Fix unitialized var
[official-gcc.git] / gcc / ChangeLog
blob5d846b1180594eaeb1cfbc258b8e09beb51a63bf
1 2013-08-14  Xinliang David Li  <davidxl@google.com>
2         * config/i386/i386.c (ix86_option_override_internal):
3         Fix potential unitialized variable error.
5 2013-08-14  Xinliang David Li  <davidxl@google.com>
7         * config/i386/i386.opt: Define two new options.
8         * config/i386/x86-tune.def: Add arch selector field in macros.
9         * config/i386/i386.h: Adjust macro definition.
10         * config/i386/i386.c (ix86_option_override_internal):
11         Refactor the code.
12         (parse_mtune_ctrl_str): New function.
13         (set_ix86_tune_features): New function.
14         (ix86_function_specific_restore): Call the new helper function.
16 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
18         PR rtl-optimization/57662
19         * sel-sched.c (code_motion_process_successors): When the current insn
20         is removed after the recursive traversal, break from the loop.
21         Add comments and debug printouts.
23 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
24             Alexandre Oliva  <aoliva@redhat.com>
26         PR target/58067
27         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
28         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
29         there also UNSPEC_PLTOFF.
31 2013-08-14  Marek Polacek  <polacek@redhat.com>
33         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
34         NUM_CONDITIONS bit positions.
36 2013-08-13  Cary Coutant  <ccoutant@google.com>
38         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
39         (attr_checksum): Hash vector contents instead of pointer.
40         (attr_checksum_ordered): Likewise.
42 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
44         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
45         when Pmode != word_mode.  Add length_address attribute.
46         (sse3_monitor_<mode>): Merge from sse3_monitor and
47         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
48         Pmode != word_mode.  Update insn length attribute.
49         * config/i386/i386.c (ix86_option_override_internal): Update
50         ix86_gen_monitor selection for merged sse3_monitor insn.
52 2013-08-13  Julian Brown  <julian@codesourcery.com>
54         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
55         perform invalid legitimization on greater-than-word-size modes for
56         TARGET_E500_DOUBLE.
58 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
60         * ira.c (setup_class_translate_array): Use aclass instead of cl
61         for classes not fully covered by allocno classes.
63 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
65         PR tree-optimization/57661
66         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
67         * tree-inline.c (tree_function_versioning): Initialize it.
68         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
69         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
70         that is not being copied.
72         PR sanitizer/56417
73         * asan.c (instrument_strlen_call): Fix typo in comment.
74         Use char * type even for the lhs of POINTER_PLUS_EXPR.
76 2013-08-13  Steve Ellcey  <sellcey@mips.com>
78         * config/mips/mips.md (prefetch): Use lw instead of ld on
79         loongson in 32bit mode.
81 2013-08-13  Nick Clifton  <nickc@redhat.com>
83         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
85 2013-08-13  Jan Hubicka  <jh@suse.cz>
87         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
88         introduced edge; fix typo in sanity check.
89         (cgraph_resolve_speculation): Export; improve diagnostic.
90         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
91         speculation at type mismatch.
92         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
93         (cgraph_resolve_speculation): Declare.
94         (symtab_can_be_discarded): New function.
95         * value-prof.c (gimple_ic_transform): Remove actual transform code.
96         * ipa-inline-transform.c (speculation_removed): New global var.
97         (clone_inlined_nodes): See if speculation can be removed.
98         (inline_call): If speculations was removed, we growths may not match.
99         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
100         (speculation_useful_p): New function.
101         (resolve_noninline_speculation): New function.
102         (inline_small_functions): Resolve useless speculations.
103         * ipa-inline.h (speculation_useful_p): Declare
104         * ipa.c (can_replace_by_local_alias): Simplify.
105         (ipa_profile): Produce speculative calls in non-lto, too;
106         add simple cost model; produce local aliases.
108 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
110         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
111         PASS_MANAGER_H.
113 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
115         * config/i386/i386.c (ix86_function_versions): Use error + inform.
117 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
119         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
120         iterator instead of X87MODEF.
122 2013-08-12  Perez Read  <netfirewall@gmail.com>
124         PR target/58132
125         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
126         operand 0 for intel asm alternative.
127         (*movabs<mode>_2): Ditto for operand 1.
129 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
131         * config/aarch64/arm_none.h
132         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
134 2013-08-12  Nick Clifton  <nickc@redhat.com>
136         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
138 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
140         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
141         expand for QI/HImode operand to produce more effictive code for
142         unsigned char(short) --> float(double) conversion.
144 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
146         * doc/invoke.texi: Mention that -ftls-model does not force the final
147         model.
149 2013-08-12  Marek Polacek  <polacek@redhat.com>
150             Marc Glisse  <marc.glisse@inria.fr>
152         PR tree-optimization/57980
153         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
154         when creating -1 constant.
156 2013-08-10  Jan Hubicka  <jh@suse.cz>
158         Workaround binutils PR14342.
159         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
160         (gimple_init_edge_profiler): Likewise.
161         (gimple_gen_ic_func_profiler): Likewise.
163 2013-08-09  Jan Hubicka  <jh@suse.cz>
165         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
167 2013-08-09  Xinliang David Li  <davidxl@google.com>
169         * config/i386/stringop.def: New file.
170         * config/i386/stringop.opt: New file.
171         * config/i386/i386-opts.h: Include stringopt.def.
172         * config/i386/i386.opt: Include stringopt.opt.
173         * config/i386/i386.c (ix86_option_override_internal):
174         Override default size based stringop inline strategies with options.
175         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
176         New function.
178 2013-08-09  Jan Hubicka  <jh@suse.cz>
180         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
182 2013-08-09  Jan Hubicka  <jh@suse.cz>
184         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
185         CGRAPH_FREQ_MAX.
186         (dump_cgraph_node): Dump profile-id.
187         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
188         and common_target_probability.
189         * lto-cgraph.c (lto_output_edge): Stream common targets.
190         (lto_output_node): Stream profile ids.
191         (input_node): Stream profile ids.
192         (input_edge): Stream common targets.
193         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
194         * ipa.c: Include value-prof.h
195         (ipa_profile_generate_summary): Turn indirect call statement histograms
196         into common targets.
197         (ipa_profile): Turn common targets into speculative edges.
199 2013-08-09  Jan Hubicka  <jh@suse.cz>
201         * cgraph.h (cgraph_node): Add profile_id.
202         * value-prof.c (cgraph_node_map): Turn into pointer_map.
203         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
204         (del_node_map): Update.
205         (find_func_by_funcdef_no): Replace by ...
206         (find_func_by_profile_id): ... this one.
207         (gimple_ic_transform): Do not remove useful histograms when
208         speculation is not done; dump info when indirect call removal
209         can happen at LTO.
210         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
211         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
212         (__gcov_indirect_call_profiler_v2): .. this one.
213         * profile.h (init_node_map): Update.
214         * coverage.c (coverage_compute_profile_id): New function.
215         * coverage.h (coverage_compute_profile_id): Declare.
216         * tree-profile.c (init_ic_make_global_vars): Make
217         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
218         (gimple_init_edge_profiler): Update prototype of
219         __gcov_indirect_call_profiler.
220         (gimple_gen_ic_func_profiler): Simplify.
221         (tree_profiling): Use init_node_map
223 2013-08-09  Jan Hubicka  <jh@suse.cz>
225         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
226         non-speculative refs.
227         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
228         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
229         (cgraph_set_call_stmt): Likewise.
230         (cgraph_create_edge_1): Fix release checking compilatoin;
231         clear lto_stmt_uid.
232         (cgraph_free_edge): Free indirect info.
233         (cgraph_turn_edge_to_speculative): New function.
234         (cgraph_speculative_call_info): New function.
235         (cgraph_make_edge_direct): Return direct edge; handle speculation.
236         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
237         (dump_cgraph_node): Dump speculation.
238         (verify_edge_count_and_frequency): Accept speculative edges.
239         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
240         (verify_cgraph_node): Handle speculation.
241         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
242         (cgraph_set_call_stmt): Update prototype.
243         (cgraph_make_edge_direct): Update prototype.
244         (cgraph_speculative_call_info): Declare.
245         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
246         to change; update call of ipa_find_references.
247         * ipa-ref.c (ipa_record_reference): Fix return value; clear
248         lto_stmt_uid and speculative flags.
249         (ipa_dump_references): Dump speculation.
250         (ipa_clone_references): Clone speculative flag.
251         (ipa_clone_referring): Likewise.
252         (ipa_clone_ref): New function.
253         (ipa_find_reference): Look into lto_stmt_uids
254         (ipa_clear_stmts_in_references): Do not clear speculative calls.
255         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
256         (ipa_find_reference): Update declaration.
257         (ipa_clone_ref): Declare.
258         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
259         stream speculative flag.
260         (lto_output_ref): Stream statements uids and speculation.
261         (input_ref): Likewise.
262         (input_edge): Stream speuclation.
263         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
264         (cgraph_set_call_stmt_including_clones): Handle speculation.
265         * ipa-inline.c (heap_edge_removal_hook): New function.
266         (inline_small_functions): Register it.
267         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
268         also initialize refs.
269         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
270         edge to change.
271         (try_make_edge_direct_simple_call): Likewise.
272         (try_make_edge_direct_simple_call): Likewise.
273         (update_indirect_edges_after_inlining): Likewise.
274         (remove_described_reference): Look proper lto_stmt_uid.
275         (propagate_controlled_uses): Likewise.
276         (propagate_controlled_uses): Liekwise.
277         * tree-inline.c (copy_bb): Copy speculative edges.
278         (redirect_all_calls): New function.
279         (copy_cfg_body): Do redirection after loop info is updated.
280         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
282 2013-08-09  Jan Hubicka  <jh@suse.cz>
284         * lto-streamer-out.c (output_function): Renumber PHIs.
285         * lto-streamer-in.c (input_function): Likewise.
287 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
289         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
290         (get_lane_unsigned): Likewise.
291         (dup_lane_scalar): Likewise.
292         (get_lane): enable for VALL.
293         * config/aarch64/aarch64-simd.md
294         (aarch64_dup_lane_scalar<mode>): Remove.
295         (aarch64_get_lane_signed<mode>): Likewise.
296         (aarch64_get_lane_unsigned<mode>): Likewise.
297         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
298         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
299         (aarch64_get_lane<mode>): Enable for all vector modes.
300         (aarch64_get_lanedi): Remove misleading constraints.
301         * config/aarch64/arm_neon.h
302         (__aarch64_vget_lane_any): Define.
303         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
304         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
305         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
306         * config/aarch64/iterators.md (VDQQH): New.
307         (VDQQHS): Likewise.
308         (vwcore): Likewise.
310 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
312         * configure.ac: Add GAS check for LEON instructions on SPARC.
313         * configure: Regenerate.
314         * config.in: Likewise.
315         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
316         sparc*-*-* block.
317         * config/sparc/sparc.opt (LEON, LEON3): New masks.
318         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
319         for LEON or LEON3.
320         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
321         (AS_LEON_FLAG): New macro.
322         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
323         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
324         Deal with LEON and LEON3 for the memory model.
325         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
326         (atomic_compare_and_swap<mode>_1): Likewise.
327         (*atomic_compare_and_swap<mode>_1): Likewise.
329 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
331         * config/arm/neon.md (vcond): Fix floating-point vector
332         comparisons against 0.
334 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
336         * lra-constraints.c (emit_spill_move): Remove assert.
337         (process_alt_operands): Add more debugging
338         output.  Increase reject for spilling into memory.  Decrease
339         reject for reloading scratch.
340         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
342 2013-08-08  Steve Ellcey  <sellcey@mips.com>
344         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
345         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
346         micromips incompatible.  Add nan2008.
347         (MULTILIB_DIRNAMES): Add nan2008.
348         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
349         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
350         and micromips incompatible.  Add nan2008.
351         (MULTILIB_DIRNAMES): Add nan2008.
352         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
354 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
356         PR rtl-optimization/58079
357         * combine.c (combine_simplify_rtx): Avoid using SUBST if
358         simplify_comparison has widened a comparison with an integer.
360 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
362         * config/arm/neon.md (movmisalign<mode>): Disable when we
363         don't allow unaligned accesses.
364         (*movmisalign<mode>_neon_store): Likewise.
365         (*movmisalign<mode>_neon_load): Likewise.
366         (*movmisalign<mode>_neon_store): Likewise.
367         (*movmisalign<mode>_neon_load): Likewise.
369 2013-08-08  Jan Hubicka  <jh@suse.cz>
371         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
372         (make_pass_rebuild_cgraph_edges): Also clear references.
373         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
374         * ipa-inline-transform.c (inline_transform): Remove all references
375         after inlining.
376         * cgraphunit.c (expand_function): Remove all references after
377         expansion.
378         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
379         (ipa_find_reference): Rewrite to iterator.
380         (remove_stmt_references): Likewise.
381         (ipa_clear_stmts_in_references): New function.
382         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
383         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
384         clear references.
385         * ipa-split.c (split_function): Remove references in split function.
387 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
389         PR target/57431
390         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
391         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
393 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
395         PR target/56979
396         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
397         suggested mode for the assignment isn't compatible with the
398         registers required.
400 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
402         PR target/58065
403         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
405 2013-08-07  Xinliang David Li  <davidxl@google.com>
407         * config/i386/i386.opt: New option -mtune-ctrl=.
408         * config/i386/x86-tune.def: New file.
409         * config/i386/i386.h: include x86-tune.def.
410         * config/i386/i386.c (ix86_option_override_internal):
411         Parsing -mtune-ctrl= option and set tune features.
413 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
415         PR other/12081
416         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
417         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
418         to gen_2arg_fn_t.
420 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
422         * rtl.h (update_alignments): Declare.
423         * final.c (grow_label_align): New function extracted from...
424         (shorten_branches): ...here.  Call it.
425         (update_alignments): New function.
426         * reorg.c (sibling_labels): New variable.
427         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
428         the new label along with it onto the sibling_labels vector.
429         (fill_simple_delay_slots): Adjust call to get_label_before.
430         (fill_slots_from_thread): Likewise.
431         (relax_delay_slots): Likewise.
432         (make_return_insns): Likewise.
433         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
435 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
437         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
438         document its semantics.
439         (diagnostic_report_diagnostic): Adjust accordingly.
441 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
443         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
444         (sparc_option_override): ...and port to new C++ pass API.
445         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
447 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
449         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
451 2013-08-06  Caroline Tice  <cmtice@google.com>
453         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
454         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
455         * tree-pass.h: Add pass_vtable_verify.
456         * varasm.c (assemble_variable): Add code to properly set the comdat
457         section and name for the .vtable_map_vars section.
458         (assemble_vtyv_preinit_initializer): New function.
459         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
460         LINK_ONCE flag.
461         * output.h: Add function decl for assemble_vtv_preinit_initializer.
462         * vtable-verify.c: New file.
463         * vtable-verify.h: New file.
464         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
465         initialiation levels.
466         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
467         * passes.def: Insert pass_vtable_verify.
468         * aclocal.m4: Reorder includes.
469         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
470         -fvtv-counts options.
471         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
472         as appropriate, if -fvtable-verify=... is used.
473         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
474         -fvtable-verify=... is used.
475         * Makefile.in (OBJS):  Add vtable-verify.o to list.
476         (vtable-verify.o): Add new build rule.
477         (GTFILES): Add vtable-verify.c to list.
478         * common.opt (fvtable-verify=): New flag.
479         (vtv_priority): Values for fvtable-verify= flag.
480         (fvtv-counts): New flag.
481         (fvtv-debug): New flag.
482         * tree.h (save_vtable_map_decl): New extern function decl.
484 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
486         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
487         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
488         (pass_data_rl78_devirt): ...new pass_data instance and...
489         (make_pass_rl78_devirt): ...new function.
490         (rl78_asm_file_start): Port pass registration to new C++ API.
492 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
494         * coretypes.h (rtl_opt_pass): Add.
495         (gcc::context): Add.
496         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
497         (epiphany_init): Port to new C++ pass API.
498         (epiphany_optimize_mode_switching): Likewise.
499         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
500         (pass_manager::get_pass_mode_switching): New.
501         (pass_manager::get_pass_peephole2): New.
502         * mode-switching.c (pass_mode_switching): Add clone method.
503         * recog.c (pass_peephole2): Add clone method.
504         (pass_split_all_insns): Add clone method.
506 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
508         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
509         (mips_option_override): ...here, porting to new C++ API for passes.
511 2013-08-06  Jan Hubicka  <jh@suse.cz>
513         * cgraph.c (cgraph_get_body): New function based on lto.c
514         implementation.
515         * cgraph.h (cgraph_get_body): Declare.
516         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
517         LTO paths.
518         * cgraphunit.c (expand_function): Get body prior expanding.
519         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
520         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
521         really need.
522         * passes.c (do_per_function_toporder): Get body.
523         * tree-inline.c (expand_call_inline): Get body prior inlining it.
524         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
526 2013-08-06  Martin Jambor  <mjambor@suse.cz>
528         PR fortran/57987
529         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
530         is not re-finalized.  Rename second parameter to no_collect.
532 2013-08-06  Martin Jambor  <mjambor@suse.cz>
534         PR middle-end/58041
535         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
536         MEM_REF has proper alignment information.
538 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
540         PR other/12081
541         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
542         class insn_gen_fn.
543         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
544         rtx (*) (rtx, ...) with insn_gen_fn.
545         * genoutput.c (output_insn_data): Cast gen_? function pointers to
546         insn_gen_fn::stored_funcptr.  Add initializer braces.
548 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
550         Rewrite how instances of passes are cloned to remove assumptions
551         about their sizes (thus allowing pass subclasses to have
552         additional data fields, albeit non-GC-managed ones at this point).
554         * passes.c (make_pass_instance): Now that passes have clone
555         methods, rewrite this function to eliminate XNEW and memcpy
556         calls that used hardcoded sizes.  Since this function no longer
557         creates pass instances, rename it to...
558         (add_pass_instance): ...this.  Document the old way that passes were
559         numbered and flagged, and rework this function to continue using it.
560         (next_pass_1): Add an initial_pass argument for use by
561         add_pass_instance.
562         (position_pass): When adding multiple instances of a pass, use
563         the pass's clone method, rather than relying on the XNEW/memcpy
564         within the former make_pass_instance (now add_pass_instance).
565         (pass_manager::pass_manager): When invoking next_pass_1, also supply
566         the initial instance of the current pass within the pass manager.
568 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
570         This is the automated part of the conversion of passes from C
571         structs to C++ classes.
573         Patch autogenerated by refactor_passes.py from
574         https://github.com/davidmalcolm/gcc-refactoring-scripts
575         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
577         * asan.c (pass_asan): Convert from a global struct to a subclass of
578         gimple_opt_pass along with...
579         (pass_data_asan): ...new pass_data instance and...
580         (make_pass_asan): ...new function.
581         (pass_asan_O0): Convert from a global struct to a subclass of
582         gimple_opt_pass along with...
583         (pass_data_asan_O0): ...new pass_data instance and...
584         (make_pass_asan_O0): ...new function.
585         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
586         subclass of rtl_opt_pass along with...
587         (pass_data_inc_dec): ...new pass_data instance and...
588         (make_pass_inc_dec): ...new function.
589         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
590         a subclass of rtl_opt_pass along with...
591         (pass_data_reorder_blocks): ...new pass_data instance and...
592         (make_pass_reorder_blocks): ...new function.
593         (pass_duplicate_computed_gotos): Convert from a global struct to a
594         subclass of rtl_opt_pass along with...
595         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
596         (make_pass_duplicate_computed_gotos): ...new function.
597         (pass_partition_blocks): Convert from a global struct to a subclass of
598         rtl_opt_pass along with...
599         (pass_data_partition_blocks): ...new pass_data instance and...
600         (make_pass_partition_blocks): ...new function.
601         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
602         struct to a subclass of rtl_opt_pass along with...
603         (pass_data_branch_target_load_optimize1): ...new pass_data instance
604         and...
605         (make_pass_branch_target_load_optimize1): ...new function.
606         (pass_branch_target_load_optimize2): Convert from a global struct to a
607         subclass of rtl_opt_pass along with...
608         (pass_data_branch_target_load_optimize2): ...new pass_data instance
609         and...
610         (make_pass_branch_target_load_optimize2): ...new function.
611         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
612         of rtl_opt_pass along with...
613         (pass_data_jump): ...new pass_data instance and...
614         (make_pass_jump): ...new function.
615         (pass_jump2): Convert from a global struct to a subclass of
616         rtl_opt_pass along with...
617         (pass_data_jump2): ...new pass_data instance and...
618         (make_pass_jump2): ...new function.
619         * cfgexpand.c (pass_expand): Convert from a global struct to a
620         subclass of rtl_opt_pass along with...
621         (pass_data_expand): ...new pass_data instance and...
622         (make_pass_expand): ...new function.
623         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
624         of rtl_opt_pass along with...
625         (pass_data_free_cfg): ...new pass_data instance and...
626         (make_pass_free_cfg): ...new function.
627         (pass_into_cfg_layout_mode): Convert from a global struct to a
628         subclass of rtl_opt_pass along with...
629         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
630         (make_pass_into_cfg_layout_mode): ...new function.
631         (pass_outof_cfg_layout_mode): Convert from a global struct to a
632         subclass of rtl_opt_pass along with...
633         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
634         (make_pass_outof_cfg_layout_mode): ...new function.
635         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
636         struct to a subclass of gimple_opt_pass along with...
637         (pass_data_build_cgraph_edges): ...new pass_data instance and...
638         (make_pass_build_cgraph_edges): ...new function.
639         (pass_rebuild_cgraph_edges): Convert from a global struct to a
640         subclass of gimple_opt_pass along with...
641         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
642         (make_pass_rebuild_cgraph_edges): ...new function.
643         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
644         subclass of gimple_opt_pass along with...
645         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
646         and...
647         (make_pass_remove_cgraph_callee_edges): ...new function.
648         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
649         struct to a subclass of rtl_opt_pass along with...
650         (pass_data_stack_adjustments): ...new pass_data instance and...
651         (make_pass_stack_adjustments): ...new function.
652         * combine.c (pass_combine): Convert from a global struct to a subclass
653         of rtl_opt_pass along with...
654         (pass_data_combine): ...new pass_data instance and...
655         (make_pass_combine): ...new function.
656         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
657         global struct to a subclass of rtl_opt_pass along with...
658         (pass_data_compare_elim_after_reload): ...new pass_data instance
659         and...
660         (make_pass_compare_elim_after_reload): ...new function.
661         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
662         of rtl_opt_pass along with...
