * lto-streamer-out.c (output_cfg): Likewise.
11155         * modulo-sched.c (sms_schedule): Likewise.
11156         * predict.c (tree_estimate_probability): Likewise.
11157         (tree_estimate_probability_driver): Likewise.
11158         (estimate_loops): Likewise.
11159         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11160         (move_sese_region_to_fn): Likewise.
11161         (debug_loop_num): Likewise.
11162         * tree-chrec.c (chrec_evaluate): Likewise.
11163         (hide_evolution_in_other_loops_than_loop): Likewise.
11164         (chrec_component_in_loop_num): Likewise.
11165         (reset_evolution_in_loop): Likewise.
11166         (evolution_function_is_invariant_rec_p): Likewise.
11167         * tree-if-conv.c (main_tree_if_conversion): Likewise.
11168         * tree-inline.c (copy_loops): Likewise.
11169         (copy_cfg_body): Likewise.
11170         (tree_function_versioning): Likewise.
11171         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11172         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11173         Likewise.
11174         (add_to_evolution_1): Likewise.
11175         (scev_const_prop): Likewise.
11176         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11177         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11178         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11179         (tree_ssa_lim_initialize): Likewise.
11180         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11181         (verify_loop_closed_ssa): Likewise.
11182         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
11183         (tree_ssa_loop_im): Likewise.
11184         (tree_ssa_loop_unswitch): Likewise.
11185         (tree_vectorize): Likewise.
11186         (check_data_deps): Likewise.
11187         (tree_ssa_loop_ivcanon): Likewise.
11188         (tree_ssa_loop_bounds): Likewise.
11189         (tree_complete_unroll): Likewise.
11190         (tree_complete_unroll_inner): Likewise.
11191         (tree_parallelize_loops): Likewise.
11192         (tree_ssa_loop_prefetch): Likewise.
11193         (tree_ssa_loop_ivopts): Likewise.
11194         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11195         * tree-vectorizer.c (vectorize_loops): Likewise.
11197 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
11199         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11200         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11201         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11202         with EABI_LINK_SPEC.
11204 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11206         PR target/44578
11207         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11208         alternative.
11210 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
11212         PR target/57097
11213         * lra-constraints.c (process_alt_operands): Discourage a bit more
11214         using memory for pseudos.  Print cost dump for alternatives.
11215         Modify cost values for conflicts with early clobbers.
11216         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11218 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11220         PR target/57098
11221         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11223 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11225         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11226         from/to S register.
11227         (movdi_aarch64): Support LDR/STR from/to D register.
11229 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11231         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11232         or higher by default.
11234 2013-04-29  Richard Biener  <rguenther@suse.de>
11236         PR middle-end/57075
11237         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11238         even if not adding abnormal edges for calls that can make
11239         abnormal gotos.
11241 2013-04-29  Richard Biener  <rguenther@suse.de>
11243         PR middle-end/57103
11244         * tree-cfg.c (move_stmt_op): Fix condition under which to update
11245         TREE_BLOCK.
11246         (move_stmt_r): Remove redundant checking.
11248 2013-04-29  Teresa Johnson  <tejohnson@google.com>
11250         PR bootstrap/57077
11251         * basic-block.h (apply_scale): New function.
11252         (apply_probability): Use apply_scale.
11253         * gimple-streamer-in.c (input_bb): Ditto.
11254         * lto-streamer-in.c (input_cfg): Ditto.
11255         * lto-cgraph.c (merge_profile_summaries): Ditto.
11256         * tree-optimize.c (execute_fixup_cfg): Ditto.
11257         * tree-inline.c (copy_bb): Update comment to use apply_scale.
11258         (copy_edges_for_bb): Ditto.
11259         (copy_cfg_body): Ditto.
11261 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11263         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11264         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11265         (tail_merge_optimize): Handle current_loops == NULL.
11267 2013-04-26  Jeff Law  <law@redhat.com>
11269         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11270         (simplify_cond_using_ranges): Generalize code to simplify
11271         COND_EXPRs where one argument is a constant and the other
11272         is an SSA_NAME created by an integral type conversion.
11274 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11276         * config/arm/arm.md (store_minmaxsi): Use only when
11277         optimize_insn_for_size_p.
11279 2013-04-29  Christian Bruel  <christian.bruel@st.com>
11281         PR target/57108
11282         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11284 2013-04-29  Richard Biener  <rguenther@suse.de>
11286         PR middle-end/57089
11287         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11288         loop tree make sure to schedule a fixup for the child as well.
11289         (expand_omp_for_generic): Properly add loops.
11290         (expand_omp_for_static_nochunk): Likewise.
11291         (expand_omp_for_static_chunk): Likewise.
11292         (expand_omp_for): For the degenerate case fixup loops.
11293         (expand_omp_sections): Fix default bb placement in loops.
11294         (expand_omp_atomic_pipeline): Properly add loops.
11296 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11298         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11300 2013-04-29  Tom de Vries  <tom@codesourcery.com>
11302         * tree-ssa-tail-merge.c: Update header comment.
11304 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11306         * config/aarch64/arm_neon.h
11307         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11308         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11309         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11310         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11311         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11312         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11313         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11314         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11316 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11318         * config/aarch64/aarch64-simd.md
11319         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11320         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11321         fix_trunc, fixuns_trunc.
11322         (ftrunc<VDQF:mode>2): New.
11323         * config/aarch64/iterators.md (optab): Add fix, fixuns.
11324         (fix_trunc_optab): New.
11326 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11328         * config/aarch64/aarch64-builtins.c
11329         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11330         iceilf, lround, iroundf.
11332 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11334         PR target/54349
11335         * config/i386/i386.h (enum ix86_tune_indices)
11336         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11337         New, split from X86_TUNE_INTER_UNIT_MOVES.
11338         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11339         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11340         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11341         (TARGET_INTER_UNIT_MOVES): Remove.
11342         * config/i386/i386.c (initial_ix86_tune_features): Update.
11343         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11344         (ix86_expand_convert_uns_didf_sse): Use
11345         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11346         (ix86_expand_vector_init_one_nonzero): Ditto.
11347         (ix86_expand_vector_init_interleave): Ditto.
11348         (inline_secondary_memory_needed): Return true for moves from SSE class
11349         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11350         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11351         * config/i386/constraints.md (Yi, Ym): Depend on
11352         TARGET_INTER_UNIT_MOVES_TO_VEC.
11353         (Yj, Yn): New constraints.
11354         * config/i386/i386.md (*movdi_internal): Change constraints of
11355         operand 1 from Yi to Yj and from Ym to Yn.
11356         (*movsi_internal): Ditto.
11357         (*movdf_internal): Ditto.
11358         (*movsf_internal): Ditto.
11359         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11360         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11361         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11362         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11363         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11364         * config/i386/sse.md (movdi_to_sse): Ditto.
11365         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11366         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11367         TARGET_INTER_UNIT_MOVES.
11368         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11369         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11370         instead of TARGET_INTER_UNIT_MOVES.
11371         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11372         operand 1 from Yi to Yj and from Ym to Yn.
11374 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11376         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11377         (float_truncate_hi_): Likewise.
11378         (float_extend_lo_): Likewise.
11379         (float_truncate_lo_): Likewise.
11380         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11381         (aarch64_float_extend_lo_v2df): Likewise.
11382         (vec_unpacks_hi_v4sf): Likewise.
11383         (aarch64_float_truncate_lo_v2sf): Likewise.
11384         (aarch64_float_truncate_hi_v4sf): Likewise.
11385         (vec_pack_trunc_v2df): Likewise.
11386         (vec_pack_trunc_df): Likewise.
11388 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11390         * config/aarch64/aarch64-builtins.c
11391         (aarch64_fold_builtin): Fold float conversions.
11392         * config/aarch64/aarch64-simd-builtins.def
11393         (floatv2si, floatv4si, floatv2di): New.
11394         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11395         * config/aarch64/aarch64-simd.md
11396         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11397         * config/aarch64/iterators.md (FLOATUORS): New.
11398         (optab): Add float, floatuns.
11399         (su_optab): Likewise.
11401 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11403         * config/aarch64/aarch64-builtins.c
11404         (aarch64_builtin_vectorized_function): Use new names for
11405         fcvt builtins.
11406         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11407         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11408         (fcvtzu): Split as...
11409         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11410         (fcvtas): Split as...
11411         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11412         (fcvtau): Split as...
11413         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11414         (fcvtps): Split as...
11415         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11416         (fcvtpu): Split as...
11417         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11418         (fcvtms): Split as...
11419         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11420         (fcvtmu): Split as...
11421         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11422         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11423         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11424         (lfrintnusf, lfrintnudf): Likewise.
11425         * config/aarch64/aarch64-simd.md
11426         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11427         define_insn.
11428         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11429         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11430         (fcvt_pattern): Likewise.
11432 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11434         * config/aarch64/aarch64-simd.md
11435         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11436         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11438 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11440         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11441         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11442         (vrnd<a,m,n,p>_f32): Implement using builtins.
11443         (vrnd<i,x><q>_f<32, 64>): New.
11445 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
11447         * config/aarch64/aarch64-builtins.c
11448         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11449         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11450         (frintz): Rename to...
11451         (btrunc): ...this.
11452         (frintp): Rename to...
11453         (ceil): ...this.
11454         (frintm): Rename to...
11455         (floor): ...this.
11456         (frinti): Rename to...
11457         (nearbyint): ...this.
11458         (frintx): Rename to...
11459         (rint): ...this.
11460         (frinta): Rename to...
11461         (round): ...this.
11462         * config/aarch64/aarch64-simd.md
11463         (aarch64_frint<frint_suffix><mode>): Delete.
11464         (<frint_pattern><mode>2): Convert to insn.
11465         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11466         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11467         (frint_pattern): Likewise.
11468         (frint_suffix): Likewise.
11470 2013-04-29  Richard Biener  <rguenther@suse.de>
11472         PR tree-optimization/57081
11473         * loop-init.c: Include tree-flow.h.
11474         (loop_optimizer_finalize): Free number of iteration estimates.
11475         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11477 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
11479         PR tree-optimization/57083
11480         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11481         non-singleton shift count range, zero extend low_bound for uns case.
11483         * config/i386/predicates.md (general_vector_operand): New predicate.
11484         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11485         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11486         if they aren't nonimmediate operands.  If their original values
11487         satisfy const_vector_equal_evenodd_p, don't shift them.
11488         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11489         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
11490         if not nonimmediate_operand.
11491         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11492         instead of register_operand.
11493         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11495 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
11497         * stor-layout.c (finalize_size_functions): Allocate a structure and
11498         reset cfun before dumping the functions.
11500 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
11502         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11504         PR target/56866
11505         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11506         use xop_pmacsdqh if uns_p.
11507         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11508         the immediate rotate count.
11510 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
11512         * rtl.h (struct rtx_def): Add comment for field jump.
11513         (LRA_SUBREG_P): New macro.
11514         * recog.c (register_operand): Check LRA_SUBREG_P.
11515         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11516         stack frame.
11517         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11518         (lra_final_code_change): Skip subreg change for operators.
11519         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11520         if there are no operand changes.
11521         * lra-constraints.c (curr_insn_set): New.
11522         (match_reload): Set LRA_SUBREG_P.
11523         (emit_spill_move): Ditto.
11524         (check_and_process_move): Use curr_insn_set. Process only single
11525         set insns.  Don't initialize sec_mem_p and change_p.
11526         (simplify_operand_subreg): Use LRA_SUBREG_P.
11527         (reg_in_class_p): New function.
11528         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11529         of #ifdef.  Add code to remove cycling.
11530         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11531         non-null disp.  Reload inner instead of disp when base and index
11532         are null.  Try to put lo_sum into register.
11533         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11534         (check_and_process_move): Move code for move cost check to
11535         simple_move_p.  Remove equiv_substitution.
11536         (simple_move_p): New function.
11537         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11538         curr_insn_set.  Call check_and_process_move only for single set
11539         insns.  Use the new function.  Move call of check_and_process_move
11540         after operand equiv substitution and address process.
11542 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11544         PR go/57045
11545         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11546         with nonlocal goto receivers or returns twice calls, ignore
11547         unininitialized values from abnormal edges to nl goto receiver
11548         or returns twice call.
11550 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11552         PR tree-optimization/57051
11553         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11554         and VEC_RSHIFT_EXPR if shift count is a multiple of element
11555         bitsize.
11557 2013-04-26  Richard Biener  <rguenther@suse.de>
11559         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11560         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
11561         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11562         (fixup_loop_arrays_after_move): New function.
11563         (move_sese_region_to_fn): Properly outline the loop tree parts
11564         of the SESE region.
11566 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
11568         * config/i386/i386.md (type, unit): Fix long lines.
11570 2013-04-26  Richard Biener  <rguenther@suse.de>
11572         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11573         (lto-streamer-out.o): Likewise.
11574         * cfgloop.c (init_loops_structure): Export, add struct function
11575         argument and adjust.
11576         (flow_loops_find): Adjust.
11577         * cfgloop.h (enum loop_estimation): Add EST_LAST.
11578         (init_loops_structure): Declare.
11579         * lto-streamer-in.c: Include cfgloop.h.
11580         (input_cfg): Input the loop tree.
11581         * lto-streamer-out.c: Include cfgloop.h.
11582         (output_cfg): Output the loop tree.
11583         (output_struct_function_base): Do not drop PROP_loops.
11585 2013-03-26  Richard Biener  <rguenther@suse.de>
11587         * tree-cfg.c (execute_build_cfg): Build the loop tree.
11588         (pass_build_cfg): Provide PROP_loops.
11589         (move_sese_region_to_fn): Remove loops that are outlined into fn
11590         for now.
11591         * tree-inline.c: Include cfgloop.h.
11592         (initialize_cfun): Do not drop PROP_loops.
11593         (copy_loops): New function.
11594         (copy_cfg_body): Copy loop structure.
11595         (tree_function_versioning): Initialize destination loop tree.
11596         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11597         (pass_parallelize_loops): Do IL verification.
11598         * loop-init.c (loop_optimizer_init): Fixup loops if required.
11599         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11600         the CFG make sure we fixup loops as well.
11601         * tree-ssa-tail-merge.c: Include cfgloop.h.
11602         (replace_block_by): When merging loop latches mark loops for fixup.
11603         * lto-streamer-out.c (output_struct_function_base): Drop
11604         PROP_loops for now.
11605         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11606         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11607         * ipa-split.c: Include cfgloop.h.
11608         (split_function): Add the new return block to the loop tree root.
11609         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11610         whether we have removed the forwarder block.
11611         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11612         * cfgloop.h (place_new_loop): Declare.
11613         * cfgloopmanip.c (place_new_loop): Export.
11614         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11615         (tree-switch-conversion.o): Likewise.
11616         (tree-complex.o): Likewise.
11617         (tree-inline.o): Likewise.
11618         (tree-ssa-tailmerge.o): Likewise.
11619         (ipa-split.o): Likewise.
11620         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11621         (tree-ssa-copy.o): Likewise.
11622         * tree-switch-conversion.c: Include cfgloop.h
11623         (process_switch): If we emit a bit-test cascade, schedule loops
11624         for fixup.
11625         * tree-complex.c: Include cfgloop.h.
11626         (expand_complex_div_wide): Properly add new basic-blocks to loops.
11627         * asan.c: Include cfgloop.h.
11628         (create_cond_insert_point): Properly add new basic-blocks to
11629         loops, schedule loop fixup.
11630         * cfgloop.c (verify_loop_structure): Check that looks are not
11631         marked for fixup.
11632         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11633         to loops.
11634         (expand_omp_for_generic): Likewise.
11635         (expand_omp_sections): Likewise.
11636         (expand_omp_atomic_pipeline): Schedule loops for fixup.
11637         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11638         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11639         is initialized, not when loops are present.
11640         * tree-parloops.c (parallelize_loops): Remove checking here.
11641         * passes.c (init_optimization_passes): Schedule a copy-propagation
11642         pass before complete unrolling of inner loops.
11644 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11646         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11647         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11648         (colorize_init): Add argument to _WIN32 version.
11649         * toplev.c: Include diagnostic-color.h.
11650         (process_options): Default to -fdiagnostics-color=auto if
11651         GCC_COLORS env var is in the environment.
11652         * common.opt (fdiagnostics-color=): Add Var and Init.
11653         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11654         env var is in the environment, the default is auto rather than never.
11656         * diagnostic.h (file_name_as_prefix): Add context argument.
11657         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
11658         the string as locus.
11659         * langhooks.c (lhd_print_error_function): Adjust caller.
11661 2013-04-25  Lawrence Crowl  <crowl@google.com>
11663         * var-tracking.c (shared_hash_def::htab):
11664         Change type to hash_table.  Update dependent calls and types.
11666 2013-04-25  Lawrence Crowl  <crowl@google.com>
11668         * Makefile.in: Update as needed below.
11670         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11671         Move declaration to after the type's method definitons.
11673         * attribs.c (htab_t scoped_attributes::attribute_hash):
11674         Change type to hash_table.  Update dependent calls and types.
11676         * bitmap.c (htab_t bitmap_desc_hash):
11677         Change type to hash_table.  Update dependent calls and types.
11679         * cselib.c (htab_t cselib_hash_table):
11680         Change type to hash_table.  Update dependent calls and types.
11682         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11683         (hash_string_slot_node): Move implementation into lto-streamer.h
11684         struct string_slot_hasher.
11685         (eq_string_slot_node): Likewise.
11687         * data-streamer-out.c: Update output_block::string_hash_table
11688         dependent calls and types.
11690         * dwarf2cfi.c (htab_t trace_index):
11691         Change type to hash_table.  Update dependent calls and types.
11693         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11694         Change type to hash_table.  Update dependent calls and types.
11695         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11696         (htab_t optimize_external_refs::map): Likewise.
11697         (htab_t output_comp_unit::extern_map): Likewise.
11698         (htab_t output_comdat_type_unit::extern_map): Likewise.
11699         (htab_t output_macinfo::macinfo_htab): Likewise.
11700         (htab_t optimize_location_lists::htab): Likewise.
11701         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11703         * except.c (htab_t ehspec_hash_type):
11704         Change type to hash_table.  Update dependent calls and types.
11705         (assign_filter_values::ttypes): Likewise.
11706         (assign_filter_values::ehspec): Likewise.
11707         (sjlj_assign_call_site_values::ar_hash): Likewise.
11708         (convert_to_eh_region_ranges::ar_hash): Likewise.
11710         * gcse.c (htab_t pre_ldst_table):
11711         Change type to hash_table.  Update dependent calls and types.
11713         * ggc-common.c (htab_t saving_htab):
11714         Change type to hash_table.  Update dependent calls and types.
11715         (htab_t loc_hash): Likewise.
11716         (htab_t ptr_hash): Likewise.
11717         (call_count): Rename ggc_call_count.
11718         (call_alloc): Rename ggc_call_alloc.
11719         (loc_descriptor): Rename make_loc_descriptor.
11720         (add_statistics): Rename ggc_add_statistics.
11722         * ggc-common.c (saving_htab):
11723         Change type to hash_table.  Update dependent calls and types.
11725         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11726         (push_gimplify_context): Likewise.
11727         (pop_gimplify_context): Likewise.
11728         (struct gimple_temp_hash_elt): Added.
11729         (struct gimplify_hasher): Likewise.
11730         (struct gimplify_ctx.temp_htab):
11731         Change type to hash_table.  Update dependent calls and types.
11733         * gimple-fold.c: Include gimplify-ctx.h.
11735         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11736         Change type to hash_table.  Update dependent calls and types.
11737         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11738         avoid potential global name collision.
11740         * gimplify.c: Include gimplify-ctx.h.
11741         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11742         (htab_t gimplify_ctx::temp_htab):
11743         Update dependent calls and types for new type hash_table.
11744         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11745         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11747         * gimplify-ctx.h: New.
11748         (struct gimple_temp_hash_elt): Move from gimplify.c.
11749         (class gimplify_hasher): New.
11750         (struct gimplify_ctx): Move from gimple.h.
11751         (htab_t gimplify_ctx::temp_htab):
11752         Change type to hash_table.  Update dependent calls and types.
11754         * graphite-clast-to-gimple.c: Include graphite-htab.h.
11755         (htab_t ivs_params::newivs_index):
11756         Change type to hash_table.  Update dependent calls and types.
11757         (htab_t ivs_params::params_index): Likewise.
11758         (htab_t print_generated_program::params_index): Likewise.
11759         (htab_t gloog::newivs_index): Likewise.
11760         (htab_t gloog::params_index): Likewise.
11762         * graphite.c: Include graphite-htab.h.
11763         4htab_t graphite_transform_loops::bb_pbb_mapping):
11764         Change type to hash_table.  Update dependent calls and types.
11766         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11767         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11768         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11770         * graphite-dependences.c: Include graphite-htab.h.
11771         (loop_is_parallel_p): Change hash table type of parameter.
11773         * graphite-htab.h: New.
11774         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11775         (extern find_pbb_via_hash): Move from graphite-poly.h.
11776         (extern loop_is_parallel_p): Move from graphite-poly.h.
11777         (extern get_loop_body_pbbs): Move from graphite-poly.h.
11779         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11780         (extern loop_is_parallel_p): Move to graphite-htab.h.
11781         (extern get_loop_body_pbbs): Move to graphite-htab.h.
11783         * haifa-sched.c (htab_t delay_htab):
11784         Change type to hash_table.  Update dependent calls and types.
11785         (htab_t delay_htab_i2): Likewise.
11787         * ira-color.c (htab_t allocno_hard_regs_htab):
11788         Change type to hash_table.  Update dependent calls and types.
11790         * ira-costs.c (htab_t cost_classes_htab):
11791         Change type to hash_table.  Update dependent calls and types.
11793         * loop-invariant.c (htab_t merge_identical_invariants::eq):
11794         Change type to hash_table.  Update dependent calls and types.
11796         * loop-iv.c (htab_t bivs):
11797         Change type to hash_table.  Update dependent calls and types.
11799         * loop-unroll.c (htab_t opt_info::insns_to_split):
11800         Change type to hash_table.  Update dependent calls and types.
11801         (htab_t opt_info::insns_with_var_to_expand): Likewise.
11803         * lto-streamer.h (struct string_slot): Move from data-streamer.h
11804         (struct string_slot_hasher): New.
11805         (htab_t output_block::string_hash_table):
11806         Change type to hash_table.  Update dependent calls and types.
11808         * lto-streamer-in.c (freeing_string_slot_hasher): New.
11809         (htab_t file_name_hash_table):
11810         Change type to hash_table.  Update dependent calls and types.
11812         * lto-streamer-out.c: Update output_block::string_hash_table dependent
11813         calls and types.
11815         * lto-streamer.c (htab_t tree_htab):
11816         Change type to hash_table.  Update dependent calls and types.
11818         * omp-low.c: Include gimplify-ctx.h.
11820         * passes.c (htab_t name_to_pass_map):
11821         Change type to hash_table.  Update dependent calls and types.
11822         (pass_traverse): Rename to passes_pass_traverse.
11824         * plugin.c (htab_t event_tab):
11825         Change type to hash_table.  Update dependent calls and types.
11827         * postreload-gcse.c (htab_t expr_table):
11828         Change type to hash_table.  Update dependent calls and types.
11829         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11831         * sese.c (debug_rename_map_1): Make extern.
11832         (htab_t copy_bb_and_scalar_dependences::rename_map):
11833         Change type to hash_table.  Update dependent calls and types.
11835         * sese.h (extern debug_rename_map): Move to .c file.
11837         * store-motion.c (htab_t store_motion_mems_table):
11838         Change type to hash_table.  Update dependent calls and types.
11840         * trans-mem.c (htab_t tm_new_mem_hash):
11841         Change type to hash_table.  Update dependent calls and types.
11843         * tree-browser.c (htab_t TB_up_ht):
11844         Change type to hash_table.  Update dependent calls and types.
11846         * tree-cfg.c (htab_t discriminator_per_locus):
11847         Change type to hash_table.  Update dependent calls and types.
11849         * tree-complex.c: Include tree-hasher.h
11850         (htab_t complex_variable_components):
11851         Change type to hash_table.  Update dependent calls and types.
11853         * tree-eh.c (htab_t finally_tree):
11854         Change type to hash_table.  Update dependent calls and types.
11856         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11857         struct int_tree_hasher.
11858         (extern int_tree_map_eq): Likewise.
11859         (uid_decl_map_hash): Removed.
11860         (extern decl_tree_map_eq): Likewise.
11862         * tree-hasher.h: New.
11863         (struct int_tree_hasher): New.
11864         (typedef int_tree_htab_type): New.
11866         * tree-inline.c: Include gimplify-ctx.h.
11868         * tree-mudflap.c: Include gimplify-ctx.h.
11870         * tree-parloops.c: Include tree-hasher.h.
11871         (htab_t eliminate_local_variables_stmt::decl_address):
11872         Change type to hash_table.  Update dependent calls and types.
11873         (htab_t separate_decls_in_region::decl_copies): Likewise.
11875         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11876         Change type to hash_table.  Update dependent calls and types.
11878         * tree-sra.c (candidates):
11879         Change type to hash_table.  Update dependent calls and types.
11881         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11882         in tree-flow.h.
11883         (int_tree_map_hash): Likewise.
11885         * tree-ssa-dom.c (htab_t avail_exprs):
11886         Change type to hash_table.  Update dependent calls and types.
11888         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11889         Change type to hash_table.  Update dependent calls and types.
11891         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11892         Change type to hash_table.  Update dependent calls and types.
11894         * tree-ssa-phiopt.c (seen_ssa_names):
11895         Change type to hash_table.  Update dependent calls and types.
11897         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11898         Change type to hash_table.  Update dependent calls and types.
11900         * tree-ssa-uncprop.c (equiv):
11901         Change type to hash_table.  Update dependent calls and types.
11903 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11905         PR rtl-optimization/57003
11906         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11907         call note_stores with kill_clobbered_value callback again after
11908         killing regs_invalidated_by_call.
11910 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11912         * config/aarch64/aarch64-simd.md
11913         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11914         (aarch64_simd_bsl<mode>): Likewise.
11915         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11917 2013-04-25  Marek Polacek  <polacek@redhat.com>
11919         PR tree-optimization/57066
11920         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11922 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11924         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11926 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11928         * config/aarch64/aarch64-builtins.c
11929         (aarch64_fold_builtin): New.
11930         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11931         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11932         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11933         * config/aarch64/arm_neon.h
11934         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11936 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11937             Tejas Belagod  <tejas.belagod@arm.com>
11939         * config/aarch64/aarch64-builtins.c
11940         (aarch64_gimple_fold_builtin): New.
11941         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11942         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11943         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11944         (addvv4sf): Update.
11945         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11947 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11949         * config/aarch64/aarch64.md
11950         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11952 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11954         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11955         (*ngcsi_uxtw): New pattern.
11957 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11958             Julian Brown  <julian@codesourcery.com>
11960         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11961         (TB_DREG): Add T_V4HF.
11962         (v4hf_UP): New macro.
11963         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11964         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11965         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11966         built-ins.
11967         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11968         (arm_vector_mode_supported_p): Handle V4HF.
11969         (arm_mangle_map): Handle V4HFmode.
11970         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11971         * config/arm/arm_neon_builtins.def: Add entries for
11972         vcvtv4hfv4sf, vcvtv4sfv4hf.
11973         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11974         (neon_vcvtv4hfv4sf): Likewise.
11975         * config/arm/neon-gen.ml: Handle half-precision floating point
11976         features.
11977         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11978         * config/arm/arm_neon.h: Regenerate.
11979         * config/arm/neon.ml (type elts): Add F16.
11980         (type vectype): Add T_float16x4, T_floatHF.
11981         (type vecmode): Add V4HF.
11982         (type features): Add Requires_FP_bit feature.
11983         (elt_width): Handle F16.
11984         (elt_class): Likewise.
11985         (elt_of_class_width): Likewise.
11986         (mode_of_elt): Refactor.
11987         (type_for_elt): Handle F16, fix error messages.
11988         (vectype_size): Handle T_float16x4.
11989         (vcvt_sh): New function.
11990         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11991         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11992         (string_of_mode): Handle V4HF.
11993         * doc/arm-neon-intrinsics.texi: Regenerate.
11995 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11997         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11998         format specifier in 'X' case.
12000 2013-04-25  Alan Modra  <amodra@gmail.com>
12002         PR target/57052
12003         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
12004         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
12005         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
12006         Repeat for many other rotate/shift and mask patterns using subregs.
12007         Name lshiftrt insns.
12008         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
12009         on WORDS_BIG_ENDIAN.
12011 2013-04-25  Alan Modra  <amodra@gmail.com>
12013         * config.gcc: Support little-endian powerpc-linux targets.
12014         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
12015         (LINK_OS_LINUX_SPEC): Define.
12016         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
12017         Preserve MASK_LITTLE_ENDIAN.
12018         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
12019         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
12020         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
12021         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
12022         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
12023         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
12024         Correct fp word order for little-endian.  Don't shift toc entries
12025         smaller than a word for little-endian.
12026         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
12027         (bswapdi2 splits): Correct low-part subreg for little-endian.
12028         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
12029         low/high where such is correct only for be.
12030         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
12031         little-endian for -mcall-aixdesc.
12033 2013-04-25  Alan Modra  <amodra@gmail.com>
12035         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
12036         replace_equiv_address_nv.
12038 2013-04-25  Alan Modra  <amodra@gmail.com>
12040         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
12042 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12044         Revert:
12045         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12046         * rtl.h (struct rtx_def): ...
12048 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12050         PR rtl-optimizations/57046
12051         * lra-constraints (split_reg): Set up lra_risky_transformations_p
12052         for multi-reg splits.
12054 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
12056         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
12058 2013-04-24  Sterling Augustine  <saugustine@google.com>
12060         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
12061         (comp_dir_string, debug_str_dwo_section): New.
12062         (DEBUG_STR_DWO_SECTION): Rename to ...
12063         (DEBUG_DWO_STR_SECTION): ... this.
12064         (DEBUG_NORM_STR_SECTION): Delete.
12065         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
12066         (DEBUG_STR_DWO_SECTION_FLAGS): New.
12067         (find_AT_string): Move most logic to ...
12068         (find_AT_string_in_table): ... here.  New.
12069         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
12070         add_skeleton_AT_string.  Delete logic.
12071         (output_skeleton_debug_sections): Remove call to
12072         add_top_level_skeleton_die_attrs.
12073         (add_comp_dir_attribute): Move logic to comp_dir_string.
12074         (dwarf2out_init): Initialize debug_str_dwo_section.
12075         (output_indirect_string): Call find_string_form.
12076         (output_indirect_strings): Rewrite.
12077         (prune_unused_types): Empty skeleton_debug_str_hash.
12078         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
12079         (dwarf2out_finish):  Call output_indirect_strings.
12081 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12083         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
12085 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12087         * rtl.h (struct rtx_def): Add comment for field jump.
12088         (LRA_SUBREG_P): New macro.
12089         * recog.c (register_operand): Check LRA_SUBREG_P.
12090         * lra.c (lra): Add note at the end of RTL code. Align non-empty
12091         stack frame.
12092         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12093         (lra_final_code_change): Skip subreg change for operators.
12094         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12095         if there are no operand changes.
12096         * lra-constraints.c (curr_insn_set): New.
12097         (match_reload): Set LRA_SUBREG_P.
12098         (emit_spill_move): Ditto.
12099         (check_and_process_move): Use curr_insn_set. Process only single
12100         set insns.  Don't initialize sec_mem_p and change_p.
12101         (simplify_operand_subreg): Use LRA_SUBREG_P.
12102         (reg_in_class_p): New function.
12103         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
12104         of #ifdef.  Add code to remove cycling.
12105         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12106         non-null disp.  Reload inner instead of disp when base and index
12107         are null.  Try to put lo_sum into register.
12108         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12109         (check_and_process_move): Move code for move cost check to
12110         simple_move_p.  Remove equiv_substitution.
12111         (simple_move_p): New function.
12112         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
12113         curr_insn_set.  Call check_and_process_move only for single set
12114         insns.  Use the new function.  Move call of check_and_process_move
12115         after operand equiv substitution and address process.
12117 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
12119         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12120         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12121         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12123 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12125         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12127 2013-04-24  Marek Polacek  <polacek@redhat.com>
12129         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12130         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12131         (select_loops_exit_conditions): Likewise.
12132         (number_of_iterations_for_all_loops): Likewise.
12133         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12134         (scev_analysis): Likewise.
12136 2013-04-02  Catherine Moore  <clm@codesourcery.com>
12137             Chao-ying Fu  <fu@mips.com>
12139         * config/mips/micromips.md (jraddiusp): New pattern.
12140         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12141         instruction if possible.
12143 2013-04-24  Alan Modra  <amodra@gmail.com>
12145         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12147 2013-04-24  Julian Brown  <julian@codesourcery.com>
12148             Chung-Lin Tang  <cltang@codesourcery.com>
12150         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12151         dependency behavior in enumeration type DIE generation. Add TODO note
12152         to comments about future DW_FORM_sdata/udata re-work of related code.
12154 2013-04-23  Lawrence Crowl  <crowl@google.com>
12156         * Makefile.in: Update as needed below.
12158         * hash-table.h (class hash_table):
12159         Correct many methods with parameter types compare_type to the correct
12160         value_type.  (Correct code was unlikely to notice the change.)
12161         (hash_table::elements_with_deleted) New.
12162         (class hashtable::iterator): New.
12163         (hashtable::begin()): New.
12164         (hashtable::end()): New.
12165         (FOR_EACH_HASH_TABLE_ELEMENT): New.
12167         * statistics.c (statistics_hashes):
12168         Change type to hash_table.  Update dependent calls and types.
12170         * tree-into-ssa.c (var_infos):
12171         Change type to hash_table.  Update dependent calls and types.
12173         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12174         Change type to hash_table.  Update dependent calls and types.
12176         * tree-ssa-loop-im.c (struct mem_ref.refs):
12177         Change type to hash_table.  Update dependent calls and types.
12179         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12180         Change type to hash_table.  Update dependent calls and types.
12182         * tree-ssa-sccvn.c (vn_tables_s::nary):
12183         Change type to hash_table.  Update dependent calls and types.
12184         (vn_tables_s::phis): Likewise.
12185         (vn_tables_s::references): Likewise.
12187         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12188         (vn_reference_eq): Update parameter and return types.
12190         * tree-ssa-structalias.c (pointer_equiv_class_table):
12191         Change type to hash_table.  Update dependent calls and types.
12192         (location_equiv_class_table): Likewise.
12194         * tree-vect-data-refs.c: Consequential changes for making
12195         peeling a hash_table.
12197         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12198         (destroy_loop_vec_info): Dependent hash_table update.
12200         * tree-vectorizer.h (peeling_htab):
12201         Change type to hash_table.  Update dependent calls and types.
12203 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
12205         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12206         to check the register content is equal or not.
12207         * lra-constraints.c (match_reload): Use lra_assign_reg_val
12208         to assign register content record.
12209         * lra-eliminations.c (update_reg_eliminate): Use
12210         lra_update_reg_val_offset to update register content offset.
12211         * lra-int.h (struct lra_reg): Add offset member.
12212         (lra_reg_val_equal_p): New static inline function.
12213         (lra_update_reg_val_offset): New static inline function.
12214         (lra_assign_reg_val): New static inline function.
12215         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12216         to assign register content record.
12217         (initialize_lra_reg_info_element): Initial offset to zero.
12219 2013-04-23  Catherine Moore  <clm@codesourcery.com>
12221         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12222         operands.  Record compression.
12224 2013-04-23  Xinliang David Li  <davidxl@google.com>
12226         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12228 2013-04-23  Richard Biener  <rguenther@suse.de>
12230         PR middle-end/57036
12231         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12232         parameter, only add abnormal goto edges from the copied body
12233         if the call could perform abnormal gotos.
12234         (copy_cfg_body): Adjust.
12236 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
12238         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12240 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
12242         * coretypes.h (gimple_stmt_iterator): Add struct to make
12243         compatible with C.
12245 2013-04-23  Richard Biener  <rguenther@suse.de>
12247         PR tree-optimization/57026
12248         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12249         from SSA names occuring in abnormal PHI nodes.
12251 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12253         * lto/lto.c (print_lto_report_1): Fix LTO report names.
12255 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12257         * lto/lto.c (print_lto_report_1): Declare early.
12258         (read_cgraph_and_symbols): Call print_lto_report_1 early.
12260 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12262         * common.opt (-flto-report-wpa): Add.
12263         * doc/invoke.texi (-flto-report-wpa): Add.
12264         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12265         (lto_main): dito.
12267 2013-04-22  Xinliang David Li  <davidxl@google.com>
12269         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12270         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12271         * Makefile.in: New dependency
12273             David Daney <ddaney.cavm@gmail.com>
12275         * configure.ac (gcc_cv_as_micromips_support):  Use the
12276         --fatal-warnings option.
12277         * configure: Regenerate.
12279 2013-04-22  Marek Polacek  <polacek@redhat.com>
12281         PR sanitizer/56990
12282         * tsan.c (instrument_expr): Don't instrument expression
12283         in case its size is zero.
12285 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
12287         PR target/57032
12288         Revert:
12289         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
12291         * config/alpha/alpha.c (TARGET_LRA_P): New define.
12293 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12295         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12296         (gimple_stmt_iterator): New typedef.
12297         * gimple.h (gimple_stmt_iterator): Rename to...
12298         (gimple_stmt_iterator_d): ... This.
12299         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12300         trees be valid for GIMPLE and GENERIC.
12301         (TARGET_GIMPLE_FOLD_BUILTIN): New.
12302         * gimple-fold.c (gimple_fold_call): Call target hook
12303         gimple_fold_builtin.
12304         * hooks.c (hook_bool_gsiptr_false): New.
12305         * hooks.h (hook_bool_gsiptr_false): New.
12306         * target.def (fold_stmt): New.
12307         * doc/tm.texi: Regenerate.
12309 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
12311         PR target/57018
12312         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12313         a set sp if no stack realignment.
12315 2013-04-22  Nick Clifton  <nickc@redhat.com>
12317         * config.gcc (tilegx-linux): Extend extra_objs rather than
12318         overwriting it.
12319         (tilepro-linux): Likewise.
12321 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12323         * config/aarch64/aarch64-builtins.c
12324         (CF): Remove.
12325         (CF0, CF1, CF2, CF3, CF4, CF10): New.
12326         (VAR<1-12>): Add MAP parameter.
12327         (BUILTIN_*): Likewise.
12328         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12329         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12330         (aarch64_ushl_n<mode>): Likewise.
12331         (aarch64_sshr_n<mode>): Likewise.
12332         (aarch64_ushr_n<mode>): Likewise.
12333         (aarch64_<maxmin><mode>): Likewise.
12334         (aarch64_sqrt<mode>): Likewise.
12335         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12336         (vshr<q>_n_*): Likewise.
12338 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
12340         * config/aarch64/aarch64-builtins.c
12341         (aarch64_simd_builtin_type_mode): Handle SF types.
12342         (sf_UP): Define.
12343         (BUILTIN_GPF): Define.
12344         (aarch64_init_simd_builtins): Handle SF types.
12345         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12346         (frecps): Likewise.
12347         (frecpx): Likewise.
12348         * config/aarch64/aarch64-simd.md
12349         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12350         (aarch64_frecpe<mode>): New.
12351         (aarch64_frecps<mode>): Likewise.
12352         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12353         (v8type): Add frecp<esx>.
12354         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12355         (aarch64_frecps<mode>): Likewise.
12356         * config/aarch64/iterators.md (FRECP): New.
12357         (frecp_suffix): Likewise.
12358         * config/aarch64/arm_neon.h
12359         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12361 2013-04-22  Christian Bruel  <christian.bruel@st.com>
12363         PR target/56995
12364         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12365         (REG_CLASS_NAMES): Idem.
12366         (REG_CLASS_CONTENTS): Idem.
12367         (REGCLASS_HAS_FP_REG): Idem.
12368         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12369         (sh_conditional_register_usage): Idem.
12371 2013-04-21  Jeff Law  <law@redhat.com>
12373         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12374         (ssa_forward_propagate_and_combine): Use it.
12376 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12378         * lra.c: Update the flow chart diagram.
12380 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
12382         PR rtl-optimization/56847
12383         * lra-constraints.c (process_alt_operands): Discourage alternative
12384         with non-matche doffsettable memory constraint fro memory with
12385         known offset.
12387 2013-04-19  Richard Biener  <rguenther@suse.de>
12389         PR tree-optimization/56982
12390         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12391         function.
12392         * gimplify.c (gimplify_call_expr): Notice special calls.
12393         (gimplify_modify_expr): Likewise.
12394         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12395         abnormal control flow receivers.
12396         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12397         in the same way as cfun->has_nonlocal_labels.
12398         (gimple_purge_dead_abnormal_call_edges): Likewise.
12399         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12400         receivers start a basic-block.
12402 2013-04-19  Richard Biener  <rguenther@suse.de>
12404         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12405         member ...
12406         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
12407         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12408         (SLP_TREE_LOAD_PERMUTATION): Add.
12409         (vect_transform_slp_perm_load): Adjust prototype.
12410         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12411         (vect_free_slp_instance): Likewise.
12412         (vect_create_new_slp_node): Likewise.
12413         (vect_supported_slp_permutation_p): Remove.
12414         (vect_slp_rearrange_stmts): Adjust.
12415         (vect_supported_load_permutation_p): Likewise.  Inline
12416         vect_supported_slp_permutation_p here.
12417         (vect_analyze_slp_instance): Compute load permutations per
12418         slp node instead of per instance.
12419         (vect_get_slp_defs): Adjust.
12420         (vect_transform_slp_perm_load): Likewise.
12421         (vect_schedule_slp_instance): Remove redundant code.
12422         (vect_schedule_slp): Remove hack for PR56270, add it ...
12423         * tree-vect-stmts.c (vectorizable_load): ... here, do not
12424         CSE loads for SLP.  Adjust.
12426 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12428         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12429         spelling in two comments.
12431 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
12433         PR target/56797
12434         * config/arm/arm.c (load_multiple_sequence): Require SP
12435         as base register for loads if SP is in the register list.
12437 2013-04-19  Martin Jambor  <mjambor@suse.cz>
12439         PR tree-optimization/56718
12440         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12441         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12442         and made public.  Adjusted all callers.
12443         (ipa_intraprocedural_devirtualization): New function.
12444         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12445         (ipa_intraprocedural_devirtualization): Likewise.
12446         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12448 2013-04-19  Richard Biener  <rguenther@suse.de>
12450         PR tree-optimization/57000
12451         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12453 2013-04-19  Terry Guo  <terry.guo@arm.com>
12455         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12456         Replace with ...
12457         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
12458         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12459         (cortex_m4_fmacs): Use new reservations.
12460         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12462 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
12464         PR rtl-optimization/56999
12465         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12466         related code.
12467         (lra_coalesce): Remove split_origin_bitmap and related code.
12468         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
12469         ranges if necessary.
12471 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
12473         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12474         New array.
12475         (ix86_expand_call): Remove clobbered_registers array and use
12476         x86_64_ms_sysv_extra_clobbered_registers instead.
12477         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12478         Declare here.
12479         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12480         predicate.
12481         * config/i386/i386.md (*call_rex64_ms_sysv): Use
12482         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
12483         (*call_value_rex64_ms_sysv): Ditto.
12485 2013-04-18  Cary Coutant  <ccoutant@google.com>
12487         * dwarf2out.c (output_pubnames): Check die_perennial_p of
12488         parent instead of die_mark.
12490 2013-04-18  Diego Novillo  <dnovillo@google.com>
12492         * gimple.c (create_gimple_tmp): New.
12493         (get_expr_type): New.
12494         (build_assign): New.
12495         (build_type_cast): New.
12496         * gimple.h (enum ssa_mode): Define.
12497         (gimple_seq_set_location): New.
12498         * asan.c (build_check_stmt): Change some gimple_build_* calls
12499         to use build_assign and build_type_cast.
12501 2013-04-18  Richard Biener  <rguenther@suse.de>
12503         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12504         handle negative step.  Remove redundant checks.
12505         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12506         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12507         for negative step and grouped loads fail to vectorize.
12509 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12511         * emit-rtl.c (reset_insn_used_flags): New function.
12512         (reset_all_used_flags): Use it.
12513         (verify_insn_sharing): New function.
12514         (verify_rtl_sharing): Fix verification for SEQUENCEs.
12516 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
12518         PR tree-optimization/56984
12519         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12520         and (x >> M) >= N don't register any assertion if N << M is the
12521         minimum value.
12523 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12525         * lower-subreg.c (resolve_simple_move): If called self-recursive,
12526         do not delete_insn insns that have not yet been emitted, only
12527         unlink them with remove_insn.
12528         * df-scan.c (df_insn_delete): Revert r197492.
12530 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
12532         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12533         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12535 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12537         * config/arm/arm.md (movsicc_insn): Convert define_insn into
12538         define_insn_and_split.
12539         (and_scc,ior_scc,negscc): Likewise.
12540         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12542 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12544         * config/arm/arm.c (use_return_insn): Return 0 for targets that
12545         can benefit from using a sequence of LDRD instructions in epilogue
12546         instead of a single LDM instruction.
12548 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12550         PR 45688
12551         * doc/extend.texi: Fix typo.
12553 2013-04-17  Richard Biener  <rguenther@suse.de>
12555         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12556         (vect_build_slp_tree): ... here.
12557         (vect_build_slp_tree_1): Compute which stmts of the SLP group
12558         match.  Remove special-casing of mismatched complex loads.
12559         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12560         re-try the match with swapped commutative operands.
12561         (vect_supported_load_permutation_p): Remove special-casing of
12562         mismatched complex loads.
12563         (vect_analyze_slp_instance): Adjust.
12565 2013-04-17  Richard Biener  <rguenther@suse.de>
12567         PR rtl-optimization/56921
12568         * cfgloop.h (struct loop): Add simple_loop_desc member.
12569         (struct niter_desc): Mark with GTY(()).
12570         (simple_loop_desc): Do not use aux field but simple_loop_desc.
12571         * loop-iv.c (get_simple_loop_desc): Likewise.
12572         (free_simple_loop_desc): Likewise.
12574         Revert
12575         2013-04-16  Richard Biener  <rguenther@suse.de>
12577         PR rtl-optimization/56921
12578         * loop-init.c (pass_rtl_move_loop_invariants): Add
12579         TODO_do_not_ggc_collect to todo_flags_finish.
12580         (pass_rtl_unswitch): Same.
12581         (pass_rtl_unroll_and_peel_loops): Same.
12582         (pass_rtl_doloop): Same.
12584 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
12586         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12587         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12588         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12589         references.
12590         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12591         * tree-streamer.c (record_common_node): Adjust reference in comment.
12593 2013-04-17  Terry Guo  <terry.guo@arm.com>
12595         * config/arm/cortex-m4.md: Add a new bypass.
12597 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12599         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12600         New pattern.
12601         (*subs_<optab><mode>_multp2): New pattern.
12602         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12603         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12605 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12607         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12608         (*subs_mul_imm_<mode>): New pattern.
12610 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
12612         PR target/56948
12613         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12614         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
12615         (vsx_movti_32bit): Change j->wa to O->wa.
12617 2013-04-16  Richard Biener  <rguenther@suse.de>
12619         PR rtl-optimization/56921
12620         * loop-init.c (pass_rtl_move_loop_invariants): Add
12621         TODO_do_not_ggc_collect to todo_flags_finish.
12622         (pass_rtl_unswitch): Same.
12623         (pass_rtl_unroll_and_peel_loops): Same.
12624         (pass_rtl_doloop): Same.
12626 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12628         * config/arm/arm.c (emit_multi_reg_push): New declaration
12629         for an existing function.
12630         (arm_emit_strd_push): New function.
12631         (arm_expand_prologue): Used here.
12632         (arm_emit_ldrd_pop): New function.
12633         (arm_expand_epilogue): Used here.
12634         (arm_get_frame_offsets): Update condition.
12635         (arm_emit_multi_reg_pop): Add a special case for load of a single
12636         register with writeback.
12638 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
12640         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12641         description.
12643 2013-04-16  Richard Biener  <rguenther@suse.de>
12645         PR tree-optimization/56756
12646         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12647         (first_mem_ref_loc): New.
12648         (execute_sm): Place the load temporarily before a previous
12649         access instead of in the latch edge to ensure its SSA dependencies
12650         are defined at points dominating the load.
12652 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
12654         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12655         correct fix by moving header and footer insn to the footer of
12656         the merged basic block.  Clear BB_END of the merged-away block.
12658         PR middle-end/43631
12659         * emit-rtl.c (make_note_raw): New function.
12660         (link_insn_into_chain): New static inline function.
12661         (add_insn): Use it.
12662         (add_insn_before, add_insn_after): Factor insn chain linking code...
12663         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12664         using link_insn_into_chain.
12665         (note_outside_basic_block_p): New helper function for emit_note_after
12666         and emit_note_before.
12667         (emit_note_after): Use nobb variant of add_insn_after if the note
12668         should not be contained in a basic block.
12669         (emit_note_before): Use nobb variant of add_insn_before if the note
12670         should not be contained in a basic block.
12671         (emit_note_copy): Use make_note_raw.
12672         (emit_note): Likewise.
12673         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12674         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12675         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12676         the moved barrier the tail of the basic block it follows.
12677         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12679 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
12681         PR tree-optimization/56962
12682         * gimple-ssa-strength-reduction.c (record_increment): Only set
12683         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12684         either rhs1 or rhs2 is equal to c->base_expr.
12686 2013-04-15  Richard Biener  <rguenther@suse.de>
12688         PR tree-optimization/56933
12689         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12690         member.
12691         (GROUP_READ_WRITE_DEPENDENCE): Remove.
12692         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12693         * tree-vect-data-refs.c (vect_analyze_group_access): Move
12694         dependence check ...
12695         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12696         ... here.
12697         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12698         GROUP_READ_WRITE_DEPENDENCE.
12700 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12702         * emit-rtl.c (reset_all_used_flags): New function.
12703         (verify_rtl_sharing): Call reset_all_used_flags before and after
12704         performing the checks.
12706 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12708         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12709         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12710         * config/arm/constraints.md (De): New constraint.
12711         * config/arm/neon.md (anddi3_neon): Delete.
12712         (neon_vand<mode>): Expand to standard anddi3 pattern.
12713         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12714         Move earlier in the file.
12715         (neon_inv_logic_op2): Likewise.
12716         (arm_anddi_operand_neon): New predicate.
12718 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12720         * configure.ac (gcc_cv_ld_as_needed): Set
12721         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12722         Use -z ignore, -z record on *-*-solaris2*.
12723         (HAVE_LD_AS_NEEDED): Update comment.
12724         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12725         * configure: Regenerate.
12726         * config.in: Regenerate.
12727         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12728         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12729         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12730         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12731         equivalents.  Fix markup.
12732         * doc/tm.texi: Regenerate.
12734 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
12736         * config/i386/i386.opt: New option mstack-protector-guard=.
12737         * config/i386/i386-opts.h: Add enum stack_protector_guard.
12738         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12739         TARGET_SSP_TLS_GUARD.
12740         * config/i386/i386.c (ix86_option_override_internal): Set
12741         ix86_stack_protector_guard.
12742         * config/i386/i386.md (stack_protect_set): Enable for
12743         TARGET_SSP_TLS_GUARD only.
12744         (stack_protect_set_<mode>): Ditto.
12745         (stack_protect_test): Ditto.
12746         (stack_protect_test_<mode>): Ditto.
12747         * doc/invoke.texi (i386 Option): Document.
12749 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
12751         PR target/56890
12752         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12753         (S_MODES): Set H_MODE bit.
12754         (SF_MODES): Set only S_MODE and SF_MODE bits.
12755         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12756         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12757         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12758         <MODE_FLOAT>: Likewise.
12760 2013-04-15  Joey Ye  <joey.ye@arm.com>
12762         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12764 2013-04-15  Joey Ye  <joey.ye@arm.com>
12766         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12767         for real far jump.
12768         (thumb_far_jump_used_p): Count instruction size and set
12769         far_jump_used.
12771 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
12773         * reorg.c (fill_simple_delay_slots): Reindent block of code.
12774         * resource.c (mark_target_live_regs): Reformat conditional block.
12776 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12778         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12779         notes, they are emitted only just before final.
12780         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12782 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12784         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12785         * cfgrtl.c (delete_insn): Call it here instead.
12786         * lra-spills.c (lra_final_code_change): Use delete_insn.
12787         * haifa-sched.c (sched_remove_insn): Likewise.
12788         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12789         returning to the nop pool.
12790         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12791         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
12793 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12795         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12796         * doc/tm.texi: Regenerated.
12798 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
12800         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12801         QImode checks.
12803 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12805         * df-core.c (df_find_def): Compare register numbers.
12806         (df_find_use): Likewise.
12808 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
12810         PR target/56903
12811         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12812         lra_in_progress for return.
12814 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12816         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12817         define_insn into define_insn_and_split and emit movsicc patterns.
12819 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12821         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12823 2013-04-12  Richard Biener  <rguenther@suse.de>
12825         * tree-pass.h (TODO_do_not_ggc_collect): New.
12826         * passes.c (execute_one_ipa_transform_pass): Honor
12827         TODO_do_not_ggc_collect.
12828         (execute_one_pass): Likewise.
12830         Revert
12831         2013-04-10  Richard Biener  <rguenther@suse.de>
12833         * passes.c (init_optimization_passes): Remove reload pass.
12834         * ira.c (do_reload): Merge into ...
12835         (ira): ... this.
12836         (rest_of_handle_reload): Remove.
12837         (pass_reload): Likewise.
12838         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12839         of reload for vzeroupper pass placement.
12841 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12843         PR tree-optimization/56918
12844         PR tree-optimization/56920
12845         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12846         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12847         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12848         use wide_mul_with_sign method.
12850 2013-04-12  Richard Biener  <rguenther@suse.de>
12852         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12853         not be considered a gimple constant.
12855 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12857         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12858         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12859         operations.
12861 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12862             Jakub Jelinek  <jakub@redhat.com>
12864         * opts.c: Include diagnostic-color.h.
12865         (common_handle_option): Handle OPT_fdiagnostics_color_.
12866         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12867         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12868         (diagnostic-color.o): New.
12869         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12870         (diagnostic_color_rule): New enum.
12871         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12872         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12873         the location string.
12874         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12875         either NULL, or color kind.
12876         * diagnostic-color.c: New file.
12877         * diagnostic-color.h: New file.
12878         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12879         arguments.
12880         * doc/invoke.texi (-fdiagnostics-color): Document.
12881         * pretty-print.h (pp_show_color): Define.
12882         (struct pretty_print_info): Add show_color field.
12883         * diagnostic.c: Include diagnostic-color.h.
12884         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12885         macros.  Colorize error:, warning: etc. strings and also the location
12886         string.
12887         (diagnostic_show_locus): Colorize the caret line.
12888         * pretty-print.c: Include diagnostic-color.h.
12889         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12890         inside of %< %> quotes or quoted through q format modifier.
12892 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12894         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12896 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12898         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12899         code in CC_NZ mode.
12900         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12901         pattern.
12903 2013-04-11  Marek Polacek  <polacek@redhat.com>
12905         PR tree-optimization/48184
12906         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12908 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12910         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12911         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12912         (skip_simple_arithmetic): Tidy up.
12913         * tree.h (skip_simple_constant_arithmetic): Declare.
12915 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12917         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12919 2013-04-11  Richard Biener  <rguenther@suse.de>
12921         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12922         generate vector constants.
12924 2013-04-11  Richard Biener  <rguenther@suse.de>
12926         PR tree-optimization/56878
12927         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12928         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12929         New function.
12930         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12931         Prefer to align the DR with the most invariant base address.
12933 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12935         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12936         comment.
12938 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12940         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12941         floating-point vector comparisons against 0.
12943 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12945         PR tree-optimization/56899
12946         * fold-const.c (extract_muldiv_1): Apply distributive law
12947         only if TYPE_OVERFLOW_WRAPS (ctype).
12949 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12951         PR target/56124
12952         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12953         loading has side effect.
12955 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12957         * config/sparc/sparc.c: Include tree-pass.h.
12958         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12959         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12960         head of file.  Change return type.  Split off gate function.
12961         (sparc_gate_work_around_errata): New function.
12962         (pass_work_around_errata): New pass definition.
12963         (insert_pass_work_around_errata) New pass insert definition to
12964         insert pass_work_around_errata just after delayed-branch scheduling.
12965         (sparc_option_override): Insert the pass.
12966         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12968 2013-04-10  David S. Miller  <davem@davemloft.net>
12970         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12971         or -mcpu=hypersparc.
12973         * target.def (cstore_mode): New hook.
12974         * target.h: Include insn-codes.h
12975         * targhooks.c: Likewise.
12976         (default_cstore_mode): New function.
12977         * targhooks.h: Declare it.
12978         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12979         * doc/tm.texi: Rebuild.
12980         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12981         target hook, rather than inspecting the insn_data.
12982         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12983         (TARGET_CSTORE_MODE): Redefine.
12984         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12985         result patterns.
12986         * config/sparc/predicates.md (cstore_result_operand): New special
12987         predicate.
12988         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12989         Use it for operand 0.
12990         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12991         (*snesi_special): Likewise.
12992         (*snesi_zero): Likewise.
12993         (*seqsi_zero): Likewise.
12994         (*sltu_insn): Likewise.
12995         (*sgeu_insn): Likewise.
12996         (*seqdi_special): Make operand 0 and comparison operation be of
12997         DImode.
12998         (*snedi_special): Likewise.
12999         (*snedi_special_vis3): Likewise.
13000         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
13001         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
13002         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
13003         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
13004         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
13005         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
13006         (*sltu_extend_sp64): Likewise.
13007         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
13008         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
13009         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
13010         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
13011         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
13013 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
13015         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
13016         (aarch64_start_file): Use the new function.
13018 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
13019             Jason Merrill  <jason@redhat.com>
13021         * common.opt: Add -gdwarf.
13022         * opts.c (common_handle_option): Handle it.
13023         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
13025 2013-04-10  Richard Biener  <rguenther@suse.de>
13027         * passes.c (execute_todo): Do not call ggc_collect conditional here.
13028         (execute_one_ipa_transform_pass): But unconditionally here.
13029         (execute_one_pass): And here.
13030         (init_optimization_passes): Remove reload pass.
13031         * tree-pass.h (TODO_ggc_collect): Remove.
13032         (pass_reload): Likewise.
13033         * ira.c (do_reload): Merge into ...
13034         (ira): ... this.
13035         (rest_of_handle_reload): Remove.
13036         (pass_reload): Likewise.
13037         * config/i386/i386.c (ix86_option_override): Refer to ira instead
13038         of reload for vzeroupper pass placement.
13039         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
13040         and todo_flags_finish of all passes.
13042 2013-04-10  Richard Biener  <rguenther@suse.de>
13044         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
13045         first_const_oprnd field, rename first_def_type to first_op_type.
13046         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
13047         (vect_get_and_check_slp_defs): Always use the type of the
13048         operand.  Allow mixed vect_external_def, vect_constant_def types.
13049         (vect_get_constant_vectors): Handle mixed vect_external_def,
13050         vect_constant_def types.
13052 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
13054         PR tree-optimization/55524
13055         * tree-ssa-math-opts.c
13056         (convert_mult_to_fma): Don't use an fms construct
13057         when we don't have an fms operation, but fnma, and it looks
13058         likely that we'll be able to use the latter.
13060 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
13062         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
13063         function.
13064         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
13065         inline fail caused by overwritable functions.
13067 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
13069         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
13070         unnecessary bits in the constant power of two case.
13072 2013-04-10  Richard Biener  <rguenther@suse.de>
13074         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
13075         broken code swapping operands.
13076         (vect_build_slp_tree): Do not compute load permutations here.
13077         (vect_analyze_slp_instance): Compute load permutations here,
13078         after building the SLP tree.
13080 2013-04-09  Christian Bruel  <christian.bruel@st.com>
13082         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
13083         of next/prev_real_insn.
13085 2013-04-09  Jan Hubicka  <jh@suse.cz>
13087         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
13088         Drop aliased parameter.
13089         (function_and_variable_visibility): Do not handle alias pairs.
13090         * cgraph.c (varpool_externally_visible_p): Update prototype.
13091         * varpool.c (varpool_add_new_variable): Update.
13093 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13095         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13097 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
13099         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13101         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13103 2013-04-09  Marek Polacek  <polacek@redhat.com>
13105         PR tree-optimization/48762
13106         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13108 2013-04-09  Richard Biener  <rguenther@suse.de>
13110         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13111         dealing with cost.
13112         (vect_build_slp_tree): Likewise.
13113         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13114         calculating the cost of a SLP instance.
13115         (vect_analyze_slp_instance): Use it from here, after building
13116         the SLP tree.
13118 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13120         PR middle-end/56883
13121         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13122         expand_omp_for_static_chunk): Use simple_p = true in
13123         force_gimple_operand_gsi calls when assigning to addressable decls.
13125 2013-04-09  Jeff Law  <law@redhat.com>
13127         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13128         when the boolean was created by converting a wider object which
13129         had a boolean range.
13131 2013-04-09  Richard Biener  <rguenther@suse.de>
13133         * tree-vectorizer.h (slp_void_p): Remove.
13134         (slp_tree): Typedef before _slp_tree declaration.
13135         (struct _slp_tree): Use a vector of slp_tree as children.
13136         (vect_get_place_in_interleaving_chain): Remove.
13137         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13138         Move ...
13139         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13140         and make static.
13141         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13142         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13143         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13144         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13145         Use slp_node instead of slp_void_p and adjust.
13147 2013-04-09  Richard Biener  <rguenther@suse.de>
13149         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13150         work that is not necessary.
13152 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13154         PR tree-optimization/56854
13155         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13156         forward into clobber stmts if it would change MEM_REF lhs into
13157         non-MEM_REF.
13159 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
13161         * tree.c (type_hash_lookup, type_hash_add): Make static.
13162         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13164 2013-04-09  Richard Biener  <rguenther@suse.de>
13166         * tree.h (unsave_expr_now): Remove.
13167         * tree-inline.c (mark_local_for_remap_r): Remove.
13168         (unsave_expr_1): Likewise.
13169         (unsave_r): Likewise.
13170         (unsave_expr_now): Likewise.
13171         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13172         (propagate_tree_value): Likewise.
13174 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
13176         * doc/rtl.texi (sequence): Rewrite documentation to match the
13177         current use of SEQUENCE rtl objects.
13178         * rtl.def (SEQUENCE): Likewise.
13180         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13181         Update documentation.
13182         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13183         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13185         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13187 2013-04-08  Teresa Johnson  <tejohnson@google.com>
13189         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13190         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13191         methods.
13192         (estimate_edge_size_and_time): Add comment to suggest using rounding
13193         methods.
13194         (estimate_node_size_and_time): Ditto.
13195         (remap_edge_change_prob): Use helper rounding divide methods.
13196         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13197         (gimple_mod_pow2_value_transform): Ditto.
13198         (gimple_mod_subtract_transform): Ditto.
13199         (gimple_ic_transform): Ditto.
13200         (gimple_stringops_transform): Ditto.
13201         * stmt.c (conditional_probability): Ditto.
13202         (emit_case_dispatch_table): Ditto.
13203         * lto-cgraph.c (merge_profile_summaries): Ditto.
13204         * tree-optimize.c (execute_fixup_cfg): Ditto.
13205         * cfgcleanup.c (try_forward_edges): Ditto.
13206         * cfgloopmanip.c (scale_loop_profile): Ditto.
13207         (loopify): Ditto.
13208         (duplicate_loop_to_header_edge): Ditto.
13209         (lv_adjust_loop_entry_edge): Ditto.
13210         * tree-vect-loop.c (vect_transform_loop): Ditto.
13211         * profile.c (compute_branch_probabilities): Ditto.
13212         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13213         * lto-streamer-in.c (input_cfg): Ditto.
13214         * gimple-streamer-in.c (input_bb): Ditto.
13215         * ipa-cp.c (update_profiling_info): Ditto.
13216         (update_specialized_profile): Ditto.
13217         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13218         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13219         rounding methods.
13220         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13221         (compute_trg_info): Ditto.
13222         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13223         (purge_dead_edges): Ditto.
13224         * loop-unswitch.c (unswitch_loop): Ditto.
13225         * cgraphclones.c (cgraph_clone_edge): Ditto.
13226         (cgraph_clone_node): Ditto.
13227         * tree-inline.c (copy_bb): Ditto.
13228         (copy_edges_for_bb): Ditto.
13229         (initialize_cfun): Ditto.
13230         (copy_cfg_body): Ditto.
13231         (expand_call_inline): Ditto.
13233 2013-04-08  Kai Tietz  <ktietz@redhat.com>
13235         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13236         TARGET_CYGWIN64 by TARGET_64BIT.
13238 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13240         * config/epiphany/epiphany.md (GPR_1): New constant.
13241         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13242         * config/epiphany/epiphany.c (gen_compare_reg):
13243         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13244         is already in place.
13245         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13246         Don't require being called during rtl expansion; If y operlaps r0,
13247         return 0.
13248         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13249         (epiphany_expand_epilogue): Likewise.
13251         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13252         Don't use CC_FPmode for ORDERED / UNORDERED.
13253         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13255         * config/epiphany/constraints.md (CnL): New constraint.
13256         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13257         * config/epiphany/predicates.md (add_operand): Allow 1024.
13259         * config/epiphany/epiphany.md (logical_op): New code iterator.
13260         (op_mnc): New code attribute.
13261         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13262         (mov_f+1, mov_f+2): New peephole2 patterns.
13264         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13265         (cstoresi4): Also allow re-use of zero result when doing a NE
13266         comparison to a non-zero operand.
13267         Use (clobber (scratch)) for first insn if the gpr output is not needed.
13269         * config/epiphany/epiphany.md (<insn_opname>v2si3):
13270         Use gen_addsi3_i / gen_subsi3_i.
13272 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13274         PR c++/34949
13275         PR c++/50243
13276         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13277         contain anything but clobbers, at most one __builtin_stack_restore,
13278         optionally debug stmts and final resx, and if it has at least one
13279         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
13280         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13281         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13282         which isn't defaut definition, remove them.