663         (pass_data_rtl_cprop): ...new pass_data instance and...
664         (make_pass_rtl_cprop): ...new function.
665         * cse.c (pass_cse): Convert from a global struct to a subclass of
666         rtl_opt_pass along with...
667         (pass_data_cse): ...new pass_data instance and...
668         (make_pass_cse): ...new function.
669         (pass_cse2): Convert from a global struct to a subclass of
670         rtl_opt_pass along with...
671         (pass_data_cse2): ...new pass_data instance and...
672         (make_pass_cse2): ...new function.
673         (pass_cse_after_global_opts): Convert from a global struct to a
674         subclass of rtl_opt_pass along with...
675         (pass_data_cse_after_global_opts): ...new pass_data instance and...
676         (make_pass_cse_after_global_opts): ...new function.
677         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
678         of rtl_opt_pass along with...
679         (pass_data_ud_rtl_dce): ...new pass_data instance and...
680         (make_pass_ud_rtl_dce): ...new function.
681         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
682         rtl_opt_pass along with...
683         (pass_data_fast_rtl_dce): ...new pass_data instance and...
684         (make_pass_fast_rtl_dce): ...new function.
685         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
686         a subclass of rtl_opt_pass along with...
687         (pass_data_df_initialize_opt): ...new pass_data instance and...
688         (make_pass_df_initialize_opt): ...new function.
689         (pass_df_initialize_no_opt): Convert from a global struct to a
690         subclass of rtl_opt_pass along with...
691         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
692         (make_pass_df_initialize_no_opt): ...new function.
693         (pass_df_finish): Convert from a global struct to a subclass of
694         rtl_opt_pass along with...
695         (pass_data_df_finish): ...new pass_data instance and...
696         (make_pass_df_finish): ...new function.
697         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
698         rtl_opt_pass along with...
699         (pass_data_rtl_dse1): ...new pass_data instance and...
700         (make_pass_rtl_dse1): ...new function.
701         (pass_rtl_dse2): Convert from a global struct to a subclass of
702         rtl_opt_pass along with...
703         (pass_data_rtl_dse2): ...new pass_data instance and...
704         (make_pass_rtl_dse2): ...new function.
705         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
706         subclass of rtl_opt_pass along with...
707         (pass_data_dwarf2_frame): ...new pass_data instance and...
708         (make_pass_dwarf2_frame): ...new function.
709         * except.c (pass_set_nothrow_function_flags): Convert from a global
710         struct to a subclass of rtl_opt_pass along with...
711         (pass_data_set_nothrow_function_flags): ...new pass_data instance
712         and...
713         (make_pass_set_nothrow_function_flags): ...new function.
714         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
715         subclass of rtl_opt_pass along with...
716         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
717         and...
718         (make_pass_convert_to_eh_region_ranges): ...new function.
719         * final.c (pass_compute_alignments): Convert from a global struct to a
720         subclass of rtl_opt_pass along with...
721         (pass_data_compute_alignments): ...new pass_data instance and...
722         (make_pass_compute_alignments): ...new function.
723         (pass_final): Convert from a global struct to a subclass of
724         rtl_opt_pass along with...
725         (pass_data_final): ...new pass_data instance and...
726         (make_pass_final): ...new function.
727         (pass_shorten_branches): Convert from a global struct to a subclass of
728         rtl_opt_pass along with...
729         (pass_data_shorten_branches): ...new pass_data instance and...
730         (make_pass_shorten_branches): ...new function.
731         (pass_clean_state): Convert from a global struct to a subclass of
732         rtl_opt_pass along with...
733         (pass_data_clean_state): ...new pass_data instance and...
734         (make_pass_clean_state): ...new function.
735         * function.c (pass_instantiate_virtual_regs): Convert from a global
736         struct to a subclass of rtl_opt_pass along with...
737         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
738         (make_pass_instantiate_virtual_regs): ...new function.
739         (pass_leaf_regs): Convert from a global struct to a subclass of
740         rtl_opt_pass along with...
741         (pass_data_leaf_regs): ...new pass_data instance and...
742         (make_pass_leaf_regs): ...new function.
743         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
744         subclass of rtl_opt_pass along with...
745         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
746         and...
747         (make_pass_thread_prologue_and_epilogue): ...new function.
748         (pass_match_asm_constraints): Convert from a global struct to a
749         subclass of rtl_opt_pass along with...
750         (pass_data_match_asm_constraints): ...new pass_data instance and...
751         (make_pass_match_asm_constraints): ...new function.
752         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
753         subclass of rtl_opt_pass along with...
754         (pass_data_rtl_fwprop): ...new pass_data instance and...
755         (make_pass_rtl_fwprop): ...new function.
756         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
757         rtl_opt_pass along with...
758         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
759         (make_pass_rtl_fwprop_addr): ...new function.
760         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
761         rtl_opt_pass along with...
762         (pass_data_rtl_pre): ...new pass_data instance and...
763         (make_pass_rtl_pre): ...new function.
764         (pass_rtl_hoist): Convert from a global struct to a subclass of
765         rtl_opt_pass along with...
766         (pass_data_rtl_hoist): ...new pass_data instance and...
767         (make_pass_rtl_hoist): ...new function.
768         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
769         subclass of gimple_opt_pass along with...
770         (pass_data_lower_cf): ...new pass_data instance and...
771         (make_pass_lower_cf): ...new function.
772         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
773         from a global struct to a subclass of gimple_opt_pass along with...
774         (pass_data_strength_reduction): ...new pass_data instance and...
775         (make_pass_strength_reduction): ...new function.
776         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
777         of rtl_opt_pass along with...
778         (pass_data_rtl_ifcvt): ...new pass_data instance and...
779         (make_pass_rtl_ifcvt): ...new function.
780         (pass_if_after_combine): Convert from a global struct to a subclass of
781         rtl_opt_pass along with...
782         (pass_data_if_after_combine): ...new pass_data instance and...
783         (make_pass_if_after_combine): ...new function.
784         (pass_if_after_reload): Convert from a global struct to a subclass of
785         rtl_opt_pass along with...
786         (pass_data_if_after_reload): ...new pass_data instance and...
787         (make_pass_if_after_reload): ...new function.
788         * init-regs.c (pass_initialize_regs): Convert from a global struct to
789         a subclass of rtl_opt_pass along with...
790         (pass_data_initialize_regs): ...new pass_data instance and...
791         (make_pass_initialize_regs): ...new function.
792         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
793         of ipa_opt_pass_d along with...
794         (pass_data_ipa_cp): ...new pass_data instance and...
795         (make_pass_ipa_cp): ...new function.
796         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
797         global struct to a subclass of gimple_opt_pass along with...
798         (pass_data_inline_parameters): ...new pass_data instance and...
799         (make_pass_inline_parameters): ...new function.
800         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
801         subclass of gimple_opt_pass along with...
802         (pass_data_early_inline): ...new pass_data instance and...
803         (make_pass_early_inline): ...new function.
804         (pass_ipa_inline): Convert from a global struct to a subclass of
805         ipa_opt_pass_d along with...
806         (pass_data_ipa_inline): ...new pass_data instance and...
807         (make_pass_ipa_inline): ...new function.
808         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
809         struct to a subclass of gimple_opt_pass along with...
810         (pass_data_local_pure_const): ...new pass_data instance and...
811         (make_pass_local_pure_const): ...new function.
812         (pass_ipa_pure_const): Convert from a global struct to a subclass of
813         ipa_opt_pass_d along with...
814         (pass_data_ipa_pure_const): ...new pass_data instance and...
815         (make_pass_ipa_pure_const): ...new function.
816         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
817         to a subclass of ipa_opt_pass_d along with...
818         (pass_data_ipa_reference): ...new pass_data instance and...
819         (make_pass_ipa_reference): ...new function.
820         * ipa-split.c (pass_split_functions): Convert from a global struct to
821         a subclass of gimple_opt_pass along with...
822         (pass_data_split_functions): ...new pass_data instance and...
823         (make_pass_split_functions): ...new function.
824         (pass_feedback_split_functions): Convert from a global struct to a
825         subclass of gimple_opt_pass along with...
826         (pass_data_feedback_split_functions): ...new pass_data instance and...
827         (make_pass_feedback_split_functions): ...new function.
828         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
829         global struct to a subclass of simple_ipa_opt_pass along with...
830         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
831         instance and...
832         (make_pass_ipa_function_and_variable_visibility): ...new function.
833         (pass_ipa_free_inline_summary): Convert from a global struct to a
834         subclass of simple_ipa_opt_pass along with...
835         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
836         (make_pass_ipa_free_inline_summary): ...new function.
837         (pass_ipa_whole_program_visibility): Convert from a global struct to a
838         subclass of ipa_opt_pass_d along with...
839         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
840         and...
841         (make_pass_ipa_whole_program_visibility): ...new function.
842         (pass_ipa_profile): Convert from a global struct to a subclass of
843         ipa_opt_pass_d along with...
844         (pass_data_ipa_profile): ...new pass_data instance and...
845         (make_pass_ipa_profile): ...new function.
846         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
847         ipa_opt_pass_d along with...
848         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
849         (make_pass_ipa_cdtor_merge): ...new function.
850         * ira.c (pass_ira): Convert from a global struct to a subclass of
851         rtl_opt_pass along with...
852         (pass_data_ira): ...new pass_data instance and...
853         (make_pass_ira): ...new function.
854         (pass_reload): Convert from a global struct to a subclass of
855         rtl_opt_pass along with...
856         (pass_data_reload): ...new pass_data instance and...
857         (make_pass_reload): ...new function.
858         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
859         subclass of rtl_opt_pass along with...
860         (pass_data_cleanup_barriers): ...new pass_data instance and...
861         (make_pass_cleanup_barriers): ...new function.
862         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
863         of rtl_opt_pass along with...
864         (pass_data_loop2): ...new pass_data instance and...
865         (make_pass_loop2): ...new function.
866         (pass_rtl_loop_init): Convert from a global struct to a subclass of
867         rtl_opt_pass along with...
868         (pass_data_rtl_loop_init): ...new pass_data instance and...
869         (make_pass_rtl_loop_init): ...new function.
870         (pass_rtl_loop_done): Convert from a global struct to a subclass of
871         rtl_opt_pass along with...
872         (pass_data_rtl_loop_done): ...new pass_data instance and...
873         (make_pass_rtl_loop_done): ...new function.
874         (pass_rtl_move_loop_invariants): Convert from a global struct to a
875         subclass of rtl_opt_pass along with...
876         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
877         (make_pass_rtl_move_loop_invariants): ...new function.
878         (pass_rtl_unswitch): Convert from a global struct to a subclass of
879         rtl_opt_pass along with...
880         (pass_data_rtl_unswitch): ...new pass_data instance and...
881         (make_pass_rtl_unswitch): ...new function.
882         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
883         subclass of rtl_opt_pass along with...
884         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
885         and...
886         (make_pass_rtl_unroll_and_peel_loops): ...new function.
887         (pass_rtl_doloop): Convert from a global struct to a subclass of
888         rtl_opt_pass along with...
889         (pass_data_rtl_doloop): ...new pass_data instance and...
890         (make_pass_rtl_doloop): ...new function.
891         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
892         a subclass of rtl_opt_pass along with...
893         (pass_data_lower_subreg): ...new pass_data instance and...
894         (make_pass_lower_subreg): ...new function.
895         (pass_lower_subreg2): Convert from a global struct to a subclass of
896         rtl_opt_pass along with...
897         (pass_data_lower_subreg2): ...new pass_data instance and...
898         (make_pass_lower_subreg2): ...new function.
899         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
900         struct to a subclass of ipa_opt_pass_d along with...
901         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
902         (make_pass_ipa_lto_gimple_out): ...new function.
903         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
904         of ipa_opt_pass_d along with...
905         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
906         (make_pass_ipa_lto_finish_out): ...new function.
907         * mode-switching.c (pass_mode_switching): Convert from a global struct
908         to a subclass of rtl_opt_pass along with...
909         (pass_data_mode_switching): ...new pass_data instance and...
910         (make_pass_mode_switching): ...new function.
911         * modulo-sched.c (pass_sms): Convert from a global struct to a
912         subclass of rtl_opt_pass along with...
913         (pass_data_sms): ...new pass_data instance and...
914         (make_pass_sms): ...new function.
915         * omp-low.c (pass_expand_omp): Convert from a global struct to a
916         subclass of gimple_opt_pass along with...
917         (pass_data_expand_omp): ...new pass_data instance and...
918         (make_pass_expand_omp): ...new function.
919         (pass_lower_omp): Convert from a global struct to a subclass of
920         gimple_opt_pass along with...
921         (pass_data_lower_omp): ...new pass_data instance and...
922         (make_pass_lower_omp): ...new function.
923         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
924         of gimple_opt_pass along with...
925         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
926         (make_pass_diagnose_omp_blocks): ...new function.
927         * passes.c (pass_early_local_passes): Convert from a global struct to
928         a subclass of simple_ipa_opt_pass along with...
929         (pass_data_early_local_passes): ...new pass_data instance and...
930         (make_pass_early_local_passes): ...new function.
931         (pass_all_early_optimizations): Convert from a global struct to a
932         subclass of gimple_opt_pass along with...
933         (pass_data_all_early_optimizations): ...new pass_data instance and...
934         (make_pass_all_early_optimizations): ...new function.
935         (pass_all_optimizations): Convert from a global struct to a subclass
936         of gimple_opt_pass along with...
937         (pass_data_all_optimizations): ...new pass_data instance and...
938         (make_pass_all_optimizations): ...new function.
939         (pass_all_optimizations_g): Convert from a global struct to a subclass
940         of gimple_opt_pass along with...
941         (pass_data_all_optimizations_g): ...new pass_data instance and...
942         (make_pass_all_optimizations_g): ...new function.
943         (pass_rest_of_compilation): Convert from a global struct to a subclass
944         of rtl_opt_pass along with...
945         (pass_data_rest_of_compilation): ...new pass_data instance and...
946         (make_pass_rest_of_compilation): ...new function.
947         (pass_postreload): Convert from a global struct to a subclass of
948         rtl_opt_pass along with...
949         (pass_data_postreload): ...new pass_data instance and...
950         (make_pass_postreload): ...new function.
951         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
952         subclass of rtl_opt_pass along with...
953         (pass_data_gcse2): ...new pass_data instance and...
954         (make_pass_gcse2): ...new function.
955         * postreload.c (pass_postreload_cse): Convert from a global struct to
956         a subclass of rtl_opt_pass along with...
957         (pass_data_postreload_cse): ...new pass_data instance and...
958         (make_pass_postreload_cse): ...new function.
959         * predict.c (pass_profile): Convert from a global struct to a subclass
960         of gimple_opt_pass along with...
961         (pass_data_profile): ...new pass_data instance and...
962         (make_pass_profile): ...new function.
963         (pass_strip_predict_hints): Convert from a global struct to a subclass
964         of gimple_opt_pass along with...
965         (pass_data_strip_predict_hints): ...new pass_data instance and...
966         (make_pass_strip_predict_hints): ...new function.
967         * recog.c (pass_peephole2): Convert from a global struct to a subclass
968         of rtl_opt_pass along with...
969         (pass_data_peephole2): ...new pass_data instance and...
970         (make_pass_peephole2): ...new function.
971         (pass_split_all_insns): Convert from a global struct to a subclass of
972         rtl_opt_pass along with...
973         (pass_data_split_all_insns): ...new pass_data instance and...
974         (make_pass_split_all_insns): ...new function.
975         (pass_split_after_reload): Convert from a global struct to a subclass
976         of rtl_opt_pass along with...
977         (pass_data_split_after_reload): ...new pass_data instance and...
978         (make_pass_split_after_reload): ...new function.
979         (pass_split_before_regstack): Convert from a global struct to a
980         subclass of rtl_opt_pass along with...
981         (pass_data_split_before_regstack): ...new pass_data instance and...
982         (make_pass_split_before_regstack): ...new function.
983         (pass_split_before_sched2): Convert from a global struct to a subclass
984         of rtl_opt_pass along with...
985         (pass_data_split_before_sched2): ...new pass_data instance and...
986         (make_pass_split_before_sched2): ...new function.
987         (pass_split_for_shorten_branches): Convert from a global struct to a
988         subclass of rtl_opt_pass along with...
989         (pass_data_split_for_shorten_branches): ...new pass_data instance
990         and...
991         (make_pass_split_for_shorten_branches): ...new function.
992         * ree.c (pass_ree): Convert from a global struct to a subclass of
993         rtl_opt_pass along with...
994         (pass_data_ree): ...new pass_data instance and...
995         (make_pass_ree): ...new function.
996         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
997         subclass of rtl_opt_pass along with...
998         (pass_data_stack_regs): ...new pass_data instance and...
999         (make_pass_stack_regs): ...new function.
1000         (pass_stack_regs_run): Convert from a global struct to a subclass of
1001         rtl_opt_pass along with...
1002         (pass_data_stack_regs_run): ...new pass_data instance and...
1003         (make_pass_stack_regs_run): ...new function.
1004         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1005         subclass of rtl_opt_pass along with...
1006         (pass_data_cprop_hardreg): ...new pass_data instance and...
1007         (make_pass_cprop_hardreg): ...new function.
1008         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1009         subclass of rtl_opt_pass along with...
1010         (pass_data_reginfo_init): ...new pass_data instance and...
1011         (make_pass_reginfo_init): ...new function.
1012         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1013         of rtl_opt_pass along with...
1014         (pass_data_regmove): ...new pass_data instance and...
1015         (make_pass_regmove): ...new function.
1016         * regrename.c (pass_regrename): Convert from a global struct to a
1017         subclass of rtl_opt_pass along with...
1018         (pass_data_regrename): ...new pass_data instance and...
1019         (make_pass_regrename): ...new function.
1020         * reorg.c (pass_delay_slots): Convert from a global struct to a
1021         subclass of rtl_opt_pass along with...
1022         (pass_data_delay_slots): ...new pass_data instance and...
1023         (make_pass_delay_slots): ...new function.
1024         (pass_machine_reorg): Convert from a global struct to a subclass of
1025         rtl_opt_pass along with...
1026         (pass_data_machine_reorg): ...new pass_data instance and...
1027         (make_pass_machine_reorg): ...new function.
1028         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1029         of rtl_opt_pass along with...
1030         (pass_data_sched): ...new pass_data instance and...
1031         (make_pass_sched): ...new function.
1032         (pass_sched2): Convert from a global struct to a subclass of
1033         rtl_opt_pass along with...
1034         (pass_data_sched2): ...new pass_data instance and...
1035         (make_pass_sched2): ...new function.
1036         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1037         to a subclass of rtl_opt_pass along with...
1038         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1039         (make_pass_stack_ptr_mod): ...new function.
1040         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1041         to a subclass of rtl_opt_pass along with...
1042         (pass_data_rtl_store_motion): ...new pass_data instance and...
1043         (make_pass_rtl_store_motion): ...new function.
1044         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1045         of gimple_opt_pass along with...
1046         (pass_data_tracer): ...new pass_data instance and...
1047         (make_pass_tracer): ...new function.
1048         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1049         to a subclass of gimple_opt_pass along with...
1050         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1051         (make_pass_diagnose_tm_blocks): ...new function.
1052         (pass_lower_tm): Convert from a global struct to a subclass of
1053         gimple_opt_pass along with...
1054         (pass_data_lower_tm): ...new pass_data instance and...
1055         (make_pass_lower_tm): ...new function.
1056         (pass_tm_init): Convert from a global struct to a subclass of
1057         gimple_opt_pass along with...
1058         (pass_data_tm_init): ...new pass_data instance and...
1059         (make_pass_tm_init): ...new function.
1060         (pass_tm_mark): Convert from a global struct to a subclass of
1061         gimple_opt_pass along with...
1062         (pass_data_tm_mark): ...new pass_data instance and...
1063         (make_pass_tm_mark): ...new function.
1064         (pass_tm_edges): Convert from a global struct to a subclass of
1065         gimple_opt_pass along with...
1066         (pass_data_tm_edges): ...new pass_data instance and...
1067         (make_pass_tm_edges): ...new function.
1068         (pass_tm_memopt): Convert from a global struct to a subclass of
1069         gimple_opt_pass along with...
1070         (pass_data_tm_memopt): ...new pass_data instance and...
1071         (make_pass_tm_memopt): ...new function.
1072         (pass_ipa_tm): Convert from a global struct to a subclass of
1073         simple_ipa_opt_pass along with...
1074         (pass_data_ipa_tm): ...new pass_data instance and...
1075         (make_pass_ipa_tm): ...new function.
1076         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1077         subclass of gimple_opt_pass along with...
1078         (pass_data_call_cdce): ...new pass_data instance and...
1079         (make_pass_call_cdce): ...new function.
1080         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1081         subclass of gimple_opt_pass along with...
1082         (pass_data_build_cfg): ...new pass_data instance and...
1083         (make_pass_build_cfg): ...new function.
1084         (pass_split_crit_edges): Convert from a global struct to a subclass of
1085         gimple_opt_pass along with...
1086         (pass_data_split_crit_edges): ...new pass_data instance and...
1087         (make_pass_split_crit_edges): ...new function.
1088         (pass_warn_function_return): Convert from a global struct to a
1089         subclass of gimple_opt_pass along with...
1090         (pass_data_warn_function_return): ...new pass_data instance and...
1091         (make_pass_warn_function_return): ...new function.
1092         (pass_warn_function_noreturn): Convert from a global struct to a
1093         subclass of gimple_opt_pass along with...
1094         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1095         (make_pass_warn_function_noreturn): ...new function.
1096         (pass_warn_unused_result): Convert from a global struct to a subclass
1097         of gimple_opt_pass along with...
1098         (pass_data_warn_unused_result): ...new pass_data instance and...
1099         (make_pass_warn_unused_result): ...new function.
1100         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1101         a subclass of gimple_opt_pass along with...
1102         (pass_data_merge_phi): ...new pass_data instance and...
1103         (make_pass_merge_phi): ...new function.
1104         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1105         a subclass of gimple_opt_pass along with...
1106         (pass_data_lower_complex): ...new pass_data instance and...
1107         (make_pass_lower_complex): ...new function.
1108         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1109         gimple_opt_pass along with...
1110         (pass_data_lower_complex_O0): ...new pass_data instance and...
1111         (make_pass_lower_complex_O0): ...new function.
1112         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1113         subclass of gimple_opt_pass along with...
1114         (pass_data_lower_eh): ...new pass_data instance and...
1115         (make_pass_lower_eh): ...new function.