13283         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13284         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13285         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13286         with MEM_REF LHS with SSA_NAME address.
13288 2013-04-08  Jeff Law  <law@redhat.com>
13290         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13292 2013-04-08  Richard Biener  <rguenther@suse.de>
13294         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13295         extra newline.
13296         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13297         determined vector type.
13298         (vect_analyze_data_refs): Likewise.
13299         (vect_get_new_vect_var): Adjust.
13300         (vect_create_destination_var): Preserve SSA name versions.
13301         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13302         not dump anything here.
13304 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13306         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13307         Add member lr_slot_known.
13308         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13309         if necessary.
13310         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13311         Remove code that sets lr_slot_offset according to what a previous
13312         version of epiphany_emit_save_restore used to do.
13313         (epiphany_emit_save_restore): When doing an lr save or restore,
13314         set/verify lr_slot_known and lr_slot_offset.
13316 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
13318         PR target/54338
13319         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13320         in ALL_REGS.
13322 2013-04-08  Richard Biener  <rguenther@suse.de>
13324         * alias.c (find_base_term): Fix thinko in previous change.
13326 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
13328         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13329         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13330         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
13331         if possible to compute val.
13332         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13333         For QImode integers don't require anything about precision.  Use
13334         const_with_all_bytes_same to find out if the constant doesn't have
13335         repeated bytes in it.
13337 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13339         * config/s390/s390.c (s390_expand_insv): Only accept insertions
13340         within mode size.
13342 2013-04-08  Marek Polacek  <polacek@redhat.com>
13344         PR rtl-optimization/48182
13345         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13346         value to 1.
13348 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13350         PR target/55487
13351         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13352         nuses, make sure we have a label.
13354 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13356         PR target/56843
13357         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13358         (rs6000_emit_swdiv_low_precision): Remove.
13359         (rs6000_emit_swdiv): Rewrite to handle between one and four
13360         iterations of Newton-Raphson generally; modify required number of
13361         iterations for some cases.
13362         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13364 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
13366         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13367         set-but-unused variable.
13369         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13370         basic blocks of released function bodies garbage-collectable.
13372         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13373         (struct rtl_opt_pass): Add TODO_df_finish.
13375         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13377 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13379         * config/arm/constraints.md (q): New constraint.
13380         * config/arm/ldrdstrd.md: New file.
13381         * config/arm/arm.md (ldrdstrd.md) New include.
13382         (arm_movdi): Use "q" instead of "r" constraint
13383         for double-word memory access.
13384         (movdf_soft_insn): Likewise.
13385         * config/arm/vfp.md (movdi_vfp): Likewise.
13386         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13387         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13388         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13389         (mem_ok_for_ldrd_strd): Likewise.
13390         (output_move_double): Update assertion.
13392 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13394         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13396 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13398         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13399         define_insn_and_split.
13400         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13402 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13404         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13405         define_insn_and_split.
13406         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13407         (shiftsi3_compare): New pattern.
13408         (rrx): New pattern.
13409         * config/arm/unspecs.md (UNSPEC_RRX): New.
13411 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13413         * config/arm/arm.md (negdi_extendsidi): New pattern.
13414         (negdi_zero_extendsidi): Likewise.
13416 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13418         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13419         define_insn_and_split.
13420         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13421         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13423 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13425         * config/arm/arm.md (arm_subdi3): Convert define_insn into
13426         define_insn_and_split.
13427         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13428         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13430 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13432         * config/arm/arm.md (subsi3_carryin): New pattern.
13433         (subsi3_carryin_const): Likewise.
13434         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13435         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13437 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13439         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13441 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
13443         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13444         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13446 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13448         * config/arm/arm.c (arm_expand_builtin): Change fcode
13449         type to unsigned int.
13451 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13453         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13455 2013-04-04  Ian Lance Taylor  <iant@google.com>
13457         * doc/standards.texi (Standards): The Go frontend supports the Go 1
13458         language standard.
13460 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
13462         PR middle-end/56729
13463         * df-scan.c (df_insn_delete): Disable failing assert.
13465 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13467         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13468         New function prototype.
13469         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13470         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13471         (arm_builtin_vectorized_function): New function.
13473 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13475         * config/arm/arm_neon_builtins.def: New file.
13476         * config/arm/arm.c (neon_builtin_data): Move contents to
13477         arm_neon_builtins.def.
13478         (enum arm_builtins): Include neon builtin definitions.
13479         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13480         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13482 2013-04-04  Marek Polacek  <polacek@redhat.com>
13484         PR tree-optimization/48186
13485         * predict.c (maybe_hot_frequency_p): Return false if
13486         HOT_BB_FREQUENCY_FRACTION is 0.
13487         (cgraph_maybe_hot_edge_p): Likewise.
13489 2013-04-04  Richard Biener  <rguenther@suse.de>
13491         PR tree-optimization/56826
13492         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13493         more accurately.
13495 2013-04-04  Richard Biener  <rguenther@suse.de>
13497         PR tree-optimization/56213
13498         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13499         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13501 2013-04-04  Richard Biener  <rguenther@suse.de>
13503         PR tree-optimization/56837
13504         * tree-loop-distribution.c (classify_partition): For non-zero
13505         values require that the value has the same precision as its
13506         mode to be useful as memset value.
13508 2013-04-03  Nick Clifton  <nickc@redhat.com>
13510         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13511         (fmssf4): Use fmsf.s on E3V5 architectures.
13512         (fnmasf4): Use fnmaf.s on E3V5 architectures.
13513         (fnmssf4): Use fnmsf.s on E3V5 architectures.
13515 2013-04-03  Jeff Law  <law@redhat.com>
13517         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13518         (lra-eliminations.o): Likewise.
13520 2013-04-03  Teresa Johnson  <tejohnson@google.com>
13522         * gcov-io.c (compute_working_sets): Moved most of body of old
13523         compute_working_sets here from profile.c.
13524         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13525         (gcov_working_set_t): Moved typedef here from basic-block.h
13526         (compute_working_set): Declare.
13527         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13528         (get_working_sets): Renamed from compute_working_set,
13529         replace most of body with call to new compute_working_sets.
13530         (get_exec_counts): Replace call to compute_working_sets
13531         to get_working_sets.
13532         * profile.h (get_working_sets): Renamed from compute_working_set.
13533         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13534         to get_working_sets.
13535         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13536         * gcov-dump.c (dump_working_sets): New function.
13538 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
13540         * hwint.c (sext_hwi, zext_hwi): New functions.
13541         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13542         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13543         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13544         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13545         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13546         (sext_hwi, zext_hwi): New functions.
13548 2013-04-03  Jeff Law  <law@redhat.com>
13550         PR tree-optimization/56799
13551         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13552         back test for widening conversion erroneously dropped in prior change.
13554 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13556         PR target/56809
13557         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13558         instead of next_real_insn.
13560 2013-04-03  Marek Polacek  <polacek@redhat.com>
13562         PR sanitizer/55702
13563         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13565 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13567         PR target/56809
13568         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13569         next_real_insn.
13570         (thumb1_output_casesi): Likewise.
13571         (thumb2_output_casesi): Likewise.
13573 2013-04-03  Richard Biener  <rguenther@suse.de>
13575         PR tree-optimization/56817
13576         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13577         Split out ...
13578         (tree_unroll_loops_completely_1): ... new function to manually
13579         walk the loop tree, properly defering outer loops of unrolled
13580         loops to later iterations.
13582 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13584         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13585         (vectorizable_load): Likewise.
13586         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13587         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13589 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13591         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13592         BIT_FIELD_REF.
13594 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13596         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13598 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
13600         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13602 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13604         PR tree-optimization/56790
13605         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13606         folding.
13608 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13610         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13611         Handle VEC_MERGE.
13612         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13613         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13614         equal arguments.
13616 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
13618         PR c/19449
13619         * tree.h (force_folding_builtin_constant_p): New decl.
13620         * builtins.c (force_folding_builtin_constant_p): New variable.
13621         (fold_builtin_constant_p): Fold immediately also if
13622         force_folding_builtin_constant_p.
13624 2013-04-03  Richard Biener  <rguenther@suse.de>
13626         PR tree-optimization/56812
13627         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13628         DRs of the same interleaving chain are independent.
13630 2013-04-02  Jason Merrill  <jason@redhat.com>
13632         * gdbinit.in (pbb): Use debug fn.
13634 2013-04-02  Lawrence Crowl  <crowl@google.com>
13636         * sese.h (struct ivtype_map_elt_s): Remove unused.
13637         (extern debug_ivtype_map): Remove unused.
13638         (extern eq_ivtype_map_elts): Remove unused.
13639         * sese.c (debug_ivtype_map): Removed unused.
13640         (debug_ivtype_map_1): Removed unused.
13641         (debug_ivtype_elt): Remove unused.
13642         (eq_ivtype_map_elts): Remove unused.
13645 2013-04-02  Kai Tietz  <ktietz@redhat.com>
13647         PR target/52790
13648         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13649         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13650         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13651         function.
13652         (legitimize_pe_coff_symbol): Likewise.
13653         (is_imported_p): New helper-function.
13654         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13655         for Windows x64 targets.
13656         (ix86_expand_prologue): Optimize for pe-coff targets.
13657         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13658         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13659         medium/large code-model.
13660         (legitimize_pic_address): Likewise.
13661         (legitimize_tls_address): Likewise.
13662         (ix86_expand_call): Likewise.
13663         (x86_output_mi_thunk): Likewise.
13664         (get_dllimport_decl): Add new beimport argument.
13665         (construct_plt_address): Don't assert for x64 pe-coff targets.
13666         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13667         targets.
13668         (SYMBOL_FLAG_STUBVAR): New macro.
13669         (SYMBOL_REF_STUBVAR_P): Likewise.
13670         * config/i386/winnt.c (stub_list): New structure.
13671         (stub_head): New local variable.
13672         (i386_pe_record_stub): New function.
13673         (i386_pe_file_end): Emit refptr-stubs.
13675 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
13677         PR rtl-optimization/56745
13678         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13679         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13681         PR c++/34949
13682         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13683         and both of them are MEM_REFs, just compare first argument for
13684         equality and attempt to deal even with differing offsets.
13686         PR c++/34949
13687         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13688         of gimple_clobber_p to be MEM_REF.
13689         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13690         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
13691         after gimplification.
13692         * asan.c (get_mem_ref_of_assignment): Don't instrument
13693         gimple_clobber_p stmts.
13694         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13695         gimple_clobber_p stmt if they have MEM_REF lhs and
13696         are dead because of another gimple_clobber_p stmt.
13697         * tree-ssa-live.c (clear_unused_block_pointer): Treat
13698         gimple_clobber_p stmts like debug stmts.
13699         (remove_unused_locals): Remove clobbers with MEM_REF lhs
13700         that refer to unused VAR_DECLs or uninitialized values.
13701         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13702         gimple_clobber_p stmts if they refer to removed parameters.
13703         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13704         formatting.
13706 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
13708         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13709         using SWI48 mode attribute.
13711 2013-04-02  Wei Mi  <wmi@google.com>
13713         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13714         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13715         *<rotate_insn><mode>3_mask in i386.md.
13717 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13719         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13721 2013-04-02  Richard Biener  <rguenther@suse.de>
13723         PR tree-optimization/56778
13724         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13725         Runtime alias tests are not supported for gather loads.
13726         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13727         stmts referenced from SSA operands before updating SSA form.
13729 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
13730             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13732         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13733         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13734         * config/arm/cortex-a53.md: New file.
13735         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13736         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13737         * config/arm/arm.c (arm_issue_rate): Likewise.
13738         * config/arm/arm-tune.md: Regenerate
13739         * config/arm/arm-tables.opt: Regenerate.
13740         * config/arm/arm-cores.def: Add cortex-a53.
13742 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
13744         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13745         non-static link.
13747 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13749         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13750         scalar load/store operations using B/H registers.
13751         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13753 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13755         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13756         scalar move.
13757         * config/aarch64/aarch64.c
13758         (aarch64_simd_scalar_immediate_valid_for_move): New.
13759         * config/aarch64/aarch64-protos.h
13760         (aarch64_simd_scalar_immediate_valid_for_move): New.
13761         * config/aarch64/constraints.md (Dh, Dq): New.
13762         * config/aarch64/iterators.md (hq): New.
13764 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
13766         * reorg.c (get_branch_condition): Deal with conditional returns.
13767         (fill_simple_delay_slots): Remove dead code dealing with jumps.
13769 2013-04-01  Wei Mi  <wmi@google.com>
13771         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13772         Truncate operand 2 using %b asm operand modifier.
13773         (*<shift_insn><mode>3_mask): Ditto.
13774         (*<rotate_insn><mode>3_mask): Ditto.
13776 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
13778         PR middle-end/56798
13779         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13781 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
13783         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13784         of next_real_insn.
13785         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13787 2013-03-30  Lawrence Crowl  <crowl@google.com>
13789         * dse.c (clear_alias_sets): Remove never set.
13790         (disqualified_clear_alias_sets): Remove never set.
13791         (clear_alias_mode_pool): Remove never set.
13792         (dse_step0): Remove condition that is never true.
13793         (canon_address): Remove condition that is never true.
13794         (dse_step7): Remove condition that is never true.
13795         (rest_of_handle_dse): Remove condition that is never true.
13796         (rest_of_handle_dse::did_global): Remove never read from above.
13797         (dse_step2_spill): Remove never called from above.
13798         (dse_step5_spill): Remove never called from above.
13800 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
13802         * doc/md.texi (Standard Names) <casesi>: Update documentation for
13803         JUMP_TABLE_DATA changes.
13804         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13805         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13806         (Insns) <jump_table_data>: New entry.
13807         * doc/tm.texi: Regenerate.
13809         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13811         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13812         for table jump at the end of a basic block using tablejump_p.
13813         * targhooks.c (default_invalid_within_doloop): Likewise.
13814         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13815         target hook implementation that is identical to the default hook.
13816         (rs6000_invalid_within_doloop): Remove.
13818         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13819         unused variable from tablejump_p call.
13821         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13822         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13823         (INSN_DELETED_P): Likewise.
13824         (emit_jump_table_data): New prototype.
13825         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13826         after 4th as unused.
13827         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13828         * sched-vis.c (print_insn): Likewise.
13829         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13830         insn for compatibility with back ends that use next_active_insn to
13831         identify jump table data.
13832         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13833         (remove_insn): Likewise.
13834         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13835         to be emitted.
13836         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13837         (emit_jump_table_data): New function.
13839         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13840         basic block, a JUMP_TABLE_DATA never is.
13841         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13842         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13843         off from code handling real insns.
13844         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13845         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13846         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13847         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13848         is not a NONDEBUG_INSN_P.
13849         * ira-costs.c (scan_one_insn): Likewise.
13850         * jump.c (mark_all_labels): Likewise.
13851         (mark_jump_label_1): Likewise.
13852         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13853         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13854         a basic block.
13855         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13856         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13857         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13858         JUMP_TABLE_DATA_P insns.
13859         (calculate_elim_costs_all_insns): Likewise.
13860         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13861         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13862         (delete_output_reload): Code style fixups.
13863         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13864         insn flags on this non-insn.
13865         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13866         as scheduling barriers, for pre-change compatibility.
13867         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13868         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13870         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13871         redundant JUMP_TABLE_DATA_P test.
13872         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13873         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13874         (frv_for_each_packet): Likewise.
13875         * config/i386/i386.c (min_insn_size): Likewise.
13876         (ix86_avoid_jump_mispredicts): Likewise.
13877         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13878         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13879         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13880         (mips16_insn_length): Robustify.
13881         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13882         (mips16_split_long_branches): Likewise.
13883         * config/pa/pa.c (pa_combine_instructions): Likewise.
13884         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13885         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13886         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13887         as contributing to pool range lengths.
13888         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13889         Remove redundant JUMP_TABLE_DATA_P test.
13890         (sh_loop_align): Likewise.
13891         (split_branches): Likewise.
13892         (sh_insn_length_adjustment): Likewise.
13893         * config/spu/spu.c (get_branch_target): Likewise.
13895 2013-03-29  Jan Hubicka  <jh@suse.cz>
13897         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13898         gcov streaming; stream hot bb threshold to ltrans.
13899         * predict.c (get_hot_bb_threshold): Break out from ....
13900         (maybe_hot_count_p): ... here.
13901         (set_hot_bb_threshold): New function.
13902         * lto-section-in.c (lto_section_name): Add profile.
13903         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13904         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13905         and data-streamer.h
13906         (histogram_entry): New structure.
13907         (histogram, histogram_pool): New global vars.
13908         (histogram_hash): New structure.
13909         (histogram_hash::hash): New method.
13910         (histogram_hash::equal): Likewise.
13911         (account_time_size): New function.
13912         (cmp_counts): New function.
13913         (dump_histogram): New function.
13914         (ipa_profile_generate_summary): New function.
13915         (ipa_profile_write_summary): New function.
13916         (ipa_profile_read_summary): New function.
13917         (ipa_profile): Decide on threshold.
13918         (pass_ipa_profile): Add ipa_profile_write_summary and
13919         ipa_profile_read_summary.
13920         * Makefile.in (ipa.o): Update dependencies.
13921         * lto-streamer.h (LTO_section_ipa_profile): New section.
13923 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13925         * tree.h (VAR_P): New.
13927 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13929         PR lto/56777
13930         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13932 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13934         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13935         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13936         (control_flow_insn_p): Likewise.
13937         * cfgrtl.c (duplicate_insn_chain): Likewise.
13938         * final.c (get_attr_length_1): Likewise.
13939         (shorten_branches): Likewise.
13940         (final_scan_insn): Likewise.
13941         * function.c (instantiate_virtual_regs): Likewise.
13942         * gcse.c (insert_insn_end_basic_block): Likewise.
13943         * ira-costs.c (scan_one_insn): Likewise.
13944         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13945         * lra.c (check_rtl): Likewise.
13946         * reload1.c (elimination_costs_in_insn): Likewise.
13947         * reorg.c (follow_jumps): Likewise.
13949         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13950         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13951         (thumb_far_jump_used_p): Likewise.
13952         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13953         (workaround_speculation): Likewise.
13954         (add_sched_insns_for_speculation): Likewise.
13955         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13956         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13957         (frv_for_each_packet): Likewise.
13958         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13959         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13960         (final_emit_insn_group_barriers): Likewise.
13961         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13962         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13963         (mips16_insn_length): Likewise.
13964         * config/pa/pa.c (pa_reorg): Likewise.
13965         (pa_combine_instructions): Likewise.
13966         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13967         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13968         (sh_reorg): Likewise.
13969         (split_branches): Likewise.
13970         * config/spu/spu.c (get_branch_target): Likewise.
13972         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13973         JUMP_TABLE_DATA_P.
13975 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13977         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13978         Fix declaration name.
13980 2013-03-28  Lawrence Crowl  <crowl@google.com>
13982         * graphds.h (struct graph.indicies): Remove unused.
13983         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13984         (SCOP_ORIGINAL_PDDRS): Remove unused.
13985         * sese.h (extern insert_loop_close_phis): Removed unused.
13986         (extern insert_guard_phis): Removed unused.
13987         (extern ivtype_map_elt_info): Removed unused.
13988         (new_ivtype_map_elt): Removed unused.
13989         * sese.c (ivtype_map_elt_info): Removed unused.
13991 2013-03-28  Lawrence Crowl  <crowl@google.com>
13993         * Makefile.in: Add several missing include dependences.
13994         (DUMPFILE_H): New.
13995         (test-dump.o): New.  This object is not added to any executable,
13996         but is present for ad-hoc testing.
13997         * bitmap.c
13998         (debug (const bitmap_head_def &)): New.
13999         (debug (const bitmap_head_def *)): New.
14000         * bitmap.h
14001         (extern debug (const bitmap_head_def &)): New.
14002         (extern debug (const bitmap_head_def *)): New.
14003         * cfg.c
14004         (debug (edge_def &)): New.
14005         (debug (edge_def *)): New.
14006         * cfghooks.c
14007         (debug (basic_block_def &)): New.
14008         (debug (basic_block_def *)): New.
14009         * dumpfile.h
14010         (dump_node (const_tree, int, FILE *)): Correct source file.
14011         * dwarf2out.c
14012         (debug (die_struct &)): New.
14013         (debug (die_struct *)): New.
14014         * dwarf2out.h
14015         (extern debug (die_struct &)): New.
14016         (extern debug (die_struct *)): New.
14017         * gimple-pretty-print.c
14018         (debug (gimple_statement_d &)): New.
14019         (debug (gimple_statement_d *)): New.
14020         * gimple-pretty-print.h
14021         (extern debug (gimple_statement_d &)): New.
14022         (extern debug (gimple_statement_d *)): New.
14023         * ira-build.c
14024         (debug (ira_allocno_copy &)): New.
14025         (debug (ira_allocno_copy *)): New.
14026         (debug (ira_allocno &)): New.
14027         (debug (ira_allocno *)): New.
14028         * ira-int.h
14029         (extern debug (ira_allocno_copy &)): New.
14030         (extern debug (ira_allocno_copy *)): New.
14031         (extern debug (ira_allocno &)): New.
14032         (extern debug (ira_allocno *)): New.
14033         * ira-lives.c
14034         (debug (live_range &)): New.
14035         (debug (live_range *)): New.
14036         * lra-int.h
14037         (debug (lra_live_range &)): New.
14038         (debug (lra_live_range *)): New.
14039         * lra-lives.c
14040         (debug (lra_live_range &)): New.
14041         (debug (lra_live_range *)): New.
14042         * omega.c
14043         (debug (omega_pb_d &)): New.
14044         (debug (omega_pb_d *)): New.
14045         * omega.h
14046         (extern debug (omega_pb_d &)): New.
14047         (extern debug (omega_pb_d *)): New.
14048         * print-rtl.c
14049         (debug (const rtx_def &)): New.
14050         (debug (const rtx_def *)): New.
14051         * print-tree.c
14052         (debug_tree (tree): Move within file.
14053         (debug_raw (const tree_node &)): New.
14054         (debug_raw (const tree_node *)): New.
14055         (dump_tree_via_hooks (const tree_node *, int)): New.
14056         (debug (const tree_node &)): New.
14057         (debug (const tree_node *)): New.
14058         (debug_verbose (const tree_node &)): New.
14059         (debug_verbose (const tree_node *)): New.
14060         (debug_head (const tree_node &)): New.
14061         (debug_head (const tree_node *)): New.
14062         (debug_body (const tree_node &)): New.
14063         (debug_body (const tree_node *)): New.
14064         (debug_vec_tree (tree): Move and reimplement in terms of dump.
14065         (debug (vec<tree, va_gc> &)): New.
14066         (debug (vec<tree, va_gc> *)): New.
14067         * rtl.h
14068         (extern debug (const rtx_def &)): New.
14069         (extern debug (const rtx_def *)): New.
14070         * sbitmap.c
14071         (debug_raw (simple_bitmap_def &)): New.
14072         (debug_raw (simple_bitmap_def *)): New.
14073         (debug (simple_bitmap_def &)): New.
14074         (debug (simple_bitmap_def *)): New.
14075         * sbitmap.h
14076         (extern debug (simple_bitmap_def &)): New.
14077         (extern debug (simple_bitmap_def *)): New.
14078         (extern debug_raw (simple_bitmap_def &)): New.
14079         (extern debug_raw (simple_bitmap_def *)): New.
14080         * sel-sched-dump.c
14081         (debug (vinsn_def &)): New.
14082         (debug (vinsn_def *)): New.
14083         (debug_verbose (vinsn_def &)): New.
14084         (debug_verbose (vinsn_def *)): New.
14085         (debug (expr_def &)): New.
14086         (debug (expr_def *)): New.
14087         (debug_verbose (expr_def &)): New.
14088         (debug_verbose (expr_def *)): New.
14089         (debug (vec<rtx> &)): New.
14090         (debug (vec<rtx> *)): New.
14091         * sel-sched-dump.h
14092         (extern debug (vinsn_def &)): New.
14093         (extern debug (vinsn_def *)): New.
14094         (extern debug_verbose (vinsn_def &)): New.
14095         (extern debug_verbose (vinsn_def *)): New.
14096         (extern debug (expr_def &)): New.
14097         (extern debug (expr_def *)): New.
14098         (extern debug_verbose (expr_def &)): New.
14099         (extern debug_verbose (expr_def *)): New.
14100         (extern debug (vec<rtx> &)): New.
14101         (extern debug (vec<rtx> *)): New.
14102         * sel-sched-ir.h
14103         (_list_iter_cond_expr): Make inline instead of static.
14104         * sreal.c
14105         (debug (sreal &)): New.
14106         (debug (sreal *)): New.
14107         * sreal.h
14108         (extern debug (sreal &)): New.
14109         (extern debug (sreal *)): New.
14110         * tree.h
14111         (extern debug_raw (const tree_node &)): New.
14112         (extern debug_raw (const tree_node *)): New.
14113         (extern debug (const tree_node &)): New.
14114         (extern debug (const tree_node *)): New.
14115         (extern debug_verbose (const tree_node &)): New.
14116         (extern debug_verbose (const tree_node *)): New.
14117         (extern debug_head (const tree_node &)): New.
14118         (extern debug_head (const tree_node *)): New.
14119         (extern debug_body (const tree_node &)): New.
14120         (extern debug_body (const tree_node *)): New.
14121         (extern debug (vec<tree, va_gc> &)): New.
14122         (extern debug (vec<tree, va_gc> *)): New.
14123         * tree-cfg.c
14124         (debug (struct loop &)): New.
14125         (debug (struct loop *)): New.
14126         (debug_verbose (struct loop &)): New.
14127         (debug_verbose (struct loop *)): New.
14128         * tree-dump.c: Add header dependence.
14129         * tree-flow.h
14130         (extern debug (struct loop &)): New.
14131         (extern debug (struct loop *)): New.
14132         (extern debug_verbose (struct loop &)): New.
14133         (extern debug_verbose (struct loop *)): New.
14134         * tree-data-ref.c
14135         (debug (data_reference &)): New.
14136         (debug (data_reference *)): New.
14137         (debug (vec<data_reference_p> &)): New.
14138         (debug (vec<data_reference_p> *)): New.
14139         (debug (vec<ddr_p> &)): New.
14140         (debug (vec<ddr_p> *)): New.
14141         * tree-data-ref.h
14142         (extern debug (data_reference &)): New.
14143         (extern debug (data_reference *)): New.
14144         (extern debug (vec<data_reference_p> &)): New.
14145         (extern debug (vec<data_reference_p> *)): New.
14146         (extern debug (vec<ddr_p> &)): New.
14147         (extern debug (vec<ddr_p> *)): New.
14148         * tree-ssa-alias.c
14149         (debug (pt_solution &)): New.
14150         (debug (pt_solution *)): New.
14151         * tree-ssa-alias.h
14152         (extern debug (pt_solution &)): New.
14153         (extern debug (pt_solution *)): New.
14154         * tree-ssa-alias.c
14155         (debug (_var_map &)): New.
14156         (debug (_var_map *)): New.
14157         (debug (tree_live_info_d &)): New.
14158         (debug (tree_live_info_d *)): New.
14159         * tree-ssa-alias.h
14160         (extern debug (_var_map &)): New.
14161         (extern debug (_var_map *)): New.
14162         (extern debug (tree_live_info_d &)): New.
14163         (extern debug (tree_live_info_d *)): New.
14165 2013-03-28  Jan Hubicka  <jh@suse.cz>
14167         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14169 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
14171         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14172         record only when desired or required.
14174 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
14176         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14177         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
14179 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14181         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14182         (*andsi3_compare0_uxtw): New pattern.
14183         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14184         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14186 2013-03-28  Jan Hubicka  <jh@suse.cz>
14188         * data-streamer-in.c (streamer_read_gcov_count): New function.
14189         * gimple-streamer-out.c: Include value-prof.h.
14190         (output_gimple_stmt): Output histogram.
14191         (output_bb): Use streamer_write_gcov_count.
14192         * value-prof.c: Include data-streamer.h
14193         (dump_histogram_value): Add HIST_TYPE_MAX.
14194         (stream_out_histogram_value): New function.
14195         (stream_in_histogram_value): New function.
14196         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14197         (stream_out_histogram_value, stream_in_histogram_value): Declare.
14198         * data-streamer-out.c (streamer_write_gcov_count): New function.
14199         (streamer_write_gcov_count_stream): New function.
14200         * lto-cgraph.c (lto_output_edge): Update counter streaming.
14201         (lto_output_node): Likewise.
14202         (input_node, input_edge): Likewise.
14203         * lto-streamer-out.c (output_cfg): Update streaming.
14204         * lto-streamer-in.c (input_cfg): Likewise.
14205         * data-streamer.h (streamer_write_gcov_count,
14206         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14207         * gimple-streamer-in.c: Include value-prof.h
14208         (input_gimple_stmt): Input histograms.
14209         (input_bb): Update profile streaming.
14211 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14213         * genmodes.c (emit_max_int): New function.
14214         (emit_insn_modes_h): Added call to emit_max_function.
14215         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14216         Added doc.
14217         * machmode.def: Fixed comment.
14219 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14221         * combine.c (try_combine): Removed useless assert.
14222         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14224 2013-03-28  Marek Polacek  <polacek@redhat.com>
14225             Richard Biener  <rguenther@suse.de>
14227         PR tree-optimization/56695
14228         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14229         build signed result of a vector comparison.
14230         * tree-cfg.c (verify_gimple_comparison): Check that a result
14231         of a vector comparison has signed type.
14233 2013-03-28  Richard Biener  <rguenther@suse.de>
14235         PR tree-optimization/37021
14236         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14237         do not restrict gaps between groups.
14238         * tree-vect-stmts.c (vectorizable_load): Properly account for
14239         a gap between groups.
14241 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
14243         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14244         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14245         is not enabled.
14247 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
14249         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14250         * doc/extend.texi (Named Address Spaces): Ditto.
14251         (Variable Attributes): Ditto.
14253 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14255         * config.build: Add support for cygwin x64 target.
14256         * config.gcc: Likewise.
14257         * config.host: Likewise.
14258         * configure.ac: Likewise
14259         * configure: Regenerated.
14261 2013-03-27  Kai Tietz  <ktietz@redhat.com>
14263         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14264         * config/i386/t-cygwin-w64: New file.
14265         * config/i386/cygwin-w64.h: New file.
14266         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14267         and add support for x64-cygwin target.
14268         (CPP_SPEC): Likewise.
14269         (CXX_WRAP_SPEC_LIST): Undefine before define.
14270         (LIBGCJ_SONAME): Use 15 as version.
14272 2013-03-27  Richard Biener  <rguenther@suse.de>
14274         PR tree-optimization/56716
14275         * tree-ssa-structalias.c (perform_var_substitution): Adjust
14276         dumping for ref nodes.
14278 2013-03-27  Martin Jambor  <mjambor@suse.cz>
14280         PR tree-optimization/55334
14281         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14282         restricted pointers to arrays.
14284 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14286         * Makefile.in (.SUFFIXES): Add .cc.
14287         (.c.o): Apply same recipe for implicit rule .cc.o.
14289 2013-03-27  Richard Biener  <rguenther@suse.de>
14291         PR tree-optimization/37021
14292         * tree-vect-data-refs.c (vect_check_strided_load): Allow
14293         REALPART/IMAGPART_EXPRs around the supported refs.
14294         * tree-ssa-structalias.c (find_func_aliases): Assume that
14295         floating-point values are not used to transfer pointers.
14297 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
14299         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14300         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14301         * gcc/doc/tm.texi: Regenerate.
14302         * gcc/targhooks.h (default_has_ifunc_p): New.
14303         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14304         * gcc/config/linux-protos.h: New file.
14305         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14306         this hook for linux which disables support of indirect functions in
14307         android.
14308         * gcc/config/linux-android.c: New file.
14309         * gcc/config/t-linux-android.c: Ditto.
14310         * gcc/config.gcc: Added new object file linux-android.o.
14311         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14312         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14313         * gcc/varasm.c (do_assemble_alias): Likewise.
14314         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14315         doesn't support indirect functions.
14316         * configure: Regenerate.
14318 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
14320         PR target/56102
14321         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14322         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14323         mult-word mode.
14325 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14327         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14329 2013-03-27  Terry Guo  <terry.guo@arm.com>
14331         * config/arm/arm-cores.def: Added core cortex-r7.
14332         * config/arm/arm-tune.md: Regenerated.
14333         * config/arm/arm-tables.opt: Regenerated.
14334         * doc/invoke.texi: Added entry for core cortex-r7.
14336 2013-03-27  Walter Lee  <walt@tilera.com>
14338         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14339         double-decrement of next_scratch_regno.
14341 2013-03-27  Walter Lee  <walt@tilera.com>
14343         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14344         input operands.
14345         (insn_v1mulus): Ditto.
14346         (insn_v2muls): Ditto.
14348 2013-03-27  Walter Lee  <walt@tilera.com>
14350         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14351         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14353 2013-03-27  Walter Lee  <walt@tilera.com>
14355         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14356         (*sibcall_value): Ditto.
14358 2013-03-27  Walter Lee  <walt@tilera.com>
14360         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14361         (insn_mnz_v8qi): ... this ...
14362         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
14363         vector equivalent.
14364         (insn_v<n>mnz): Replaced by ...
14365         (insn_v1mnz): ... this ...
14366         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
14367         equivalent.
14368         (insn_mz_<mode>): Replaced by ...
14369         (insn_mz_v8qi): ... this ...
14370         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
14371         vector equivalent.
14372         (insn_v<n>mz): Replaced by ...
14373         (insn_v1mz): ... this ...
14374         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
14375         equivalent.
14377 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
14379         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14381 2013-03-26  Roland McGrath  <mcgrathr@google.com>
14383         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14384         than fprintf with a non-constant, non-format string.
14386 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14388         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14389         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
14390         operand 0 predicate.
14391         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14392         attribute.  Use general_x64nomem_operand as operand 1 predicate.
14393         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14394         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14395         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14396         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
14397         (mov<mode>_insv_1): Remove expander.  Merge insn with
14398         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14399         Use general_x64nomem_operand as operand 1 predicate.
14400         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14401         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14402         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14403         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14404         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
14405         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14406         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
14407         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14408         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
14409         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14410         (general_x64nomem_operand): Ditto.
14412 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14414         * config/rtems.opt: Add -pthread option.
14416 2013-03-26  Richard Biener  <rguenther@suse.de>
14418         * alias.c (find_base_term): Avoid redundant and not used recursion.
14419         (base_alias_check): Get the initial base term from the caller.
14420         (true_dependence_1): Compute and pass base terms to base_alias_check.
14421         (write_dependence_p): Likewise.
14422         (may_alias_p): Likewise.
14424 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
14426         * config/aarch64/aarch64.c (aarch64_classify_address): Support
14427         PC-relative load in SI modes and above only.
14429 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
14431         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14432         * config/arm/iwmmxt.md (WCGR0): Update.
14433         (WCGR1, WCGR2, WCGR3): Likewise.
14435 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
14437         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14438         Use x64 and nox64 isa attributes.
14440 2013-03-26  Richard Biener  <rguenther@suse.de>
14442         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14443         alignment computations and rely on get_object_alignment_1
14444         for the !TYPE_P case.
14445         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14447 2013-03-26  Walter Lee  <walt@tilera.com>
14449         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14450         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14452 2013-03-25  Jeff Law  <law@redhat.com>
14454         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14455         check for INTEGRAL_TYPE_P that was missing due to checking in
14456         wrong version of prior patch.
14458 2013-03-25  Walter Lee  <walt@tilera.com>
14460         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14461         TILEGX_INSN_SHUFFLEBYTES1.
14462         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14463         shufflebytes1.
14464         (tilegx_builtins): Ditto.
14465         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14467 2013-03-25  Walter Lee  <walt@tilera.com>
14469         * config/tilegx/tilegx.md (floatsisf2): New pattern.
14470         (floatunssisf2): New pattern.
14471         (floatsidf2): New pattern.
14472         (floatunssidf2): New pattern.
14474 2013-03-25  Walter Lee  <walt@tilera.com>
14476         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14477         tests for constraint J, K, N, P.
14479 2013-03-25  Walter Lee  <walt@tilera.com>
14481         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14482         Use indirect/pcrel encoding.
14483         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14484         Ditto.
14486 2013-03-25  Steve Ellcey  <sellcey@mips.com>
14488         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14489         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14490         * config/mips/mips.c (mips_option_override): Set IMADD default.
14491         * config/mips/mips.h (PTF_AVOID_IMADD): New.
14492         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14493         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14494         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14495         * doc/invoke.texi (-mimadd/-mno-imadd): New.
14497 2013-03-25  Jeff Law  <law@redhat.com>
14499         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14500         slightly to avoid creating and folding useless trees.  Simplify
14501         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14503 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
14505         * config/i386/i386.md (*zero_extendsidi2): Merge with
14506         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
14507         * config/i386/predicates.md (x86_64_zext_operand): Rename from
14508         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
14509         targets.  Clarify comment.
14511 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14513         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14514         pass-through jump functions differently.
14515         (ipa_read_jump_function): Likewise.  Also use setter functions to set
14516         up jump functions.
14518 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14520         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14521         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14522         process it.
14523         (ipa_get_indirect_edge_target): New function.
14524         (devirtualization_time_bonus): New parameter known_aggs, pass it to
14525         ipa_get_indirect_edge_target.  Update all callers.
14526         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
14527         ipa_get_indirect_edge_target_1 instead of calling
14528         ipa_get_indirect_edge_target.
14529         (create_specialized_node): Pass aggvlas to
14530         ipcp_discover_new_direct_edges.
14532 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14534         * config/arm/arm.md (f_sels, f_seld): New types.
14535         (*cmov<mode>): New pattern.
14536         * config/arm/predicates.md (arm_vsel_comparison_operator): New
14537         predicate.
14539 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14541         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14542         POSIX-printf for mingw-hosted  builds.
14544 2013-03-25  Richard Biener  <rguenther@suse.de>
14546         PR middle-end/56694
14547         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14548         must-not-throw stmt location.
14550 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14552         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14553         Emit load-acquire versions when acq is true.
14554         (arm_emit_store_exclusive): Add rel parameter.
14555         Emit store-release versions when rel is true.
14556         (arm_split_compare_and_swap): Use acquire-release instructions
14557         instead.
14558         of barriers when appropriate.
14559         (arm_split_atomic_op): Likewise.
14560         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14561         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14562         (VUNSPEC_SLX): Likewise.
14563         (VUNSPEC_LDA): Likewise.
14564         (VUNSPEC_STL): Likewise.
14565         * config/arm/sync.md (atomic_load<mode>): New pattern.
14566         (atomic_store<mode>): Likewise.
14567         (arm_load_acquire_exclusive<mode>): Likewise.
14568         (arm_load_acquire_exclusivesi): Likewise.
14569         (arm_load_acquire_exclusivedi): Likewise.
14570         (arm_store_release_exclusive<mode>): Likewise.
14572 2013-03-25  Catherine Moore  <clm@codesourcery.com>
14574         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14575         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14576         * config/mip/predicates.md (lwsp_swsp_operand,
14577         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14578         sb16_operand, db4_operand, db7_operand, ib3_operand,
14579         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14580         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14581         andi16_operand): New predicates.
14582         * config/mips/mips.md (compression): New attribute.
14583         (enabled): New attribute.
14584         (length): Consider compression in computing length.
14585         (shift_compression): New code attribute.
14586         (*add<mode>3): New operands. Record compression.
14587         (sub<mode>3): Likewise.
14588         (one_cmpl<mode>2): Likewise.
14589         (*and<mode>3): Likewise.
14590         (*ior<mode>3): Likewise.
14591         (unnamed pattern for xor): Likewise.
14592         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14593         (*<optab><mode>3): Likewise.
14594         (*mov<mode>_internal: Likewise.
14595         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14596         (mips_unsigned_immediate_p): New.
14597         (umips_lwsp_swsp_address_p): New.
14598         (m16_based_address_p): New.
14599         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14600         (mips_unsigned_immediate_p): New prototype.
14601         (lwsp_swsp_address_p): New prototype.
14602         (m16_based_address_p): New prototype.
14603         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14604         (mips_signed_immediate_p): New function.
14605         (m16_based_address_p): New function.
14606         (lwsp_swsp_address_p): New function.
14607         (mips_print_operand_punctuation): Recognize short delay slot insns
14608         for microMIPS.add<mode>3"
14610 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14612         PR target/56720
14613         * config/arm/iterators.md (v_cmp_result): New mode attribute.
14614         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14616 2013-03-25  Richard Biener  <rguenther@suse.de>
14618         PR tree-optimization/56689
14619         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14620         any edge.
14622 2013-03-25  Richard Biener  <rguenther@suse.de>
14624         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14625         of bitmap.
14626         (memory_references): Likewise.
14627         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14628         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14629         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14630         (gather_mem_refs_in_loops): Fold into ...
14631         (analyze_memory_references): ... this.  Move initialization
14632         to tree_ssa_lim_initialize.
14633         (fill_always_executed_in): Rename to ...
14634         (fill_always_executed_in_1): ... this.
14635         (fill_always_executed_in): Move contains_call computation to
14636         this new function from ...
14637         (tree_ssa_lim_initialize): ... here.
14638         (tree_ssa_lim): Call fill_always_executed_in.
14640 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14642         * postreload.c (reload_combine): Fix code detecting returns.
14644 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14646         * function.c (emit_use_return_register_into_block): On cc0 targets,
14647         do not emit the sequence between cc0 setter and user.
14649 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14651         * config/i386/predicates.md (local_symbolic_operand): Interpret
14652         dll-imported symbols as none-local.
14654 2013-03-25  Richard Biener  <rguenther@suse.de>
14656         * tree-ssa-loop-im.c (struct depend): Remove.
14657         (struct lim_aux_data): Make depends a vec of gimples.
14658         (free_lim_aux_data): Adjust.
14659         (add_dependency): Likewise.
14660         (set_level): Likewise.
14662 2013-03-25  Richard Biener  <rguenther@suse.de>
14664         PR middle-end/56434
14665         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14666         the pointer returned by calls with ECF_MALLOC set.
14668 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14670         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14672 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14674         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14675         using MMXMODE mode iterator.
14676         (*move<mode>_internal): Merge with *movv2sf_internal and
14677         *movv2sf_internal_rex64 using MMXMODE mode iterator.
14679 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14681         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14682         (record_last_mem_set_info): Likewise.
14684         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14685         of XNEWVEC followed by memset.
14686         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14688 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14690         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14691         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14692         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14693         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14694         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14695         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14696         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14697         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14698         BARRIER_P instead of GET_CODE.
14700 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14702         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14703         inaccuracy in the probing code.
14705         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14706         (ctrapdi4): Likewise.
14708 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14710         * calls.c (expand_call): Add missing guard to code handling return
14711         of non-BLKmode structures in MSB.
14712         * function.c (expand_function_end): Likewise.
14714 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14716         * combine.c (try_combine): Adjust comment.  Do not add the set of
14717         insn #0 if the destination indirectly is set or dies in insn #2.
14718         Tidy up code to distribute a new note.
14720 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14722         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14723         also for alternatives 16 and 17.
14725 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14727         * config/i386/sse.md (*mov<mode>_internal): Merge with
14728         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
14729         Emit insn template depending on type attribute.  Use
14730         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14731         movd instead of movq mnemonic for interunit moves.  Rewrite mode
14732         attribute calculation.  Remove unit attribute calculation.
14733         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14734         Set prefix_data16 attribute for DImode ssemov types.
14735         Use Ym instead of y for SSE-MMX conversion alternatives.
14736         Reorder operand constraints.
14738 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
14740         * df.h (df_insn_delete): Adjust prototype.
14741         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14742         and let it decide whether mark the basic block dirty.
14743         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14744         * df-scan.c (df_insn_info_delete): New helper function, split
14745         off from df_insn_delete.
14746         (df_scan_free_bb_info): Use it.
14747         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14748         Likewise.
14749         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
14750         that the insn is actually an insn and it has a non-NULL basic block.
14751         Do not mark basic block dirty if only deleting a DEBUG_INSN.
14753 2013-03-22  Richard Biener  <rguenther@suse.de>
14755         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14756         dep_ref members.
14757         (mem_ref_alloc): Do not allocate them.
14758         (refs_independent_p): Do not query or maintain a cache.
14760 2013-03-22  Richard Biener  <rguenther@suse.de>
14762         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14763         (gather_mem_refs_in_loops): Do not compute it.
14764         (analyze_memory_references): Do not allocate it.
14765         (tree_ssa_lim_finalize): Do not free it.
14766         (for_all_locs_in_loop): Do not query all_refs_in_loop.
14768 2013-03-22  Richard Biener  <rguenther@suse.de>
14770         * is-a.h (as_a): Use gcc_checking_assert.
14772 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
14774         * config/aarch64/aarch64.c (aarch64_print_operand): New
14775         format specifier for printing a constant in hex.
14776         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14777         format specifier for printing second operand.
14779 2013-03-22  Richard Biener  <rguenther@suse.de>
14781         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14782         bitmaps.
14783         (gather_mem_refs_in_loops): Perform store accumulation here.
14784         (create_vop_ref_mapping_loop): Remove.
14785         (create_vop_ref_mapping): Likewise.
14786         (analyze_memory_references): Initialize refs_stored_in_loop.
14787         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14788         (record_indep_loop): Remove.
14789         (record_dep_loop): New function.
14790         (ref_indep_loop_p_1): Adjust to only walk over references
14791         in the loop, not its subloops.
14792         (ref_indep_loop_p): Rename to ...
14793         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14794         maintaining a more fine-grained cache.
14795         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14796         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14798 2013-03-22  Richard Biener  <rguenther@suse.de>
14800         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14801         (struct mem_ref): Make accesses_in_loop a vec of a vec of
14802         aggregate mem_ref_loc.
14803         (free_mem_ref_locs): Inline into ...
14804         (memref_free): ... this and adjust.
14805         (mem_ref_alloc): Adjust.
14806         (mem_ref_locs_alloc): Remove.
14807         (record_mem_ref_loc): Adjust.
14808         (get_all_locs_in_loop): Rewrite into ...
14809         (for_all_locs_in_loop): ... this iterator.
14810         (rewrite_mem_ref_loc): New functor.
14811         (rewrite_mem_refs): Use for_all_locs_in_loop.
14812         (sm_set_flag_if_changed): New functor.
14813         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14814         (ref_always_accessed): New functor.
14815         (ref_always_accessed_p): Use for_all_locs_in_loop.
14817 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14819         * tree-pass.h (PROP_gimple_lvec): New.
14820         * passes.c (dump_properties): Handle PROP_gimple_lvec.
14821         (init_optimization_passes): Move pass_lower_vector.
14822         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14823         PROP_gimple_lvec.
14824         (pass_lower_vector): Provide PROP_gimple_lvec.
14825         (pass_lower_vector_ssa): Likewise.
14826         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14828 2013-03-21  Mark Wielaard  <mjw@redhat.com>
14830         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14832 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14834         * config/i386/i386.md (*movdi_internal): Disparage slightly
14835         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14836         conversion alternatives.
14838 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14840         PR middle-end/48087
14841         * diagnostic.def (DK_WERROR): New kind.
14842         * diagnostic.h (werrorcount): Define.
14843         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14844         promoted to DK_ERROR, increment DK_WERROR counter instead of
14845         DK_ERROR counter.
14846         * toplev.c (toplev_main): Call print_ignored_options even if
14847         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14848         even if just werrorcount is non-zero.
14850         PR debug/55608
14851         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14852         on failure.
14853         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14854         (string_cst_pool_decl): New function.
14855         (optimize_one_addr_into_implicit_ptr): New function.
14856         (resolve_addr_in_expr): Optimize DWARF location expression
14857         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14858         which doesn't live in memory, but has DW_AT_location or
14859         DW_AT_const_value, or refers to a string literal, into
14860         DW_OP_GNU_implicit_pointer.
14861         (optimize_location_into_implicit_ptr): New function.
14862         (resolve_addr): If removing DW_AT_location of a variable because
14863         it was DW_OP_addr of address of the variable, but the variable doesn't
14864         live in memory, try to emit const value attribute for the initializer.
14866 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14868         * tree.h (VECTOR_TYPE_P): New macro.
14869         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14870         TYPE_MODE): Use it.
14871         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14872         VEC_COND_EXPR cannot be lvalues.
14873         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14875 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14877         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14878         Restrict the transformation to equal modes.
14880 2013-03-21  Richard Biener  <rguenther@suse.de>
14882         PR tree-optimization/39326
14883         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14884         (MEM_ANALYZABLE): Adjust.
14885         (record_mem_ref_loc): Move bitmap ops ...
14886         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14887         unanalyzable refs, do not record locations for it.
14888         (analyze_memory_references): Allocate ref zero as shared
14889         unanalyzable ref.
14890         (refs_independent_p): Do not test for unanalyzed mems here.
14891         (ref_indep_loop_p_1): Special-case disambiguation against
14892         the unanalyzed ref.
14893         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14895 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14897         * config/arm/arm-protos.h (tune_params): Add
14898         prefer_neon_for_64bits field.
14899         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14900         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14901         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14902         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14903         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14904         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14905         (arm_option_override): Handle -mneon-for-64bits new option.
14906         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14907         (prefer_neon_for_64bits): Declare new variable.
14908         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14909         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14910         (arch_enabled): Handle new arch types. Remove support for onlya8
14911         and nota8.
14912         (one_cmpldi2): Use new arch names.
14913         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14914         * config/arm/arm.opt (mneon-for-64bits): Add option.
14915         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14916         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14917         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14918         of onlya8.
14919         * doc/invoke.texi (-mneon-for-64bits): Document.
14921 2013-03-21  Richard Biener  <rguenther@suse.de>
14923         PR tree-optimization/39326
14924         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14925         (sort_bbs_in_loop_postorder_cmp): New function.
14926         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14928 2013-03-21  Richard Biener  <rguenther@suse.de>
14930         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14931         (vect_insert_into_interleaving_chain): Likewise.
14932         (vect_drs_dependent_in_basic_block): Inline ...
14933         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14934         split out from ...
14935         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14936         (vect_check_interleaving): Simplify.
14937         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14938         (vect_slp_analyze_data_ref_dependences): ... this new function.
14939         (dr_group_sort_cmp): New function.
14940         (vect_analyze_data_ref_accesses): Compute data-reference groups
14941         here instead of in vect_analyze_data_ref_dependence.  Use
14942         a more efficient algorithm.
14943         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14944         vect_slp_analyze_data_ref_dependences.  Call
14945         vect_analyze_data_ref_accesses earlier.
14946         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14947         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14948         (vect_slp_analyze_data_ref_dependences): New prototype.
14950 2013-03-21  Richard Biener  <rguenther@suse.de>
14952         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14953         ref is stored in the loop.
14954         (find_refs_for_sm): Walk only over all stores.
14955         (store_motion_loop): Allocate from lim_bitmap_obstack.
14956         (store_motion): Likewise.
14958 2013-03-21  Richard Biener  <rguenther@suse.de>
14960         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14961         Update virtual SSA form.
14963 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14965         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14966         * configure: Regenerate.
14967         * config.in: Regenerate.
14968         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14969         if !HAVE_LD_EH_FRAME_CIEV3.
14971 2013-03-21  Richard Biener  <rguenther@suse.de>
14973         * tree-cfg.c (verify_expr_no_block): New function.
14974         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14975         nor DECL_VALUE_EXPR have locations with associated blocks.
14976         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14977         (clear_unused_block_pointer): Remove code dealing with
14978         blocks in DECL_DEBUG_EXPR locations.
14980 2013-03-21  Richard Biener  <rguenther@suse.de>
14982         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14983         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14984         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14985         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14986         instead of DECL_DEBUG_EXPR_IS_FROM.
14987         * gimplify.c (gimplify_modify_expr): Likewise.
14988         * tree-cfg.c (verify_expr_location_1): Likewise.
14989         * tree-complex.c (create_one_component_var): Likewise.
14990         * tree-sra.c (create_access_replacement): Likewise.
14991         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14992         (clear_unused_block_pointer): Likewise.
14993         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14994         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14995         * var-tracking.c (var_debug_decl): Likewise.
14996         (track_expr_p): Likewise.
14997         * tree-inline.c (add_local_variables): Likewise.  Set
14998         DECL_HAS_DEBUG_EXPR_P after copying it.
14999         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
15000         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
15002 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
15004         PR bootstrap/56656
15005         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
15006         * configure: Regenerate.
15007         * config.in: Regenerate.
15008         * config/i386/i386.md (*movdf_internal): Use
15009         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
15010         movd instead of movq mnemonic for interunit moves.
15011         (*movdi_internal): Ditto.
15013 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15015         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
15016         (abd<mode>_3): New pattern.
15017         (aba<mode>_3): New pattern.
15018         (fabd<mode>_3): New pattern.
15020 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15022         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
15023         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
15024         occurrence of REGISTER_PREFIX as its empty string.
15026 2013-03-20  Jeff Law  <law@redhat.com>
15028         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
15029         addititional equivalences for equality comparisons between an SSA_NAME
15030         and a constant where the SSA_NAME was set from a widening conversion.
15032 2013-03-20  Walter Lee  <walt@tilera.com>
15034         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
15036 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15038         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
15039         depending on type attribute.
15040         (*movti_internal): Ditto.
15041         (*movtf_internal): Ditto.
15042         (*movxf_internal): Ditto.
15043         (*movdf_internal): Ditto.
15044         (*movsf_internal): Ditto.
15046 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15048         * config/i386/i386.md (*movti_internal): Set prefix attribute to
15049         maybe_vex for sselog1 and ssemov types.
15050         (*movdi_internal): Reorder operand constraints.
15051         (*movsi_internal): Ditto.  Set prefix attribute to
15052         maybe_vex for sselog1 and ssemov types.
15053         (*movtf_internal): Set prefix attribute to maybe_vex
15054         for sselog1 and ssemov types.
15055         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
15056         DImode ssemov types.  Reorder operand constraints.
15057         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
15058         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
15059         attribute for SImode ssemov types.  Reorder operand constraints.
15061 2013-03-20  Martin Jambor  <mjambor@suse.cz>
15063         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
15064         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
15066 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
15068         * config/rs6000/predicates.md (indexed_address, update_address_mem
15069         update_indexed_address_mem): New predicates.
15070         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
15071         attribute for load/store instructions.
15072         * config/rs6000/dfp.md (movsd_store): Likewise.
15073         (movsd_load): Likewise.
15074         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
15075         (unnamed HI->DI extend define_insn): Likewise.
15076         (unnamed SI->DI extend define_insn): Likewise.
15077         (unnamed QI->SI extend define_insn): Likewise.
15078         (unnamed QI->HI extend define_insn): Likewise.
15079         (unnamed HI->SI extend define_insn): Likewise.
15080         (unnamed HI->SI extend define_insn): Likewise.
15081         (extendsfdf2_fpr): Likewise.
15082         (movsi_internal1): Likewise.
15083         (movsi_internal1_single): Likewise.
15084         (movhi_internal): Likewise.
15085         (movqi_internal): Likewise.
15086         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
15087         attribute for load/store instructions.
15088         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
15089         instructions.
15090         (mov<mode>_softfloat): Likewise.
15091         (mov<mode>_hardfloat32): Likewise.
15092         (mov<mode>_hardfloat64): Likewise.
15093         (mov<mode>_softfloat64): Likewise.
15094         (movdi_internal32): Likewise.
15095         (movdi_internal64): Likewise.
15096         (probe_stack_<mode>): Likewise.
15098 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
15100         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15101         floating point, and decimal floating point to reload iterator.
15103         * config/rs6000/constraints.md (wl constraint): New constraints to
15104         return FLOAT_REGS if certain options are used to reduce the number
15105         of separate patterns that exist in the file.
15106         (wx constraint): Likewise.
15107         (wz constraint): Likewise.
15109         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15110         -mdebug=reg, print wg, wl, wx, and wz constraints.
15111         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15112         Initialize the reload functions for 64-bit binary/decimal floating
15113         point types.
15114         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15115         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15116         create the buffer on the stack to overcome not having a 32-bit
15117         load and store.
15118         (rs6000_emit_move): Likewise.
15119         (rs6000_secondary_memory_needed_rtx): Likewise.
15120         (rs6000_alloc_sdmode_stack_slot): Likewise.
15121         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15122         via xxlxor, just like DFmode 0.0.
15124         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15125         define as 1 if we are running on a power7 or newer.
15126         (enum r6000_reg_class_enum): Add new constraints.
15128         * config/rs6000/dfp.md (movsd): Delete, combine with binary
15129         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15130         with other moves by using conditional constraits (wg).  Use LFIWZX
15131         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
15132         (movsd splitter): Likewise.
15133         (movsd_hardfloat): Likewise.
15134         (movsd_softfloat): Likewise.
15136         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15137         binary and decimal floating point moves.
15138         (fmove_ok): New attributes to combine binary and decimal floating
15139         point moves, and to combine power6x (mfpgpr) moves along normal
15140         floating moves.
15141         (real_value_to_target): Likewise.
15142         (f32_lr): Likewise.
15143         (f32_lm): Likewise.
15144         (f32_li): Likewise.
15145         (f32_sr): Likewise.
15146         (f32_sm): Likewise.
15147         (f32_si): Likewise.
15148         (movsf): Combine binary and decimal floating point moves.  Combine
15149         power6x (mfpgpr) moves with other moves by using conditional
15150         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
15151         (mov<mode> for SFmode/SDmode); Likewise.
15152         (SFmode/SDmode splitters): Likewise.
15153         (movsf_hardfloat): Likewise.
15154         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15155         (movsf_softfloat): Likewise.
15156         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15158         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15159         wx and wz constraints.
15161         * config/rs6000/constraints.md (wg constraint): New constraint to
15162         return FLOAT_REGS if -mmfpgpr (power6x) was used.
15164         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15165         constraint.
15167         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15168         -mdebug=reg, print wg, wl, wx, and wz constraints.
15169         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15170         Initialize the reload functions for 64-bit binary/decimal floating
15171         point types.
15172         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15173         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15174         create the buffer on the stack to overcome not having a 32-bit
15175         load and store.
15176         (rs6000_emit_move): Likewise.
15177         (rs6000_secondary_memory_needed_rtx): Likewise.
15178         (rs6000_alloc_sdmode_stack_slot): Likewise.
15179         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15180         via xxlxor, just like DFmode 0.0.
15182         * config/rs6000/dfp.md (movdd): Delete, combine with binary
15183         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15184         with other moves by using conditional constraits (wg).  Use LFIWZX
15185         and STFIWX for loading SDmode on power7.
15186         (movdd splitters): Likewise.
15187         (movdd_hardfloat32): Likewise.
15188         (movdd_softfloat32): Likewise.
15189         (movdd_hardfloat64_mfpgpr): Likewise.
15190         (movdd_hardfloat64): Likewise.
15191         (movdd_softfloat64): Likewise.
15193         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15194         64-bit binary and decimal floating point moves.
15195         (FMOVE64X): Likewise.
15196         (movdf): Combine 64-bit binary and decimal floating point moves.
15197         Combine power6x (mfpgpr) moves with other moves by using
15198         conditional constraits (wg).
15199         (mov<mode> for DFmode/DDmode): Likewise.
15200         (DFmode/DDmode splitters): Likewise.
15201         (movdf_hardfloat32): Likewise.
15202         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15203         (movdf_softfloat32): Likewise.
15204         (movdf_hardfloat64_mfpgpr): Likewise.
15205         (movdf_hardfloat64): Likewise.
15206         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15207         (movdf_softfloat64): Likewise.
15208         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15209         (reload_<mode>_load): Move to later in the file so they aren't in
15210         the middle of the floating point move insns.
15211         (reload_<mode>_store): Likewise.
15213         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15214         constraint.
15216         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15217         constraint if -mdebug=reg.
15218         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15219         Enable using dd reload support if needed.
15221         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15222         binary and decimal floating point moves in rs6000.md.
15223         (movtd_internal): Likewise.
15225         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15226         decimal floating point moves.
15227         (movtf): Likewise.
15228         (movtf_internal): Likewise.
15229         (mov<mode>_internal, TDmode/TFmode): Likewise.
15230         (movtf_softfloat): Likewise.
15231         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15233         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15234         movdi_internal64, using wg constraint for move direct operations.
15235         (movdi_internal64): Likewise.
15237         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15238         MODES_TIEABLE_P for selected modes.  Print the numerical value of
15239         the various virtual registers. Use GPR/FPR first/last values,
15240         instead of hard coding the register numbers.  Print which modes
15241         have reload functions registered.
15242         (rs6000_option_override_internal): If -mdebug=reg, trace the options
15243         settings before/after setting cpu, target and subtarget settings.
15244         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15245         and for secondary reload failures in rs6000_secondary_reload_inner.
15246         (rs6000_secondary_reload_fail): Likewise.
15247         (rs6000_secondary_reload_inner): Likewise.
15249         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15250         macros for first/last GPR and FPR registers.
15251         (LAST_GPR_REGNO): Likewise.
15252         (FIRST_FPR_REGNO): Likewise.
15253         (LAST_FPR_REGNO): Likewise.
15255         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15256         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15257         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15258         (vcond<mode><mode>): Likewise.
15259         (vcondu<mode><mode>): Likewise.
15260         (vector_gtu<mode>): Likewise.
15261         (vector_gte<mode>): Likewise.
15262         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15263         to prevent the compiler from converting DImode operations to TImode.
15264         (ior<mode>3): Likewise.
15265         (and<mode>3): Likewise.
15266         (one_cmpl<mode>2): Likewise.
15267         (nor<mode>3): Likewise.
15268         (andc<mode>3): Likewise.
15270         * config/rs6000/constraints.md (wt constraint): New constraint
15271         that returns VSX_REGS if TImode is allowed in VSX registers.
15273         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15274         constant under VSX.
15276         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15277         similar to TImode, but it is restricted to being in the GPRs.
15279         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15280         TImode to occupy a single VSX register.
15282         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15283         -mvsx-timode for power7/power8.
15284         (power7 cpu): Likewise.
15285         (power8 cpu): Likewise.
15287         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15288         sure that TFmode/TDmode take up two registers if they are ever
15289         allowed in the upper VSX registers.
15290         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15291         registers.
15292         (rs6000_init_hard_regno_mode_ok): Likewise.
15293         (rs6000_debug_reg_global): Add debugging for PTImode and wt
15294         constraint.  Print if LRA is turned on.
15295         (rs6000_option_override_internal): Give an error if -mvsx-timode
15296         and VSX is not enabled.
15297         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
15298         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15299         to reg+offset addressing.  Use PTImode when checking offset
15300         addresses for validity.
15301         (reg_offset_addressing_ok_p): Likewise.
15302         (rs6000_legitimate_offset_address_p): Likewise.
15303         (rs6000_legitimize_address): Likewise.
15304         (rs6000_legitimize_reload_address): Likewise.
15305         (rs6000_legitimate_address_p): Likewise.
15306         (rs6000_eliminate_indexed_memrefs): Likewise.
15307         (rs6000_emit_move): Likewise.
15308         (rs6000_secondary_reload): Likewise.
15309         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
15310         reloads to fpr registers to continue to use reg+offset addressing,
15311         but 64-bit reloads to altivec registers need reg+reg addressing.
15312         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15313         it.  Treat LO_SUM like a PLUS operation.
15314         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15315         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15316         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15317         registers to share a register with a smaller sized type, since VSX
15318         puts scalars in the upper 64-bits.
15319         (print_operand): Add support for PTImode.
15320         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15321         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15322         registers, but don't have arithmetic support.
15323         (rs6000_memory_move_cost): Add test for VSX.
15324         (rs6000_opt_masks): Add -mvsx-timode.
15326         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15327         for TImode.
15328         (VSs): Likewise.
15329         (VSr): Use wt constraint for TImode.
15330         (VSv): Drop TImode support.
15331         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15332         (vsx_movti_64bit): Likewise.
15333         (vsx_movti_32bit): Likewise.
15334         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15335         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15336         one '?' on the appropriate output constraint.  Do not allow TImode
15337         logical operations on 32-bit systems.
15338         (vsx_ior<mode>3): Likewise.
15339         (vsx_xor<mode>3): Likewise.
15340         (vsx_one_cmpl<mode>2): Likewise.
15341         (vsx_nor<mode>3): Likewise.
15342         (vsx_andc<mode>3): Likewise.
15343         (vsx_concat_<mode>): Likewise.
15344         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15346         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15347         OPTION_MASK_VSX_TIMODE.
15348         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15349         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15351         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15352         (TI2 iterator): New iterator for TImode, PTImode.
15353         (wd mode attribute): Add values for vector types.
15354         (movti_string): Replace TI move operations with operations for TImode
15355         and PTImode.  Add support for TImode being allowed in VSX registers.
15356         (mov<mode>_string, TImode/PTImode): Likewise.
15357         (movti_ppc64): Likewise.
15358         (mov<mode>_ppc64, TImode/PTImode): Likewise.
15359         (TI mode splitters): Likewise.
15361         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15362         constraint.
15364 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
15366         PR tree-optimization/56355
15367         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15368         Also handle integers with undefined overflow.
15370 2013-03-20  Catherine Moore  <clm@codesourcery.com>
15371             Maciej W. Rozycki  <macro@codesourcery.com>
15372             Tom de Vries  <tom@codesourcery.com>
15373             Nathan Sidwell  <nathan@codesourcery.com>
15374             Iain Sandoe  <iain@codesourcery.com>
15375             Nathan Froyd  <froydnj@codesourcery.com>
15376             Chao-ying Fu  <fu@mips.com>
15378         * doc/extend.texi: (micromips, nomicromips, nocompression):
15379         Document new function attributes.
15380         * doc/invoke.texi (minterlink-compressed, mmicromips,
15381         m14k, m14ke, m14kec): Document new options.
15382         (minterlink-mips16): Update documentation.
15383         * doc/md.texi (ZC, ZD): Document new constraints.
15384         * configure.ac (gcc_cv_as_micromips): Check if linker
15385         supports the .set micromips directive.
15386         * configure: Regenerate.
15387         * config.in: Regenerate.