1116         (pass_refactor_eh): Convert from a global struct to a subclass of
1117         gimple_opt_pass along with...
1118         (pass_data_refactor_eh): ...new pass_data instance and...
1119         (make_pass_refactor_eh): ...new function.
1120         (pass_lower_resx): Convert from a global struct to a subclass of
1121         gimple_opt_pass along with...
1122         (pass_data_lower_resx): ...new pass_data instance and...
1123         (make_pass_lower_resx): ...new function.
1124         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1125         of gimple_opt_pass along with...
1126         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1127         (make_pass_lower_eh_dispatch): ...new function.
1128         (pass_cleanup_eh): Convert from a global struct to a subclass of
1129         gimple_opt_pass along with...
1130         (pass_data_cleanup_eh): ...new pass_data instance and...
1131         (make_pass_cleanup_eh): ...new function.
1132         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1133         to a subclass of simple_ipa_opt_pass along with...
1134         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1135         (make_pass_ipa_lower_emutls): ...new function.
1136         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1137         a subclass of gimple_opt_pass along with...
1138         (pass_data_if_conversion): ...new pass_data instance and...
1139         (make_pass_if_conversion): ...new function.
1140         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1141         subclass of gimple_opt_pass along with...
1142         (pass_data_build_ssa): ...new pass_data instance and...
1143         (make_pass_build_ssa): ...new function.
1144         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1145         global struct to a subclass of gimple_opt_pass along with...
1146         (pass_data_loop_distribution): ...new pass_data instance and...
1147         (make_pass_loop_distribution): ...new function.
1148         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1149         subclass of gimple_opt_pass along with...
1150         (pass_data_mudflap_1): ...new pass_data instance and...
1151         (make_pass_mudflap_1): ...new function.
1152         (pass_mudflap_2): Convert from a global struct to a subclass of
1153         gimple_opt_pass along with...
1154         (pass_data_mudflap_2): ...new pass_data instance and...
1155         (make_pass_mudflap_2): ...new function.
1156         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1157         subclass of gimple_opt_pass along with...
1158         (pass_data_mudflap_1): ...new pass_data instance and...
1159         (make_pass_mudflap_1): ...new function.
1160         (pass_mudflap_2): Convert from a global struct to a subclass of
1161         gimple_opt_pass along with...
1162         (pass_data_mudflap_2): ...new pass_data instance and...
1163         (make_pass_mudflap_2): ...new function.
1164         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1165         gimple_opt_pass along with...
1166         (pass_data_nrv): ...new pass_data instance and...
1167         (make_pass_nrv): ...new function.
1168         (pass_return_slot): Convert from a global struct to a subclass of
1169         gimple_opt_pass along with...
1170         (pass_data_return_slot): ...new pass_data instance and...
1171         (make_pass_return_slot): ...new function.
1172         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1173         to a subclass of gimple_opt_pass along with...
1174         (pass_data_object_sizes): ...new pass_data instance and...
1175         (make_pass_object_sizes): ...new function.
1176         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1177         global struct to a subclass of gimple_opt_pass along with...
1178         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1179         and...
1180         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1181         (pass_fixup_cfg): Convert from a global struct to a subclass of
1182         gimple_opt_pass along with...
1183         (pass_data_fixup_cfg): ...new pass_data instance and...
1184         (make_pass_fixup_cfg): ...new function.
1185         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1186         (make_pass_mudflap_1): ...new function.
1187         (pass_mudflap_2): Replace declaration with that of...
1188         (make_pass_mudflap_2): ...new function.
1189         (pass_asan): Replace declaration with that of...
1190         (make_pass_asan): ...new function.
1191         (pass_asan_O0): Replace declaration with that of...
1192         (make_pass_asan_O0): ...new function.
1193         (pass_tsan): Replace declaration with that of...
1194         (make_pass_tsan): ...new function.
1195         (pass_tsan_O0): Replace declaration with that of...
1196         (make_pass_tsan_O0): ...new function.
1197         (pass_lower_cf): Replace declaration with that of...
1198         (make_pass_lower_cf): ...new function.
1199         (pass_refactor_eh): Replace declaration with that of...
1200         (make_pass_refactor_eh): ...new function.
1201         (pass_lower_eh): Replace declaration with that of...
1202         (make_pass_lower_eh): ...new function.
1203         (pass_lower_eh_dispatch): Replace declaration with that of...
1204         (make_pass_lower_eh_dispatch): ...new function.
1205         (pass_lower_resx): Replace declaration with that of...
1206         (make_pass_lower_resx): ...new function.
1207         (pass_build_cfg): Replace declaration with that of...
1208         (make_pass_build_cfg): ...new function.
1209         (pass_early_tree_profile): Replace declaration with that of...
1210         (make_pass_early_tree_profile): ...new function.
1211         (pass_cleanup_eh): Replace declaration with that of...
1212         (make_pass_cleanup_eh): ...new function.
1213         (pass_sra): Replace declaration with that of...
1214         (make_pass_sra): ...new function.
1215         (pass_sra_early): Replace declaration with that of...
1216         (make_pass_sra_early): ...new function.
1217         (pass_early_ipa_sra): Replace declaration with that of...
1218         (make_pass_early_ipa_sra): ...new function.
1219         (pass_tail_recursion): Replace declaration with that of...
1220         (make_pass_tail_recursion): ...new function.
1221         (pass_tail_calls): Replace declaration with that of...
1222         (make_pass_tail_calls): ...new function.
1223         (pass_tree_loop): Replace declaration with that of...
1224         (make_pass_tree_loop): ...new function.
1225         (pass_tree_loop_init): Replace declaration with that of...
1226         (make_pass_tree_loop_init): ...new function.
1227         (pass_lim): Replace declaration with that of...
1228         (make_pass_lim): ...new function.
1229         (pass_tree_unswitch): Replace declaration with that of...
1230         (make_pass_tree_unswitch): ...new function.
1231         (pass_predcom): Replace declaration with that of...
1232         (make_pass_predcom): ...new function.
1233         (pass_iv_canon): Replace declaration with that of...
1234         (make_pass_iv_canon): ...new function.
1235         (pass_scev_cprop): Replace declaration with that of...
1236         (make_pass_scev_cprop): ...new function.
1237         (pass_empty_loop): Replace declaration with that of...
1238         (make_pass_empty_loop): ...new function.
1239         (pass_record_bounds): Replace declaration with that of...
1240         (make_pass_record_bounds): ...new function.
1241         (pass_graphite): Replace declaration with that of...
1242         (make_pass_graphite): ...new function.
1243         (pass_graphite_transforms): Replace declaration with that of...
1244         (make_pass_graphite_transforms): ...new function.
1245         (pass_if_conversion): Replace declaration with that of...
1246         (make_pass_if_conversion): ...new function.
1247         (pass_loop_distribution): Replace declaration with that of...
1248         (make_pass_loop_distribution): ...new function.
1249         (pass_vectorize): Replace declaration with that of...
1250         (make_pass_vectorize): ...new function.
1251         (pass_slp_vectorize): Replace declaration with that of...
1252         (make_pass_slp_vectorize): ...new function.
1253         (pass_complete_unroll): Replace declaration with that of...
1254         (make_pass_complete_unroll): ...new function.
1255         (pass_complete_unrolli): Replace declaration with that of...
1256         (make_pass_complete_unrolli): ...new function.
1257         (pass_parallelize_loops): Replace declaration with that of...
1258         (make_pass_parallelize_loops): ...new function.
1259         (pass_loop_prefetch): Replace declaration with that of...
1260         (make_pass_loop_prefetch): ...new function.
1261         (pass_iv_optimize): Replace declaration with that of...
1262         (make_pass_iv_optimize): ...new function.
1263         (pass_tree_loop_done): Replace declaration with that of...
1264         (make_pass_tree_loop_done): ...new function.
1265         (pass_ch): Replace declaration with that of...
1266         (make_pass_ch): ...new function.
1267         (pass_ccp): Replace declaration with that of...
1268         (make_pass_ccp): ...new function.
1269         (pass_phi_only_cprop): Replace declaration with that of...
1270         (make_pass_phi_only_cprop): ...new function.
1271         (pass_build_ssa): Replace declaration with that of...
1272         (make_pass_build_ssa): ...new function.
1273         (pass_build_alias): Replace declaration with that of...
1274         (make_pass_build_alias): ...new function.
1275         (pass_build_ealias): Replace declaration with that of...
1276         (make_pass_build_ealias): ...new function.
1277         (pass_dominator): Replace declaration with that of...
1278         (make_pass_dominator): ...new function.
1279         (pass_dce): Replace declaration with that of...
1280         (make_pass_dce): ...new function.
1281         (pass_dce_loop): Replace declaration with that of...
1282         (make_pass_dce_loop): ...new function.
1283         (pass_cd_dce): Replace declaration with that of...
1284         (make_pass_cd_dce): ...new function.
1285         (pass_call_cdce): Replace declaration with that of...
1286         (make_pass_call_cdce): ...new function.
1287         (pass_merge_phi): Replace declaration with that of...
1288         (make_pass_merge_phi): ...new function.
1289         (pass_split_crit_edges): Replace declaration with that of...
1290         (make_pass_split_crit_edges): ...new function.
1291         (pass_pre): Replace declaration with that of...
1292         (make_pass_pre): ...new function.
1293         (pass_profile): Replace declaration with that of...
1294         (make_pass_profile): ...new function.
1295         (pass_strip_predict_hints): Replace declaration with that of...
1296         (make_pass_strip_predict_hints): ...new function.
1297         (pass_lower_complex_O0): Replace declaration with that of...
1298         (make_pass_lower_complex_O0): ...new function.
1299         (pass_lower_complex): Replace declaration with that of...
1300         (make_pass_lower_complex): ...new function.
1301         (pass_lower_vector): Replace declaration with that of...
1302         (make_pass_lower_vector): ...new function.
1303         (pass_lower_vector_ssa): Replace declaration with that of...
1304         (make_pass_lower_vector_ssa): ...new function.
1305         (pass_lower_omp): Replace declaration with that of...
1306         (make_pass_lower_omp): ...new function.
1307         (pass_diagnose_omp_blocks): Replace declaration with that of...
1308         (make_pass_diagnose_omp_blocks): ...new function.
1309         (pass_expand_omp): Replace declaration with that of...
1310         (make_pass_expand_omp): ...new function.
1311         (pass_expand_omp_ssa): Replace declaration with that of...
1312         (make_pass_expand_omp_ssa): ...new function.
1313         (pass_object_sizes): Replace declaration with that of...
1314         (make_pass_object_sizes): ...new function.
1315         (pass_strlen): Replace declaration with that of...
1316         (make_pass_strlen): ...new function.
1317         (pass_fold_builtins): Replace declaration with that of...
1318         (make_pass_fold_builtins): ...new function.
1319         (pass_stdarg): Replace declaration with that of...
1320         (make_pass_stdarg): ...new function.
1321         (pass_early_warn_uninitialized): Replace declaration with that of...
1322         (make_pass_early_warn_uninitialized): ...new function.
1323         (pass_late_warn_uninitialized): Replace declaration with that of...
1324         (make_pass_late_warn_uninitialized): ...new function.
1325         (pass_cse_reciprocals): Replace declaration with that of...
1326         (make_pass_cse_reciprocals): ...new function.
1327         (pass_cse_sincos): Replace declaration with that of...
1328         (make_pass_cse_sincos): ...new function.
1329         (pass_optimize_bswap): Replace declaration with that of...
1330         (make_pass_optimize_bswap): ...new function.
1331         (pass_optimize_widening_mul): Replace declaration with that of...
1332         (make_pass_optimize_widening_mul): ...new function.
1333         (pass_warn_function_return): Replace declaration with that of...
1334         (make_pass_warn_function_return): ...new function.
1335         (pass_warn_function_noreturn): Replace declaration with that of...
1336         (make_pass_warn_function_noreturn): ...new function.
1337         (pass_cselim): Replace declaration with that of...
1338         (make_pass_cselim): ...new function.
1339         (pass_phiopt): Replace declaration with that of...
1340         (make_pass_phiopt): ...new function.
1341         (pass_forwprop): Replace declaration with that of...
1342         (make_pass_forwprop): ...new function.
1343         (pass_phiprop): Replace declaration with that of...
1344         (make_pass_phiprop): ...new function.
1345         (pass_tree_ifcombine): Replace declaration with that of...
1346         (make_pass_tree_ifcombine): ...new function.
1347         (pass_dse): Replace declaration with that of...
1348         (make_pass_dse): ...new function.
1349         (pass_nrv): Replace declaration with that of...
1350         (make_pass_nrv): ...new function.
1351         (pass_rename_ssa_copies): Replace declaration with that of...
1352         (make_pass_rename_ssa_copies): ...new function.
1353         (pass_sink_code): Replace declaration with that of...
1354         (make_pass_sink_code): ...new function.
1355         (pass_fre): Replace declaration with that of...
1356         (make_pass_fre): ...new function.
1357         (pass_check_data_deps): Replace declaration with that of...
1358         (make_pass_check_data_deps): ...new function.
1359         (pass_copy_prop): Replace declaration with that of...
1360         (make_pass_copy_prop): ...new function.
1361         (pass_vrp): Replace declaration with that of...
1362         (make_pass_vrp): ...new function.
1363         (pass_uncprop): Replace declaration with that of...
1364         (make_pass_uncprop): ...new function.
1365         (pass_return_slot): Replace declaration with that of...
1366         (make_pass_return_slot): ...new function.
1367         (pass_reassoc): Replace declaration with that of...
1368         (make_pass_reassoc): ...new function.
1369         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1370         (make_pass_rebuild_cgraph_edges): ...new function.
1371         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1372         (make_pass_remove_cgraph_callee_edges): ...new function.
1373         (pass_build_cgraph_edges): Replace declaration with that of...
1374         (make_pass_build_cgraph_edges): ...new function.
1375         (pass_local_pure_const): Replace declaration with that of...
1376         (make_pass_local_pure_const): ...new function.
1377         (pass_tracer): Replace declaration with that of...
1378         (make_pass_tracer): ...new function.
1379         (pass_warn_unused_result): Replace declaration with that of...
1380         (make_pass_warn_unused_result): ...new function.
1381         (pass_diagnose_tm_blocks): Replace declaration with that of...
1382         (make_pass_diagnose_tm_blocks): ...new function.
1383         (pass_lower_tm): Replace declaration with that of...
1384         (make_pass_lower_tm): ...new function.
1385         (pass_tm_init): Replace declaration with that of...
1386         (make_pass_tm_init): ...new function.
1387         (pass_tm_mark): Replace declaration with that of...
1388         (make_pass_tm_mark): ...new function.
1389         (pass_tm_memopt): Replace declaration with that of...
1390         (make_pass_tm_memopt): ...new function.
1391         (pass_tm_edges): Replace declaration with that of...
1392         (make_pass_tm_edges): ...new function.
1393         (pass_split_functions): Replace declaration with that of...
1394         (make_pass_split_functions): ...new function.
1395         (pass_feedback_split_functions): Replace declaration with that of...
1396         (make_pass_feedback_split_functions): ...new function.
1397         (pass_strength_reduction): Replace declaration with that of...
1398         (make_pass_strength_reduction): ...new function.
1399         (pass_ipa_lower_emutls): Replace declaration with that of...
1400         (make_pass_ipa_lower_emutls): ...new function.
1401         (pass_ipa_function_and_variable_visibility): Replace declaration with
1402         that of...
1403         (make_pass_ipa_function_and_variable_visibility): ...new function.
1404         (pass_ipa_tree_profile): Replace declaration with that of...
1405         (make_pass_ipa_tree_profile): ...new function.
1406         (pass_early_local_passes): Replace declaration with that of...
1407         (make_pass_early_local_passes): ...new function.
1408         (pass_ipa_whole_program_visibility): Replace declaration with that
1409         of...
1410         (make_pass_ipa_whole_program_visibility): ...new function.
1411         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1412         (make_pass_ipa_lto_gimple_out): ...new function.
1413         (pass_ipa_increase_alignment): Replace declaration with that of...
1414         (make_pass_ipa_increase_alignment): ...new function.
1415         (pass_ipa_inline): Replace declaration with that of...
1416         (make_pass_ipa_inline): ...new function.
1417         (pass_ipa_free_lang_data): Replace declaration with that of...
1418         (make_pass_ipa_free_lang_data): ...new function.
1419         (pass_ipa_free_inline_summary): Replace declaration with that of...
1420         (make_pass_ipa_free_inline_summary): ...new function.
1421         (pass_ipa_cp): Replace declaration with that of...
1422         (make_pass_ipa_cp): ...new function.
1423         (pass_ipa_reference): Replace declaration with that of...
1424         (make_pass_ipa_reference): ...new function.
1425         (pass_ipa_pure_const): Replace declaration with that of...
1426         (make_pass_ipa_pure_const): ...new function.
1427         (pass_ipa_pta): Replace declaration with that of...
1428         (make_pass_ipa_pta): ...new function.
1429         (pass_ipa_lto_finish_out): Replace declaration with that of...
1430         (make_pass_ipa_lto_finish_out): ...new function.
1431         (pass_ipa_tm): Replace declaration with that of...
1432         (make_pass_ipa_tm): ...new function.
1433         (pass_ipa_profile): Replace declaration with that of...
1434         (make_pass_ipa_profile): ...new function.
1435         (pass_ipa_cdtor_merge): Replace declaration with that of...
1436         (make_pass_ipa_cdtor_merge): ...new function.
1437         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1438         of...
1439         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1440         (pass_init_datastructures): Replace declaration with that of...
1441         (make_pass_init_datastructures): ...new function.
1442         (pass_fixup_cfg): Replace declaration with that of...
1443         (make_pass_fixup_cfg): ...new function.
1444         (pass_expand): Replace declaration with that of...
1445         (make_pass_expand): ...new function.
1446         (pass_instantiate_virtual_regs): Replace declaration with that of...
1447         (make_pass_instantiate_virtual_regs): ...new function.
1448         (pass_rtl_fwprop): Replace declaration with that of...
1449         (make_pass_rtl_fwprop): ...new function.
1450         (pass_rtl_fwprop_addr): Replace declaration with that of...
1451         (make_pass_rtl_fwprop_addr): ...new function.
1452         (pass_jump): Replace declaration with that of...
1453         (make_pass_jump): ...new function.
1454         (pass_jump2): Replace declaration with that of...
1455         (make_pass_jump2): ...new function.
1456         (pass_lower_subreg): Replace declaration with that of...
1457         (make_pass_lower_subreg): ...new function.
1458         (pass_cse): Replace declaration with that of...
1459         (make_pass_cse): ...new function.
1460         (pass_fast_rtl_dce): Replace declaration with that of...
1461         (make_pass_fast_rtl_dce): ...new function.
1462         (pass_ud_rtl_dce): Replace declaration with that of...
1463         (make_pass_ud_rtl_dce): ...new function.
1464         (pass_rtl_dce): Replace declaration with that of...
1465         (make_pass_rtl_dce): ...new function.
1466         (pass_rtl_dse1): Replace declaration with that of...
1467         (make_pass_rtl_dse1): ...new function.
1468         (pass_rtl_dse2): Replace declaration with that of...
1469         (make_pass_rtl_dse2): ...new function.
1470         (pass_rtl_dse3): Replace declaration with that of...
1471         (make_pass_rtl_dse3): ...new function.
1472         (pass_rtl_cprop): Replace declaration with that of...
1473         (make_pass_rtl_cprop): ...new function.
1474         (pass_rtl_pre): Replace declaration with that of...
1475         (make_pass_rtl_pre): ...new function.
1476         (pass_rtl_hoist): Replace declaration with that of...
1477         (make_pass_rtl_hoist): ...new function.
1478         (pass_rtl_store_motion): Replace declaration with that of...
1479         (make_pass_rtl_store_motion): ...new function.
1480         (pass_cse_after_global_opts): Replace declaration with that of...
1481         (make_pass_cse_after_global_opts): ...new function.
1482         (pass_rtl_ifcvt): Replace declaration with that of...
1483         (make_pass_rtl_ifcvt): ...new function.
1484         (pass_into_cfg_layout_mode): Replace declaration with that of...
1485         (make_pass_into_cfg_layout_mode): ...new function.
1486         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1487         (make_pass_outof_cfg_layout_mode): ...new function.
1488         (pass_loop2): Replace declaration with that of...
1489         (make_pass_loop2): ...new function.
1490         (pass_rtl_loop_init): Replace declaration with that of...
1491         (make_pass_rtl_loop_init): ...new function.
1492         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1493         (make_pass_rtl_move_loop_invariants): ...new function.
1494         (pass_rtl_unswitch): Replace declaration with that of...
1495         (make_pass_rtl_unswitch): ...new function.
1496         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1497         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1498         (pass_rtl_doloop): Replace declaration with that of...
1499         (make_pass_rtl_doloop): ...new function.
1500         (pass_rtl_loop_done): Replace declaration with that of...
1501         (make_pass_rtl_loop_done): ...new function.
1502         (pass_web): Replace declaration with that of...
1503         (make_pass_web): ...new function.
1504         (pass_cse2): Replace declaration with that of...
1505         (make_pass_cse2): ...new function.
1506         (pass_df_initialize_opt): Replace declaration with that of...
1507         (make_pass_df_initialize_opt): ...new function.
1508         (pass_df_initialize_no_opt): Replace declaration with that of...
1509         (make_pass_df_initialize_no_opt): ...new function.
1510         (pass_reginfo_init): Replace declaration with that of...
1511         (make_pass_reginfo_init): ...new function.
1512         (pass_inc_dec): Replace declaration with that of...
1513         (make_pass_inc_dec): ...new function.
1514         (pass_stack_ptr_mod): Replace declaration with that of...
1515         (make_pass_stack_ptr_mod): ...new function.
1516         (pass_initialize_regs): Replace declaration with that of...
1517         (make_pass_initialize_regs): ...new function.
1518         (pass_combine): Replace declaration with that of...
1519         (make_pass_combine): ...new function.
1520         (pass_if_after_combine): Replace declaration with that of...
1521         (make_pass_if_after_combine): ...new function.
1522         (pass_ree): Replace declaration with that of...
1523         (make_pass_ree): ...new function.
1524         (pass_partition_blocks): Replace declaration with that of...
1525         (make_pass_partition_blocks): ...new function.
1526         (pass_match_asm_constraints): Replace declaration with that of...
1527         (make_pass_match_asm_constraints): ...new function.
1528         (pass_regmove): Replace declaration with that of...
1529         (make_pass_regmove): ...new function.
1530         (pass_split_all_insns): Replace declaration with that of...