15388         * config/mips/mips-tables.opt: Regenerate.
15389         * config/mips/micromips.md: New file.
15390         * constraints.md (ZC, ZD): New constraints.
15391         * config/mips/predicates.md (movep_src_register): New predicate.
15392         (movep_src_operand): New predicate.
15393         (non_volatile_mem_operand): New predicate.
15394         * config/mips/mips.md (multimem): New type.
15395         (length): Differentiate between 17-bit and 18-bit branch offsets.
15396         (MOVEP1, MOVEP2): New mode iterator.
15397         (mov_<load>l): Use ZC constraint.
15398         (mov_<load>r): Likewise.
15399         (mov_<store>l): Likewise.
15400         (mov_<store>r): Likewise.
15401         (*branch_equality<mode>_inverted): Add microMIPS support.
15402         (*branch_equality<mode>): Likewise.
15403         (*jump_absolute): Likewise.
15404         (indirect_jump_<mode>): Likewise.
15405         (tablejump_<mode>): Likewise.
15406         (<optab>_internal): Likewise.
15407         (sibcall_internal): Likewise.
15408         (sibcall_value_internal): Likewise.
15409         (prefetch): Use constraint ZD.
15410         * config/mips/mips.opt (minterlink-compressed): New option.
15411         (minterlink-mips16): Now an alias for minterlink-compressed.
15412         (mmicromips): New option.
15413         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15414         (compare_and_swap_12): Likewise.
15415         (sync_add<mode>): Likewise.
15416         (sync_<optab>_12): Likewise.
15417         (sync_old_<optab>_12): Likewise.
15418         (sync_new_<optab>_12): Likewise.
15419         (sync_nand_12): Likewise.
15420         (sync_old_nand_12): Likewise.
15421         (sync_new_nand_12): Likewise.
15422         (sync_sub<mode>): Likewise.
15423         (sync_old_add<mode>): Likewise.
15424         (sync_old_sub<mode>): Likewise.
15425         (sync_new_add<mode>): Likewise.
15426         (sync_new_sub<mode>): Likewise.
15427         (sync_<optab><mode>): Likewise.
15428         (sync_old_<optab><mode>): Likewise.
15429         (sync_new_<optab><mode>): Likewise.
15430         (sync_nand<mode>): Likewise.
15431         (sync_old_nand<mode>): Likewise.
15432         (sync_new_nand<mode>): Likewise.
15433         (sync_lock_test_and_set<mode>): Likewise.
15434         (test_and_set_12): Likewise.
15435         (atomic_compare_and_swap<mode>): Likewise.
15436         (atomic_exchange<mode>_llsc): Likewise.
15437         (atomic_fetch_add<mode>_llsc): Likewise.
15438         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15439         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15440         (umips_save_restore_pattern_p): Likewise.
15441         (umips_load_store_pair_p): Likewise.
15442         (umips_output_load_store_pair): Likewise.
15443         (umips_movep_target_p): Likewise.
15444         (umips_12bit_offset_address_p): Likewise.
15445         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15446         (mips_base_mips16): Rename this...
15447         (mips_base_compression_flags): ...to this. Update all uses.
15448         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15449         (mips_mips16_decl_p): Delete.
15450         (mips_nomips16_decl_p): Delete.
15451         (mips_get_compress_on_flags): New function.
15452         (mips_get_compress_off_flags): New function.
15453         (mips_get_compress_mode): New function.
15454         (mips_get_compress_on_name): New function.
15455         (mips_get_compress_off_name): New function.
15456         (mips_insert_attributes): Support multiple compression types.
15457         (mips_merge_decl_attributes): Likewise.
15458         (umips_12bit_offset_address_p): New function.
15459         (mips_start_function_definition): Emit .set micromips directive.
15460         (mips_call_may_need_jalx_p): New function.
15461         (mips_function_ok_for_sibcall): Add microMIPS support.
15462         (mips_print_operand_punctuation): Support short delay slots and
15463         compact jumps.
15464         (umips_swm_mask, umips_swm_encoding): New.
15465         (umips_build_save_restore): New function.
15466         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15467         (was_mips16_p): Remove.
15468         (old_compression_mode): New.
15469         (mips_set_compression_mode): New function.
15470         (mips_set_current_function): Add microMIPS support.
15471         (mips_option_override): Likewise.
15472         (umips_save_restore_pattern_p): New function.
15473         (umips_output_save_restore): New function.
15474         (umips_load_store_pair_p_1): New function.
15475         (umips_load_store_pair_p): New function.
15476         (umips_output_load_store_pair_1): New function.
15477         (umips_output_load_store_pair): New function.
15478         (umips_movep_target_p) New function.
15479         (mips_prepare_pch_save): Add microMIPS support.
15480         * config/mips/mips.h (TARGET_COMPRESSION): New.
15481         (TARGET_CPU_CPP_BUILTINS): Update macro
15482         to use new compression flags and to support microMIPS.
15483         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15484         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15485         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15486         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15487         (ASM_SPEC): Support mmicromips and mno-micromips.
15488         (M16STORE_REG_P): New macro.
15489         (MIPS_CALL): Support TARGET_MICROMIPS.
15490         (MICROMIPS_J): New macro.
15491         (mips_base_mips16): Rename this...
15492         (mips_base_compression_flags): ...to this.
15493         (UMIPS_12BIT_OFFSET_P): New macro.
15494         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15495         (MULTILIB_DIRNAMES): Likewise.
15496 2013-03-20  Richard Biener  <rguenther@suse.de>
15498         PR tree-optimization/56661
15499         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15500         the result does not have to be distinct.
15502 2013-03-20  Richard Biener  <rguenther@suse.de>
15504         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15505         remap_gimple_op_r.
15507 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15508             Steven Bosscher  <steven@gcc.gnu.org>
15510         PR rtl-optimization/56605
15511         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15513 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15515         PR bootstrap/56656
15516         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15517         that require movd instead of movq.
15519 2013-03-20  Richard Biener  <rguenther@suse.de>
15521         * tree-ssa-structalias.c (struct variable_info): Add pointer
15522         to the first field of an aggregate with sub-vars.  Make
15523         this and the pointer to the next subfield its ID.
15524         (vi_next): New function.
15525         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15526         storedanything_id, integer_id): Increment by one.
15527         (new_var_info, get_call_vi, lookup_call_clobber_vi,
15528         get_call_clobber_vi): Adjust.
15529         (solution_set_expand): Simplify and speedup.
15530         (solution_set_add): Inline into ...
15531         (set_union_with_increment): ... this.  Adjust accordingly.
15532         (do_sd_constraint): Likewise.
15533         (do_ds_constraint): Likewise.
15534         (do_complex_constraint): Simplify.
15535         (build_pred_graph): Adjust.
15536         (solve_graph): Likewise.  Simplify and speedup.
15537         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15538         get_constraint_for_component_ref, get_constraint_for_1,
15539         first_vi_for_offset, first_or_preceding_vi_for_offset,
15540         create_function_info_for, create_variable_info_for_1,
15541         create_variable_info_for, intra_create_variable_infos): Adjust.
15542         (init_base_vars): Push NULL for ID zero.
15543         (compute_points_to_sets): Adjust.
15545 2013-03-20  Richard Biener  <rguenther@suse.de>
15547         * cfgloop.c (verify_loop_structure): Streamline and avoid
15548         ICEing on corrupt loop tree.
15549         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15550         loop tree.
15552 2013-03-20  Richard Biener  <rguenther@suse.de>
15554         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15555         check whether an SSA update is needed.
15557 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
15559         * config/mips/constraints.md (T): Rename to...
15560         (Yf): ...this.
15561         (U): Rename to...
15562         (Yd): ...this.
15563         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15564         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15566 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15568         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15569         (*subsi3_carryin_uxtw): Likewise.
15571 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15573         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15574         (*rorsi3_insn_uxtw): Likewise.
15576 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15578         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15579         (*extrsi5_insn_uxtw): Likewise.
15581 2013-03-19  Richard Biener  <rguenther@suse.de>
15583         PR tree-optimization/56273
15584         * passes.c (init_optimization_passes): Move second VRP after DOM.
15586 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
15588         * config/i386/i386.md (*movti_internal): Merge from
15589         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
15590         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
15591         nox64 isa attributes.
15593 2013-03-18  Richard Biener  <rguenther@suse.de>
15595         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15596         (unite): Likewise.
15597         (merge_node_constraints): Likewise.
15598         (build_succ_graph): Likewise.
15599         (valid_graph_edge): Inline into single caller.
15600         (unify_nodes): Likewise.  Use bitmap_set_bit return value
15601         and cache varinfo.
15602         (scc_visit): Fix formatting and variable use.
15603         (do_sd_constraint): Use gcc_checking_assert.
15604         (do_ds_constraint): Likewise.
15605         (do_complex_constraint): Likewise.
15606         (condense_visit): Likewise.  Cleanup.
15607         (dump_pred_graph): New function.
15608         (perform_var_substitution): Dump the pred-graph before
15609         variable substitution.
15610         (find_equivalent_node): Use gcc_checking_assert.
15611         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15613 2013-03-18  Richard Biener  <rguenther@suse.de>
15615         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15616         Remove cond_expr_stmt_list argument and do not gimplify the
15617         built expression.
15618         (vect_loop_versioning): Adjust.
15619         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15620         Cleanup to use less temporaries.
15621         (vect_create_data_ref_ptr): Cleanup.
15623 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15625         PR tree-optimization/56635
15626         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15627         require types_compatible_p types.
15629 2013-03-18  Nick Clifton  <nickc@redhat.com>
15631         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15632         spurious backslash.
15634         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15635         Add missing line to comment describing function.
15637 2013-03-18  Richard Biener  <rguenther@suse.de>
15639         PR tree-optimization/56210
15640         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15641         Handle string / character search functions.
15642         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15644 2013-03-18  Richard Biener  <rguenther@suse.de>
15646         PR middle-end/56483
15647         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15648         and implement properly.
15649         * gimple.h (gimple_cond_single_var_p): Remove.
15651 2013-03-18  Richard Biener  <rguenther@suse.de>
15653         * tree-data-ref.h (find_data_references_in_loop): Declare.
15654         * tree-data-ref.c (get_references_in_stmt): Use a stack
15655         vector pre-allocated in the callers.
15656         (find_data_references_in_stmt): Adjust.
15657         (graphite_find_data_references_in_stmt): Likewise.
15658         (create_rdg_vertices): Likewise.
15659         (find_data_references_in_loop): Export.
15660         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15661         Compute dependences here...
15662         (vect_analyze_data_refs): ...not here.  When we encounter
15663         a non-vectorizable data reference in basic-block vectorization
15664         truncate the data reference vector.  Do not bother to
15665         fixup data-dependence information for gather loads.
15666         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15667         of data references, as reported.
15669 2013-03-18  Richard Biener  <rguenther@suse.de>
15671         PR tree-optimization/3713
15672         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
15673         has_constants and expr.
15674         (stmt_has_constants): Properly valueize SSA names when deciding
15675         whether the stmt has constants.
15677 2013-03-18  Richard Biener  <rguenther@suse.de>
15679         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15680         whole function when there is nothing to do.
15681         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15682         * tree-vectorizer.c (vectorize_loops): Update virtual and
15683         loop-closed SSA once.
15684         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15686 2013-03-18  Richard Biener  <rguenther@suse.de>
15688         PR middle-end/56113
15689         * domwalk.c (bb_postorder): New global static.
15690         (cmp_bb_postorder): New function.
15691         (walk_dominator_tree): Replace scheme imposing an order for
15692         visiting dominator sons by one sorting them at the time they
15693         are pushed on the stack.
15695 2013-03-18  Richard Biener  <rguenther@suse.de>
15697         PR tree-optimization/39326
15698         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15699         (struct mem_ref): Replace mem member with ao_ref typed member.
15700         (MEM_ANALYZABLE): Adjust.
15701         (memref_eq): Likewise.
15702         (mem_ref_alloc): Likewise.
15703         (gather_mem_refs_stmt): Likewise.
15704         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15705         (execute_sm_if_changed_flag_set): Adjust.
15706         (execute_sm): Likewise.
15707         (ref_always_accessed_p): Likewise.
15708         (refs_independent_p): Likewise.
15709         (can_sm_ref_p): Likewise.
15711 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15713         PR c/56566
15714         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15715         return 1 even for !unsignedp.
15717 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15719         * config/i386/i386.md (isa): Add x64 and nox64.
15720         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15721         (*pushtf): Enable *roF alternative for x64 isa only.
15722         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
15723         mode attribute of integer alternatives to DImode for TARGET_64BIT.
15724         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
15725         (*movtf_internal): Merge from *movtf_internal_rex64 and
15726         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
15727         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
15728         nox64 isa attributes.
15729         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
15730         nox64 isa attributes.
15731         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15733 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15735         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15737 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
15739         PR target/56640
15740         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15741         class names.  Remove trailing comma after "ALL_REGS".
15743 2013-03-16  Jan Hubicka  <jh@suse.cz>
15745         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15746         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15747         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15748         of cgraph_get_create_node.
15749         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15751 2013-03-16  Jason Merrill  <jason@redhat.com>
15753         PR debug/49090
15754         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15755         with DW_AT_default_value.
15757 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
15759         * BASE-VER: Set to 4.9.0.
15761 2013-03-14  Andi Kleen  <ak@linux.intel.com>
15763         PR target/56619
15764         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15765         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15766         Document _x* TSX intrinsics.
15768 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
15769             David Holsgrove  <david.holsgrove@xilinx.com>
15771         * configure.ac: Add MicroBlaze TLS support detection.
15772         * configure: Regenerate.
15773         * config/microblaze/microblaze-protos.h
15774         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15775         symbol_mentioned_p, label_mentioned_p): Add prototypes.
15776         * config/microblaze/microblaze.c (microblaze_address_type): Add
15777         ADDRESS_TLS and tls_reloc address types.
15778         (microblaze_address_info): Add tls_reloc.
15779         (TARGET_HAVE_TLS): Define.
15780         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15781          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15782          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15783         load_tls_operand,  microblaze_call_tls_get_addr,
15784         microblaze_legitimize_tls_address): New functions.
15785         (microblaze_classify_unspec): Handle UNSPEC_TLS.
15786         (get_base_reg): Use microblaze_tls_symbol_p.
15787         (microblaze_classify_address): Handle TLS.
15788         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15789         label_mentioned_p and microblaze_tls_referenced_p.
15790         (microblaze_legitimize_address): Handle TLS.
15791         (microblaze_address_insns): Handle ADDRESS_TLS.
15792         (pic_address_needs_scratch): Handle TLS.
15793         (print_operand_address): Handle TLS.
15794         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15795         (microblaze_expand_move): Handle TLS.
15796         (microblaze_legitimate_constant_p): Check
15797         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15798         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15799         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15800         (PIC_OFFSET_TABLE_REGNUM): Set.
15801         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15802         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15803         (addsi3, movsi_internal2, movdf_internal): Update constraints
15804         * config/microblaze/predicates.md (arith_plus_operand): Define
15805         (move_operand): Redefine as move_src_operand,
15806         check microblaze_tls_referenced_p.
15808 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15810         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15811         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15813 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15815         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15816         CC mode for AND.
15818 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
15820         PR tree-optimization/53265
15821         * common.opt (Waggressive-loop-optimizations): New option.
15822         * tree-ssa-loop-niter.c: Include tree-pass.h.
15823         (do_warn_aggressive_loop_optimizations): New function.
15824         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
15825         if number_of_latch_executions returned constant.
15826         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15827         early.  If number_of_latch_executions returned constant, set
15828         nb_iterations_upper_bound back to it.
15829         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15830         field.
15831         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15832         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15834         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15835         (MULTILIB_OSDIRNAMES): Set.
15836         * genmultilib: If defaultosdirname doesn't start with :: , set
15837         defaultosdirname2 instead, clear it and emit two . multilib_raw
15838         entries instead of just one.
15840 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15842         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15843         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15844         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15845         (SUBTARGET_OVERRIDE_OPTIONS): New.
15847 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15849         PR target/49880
15850         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15851         (musermode): Convert to Var(TARGET_USERMODE).
15852         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15853         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15854         * config/sh/sh.c (sh_option_override): Use
15855         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15856         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15857         condition.
15858         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15859         TARGET_SH4.
15860         (udivsi3_i4_single, divsi3_i4_single): Use
15861         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15863 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15865         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15866         default setting.
15868 2013-03-13  Richard Biener  <rguenther@suse.de>
15870         PR tree-optimization/56608
15871         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15872         calls when vectorizing basic-blocks.
15874 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15876         PR plugins/45078
15877         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15878         tm_file.
15880 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15882         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15884 2013-03-11  Jan Hubicka  <jh@suse.cz>
15886         PR lto/56557
15887         * lto-streamer-out.c (output_symbol_p): Skip references from
15888         constructors of external variables.
15890 2013-03-11  Jan Hubicka  <jh@suse.cz>
15892         PR middle-end/56571
15893         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15894         from pseudos.
15895         * emit-rtl.c (verify_rtx_sharing): Likewise.
15896         (copy_insn_1): Likewise.
15897         * rtl.c (copy_rtx): Likewise.
15899 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15901         PR target/56591
15902         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15903         output_operand_lossage message.
15905 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15907         PR target/56470
15908         * arm.c (shift_op): Validate RTL pattern on the fly.
15909         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15910         the RTL.
15912 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15914         PR target/56347
15915         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15916         new call patterns that clobber %fr12.
15917         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15918         split and postreload patterns.
15919         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15920         registers %fr12 and %fr12R as call used.
15922 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15924         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15925         (canon_address, record_store, replace_read, check_mem_read_rtx,
15926         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15927         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15928         rest_of_handle_dse): Likewise.
15930 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15932         PR middle-end/56524
15933         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15934         Add base_optabs.
15935         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15936         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15937         (save_optabs_if_changed): Replace with...
15938         (init_tree_optimization_optabs): ...this.
15939         * optabs.c (save_optabs_if_changed): Rename to...
15940         (init_tree_optimization_optabs): ...this.  Take the optimization node
15941         as argument.  Do nothing if the base optabs are already correct.
15942         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15943         to recompute optabs.
15944         * function.h (function): Remove optabs field.
15945         * function.c (invoke_set_current_function_hook): Call
15946         init_tree_optimization_optabs.  Use the result to initialize
15947         this_fn_optabs.
15949 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15951         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15952         if GTMA_HAS_NO_INSTRUMENTATION.
15953         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15954         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15955         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15956         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15957         GTMA_HAS_NO_INSTRUMENTATION.
15959 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15961         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15962         libasan_preinit.o.
15964 2013-03-08  Marek Polacek  <polacek@redhat.com>
15965             Jakub Jelinek  <jakub@redhat.com>
15967         PR tree-optimization/56478
15968         * predict.c (is_comparison_with_loop_invariant_p): Change the
15969         type of loop_step to tree.
15970         (predict_loops): Adjust.
15971         (predict_iv_comparison): Perform the computations on double_ints.
15973 2013-03-08  Richard Biener  <rguenther@suse.de>
15975         PR tree-optimization/56570
15976         * tree-cfg.c (verify_expr_location_1): Verify locations for
15977         DECL_DEBUG_EXPR.
15978         * tree-sra.c (create_access_replacement): Strip locations
15979         from DECL_DEBUG_EXPRs.
15981 2013-03-08  Richard Biener  <rguenther@suse.de>
15983         * tree-inline.c (expand_call_inline): Do not associate
15984         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15985         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15987 2013-03-08  Richard Biener  <rguenther@suse.de>
15989         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15990         or block changes with -Og.  Fix for location / block encoding
15991         changes and PHI arguments with locations.
15993 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15995         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15996         for all counters.
15997         (struct output_info): Likewise.
15998         (register_overhead): Remove bad gcc_assert.
15999         (bitmap_find_bit): If there is only a single bitmap element, do not
16000         count a miss as a search.
16001         (print_statistics): Update for counter type changes.
16002         (dump_bitmap_statistics): Likewise.  Print headers such that they
16003         are properly lined up with the printed counters.
16005 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
16007         PR tree-optimization/56559
16008         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
16009         check that it has only a single use.
16011 2013-03-07  Richard Biener  <rguenther@suse.de>
16013         * doc/invoke.texi (fwhole-program): Discourage use in combination
16014         with -flto.
16016 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
16018         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
16020         PR tree-optimization/56539
16021         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
16022         instead of GSI_CONTINUE_LINKING as last argument to
16023         force_gimple_operand_gsi.  Adjust function comment.
16025         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
16026         aarch64-cores.def.
16028         PR middle-end/56548
16029         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
16030         promoted mode, convert the result back to the original mode.
16032 2013-03-06  Richard Biener  <rguenther@suse.de>
16034         PR middle-end/56294
16035         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
16036         (insert_updated_phi_nodes_compare_uids): New function.
16037         (update_ssa): Sort symbols_to_rename after UID before
16038         traversing it to insert PHI nodes.
16040 2013-03-06  Richard Biener  <rguenther@suse.de>
16042         PR middle-end/50494
16043         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
16044         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
16046         Revert
16047         2013-02-13  Richard Biener  <rguenther@suse.de>
16049         PR lto/50494
16050         * varasm.c (output_constant_def_1): Get the decl representing
16051         the constant as argument.
16052         (output_constant_def): Wrap output_constant_def_1.
16053         (make_decl_rtl): Use output_constant_def_1 with the decl
16054         representing the constant.
16055         (build_constant_desc): Optionally re-use a decl already
16056         representing the constant.
16057         (tree_output_constant_def): Adjust.
16059 2013-03-06  Joey Ye  <joey.ye@arm.com>
16061         PR lto/50293
16062         * gcc.c (convert_white_space): New function.
16063         (main): Handles white space in function name.
16065 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
16067         PR target/56529
16068         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
16069         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
16070         to SH_DIV_CALL_TABLE for TARGET_SH2.
16071         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
16072         list.
16073         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
16074         call-table options.
16076 2013-03-05  Sterling Augustine  <saugustine@google.com>
16077             Cary Coutant  <ccoutant@google.com>
16079         PR debug/55364
16080         * dwarf2out.c (resolve_addr): Don't call
16081         remove_loc_list_addr_table_entries a second time for the same
16082         expression.
16084 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16086         PR debug/56510
16087         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
16088         (avoid_complex_debug_insns): New function.
16089         (expand_debug_locations): Call it.
16091         PR rtl-optimization/56484
16092         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16093         lifetimes of hard registers on small register class machines.
16095 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16097         * config/microblaze/microblaze-protos.h: Rename
16098         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16099         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16100         fast_interrupt.
16101         (microblaze_fast_interrupt_function_p): New function.
16102         (microblaze_is_interrupt_handler): Rename to
16103         microblaze_is_interrupt_variant and add fast_interrupt check.
16104         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16105         (save_restore_insns): Likewise.
16106         (compute_frame_size): Likewise.
16107         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16108         (microblaze_globalize_label): Likewise.
16109         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16110         * config/microblaze/microblaze.md: Use wrapper
16111         microblaze_is_interrupt_variant.
16113 2013-03-05  Kai Tietz  <ktietz@redhat.com>
16115         * sdbout.c (sdbout_one_type): Switch to current function's section
16116         supporting cold/hot.
16118 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16120         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16121         -mxl-reorder.
16123 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16125         PR middle-end/56461
16126         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16127         if VALGRIND_GET_VBITS is defined, temporarily make object
16128         memory all defined, and restore previous valgrind addressability
16129         and definability afterwards.  Free this_object at the end.
16131         PR middle-end/56461
16132         * lra.c (lra): Call lra_clear_live_ranges if live_p,
16133         right before calling lra_create_live_ranges, also call it
16134         when clearing live_p.  Only call lra_clear_live_ranges
16135         at the end if live_p.
16137         PR middle-end/56461
16138         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16140 2013-03-05  Richard Biener  <rguenther@suse.de>
16142         PR tree-optimization/56521
16143         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16144         value-id.
16146 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
16148         PR c++/55135
16149         * except.h (remove_unreachable_eh_regions): New prototype.
16150         * except.c (remove_eh_handler_splicer): New function, split out
16151         of remove_eh_handler.
16152         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
16153         warning about running it on many EH regions one at a time.
16154         (remove_unreachable_eh_regions_worker): New function, walk the
16155         EH tree in depth-first order and remove non-marked regions.
16156         (remove_unreachable_eh_regions): New function.
16157         * tree-eh.c (mark_reachable_handlers): New function, split out
16158         from remove_unreachable_handlers.
16159         (remove_unreachable_handlers): Use mark_reachable_handlers and
16160         remove_unreachable_eh_regions.
16161         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16162         and remove_unreachable_eh_regions.
16164 2013-03-05  Richard Biener  <rguenther@suse.de>
16166         PR middle-end/56525
16167         * loop-init.c (fix_loop_structure): Remove loops in two stages,
16168         not freeing them until the end.
16170 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16172         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16174 2013-03-05  Richard Biener  <rguenther@suse.de>
16176         PR tree-optimization/56270
16177         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16178         of loads after scheduling an SLP instance.
16180 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16182         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16183         tic6x.exp.
16184         (check_gcc_parallelize): Run guality.exp as a separate job from
16185         vect.exp with unsorted.exp and $(dg_target_exps) separately from
16186         struct-layout-1.exp with stackalign.exp.
16188         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16190         PR middle-end/56461
16191         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16192         load_index sbitmap even if some bit in it isn't set.
16194         PR middle-end/56461
16195         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16196         (discover_iteration_bound_by_body_walk): Change queues to
16197         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
16198         spelling in comment.  Call safe_push on queues[bound_index] directly.
16199         Release queues[queue_index] in every iteration unconditionally.
16200         Release bounds vector.
16202         PR middle-end/56461
16203         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16204         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16205         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16206         inner_phis vector.
16208 2013-03-05  Richard Biener  <rguenther@suse.de>
16210         PR lto/56515
16211         * tree-inline.c (remap_blocks_to_null): New function.
16212         (expand_call_inline): When expanding a call stmt without
16213         an associated block inline remap all callee blocks to NULL.
16215 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16217         PR rtl-optimization/56494
16218         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16219         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16220         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16222         PR middle-end/56461
16223         * sel-sched-ir.c (free_sched_pools): Release
16224         succs_info_pool.stack[succs_info_pool.max_top] vectors too
16225         if succs_info_pool.max_top isn't -1.
16227         PR bootstrap/56509
16228         * opts.c (opts_obstack, opts_concat): Moved to...
16229         * opts-common.c (opts_obstack, opts_concat): ... here.
16231 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16233         PR middle-end/56461
16234         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16236 2013-03-04  Martin Jambor  <mjambor@suse.cz>
16238         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16239         all appropriate places.
16241 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
16243         PR tree-optimization/56424
16244         * ipa-split.c (split_function): Do not set the RSO flag if result is
16245         not by reference and its type is a register type.
16247 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
16249         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16250         (microblaze_legitimate_pic_operand): Likewise
16251         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16252         new function microblaze_legitimate_pic_operand
16253         * config/microblaze/microblaze-protos.h
16254         (microblaze_legitimate_pic_operand): Declare.
16256 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16258         * config/microblaze/predicates.md (call_insn_simple_operand):
16259         New predicate for supported rtx code types.
16260         * config/microblaze/microblaze.md (call_internal1): Use
16261         call_insn_simple_operand predicate.
16263 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16265         PR middle-end/56461
16266         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16267         partitions.ordered_remove.
16269         PR middle-end/56461
16270         * tree-vect-stmts.c (vectorizable_conversion): Don't call
16271         vec_oprnds0.create (1) for modifier == NONE.
16273         PR middle-end/56461
16274         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16275         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16276         vec_oprnds1 right before pushing anything to it for
16277         scalar_shift_arg.
16279         PR middle-end/56461
16280         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16281         set nbbs to 0 instead of having separate code path.
16282         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16283         instead of false as last argument if returning NULL.
16285 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
16287         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16288         the attribute is now called "target" instead of "option".
16289         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16290         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
16291         attribute/pragma name for TARGET_OPTION_VALID_P and
16292         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
16293         * doc/tm.texi: Regenerated.
16295 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
16297         * config/microblaze/microblaze.c:
16298         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16299         * config/microblaze/microblaze.h: Add -mxl-reorder to
16300         DRIVER_SELF_SPECS.
16301         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16302         instructions emitted if TARGET_REORDER.
16303         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16304         or 0 for -m/-mno case, but initialises as 2 to detect default use case
16305         separately.
16307 2013-03-01  Xinliang David Li  <davidxl@google.com>
16309         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16310         walk length.
16312 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
16314         PR middle-end/56461
16315         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16316         vector even when returning true.  Fix up function comment formatting.
16318         PR middle-end/56461
16319         * ira-build.c (ira_loop_nodes_count): New variable.
16320         (create_loop_tree_nodes): Initialize it.
16321         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16323         PR middle-end/56461
16324         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16325         method on dr_chain and result_chain.
16326         * tree-vect-stmts.c (vectorizable_store): Only call
16327         result_chain.create if j == 0.
16329         PR middle-end/56461
16330         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16331         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16332         before overwriting it.
16334 2013-03-01  Tobias Burnus  <burnus@net-b.de>
16336         * doc/extended.texi (C Extensions): Change order in @menu
16337         to match @node.
16338         (Other MIPS Built-in Functions): Move last MIPS entry before
16339         "picoChip Built-in Functions".
16340         (SH Built-in Functions): Move after RX Built-in Functions.
16341         * doc/gcc.texi (Introduction): Change order in @menu
16342         to match @node.
16343         * doc/md.texi (Constraints): Ditto.
16344         * gty.texi (Type Information): Ditto.
16345         (User-provided marking routines for template types): Make
16346         subsection.
16347         * doc/invoke.texi (AArch64 Options): Move before
16348         "Adapteva Epiphany Options".
16350 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
16351             Jakub Jelinek  <jakub@redhat.com>
16353         PR sanitizer/56454
16354         * asan.c (gate_asan): Lookup no_sanitize_address instead of
16355         no_address_safety_analysis attribute.
16356         * doc/extend.texi (no_address_safety_attribute): Rename to
16357         no_sanitize_address attribute, mention no_address_safety_analysis
16358         attribute as deprecated alias.
16360 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
16362         PR middle-end/56461
16363         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16364         type to vec<vec<tree> > *.
16365         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
16366         to be vec<tree> instead of vec<tree> *, set vec_defs
16367         to vNULL and call vec_defs.create (number_of_vects), adjust other
16368         uses of vec_defs.
16369         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16370         vectorizable_condition): Adjust vect_get_slp_defs callers.
16372 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16374         * config/aarch64/aarch64.c
16375         (aarch64_float_const_representable): Remove unused variable.
16377 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16379         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16381 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16383         * config/aarch64/aarch64-builtins.c
16384         (aarch64_init_simd_builtins): Make static.
16386 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
16388         * config/aarch64/aarch64.c
16389         (aarch64_simd_make_constant): Make static.
16391 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16393         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16394         with no initialization to the RHS of debug statements.
16396 2013-02-28  Martin Jambor  <mjambor@suse.cz>
16398         PR tree-optimization/56294
16399         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16400         Adjust dumping.
16401         (get_access_replacement): Do not call create_access_replacement.
16402         Assert a replacement exists.
16403         (get_repl_default_def_ssa_name): Create the replacement declaration
16404         itself.
16406 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16408         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16409         final_end_function.
16411 2013-02-28  Marek Polacek  <polacek@redhat.com>
16413         PR rtl-optimization/56466
16414         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16415         if we're changing a loop.
16416         (peel_loops_completely): Likewise.
16418 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
16420         PR c++/55813
16421         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16423 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16425         PR target/56445
16426         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16427         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16428         INTX_FTYPE_FX, FX_FTYPE_INTX.
16429         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16431 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16433         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16434         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16435         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16436         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16437         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16438         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16439         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16440         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16441         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16442         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16443         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16444         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16445         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16446         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16447         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16448         (avrxmega6): Increase max flash segments from 5 to 6.
16449         * config/avr/t-multilib: Regenerate.
16450         * config/avr/avr-tables.opt: Regenerate.
16451         * doc/avr-mmcu.texi: Regenerate.
16453 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16455         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16456         (avr_device_to_arch): Rename to avr_device_to_ld.
16457         (avr_device_to_as): New prototype.
16458         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16459         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16460         * config/avr/driver-avr.c (avr_device_to_as): New.
16461         (avr_device_to_arch): Rename to avr_device_to_ld.
16463 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16465         PR middle-end/56461
16466         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16467         method on dr_chain and result_chain.
16469         PR middle-end/56461
16470         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16471         pointer_set_destroy on not_executed_last_iteration.
16473         PR middle-end/56461
16474         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16476         PR middle-end/56461
16477         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16478         FOR_EACH_DEFINED_FUNCTION when freeing state.
16480         PR middle-end/56461
16481         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16482         pool_free.
16483         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16484         overwriting it.
16486         PR middle-end/56461
16487         * ipa-cp.c (decide_whether_version_node): Call vec_free on
16488         known_aggs[i].items and release known_aggs vector.
16490         PR middle-end/56461
16491         * ipa-reference.c (propagate): Free node_info even for alias nodes.
16493 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16495         * config/microblaze/microblaze.c (microblaze_emit_compare):
16496         Use xor for EQ/NE comparisions.
16497         * config/microblaze/microblaze.md (cstoresf4): Add constraints
16498         (cbranchsf4): Adjust operator to comparison_operator.
16500 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16502         PR middle-end/56461
16503         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16504         vector.
16505         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16506         vec_safe_push, always update *slot.
16507         (redirect_edge_var_map_clear): Use vec_free.
16508         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16509         (free_var_map_entry): Use vec_free.
16510         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16511         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16513 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
16515         PR middle-end/45472
16516         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16517         when the may_trap_p bit of the exprs being merged differs.
16518         Reorder tests for speculativeness in the logical and operator.
16520 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16522         * incpath.c (add_standard_paths): Use reconcat instead of concat
16523         where appropriate and avoid leaking memory.
16525         * opts.h: Include obstack.h.
16526         (opts_concat): New prototype.
16527         (opts_obstack): New declaration.
16528         * opts.c (opts_concat): New function.
16529         (opts_obstack): New variable.
16530         (init_options_struct): Call gcc_init_obstack on opts_obstack.
16531         (finish_options): Use opts_concat instead of concat
16532         and XOBNEWVEC instead of XNEWVEC.
16533         * opts-common.c (generate_canonical_option, decode_cmdline_option,
16534         generate_option): Likewise.
16535         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16536         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16538         PR target/56455
16539         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16540         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16542 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16544         PR middle-end/56461
16545         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16547 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16549         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16550         (arm_block_move_unaligned_straight): Likewise.
16551         (arm_adjust_block_mem): Likewise.
16553 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16555         PR target/48901
16556         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16557         temp, cond and label.
16558         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16560         PR target/52500
16561         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16562         * config/c6x/c6x.h (dbx_register_map): Update declaration.
16564         PR target/52501
16565         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16566         of prologue/epilogue functions.
16568         PR target/52550
16569         * config/tilegx/tilegx.c (tilegx_expand_prologue):
16570         Remove unused variable cfa_offset.
16571         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16573         PR target/54639
16574         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16575         type promotion to unsigned.
16577         PR target/54640
16578         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16579         for HOST_WIDE_INT of 32 bit / same size as int.
16580         (arm_block_move_unaligned_straight): Likewise.
16581         (arm_adjust_block_mem): Likewise.
16583         PR target/54662
16584         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16585         ALL_CFLAGS.
16587 2013-02-26  Marek Polacek  <polacek@redhat.com>
16589         PR tree-optimization/56426
16590         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16592 2013-02-26  Richard Biener  <rguenther@suse.de>
16594         PR target/56444
16595         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16596         unused variable loops.
16598 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16600         PR tree-optimization/56448
16601         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16602         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16603         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16604         later operands of the references, or even first operand for
16605         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16607         PR tree-optimization/56443
16608         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16609         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16610         to type_for_mode langhook.
16612 2013-02-25  Matt Turner  <mattst88@gmail.com>
16614         * doc/invoke.texi: Document r4700.
16616 2013-02-25  Richard Biener  <rguenther@suse.de>
16618         PR tree-optimization/56175
16619         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16620         split out from ...
16621         (simplify_bitwise_binary): ... here.  Also guard the conversion
16622         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16624 2013-02-25  Catherine Moore  <clm@codesourcery.com>
16626         Revert:
16627         2013-02-24  Catherine Moore  <clm@codesourcery.com>
16628             Maciej W. Rozycki  <macro@codesourcery.com>
16629             Tom de Vries  <tom@codesourcery.com>
16630             Nathan Sidwell  <nathan@codesourcery.com>
16631             Iain Sandoe  <iain@codesourcery.com>
16632             Nathan Froyd  <froydnj@codesourcery.com>
16633             Chao-ying Fu  <fu@mips.com>
16635         * doc/extend.texi: (micromips, nomicromips, nocompression):
16636         Document new function attributes.
16637         * doc/invoke.texi (minterlink-compressed, mmicromips,
16638         m14k, m14ke, m14kec): Document new options.
16639         (minterlink-mips16): Update documentation.
16640         * doc/md.texi (ZC, ZD): Document new constraints.
16641         * configure.ac (gcc_cv_as_micromips): Check if linker
16642         supports the .set micromips directive.
16643         * configure: Regenerate.
16644         * config.in: Regenerate.
16645         * config/mips/mips-tables.opt: Regenerate.
16646         * config/mips/micromips.md: New file.
16647         * constraints.md (ZC, AD): New constraints.
16648         * config/mips/predicates.md (movep_src_register): New predicate.
16649         (movep_src_operand): New predicate.
16650         (non_volatile_mem_operand): New predicate.
16651         * config/mips/mips.md (multimem): New type.
16652         (length): Differentiate between 17-bit and 18-bit branch offsets.
16653         (MOVEP1, MOVEP2): New mode iterator.
16654         (mov_<load>l): Use ZC constraint.
16655         (mov_<load>r): Likewise.
16656         (mov_<store>l): Likewise.
16657         (mov_<store>r): Likewise.
16658         (*branch_equality<mode>_inverted): Add microMIPS support.
16659         (*branch_equality<mode>): Likewise.
16660         (*jump_absolute): Likewise.
16661         (indirect_jump_<mode>): Likewise.
16662         (tablejump_<mode>): Likewise.
16663         (<optab>_internal): Likewise.
16664         (sibcall_internal): Likewise.
16665         (sibcall_value_internal): Likewise.
16666         (prefetch): Use constraint ZD.
16667         * config/mips/mips.opt (minterlink-compressed): New option.
16668         (minterlink-mips16): Now an alias for minterlink-compressed.
16669         (mmicromips): New option.
16670         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16671         (compare_and_swap_12): Likewise.
16672         (sync_add<mode>): Likewise.
16673         (sync_<optab>_12): Likewise.
16674         (sync_old_<optab>_12): Likewise.
16675         (sync_new_<optab>_12): Likewise.
16676         (sync_nand_12): Likewise.
16677         (sync_old_nand_12): Likewise.
16678         (sync_new_nand_12): Likewise.
16679         (sync_sub<mode>): Likewise.
16680         (sync_old_add<mode>): Likewise.
16681         (sync_old_sub<mode>): Likewise.
16682         (sync_new_add<mode>): Likewise.
16683         (sync_new_sub<mode>): Likewise.
16684         (sync_<optab><mode>): Likewise.
16685         (sync_old_<optab><mode>): Likewise.
16686         (sync_new_<optab><mode>): Likewise.
16687         (sync_nand<mode>): Likewise.
16688         (sync_old_nand<mode>): Likewise.
16689         (sync_new_nand<mode>): Likewise.
16690         (sync_lock_test_and_set<mode>): Likewise.
16691         (test_and_set_12): Likewise.
16692         (atomic_compare_and_swap<mode>): Likewise.
16693         (atomic_exchange<mode>_llsc): Likewise.
16694         (atomic_fetch_add<mode>_llsc): Likewise.
16695         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16696         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16697         (umips_save_restore_pattern_p): Likewise.
16698         (umips_load_store_pair_p): Likewise.
16699         (umips_output_load_store_pair): Likewise.
16700         (umips_movep_target_p): Likewise.
16701         (umips_12bit_offset_address_p): Likewise.
16702         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16703         (mips_base_mips16): Rename this...
16704         (mips_base_compression_flags): ...to this. Update all uses.
16705         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16706         (mips_mips16_decl_p): Delete.
16707         (mips_nomips16_decl_p): Delete.
16708         (mips_get_compress_on_flags): New function.
16709         (mips_get_compress_off_flags): New function.
16710         (mips_get_compress_mode): New function.
16711         (mips_get_compress_on_name): New function.
16712         (mips_get_compress_off_name): New function.
16713         (mips_insert_attributes): Support multiple compression types.
16714         (mips_merge_decl_attributes): Likewise.
16715         (umips_12bit_offset_address_p): New function.
16716         (mips_start_function_definition): Emit .set micromips directive.
16717         (mips_call_may_need_jalx_p): New function.
16718         (mips_function_ok_for_sibcall): Add microMIPS support.
16719         (mips_print_operand_punctuation): Support short delay slots and
16720         compact jumps.
16721         (umips_swm_mask, umips_swm_encoding): New.
16722         (umips_build_save_restore): New function.
16723         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16724         (was_mips16_p): Remove.
16725         (old_compression_mode): New.
16726         (mips_set_compression_mode): New function.
16727         (mips_set_current_function): Add microMIPS support.
16728         (mips_option_override): Likewise.
16729         (umips_save_restore_pattern_p): New function.
16730         (umips_output_save_restore): New function.
16731         (umips_load_store_pair_p_1): New function.
16732         (umips_load_store_pair_p): New function.
16733         (umips_output_load_store_pair_1): New function.
16734         (umips_output_load_store_pair): New function.
16735         (umips_movep_target_p) New function.
16736         (mips_prepare_pch_save): Add microMIPS support.
16737         * config/mips/mips.h (TARGET_COMPRESSION): New.
16738         (TARGET_CPU_CPP_BUILTINS): Update macro
16739         to use new compression flags and to support microMIPS.
16740         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16741         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16742         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16743         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16744         (ASM_SPEC): Support mmicromips and mno-micromips.
16745         (M16STORE_REG_P): New macro.
16746         (MIPS_CALL): Support TARGET_MICROMIPS.
16747         (MICROMIPS_J): New macro.
16748         (mips_base_mips16): Rename this...
16749         (mips_base_compression_flags): ...to this.
16750         (UMIPS_12BIT_OFFSET_P): New macro.
16751         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16752         (MULTILIB_DIRNAMES): Likewise.
16754 2013-02-25  Tom de Vries  <tom@codesourcery.com>
16756         PR rtl-optimization/56131
16757         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16758         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16759         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16761 2013-02-25  Tobias Burnus  <burnus@net-b.de>
16763         * doc/invoke.texi (-fsanitize=): Move from optimization
16764         to debugging options.
16766 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16768         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16770 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16771             Alexander Monakov  <amonakov@ispras.ru>
16773         PR middle-end/56077
16774         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16775         flush pending lists also on non-jumps.  Adjust comment.
16777 2013-02-24  Catherine Moore  <clm@codesourcery.com>
16778             Maciej W. Rozycki  <macro@codesourcery.com>
16779             Tom de Vries  <tom@codesourcery.com>
16780             Nathan Sidwell  <nathan@codesourcery.com>
16781             Iain Sandoe  <iain@codesourcery.com>
16782             Nathan Froyd  <froydnj@codesourcery.com>
16783             Chao-ying Fu  <fu@mips.com>
16785         * doc/extend.texi: (micromips, nomicromips, nocompression):
16786         Document new function attributes.
16787         * doc/invoke.texi (minterlink-compressed, mmicromips,
16788         m14k, m14ke, m14kec): Document new options.
16789         (minterlink-mips16): Update documentation.
16790         * doc/md.texi (ZC, ZD): Document new constraints.
16791         * configure.ac (gcc_cv_as_micromips): Check if linker
16792         supports the .set micromips directive.
16793         * configure: Regenerate.
16794         * config.in: Regenerate.
16795         * config/mips/mips-tables.opt: Regenerate.
16796         * config/mips/micromips.md: New file.
16797         * constraints.md (ZC, AD): New constraints.
16798         * config/mips/predicates.md (movep_src_register): New predicate.
16799         (movep_src_operand): New predicate.
16800         (non_volatile_mem_operand): New predicate.
16801         * config/mips/mips.md (multimem): New type.
16802         (length): Differentiate between 17-bit and 18-bit branch offsets.
16803         (MOVEP1, MOVEP2): New mode iterator.
16804         (mov_<load>l): Use ZC constraint.
16805         (mov_<load>r): Likewise.
16806         (mov_<store>l): Likewise.
16807         (mov_<store>r): Likewise.
16808         (*branch_equality<mode>_inverted): Add microMIPS support.
16809         (*branch_equality<mode>): Likewise.
16810         (*jump_absolute): Likewise.
16811         (indirect_jump_<mode>): Likewise.
16812         (tablejump_<mode>): Likewise.
16813         (<optab>_internal): Likewise.
16814         (sibcall_internal): Likewise.
16815         (sibcall_value_internal): Likewise.
16816         (prefetch): Use constraint ZD.
16817         * config/mips/mips.opt (minterlink-compressed): New option.
16818         (minterlink-mips16): Now an alias for minterlink-compressed.
16819         (mmicromips): New option.
16820         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16821         (compare_and_swap_12): Likewise.
16822         (sync_add<mode>): Likewise.
16823         (sync_<optab>_12): Likewise.
16824         (sync_old_<optab>_12): Likewise.
16825         (sync_new_<optab>_12): Likewise.
16826         (sync_nand_12): Likewise.
16827         (sync_old_nand_12): Likewise.
16828         (sync_new_nand_12): Likewise.
16829         (sync_sub<mode>): Likewise.
16830         (sync_old_add<mode>): Likewise.
16831         (sync_old_sub<mode>): Likewise.
16832         (sync_new_add<mode>): Likewise.
16833         (sync_new_sub<mode>): Likewise.
16834         (sync_<optab><mode>): Likewise.
16835         (sync_old_<optab><mode>): Likewise.
16836         (sync_new_<optab><mode>): Likewise.
16837         (sync_nand<mode>): Likewise.
16838         (sync_old_nand<mode>): Likewise.
16839         (sync_new_nand<mode>): Likewise.
16840         (sync_lock_test_and_set<mode>): Likewise.
16841         (test_and_set_12): Likewise.
16842         (atomic_compare_and_swap<mode>): Likewise.
16843         (atomic_exchange<mode>_llsc): Likewise.
16844         (atomic_fetch_add<mode>_llsc): Likewise.
16845         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16846         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16847         (umips_save_restore_pattern_p): Likewise.
16848         (umips_load_store_pair_p): Likewise.
16849         (umips_output_load_store_pair): Likewise.
16850         (umips_movep_target_p): Likewise.
16851         (umips_12bit_offset_address_p): Likewise.
16852         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16853         (mips_base_mips16): Rename this...
16854         (mips_base_compression_flags): ...to this. Update all uses.
16855         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16856         (mips_mips16_decl_p): Delete.
16857         (mips_nomips16_decl_p): Delete.
16858         (mips_get_compress_on_flags): New function.
16859         (mips_get_compress_off_flags): New function.
16860         (mips_get_compress_mode): New function.
16861         (mips_get_compress_on_name): New function.
16862         (mips_get_compress_off_name): New function.
16863         (mips_insert_attributes): Support multiple compression types.
16864         (mips_merge_decl_attributes): Likewise.
16865         (umips_12bit_offset_address_p): New function.
16866         (mips_start_function_definition): Emit .set micromips directive.
16867         (mips_call_may_need_jalx_p): New function.
16868         (mips_function_ok_for_sibcall): Add microMIPS support.
16869         (mips_print_operand_punctuation): Support short delay slots and
16870         compact jumps.
16871         (umips_swm_mask, umips_swm_encoding): New.
16872         (umips_build_save_restore): New function.
16873         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16874         (was_mips16_p): Remove.
16875         (old_compression_mode): New.
16876         (mips_set_compression_mode): New function.
16877         (mips_set_current_function): Add microMIPS support.
16878         (mips_option_override): Likewise.
16879         (umips_save_restore_pattern_p): New function.
16880         (umips_output_save_restore): New function.
16881         (umips_load_store_pair_p_1): New function.
16882         (umips_load_store_pair_p): New function.
16883         (umips_output_load_store_pair_1): New function.
16884         (umips_output_load_store_pair): New function.
16885         (umips_movep_target_p) New function.
16886         (mips_prepare_pch_save): Add microMIPS support.
16887         * config/mips/mips.h (TARGET_COMPRESSION): New.
16888         (TARGET_CPU_CPP_BUILTINS): Update macro
16889         to use new compression flags and to support microMIPS.
16890         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16891         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16892         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16893         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16894         (ASM_SPEC): Support mmicromips and mno-micromips.
16895         (M16STORE_REG_P): New macro.
16896         (MIPS_CALL): Support TARGET_MICROMIPS.
16897         (MICROMIPS_J): New macro.
16898         (mips_base_mips16): Rename this...
16899         (mips_base_compression_flags): ...to this.
16900         (UMIPS_12BIT_OFFSET_P): New macro.
16901         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16902         (MULTILIB_DIRNAMES): Likewise.
16904 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16906         PR target/52555
16907         * target-globals.c (save_target_globals): For init_reg_sets and
16908         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16910 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16912         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16913         * config/aarch64/t-aarch64
16914         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16916 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16918         PR inline-asm/56148
16919         * lra-constraints.c (process_alt_operands): Reload operand
16920         conflicting with earlier clobber only if no more other conflicting
16921         operands.
16923 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16925         PR sanitizer/56393
16926         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16927         if not linking a shared library.
16929 2013-02-22  Seth LaForge  <sethml@google.com>
16931         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16933 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16935         * config/arm/arm.md (split for extendsidi): Update condition.
16936         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16937         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16938         (qhs_zextenddi_cstr): Likewise.
16940 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16942         PR middle-end/56420
16943         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16944         avoid signed wrapping.
16945         (expand_mult): Handle properly multiplication by
16946         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16947         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16948         in the compiler if coeff is HOST_WIDE_INT_MIN.
16949         (expand_divmod): Don't make ext_op1 static, change it's type to
16950         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16952         PR rtl-optimization/50339
16953         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16954         field.
16955         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16956         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16957         into splitting_ashiftrt field.
16958         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16959         ASHIFTRT.
16960         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16961         choices.
16963 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16965         PR middle-end/56108
16966         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16967         are sure to go irrevocable.
16969 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16971         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16972         scalars are valid operands.
16974 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16976         PR tree-optimization/56310
16977         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16978         only matching indices and non-negative final offsets.
16979         (intersect_aggregates_with_edge): Pass src_idx to
16980         agg_replacements_to_vector.  Pass src_idx insstead of index to
16981         intersect_with_agg_replacements.
16983 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16985         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16986         instead of hard-wired defaults.
16988 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16990         * doc/invoke.texi (MIPS Options): Update documentation of the
16991         floating-point multiply-accumulate instruction restrictions.
16993 2013-02-21  Kostya Serebryany  <kcc@google.com>
16995         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16996         asan_shadow_offset on x86_64 linux.
16998 2013-02-21  Richard Biener  <rguenther@suse.de>
17000         PR tree-optimization/56415
17001         Revert
17002         2013-02-11  Richard Biener  <rguenther@suse.de>
17004         PR tree-optimization/56273
17005         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17006         first VRP run.
17008 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
17010         PR bootstrap/56258
17011         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
17012         instead of @itemx.
17014         PR inline-asm/56405
17015         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
17016         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
17018 2013-02-20  Jan Hubicka  <jh@suse.cz>
17020         PR tree-optimization/56265
17021         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
17022         when target is referenced for first time.
17024 2013-02-20  Richard Biener  <rguenther@suse.de>
17026         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
17027         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
17028         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
17029         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
17030         not return anything.
17031         (rename_ssa_copies): Do not remove unused locals.
17032         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
17033         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
17034         * passes.c (execute_function_todo): Do not schedule unused locals
17035         removal if cleanup_tree_cfg did something.
17036         * tree-ssa-live.c (remove_unused_locals): Dump statistics
17037         about the number of removed locals.
17039 2013-02-20  Richard Biener  <rguenther@suse.de>
17041         PR tree-optimization/56398
17042         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
17044 2013-02-20  Martin Jambor  <mjambor@suse.cz>
17046         PR tree-optimization/55334
17047         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
17048         restricted pointers to arrays.
17050 2013-02-20  Richard Biener  <rguenther@suse.de>
17051             Jakub Jelinek  <jakub@redhat.com>
17053         PR tree-optimization/56396
17054         * tree-ssa-ccp.c (n_const_val): New static variable.
17055         (get_value): Return NULL for SSA names we don't have a lattice
17056         entry for.
17057         (ccp_initialize): Initialize n_const_val.
17058         * tree-ssa-copy.c (n_copy_of): New static variable.
17059         (init_copy_prop): Initialize n_copy_of.
17060         (get_value): Return NULL_TREE for SSA names we don't have a
17061         lattice entry for.
17063 2013-02-20  Martin Jambor  <mjambor@suse.cz>
17065         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
17067 2013-02-20  Richard Biener  <rguenther@suse.de>
17069         * genpreds.c (write_lookup_constraint): Do not compare first
17070         letter of the constraint again.
17072 2013-02-20  Richard Biener  <rguenther@suse.de>
17074         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
17075         and ceil_log2.
17076         (get_use_iv_cost): Terminate hashtable walk when coming across
17077         an empty entry.
17079 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
17081         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
17082         reassociation for avx2 targets.
17084 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17086         * config/microblaze/microblaze.c: microblaze_has_clz = 0
17087         Add version check for v8.10.a to enable microblaze_has_clz
17088         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
17089         version and TARGET_PATTERN_COMPARE check
17090         * config/microblaze/microblaze.md: New clzsi2 instruction
17092 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17094         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17095         function before branching.
17097 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17099         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17100         DUMP_INSN_RTX_UID.
17101         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17103 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17105         PR middle-end/55889
17106         * sel-sched.c: Include ira.h.
17107         (implicit_clobber_conflict_p): New function.
17108         (moveup_expr): Use it.
17109         * Makefile.in (sel-sched.o): Depend on ira.h.
17111 2013-02-19  Richard Biener  <rguenther@suse.de>
17113         PR tree-optimization/56384
17114         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17115         (vn_hash_type): Split out from ...
17116         (vn_hash_constant_with_type): ... here.
17117         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17118         (vn_phi_eq): Compare types from vn_phi_s structure.
17119         (vn_phi_lookup): Populate vn_phi_s type.
17120         (vn_phi_insert): Likewise.
17122 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
17124         PR tree-optimization/56350
17125         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17126         if haven't found reduction or nested cycle operand, rather than
17127         asserting we must find it.
17129         PR tree-optimization/56381
17130         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17131         to fold_build3.
17133 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
17134             Jakub Jelinek  <jakub@redhat.com>
17136         PR target/52555
17137         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17138         (swap_optab_enable): Same.
17139         (init_all_optabs): Use argument instead of global.
17140         * tree.h (struct tree_optimization_option): New field target_optabs.
17141         * expr.h (init_all_optabs): Add argument to prototype.
17142         (TREE_OPTIMIZATION_OPTABS): New.
17143         (save_optabs_if_changed): Protoize.
17144         * optabs.h: Declare this_fn_optabs.
17145         * optabs.c (save_optabs_if_changed): New.
17146         Declare this_fn_optabs.
17147         (init_optabs): Add argument to init_all_optabs() call.
17148         * function.c (invoke_set_current_function_hook): Handle per
17149         function optabs.
17150         * function.h (struct function): New field optabs.
17151         * config/mips/mips.c (mips_set_mips16_mode): Handle when
17152         optimization_current_node has changed.
17153         * target-globals.h (save_target_globals_default_opts): Protoize.
17154         * target-globals.c (save_target_globals_default_opts): New.
17156 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17158         PR target/56347
17159         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17160         registers %fr12 and %fr12R as call used.
17162         PR target/56214
17163         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17164         and HImode, require all displacements to be an integer multiple of
17165         their mode size.
17166         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17167         only allow QImode and HImode when reload is in progress and strict is
17168         true.  Likewise for symbolic addresses.  Use base14_operand to check
17169         displacements in REG+BASE addresses.
17171 2013-02-18  Richard Biener  <rguenther@suse.de>
17173         PR tree-optimization/56366
17174         * tree-vect-loop.c (get_initial_def_for_induction): Properly
17175         handle sign-conversion of outer-loop initial induction value.
17177 2013-02-18  Richard Biener  <rguenther@suse.de>
17179         PR middle-end/56349
17180         * cfghooks.c (merge_blocks): If we merge a latch into another
17181         block adjust references to it.
17182         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17183         (verify_loop_structure): Verify that a recorded latch is in fact
17184         a latch.
17186 2013-02-18  Richard Biener  <rguenther@suse.de>
17188         PR tree-optimization/56321
17189         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17190         order SSA name release and virtual operand unlinking.
17192 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17194         * config/microblaze/microblaze.md (save_stack_block): Define.
17195         (restore_stack_block): Likewise.
17197 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17199         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17200         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17201         * config/microblaze/microblaze.c (microblaze_option_override):
17202         Bail out early for PIC modes when target does not support PIC.
17204 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17206         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17207         Replace with a microblaze version.
17208         (microblaze_trampoline_init): Adapt for microblaze.
17209         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17210         microblaze.
17212 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
17213             Dodji Seketeli  <dodji@redhat.com>
17215         PR asan/56330
17216         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17217         (instrument_mem_region_access): Do not forget to always put
17218         instrumentation of the of 'base' and 'base + len' in a "if (len !=
17219         0) statement, even for cases where either 'base' or 'base + len'
17220         are not instrumented -- because they have been previously
17221         instrumented.  Simplify the logic by putting all the statements
17222         instrument 'base + len' inside a sequence, and then insert that
17223         sequence right before the current insertion point.  Then, to
17224         instrument 'base + len', just get an iterator on that statement.
17225         And do not forget to update the pointer to iterator the function
17226         received as argument.
17228 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
17230         PR rtl-optimization/56348
17231         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17233 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
17235         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17236         (clean_graph_dump_file): Pass base to start_graph_dump.
17238 2013-02-14  Richard Henderson  <rth@redhat.com>
17240         PR target/55941
17241         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17243 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
17245         * collect2-aix.h: Define F_LOADONLY.
17247 2013-02-14  Richard Biener  <rguenther@suse.de>
17249         PR lto/50494
17250         * varasm.c (output_constant_def_1): Get the decl representing
17251         the constant as argument.
17252         (output_constant_def): Wrap output_constant_def_1.
17253         (make_decl_rtl): Use output_constant_def_1 with the decl
17254         representing the constant.
17255         (build_constant_desc): Optionally re-use a decl already
17256         representing the constant.
17257         (tree_output_constant_def): Adjust.
17259 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
17261         Fix an asan crash
17262         * asan.c (instrument_builtin_call):  Really put the length of the
17263         second source argument into src1_len.
17265 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17267         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17268         argument.  If it is false, don't create edge from then_bb to
17269         fallthru_bb.
17270         (insert_if_then_before_iter): Pass true to it.
17271         (build_check_stmt): Pass false to it.
17272         (transform_statements): Flush hash table only on extended basic
17273         block boundaries, rather than at the beginning of every bb.
17274         Don't flush hash table on nonfreeing_call_p calls.
17275         * tree-flow.h (nonfreeing_call_p): New prototype.
17276         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17278 2013-02-13  David S. Miller  <davem@davemloft.net>
17280         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17282 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
17284         PR target/56184
17285         * ira.c (max_regno_before_ira): Move from ...
17286         (ira): ... here.
17287         (fix_reg_equiv_init): Use max_regno_before_ira instead of
17288         vec_safe_length.
17290 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
17292         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17294 2013-02-13  Richard Biener  <rguenther@suse.de>
17296         PR lto/56295
17297         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17298         globals in MEM_REFs.
17300 2013-02-13  Richard Biener  <rguenther@suse.de>
17302         * loop-init.c (loop_optimizer_init): Clear loop state when
17303         re-initializing preserved loops.
17304         * loop-unswitch.c (unswitch_single_loop): Return whether
17305         we unswitched the loop.  Do not verify loop state here.
17306         (unswitch_loops): When we unswitched a loop discover new loops.
17308 2013-02-13  Kostya Serebryany  <kcc@google.com>
17310         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17311         on x86_64 linux.
17312         * sanitizer.def: Rename __asan_init to __asan_init_v1.
17314 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
17316         Avoid instrumenting duplicated memory access in the same basic block
17317         * Makefile.in (asan.o): Add new dependency on hash-table.h
17318         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17319         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17320         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17321         (free_mem_ref_resources, has_mem_ref_been_instrumented)
17322         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17323         (get_mem_ref_of_assignment): New functions.
17324         (get_mem_refs_of_builtin_call): Extract from
17325         instrument_builtin_call and tweak a little bit to make it fit with
17326         the new signature.
17327         (instrument_builtin_call): Use the new
17328         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
17329         of is_gimple_builtin_call.
17330         (instrument_derefs, instrument_mem_region_access): Insert the
17331         instrumented memory reference into the hash table.
17332         (maybe_instrument_assignment): Renamed instrument_assignment into
17333         this, and change it to advance the iterator when instrumentation
17334         actually happened and return true in that case.  This makes it
17335         homogeneous with maybe_instrument_assignment, and thus give a
17336         chance to callers to be more 'regular'.
17337         (transform_statements): Clear the memory reference hash table
17338         whenever we enter a new BB, when we cross a function call, or when
17339         we are done transforming statements.  Use
17340         maybe_instrument_assignment instead of instrumentation.  No more
17341         need to special case maybe_instrument_assignment and advance the
17342         iterator after calling it; it's now handled just like
17343         maybe_instrument_call.  Update comment.
17345 2013-02-13  Richard Biener  <rguenther@suse.de>
17347         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17348         Fix loop discovery code.
17350 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
17352         PR inline-asm/56148
17353         * lra-constraints.c (process_alt_operands): Match early clobber
17354         operand with itself.  Check conflicts with earlyclobber only if
17355         the operand is not reloaded.  Prefer to reload conflicting operand
17356         if earlyclobber and matching operands are the same.
17358 2013-02-12  Richard Biener  <rguenther@suse.de>
17360         PR lto/56297
17361         * lto-streamer-out.c (write_symbol): Do not output symbols
17362         for hard register variables.
17364 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
17366         PR target/54222
17367         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17368         (umulsidi3_insn, mulsidi3_insn): New insns.
17370 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
17372         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17373         (struct tune_params): Add vec_costs field.
17374         * config/arm/arm.c (arm_builtin_vectorization_cost)
17375         (arm_add_stmt_cost): New functions.
17376         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17377         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17378         (arm_default_vec_cost): New struct of type cpu_vec_costs.
17379         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17380         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17381         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17382         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17384 2013-02-12  Richard Biener  <rguenther@suse.de>
17386         PR lto/56295
17387         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17388         decls again if possible.
17390 2013-02-12  Richard Biener  <rguenther@suse.de>
17392         PR middle-end/56288
17393         * tree-ssa.c (verify_ssa_name): Fix check, move
17394         SSA_NAME_IN_FREE_LIST check up.
17396 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
17397             Steven Bosscher   <steven@gcc.gnu.org>
17399         PR rtl-optimization/56151
17400         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17401         equal to op0 or op1, and last_insn pattern is CODE operation
17402         with MEM dest and one of the operands matches that MEM.
17404 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
17406         * doc/extend.texi: Document Function Multiversioning and "default"
17407         parameter string to target attribute.
17408         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17409         target attribute parameter is "default".
17410         (ix86_compare_version_priority): Remove checks for target attribute.
17411         (ix86_mangle_function_version_assembler_name): Change error to sorry.
17412         Remove check for target attribute equal to NULL. Add assert.
17413         (ix86_generate_version_dispatcher_body): Change error to sorry.
17415 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
17416             Jack Howarth  <howarth@bromo.med.uc.edu>
17417             Patrick Marlier  <patrick.marlier@gmail.com>
17419         PR libitm/55693
17420         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17421         define ENDFILE_SPEC as TM_DESTRUCTOR.
17422         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17424 2013-02-11  Alexander Potapenko  <glider@google.com>
17425             Jack Howarth  <howarth@bromo.med.uc.edu>
17426             Jakub Jelinek  <jakub@redhat.com>
17428         PR sanitizer/55617
17429         * config/darwin.c (cdtor_record): Rename ctor_record.
17430         (sort_cdtor_records): Rename sort_ctor_records.
17431         (finalize_dtors): New routine to sort destructors by
17432         priority before use in assemble_integer.
17433         (machopic_asm_out_destructor): Use finalize_dtors if needed.
17435 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
17437         PR rtl-optimization/56275
17438         * simplify-rtx.c (avoid_constant_pool_reference): Check that
17439         offset is non-negative and less than cmode size before
17440         calling simplify_subreg.
17442 2013-02-11  Richard Biener  <rguenther@suse.de>
17444         PR tree-optimization/56264
17445         * cfgloop.h (fix_loop_structure): Adjust prototype.
17446         * loop-init.c (fix_loop_structure): Return the number of
17447         newly discovered loops.
17448         * tree-cfgcleanup.c (repair_loop_structures): When new loops
17449         are discovered, do a full loop-closed SSA rewrite.
17451 2013-02-11  Richard Biener  <rguenther@suse.de>
17453         PR tree-optimization/56273
17454         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17455         first VRP run.
17456         (check_array_ref): Fix missing newline in dumps.
17457         (search_for_addr_array): Likewise.
17459 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
17461         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17463 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
17465         PR target/56256
17466         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17468 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
17470         PR rtl-optimization/56246
17471         * lra-constraints.c (simplify_operand_subreg): Try to reuse
17472         reload pseudo.
17473         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17474         constraints are satisfied.
17476 2013-02-08  Jeff Law  <law@redhat.com>
17478         PR debug/53948
17479         * emit-rtl.c (reg_is_parm_p): New function.
17480         * regs.h (reg_is_parm_p): New prototype.
17481         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17482         callee-clobbered registers.