1531         (make_pass_split_all_insns): ...new function.
1532         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1533         (make_pass_fast_rtl_byte_dce): ...new function.
1534         (pass_lower_subreg2): Replace declaration with that of...
1535         (make_pass_lower_subreg2): ...new function.
1536         (pass_mode_switching): Replace declaration with that of...
1537         (make_pass_mode_switching): ...new function.
1538         (pass_sms): Replace declaration with that of...
1539         (make_pass_sms): ...new function.
1540         (pass_sched): Replace declaration with that of...
1541         (make_pass_sched): ...new function.
1542         (pass_ira): Replace declaration with that of...
1543         (make_pass_ira): ...new function.
1544         (pass_reload): Replace declaration with that of...
1545         (make_pass_reload): ...new function.
1546         (pass_clean_state): Replace declaration with that of...
1547         (make_pass_clean_state): ...new function.
1548         (pass_branch_prob): Replace declaration with that of...
1549         (make_pass_branch_prob): ...new function.
1550         (pass_value_profile_transformations): Replace declaration with that
1551         of...
1552         (make_pass_value_profile_transformations): ...new function.
1553         (pass_postreload_cse): Replace declaration with that of...
1554         (make_pass_postreload_cse): ...new function.
1555         (pass_gcse2): Replace declaration with that of...
1556         (make_pass_gcse2): ...new function.
1557         (pass_split_after_reload): Replace declaration with that of...
1558         (make_pass_split_after_reload): ...new function.
1559         (pass_branch_target_load_optimize1): Replace declaration with that
1560         of...
1561         (make_pass_branch_target_load_optimize1): ...new function.
1562         (pass_thread_prologue_and_epilogue): Replace declaration with that
1563         of...
1564         (make_pass_thread_prologue_and_epilogue): ...new function.
1565         (pass_stack_adjustments): Replace declaration with that of...
1566         (make_pass_stack_adjustments): ...new function.
1567         (pass_peephole2): Replace declaration with that of...
1568         (make_pass_peephole2): ...new function.
1569         (pass_if_after_reload): Replace declaration with that of...
1570         (make_pass_if_after_reload): ...new function.
1571         (pass_regrename): Replace declaration with that of...
1572         (make_pass_regrename): ...new function.
1573         (pass_cprop_hardreg): Replace declaration with that of...
1574         (make_pass_cprop_hardreg): ...new function.
1575         (pass_reorder_blocks): Replace declaration with that of...
1576         (make_pass_reorder_blocks): ...new function.
1577         (pass_branch_target_load_optimize2): Replace declaration with that
1578         of...
1579         (make_pass_branch_target_load_optimize2): ...new function.
1580         (pass_leaf_regs): Replace declaration with that of...
1581         (make_pass_leaf_regs): ...new function.
1582         (pass_split_before_sched2): Replace declaration with that of...
1583         (make_pass_split_before_sched2): ...new function.
1584         (pass_compare_elim_after_reload): Replace declaration with that of...
1585         (make_pass_compare_elim_after_reload): ...new function.
1586         (pass_sched2): Replace declaration with that of...
1587         (make_pass_sched2): ...new function.
1588         (pass_stack_regs): Replace declaration with that of...
1589         (make_pass_stack_regs): ...new function.
1590         (pass_stack_regs_run): Replace declaration with that of...
1591         (make_pass_stack_regs_run): ...new function.
1592         (pass_df_finish): Replace declaration with that of...
1593         (make_pass_df_finish): ...new function.
1594         (pass_compute_alignments): Replace declaration with that of...
1595         (make_pass_compute_alignments): ...new function.
1596         (pass_duplicate_computed_gotos): Replace declaration with that of...
1597         (make_pass_duplicate_computed_gotos): ...new function.
1598         (pass_variable_tracking): Replace declaration with that of...
1599         (make_pass_variable_tracking): ...new function.
1600         (pass_free_cfg): Replace declaration with that of...
1601         (make_pass_free_cfg): ...new function.
1602         (pass_machine_reorg): Replace declaration with that of...
1603         (make_pass_machine_reorg): ...new function.
1604         (pass_cleanup_barriers): Replace declaration with that of...
1605         (make_pass_cleanup_barriers): ...new function.
1606         (pass_delay_slots): Replace declaration with that of...
1607         (make_pass_delay_slots): ...new function.
1608         (pass_split_for_shorten_branches): Replace declaration with that of...
1609         (make_pass_split_for_shorten_branches): ...new function.
1610         (pass_split_before_regstack): Replace declaration with that of...
1611         (make_pass_split_before_regstack): ...new function.
1612         (pass_convert_to_eh_region_ranges): Replace declaration with that
1613         of...
1614         (make_pass_convert_to_eh_region_ranges): ...new function.
1615         (pass_shorten_branches): Replace declaration with that of...
1616         (make_pass_shorten_branches): ...new function.
1617         (pass_set_nothrow_function_flags): Replace declaration with that of...
1618         (make_pass_set_nothrow_function_flags): ...new function.
1619         (pass_dwarf2_frame): Replace declaration with that of...
1620         (make_pass_dwarf2_frame): ...new function.
1621         (pass_final): Replace declaration with that of...
1622         (make_pass_final): ...new function.
1623         (pass_rtl_seqabstr): Replace declaration with that of...
1624         (make_pass_rtl_seqabstr): ...new function.
1625         (pass_release_ssa_names): Replace declaration with that of...
1626         (make_pass_release_ssa_names): ...new function.
1627         (pass_early_inline): Replace declaration with that of...
1628         (make_pass_early_inline): ...new function.
1629         (pass_inline_parameters): Replace declaration with that of...
1630         (make_pass_inline_parameters): ...new function.
1631         (pass_update_address_taken): Replace declaration with that of...
1632         (make_pass_update_address_taken): ...new function.
1633         (pass_convert_switch): Replace declaration with that of...
1634         (make_pass_convert_switch): ...new function.
1635         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1636         to a subclass of simple_ipa_opt_pass along with...
1637         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1638         (make_pass_ipa_tree_profile): ...new function.
1639         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1640         subclass of gimple_opt_pass along with...
1641         (pass_data_sra_early): ...new pass_data instance and...
1642         (make_pass_sra_early): ...new function.
1643         (pass_sra): Convert from a global struct to a subclass of
1644         gimple_opt_pass along with...
1645         (pass_data_sra): ...new pass_data instance and...
1646         (make_pass_sra): ...new function.
1647         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1648         gimple_opt_pass along with...
1649         (pass_data_early_ipa_sra): ...new pass_data instance and...
1650         (make_pass_early_ipa_sra): ...new function.
1651         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1652         subclass of gimple_opt_pass along with...
1653         (pass_data_ccp): ...new pass_data instance and...
1654         (make_pass_ccp): ...new function.
1655         (pass_fold_builtins): Convert from a global struct to a subclass of
1656         gimple_opt_pass along with...
1657         (pass_data_fold_builtins): ...new pass_data instance and...
1658         (make_pass_fold_builtins): ...new function.
1659         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1660         subclass of gimple_opt_pass along with...
1661         (pass_data_copy_prop): ...new pass_data instance and...
1662         (make_pass_copy_prop): ...new function.
1663         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1664         global struct to a subclass of gimple_opt_pass along with...
1665         (pass_data_rename_ssa_copies): ...new pass_data instance and...
1666         (make_pass_rename_ssa_copies): ...new function.
1667         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1668         subclass of gimple_opt_pass along with...
1669         (pass_data_dce): ...new pass_data instance and...
1670         (make_pass_dce): ...new function.
1671         (pass_dce_loop): Convert from a global struct to a subclass of
1672         gimple_opt_pass along with...
1673         (pass_data_dce_loop): ...new pass_data instance and...
1674         (make_pass_dce_loop): ...new function.
1675         (pass_cd_dce): Convert from a global struct to a subclass of
1676         gimple_opt_pass along with...
1677         (pass_data_cd_dce): ...new pass_data instance and...
1678         (make_pass_cd_dce): ...new function.
1679         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1680         subclass of gimple_opt_pass along with...
1681         (pass_data_dominator): ...new pass_data instance and...
1682         (make_pass_dominator): ...new function.
1683         (pass_phi_only_cprop): Convert from a global struct to a subclass of
1684         gimple_opt_pass along with...
1685         (pass_data_phi_only_cprop): ...new pass_data instance and...
1686         (make_pass_phi_only_cprop): ...new function.
1687         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1688         subclass of gimple_opt_pass along with...
1689         (pass_data_dse): ...new pass_data instance and...
1690         (make_pass_dse): ...new function.
1691         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1692         a subclass of gimple_opt_pass along with...
1693         (pass_data_forwprop): ...new pass_data instance and...
1694         (make_pass_forwprop): ...new function.
1695         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1696         struct to a subclass of gimple_opt_pass along with...
1697         (pass_data_tree_ifcombine): ...new pass_data instance and...
1698         (make_pass_tree_ifcombine): ...new function.
1699         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1700         subclass of gimple_opt_pass along with...
1701         (pass_data_ch): ...new pass_data instance and...
1702         (make_pass_ch): ...new function.
1703         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1704         subclass of gimple_opt_pass along with...
1705         (pass_data_tree_loop): ...new pass_data instance and...
1706         (make_pass_tree_loop): ...new function.
1707         (pass_tree_loop_init): Convert from a global struct to a subclass of
1708         gimple_opt_pass along with...
1709         (pass_data_tree_loop_init): ...new pass_data instance and...
1710         (make_pass_tree_loop_init): ...new function.
1711         (pass_lim): Convert from a global struct to a subclass of
1712         gimple_opt_pass along with...
1713         (pass_data_lim): ...new pass_data instance and...
1714         (make_pass_lim): ...new function.
1715         (pass_tree_unswitch): Convert from a global struct to a subclass of
1716         gimple_opt_pass along with...
1717         (pass_data_tree_unswitch): ...new pass_data instance and...
1718         (make_pass_tree_unswitch): ...new function.
1719         (pass_predcom): Convert from a global struct to a subclass of
1720         gimple_opt_pass along with...
1721         (pass_data_predcom): ...new pass_data instance and...
1722         (make_pass_predcom): ...new function.
1723         (pass_vectorize): Convert from a global struct to a subclass of
1724         gimple_opt_pass along with...
1725         (pass_data_vectorize): ...new pass_data instance and...
1726         (make_pass_vectorize): ...new function.
1727         (pass_graphite): Convert from a global struct to a subclass of
1728         gimple_opt_pass along with...
1729         (pass_data_graphite): ...new pass_data instance and...
1730         (make_pass_graphite): ...new function.
1731         (pass_graphite_transforms): Convert from a global struct to a subclass
1732         of gimple_opt_pass along with...
1733         (pass_data_graphite_transforms): ...new pass_data instance and...
1734         (make_pass_graphite_transforms): ...new function.
1735         (pass_check_data_deps): Convert from a global struct to a subclass of
1736         gimple_opt_pass along with...
1737         (pass_data_check_data_deps): ...new pass_data instance and...
1738         (make_pass_check_data_deps): ...new function.
1739         (pass_iv_canon): Convert from a global struct to a subclass of
1740         gimple_opt_pass along with...
1741         (pass_data_iv_canon): ...new pass_data instance and...
1742         (make_pass_iv_canon): ...new function.
1743         (pass_scev_cprop): Convert from a global struct to a subclass of
1744         gimple_opt_pass along with...
1745         (pass_data_scev_cprop): ...new pass_data instance and...
1746         (make_pass_scev_cprop): ...new function.
1747         (pass_record_bounds): Convert from a global struct to a subclass of
1748         gimple_opt_pass along with...
1749         (pass_data_record_bounds): ...new pass_data instance and...
1750         (make_pass_record_bounds): ...new function.
1751         (pass_complete_unroll): Convert from a global struct to a subclass of
1752         gimple_opt_pass along with...
1753         (pass_data_complete_unroll): ...new pass_data instance and...
1754         (make_pass_complete_unroll): ...new function.
1755         (pass_complete_unrolli): Convert from a global struct to a subclass of
1756         gimple_opt_pass along with...
1757         (pass_data_complete_unrolli): ...new pass_data instance and...
1758         (make_pass_complete_unrolli): ...new function.
1759         (pass_parallelize_loops): Convert from a global struct to a subclass
1760         of gimple_opt_pass along with...
1761         (pass_data_parallelize_loops): ...new pass_data instance and...
1762         (make_pass_parallelize_loops): ...new function.
1763         (pass_loop_prefetch): Convert from a global struct to a subclass of
1764         gimple_opt_pass along with...
1765         (pass_data_loop_prefetch): ...new pass_data instance and...
1766         (make_pass_loop_prefetch): ...new function.
1767         (pass_iv_optimize): Convert from a global struct to a subclass of
1768         gimple_opt_pass along with...
1769         (pass_data_iv_optimize): ...new pass_data instance and...
1770         (make_pass_iv_optimize): ...new function.
1771         (pass_tree_loop_done): Convert from a global struct to a subclass of
1772         gimple_opt_pass along with...
1773         (pass_data_tree_loop_done): ...new pass_data instance and...
1774         (make_pass_tree_loop_done): ...new function.
1775         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1776         struct to a subclass of gimple_opt_pass along with...
1777         (pass_data_cse_reciprocals): ...new pass_data instance and...
1778         (make_pass_cse_reciprocals): ...new function.
1779         (pass_cse_sincos): Convert from a global struct to a subclass of
1780         gimple_opt_pass along with...
1781         (pass_data_cse_sincos): ...new pass_data instance and...
1782         (make_pass_cse_sincos): ...new function.
1783         (pass_optimize_bswap): Convert from a global struct to a subclass of
1784         gimple_opt_pass along with...
1785         (pass_data_optimize_bswap): ...new pass_data instance and...
1786         (make_pass_optimize_bswap): ...new function.
1787         (pass_optimize_widening_mul): Convert from a global struct to a
1788         subclass of gimple_opt_pass along with...
1789         (pass_data_optimize_widening_mul): ...new pass_data instance and...
1790         (make_pass_optimize_widening_mul): ...new function.
1791         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1792         subclass of gimple_opt_pass along with...
1793         (pass_data_phiopt): ...new pass_data instance and...
1794         (make_pass_phiopt): ...new function.
1795         (pass_cselim): Convert from a global struct to a subclass of
1796         gimple_opt_pass along with...
1797         (pass_data_cselim): ...new pass_data instance and...
1798         (make_pass_cselim): ...new function.
1799         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1800         subclass of gimple_opt_pass along with...
1801         (pass_data_phiprop): ...new pass_data instance and...
1802         (make_pass_phiprop): ...new function.
1803         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1804         subclass of gimple_opt_pass along with...
1805         (pass_data_pre): ...new pass_data instance and...
1806         (make_pass_pre): ...new function.
1807         (pass_fre): Convert from a global struct to a subclass of
1808         gimple_opt_pass along with...
1809         (pass_data_fre): ...new pass_data instance and...
1810         (make_pass_fre): ...new function.
1811         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1812         subclass of gimple_opt_pass along with...
1813         (pass_data_reassoc): ...new pass_data instance and...
1814         (make_pass_reassoc): ...new function.
1815         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1816         subclass of gimple_opt_pass along with...
1817         (pass_data_sink_code): ...new pass_data instance and...
1818         (make_pass_sink_code): ...new function.
1819         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1820         subclass of gimple_opt_pass along with...
1821         (pass_data_strlen): ...new pass_data instance and...
1822         (make_pass_strlen): ...new function.
1823         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1824         struct to a subclass of gimple_opt_pass along with...
1825         (pass_data_build_alias): ...new pass_data instance and...
1826         (make_pass_build_alias): ...new function.
1827         (pass_build_ealias): Convert from a global struct to a subclass of
1828         gimple_opt_pass along with...
1829         (pass_data_build_ealias): ...new pass_data instance and...
1830         (make_pass_build_ealias): ...new function.
1831         (pass_ipa_pta): Convert from a global struct to a subclass of
1832         simple_ipa_opt_pass along with...
1833         (pass_data_ipa_pta): ...new pass_data instance and...
1834         (make_pass_ipa_pta): ...new function.
1835         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1836         subclass of gimple_opt_pass along with...
1837         (pass_data_uncprop): ...new pass_data instance and...
1838         (make_pass_uncprop): ...new function.
1839         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1840         global struct to a subclass of gimple_opt_pass along with...
1841         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1842         (make_pass_late_warn_uninitialized): ...new function.
1843         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1844         to a subclass of gimple_opt_pass along with...
1845         (pass_data_init_datastructures): ...new pass_data instance and...
1846         (make_pass_init_datastructures): ...new function.
1847         (pass_early_warn_uninitialized): Convert from a global struct to a
1848         subclass of gimple_opt_pass along with...
1849         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1850         (make_pass_early_warn_uninitialized): ...new function.
1851         (pass_update_address_taken): Convert from a global struct to a
1852         subclass of gimple_opt_pass along with...
1853         (pass_data_update_address_taken): ...new pass_data instance and...
1854         (make_pass_update_address_taken): ...new function.
1855         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1856         struct to a subclass of gimple_opt_pass along with...
1857         (pass_data_release_ssa_names): ...new pass_data instance and...
1858         (make_pass_release_ssa_names): ...new function.
1859         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1860         subclass of gimple_opt_pass along with...
1861         (pass_data_stdarg): ...new pass_data instance and...
1862         (make_pass_stdarg): ...new function.
1863         * tree-switch-conversion.c (pass_convert_switch): Convert from a
1864         global struct to a subclass of gimple_opt_pass along with...
1865         (pass_data_convert_switch): ...new pass_data instance and...
1866         (make_pass_convert_switch): ...new function.
1867         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1868         to a subclass of gimple_opt_pass along with...
1869         (pass_data_tail_recursion): ...new pass_data instance and...
1870         (make_pass_tail_recursion): ...new function.
1871         (pass_tail_calls): Convert from a global struct to a subclass of
1872         gimple_opt_pass along with...
1873         (pass_data_tail_calls): ...new pass_data instance and...
1874         (make_pass_tail_calls): ...new function.
1875         * tree-vect-generic.c (pass_lower_vector): Convert from a global
1876         struct to a subclass of gimple_opt_pass along with...
1877         (pass_data_lower_vector): ...new pass_data instance and...
1878         (make_pass_lower_vector): ...new function.
1879         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1880         gimple_opt_pass along with...
1881         (pass_data_lower_vector_ssa): ...new pass_data instance and...
1882         (make_pass_lower_vector_ssa): ...new function.
1883         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1884         to a subclass of gimple_opt_pass along with...
1885         (pass_data_slp_vectorize): ...new pass_data instance and...
1886         (make_pass_slp_vectorize): ...new function.
1887         (pass_ipa_increase_alignment): Convert from a global struct to a
1888         subclass of simple_ipa_opt_pass along with...
1889         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1890         (make_pass_ipa_increase_alignment): ...new function.
1891         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1892         gimple_opt_pass along with...
1893         (pass_data_vrp): ...new pass_data instance and...
1894         (make_pass_vrp): ...new function.
1895         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1896         subclass of simple_ipa_opt_pass along with...
1897         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1898         (make_pass_ipa_free_lang_data): ...new function.
1899         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1900         gimple_opt_pass along with...
1901         (pass_data_tsan): ...new pass_data instance and...
1902         (make_pass_tsan): ...new function.
1903         (pass_tsan_O0): Convert from a global struct to a subclass of
1904         gimple_opt_pass along with...
1905         (pass_data_tsan_O0): ...new pass_data instance and...
1906         (make_pass_tsan_O0): ...new function.
1907         * var-tracking.c (pass_variable_tracking): Convert from a global
1908         struct to a subclass of rtl_opt_pass along with...
1909         (pass_data_variable_tracking): ...new pass_data instance and...
1910         (make_pass_variable_tracking): ...new function.
1911         * web.c (pass_web): Convert from a global struct to a subclass of
1912         rtl_opt_pass along with...
1913         (pass_data_web): ...new pass_data instance and...
1914         (make_pass_web): ...new function.
1915         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1916         declaration with that of...
1917         (make_pass_mode_switch_use): ...new function.
1918         (pass_resolve_sw_modes): Replace declaration with that of...
1919         (make_pass_resolve_sw_modes): ...new function.
1920         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1921         from a global struct to a subclass of rtl_opt_pass along with...
1922         (pass_data_mode_switch_use): ...new pass_data instance and...
1923         (make_pass_mode_switch_use): ...new function.
1924         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1925         from a global struct to a subclass of rtl_opt_pass along with...
1926         (pass_data_resolve_sw_modes): ...new pass_data instance and...
1927         (make_pass_resolve_sw_modes): ...new function.
1928         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1929         struct to a subclass of rtl_opt_pass along with...
1930         (pass_data_insert_vzeroupper): ...new pass_data instance and...
1931         (make_pass_insert_vzeroupper): ...new function.
1932         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1933         global struct to a subclass of rtl_opt_pass along with...
1934         (pass_data_work_around_errata): ...new pass_data instance and...
1935         (make_pass_work_around_errata): ...new function.
1936         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1937         struct to a subclass of rtl_opt_pass along with...
1938         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1939         (make_pass_mips_machine_reorg2): ...new function.
1941 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1943         * passes.c (pass_manager::operator new): New.
1945 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1947         Handwritten part of conversion of passes to C++ classes.
1949         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1950         (toplev.o): Add dep on PASS_MANAGER_H.
1951         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1952         of early local pases to reflect this moving from a global to a
1953         member of gcc::pass_manager.
1954         (cgraph_add_new_function): Likewise.
1955         * lto-cgraph.c (lto_output_node): Update for conversion of
1956         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1957         * passes.c (opt_pass::clone): New.
1958         (opt_pass::gate): New.
1959         (opt_pass::execute): New.
1960         (opt_pass::opt_pass): New.
1961         (pass_manager::execute_early_local_passes): New.
1962         (pass_manager::execute_pass_mode_switching): new.
1963         (finish_optimization_passes): Convert to...
1964         (pass_manager::finish_optimization_passes): ...this.
1965         (finish_optimization_passes): Update for conversion of passes to
1966         C++ classes.
1967         (register_dump_files_1): Use has_gate since we cannot portably
1968         check a vtable entry against NULL.
1969         (dump_one_pass): Likewise.
1970         (ipa_write_summaries_2): Likewise.
1971         (ipa_write_optimization_summaries_1): Likewise.
1972         (ipa_read_summaries_1): Likewise.
1973         (ipa_read_optimization_summaries_1): Likewise.
1974         (execute_ipa_stmt_fixups): Likewise.