17484 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
17486         PR target/56043
17487         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17488         If there is no implicit builtin declaration, just return NULL.
17490 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17492         * config/i386/sse.md (FMAMODEM): New mode iterator.
17493         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17494         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17496 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17498         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17499         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17500         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17502 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17504         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17505         (microblaze*-*-elf): Likewise.
17506         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17507         LINK_SPEC.
17508         * config/microblaze/microblaze-c.c: Add builtin defines for
17509         _LITTLE_ENDIAN and _BIG_ENDIAN.
17510         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17511         add to TARGET_DEFAULT flags.
17512         Expand ASM_SPEC and LINK_SPEC.
17513         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17514         * config/microblaze/microblaze.md: Update extendsidi2 and
17515         movdi_internal instructions to use low-order / high-order reg
17516         print_operands.
17517         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17518         options and inversemask / mask of LITTLE_ENDIAN.
17519         * config/microblaze/t-microblaze: Expand multilib options to
17520         include mlittle-endian (le) and update exceptions patterns.
17522 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
17524         PR rtl-optimization/56195
17525         * lra-constraints.c (get_reload_reg): Don't reuse regs
17526         if they have smaller mode than requested, if they have
17527         wider mode than requested, try to return a SUBREG.
17529         PR tree-optimization/56250
17530         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17531         if type is unsigned and code isn't MULT_EXPR.
17533 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17535         PR tree-optimization/56064
17536         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17537         bits according to mode.
17538         * fixed-value.h (fixed_from_double_int)
17539         (const_fixed_from_double_int): Adjust comments.
17541 2013-02-08  Richard Biener  <rguenther@suse.de>
17543         PR lto/56231
17544         * lto-streamer.h (struct data_in): Remove current_file, current_line
17545         and current_col members.
17546         * lto-streamer-out.c (lto_output_location): Stream changed bits
17547         en-block for efficiency.
17548         * lto-streamer-in.c (clear_line_info): Remove.
17549         (lto_input_location): Cache current file, line and column
17550         globally via local statics.  Read changed bits en-block.
17551         (input_function): Do not call clear_line_info.
17552         (lto_read_body): Likewise.
17553         (lto_input_toplevel_asms): Likewise.
17555 2013-02-08  Michael Matz  <matz@suse.de>
17557         PR tree-optimization/52448
17558         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17559         (nt_call_phase): New static.
17560         (add_or_mark_expr): Only mark accesses with newer phase than any
17561         call seen.
17562         (nonfreeing_call_p): New.
17563         (nt_init_block): Update nt_call_phase, mark blocks as visited.
17564         (nt_fini_block): Keep blocks marked as visited.
17565         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17567 2013-02-08  Richard Biener  <rguenther@suse.de>
17569         * ira.c (ira): Free broken dominator information.
17571 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17573         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17575 2013-02-08  Marek Polacek  <polacek@redhat.com>
17577         * cfgloop.c (verify_loop_structure): Add more checking of headers.
17579 2013-02-08  Richard Biener  <rguenther@suse.de>
17581         PR middle-end/56181
17582         * cfgloop.h (flow_loops_find): Adjust.
17583         (bb_loop_header_p): Declare.
17584         * cfgloop.c (bb_loop_header_p): New function split out from ...
17585         (flow_loops_find): ... here.  Adjust function signature,
17586         support incremental loop structure update.
17587         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
17588         * cfgloopmanip.c (fix_loop_structure): Move ...
17589         * loop-init.c (fix_loop_structure): ... here.
17590         (apply_loop_flags): Split out from ...
17591         (loop_optimizer_init): ... here.
17592         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
17593         in incremental mode, only remove dead loops here.
17595 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17597         PR target/54222
17598         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17599         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17600         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17601         (*round<mode>3.libgcc): New insns for fixed-modes.
17602         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17603         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17604         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17605         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17606         implementations.  Define to __builtin_avr_absFX,
17607         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17608         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17609         __builtin_avr_countlsFX, respectively.
17610         * config/avr/avr-c.c (target.h): Include it.
17611         (enum avr_builtin_id): New enum.
17612         (avr_resolve_overloaded_builtin): New static function.
17613         (avr_register_target_pragmas): Use it to set
17614         targetm.resolve_overloaded_builtin.
17615         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17616         tree nodes used by DEF_BUILTIN.
17617         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17618         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17619         <AVR_BUILTIN_xxBITS>: Same.
17621 2013-02-08  Richard Biener  <rguenther@suse.de>
17623         * cfgloop.c (verify_loop_structure): Properly handle
17624         a loop exiting to another loop header.
17625         * ira-int.h (ira_loops): Remove.
17626         * ira.c (ira_loops): Remove.
17627         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17628         (do_reload): Use loop_optimizer_finalize.
17629         * ira-build.c (create_loop_tree_nodes): Use get_loops and
17630         number_of_loops to access the loop tree.
17631         (more_one_region_p): Likewise.
17632         (finish_loop_tree_nodes): Likewise.
17633         (rebuild_regno_allocno_maps): Likewise.
17634         (mark_loops_for_removal): Likewise.
17635         (mark_all_loops_for_removal): Likewise.
17636         (remove_unnecessary_regions): Likewise.
17637         (ira_build): Likewise.
17638         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17640 2013-02-08  Richard Biener  <rguenther@suse.de>
17642         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17643         * ipa-pure-const.c (analyze_function): Avoid calling
17644         mark_irreducible_loops twice.
17645         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17647 2013-02-07  David S. Miller  <davem@davemloft.net>
17649         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17650         on 'reg'.
17651         * var-tracking.c (vt_add_function_parameter): Test the presence of
17652         HAVE_window_save properly and do not remap argument registers when
17653         we have a leaf function.
17655 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
17657         PR bootstrap/56227
17658         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17659         instead of "ll".
17660         * config/i386/i386.c (ix86_print_operand): Ditto.
17662 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17664         * lra-constraints.c (process_alt_operands): Fix recently added comment.
17666 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17668         PR rtl-optimization/56225
17669         * lra-constraints.c (process_alt_operands): Check that reload hard
17670         reg can hold value for strict_low_part.
17672 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17674         PR debug/56154
17675         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17676         dwarf2out_end_function.
17677         (in_first_function_p, maybe_at_text_label_p,
17678         first_loclabel_num_not_at_text_label): New variables.
17679         (dwarf2out_var_location): In the first function find out
17680         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17681         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17682         functions.
17684 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
17686         PR rtl-optimization/56178
17687         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17688         SUBREG of a register.  Tidy up related block of code.
17689         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17690         note if the source is a register or a SUBREG of a register.
17692 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17694         PR target/56228
17695         * config/rs6000/rs6000.md (ptrm): New mode attr.
17696         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17697         call_value_indirect_aix<pttrsize>,
17698         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17699         m in constraints.
17701 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17703         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17704         if -bnortl. Convert to strcmp and strncmp.
17706 2013-02-07  Alan Modra  <amodra@gmail.com>
17708         PR target/54009
17709         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17710         addresses won't wrap when offsetting.
17711         (rs6000_secondary_reload): Provide secondary reloads needed for
17712         wrapping LO_SUM addresses.
17714 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
17716         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17717         MACH, just __MACH__.
17719 2013-02-06  Richard Biener  <rguenther@suse.de>
17721         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17722         instead of calling fix_loop_structure.
17724 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
17726         PR middle-end/56217
17727         * omp-low.c (use_pointer_for_field): Return false if
17728         lower_send_shared_vars doesn't generate any copy-out code.
17730 2013-02-06  Tom de Vries  <tom@codesourcery.com>
17732         PR rtl-optimization/56131
17733         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17734         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
17735         of the label is NULL.  Add comment.
17737 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
17739         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17741         PR sanitizer/55374
17742         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17743         (STATIC_LIBTSAN_LIBS): Likewise.
17744         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17745         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17746         is defined, don't add anything else beyond that.
17747         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17748         (LINK_COMMAND_SPEC): Use them.
17750         PR tree-optimization/56205
17751         * tree-stdarg.c (check_all_va_list_escapes): Return true if
17752         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17753         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17755 2013-02-05  Richard Biener  <rguenther@suse.de>
17757         PR tree-optimization/53342
17758         PR tree-optimization/53185
17759         * tree-vectorizer.h (vect_check_strided_load): Remove.
17760         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17761         not disallow peeling for vectorized strided loads.
17762         (vect_check_strided_load): Make static and simplify.
17763         (vect_analyze_data_refs): Adjust.
17764         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17765         correctly when vectorizing strided loads.
17767 2013-02-05  Richard Biener  <rguenther@suse.de>
17769         * doc/install.texi: Refer to ISL, not PPL.
17771 2013-02-05  Jan Hubicka  <jh@suse.cz>
17773         PR tree-optimization/55789
17774         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17776 2013-02-05  Jan Hubicka  <jh@suse.cz>
17778         PR tree-optimization/55789
17779         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17780         the dead call anyway.
17782 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
17784         PR sanitizer/55374
17785         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17787 2013-02-04  Alexander Potapenko  <glider@google.com>
17788             Jack Howarth  <howarth@bromo.med.uc.edu>
17789             Jakub Jelinek  <jakub@redhat.com>
17791         PR sanitizer/55617
17792         * config/darwin.c (sort_ctor_records): Stabilized qsort
17793         on constructor priority by using original position.
17794         (finalize_ctors): New routine to sort constructors by
17795         priority before use in assemble_integer.
17796         (machopic_asm_out_constructor): Use finalize_ctors if needed.
17798 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
17800         PR libstdc++/54314
17801         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17802         about visibility on artificial decls.
17803         * config/sol2.c (solaris_assemble_visibility): Likewise.
17805 2013-02-04  Kai Tietz  <ktietz@redhat.com>
17807         PR target/56186
17808         * config/i386/i386.c (function_value_ms_64): Add additional valtype
17809         argument and improve checking of return-argument types for 16-byte
17810         modes.
17811         (ix86_function_value_1): Add additional valtype argument on call
17812         of function_value_64.
17813         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17814         handling infunction_value_64 function.
17816 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
17818         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17820 2013-02-04  Richard Biener  <rguenther@suse.de>
17822         PR tree-optimization/56188
17823         * tree-ssa-structalias.c (label_visit): Consider case with
17824         initially non-empty points-to set.
17825         (perform_var_substitution): Dump node mapping and clean up.
17827 2013-02-04  Richard Guenther  <rguenther@suse.de>
17829         PR lto/56168
17830         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17831         node prevail as last resort.
17832         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17833         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17835 2013-02-04  Richard Biener  <rguenther@suse.de>
17837         PR tree-optimization/56113
17838         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17839         Merge into ...
17840         (equiv_class_lookup_or_add): ... this.
17841         (label_visit): Adjust and fix error in previous patch.
17842         (perform_var_substitution): Adjust.
17844 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17846         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17847         * config/sh/sh4-300.md: Likewise.
17848         * config/sh/sh4a.md: Likewise.
17849         * config/sh/constraints.md: Likewise.
17850         * config/sh/sh.md: Likewise.
17851         * config/sh/netbsd-elf.h: Likewise.
17852         * config/sh/predicates.md: Likewise.
17853         * config/sh/sh-protos.h: Likewise.
17854         * config/sh/ushmedia.h: Likewise.
17855         * config/sh/linux.h: Likewise.
17856         * config/sh/sh.c: Likewise.
17857         * config/sh/superh.h: Likewise.
17858         * config/sh/elf.h: Likewise.
17859         * config/sh/sh4.md: Likewise.
17860         * config/sh/sh.h: Likewise.
17862 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17864         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17865         constraint to match_test floating_point_store_memory_operand().
17866         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17867         (base14_operand): New.
17868         (floating_point_store_memory_operand): New.
17869         (integer_store_memory_operand): Revise to use base14_operand and
17870         reg_plus_base_memory_operand.
17871         (move_dest_operand): Allow symbolic_memory_operands.
17872         (symbolic_memory_operand): Check for LO_SOM.
17873         (symbolic_operand): Change default case to break.
17874         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17875         CONST_DOUBLE values to be reloaded by putting them into memory when
17876         the destination is a floating point register.
17877         (movdf): Remove code to handle CONST_DOUBLE.
17878         (movsf): Likewise.
17879         (reload_indf_r1): New.
17880         (reload_insf_r1): New.
17881         Consistently use "Q" and "T" constraints with integer and floating
17882         point move instructions, respectively.
17883         (movdi): Remove FAIL.
17884         Change predicate for source operand unamed DImode move from
17885         general_operand to move_src_operand.
17886         (umulsidi3): Change predicate for destination operand to
17887         register_operand.
17888         Likewise for similar unamed patterns.
17889         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17890         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17891         (hppa_legitimize_address): Simplify mask calculation.
17892         (pa_emit_move_sequence): Revised handling of secondary reloads from
17893         REG+D addresses for floating point loads and stores.  Directly handle
17894         loading CONST0_RTX (mode) to a floating point register.
17895         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17896         to floating point registers.  Don't restrict secondary reloads to
17897         floating point registers to integer modes.  Revise some comments and
17898         cleanup some code.
17899         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17900         (pa_legitimate_address_p): New.
17901         (pa_legitimize_reload_address): New.
17902         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17903         (STRICT_REG_OK_FOR_BASE_P): New.
17904         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17905         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17907 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17908             Andrew Dixie  <andrewd@gentrack.com>
17910         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17911         flag set.
17913 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17915         * expmed.c (extract_bit_field_1): Pass the full width of the
17916         structure to get_best_reg_extraction_insn.
17918 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17920         PR target/54601
17921         * configure.ac (use_cxa_atexit): Add AIX.
17922         * configure: Regenerate.
17924         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17926 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17928         PR debug/54793
17929         * final.c (need_profile_function): New variable.
17930         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17931         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17932         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17933         notes, targetm.asm_out.function_prologue doesn't emit anything,
17934         HAVE_prologue and profiler should be emitted before prologue,
17935         set need_profile_function instead of emitting it.
17936         (final_scan_insn): If need_profile_function, emit
17937         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17938         NOTE_INSN_FUNCTION_BEG note.
17940 2013-02-01  Richard Henderson  <rth@redhat.com>
17942         * config/rs6000/rs6000.md (smulditi3): New.
17943         (umulditi3): New.
17945         * config/alpha/alpha.md (umulditi3): New.
17947 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17949         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17950         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17952 2013-02-01  Richard Biener  <rguenther@suse.de>
17954         PR tree-optimization/56113
17955         * tree-ssa-structalias.c (label_visit): Reduce work for
17956         single-predecessor nodes.
17958 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17960         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17961         range isn't testing for zero.
17963 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17965         PR middle-end/56113
17966         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17968 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17969             Nick Clifton  <nickc@redhat.com>
17971         * config/v850/constraints.md (Q): Define as a memory constraint.
17972         * config/v850/predicates.md (label_ref_operand): New predicate.
17973         (e3v5_shift_operand): New predicate.
17974         (ior_operator): New predicate.
17975         * config/v850/t-v850: Add e3v5 multilib.
17976         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17977         (v850_gen_movdi): Prototype.
17978         * config/v850/v850.c: Add support for e3v5 architecture.
17979         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17980         TARGET_V850E_UP.
17981         (construct_save_jarl): Add e3v5 long JARL support.
17982         (v850_adjust_insn_length): New function.  Adjust length of call
17983         insns when using e3v5 instructions.
17984         (v850_gen_movdi): New function: Generate instructions to move a
17985         DImode value.
17986         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17987         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17988         (TARGET_USE_FPU): Enable for e3v5.
17989         (CONST_OK_FOR_W): New macro.
17990         (ADJUST_INSN_LENGTH): Define.
17991         * config/v850/v850.md (UNSPEC_LOOP): Define.
17992         (attr cpu): Add v850e3v5.
17993         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17994         (movdi): New pattern.
17995         (movdi_internal): New pattern.
17996         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17997         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17998         (cstoresf4): Likewise.
17999         (cstoredf4): Likewise.
18000         (insv): New pattern.
18001         (rotlso3_a): New pattern.
18002         (rotlsi3_b): New pattern
18003         (rotlsi3_v850e3v5): New pattern.
18004         (doloop_begin): New pattern.
18005         (fix_loop_counter): New pattern.
18006         (doloop_end): New pattern.
18007         (branch_normal): Add e3v5 long branch support.
18008         (branch_invert): Likewise.
18009         (branch_z_normal): Likewise.
18010         (branch_z_invert): Likewise.
18011         (branch_nz_normal): Likewise.
18012         (branch_nz_invert): Likewise.
18013         (call_internal_short): Add e3v5 register-indirect JARL support.
18014         (call_internal_long): Likewise.
18015         (call_value_internal_short): Likewise.
18016         (call_value_internal_long): Likewise.
18017         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
18018         (mloop): New option.
18019         * config.gcc: Add support for configuring v840e3v5 target.
18020         * doc/invoke.texi: Document new v850 specific command line options.
18022 2013-01-31  Paul Koning  <ni1d@arrl.net>
18024         PR debug/55059
18025         PR debug/54508
18026         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
18027         children if parent is a class.
18028         (prune_unused_types_prune): Don't add DW_AT_declaration.
18030 2013-01-31  Richard Biener  <rguenther@suse.de>
18032         PR tree-optimization/56157
18033         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
18034         match up operand with SLP child.
18036 2013-01-31  Jason Merrill  <jason@redhat.com>
18038         PR debug/54410
18039         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
18040         parameters the first time.
18041         (gen_scheduled_generic_parms_dies): Check completeness here.
18043 2013-01-31  Richard Biener  <rguenther@suse.de>
18045         PR middle-end/53073
18046         * common.opt (faggressive-loop-optimizations): New flag,
18047         enabled by default.
18048         * doc/invoke.texi (faggressive-loop-optimizations): Document.
18049         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
18050         infer_loop_bounds_from_undefined by it.
18052 2013-01-31  Richard Biener  <rguenther@suse.de>
18054         PR tree-optimization/56150
18055         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
18056         visit virtual operands.
18057         (find_uses_to_rename_bb): Likewise.
18059 2013-01-31  Richard Biener  <rguenther@suse.de>
18061         PR tree-optimization/56150
18062         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
18063         mixed store non-store stmts.
18065 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
18067         PR sanitizer/55374
18068         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
18069         LIBASAN_EARLY_SPEC is defined.
18070         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
18071         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
18072         before %o.
18073         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
18075         PR c++/55742
18076         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
18077         invalid args instead of ICEing on it.
18078         (ix86_valid_target_attribute_tree): Return error_mark_node if
18079         ix86_valid_target_attribute_inner_p failed.
18080         (ix86_valid_target_attribute_p): Return false only if
18081         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
18082         target("default") attribute.
18083         (sorted_attr_string): Change argument from const char * to tree,
18084         merge in all target attribute arguments rather than just one.
18085         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
18086         instead of free.  Avoid using strcat.
18087         (ix86_mangle_function_version_assembler_name): Mangle
18088         target("default") as if no target attribute is present.  Adjust
18089         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
18090         instead of xmalloc and XDELETEVEC instead of free.
18091         (ix86_function_versions): Don't return true if one of the decls
18092         doesn't have target attribute.  If they don't and one of the decls
18093         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
18094         sorted_attr_string caller.  Use XDELETEVEC instead of free.
18095         (ix86_supports_function_versions): Remove.
18096         (make_name): Fix up formatting.
18097         (make_dispatcher_decl): Remove resolver_name and its initialization.
18098         Avoid leaking memory.
18099         (is_function_default_version): Return true if there is
18100         target("default") attribute rather than no target attribute at all.
18101         (make_resolver_func): Avoid leaking memory.
18102         (ix86_generate_version_dispatcher_body): Likewise.
18103         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18104         * target.def (supports_function_versions): Remove.
18105         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18106         * doc/tm.texi: Regenerated.
18108 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
18110         PR rtl-optimization/56144
18111         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18112         for values with side effects.
18114 2013-01-30  Richard Biener  <rguenther@suse.de>
18116         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18117         (sparseset_pop): Likewise.
18118         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
18119         to be able to use quick_push in the worker loop.
18121 2013-01-30  Marek Polacek  <polacek@redhat.com>
18123         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18125 2013-01-30  Richard Biener  <rguenther@suse.de>
18127         PR lto/56147
18128         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18130 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
18132         PR tree-optimization/56064
18133         * fixed-value.c (fixed_from_double_int): New function.
18134         * fixed-value.h (fixed_from_double_int): New prototype.
18135         (const_fixed_from_double_int): New static inline function.
18136         * fold-const.c (native_interpret_fixed): New static function.
18137         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18138         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18139         (native_encode_fixed): New static function.
18140         (native_encode_expr) <FIXED_CST>: Use it.
18141         (native_interpret_int): Move double_int worker code to...
18142         * double-int.c (double_int::from_buffer): ...this new static method.
18143         * double-int.h (double_int::from_buffer): Prototype it.
18145 2013-01-30  Richard Biener  <rguenther@suse.de>
18147         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18148         New pointer-map and obstack.
18149         (init_alias_vars): Allocate pointer-map and obstack.
18150         (delete_points_to_sets): Free them.
18151         (find_what_var_points_to): Cache result.
18152         (find_what_p_points_to): Adjust for changed interface of
18153         find_what_var_points_to.
18154         (compute_points_to_sets): Likewise.
18155         (ipa_pta_execute): Likewise.
18157 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18159         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18160         * configure: Regenerate.
18161         * config.in: Regenerate.
18162         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18163         #nobits/#progbits if supported.
18165 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
18167         PR target/56121
18168         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18169         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18170         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18172 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18174         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18175         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18177 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18179         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18180         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18182 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18184         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18185         declaration.
18186         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18187         * config/arm/cortex-a7.md: New bypasses using
18188         arm_mac_accumulator_is_result.
18190 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18192         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
18193         (cortex_a7_neon_mla): Likewise.
18194         (cortex_a7_fpfmad): New reservation.
18195         (cortex_a7_fpmacs): Use ffmas and update required units.
18196         (cortex_a7_fpmuld): Update required units and latency.
18197         (cortex_a7_fpmacd): Likewise.
18198         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18199         (cortex_a7_neon). Likewise.
18200         (bypass) Update participating units.
18202 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18204         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18205         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18206         from fmac to ffma.
18207         * config/arm/vfp11.md (vfp_farith): Use ffmas.
18208         (vfp_fmul): Use ffmad.
18209         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18210         (cortex_r4_fmacd): Use ffmad.
18211         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18212         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
18213         (cortex_a9_fmacd): Use ffmad.
18214         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18215         (cortex_a8_vfp_macd): Use ffmad.
18216         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18217         (cortex_a5_fpmacd): Use ffmad.
18218         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18219         (cortex_a15_vfp_macd): Use ffmad.
18220         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18222 2013-01-29  Jason Merrill  <jason@redhat.com>
18224         PR libstdc++/54314
18225         * varasm.c (default_assemble_visibility): Don't warn about
18226         visibility on artificial decls.
18228 2013-01-29  Richard Biener  <rguenther@suse.de>
18230         PR tree-optimization/56113
18231         * tree-ssa-structalias.c (equiv_class_lookup): Also return
18232         the bitmap leader.
18233         (label_visit): Free duplicate bitmaps and record the leader instead.
18234         (perform_var_substitution): Adjust.
18236 2013-01-29  Richard Biener  <rguenther@suse.de>
18238         PR tree-optimization/55270
18239         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18240         the CFG, schedule loops for fixup.
18242 2013-01-29  Nick Clifton  <nickc@redhat.com>
18244         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18245         SP_REG.
18247 2013-01-28  Leif Ekblad  <leif@rdos.net>
18249         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18250         * config/i386/i386.h (TARGET_RDOS): New macro.
18251         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18252         * config/i386/i386.c (ix86_option_override_internal): For 64bit
18253         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18254         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18255         DEFAULT_LARGE_SECTION_THRESHOLD.
18256         * config/i386/i386.md (R14_REG, R15_REG): New constants.
18257         * config/i386/rdos.h: New file.
18258         * config/i386/rdos64.h: New file.
18260 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
18262         PR other/54814
18263         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18264         TEST_HARD_REG_BIT.
18266 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18268         PR rtl-optimization/56117
18269         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18270         call cselib_lookup_from_insn on the MEM before calling
18271         add_insn_mem_dependence.
18273 2013-01-28  Richard Biener  <rguenther@suse.de>
18275         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18276         to a stmt that didn't have one.
18277         (copy_phis_for_bb): Likewise for PHI arguments.
18278         (copy_debug_stmt): Likewise for debug stmts.
18280 2013-01-28  Richard Biener  <rguenther@suse.de>
18282         PR tree-optimization/56034
18283         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18284         (partition_builtin_p): Adjust.
18285         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
18286         it is the last partition.
18287         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18288         up the vertex for the definition.
18289         (classify_partition): Classify whether a partition is a
18290         PKIND_REDUCTION, thus has uses outside of the loop.
18291         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18292         Merge all PKIND_REDUCTION partitions into the last partition.
18293         (tree_loop_distribution): Seed partitions from reductions as well.
18295 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
18297         PR tree-optimization/56125
18298         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18299         pow(x,c) into sqrt(x) * powi(x, n/2) or
18300         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18301         optimizing for size.
18302         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18303         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18304         integer.
18306         PR tree-optimization/56094
18307         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18308         to UNKNOWN_LOCATION while gimplifying expr.
18310 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
18312         PR target/56114
18313         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18314         operand 0 in movabs insn template for -masm=intel asm alternative.
18315         (*movabs<mode>_2): Ditto for operand 1.
18317 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
18319         PR target/54663
18320         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18321         of microblaze-c.o
18323 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18325         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18326         tm_file.
18328 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18330         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18331         Undef to avoid warning.
18333 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
18335         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18336         * configure: Regenerate.
18338 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
18340         PR tree-optimization/56098
18341         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18342         for stmts with volatile ops.
18343         (cond_store_replacement): Don't optimize if assign has volatile ops.
18344         (cond_if_else_store_replacement_1): Don't optimize if either
18345         then_assign or else_assign have volatile ops.
18346         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18347         volatile ops.
18349 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18351         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18353 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18355         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18356         missing ':' in asm example.
18358 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
18360         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18361         entries into lane and laneq entries.
18362         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18363         Remove AdvSIMD scalar modes.
18364         (aarch64_sq<r>dmulh_laneq<mode>): New.
18365         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18366         modes.
18367         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18368         builtin implementations to relfect changes in RTL in aarch64-simd.md.
18369         * config/aarch64/iterators.md (VCOND): New.
18370         (VCONQ): New.
18372 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
18374         PR target/54222
18375         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18376         Add NULL LIBNAME argument to existing definitions.
18377         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18378         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18379         * config/avr/avr.c (DEF_BUILTIN): Same.
18380         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18381         (avr_expand_builtin): Expand to a vanilla call if a libgcc
18382         implementation is available (DECL_ASSEMBLER_NAME is set).
18383         (avr_fold_absfx): New static function.
18384         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18385         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18386         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18387         AVR_BUILTIN_ABSLLK.
18388         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18389         (abshk, absk, abslk, absllk): Provide as static inline functions.
18391 2013-01-25  Marek Polacek  <polacek@redhat.com>
18393         PR tree-optimization/56035
18394         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18396 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
18398         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18399         (*movtf_internal_rex64): Add (!o,C) alternative
18400         (*movxf_internal_rex64): Ditto.
18401         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18403 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
18405         * doc/invoke.texi: fix typo.
18406         * doc/objc.texi: fix typo.
18408 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
18410         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18411         for the first two alternatives.
18413 2013-01-24  Diego Novillo  <dnovillo@google.com>
18415         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
18416         (ggc-zone.o): Remove.
18417         * configure.ac: Remove option --with-gc.
18418         * configure: Re-generate.
18419         * doc/install.texi: Remove documentation for --with-gc.
18420         * gengtype.c (write_enum_defn): Remove.  Update all users.
18421         (write_Types_process_field): Remove generation of gt_e_* argument.
18422         (output_type_enum): Remove.  Update all users.
18423         (write_enum_defn): Remove.  Update all users.
18424         (enum alloc_zone): Remove.  Update all users.
18425         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18426         * ggc-common.c (ggc_splay_alloc): Remove first argument.
18427         Update all callers.
18428         (struct ptr_data): Remove field TYPE.  Update all users.
18429         (gt_pch_note_object): Remove argument TYPE.  Update all users.
18430         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18431         Update all users.
18432         * ggc-none.c (ggc_alloc_typed_stat): Remove.
18433         (struct alloc_zone): Remove.
18434         (ggc_internal_alloc_zone_stat): Remove.
18435         (ggc_internal_cleared_alloc_zone_stat): Remove.
18436         * ggc-page.c (ggc_alloc_typed_stat): Remove.
18437         (ggc_pch_count_object): Remove last argument.  Update all users.
18438         (ggc_pch_alloc_object): Remove last argument.  Update all users.
18439         (struct alloc_zone): Remove.
18440         * ggc-zone.c: Remove.
18441         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
18442         (struct alloc_zone): Remove.
18443         (ggc_alloc_typed_stat): Remove.
18444         (ggc_alloc_typed): Remove.
18445         (ggc_splay_alloc): Remove first argument.
18446         (rtl_zone): Remove.  Update all users.
18447         (tree_zone): Remove.  Update all users.
18448         (tree_id_zone): Remove.  Update all users.
18449         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
18450         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
18451         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
18452         * tree-ssanames.c: Remove references to zone allocator in comments.
18454 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
18456         * config/avr/avr.c (avr_out_fract): Make register numbers that
18457         might be outside of source operand signed.
18459 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
18461         * config/i386/constraints.md (Yf): New constraint.
18462         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18463         of f constraint to conditionaly disable x87 register preferences.
18464         (*movdf_internal): Ditto.
18465         (*movsf_internal): Ditto.
18467 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
18469         PR inline-asm/55934
18470         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18471         that have operands with impossible constraints.
18472         Add a FIXME for a speed-up opportunity.
18473         * lra-constraints.c (process_alt_operands): Verify that a class
18474         selected from constraints on asms is valid for the operand mode.
18475         (curr_insn_transform): Remove incorrect comment.
18477 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
18479         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18480         TOC operand is a valid symbol ref in the constant pool.
18482 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18484         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18486 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
18488         PR target/54222
18489         * config/avr/stdfix.h: New file.
18490         * t-avr (stdfix-gcc.h): New rule to build it.
18491         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18493 2013-01-23  Kostya Serebryany  <kcc@google.com>
18495         * config/darwin.h: remove dependency on
18496         CoreFoundation (asan on Mac OS).
18498 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
18500         PR target/49069
18501         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18502         instead of cmpdi_operand for first comparison operand.
18503         Don't assert that comparison operands aren't both constants.
18505 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18507         * doc/install.texi (Downloading the Source): Update references to
18508         downloading separate components.
18510 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18512         * doc/extend.texi (__int128): Improve grammar.
18514 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
18516         PR target/56028
18517         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18518         alternative to (o,r).
18519         (*movdi_internal_rex64): Remove (!o,n) alternative.
18520         (DImode immediate->memory splitter): Remove.
18521         (DImode immediate->memory peephole2): Remove.
18522         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18523         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18524         alternative to (!o,*r).
18525         (*movtf_internal_sse): New pattern.
18526         (*movxf_internal_rex64): New pattern.
18527         (*movxf_internal): Disable for TARGET_64BIT.
18528         (*movdf_internal_rex64): Remove (!o,F) alternative.
18530 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
18532         PR middle-end/56074
18533         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18534         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18535         * tree-vect-loop-manip.c (find_loop_location): Also ignore
18536         stmt locations where LOCATION_LOCUS of the stmt location is
18537         UNKNOWN_LOCATION or BUILTINS_LOCATION.
18539         PR target/55686
18540         * config/i386/i386.md (UNSPEC_STOS): New.
18541         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18542         *strsetqi_1): Add UNSPEC_STOS.
18544 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
18546         PR c++/56067
18547         * doc/invoke.texi: Remove left over -Wsynth example.
18549 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18551         PR tree-optimization/56051
18552         * fold-const.c (fold_binary_loc): Don't fold
18553         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18554         a narrowing conversion, or widening conversion from signed
18555         to unsigned.
18557 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18559         PR rtl-optimization/56023
18560         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18561         dependent on debug instruction.
18563 2013-01-21  Martin Jambor  <mjambor@suse.cz>
18565         PR middle-end/56022
18566         * function.c (allocate_struct_function): Call
18567         invoke_set_current_function_hook earlier.
18569 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18571         * reload1.c (init_reload): Only initialize reload_obstack
18572         during the first call.
18574 2013-01-21  Marek Polacek  <polacek@redhat.com>
18576         * cfgloop.c (verify_loop_structure): Fix up grammar.
18578 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
18580         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18581         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18583 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18585         PR target/56058
18586         * config/arm/marvell-pj4.md: Update copyright year.
18587         Fix up use of alu to alu_reg and simple_alu_imm.
18589 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18591         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18593 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
18595         PR target/55433
18596         * lra-constraints.c (curr_insn_transform): Don't reuse original
18597         insn for secondary memory move when memory mode should be different.
18599 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18601         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18602         atomic_storedi_1): New patterns.
18604 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18606         btver2 pipeline descriptions.
18607         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18608         descriptions.
18609         * config/i386/i386.md (btver2_decode): New type attributes.