1975         (pass_manager::pass_manager): Rewrite pass-creation, invoking
1976         pass-creation functions rather than wiring up globals, and
1977         storing the results in fields of pass_manager generated using
1978         pass-instances.def.
1979         (pass_manager::dump_profile_report): Update for conversion of
1980         passes to C++ classes.
1981         (pass_manager::execute_ipa_summary_passes): Likewise.
1982         (execute_one_ipa_transform_pass): Likewise.
1983         (execute_one_pass): Use has_gate and has_execute since we cannot
1984         portably check a vtable entry against NULL.
1985         * pass_manager.h (pass_manager::finish_optimization_passes): New.
1986         (pass_manager): Use pass-instances.def to add fields for the
1987         various pass instances.
1988         * toplev.c (finalize): Update for move of
1989         finish_optimization_passes to a method of gcc::pass_manager.
1990         * toplev.h (finish_optimization_passes): Move to method of class
1991         pass_manager.
1992         * tree-pass.h (struct pass_data): New.
1993         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
1994         (opt_pass::gate): Convert to virtual function.
1995         (opt_pass::~opt_pass): New.
1996         (opt_pass::clone): New.
1997         (opt_pass::execute): Convert to virtual function.
1998         (opt_pass::opt_pass): New.
1999         (opt_pass::ctxt_): new.
2000         (gimple_opt_pass): Convert to subclass of opt_pass.
2001         (gimple_opt_pass::gimple_opt_pass): New.
2002         (rtl_opt_pass): Convert to subclass of opt_pass.
2003         (rtl_opt_pass::rtl_opt_pass): New.
2004         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2005         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2006         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2007         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2008         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2009         invocation of pass_mode_switching to reflect this moving from a
2010         global to a member of gcc::pass_manager.
2011         (ix86_option_override): Rework how pass_insert_vzeroupper is
2012         added to the pass_manager to reflect autogenerated changes.
2013         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2015 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2017         PR rtl-optimization/57708
2018         * recog.c (peep2_find_free_register): Validate all regs in a
2019         multi-reg mode.
2021 2013-08-05  Jan Hubicka  <jh@suse.cz>
2023         PR lto/57602
2024         * cgraph.c (verify_cgraph_node): Accept local flags from other
2025         partitions.
2026         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2027         (function_and_variable_visibility): Likewise.
2028         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2030 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2032         * graph.c (init_graph_slim_pretty_print): Remove.
2033         (print_graph_cfg): Do not call it.  Use local pretty printer.
2034         (start_graph_dump): Likewise.
2036 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2038         * gimple-pretty-print.c (buffer): Remove.
2039         (initialized): Likewise.
2040         (maybe_init_pretty_print): Likewise.
2041         (print_gimple_stmt): Do not call it.  Use non-static local
2042         pretty_printer variable.
2043         (print_gimple_expr): Likewise.
2044         (print_gimple_seq): Likewise.
2045         (gimple_dump_bb): Likewise.
2047 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2049         * asan.c (asan_pp): Remove.
2050         (asan_pp_initialized): Likewise.
2051         (asan_pp_initialize): Likewise.
2052         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2053         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2054         (asan_add_global): Likewise.
2056 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2058         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2059         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2060         * pretty-print.c (pp_formatted_text_data): Likewise.
2061         (pp_indent): Rename from pp_base_indent.
2062         (pp_format): Rename from pp_base_format.
2063         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2064         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2065         (pp_flush): Rename from pp_base_flush.
2066         (pp_set_line_maximum_length): Rename from
2067         pp_base_set_line_maximum_length.
2068         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2069         (pp_set_prefix): Rename from pp_base_set_prefix.
2070         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2071         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2072         (pp_append_text): Rename from pp_base_append_text.
2073         (pp_formatted_text): Rename from pp_base_formatted_text.
2074         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2075         (pp_remaining_character_count_for_line): Rename from
2076         pp_base_remaining_character_count_for_line.
2077         (pp_newline): Rename from pp_base_newline.
2078         (pp_character): Rename from pp_base_character.
2079         (pp_string): Rename from pp_base_string.
2080         (pp_maybe_space): Rename from pp_base_maybe_space.
2081         * asan.c (asan_pp_string): Adjust.
2082         (asan_emit_stack_protection): Likewise.
2083         (asan_add_global): Likewise.
2084         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2085         * tree-mudflap.c (mf_varname_tree): Likewise.
2086         * tree-pretty-print.c (pp_tree_identifier): Rename from
2087         pp_base_tree_identifier.
2088         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2089         Declare as function.
2091 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2093         * pretty-print.h (pp_bar_bar): New.
2094         (pp_ampersand_ampersand): Likewise.
2095         (pp_less_equal): Likewise.
2096         (pp_greater_equal): Likewise.
2097         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2098         printer functions instead of pp_string or operators and punctuators.
2099         (dump_gimple_call): Likewise.
2100         (dump_gimple_omp_for): Likewise.
2101         (dump_gimple_transaction): Likewise.
2102         (dump_gimple_phi): Likewise.
2103         (pp_gimple_stmt_1): Likewise.
2104         * sched-vis.c (print_insn): Likewise.
2105         * tree-mudflap.c (mf_varname_tree): Likewise.
2106         * tree-pretty-print.c (dump_block_node): Likewise.
2107         (dump_generic_node): Likewise.
2109 2013-08-02  Jan Hubicka  <jh@suse.cz>
2111         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2112         boundaries.
2113         * lto-streamer-out.c (tree_is_indexable): Results decls and
2114         parm decls are not indexable.
2115         (DFS_write_tree_body): Do not follow args and results.
2116         (hash_tree): Likewise.
2117         (output_functions): Rearrange so struct function is needed
2118         only when real body is output; be able to also ouptut abstract
2119         functions; output DECL_ARGUMENTS and DECL_RESULT.
2120         (lto_output): When not in WPA, ale store abstract functions.
2121         (write_symbol): Do not care about RESULT_DECL.
2122         (output_symbol_p): Handle correctly sbtract decls.
2123         * lto-streamer-in.c (input_function): Rearrange so struct
2124         function can be NULL at entry; allow streaming of
2125         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2126         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2127         sanity check during LTO.
2128         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2129         RESULT_DECl and DECL_ARGUMENTS.
2130         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2131         Likewise.
2133 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2135         * pretty-print.h (pp_underscore): New.
2136         (pp_comma): Tidy.
2137         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2138         printer functions instead of pp_character.
2139         (dump_binary_rhs): Likewise.
2140         (dump_ternary_rhs): Likewise.
2141         (dump_gimple_call_args): Likewise.
2142         (pp_points_to_solution): Likewise.
2143         (dump_gimple_call): Likewise.
2144         (dump_gimple_switch): Likewise.
2145         (dump_gimple_cond): Likewise.
2146         (dump_gimple_bind): Likewise.
2147         (dump_gimple_try): Likewise.
2148         (dump_gimple_omp_for): Likewise.
2149         (dump_gimple_omp_continue): Likewise.
2150         (dump_gimple_omp_single): Likewise.
2151         (dump_gimple_omp_sections): Likewise.
2152         (dump_gimple_omp_block): Likewise.
2153         (dump_gimple_omp_critical): Likewise.
2154         (dump_gimple_transaction): Likewise.
2155         (dump_gimple_asm): Likewise.
2156         (dump_gimple_phi): Likewise.
2157         (dump_gimple_omp_parallel): Likewise.
2158         (dump_gimple_omp_task): Likewise.
2159         (dump_gimple_omp_atomic_load): Likewise.
2160         (dump_gimple_omp_atomic_store): Likewise.
2161         (dump_gimple_mem_ops): Likewise.
2162         (pp_gimple_stmt_1): Likewise.
2163         (pp_cfg_jump): Likewise.
2164         (dump_implicit_edges): Likewise.
2165         (gimple_dump_bb_for_graph): Likewise.
2166         * graph.c (draw_cfg_node): Likewise.
2167         * langhooks.c (lhd_print_error_function): Likewise.
2168         * sched-vis.c (print_exp): Likewise.
2169         (print_value): Likewise.
2170         (print_pattern): Likewise.
2171         (print_insn): Likewise.
2172         (rtl_dump_bb_for_graph): Likewise.
2173         * tree-pretty-print.c (dump_function_declaration): Likewise.
2174         (dump_array_domain): Likewise.
2175         (dump_omp_clause): Likewise.
2176         (dump_location): Likewise.
2177         (dump_generic_node): Likewise.
2178         (print_struct_decl): Likewise.
2179         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2181 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2183         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2184         candidate table when replacing a candidate statement.
2185         (replace_rhs_if_not_dup): Likewise.
2186         (replace_one_candidate): Likewise.
2188 2013-08-02  Jan Hubicka  <jh@suse.cz>
2189             Martin Liska  <marxin.liska@gmail.com>
2191         * cgraphunit.c (add_new_function): Fix logic when adding from
2192         late IPA pass.
2193         (assemble_thunk): Rename to ...
2194         (expand_thunk); .. this one; export; get it working with
2195         general functions; make produced gimple valid.
2196         * cgraph.h (expand_thunk): Declare.
2198 2013-08-02  Jan Hubicka  <jh@suse.cz>
2200         * ipa-cp.c (gather_context_independent_values): Use
2201         ipa_get_param_move_cost.
2202         (get_replacement_map): Remove PARAM; move parameter folding
2203         into tree-inline.c
2204         (create_specialized_node): Update.
2205         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2206         assert that we have gimple body; update move_cost.
2207         (count_formal_params): Assert that we have gimple body.
2208         (ipa_dump_param): New function.
2209         (ipa_alloc_node_params): Break out from ...
2210         (ipa_initialize_node_params): ... here.
2211         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2212         (ipa_write_node_info): Stream move costs.
2213         (ipa_read_node_info): Read move costs.
2214         (ipa_update_after_lto_read): Do not recompute node params.
2215         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2216         (ipa_get_param): Check we are not in WPA.
2217         (ipa_get_param_move_cost): New.
2218         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2219         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2220         parm numbers to be present.
2222 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2224         PR rtl-optimization/58048
2225         * lra-constraints.c (process_alt_operands): Don't check asm
2226         operand on register.
2228 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2230         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2231         the implied StoreLoad barrier for atomic operations if before.
2233 2013-08-02  Jan Hubicka  <jh@suse.cz>
2234             Martin Liska  <marxin.liska@gmail.com>
2236         * cgraph.c (cgraph_function_body_availability): Do not check
2237         cgraph flags.
2238         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2239         symtab_node_availability): Declare.
2240         * ipa.c (can_replace_by_local_alias): New.
2241         (function_and_variable_visibility): Use it.
2242         * symtab.c (symtab_for_node_and_aliases,
2243         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2245 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2247         PR rtl-optimization/57963
2248         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2249         (lra_constraints): Use them.
2251 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2253         * config/arm/types.md (define_attr "type"): Add "load_acq"
2254         and "store_rel".
2255         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2256         changes.
2257         (cortex_a53_store1): Likewise.
2259 2013-08-01  Jan Hubicka  <jh@suse.cz>
2261         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2262         partitions are not needed.
2264 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2266         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2267         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2268         MAYBE_NON_Q_CLASS_P where appropriate.
2270 2013-08-01  Jan Hubicka  <jh@suse.cz>
2272         * cgraph.h (release_function_body): Declare.
2273         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2274         of unused delcarations.
2276 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2278         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2279         RTL form when subtracting a constant.
2281 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2283         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2284         Generate canonical plus rtx with negated immediate instead of minus
2285         where appropriate.
2286         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2288 2013-08-01  Jan Hubicka  <jh@suse.cz>
2290         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2291         (cgraph_release_function_body): Likewise.
2292         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2293         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2294         to used_as_abstract_origin.
2295         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2296         symbols used as abstract origins.
2297         * cgraphunit.c (analyze_functions): Update.
2298         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2299         used_as_abstract_origin.
2300         * tree-inline.c (tree_function_versioning): Update
2301         used_as_abstract_origin; be ready for DECL_RESULT and
2302         DECL_ARGUMENTS to be NULL.
2304         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2305         for abstract functions.
2306         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2307         real symbols.
2309 2013-08-01  Jan Hubicka  <jh@suse.cz>
2311         * profile.c (compute_value_histograms): Fix thinko.
2313 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2315         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2316         aarch-common-protos.h to extra_headers.
2317         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2318         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2319         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2321 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2323         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2324         Include "../arm/types.md".  Define "type" attribute for all patterns.
2325         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2326         attribute changes.
2328 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2330         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2331         to support power8 load fusion.
2332         (fusion_gpr_mem_load): Likewise.
2334         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2336         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2337         declarations for power8 load fusion.
2338         (emit_fusion_gpr_load): Likewise.
2340         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2341         tuning for power8, turn on fusion mode by default.  Turn on sign
2342         extending fusion mode if normal fusion mode is on, and we are at
2343         -O2 or -O3.
2344         (fusion_gpr_load_p): New function, return true if we can fuse an
2345         addis instruction with a dependent load to a GPR.
2346         (emit_fusion_gpr_load): Emit the instructions for power8 load
2347         fusion to GPRs.
2349         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2350         (VSX load fusion peepholes): New peepholes to fuse together an
2351         addi instruction with a VSX load instruction.
2353         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2354         peepholes to fuse an addis instruction with a load to a GPR base
2355         register.  If we are supporting sign extending fusions, convert
2356         sign extending loads to zero extending loads and add an explicit
2357         sign extension.
2359 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2361         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2362         aarch-common-protos.h to extra_headers.
2363         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2364         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2365         (arm_early_store_addr_dep): Likewise.
2366         (arm_no_early_alu_shift_dep): Likewise.
2367         (arm_no_early_alu_shift_value_dep): Likewise.
2368         (arm_no_early_mul_dep): Likewise.
2369         (arm_no_early_store_addr_dep): Likewise.
2370         (arm_mac_accumulator_is_mul_result): Likewise.
2371         (arm_mac_accumulator_is_result): Likewise.
2372         * config/arm/aarch-common.c: ... here.  New file.
2373         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2374         here to ...
2375         (arm_early_store_addr_dep): Likewise.
2376         (arm_no_early_alu_shift_dep): Likewise.
2377         (arm_no_early_alu_shift_value_dep): Likewise.
2378         (arm_no_early_mul_dep): Likewise.
2379         (arm_no_early_store_addr_dep): Likewise.
2380         (arm_mac_accumulator_is_mul_result): Likewise.
2381         (arm_mac_accumulator_is_result): Likewise.
2382         * config/arm/aarch-common-protos.h: ... here.  New file.
2383         * config/arm/t-arm (aarch-common.o): Define.
2385 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2387         * config/arm/arm.md: Include new file "types.md".
2388         (define_attr "type"): Move from here to ...
2389         (define_attr "mul32"): Likewise.
2390         (define_attr "mul64"): Likewise.
2391         * config/arm/types.md: ... here.  New file.
2393 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2395         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2396         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2398 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2400         * gen-pass-instances.awk: Fix offset of substr().
2402 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2404         * Makefile.in (pass-instances.def): New.
2405         (passes.o): Replace dependency on passes.def with one on
2406         pass-instances.def
2408         * gen-pass-instances.awk: New.
2410         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2411         rather than passes.def, updating local definition of NEXT_PASS
2412         macro to add an extra NUM parameter (currently unused).
2414 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2416         * Makefile.in (PASS_MANAGER_H): New.
2417         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2418         (passes.o): Likewise.
2419         (statistics.o): Likewise.
2420         (cgraphunit.o): Likewise.
2421         (context.o): Depend on PASS_MANAGER_H.
2423         * pass_manager.h: New.
2425         * cgraphunit.c (cgraph_add_new_function): Update for moves
2426         of globals to fields of pass_manager.
2427         (analyze_function): Likewise.
2428         (expand_function): Likewise.
2429         (ipa_passes): Likewise.
2430         (compile): Likewise.
2432         * context.c (context::context): New.
2433         * context.h  (context::context): New.
2434         (context::get_passes): New.
2435         (context::passes_): New.
2437         * lto-cgraph.c (input_node): Update for moves of globals to
2438         fields of pass_manager.
2440         * passes.c (all_passes): Remove, in favor of a field of the
2441         same name within the new class pass_manager.
2442         (all_small_ipa_passes): Likewise.
2443         (all_lowering_passes): Likewise.
2444         (all_regular_ipa_passes): Likewise.
2445         (all_late_ipa_passes): Likewise.
2446         (all_lto_gen_passes): Likewise.
2447         (passes_by_id): Likewise.
2448         (passes_by_id_size): Likewise.
2449         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2450         the new class pass_manager.
2451         (set_pass_for_id): Convert to...
2452         (pass_manager::set_pass_for_id): ...method.
2453         (get_pass_for_id): Convert to...
2454         (pass_manager::get_pass_for_id): ...method.
2455         (register_one_dump_file): Move body of implementation into...
2456         (pass_manager::register_one_dump_file): ...here.
2457         (register_dump_files_1): Convert to...
2458         (pass_manager::register_dump_files_1): ...method.
2459         (register_dump_files): Convert to...
2460         (pass_manager::register_dump_files): ...method.
2461         (create_pass_tab): Update for moves of globals to fields of
2462         pass_manager.
2463         (dump_passes): Move body of implementation into...
2464         (pass_manager::dump_passes): ...here.
2465         (register_pass): Move body of implementation into...
2466         (pass_manager::register_pass): ...here.
2467         (init_optimization_passes): Convert into...
2468         (pass_manager::pass_manager): ...constructor for new
2469         pass_manager class, and initialize the pass_lists array.
2470         (check_profile_consistency): Update for moves of globals to
2471         fields of pass_manager.
2472         (dump_profile_report): Move body of implementation into...
2473         (pass_manager::dump_profile_report): ...here.
2474         (ipa_write_summaries_1): Update for moves of pass lists from
2475         being globals to fields of pass_manager.
2476         (ipa_write_optimization_summaries): Likewise.
2477         (ipa_read_summaries):  Likewise.
2478         (ipa_read_optimization_summaries): Likewise.
2479         (execute_all_ipa_stmt_fixups): Likewise.
2481         * statistics.c (statistics_fini): Update for moves of globals to
2482         fields of pass_manager.
2484         * toplev.c (general_init): Replace call to
2485         init_optimization_passes with construction of the pass_manager
2486         instance.
2488         * tree-pass.h (all_passes): Remove, in favor of a field of the
2489         same name within the new class pass_manager.
2490         (all_small_ipa_passes): Likewise.
2491         (all_lowering_passes): Likewise.
2492         (all_regular_ipa_passes): Likewise.
2493         (all_lto_gen_passes): Likewise.
2494         (all_late_ipa_passes): Likewise.
2495         (passes_by_id): Likewise.
2496         (passes_by_id_size): Likewise.
2497         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2498         the new class pass_manager.
2499         (get_pass_for_id): Remove.
2501 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2503         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2504         configs.
2506 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2508         * arm.md (mulhi3): New expand pattern.
2510 2013-07-30  Jan Hubicka  <jh@suse.cz>
2511             Martin Liska  <marxin.liska@gmail.com>
2513         * profile.c (compute_value_histograms): Do not ICE when
2514         there is mismatch only on some counters.
2516 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2518         PR rtl-optimization/57637
2519         * function.c (move_insn_for_shrink_wrap): Also check the
2520         GEN set of the LIVE problem for the liveness analysis
2521         if it exists, otherwise give up.
2523 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2525         PR tree-optimization/57993
2526         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2527         replaced statement in the candidate table.
2528         (phi_add_costs): Return infinite cost when the hidden basis does
2529         not dominate all phis on which the candidate is dependent.
2530         (replace_one_candidate): Record replaced statement in the
2531         candidate table.
2533 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2535         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2536         (ashlv2si3): New expander.
2537         (*ashlv2si3_i): New define_insn_and_split.
2538         * predicates.md (float_operation): Allow patterns with three
2539         basic sub-patterns.
2541         PR rtl-optimization/58021
2542         * mode-switching.c (create_pre_exit): Always split off preceding
2543         insns if we are not at the basic block head.
2545 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2547         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2548         (UCLIBC_DYNAMIC_LINKER): New macro.
2549         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2550         `-mnan=2008'.
2551         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2552         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2553         `-mnan=2008'.
2554         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2555         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2556         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2557         for SF and DF modes.  Use ieee_quad_format for TF mode.
2558         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2559         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2560         (mips_option_override): Handle `-mnan=legacy'.
2561         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2562         `-mabs=2008' and `-mnan=2008'.
2563         (OPTION_DEFAULT_SPECS): Add "nan" default.
2564         (ASM_SPEC): Handle `-mnan='.
2565         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2566         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2567         comment accordingly.
2568         (neg<mode>2): Likewise.
2569         * config/mips/mips.opt (mabs, mnan): New options.
2570         * doc/install.texi (Configuration): Document `--with-nan=' option.
2571         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2572         `-mnan=' options.
2573         (MIPS Options): Document them.
2574         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2575         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2576         * configure: Regenerate.
2577         * config.in: Regenerate.
2579 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2581         * config/i386/i386.md (float post-reload splitters): Do not check
2582         for subregs of SSE registers.
2584 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2585             H.J. Lu  <hongjiu.lu@intel.com>
2587         PR target/57954
2588         PR target/57988
2589         * config/i386/i386.md (post-reload splitter
2590         to avoid partial SSE reg dependency stalls): New pattern.
2592 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2594         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2596 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2598         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2599         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2601 2013-07-26  Cary Coutant  <ccoutant@google.com>
2603         * dwarf2out.c (die_checksum_ordered): Don't include template
2604         instantiations in signature.
2605         (is_template_parameter): New function.
2606         (is_template_instantiation): New function.
2607         (generate_skeleton_bottom_up): Don't include template instantiations
2608         in type unit DIE.
2609         (generate_skeleton): Likewise.
2610         (break_out_comdat_types): Move recursive call to break out nested
2611         types earlier.
2612         (prune_unused_types_mark_generic_parms_dies): Call
2613         is_template_parameter.
2615 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2617         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2618         uses vector registers.
2619         * config/aarch64/iterators.md: Add attributes rtn and vas.
2621 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2622             Richard Earnshaw  <richard.earnshaw@arm.com>
2624         * combine.c (simplify_comparison): Re-canonicalize operands
2625         where appropriate.
2626         * config/arm/arm.md (movcond_addsi): New splitter.
2628 2013-07-25  Sterling Augustine  <saugustine@google.com>
2630         * dwarf2out.c (size_of_pubnames): Move code to...
2631         (include_pubname_in_output): ...here.  New.
2632         (want_pubnames): Rearrange.
2633         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2635 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2637         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2639 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2641         PR target/38836
2642         * doc/extend.texi: Remove obsolete builtins. Fix
2643         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2645 2013-07-25  Jan Hubicka  <jh@suse.cz>
2647         * cgraph.c (release_function_body): Break out from ...
2648         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2649         and DECL_ARGUMENTS.
2650         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2651         old_tree in the map.
2652         (create_specialized_node): Update.
2653         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2654         into index.
2655         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2656         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2657         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2658         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2659         DECL_RESULT.
2661 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2663         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2664         addsi3_carryin_alt2_<optab>): Correct output template.
2666 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2668         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2669         Adjust for arm_restrict_it.
2670         Remove trailing whitespace.
2672 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
2674         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2675         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2677         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2679 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
2681         PR rtl-optimization/57960
2682         * lra-constraints.c (process_alt_operands): Use the right mode
2683         when checking strict_low.
2685 2013-07-25  Jan Hubicka  <jh@suse.cz>
2687         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2688         * cgraph.c (cgraph_remove_node): Do not release function body
2689         when in cgraph streaming.
2690         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2691         in other partitions are not considered reachable; fix handling of
2692         clones.
2694 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2696         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2698 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2700         PR target/19599
2701         PR target/57731
2702         PR target/57837
2703         * config/arm/arm.md ("*sibcall_insn): Replace use of
2704         Ss with US. Adjust output for v5 and v4t.
2705         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2707         * config/arm/constraints.md ("Ss"): Rename to US.
2709 2013-07-25  Terry Guo  <terry.guo@arm.com>
2711         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2712         shift_add/shift_sub0/shift_sub1 RTXs.
2714 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
2715             Anton Blanchard <anton@au1.ibm.com>
2717         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2718         (altivec_vpks<VI_char>ss): Likewise.
2719         (altivec_vpks<VI_char>us): Likewise.
2720         (altivec_vpku<VI_char>us): Likewise.
2721         (altivec_vpku<VI_char>um): Likewise.
2723 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
2725         Introduce context class.
2727         * Makefile.in (CONTEXT_H): New.
2728         (OBJS): Add context.o.
2729         (toplev.o): Add CONTEXT_H to dependencies.
2730         (context.o): New.
2732         * toplev.c (general_init): Create the singleton gcc::context instance.
2734         * context.c: New.
2736         * context.h: New.
2738 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2740         PR rtl-optimization/57968
2741         * mode-switching.c (create_pre_exit): Allow instructions that
2742         don't set a return register to need a non-exit mode.
2744 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2745             Anton Blanchard <anton@au1.ibm.com>
2747         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2748         operands to vperm for little endian.
2749         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2750         of lvsl to create the control mask for a vperm for little endian.
2752 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2753             Anton Blanchard <anton@au1.ibm.com>
2755         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2756         two operands for little-endian.
2758 2013-07-23  Steve Ellcey  <sellcey@mips.com>
2760         * config/mips/mips.c (mips_case_values_threshold): New.
2761         (TARGET_CASE_VALUES_THRESHOLD): Define.
2763 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2764             Anton Blanchard <anton@au1.ibm.com>
2766         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2767         selection of field for vector splat in little endian mode.
2769 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
2771         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2772         expanders to rs6000.md.
2773         (ior<mode>3): Likewise.
2774         (and<mode>3): Likewise.
2775         (one_cmpl<mode>2): Likewise.
2776         (nor<mode>3): Likewise.
2777         (andc<mode>3): Likewise.
2778         (eqv<mode>3): Likewise.
2779         (nand<mode>3): Likewise.
2780         (orc<mode>3): Likewise.
2782         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2783         declaration.
2785         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2786         to split multi-word logical operations.
2787         (rs6000_split_logical_di): Likewise.
2788         (rs6000_split_logical): Likewise.
2790         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2791         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2792         and allow TImode operations in 32-bit.
2793         (vsx_and<mode>3_64bit): Likewise.
2794         (vsx_ior<mode>3_32bit): Likewise.
2795         (vsx_ior<mode>3_64bit): Likewise.
2796         (vsx_xor<mode>3_32bit): Likewise.
2797         (vsx_xor<mode>3_64bit): Likewise.
2798         (vsx_one_cmpl<mode>2_32bit): Likewise.
2799         (vsx_one_cmpl<mode>2_64bit): Likewise.
2800         (vsx_nor<mode>3_32bit): Likewise.
2801         (vsx_nor<mode>3_64bit): Likewise.
2802         (vsx_andc<mode>3_32bit): Likewise.
2803         (vsx_andc<mode>3_64bit): Likewise.
2804         (vsx_eqv<mode>3_32bit): Likewise.
2805         (vsx_eqv<mode>3_64bit): Likewise.
2806         (vsx_nand<mode>3_32bit): Likewise.
2807         (vsx_nand<mode>3_64bit): Likewise.
2808         (vsx_orc<mode>3_32bit): Likewise.
2809         (vsx_orc<mode>3_64bit): Likewise.
2811         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2812         logical types in GPRs.
2814         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2815         logical insns to rs6000.md, and allow TImode operations in
2816         32-bit.
2817         (altivec_ior<mode>3): Likewise.
2818         (altivec_xor<mode>3): Likewise.
2819         (altivec_one_cmpl<mode>2): Likewise.
2820         (altivec_nor<mode>3): Likewise.
2821         (altivec_andc<mode>3): Likewise.
2823         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2824         attributes for moving the 128-bit logical operations into
2825         rs6000.md.
2826         (BOOL_REGS_OUTPUT): Likewise.
2827         (BOOL_REGS_OP1): Likewise.
2828         (BOOL_REGS_OP2): Likewise.
2829         (BOOL_REGS_UNARY): Likewise.
2830         (BOOL_REGS_AND_CR0): Likewise.
2831         (one_cmpl<mode>2): Add support for DI logical operations on
2832         32-bit, splitting the operations to 32-bit.
2833         (anddi3): Likewise.
2834         (iordi3): Likewise.
2835         (xordi3): Likewise.
2836         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2837         changes to combine the 32/64-bit code, allow logical operations on
2838         TI mode in 32-bit, and to use similar match_operator patterns like
2839         scalar mode uses.  Combine the Altivec and VSX code for logical
2840         operations, and move it here.
2841         (ior<mode>3, 128-bit types): Likewise.
2842         (xor<mode>3, 128-bit types): Likewise.
2843         (one_cmpl<mode>3, 128-bit types): Likewise.
2844         (nor<mode>3, 128-bit types): Likewise.
2845         (andc<mode>3, 128-bit types): Likewise.
2846         (eqv<mode>3, 128-bit types): Likewise.
2847         (nand<mode>3, 128-bit types): Likewise.
2848         (orc<mode>3, 128-bit types): Likewise.
2849         (and<mode>3_internal): Likewise.
2850         (bool<mode>3_internal): Likewise.
2851         (boolc<mode>3_internal1): Likewise.
2852         (boolc<mode>3_internal2): Likewise.
2853         (boolcc<mode>3_internal1): Likewise.
2854         (boolcc<mode>3_internal2): Likewise.
2855         (eqv<mode>3_internal1): Likewise.
2856         (eqv<mode>3_internal2): Likewise.
2857         (one_cmpl1<mode>3_internal): Likewise.
2859 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2861         * config/microblaze/microblaze.c (microblaze_expand_prologue):
2862         Rename flag_stack_usage to flag_stack_usage_info.
2864 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2866         * config/microblaze/sync.md: New file.
2867         * config/microblaze/microblaze.md: Include sync.md
2868         * config/microblaze/microblaze.c: Add print_operand 'y'.
2869         * config/microblaze/constraints.md: Add memory_contraint
2870         'Q' which is a single register.
2872 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
2874         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2876 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2878         * reload.c (find_reloads): Exit loop once we find this operand
2879         cannot be reloaded somehow for this alternative.
2881         * reload.c (find_reloads): Exit loop once we find a hard register.
2883         * rtlanal.c (computed_jump_p): Exit loop once we find label
2884         reference is used.
2886         * i386.c (ix86_pad_returns): Exit loop after setting replace.
2888         * cfgloopmanip.c (remove_path): Exit loop after setting
2889         irred_invalidated.
2891         * gensupport.c (subst_dup): Avoid loop if code is not
2892         MATCH_DUP nor MATCH_OP_DUP.
2894 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
2896         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2898 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2900         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2901         true for SP_REGNUM if mode == ptr_mode.
2902         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2903         with value R0_REGNUM + 31.
2905 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2907         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2908         pad pointer-typed argument downward.
2910 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2912         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2913         and __ILP32__ when the ILP32 model is in use.
2915 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2917         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2918         (aarch64_load_symref_appropriately): In the case of
2919         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2920         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2921         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2922         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2923         if 'mode' doesn't equal to 'ptr_mode'.
2924         (aarch64_output_mi_thunk): Add an assertion on the alignment of
2925         'vcall_offset'; change to call aarch64_emit_move differently depending
2926         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2927         to calculate the upper bound of 'vcall_offset'.
2928         (aarch64_cannot_force_const_mem): Change to also return true if
2929         mode != ptr_mode.
2930         (aarch64_legitimize_reload_address): In the case of large
2931         displacements, add new local variable 'xmode' and an assertion
2932         based on it; change to use 'xmode' to generate the new rtx and
2933         reload.
2934         (aarch64_asm_trampoline_template): Change to generate the template
2935         differently depending on TARGET_ILP32 or not; change to use
2936         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2937         (aarch64_trampoline_size): Removed.
2938         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2939         and replace immediate literals with it.  Change to use 'ptr_mode'
2940         instead of 'DImode' and call convert_memory_address if the mode
2941         of 'fnaddr' doesn't equal to 'ptr_mode'.
2942         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2943         to output symbol.
2944         (aarch64_elf_asm_destructor): Likewise.
2945         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2946         on TARGET_ILP32 instead of aarch64_trampoline_size.
2947         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2948         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2949         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2950         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
2951         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2952         (storewb_pair<GPI:mode>_<P:mode>): ... this.
2953         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2954         depending on the value of 'mode'.
2955         (add_losym_<mode>): New.
2956         (ldr_got_small_<mode>): New, based on ldr_got_small.
2957         (ldr_got_small): Remove.
2958         (ldr_got_small_sidi): New.
2959         * config/aarch64/iterators.md (P): New.
2960         (PTR): Change to 'ptr_mode' in the condition.
2962 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2964         * config.gcc (aarch64*-*-*): Support --with-abi.
2965         (aarch64*-*-elf): Support --with-multilib-list.
2966         (aarch64*-*-linux*): Likewise.
2967         (supported_defaults): Add abi to aarch64*-*-*.
2968         * configure.ac: Mention AArch64 for --with-multilib-list.
2969         * configure: Re-generated.
2970         * config/aarch64/biarchilp32.h: New file.
2971         * config/aarch64/biarchlp64.h: New file.
2972         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2973         (ABI_SPEC): Ditto.
2974         (MULTILIB_DEFAULTS): Ditto.
2975         (DRIVER_SELF_SPECS): Ditto.
2976         (ASM_SPEC): Update to also substitute -mabi.
2977         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2978         file whose name depends on -mabi= and -mbig-endian.
2979         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2980         TARGET_ILP32.
2981         (POINTER_SIZE): New define.
2982         (POINTERS_EXTEND_UNSIGNED): Ditto.
2983         (enum aarch64_abi_type): New enumeration tag.
2984         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2985         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2986         (TARGET_ILP32): New define.
2987         * config/aarch64/aarch64.opt (mabi): New.
2988         (aarch64_abi): New.
2989         (ilp32, lp64): New values for -mabi.
2990         * config/aarch64/t-aarch64 (comma): New define.
2991         (MULTILIB_OPTIONS): Ditto.
2992         (MULTILIB_DIRNAMES): Ditto.
2993         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2994         * doc/invoke.texi: Document -mabi for AArch64.
2996 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
2998         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3000 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3001             Anton Blanchard <anton@au1.ibm.com>
3003         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3004         endianness when selecting field to splat.
3006 2013-07-22  Eric Christopher  <echristo@gmail.com>
3008         * dwarf2out.c (die_odr_checksum): New function to use
3009         CHECKSUM_ macros and ULEB128 for DIE tag.
3010         (generate_type_signature): Use.
3012 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3014         * config.gcc (sparc*-*-*): Accept leon3 processor.
3015         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3016         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3017         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3018         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3019         (mfix-ut699): Adjust comment.
3020         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3021         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3022         (CPP_CPU_SPEC): Likewise.
3023         (ASM_CPU_SPEC): Likewise.
3024         * config/sparc/sparc.c (leon3_cost): New constant.
3025         (sparc_option_override): Add leon3 support.
3026         (mem_ref): New function.
3027         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3028         (sparc_do_work_around_errata): Look into the instruction in the delay
3029         slot and adjust accordingly.  Add fix for the data cache nullify issues
3030         of the UT699.  Change insertion position for the NOP.
3031         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3032         (leon3_load): New reservation.
3033         (leon_store): Bump latency to 2.
3034         (grfpu): New automaton.
3035         (grfpu_alu): New unit.
3036         (grfpu_ds): Likewise.
3037         (leon_fp_alu): Adjust.
3038         (leon_fp_mult): Delete.
3039         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3040         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3041         * config/sparc/sparc.md (cpu): Add leon3.
3042         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3043         (swapsi): Likewise.
3044         (atomic_test_and_set): Likewise.
3045         (ldstub): Likewise.
3047 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3049         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3050         default for R5900 targets.
3051         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3052         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3053         * config/mips/mips.c (mips_option_override): Report an error for
3054         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3055         for -march=r5900 -mhard-float.
3057 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3059         * df-problems.c (can_move_insns_across): Exit loop once we
3060         find a non-fixed, non-global register.
3062         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3063         setting can_throw.
3065         * omega.c (omega_eliminate_red): Break after setting red_found.
3066         (omega_problem_has_red_equations): Similarly after setting found.
3067         (omega_query_variable): Similarly after setting coupled.
3069 2013-07-22  Marek Polacek  <polacek@redhat.com>
3071         * gimplify.c: Don't include gimple.h twice.
3073 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3075         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3076         instead of TARGET_THUMB1.
3077         (Pz): New constraint.
3078         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3079         encodings.
3080         (compare_negsi_si): Likewise.
3081         (compare_addsi2_op0): Likewise.
3082         (compare_addsi2_op1): Likewise.
3083         (addsi3_carryin_<optab>): Likewise.
3084         (addsi3_carryin_alt2_<optab>): Likewise.
3085         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3086         for arm_restrict_it.
3087         (subsi3_carryin): Likewise.
3088         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3089         (minmax_arithsi): Disable for arm_restrict_it.
3090         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3091         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3092         (satsi_<SAT:code>_shift): Likewise.
3093         (arm_shiftsi3): Add alternative for 16-bit encoding.
3094         (arm32_movhf): Disable for arm_restrict_it.
3095         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3096         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3098 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3100         * config/arm/arm.md (attribute "insn"): Delete.
3101         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3102         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3103         (not_shiftsi): Update for attribute change.
3104         (not_shiftsi_compare0): Likewise.
3105         (not_shiftsi_compare0_scratch): Likewise.
3106         (arm_one_cmplsi2): Likewise.
3107         (thumb1_one_cmplsi2): Likewise.
3108         (notsi_compare0): Likewise.
3109         (notsi_compare0_scratch): Likewise.
3110         (thumb1_movdi_insn): Likewise.
3111         (arm_movsi_insn): Likewise.
3112         (movhi_insn_arch4): Likewise.
3113         (movhi_bytes): Likewise.
3114         (arm_movqi_insn): Likewise.
3115         (thumb1_movqi_insn): Likewise.
3116         (arm32_movhf): Likewise.
3117         (thumb1_movhf): Likewise.
3118         (arm_movsf_soft_insn): Likewise.
3119         (thumb1_movsf_insn): Likewise.
3120         (thumb_movdf_insn): Likewise.
3121         (movsicc_insn): Likewise.
3122         (movsfcc_soft_insn): Likewise.
3123         (and_scc): Likewise.
3124         (cond_move): Likewise.
3125         (if_move_not): Likewise.
3126         (if_not_move): Likewise.
3127         (if_shift_move): Likewise.
3128         (if_move_shift): Likewise.
3129         (if_shift_shift): Likewise.
3130         (if_not_arith): Likewise.
3131         (if_arith_not): Likewise.
3132         (cond_move_not): Likewise.
3133         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3134         (neon_mov<mode>): Likewise.
3135         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3136         (thumb2_movsi_vfp): Likewise.
3137         (movsf_vfp): Likewise.
3138         (thumb2_movsf_vfp): Likewise.
3139         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3140         change.
3141         (cortexa7_older_only): Likewise.
3142         (cortexa7_younger): Likewise.
3143         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3144         (1020alu_shift_op): Likewise.
3145         (1020alu_shift_reg_op): Likewise.
3146         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3147         (alu_shift_op): Likewise.
3148         (alu_shift_reg_op): Likewise.
3149         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3150         (11_alu_shift_op): Likewise.
3151         (11_alu_shift_reg_op): Likewise.
3152         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3153         (9_alu_shift_reg_op): Likewise.
3154         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3155         change.
3156         (cortex_a15_alu_shift): Likewise.
3157         (cortex_a15_alu_shift_reg): Likewise.
3158         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3159         (cortex_a5_alu_shift): Likewise.
3160         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3161         change.
3162         (cortex_a53_alu_shift): Likewise.
3163         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3164         change.
3165         (cortex_a7_alu_reg): Likewise.
3166         (cortex_a7_alu_shift): Likewise.
3167         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3168         (cortex_a8_alu_shift): Likewise.
3169         (cortex_a8_alu_shift_reg): Likewise.
3170         (cortex_a8_mov): Likewise.
3171         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3172         (cortex_a9_dp_shift): Likewise.
3173         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3174         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3175         (cortex_r4_mov): Likewise.
3176         (cortex_r4_alu_shift): Likewise.
3177         (cortex_r4_alu_shift_reg): Likewise.
3178         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3179         (526_alu_shift_op): Likewise.
3180         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3181         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3182         (626te_alu_shift_op): Likewise.
3183         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3184         (726te_alu_op): Likewise.
3185         (726te_alu_shift_op): Likewise.
3186         (726te_alu_shift_reg_op): Likewise.
3187         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3188         (mp626_alu_shift_op): Likewise.
3189         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3190         (pj4_alu_e1_conds): Likewise.
3191         (pj4_alu): Likewise.
3192         (pj4_alu_conds): Likewise.
3193         (pj4_shift): Likewise.
3194         (pj4_shift_conds): Likewise.
3195         (pj4_alu_shift): Likewise.
3196         (pj4_alu_shift_conds): Likewise.
3198 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3200         * config/arm/predicates.md (shiftable_operator_strict_it):
3201         New predicate.
3202         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3203         Disable cond_exec version for arm_restrict_it.
3204         (thumb2_smaxsi3): Convert to generate cond_exec.
3205         (thumb2_sminsi3): Likewise.
3206         (thumb32_umaxsi3): Likewise.
3207         (thumb2_uminsi3): Likewise.
3208         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3209         (thumb2_neg_abssi2): Likewise.
3210         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3211         (thumb2_movsicc_insn): Adjust alternatives.
3212         (thumb2_mov_negscc): Disable for arm_restrict_it.
3213         (thumb2_mov_negscc_strict_it): New pattern.
3214         (thumb2_mov_notscc_strict_it): New pattern.
3215         (thumb2_mov_notscc): Disable for arm_restrict_it.
3216         (thumb2_ior_scc): Likewise.
3217         (thumb2_ior_scc_strict_it): New pattern.
3218         (thumb2_cond_move): Adjust for arm_restrict_it.
3219         (thumb2_cond_arith): Disable for arm_restrict_it.
3220         (thumb2_cond_arith_strict_it): New pattern.
3221         (thumb2_cond_sub): Adjust for arm_restrict_it.
3222         (thumb2_movcond): Likewise.
3223         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3224         (thumb2_zero_extendhisi2_v6): Likewise.
3225         (thumb2_zero_extendqisi2_v6): Likewise.
3226         (orsi_notsi_si): Likewise.
3227         (orsi_not_shiftsi_si): Likewise.
3229 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3231         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3232         instruction sequence is 1 byte shorter.
3234 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3236         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3237         it is not needed after split.
3239 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3241         PR target/51784
3242         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3243         second label for nonlocal goto receivers. Don't output pic base labels
3244         unless we're producing PIC; mark that action unreachable().
3245         (ix86_save_reg): If the function contains a nonlocal label, save the
3246         PIC base reg.
3247         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3248         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3249         (update_pic_label_number_if_needed): New.
3250         (machopic_output_function_base_name): Adjust for nonlocal receiver
3251         case.
3252         (machopic_should_output_picbase_label): New.
3253         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3254         (nonlocal_goto_receiver): New insn and split.
3256 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3258         * config/aarch64/aarch64-builtins.c
3259         (aarch64_fold_builtin): Fold abs in all modes.
3260         * config/aarch64/aarch64-simd-builtins.def
3261         (abs): Enable for all modes.
3262         * config/aarch64/arm_neon.h
3263         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3264         (vabs_f64): Add missing intrinsic.
3266 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3268         * config/aarch64/arm_neon.h (vabs_s64): New function
3270 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3272         PR target/57516
3273         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3274         * config/avr/avr.md (adjust_len): Add `round'.
3275         * config/avr/avr-protos.h (avr_out_round): New prototype.
3276         (avr_out_plus): Add `out_label' argument.
3277         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3278         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3279         Handle the case where `insn' is just a pattern.
3280         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3281         (avr_out_round): New function.
3282         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3284 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3286         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3287         Add check for flag_stack_usage to handle -fstack-usage support
3289 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3291         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3292         interaction for new Power8 flags and VSX.
3294 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3296         PR middle-end/57698
3297         * tree-inline.c (expand_call_inline): Emit errors during
3298         early_inlining only if optimization is not turned on.
3300 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3302         * passes.def: New.
3304         * passes.c (init_optimization_passes): Move the construction of
3305         the pass hierarchy into a new passes.def file.
3307         * Makefile.in (passes.o): Add dependency on passes.def.
3309 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3311         * passes.c (init_optimization_passes): Introduce macros for
3312         constructing the tree of passes (INSERT_PASSES_AFTER,
3313         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3314         TERMINATE_PASS_LIST).
3316 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3317             Wei Mi  <wmi@google.com>
3319         PR rtl-optimization/57878
3320         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3321         top.
3322         (reload_pseudo_compare_func): Check nregs first for reload
3323         pseudos.