18610         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18611         type attributes.
18612         * config/i386/btver2.md: New file describing btver2 pipelines.
18614 2013-01-19  Andrew Pinski  <apinski@cavium.com>
18616         PR tree-optimization/52631
18617         * tree-ssa-sccvn (visit_use): Before looking up the original
18618         statement, try looking up the simplified expression.
18620 2013-01-19  Anthony Green  <green@moxielogic.com>
18622         * config/moxie/moxie.c (moxie_expand_prologue): Set
18623         current_function_static_stack_size.
18625 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18627         PR tree-optimization/56029
18628         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18629         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18631 2013-01-18  Sharad Singhai  <singhai@google.com>
18633         PR tree-optimization/55995
18634         * dumpfile.c (dump_loc): Print location only if available.
18635         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18637 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
18639         PR target/55433
18640         * lra-constraints.c (curr_insn_transform): Reuse original insn for
18641         secondary memory move.
18642         (inherit_reload_reg): Use rclass instead of cl for
18643         check_secondary_memory_needed_p.
18645 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18647         PR middle-end/56015
18648         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18649         the case where writing real complex part of target modifies op1.
18651 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
18653         * config/aarch64/aarch64-simd.md
18654         (aarch64_vcond_internal<mode>): Handle unordered cases.
18655         * config/aarch64/iterators.md (v_cmp_result): New.
18657 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
18658             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18660         * config/arm/marvell-pj4.md: New file.
18661         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18662         * config/arm/arm.md (generic_sched): Add marvell_pj4.
18663         (generic_vfp): Likewise.
18664         * config/arm/arm-cores.def: Add marvell-pj4.
18665         * config/arm/arm-tune.md: Regenerate.
18666         * config/arm/arm-tables.opt: Regenerate.
18667         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18668         * doc/invoke.texi: Document marvell-pj4.
18670 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
18672         * config/aarch64/arm_neon.h: Map scalar types to standard types.
18674 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18676         PR debug/54114
18677         PR debug/54402
18678         PR debug/49888
18679         * var-tracking.c (negative_power_of_two_p): New.
18680         (global_get_addr_cache, local_get_addr_cache): New.
18681         (get_addr_from_global_cache, get_addr_from_local_cache): New.
18682         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
18683         heading comment.
18684         (vt_stack_offset_p): Remove.
18685         (vt_canon_true_dep): Always canonicalize loc's address.
18686         (clobber_overlapping_mems): Make sure we have a MEM.
18687         (local_get_addr_clear_given_value): New.
18688         (val_reset): Clear local cached entries.
18689         (compute_bb_dataflow): Create and release the local cache.
18690         Disable duplicate MEMs clobbering.
18691         (emit_notes_in_bb): Clobber MEMs likewise.
18692         (vt_emit_notes): Create and release the local cache.
18693         (vt_initialize, vt_finalize): Create and release the global
18694         cache, respectively.
18695         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18697 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18699         PR libmudflap/53359
18700         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18701         not found in the symtab.
18703 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18705         PR debug/56006
18706         PR rtl-optimization/55547
18707         PR rtl-optimization/53827
18708         PR debug/53671
18709         PR debug/49888
18710         * alias.c (offset_overlap_p): New, factored out of...
18711         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
18712         the conservative special case for symbolic constants.  Don't
18713         adjust zero sizes on alignment.
18715 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
18717         PR rtl-optimization/52573
18718         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18719         REG_UNUSED for the same register.
18721 2013-01-17  Richard Biener  <rguenther@suse.de>
18722             Marek Polacek  <polacek@redhat.com>
18724         PR rtl-optimization/55833
18725         * loop-unswitch.c (unswitch_loops): Move loop verification...
18726         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
18727         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18728         Set it to true when we're removing a loop from hierarchy tree in
18729         an irreducible region.
18730         (fix_bb_placements): Adjust caller.
18731         (fix_loop_placements): Likewise.
18733 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
18735         * config/avr/builtins.def (DEF_BUILTIN): Factor out
18736         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18737         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18738         Remove ID.  Adjust comments.
18739         * config/avr/avr-c.c (avr_builtin_name): Remove.
18740         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18741         * config/avr/avr.c (avr_tolower): New static function.
18742         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
18743         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18744         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18745         default expansion.
18747 2013-01-17  Jan Hubicka  <jh@suse.cz>
18749         PR tree-optimization/55273
18750         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18752 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
18754         PR target/55981
18755         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18756         store through atomic_store<mode>_1.
18757         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18759 2013-01-17  Martin Jambor  <mjambor@suse.cz>
18761         PR tree-optimizations/55264
18762         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18763         for virtual methods.
18764         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18765         virtual methods before inlining is over.
18766         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18767         virtual functions.
18768         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18769         non-virtual.
18771 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
18773         PR rtl-optimization/56005
18774         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18775         pending reads for prefetch.
18777 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
18779         * config/aarch64/aarch64.md
18780         (*cstoresi_neg_uxtw): New pattern.
18781         (*cmovsi_insn_uxtw): New pattern.
18782         (*<optab>si3_uxtw): New pattern.
18783         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18784         (*<optab>si3_insn_uxtw): New pattern.
18785         (*bswapsi2_uxtw): New pattern.
18787 2013-01-16  Richard Biener  <rguenther@suse.de>
18789         * tree-inline.c (tree_function_versioning): Remove set but
18790         never used variable.
18792 2013-01-16  Richard Biener  <rguenther@suse.de>
18794         PR tree-optimization/55964
18795         * tree-flow.h (rename_variables_in_loop): Remove.
18796         (rename_variables_in_bb): Likewise.
18797         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18798         (copy_loop_before): Adjust and delete update-ssa status.
18799         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18800         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
18801         (rename_variables_in_loop): Remove.
18802         (slpeel_update_phis_for_duplicate_loop): Likewise.
18803         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18804         use available cfg machinery instead of duplicating it.
18805         Update PHI nodes and perform poor-mans SSA update here.
18806         (slpeel_tree_peel_loop_to_edge): Adjust.
18808 2013-01-16  Richard Biener  <rguenther@suse.de>
18810         PR tree-optimization/54767
18811         PR tree-optimization/53465
18812         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18813         (vrp_visit_phi_node): For PHI arguments coming via backedges
18814         drop all symbolical range information.
18815         (execute_vrp): Compute backedges.
18817 2013-01-16  Richard Biener  <rguenther@suse.de>
18819         * doc/install.texi: Update CLooG and ISL requirements to
18820         0.18.0 and 0.11.1.
18822 2013-01-16  Christian Bruel  <christian.bruel@st.com>
18824         PR target/55301
18825         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18826         (broken_move): Handle UNSPECV_SP_SWITCH_B.
18827         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18829 2013-01-16  DJ Delorie  <dj@redhat.com>
18831         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18832         (UNSPECV_SP_SWITCH_E): New.
18833         (sp_switch_1): Change to an unspec.
18834         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18835         replace $r15.
18837 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18839         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18840         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18841         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18842         (expand_mem_thread_fence): Ditto.
18843         (expand_mem_signal_fence): Ditto.
18844         (expand_atomic_load): Ditto.
18845         (expand_atomic_store): Ditto.
18847 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18849         PR rtl-optimization/55547
18850         PR rtl-optimization/53827
18851         PR debug/53671
18852         PR debug/49888
18853         * alias.c (memrefs_conflict_p): Set sizes to negative after
18854         AND adjustments.
18856 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18858         PR target/55940
18859         * function.c (thread_prologue_and_epilogue_insns): Always
18860         add crtl->drap_reg to set_up_by_prologue.set, even if
18861         stack_realign_drap is false.
18863 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18865         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18866         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18867         *call): Fix indention.
18869 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18871         PR target/55876
18872         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18873         Update comment.
18875 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18877         PR rtl-optimization/55153
18878         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18880 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18882         PR tree-optimization/55920
18883         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18884         accesses as grp_to_be_debug_replaced.
18886 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18888         PR tree-optimization/55920
18889         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18890         there is non-useless type conversion needed from debug rhs to lhs,
18891         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18893 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18894             Mikael Pettersson  <mikpe@it.uu.se>
18896         PR target/43961
18897         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18898         Thumb.
18899         (ASM_OUTPUT_CASE_LABEL): Remove.
18900         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18901         * final.c (shorten_branches): Update alignment of labels before
18902         jump tables if CASE_VECTOR_SHORTEN_MODE.
18904 2013-01-15  Richard Biener  <rguenther@suse.de>
18906         PR bootstrap/55961
18907         * system.h: Do not include gmp.h for building host tools.
18909 2013-01-15  Richard Biener  <rguenther@suse.de>
18911         PR middle-end/55882
18912         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18913         account for bitpos when computing alignment.
18915 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18917         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18918         (ix86_target_macros_internal): Likewise.
18920         * config/i386/i386.c (m_CORE2I7): Removed.
18921         (m_CORE_HASWELL): New macro.
18922         (m_CORE_ALL): Likewise.
18923         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18924         (initial_ix86_arch_features): Likewise.
18925         (processor_target_table): Initializations for Core avx2.
18926         (cpu_names): New names "core-avx2".
18927         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18928         PROCESSOR_CORE_HASWELL.
18929         (ix86_issue_rate): New case.
18930         (ia32_multipass_dfa_lookahead): Likewise.
18931         (ix86_sched_init_global): Likewise.
18933         * config/i386/i386.h (TARGET_HASWELL): New macro.
18934         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18935         (processor_type): New PROCESSOR_HASWELL.
18937 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18939         PR tree-optimization/55955
18940         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18941         *SHIFT_EXPR and *ROTATE_EXPR codes.
18943         PR tree-optimization/48766
18944         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18945         -ftrapv disable -fwrapv.
18947 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18949         PR target/55974
18950         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18951         etc. to 1 and not to __flash.
18952         Use LL suffix for __INT24_MAX__ with -mint8.
18953         Use ULL suffix for __UINT24_MAX__ with -mint8.
18955 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18957         * config/avr/avr-arch.h
18958         (struct base_arch_s): Use typedef avr_arch_t instead.
18959         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18960         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18961         * config/avr/avr.c: Same.
18962         * config/avr/avr-devices.c: Same.
18963         * config/avr/driver-avr.c: Same.
18964         * config/avr/gen-avr-mmcu-texi.c: Same.
18965         * config/avr/avr-mcus.def: Adjust comment.
18967 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18969         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18970         * config/aarch64/iterators.md (VALLDI): New.
18972 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18973             Andi Kleen  <ak@linux.intel.com>
18975         PR target/55948
18976         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18977         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18978         memmodel flag.
18980 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18982         * config/avr/avr-stdint.h: Remove trailing blanks.
18983         * config/avr/avr-log.h: Same.
18984         * config/avr/avr-arch.h: Same.
18985         * config/avr/avr-devices.c: Same.
18986         * config/avr/avr-dimode.md: Same.
18987         * config/avr/predicates.md: Same.
18988         * config/avr/avr-c.c: Same.  And fix typo.
18990         * config/avr/avr-protos.h: Same.  And:
18991         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18992         (init_cumulative_args): Rename to avr_init_cumulative_args.
18993         (expand_prologue): Rename to avr_expand_prologue.
18994         (expand_epilogue): Rename to avr_expand_epilogue.
18995         (adjust_insn_length): Rename to avr_adjust_insn_length.
18996         (notice_update_cc): Rename to avr_notice_update_cc.
18997         (final_prescan_insn): Rename to avr_final_prescan_insn.
18998         * config/avr/avr.c: Same.
18999         * config/avr/avr.h: Same.
19000         * config/avr/avr.md: Remove trailing blanks.
19001         (prologue): Use avr_expand_prologue.
19002         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
19004 2013-01-14  Richard Biener  <rguenther@suse.de>
19006         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
19007         verify_location, collect_subblocks): New functions.
19008         (verify_gimple_in_cfg): Verify that locations only reference
19009         BLOCKs in the functions BLOCK tree.
19011 2013-01-14  Richard Biener  <rguenther@suse.de>
19013         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
19014         PHI argument.
19015         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
19016         unshare reference.
19017         (insert_out_of_ssa_copy_on_edge): Likewise.
19018         (rewrite_close_phi_out_of_ssa): Likewise.
19019         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
19020         debug expressions.
19021         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
19022         propagated constants.
19023         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
19024         can not be shared.
19026 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
19028         * config/avr/avr-modes.def: Add GPL copyright notice.
19030 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
19032         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
19033         MEMMODEL_MASK to determine memory model.
19034         (atomic_store<mode>): Ditto from operands[2].
19035         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
19037 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
19039         PR fortran/55935
19040         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
19041         (fold_gimple_assign): Don't call unshare_expr here.
19042         (fold_ctor_reference): Call unshare_expr.
19044 2013-01-13  Terry Guo  <terry.guo@arm.com>
19046         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
19047         * doc/fragments.texi: Document MULTILIB_REUSE.
19048         * gcc.c (multilib_reuse): New internal spec.
19049         (set_multilib_dir): Also search multilib from multilib_reuse.
19050         * genmultilib (tmpmultilib3): Refactor code.
19051         (tmpmultilib4): Ditto.
19052         (multilib_reuse): New multilib argument.
19054 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
19056         * Makefile.in: Update copyright.
19058 2013-01-12  Tom de Vries  <tom@codesourcery.com>
19060         PR middle-end/55890
19061         * calls.c (expand_call): Check if arg_nr is valid.
19063 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
19065         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
19066         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
19067         documentation.  Add missing '__' in front of
19068         __builtin_ia32_packssdw256.
19070 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19072         PR target/55719
19073         * config/s390/s390.c (s390_preferred_reload_class): Do not return
19074         NO_REGS for larl operands.
19075         (s390_reload_larl_operand): Use s390_load_address instead of
19076         emit_move_insn.
19078 2013-01-11  Richard Biener  <rguenther@suse.de>
19080         * tree-cfg.c (verify_node_sharing_1): Split out from ...
19081         (verify_node_sharing): ... here.
19082         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
19084 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
19086         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
19087         Substitute TREECHECKING.
19088         * configure: Regenerate.
19089         * Makefile.in (TREECHECKING): New.
19091 2013-01-11  Richard Guenther  <rguenther@suse.de>
19093         PR tree-optimization/44061
19094         * tree-vrp.c (extract_range_basic): Compute zero as
19095         value-range for __builtin_constant_p of function parameters.
19097 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
19099         Update copyright years.
19101 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
19103         PR rtl-optimization/55672
19104         * lra-eliminations.c (mark_not_eliminable): Permit addition with
19105         const to be eliminable.
19107 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
19109         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19110         * configure: Regenerate.
19112 2013-01-10  Richard Biener  <rguenther@suse.de>
19114         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19116 2013-01-10  Richard Biener  <rguenther@suse.de>
19118         PR bootstrap/55792
19119         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19120         locations for virtual PHI arguments.
19121         (rewrite_update_phi_arguments): Likewise.
19123 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
19125         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19126         on to assembler.
19128 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
19130         PR tree-optimization/55921
19131         * tree-complex.c (expand_complex_asm): New function.
19132         (expand_complex_operations_1): Call it for GIMPLE_ASM.
19134 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19136         PR target/55718
19137         * config/s390/s390.c (s390_symref_operand_p)
19138         (s390_loadrelative_operand_p): Merge the two functions.
19139         (s390_check_qrst_address, print_operand_address): Add parameters
19140         to s390_loadrelative_operand_p invokation.
19141         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19142         (s390_reload_larl_operand, s390_secondary_reload): Use
19143         s390_loadrelative_operand_p instead of s390_symref_operand_p.
19144         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19146 2013-01-09  Mike Stump  <mikestump@comcast.net>
19148         * dse.c (record_store): Remove unnecessary assert.
19150 2013-01-09  Jan Hubicka  <jh@suse.cz>
19152         PR tree-optimization/55569
19153         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19154         * cfgloop.h (scale_loop_profile): Likewise.
19156 2013-01-09  Jan Hubicka  <jh@suse.cz>
19158         PR lto/45375
19159         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19160         functions.
19161         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19163 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
19165         PR middle-end/55114
19166         * expr.h (maybe_emit_group_store): Declare.
19167         * expr.c (maybe_emit_group_store): New function.
19168         * builtins.c (expand_builtin_int_roundingfn): Call it.
19169         (expand_builtin_int_roundingfn_2): Likewise.
19171 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
19173         PR rtl-optimization/55829
19174         * lra-constraints.c (match_reload): Add code for absent output.
19175         (curr_insn_transform): Add code for reloads of matched inputs
19176         without output.
19178 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
19180         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19181         attribute of movddup insn to DF.
19182         (*vec_interleave_lowv2df): Ditto.
19183         (vec_dupv2df): Ditto.
19185 2013-01-09  Jan Hubicka  <jh@suse.cz>
19187         PR tree-optimiation/55875
19188         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19189         EVERY_ITERATION parameter.
19190         (number_of_iterations_exit): Check if exit is executed every iteration.
19191         (idx_infer_loop_bounds): Similarly here.
19192         (n_of_executions_at_most): Simplify
19193         to only test for cases where statement is dominated by the
19194         particular bound; handle correctly the "postdominance" test.
19195         (scev_probably_wraps_p): Use max loop iterations info
19196         as a global bound first.
19198 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
19199             Nick Clifton  <nickc@redhat.com>
19201         * config/v850/v850.md (cbranchsf4): New pattern.
19202         (cstoresf4): New pattern.
19203         (cbranchdf4): New pattern.
19204         (cstoredf4): New pattern.
19205         (movsicc): Disallow floating point comparisons.
19206         (cmpsf_le_insn): Fix order of operators.
19207         (cmpsf_lt_insn): Likewise.
19208         (cmpsf_eq_insn): Likewise.
19209         (cmpdf_le_insn): Likewise.
19210         (cmpdf_lt_insn): Likewise.
19211         (cmpdf_eq_insn): Likewise.
19212         (cmpsf_ge_insn): Use LE comparison.
19213         (cmpdf_ge_insn): Likewise.
19214         (cmpsf_gt_insn): Use LT comparison.
19215         (cmpdf_gt_insn): Likewise.
19216         (cmpsf_ne_insn): Delete pattern.
19217         (cmpdf_ne_insn): Delete pattern.
19218         * config/v850/v850.c (v850_gen_float_compare): Use
19219         gen_cmpdf_eq_insn for NE comparison.
19220         (v850_float_z_comparison_operator)
19221         (v850_float_nz_comparison_operator): Move from here ...
19222         * config/v850/predicates.md: ... to here.  Move GT and GE
19223         comparisons into v850_float_z_comparison_operator.
19224         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19225         Delete prototype.
19226         (v850_float_nz_comparison_operator): Likewise.
19228 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19230         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19231         with calls to gen_insvsi/gen_insvdi.
19233 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19235         * config/i386/i386.c (initial_ix86_tune_features): Set up
19236         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19238 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
19239             Jakub Jelinek  <jakub@redhat.com>
19241         PR tree-optimization/48189
19242         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19243         If nitercst is 0, don't predict the exit edge.
19245 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19247         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19248         in asm_fprintf with reg_names.
19249         (aarch64_print_operand_address): Likewise.
19250         (aarch64_return_addr): Likewise.
19251         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19253 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19255         * config/pa/pa.h (VAL_U6_BITS_P): Define.
19256         (INT_U6_BITS): Likewise.
19257         * config/pa/predicates.md (uint6_operand): New predicate.
19258         (shift5_operand, shift6_operand): Likewise.
19259         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19260         arith32_operand.
19261         (lshrdi3): Use shift6_operand.
19262         (shrpsi4, shrpdi4): New insn patterns.
19263         (extzv): Delete expander.
19264         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
19265         predicates in unamed zero extract patterns.  Tighten common constraint.
19266         (extv): Delete expander.
19267         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
19268         predicates in unamed sign extract patterns.  Tighten common constraint.
19269         (insv): Delete expander.
19270         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
19271         predicates in unamed insert patterns.  Tighten common constraint.
19272         Change uint32_operand predicate to uint6_operand predicate in unamed
19273         DImode pattern to insert constant values of type 1...1xxxx.
19275 2013-01-04  Jan Hubicka  <jh@suse.cz>
19277         PR tree-optimization/55823
19278         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19279         issue.
19281 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19282             Uros Bizjak  <ubizjak@gmail.com>
19284         PR rtl-optimization/55845
19285         * df-problems.c (can_move_insns_across): Stop scanning at
19286         volatile_insn_p source instruction or give up if
19287         across_from .. across_to range contains any volatile_insn_p
19288         instructions.
19290 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19292         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19293         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19294         Declare.
19295         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19296         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19298 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19300         PR fortran/55341
19301         * asan.c (asan_clear_shadow): New function.
19302         (asan_emit_stack_protection): Use it.
19304 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
19306         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19307         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19308         with tab instead of space.
19310 2013-01-08  Nick Clifton  <nickc@redhat.com>
19312         * config/rl78/rl78.c (rl78_expand_prologue): Always select
19313         register bank 0 at the start of an interrupt handler.
19314         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19315         MDBH registers.
19317 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19319         * config/aarch64/aarch64-simd.md
19320         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19321         (aarch64_simd_bsl): Likewise.
19322         (aarch64_vcond_internal<mode>): Likewise.
19323         (vcond<mode><mode>): Likewise.
19324         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19325         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19327 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
19329         * config/aarch64/aarch64-builtins.c
19330         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19332 2013-01-08  Martin Jambor  <mjambor@suse.cz>
19334         PR debug/55579
19335         * tree-sra.c (analyze_access_subtree): Return true also after
19336         potentially creating a debug-only replacement.
19338 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19340         PR middle-end/55890
19341         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19343         PR tree-optimization/54120
19344         * tree-vrp.c (range_fits_type_p): Don't allow
19345         src_precision < precision from signed vr to unsigned_p
19346         if vr->min or vr->max is negative.
19347         (simplify_float_conversion_using_ranges): Test can_float_p
19348         against CODE_FOR_nothing.
19350 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
19351             Richard Biener  <rguenther@suse.de>
19353         PR middle-end/55851
19354         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19355         types instead of just INTEGER_TYPE types.
19357 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
19359         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19360         TF_SIZE): Define.
19362 2013-01-07  Steve Ellcey  <sellcey@mips.com>
19364         PR target/42661
19365         * config/mips/mips.opt: Change mad to mmad to match documentation.
19367 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19369         PR target/55897
19370         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19371         .progmemx.data now.
19373 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19375         PR target/55897
19376         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19377         (avr_addrspace_t): Add .section_name field.
19378         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19379         array size.
19380         (avr_addrspace): Same.  Initialize .section_name.  Remove last
19381         NULL entry.  Put __memx into .progmemx.data.
19382         (progmem_section_prefix): Remove.
19383         (avr_asm_init_sections): No need to initialize progmem_section.
19384         (avr_asm_named_section): Use avr_addrspace[].section_name to get
19385         section name prefix.
19386         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
19387         retrieve the progmem section.
19388         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19389         boundary to run over avr_addrspace[].
19390         (avr_register_target_pragmas): Ditto.
19392 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
19394         * varasm.c (output_constant_def_contents): For asan_protect_global
19395         protected strings, adjust DECL_ALIGN if needed, before testing for
19396         anchored symbols.
19397         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19398         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
19399         normal decls.
19400         (output_object_block): For asan protected decls, emit asan padding
19401         after their contents.
19402         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19403         (asan_finish_file): Test it here instead.
19405 2013-01-07  Nick Clifton  <nickc@redhat.com>
19406             Matthias Klose  <doko@debian.org>
19407             Doug Kwan  <dougkwan@google.com>
19408             H.J. Lu  <hongjiu.lu@intel.com>
19410         PR driver/55470
19411         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19413         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19415         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19417         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19419         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19421 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19423         PR target/54461
19424         * doc/install.texi (Cross-Compiler-Specific Options): Document
19425         --with-avrlibc.
19427 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
19429         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19430         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19431         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19432         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19433         vqmovun_high_s64): Fix source operand number and update copyright.
19435 2013-01-07  Richard Biener  <rguenther@suse.de>
19437         PR middle-end/55890
19438         * gimple.h (gimple_call_builtin_p): New overload.
19439         * gimple.c (validate_call): New function.
19440         (gimple_call_builtin_p): Likewise.
19441         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19442         Use gimple_call_builtin_p.
19443         (find_func_clobbers): Likewise.
19444         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19445         (strlen_optimize_stmt): Likewise.
19447 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19449         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19450         (vld1q_dup_*): Likewise.
19451         (vld1_*): Likewise.
19452         (vld1q_*): Likewise.
19453         (vld1_lane_*): Likewise.
19454         (vld1q_lane_*): Likewise.
19456 2013-01-07  Richard Biener  <rguenther@suse.de>
19458         * lto-streamer.h (LTO_minor_version): Bump to 2.
19460 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19462         * config/aarch64/aarch64-protos.h
19463         (aarch64_const_double_zero_rtx_p): Rename to...
19464         (aarch64_float_const_zero_rtx_p): ...this.
19465         (aarch64_float_const_representable_p): New.
19466         (aarch64_output_simd_mov_immediate): Likewise.
19467         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19468         move immediate case.
19469         * config/aarch64/aarch64.c
19470         (aarch64_const_double_zero_rtx_p): Rename to...
19471         (aarch64_float_const_zero_rtx_p): ...this.
19472         (aarch64_print_operand): Allow printing of new constants.
19473         (aarch64_valid_floating_const): New.
19474         (aarch64_legitimate_constant_p): Check for valid floating-point
19475         constants.
19476         (aarch64_simd_valid_immediate): Likewise.
19477         (aarch64_vect_float_const_representable_p): New.
19478         (aarch64_float_const_representable_p): Likewise.
19479         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19480         (aarch64_output_simd_mov_immediate): New.
19481         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19482         (*movdf_aarch64): Likewise.
19483         * config/aarch64/constraints.md (Ufc): New.
19484         (Y): call aarch64_float_const_zero_rtx.
19485         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19487 2013-01-07  Richard Biener  <rguenther@suse.de>
19489         PR tree-optimization/55888
19490         PR tree-optimization/55862
19491         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19492         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19493         not if it is contained therein.
19495 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19497         * config/avr/t-avr: Typo.
19499 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19501         PR55243
19502         * config/avr/t-avr: Don't automatically rebuild
19503         $(srcdir)/config/avr/t-multilib
19504         $(srcdir)/config/avr/avr-tables.opt
19505         $(srcdir)/doc/avr-mmcu.texi
19506         (avr-mcus): New phony target to build them on request.
19507         (s-avr-mlib, s-avr-mmcu-texi): Remove.
19508         * avr/avr-mcus.def: Adjust comments.
19510 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
19512         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19514 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19516         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19518 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19520         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19522 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
19524         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19525         to generate profiling.
19526         * config/rs6000/aix64.h (LIB_SPEC): Same.
19528 2013-01-04  Andrew Pinski  <apinski@cavium.com>
19530         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19531         New function.
19532         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19534 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
19536         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19537         unconditionally.
19538         (ix86_expand_move): Ditto.
19539         (ix86_zero_extend_to_Pmode): Ditto.
19540         (ix86_expand_call): Ditto.
19541         (ix86_expand_special_args_builtin): Ditto.
19542         (ix86_expand_builtin): Ditto.
19544 2013-01-04  Richard Biener  <rguenther@suse.de>
19546         PR tree-optimization/55862
19547         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19548         translating them through PHI nodes.
19550 2013-01-04  Martin Jambor  <mjambor@suse.cz>
19552         PR tree-optimization/55755
19553         * tree-sra.c (sra_modify_assign): Do not check that an access has no
19554         children when trying to avoid producing a VIEW_CONVERT_EXPR.
19556 2013-01-04  Marek Polacek  <polacek@redhat.com>
19558         PR middle-end/55859
19559         * opts.c (default_options_optimization): Clarify error message.
19561 2013-01-04  Richard Biener  <rguenther@suse.de>
19563         PR middle-end/55863
19564         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19565         reassociation.
19567 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19569         PR target/53789
19570         * config/pa/pa.md (movsi): Revert previous change.
19571         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19572         references.
19574 2013-01-03  Richard Henderson  <rth@redhat.com>
19576         * config/i386/i386.c (ix86_expand_move): Always assign to op1
19577         after eliminating TLS symbols.
19579 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
19581         PR bootstrap/50167
19582         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19583         * graphite-poly.c (debug_gmp_value): Likewise.
19585 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
19587         PR target/55712
19588         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19589         selected code model, define __code_mode_small__, __code_model_medium__,
19590         __code_model_large__, __code_model_32__ or __code_model_kernel__.
19591         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19592         xchg temporary register with %k.  Declare temporary register as
19593         early clobbered.
19594         [__x86_64__]: For medium and large code models, preserve %rbx register.
19596 2013-01-03  Richard Biener  <rguenther@suse.de>
19598         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19599         (dump_subscript): Adjust.
19600         (finalize_ddr_dependent): Do not dump redundant info.
19601         (analyze_siv_subscript): Adjust.
19602         (subscript_dependence_tester): Likewise.
19603         (compute_affine_dependence): Likewise.
19605 2013-01-03  Richard Biener  <rguenther@suse.de>
19607         Revert
19608         2013-01-03  Richard Biener  <rguenther@suse.de>
19610         PR tree-optimization/55857
19611         * tree-vect-stmts.c (vectorizable_load): Do not setup
19612         re-alignment for invariant loads.
19614         2013-01-02  Richard Biener  <rguenther@suse.de>
19616         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19617         invariant load do not generate a vector load from the scalar location.
19619 2013-01-03  Richard Biener  <rguenther@suse.de>
19621         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19622         for not vectorizing.
19623         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19624         not build INDIRECT_REFs, call get_name once only.
19625         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
19626         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19628 2013-01-03  Richard Biener  <rguenther@suse.de>
19630         PR tree-optimization/55857
19631         * tree-vect-stmts.c (vectorizable_load): Do not setup
19632         re-alignment for invariant loads.
19634 2013-01-03  Richard Biener  <rguenther@suse.de>
19636         PR lto/55848
19637         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19638         prefer a built-in decl.
19640 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19642         * gcc.c (process_command): Update copyright notice dates.
19643         * gcov.c (print_version): Likewise.
19644         * gcov-dump.c (print_version): Likewise.
19646         PR rtl-optimization/55838
19647         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19648         iv0.step, iv1.step and step.
19650 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19651             Marc Glisse  <marc.glisse@inria.fr>
19653         PR tree-optimization/55832
19654         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19655         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19656         integer_{one,zero}_node.
19658 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19660         PR debug/54402
19661         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19662         * var-tracking.c (reverse_op): Don't add reverse ops to
19663         VALUEs that have already
19664         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19666 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
19668         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19670 2013-01-02  Teresa Johnson  <tejohnson@google.com>
19672         * dumpfile.c (dump_loc): Print filename with location.
19673         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19674         new location_t parameter to emit complete unroll message with
19675         new dump framework.
19676         (canonicalize_loop_induction_variables): Compute loops location
19677         and pass to try_unroll_loop_completely.
19678         * loop-unroll.c (report_unroll_peel): New function.
19679         (peel_loops_completely): Use new dump format with location
19680         for main dumpfile message, and invoke report_unroll_peel on success.
19681         (decide_unrolling_and_peeling): Ditto.
19682         (decide_peel_once_rolling): Remove old dumpfile message subsumed
19683         by report_unroll_peel.
19684         (decide_peel_completely): Ditto.
19685         (decide_unroll_constant_iterations): Ditto.
19686         (decide_unroll_runtime_iterations): Ditto.
19687         (decide_peel_simple): Ditto.
19688         (decide_unroll_stupid): Ditto.
19689         * cfgloop.c (get_loop_location): New function.
19690         * cfgloop.h (get_loop_location): Declare.
19692 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19694         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19695         NULL.
19697 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19699         PR middle-end/55198
19700         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19701         BLKmode objects when EXPAND_MEMORY is specified.
19703 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19705         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19706         in loop predicate.
19707         (fold_builtin_cpu): Do not share cpu model decls across statements.
19709 2013-01-02  Jason Merrill  <jason@redhat.com>
19711         PR c++/55804
19712         * tree.c (build_array_type_1): Revert earlier change.
19714 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
19716         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19717         "cortex-a57".
19718         * config/aarch64/aarch64-tune.md: Re-generate.
19720 2013-01-02  Richard Biener  <rguenther@suse.de>
19722         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19723         invariant load do not generate a vector load from the scalar location.
19725 2013-01-02  Richard Biener  <rguenther@suse.de>
19727         PR bootstrap/55784
19728         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19729         * configure: Regenerate.
19731 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19733         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19734         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19735         (expand_builtin_int_roundingfn_2): Keep the original target around
19736         for the fallback case.
19738 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19740         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19741         to be clear for sign changes.
19743 2013-01-01  Jan Hubicka  <jh@suse.cz>
19745         * ipa-inline-analysis.c: Fix formatting.
19747 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
19749         PR tree-optimization/55831
19750         * tree-vect-loop.c (get_initial_def_for_induction): Use
19751         gsi_after_labels instead of gsi_start_bb.
19753 Copyright (C) 2013 Free Software Foundation, Inc.
19755 Copying and distribution of this file, with or without modification,
19756 are permitted in any medium without royalty provided the copyright
19757 notice and this notice are preserved.