3325 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3327         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3329 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3331         * read-rtl.c (validate_const_int): Once an invalid character is
3332         seen, quit the loop.
3334         * gengtype.c (write_roots): Similarly once we find the "deletable"
3335         or "if_marked" option.
3337 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3339         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3340         "xtab" and "sat".  Move value "clz" from here to ...
3341         (attriubte "type"): ... here.
3342         (satsi_<SAT:code>): Delete "insn" attribute.
3343         (satsi_<SAT:code>_shift): Likewise.
3344         (arm_zero_extendqisi2addsi): Likewise.
3345         (arm_extendqisi2addsi): Likewise.
3346         (clzsi2): Update for attribute changes.
3347         (rbitsi2): Likewise.
3348         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3349         attribute.
3350         (arm_usatsihi): Likewise.
3351         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3353 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3355         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3356         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3357         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3358         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3359         in alphabetical order.
3360         (attribute "core_cycles"): Update for attribute changes.
3361         (arm_addsi3): Likewise.
3362         (addsi3_compare0): Likewise.
3363         (addsi3_compare0_scratch): Likewise.
3364         (addsi3_compare_op1): Likewise.
3365         (addsi3_compare_op2): Likewise.
3366         (compare_addsi2_op0): Likewise.
3367         (compare_addsi2_op1): Likewise.
3368         (addsi3_carryin_shift_<optab>): Likewise.
3369         (subsi3_carryin_shift): Likewise.
3370         (rsbsi3_carryin_shift): Likewise.
3371         (arm_subsi3_insn): Likewise.
3372         (subsi3_compare0): Likewise.
3373         (subsi3_compare): Likewise.
3374         (arm_andsi3_insn): Likewise.
3375         (thumb1_andsi3_insn): Likewise.
3376         (andsi3_compare0): Likewise.
3377         (andsi3_compare0_scratch): Likewise.
3378         (zeroextractsi_compare0_scratch
3379         (andsi_not_shiftsi_si): Likewise.
3380         (iorsi3_insn): Likewise.
3381         (iorsi3_compare0): Likewise.
3382         (iorsi3_compare0_scratch): Likewise.
3383         (arm_xorsi3): Likewise.
3384         (thumb1_xorsi3_insn): Likewise.
3385         (xorsi3_compare0): Likewise.
3386         (xorsi3_compare0_scratch): Likewise.
3387         (satsi_<SAT:code>_shift): Likewise.
3388         (rrx): Likewise.
3389         (arm_shiftsi3): Likewise.
3390         (shiftsi3_compare0): Likewise.
3391         (not_shiftsi): Likewise.
3392         (not_shiftsi_compare0): Likewise.
3393         (not_shiftsi_compare0_scratch): Likewise.
3394         (arm_one_cmplsi2): Likewise.
3395         (thumb_one_complsi2): Likewise.
3396         (notsi_compare0): Likewise.
3397         (notsi_compare0_scratch): Likewise.
3398         (thumb1_zero_extendhisi2): Likewise.
3399         (arm_zero_extendhisi2): Likewise.
3400         (arm_zero_extendhisi2_v6): Likewise.
3401         (arm_zero_extendhisi2addsi): Likewise.
3402         (thumb1_zero_extendqisi2): Likewise.
3403         (thumb1_zero_extendqisi2_v6): Likewise.
3404         (arm_zero_extendqisi2): Likewise.
3405         (arm_zero_extendqisi2_v6): Likewise.
3406         (arm_zero_extendqisi2addsi): Likewise.
3407         (thumb1_extendhisi2): Likewise.
3408         (arm_extendhisi2): Likewise.
3409         (arm_extendhisi2_v6): Likewise.
3410         (arm_extendqisi): Likewise.
3411         (arm_extendqisi_v6): Likewise.
3412         (arm_extendqisi2addsi): Likewise.
3413         (thumb1_extendqisi2): Likewise.
3414         (thumb1_movdi_insn): Likewise.
3415         (arm_movsi_insn): Likewise.
3416         (movsi_compare0): Likewise.
3417         (movhi_insn_arch4): Likewise.
3418         (movhi_bytes): Likewise.
3419         (arm_movqi_insn): Likewise.
3420         (thumb1_movqi_insn): Likewise.
3421         (arm32_movhf): Likewise.
3422         (thumb1_movhf): Likewise.
3423         (arm_movsf_soft_insn): Likewise.
3424         (thumb1_movsf_insn): Likewise.
3425         (movdf_soft_insn): Likewise.
3426         (thumb_movdf_insn): Likewise.
3427         (arm_cmpsi_insn): Likewise.
3428         (cmpsi_shiftsi): Likewise.
3429         (cmpsi_shiftsi_swp): Likewise.
3430         (arm_cmpsi_negshiftsi_si): Likewise.
3431         (movsicc_insn): Likewise.
3432         (movsfcc_soft_insn): Likewise.
3433         (arith_shiftsi): Likewise.
3434         (arith_shiftsi_compare0
3435         (arith_shiftsi_compare0_scratch
3436         (sub_shiftsi): Likewise.
3437         (sub_shiftsi_compare0
3438         (sub_shiftsi_compare0_scratch
3439         (and_scc): Likewise.
3440         (cond_move): Likewise.
3441         (if_plus_move): Likewise.
3442         (if_move_plus): Likewise.
3443         (if_move_not): Likewise.
3444         (if_not_move): Likewise.
3445         (if_shift_move): Likewise.
3446         (if_move_shift): Likewise.
3447         (if_shift_shift): Likewise.
3448         (if_not_arith): Likewise.
3449         (if_arith_not): Likewise.
3450         (cond_move_not): Likewise.
3451         (thumb1_ashlsi3): Set type attribute.
3452         (thumb1_ashrsi3): Likewise.
3453         (thumb1_lshrsi3): Likewise.
3454         (thumb1_rotrsi3): Likewise.
3455         (shiftsi3_compare0_scratch): Likewise.
3456         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3457         (neon_mov<mode>): Likewise.
3458         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3459         attribute changes.
3460         (thumb2_movsi_insn): Likewise.
3461         (thumb2_cmpsi_neg_shiftsi): Likewise.
3462         (thumb2_extendqisi_v6): Likewise.
3463         (thumb2_zero_extendhisi2_v6): Likewise.
3464         (thumb2_zero_extendqisi2_v6): Likewise.
3465         (thumb2_shiftsi3_short): Likewise.
3466         (thumb2_addsi3_compare0_scratch): Likewise.
3467         (orsi_not_shiftsi_si): Likewise.
3468         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3469         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3470         changes.
3471         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3472         (1020alu_shift_op): Likewise.
3473         (1020alu_shift_reg_op): Likewise.
3474         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3475         (alu_shift_op): Likewise.
3476         (alu_shift_reg_op): Likewise.
3477         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3478         (11_alu_shift_op): Likewise.
3479         (11_alu_shift_reg_op): Likewise.
3480         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3481         (9_alu_shift_reg_op): Likewise.
3482         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3483         changes.
3484         (cortex_a15_alu_shift): Likewise.
3485         (cortex_a15_alu_shift_reg): Likewise.
3486         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3487         changes.
3488         (cortex_a5_alu_shift): Likewise.
3489         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3490         changes.
3491         (cortex_a53_alu_shift): Likewise.
3492         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3493         changes.
3494         (cortex_a7_alu_reg): Likewise.
3495         (cortex_a7_alu_shift): Likewise.
3496         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3497         changes.
3498         (cortex_a8_alu_shift): Likewise.
3499         (cortex_a8_alu_shift_reg): Likewise.
3500         (cortex_a8_mov): Likewise.
3501         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3502         (cortex_a9_dp_shift): Likewise.
3503         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3504         changes.
3505         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3506         changes.
3507         (cortex_r4_mov): Likewise.
3508         (cortex_r4_alu_shift): Likewise.
3509         (cortex_r4_alu_shift_reg): Likewise.
3510         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3511         (526_alu_shift_op): Likewise.
3512         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3513         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3514         (626te_alu_shift_op): Likewise.
3515         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3516         (726te_alu_op): Likewise.
3517         (726te_alu_shift_op): Likewise.
3518         (726te_alu_shift_reg_op): Likewise.
3519         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3520         (mp626_alu_shift_op): Likewise.
3521         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3522         (pj4_alu_e1_conds): Likewise.
3523         (pj4_alu): Likewise.
3524         (pj4_alu_conds): Likewise.
3525         (pj4_shift): Likewise.
3526         (pj4_shift_conds): Likewise.
3527         (pj4_alu_shift): Likewise.
3528         (pj4_alu_shift_conds): Likewise.
3529         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3530         changes.
3531         (cortexa7_older_only): Likewise.
3532         (cortexa7_younger): Likewise.
3534 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3536         * ipa-pure-const.c (generate_summary): Rename to...
3537         (pure_const_generate_summary): ... this.
3539 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3541         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3543 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3545         PR target/57909
3546         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3547         usage in HI mode.
3549 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3551         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3552         enabled without -march=zEC12.
3553         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3554         flags to be set.
3556 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3558         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3559         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3560         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3561         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3562         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3563         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3564         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3565         ISA_HAS_FP4.
3566         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3567         and ISA_HAS_NMADD3_NMSUB3.
3568         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3569         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3570         (nmsub4<mode>, nmsub3<mode>): Likewise.
3571         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3573 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3575         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3576         TARGET_MIPS5400 checking.
3578 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3579             Peter Bergner  <bergner@vnet.ibm.com>
3581         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3582         registers in the comment.
3583         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3584         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3585         rather than FIRST_PSEUDO_REGISTERS.
3587 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3589         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3590         enable extra ISA flags with TARGET_HTM.
3592 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3594         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3595         Fix comment typos.
3597 2013-07-15  Cong Hou  <congh@google.com>
3599         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3600         in compare function for sorting.
3602 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3604         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3605         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3606         * config/rs6000/rs6000.opt: Add -mhtm option.
3607         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3608         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3609         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3610         __HTM__ if the HTM instructions are available.
3611         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3612         htm_spr_reg_operand): New define_predicates.
3613         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3614         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3615         Include htm.md.
3616         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3617         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3618         HTM builtin functions.
3619         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3620         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3621         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3622         (rs6000_builtin_mask_calculate): Likewise.
3623         (rs6000_option_override_internal): Likewise.
3624         (bdesc_htm): Add new HTM builtin support.
3625         (htm_spr_num): New function.
3626         (htm_spr_regno): Likewise.
3627         (rs6000_htm_spr_icode): Likewise.
3628         (htm_expand_builtin): Likewise.
3629         (htm_init_builtins): Likewise.
3630         (rs6000_expand_builtin): Add support for HTM builtin functions.
3631         (rs6000_init_builtins): Likewise.
3632         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3633         option.
3634         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3635         (TARGET_HTM, MASK_HTM): Define macros.
3636         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3637         (FIXED_REGISTERS): Likewise.
3638         (CALL_USED_REGISTERS): Likewise.
3639         (CALL_REALLY_USED_REGISTERS): Likewise.
3640         (REG_ALLOC_ORDER): Likewise.
3641         (enum reg_class): Likewise.
3642         (REG_CLASS_NAMES): Likewise.
3643         (REG_CLASS_CONTENTS): Likewise.
3644         (REGISTER_NAMES): Likewise.
3645         (ADDITIONAL_REGISTER_NAMES): Likewise.
3646         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3647         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3648         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3649         * config/rs6000/htm.md: New file.
3650         * config/rs6000/htmintrin.h: New file.
3651         * config/rs6000/htmxlintrin.h: New file.
3653 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3655         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3656         Define SYMBOL_TINY_GOT, update comment.
3657         * config/aarch64/aarch64.c
3658         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3659         (aarch64_expand_mov_immediate): Likewise.
3660         (aarch64_print_operand): Likewise.
3661         (aarch64_classify_symbol): Likewise.
3662         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3663         (ldr_got_tiny): Define.
3665 2013-07-13  Tobias Grosser  <tobias@grosser.es>
3667         PR tree-optimization/54094
3668         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3669           scheduling dimension for the parallelism check from the polyhedral
3670           information in the AST.
3671         * graphite-dependences.c (carries_deps): Do not assume the schedule is
3672           in 2D + 1 form.
3674 2013-07-13  Jason Merrill  <jason@redhat.com>
3676         * print-tree.c (debug_vec_tree): Use debug_raw.
3677         (debug_raw (vec<tree, va_gc> &)): New.
3678         (debug_raw (vec<tree, va_gc> *)): New.
3679         * tree.h: Declare them.
3681 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
3683         * ifcvt.c (ifcvt_after_combine): New static variable.
3684         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3685         for size.
3686         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
3687         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3688         rest_of_handle_if_after_reload): Pass new argument for if_convert.
3690 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
3692         * config/mips/mips.c (mips_expand_call): Remove empty statement.
3694 2013-07-12  Michael Matz  <matz@suse.de>
3696         PR middle-end/55771
3697         * convert.c (convert_to_real): Reject non-float inner types.
3699 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
3701         * config/aarch64/aarch64-protos.h
3702         (aarch64_simd_immediate_valid_for_move): Remove.
3703         * config/aarch64/aarch64.c (simd_immediate_info): New member.
3704         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3705         cases.
3706         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3708 2013-07-11  Steve Ellcey  <sellcey@mips.com>
3710         * config/mips/mips.c (mips_conditional_register_usage): Do not
3711         use t[0-7] registers in MIPS16 mode when optimizing for size.
3713 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
3715         * config/i386/i386.c (dispatch_function_versions): Fix array
3716         indexing of function_version_info to match actual_versions.
3718 2013-07-11  Teresa Johnson  <tejohnson@google.com>
3720         * vec.h (struct va_gc): Move release out-of-line.
3721         (va_gc::release): Call ggc_free on released vec.
3723 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3725         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3726         Require GOT register as additional operand in UNSPEC.
3727         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3728         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3729         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3730         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3731         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3732         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3733         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3735 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
3737         PR target/57631
3738         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3739         name seen by assembler/linker rather if available.
3741 2013-07-11  Andreas Schwab  <schwab@suse.de>
3743         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3745 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
3747         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
3749 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
3751         * doc/tm.texi.in: Move hook documentation to ....
3752         * target.def: ... here.
3754         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3755         text on @hook line.
3756         * doc/tm.texi: Regenerate.
3758 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
3760         PR c++/57869
3761         * doc/invoke.texi: Document -Wconditionally-supported.
3763 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3765         PR target/57844
3766         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3767         of my_fp.
3769 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3771         PR target/57506
3772         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3773         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3774         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3775         Remove duplicate devices.
3776         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3777         * config/avr/t-multilib: Regenerate.
3778         * config/avr/avr-tables.opt: Regenerate.
3779         * doc/avr-mmcu.texi: Regenerate.
3781 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3783         PR target/56987
3784         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3786 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
3788         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3789         the cost of MULT when optimizing for size.
3791 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3793         * config/cr16/cr16-protos.h: Don't include target.h.
3795 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
3797         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3798         adjust register size for TDmode and TFmode for VSX registers.
3800 2013-07-08  Kai Tietz  <ktietz@redhat.com>
3802         PR target/56892
3803         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3804         hook_bool_const_tree_true.
3806 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3808         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3809         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3810         * config/s390/s390.md: Define FPR*_REGNUM constants.
3811         Fix FPR2_REGNUM constant (18 -> 17).
3812         ("*trunc<BFP:mode><DFP_ALL:mode>2")
3813         ("*trunc<DFP_ALL:mode><BFP:mode>2")
3814         ("trunc<BFP:mode><DFP_ALL:mode>2")
3815         ("trunc<DFP_ALL:mode><BFP:mode>2")
3816         ("*extend<BFP:mode><DFP_ALL:mode>2")
3817         ("*extend<DFP_ALL:mode><BFP:mode>2")
3818         ("extend<BFP:mode><DFP_ALL:mode>2")
3819         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3820         FPR4_REGNUM.
3822 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
3824         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
3826 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3828         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3829         and cfun_fpr_bit_p to cfun_fpr_save_p.
3830         (s390_frame_area, s390_register_info, s390_frame_info)
3831         (s390_emit_prologue, s390_emit_epilogue)
3832         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3833         register numbers.
3834         * config/s390/s390.h: Define *_REGNUM macros for floating point
3835         register numbers.
3837 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
3839         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3841 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
3843         PR rtl-optimization/57786
3844         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3845         and break out of the loop when it is set to false.
3847 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
3849         PR target/57819
3850         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3851         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3852         (const_int 63)) 0)).
3853         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3854         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3855         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3857         PR rtl-optimization/57829
3858         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3859         mask bits outside of mode are just sign-extension from mode to HWI.
3861 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
3863         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3864         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3865         adjust_address instead of change_address to keep info about alignment.
3866         (emit_strmov): Remove.
3867         (emit_memmov): New function.
3868         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3869         (expand_movmem_epilogue): Likewise and return updated rtx for
3870         destination.
3871         (expand_constant_movmem_prologue): Likewise and return updated rtx for
3872         destination and source.
3873         (decide_alignment): Refactor, handle vector_loop.
3874         (ix86_expand_movmem): Likewise.
3875         (ix86_expand_setmem): Likewise.
3876         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3878 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
3880         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3881         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3883 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3885         * config/i386/sse.md (sse_movlhps): Change alternative 3
3886         of operand 2 to "m".
3888 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3890         PR target/57807
3891         * config/i386/sse.md (iptr): New mode attribute.
3892         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3893         (<sse>_vm<plusminus_insn><mode>3): Ditto.
3894         (<sse>_vmmul<mode>3): Ditto.
3895         (<sse>_vmdiv<mode>3): Ditto.
3896         (sse_vmrcpv4sf2): Ditto.
3897         (<sse>_vmsqrt<mode>2): Ditto.
3898         (sse_vmrsqrtv4sf2): Ditto.
3899         (<sse>_vm<code><mode>3): Ditto.
3900         (avx_vmcmp<mode>3): Ditto.
3901         (<sse>_vmmaskcmp<mode>3): Ditto.
3902         (<sse>_comi): Ditto.
3903         (<sse>_ucomi): Ditto.
3904         (*xop_vmfrcz_<mode>): Ditto.
3905         (*fmai_fmadd_<mode>): Ditto.
3906         (*fmai_fmsub_<mode>): Ditto.
3907         (*fmai_fnmadd_<mode>): Ditto.
3908         (*fmai_fnmsub_<mode>): Ditto.
3909         (*fma4i_vmfmadd_<mode>): Ditto.
3910         (*fma4i_vmfmsub_<mode>): Ditto.
3911         (*fma4i_vmfnmadd_<mode>): Ditto.
3912         (*fma4i_vmfnmsub_<mode>): Ditto.
3913         (*xop_vmfrcz_<mode>): Ditto.
3914         (sse_cvtps2pi): Ditto.
3915         (sse_cvttps2pi): Ditto.
3916         (sse_cvtss2si): Ditto.
3917         (sse_cvtss2si_2): Ditto.
3918         (sse_cvtss2siq_2): Ditto.
3919         (sse_cvttss2si): Ditto.
3920         (sse_cvttss2siq): Ditto.
3921         (sse_cvtsd2si): Ditto.
3922         (sse_cvtsd2si_2): Ditto.
3923         (sse_cvtsd2siq_2): Ditto.
3924         (sse_cvttsd2si): Ditto.
3925         (sse_cvttsd2siq): Ditto.
3926         (sse_cvtsd2ss): Ditto.
3927         (sse_cvtss2sd): Ditto.
3928         (avx2_pbroadcast<mode>): Ditto.
3929         (avx2_pbroadcast<mode>_1): Ditto.
3930         (*avx_vperm_broadcast_v4sf): Ditto.
3932         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3933         (sse_movlhps): Ditto.
3934         (sse_storehps): Ditto.
3935         (sse_loadhps): Ditto.
3936         (sse_storelps): Ditto.
3937         (sse_loadlps): Ditto.
3938         (*vec_concatv4sf): Ditto.
3939         (*vec_interleave_highv2df): Ditto.
3940         (*vec_interleave_lowv2df): Ditto.
3941         (*vec_extractv2df_1_sse): Ditto.
3942         (*vec_extractv2df_0_sse): Ditto.
3943         (sse2_storelpd): Ditto.
3944         (sse2_loadlpd): Ditto.
3945         (sse2_movsd): Ditto.
3946         (*vec_concatv4si): Ditto.
3947         (vec_concatv2di): Ditto.
3949         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3950         for Intel asm dialect.
3951         (mmx_punpcklwd): Ditto.
3952         (mmx_punpckldq): Ditto.
3954         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3955         for intel assembler dialect.
3957 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
3959         PR target/29776
3960         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3961         for BUILT_IN_C{LZ,LRSB}*.
3962         * tree.h (CASE_INT_FN): Add FN##IMAX case.
3963         * tree-vrp.c (extract_range_basic): Handle
3964         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
3965         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3966         fall thru to code calling set_value*.
3967         * builtins.c (expand_builtin): Remove *IMAX cases.
3968         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3969         if width is bigger than 2*HWI.
3971 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
3973         PR rtl-optimization/55342
3974         * lra-int.h (lra_subreg_reload_pseudos): New.
3975         * lra.c: Add undoing optional reloads to the block diagram.
3976         (lra_subreg_reload_pseudos): New.
3977         (lra_optional_reload_pseudos): Change comments.
3978         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
3979         lra_optional_reload_pseudos after undo transformations.
3980         * lra-assigns.c (pseudo_prefix_title): New.
3981         (lra_setup_reg_renumber): Use it.
3982         (spill_for): Ditto.  Check subreg reload pseudos too.
3983         (assign_by_spills): Consider subreg reload pseudos too.
3984         * lra-constraints.c (simplify_operand_subreg): Use
3985         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3986         (curr_insn_transform): Recognize and do optional reloads.
3987         (undo_optional_reloads): New.
3988         (lra_undo_inheritance): Call undo_optional_reloads.
3990 2013-07-05  Thomas Quinot  <quinot@adacore.com>
3992         * tree-complex.c (expand_complex_operations_1): Fix typo.
3994 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
3996         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3997         (tune_params): New member 'const vec_costs'.
3998         * config/aarch64/aarch64.c (generic_vector_cost): New.
3999         (generic_tunings): New member 'generic_vector_cost'.
4000         (aarch64_builtin_vectorization_cost): New.
4001         (aarch64_add_stmt_cost): New.
4002         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4003         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4005 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4007         PR target/57777
4008         * config/i386/predicates.md (vsib_address_operand): Disallow
4009         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4011 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4013         PR middle-end/55030
4014         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4015         expand_builtin_setjmp_receiver.
4016         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4017         with NULL for the label parameter.
4018         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4019         the frame-pointer.  Adjust comments.
4020         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4021         only if LABEL is non-NULL.
4023 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4025         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4026         (ARM_ABI_AAPCS64): Ditto.
4027         (arm_abi): Ditto.
4028         (ARM_DEFAULT_ABI): Ditto.
4030 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4032         * config/aarch64/aarch64-builtins.c
4033         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4034         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4035         (st1): Likewise.
4036         * config/aarch64/aarch64-simd.md
4037         (aarch64_ld1<VALL:mode>): New.
4038         (aarch64_st1<VALL:mode>): Likewise.
4039         * config/aarch64/arm_neon.h
4040         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4042 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4044         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4045         target ISA is AVX.
4046         (ix86_option_override_internal):Turn on all -mavx target flags by
4047         default as they are dependent on AVX anyway.
4049 2013-07-02  Cary Coutant  <ccoutant@google.com>
4051         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4052         deterministic hash.
4053         (loc_checksum_ordered): Likewise.
4054         (hash_loc_operands): Remove inline keyword.
4056 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4058         PR tree-optimization/57741
4059         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4060         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4061         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4062         Allow REAL_CST step_exprs if flag_associative_math.
4063         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4065 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4067         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4069 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4071         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4073 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4075         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4076         encoding.
4077         (iorsi3_insn): Likewise.
4078         (arm_xorsi3): Likewise.
4080 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4082         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4083         to ...
4084         (attribute "type"): ... here, and prefix with "wmmx_".
4085         (attribute "core_cycles"): Update for attribute changes.
4086         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4087         (tbcstv4hi): Likewise.
4088         (tbcstv2si): Likewise.
4089         (iwmmxt_iordi3): Likewise.
4090         (iwmmxt_xordi3): Likewise.
4091         (iwmmxt_anddi3): Likewise.
4092         (iwmmxt_nanddi3): Likewise.
4093         (iwmmxt_arm_movdi): Likewise.
4094         (iwmmxt_movsi_insn): Likewise.
4095         (mov<mode>_internal): Likewise.
4096         (and<mode>3_iwmmxt): Likewise.
4097         (ior<mode>3_iwmmxt): Likewise.
4098         (xor<mode>3_iwmmxt): Likewise.
4099         (add<mode>3_iwmmxt): Likewise.
4100         (ssaddv8qi3): Likewise.
4101         (ssaddv4hi3): Likewise.
4102         (ssaddv2si3): Likewise.
4103         (usaddv8qi3): Likewise.
4104         (usaddv4hi3): Likewise.
4105         (usaddv2si3): Likewise.
4106         (sub<mode>3_iwmmxt): Likewise.
4107         (sssubv8qi3): Likewise.
4108         (sssubv4hi3): Likewise.
4109         (sssubv2si3): Likewise.
4110         (ussubv8qi3): Likewise.
4111         (ussubv4hi3): Likewise.
4112         (ussubv2si3): Likewise.
4113         (mulv4hi3_iwmmxt): Likewise.
4114         (smulv4hi3_highpart): Likewise.
4115         (umulv4hi3_highpart): Likewise.
4116         (iwmmxt_wmacs): Likewise.
4117         (iwmmxt_wmacsz): Likewise.
4118         (iwmmxt_wmacu): Likewise.
4119         (iwmmxt_wmacuz): Likewise.
4120         (iwmmxt_clrdi): Likewise.
4121         (iwmmxt_clrv8qi): Likewise.
4122         (iwmmxt_clr4hi): Likewise.
4123         (iwmmxt_clr2si): Likewise.
4124         (iwmmxt_uavgrndv8qi3): Likewise.
4125         (iwmmxt_uavgrndv4hi3): Likewise.
4126         (iwmmxt_uavgv8qi3): Likewise.
4127         (iwmmxt_uavgv4hi3): Likewise.
4128         (iwmmxt_tinsrb): Likewise.
4129         (iwmmxt_tinsrh): Likewise.
4130         (iwmmxt_tinsrw): Likewise.
4131         (iwmmxt_textrmub): Likewise.
4132         (iwmmxt_textrmsb): Likewise.
4133         (iwmmxt_textrmuh): Likewise.
4134         (iwmmxt_textrmsh): Likewise.
4135         (iwmmxt_textrmw): Likewise.
4136         (iwmxxt_wshufh): Likewise.
4137         (eqv8qi3): Likewise.
4138         (eqv4hi3): Likewise.
4139         (eqv2si3): Likewise.
4140         (gtuv8qi3): Likewise.
4141         (gtuv4hi3): Likewise.
4142         (gtuv2si3): Likewise.
4143         (gtv8qi3): Likewise.
4144         (gtv4hi3): Likewise.
4145         (gtv2si3): Likewise.
4146         (smax<mode>3_iwmmxt): Likewise.
4147         (umax<mode>3_iwmmxt): Likewise.
4148         (smin<mode>3_iwmmxt): Likewise.
4149         (umin<mode>3_iwmmxt): Likewise.
4150         (iwmmxt_wpackhss): Likewise.
4151         (iwmmxt_wpackwss): Likewise.
4152         (iwmmxt_wpackdss): Likewise.
4153         (iwmmxt_wpackhus): Likewise.
4154         (iwmmxt_wpackwus): Likewise.
4155         (iwmmxt_wpackdus): Likewise.
4156         (iwmmxt_wunpckihb): Likewise.
4157         (iwmmxt_wunpckihh): Likewise.
4158         (iwmmxt_wunpckihw): Likewise.
4159         (iwmmxt_wunpckilb): Likewise.
4160         (iwmmxt_wunpckilh): Likewise.
4161         (iwmmxt_wunpckilw): Likewise.
4162         (iwmmxt_wunpckehub): Likewise.
4163         (iwmmxt_wunpckehuh): Likewise.
4164         (iwmmxt_wunpckehuw): Likewise.
4165         (iwmmxt_wunpckehsb): Likewise.
4166         (iwmmxt_wunpckehsh): Likewise.
4167         (iwmmxt_wunpckehsw): Likewise.
4168         (iwmmxt_wunpckelub): Likewise.
4169         (iwmmxt_wunpckeluh): Likewise.
4170         (iwmmxt_wunpckeluw): Likewise.
4171         (iwmmxt_wunpckelsb): Likewise.
4172         (iwmmxt_wunpckelsh): Likewise.
4173         (iwmmxt_wunpckelsw): Likewise.
4174         (ror<mode>3): Likewise.
4175         (ashr<mode>3_iwmmxt): Likewise.
4176         (lshr<mode>3_iwmmxt): Likewise.
4177         (ashl<mode>3_iwmmxt): Likewise.
4178         (ror<mode>3_di): Likewise.
4179         (ashr<mode>3_di): Likewise.
4180         (lshr<mode>3_di): Likewise.
4181         (ashl<mode>3_di): Likewise.
4182         (iwmmxt_wmadds): Likewise.
4183         (iwmmxt_wmaddu): Likewise.
4184         (iwmmxt_tmia): Likewise.
4185         (iwmmxt_tmiaph): Likewise.
4186         (iwmmxt_tmiabb): Likewise.
4187         (iwmmxt_tmiatb): Likewise.
4188         (iwmmxt_tmiabt): Likewise.
4189         (iwmmxt_tmiatt): Likewise.
4190         (iwmmxt_tmovmskb): Likewise.
4191         (iwmmxt_tmovmskh): Likewise.
4192         (iwmmxt_tmovmskw): Likewise.
4193         (iwmmxt_waccb): Likewise.
4194         (iwmmxt_wacch): Likewise.
4195         (iwmmxt_waccw): Likewise.
4196         (iwmmxt_waligni): Likewise.
4197         (iwmmxt_walignr): Likewise.
4198         (iwmmxt_walignr0): Likewise.
4199         (iwmmxt_walignr1): Likewise.
4200         (iwmmxt_walignr2): Likewise.
4201         (iwmmxt_walignr3): Likewise.
4202         (iwmmxt_wsadb): Likewise.
4203         (iwmmxt_wsadh): Likewise.
4204         (iwmmxt_wsadbz): Likewise.
4205         (iwmmxt_wsadhz): Likewise.
4206         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4207         (iwmmxt_wabsdiffb): Likewise.
4208         (iwmmxt_wabsdiffh): Likewise.
4209         (iwmmxt_wabsdiffw): Likewise.
4210         (iwmmxt_waddsubhx): Likewise
4211         (iwmmxt_wsubaddhx): Likewise.
4212         (addc<mode>3): Likewise.
4213         (iwmmxt_avg4): Likewise.
4214         (iwmmxt_avg4r): Likewise.
4215         (iwmmxt_wmaddsx): Likewise.
4216         (iwmmxt_wmaddux): Likewise.
4217         (iwmmxt_wmaddsn): Likewise.
4218         (iwmmxt_wmaddun): Likewise.
4219         (iwmmxt_wmulwsm): Likewise.
4220         (iwmmxt_wmulwum): Likewise.
4221         (iwmmxt_wmulsmr): Likewise.
4222         (iwmmxt_wmulumr): Likewise.
4223         (iwmmxt_wmulwsmr): Likewise.
4224         (iwmmxt_wmulwumr): Likewise.
4225         (iwmmxt_wmulwl): Likewise.
4226         (iwmmxt_wqmulm): Likewise.
4227         (iwmmxt_wqmulwm): Likewise.
4228         (iwmmxt_wqmulmr): Likewise.
4229         (iwmmxt_wqmulwmr): Likewise.
4230         (iwmmxt_waddbhusm): Likewise.
4231         (iwmmxt_waddbhusl): Likewise.
4232         (iwmmxt_wqmiabb): Likewise.
4233         (iwmmxt_wqmiabt): Likewise.
4234         (iwmmxt_wqmiatb): Likewise.
4235         (iwmmxt_wqmiatt): Likewise.
4236         (iwmmxt_wqmiabbn): Likewise.
4237         (iwmmxt_wqmiabtn): Likewise.
4238         (iwmmxt_wqmiatbn): Likewise.
4239         (iwmmxt_wqmiattn): Likewise.
4240         (iwmmxt_wmiabb): Likewise.
4241         (iwmmxt_wmiabt): Likewise.
4242         (iwmmxt_wmiatb): Likewise.
4243         (iwmmxt_wmiatt): Likewise.
4244         (iwmmxt_wmiabbn): Likewise.
4245         (iwmmxt_wmiabtn): Likewise.
4246         (iwmmxt_wmiatbn): Likewise.
4247         (iwmmxt_wmiattn): Likewise.
4248         (iwmmxt_wmiawbb): Likewise.
4249         (iwmmxt_wmiawbt): Likewise.
4250         (iwmmxt_wmiawtb): Likewise.
4251         (iwmmxt_wmiawtt): Likewise.
4252         (iwmmxt_wmiawbbn): Likewise.
4253         (iwmmxt_wmiawbtn): Likewise.
4254         (iwmmxt_wmiawtbn): Likewise.
4255         (iwmmxt_wmiawttn): Likewise.
4256         (iwmmxt_wmerge): Likewise.
4257         (iwmmxt_tandc<mode>3): Likewise.
4258         (iwmmxt_torc<mode>3): Likewise.
4259         (iwmmxt_torvsc<mode>3): Likewise.
4260         (iwmmxt_textrc<mode>3): Likewise.
4261         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4262         (wmmxt_pack): Likewise.
4263         (wmmxt_mult_c1): Likewise.
4264         (wmmxt_mult_c2): Likewise.
4265         (wmmxt_alu_c1): Likewise.
4266         (wmmxt_alu_c2): Likewise.
4267         (wmmxt_alu_c3): Likewise.
4268         (wmmxt_transfer_c1): Likewise.
4269         (wmmxt_transfer_c2): Likewise.
4270         (wmmxt_transfer_c3): Likewise.
4271         (marvell_f_iwmmxt_wstr): Likewise.
4272         (marvell_f_iwmmxt_wldr): Likewise.
4274 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4276         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4278 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4280         Revert:
4281         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4282         * lra-constraints.c (need_for_split_p): Check call used hard regs
4283         living through calls.
4285         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4286         call used regs for call insn.
4288 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4290         PR target/57736
4291         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4292         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4294 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4296         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4298 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4300         * lra-constraints.c (need_for_split_p): Check call used hard regs
4301         living through calls.
4303 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4305         PR target/57744
4306         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4307         to tie with any other modes.  Eliminate Altivec vector mode tests,
4308         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4309         code, to return 0 if testing MODE2 for a condition, if we've
4310         already tested MODE1 for the same condition.
4312 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4314         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4315         layout.
4317 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4319         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4320         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4322 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4324         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4325         Define.
4326         (aarch64_symbolic_constant_p): Remove.
4327         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4328         static.  Fix line length and white space.
4329         (aarch64_symbolic_constant_p): Remove.
4330         * config/aarch64/predicates.md (aarch64_valid_symref):
4331         Use aarch64_classify_symbol_expression.
4333 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4335         * config/arm/constraints.md (Ts): New constraint.
4336         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4337         16-bit encodings.
4338         (compare_scc): Use "Ts" constraint for operand 0.
4339         (ior_scc_scc): Likewise.
4340         (and_scc_scc): Likewise.
4341         (and_scc_scc_nodom): Likewise.
4342         (ior_scc_scc_cmp): Likewise for operand 7.
4343         (and_scc_scc_cmp): Likewise.
4344         * config/arm/thumb2.md (thumb2_movsi_insn):
4345         Add alternatives for 16-bit encodings.
4346         (thumb2_movhi_insn): Likewise.
4347         (thumb2_movsicc_insn): Likewise.
4348         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4349         (thumb2_negscc): Use "Ts" constraint.
4350         Move mvn instruction outside cond_exec block.
4351         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4352         for 16-bit encodings.
4354 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4356         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4357         encoding.
4358         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4359         (mulsi3subsi): Likewise.
4360         (mulsidi3adddi): Likewise.
4361         (mulsidi3_v6): Likewise.
4362         (umulsidi3_v6): Likewise.
4363         (umulsidi3adddi_v6): Likewise.
4364         (smulsi3_highpart_v6): Likewise.
4365         (umulsi3_highpart_v6): Likewise.
4366         (mulhisi3tb): Likewise.
4367         (mulhisi3bt): Likewise.
4368         (mulhisi3tt): Likewise.
4369         (maddhisi4): Likewise.
4370         (maddhisi4tb): Likewise.
4371         (maddhisi4tt): Likewise.
4372         (maddhidi4): Likewise.
4373         (maddhidi4tb): Likewise.
4374         (maddhidi4tt): Likewise.
4375         (zeroextractsi_compare0_scratch): Likewise.
4376         (insv_zero): Likewise.
4377         (insv_t2): Likewise.
4378         (anddi_notzesidi_di): Likewise.
4379         (anddi_notsesidi_di): Likewise.
4380         (andsi_notsi_si): Likewise.
4381         (iordi_zesidi_di): Likewise.
4382         (xordi_zesidi_di): Likewise.
4383         (andsi_iorsi3_notsi): Likewise.
4384         (smax_0): Likewise.
4385         (smax_m1): Likewise.
4386         (smin_0): Likewise.
4387         (not_shiftsi): Likewise.
4388         (unaligned_loadsi): Likewise.
4389         (unaligned_loadhis): Likewise.
4390         (unaligned_loadhiu): Likewise.
4391         (unaligned_storesi): Likewise.
4392         (unaligned_storehi): Likewise.
4393         (extv_reg): Likewise.
4394         (extzv_t2): Likewise.
4395         (divsi3): Likewise.
4396         (udivsi3): Likewise.
4397         (arm_zero_extendhisi2addsi): Likewise.
4398         (arm_zero_extendqisi2addsi): Likewise.
4399         (compareqi_eq0): Likewise.
4400         (arm_extendhisi2_v6): Likewise.
4401         (arm_extendqisi2addsi): Likewise.
4402         (arm_movt): Likewise.
4403         (thumb2_ldrd): Likewise.
4404         (thumb2_ldrd_base): Likewise.
4405         (thumb2_ldrd_base_neg): Likewise.
4406         (thumb2_strd): Likewise.
4407         (thumb2_strd_base): Likewise.
4408         (thumb2_strd_base_neg): Likewise.
4409         (arm_negsi2): Add alternative for 16-bit encoding.
4410         (arm_one_cmplsi2): Likewise.
4412 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4414         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4415         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4416         (movdfcc): Likewise.
4417         * config/arm/vfp.md (*thumb2_movsf_vfp):
4418         Disable predication for arm_restrict_it.
4419         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4420         (*thumb2_movdfcc_vfp): Likewise.
4421         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4422         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4423         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4424         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4425         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4426         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4427         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4428         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4429         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4430         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4431         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4432         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4433         Disable predication for arm_restrict_it.
4435 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4437         * config/i386/bmiintrin.h (_bextr_u32): New.
4438         (_bextr_u64): Ditto.
4440 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4442         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4443         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4444         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4445         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4446         * config/mips/n32-elf.h: ...this new file.
4448 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4450         PR target/57224
4451         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4452         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4454 2013-06-27  Catherine Moore <clm@codesourcery.com>
4456         * config/mips/mips-tables.opt: Regenerate.
4457         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4458         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4459         * doc/invoke.texi: Add -m14kc.
4461 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4463         PR target/57623
4464         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4465         constraints of operand 1 and 2.
4467         PR target/57623
4468         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4469         to match RTL canonicalization.  Swap predicates and
4470         constraints of operand 1 and 2.
4472 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4474         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4475         Process OP_INOUT regs for splitting too.
4477 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4479         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4480         decl before the loop, initialize to NULL.
4481         (vectorizable_load): Initialize ptr_incr to NULL.
4483 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4485         PR lto/57208
4486         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4487         * ipa-ref.c (ipa_maybe_record_reference): New function.
4488         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4489         * ipa-cp.c (create_specialized_node): Record potential references from
4490         aggvals.
4491         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4493 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4495         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4496         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4497         'mode' to force_reg.
4498         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4499         (aarch64_expand_mov_immediate): Likewise.
4501 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4503         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4504         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4506 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4508         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4509         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4510         (struct machine_function): Add tbegin_p.
4511         (s390_canonicalize_comparison): Fold CC mode compares to
4512         conditional jump if possible.
4513         (s390_emit_jump): Return the emitted jump.
4514         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4515         Handle CCRAWmode compares.
4516         (s390_option_override): Default to -mhtm if available.
4517         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4518         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4519         FPRs instead of df_regs_ever_live_p.
4520         (s390_optimize_nonescaping_tx): New function.
4521         (s390_init_frame_layout): Extend clobbered_regs array to cover
4522         FPRs as well.
4523         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4524         (s390_expand_tbegin): New function.
4525         (enum s390_builtin): New enum definition.
4526         (code_for_builtin): New array definition.
4527         (s390_init_builtins): New function.
4528         (s390_expand_builtin): New function.
4529         (TARGET_INIT_BUILTINS): Define.
4530         (TARGET_EXPAND_BUILTIN): Define.
4531         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4532         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4533         (s390_alc_comparison): Likewise.
4534         * config/s390/s390-modes.def: Add CCRAWmode.
4535         * config/s390/s390.h (processor_flags): Add PF_TX.
4536         (TARGET_CPU_HTM): Define macro.
4537         (TARGET_HTM): Define macro.
4538         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4539         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4540         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4541         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4542         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4543         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4544         ("*cc_to_int"): Move up.
4545         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4546         constants other than 0.
4547         ("*ccraw_to_int"): New insn and splitter definition.
4548         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4549         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4550         ("tx_assist"): New expander.
4551         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4552         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4553         * config/s390/s390.opt: Add -mhtm option.
4554         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4555         * config/s390/htmxlintrin.h: New file.
4556         * config/s390/htmintrin.h: New file.
4557         * config/s390/s390intrin.h: New file.
4558         * doc/extend.texi: Document htm builtins.
4559         * config.gcc: Add the new header files to extra_headers.
4561 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4563         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4564         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4566 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4567             Pat Haugen <pthaugen@us.ibm.com>
4568             Peter Bergner <bergner@vnet.ibm.com>
4570         * config/rs6000/power8.md: New.
4571         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4572         setting for power8 entry.
4573         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4574         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4575         test for Power4/Power5 only.
4576         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4577         support.
4578         (force_new_group): Adjust comment.
4579         * config/rs6000/rs6000.md: Include power8.md.
4581 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4583         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4584         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4585         declaration.
4586         (tune_params): Update comment.
4587         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4588         (arm_max_conditional_execute): New function.
4589         (thumb2_final_prescan_insn): Use max_insn_skipped and
4590         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4592 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4594         PR tree-optimization/57705
4595         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4596         SSA_NAME step, provided that it is not defined inside the loop.
4597         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4598         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4600 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4602         PR middle-end/57670
4603         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4604         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4605         calls in the dump.
4606         (ipa_note_param_call): Initialize member_ptr flag.
4607         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4608         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4609         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4610         (ipa_read_indirect_edge_info): Likewise.
4612 2013-06-25  Richard Biener  <rguenther@suse.de>
4614         PR middle-end/56977
4615         * passes.c (init_optimization_passes): Move pass_fold_builtins
4616         and pass_dce earlier with -Og.
4618 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4620         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4621         <BIT_FIELD_REF>: Remove trailing TAB.
4622         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4623         remove blank line.
4625 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4627         PR tree-optimization/57358
4628         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4629         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4630         (ipa_analyze_params_uses): Generate pessimistic info when true.
4632 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4634         PR tree-optimization/57539
4635         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4636         global.inlined_to of the new node to it.  All callers changed.
4637         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4638         inlining_into, pass it to cgraph_clone_node.
4639         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4640         ipa_free_edge_args_substructures.
4641         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4642         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4643         Assert we have found an rdesc in the rdesc list.
4645 2013-06-24  Richard Biener  <rguenther@suse.de>
4647         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4648         (pointer_set_lookup): Declare.
4649         (class pointer_map): New template class implementing a
4650         generic pointer to T map.
4651         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4652         pointer_map<T>::contains, pointer_map<T>::insert,
4653         pointer_map<T>::traverse): New functions.
4654         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4655         (pointer_set_lookup): New function.
4656         (pointer_set_contains): Use pointer_set_lookup.
4657         (pointer_set_insert): Likewise.
4658         (insert_aux): Remove.
4659         (struct pointer_map_t): Embed a pointer_set_t.
4660         (pointer_map_create): Adjust.
4661         (pointer_map_destroy): Likewise.
4662         (pointer_map_contains): Likewise.
4663         (pointer_map_insert): Likewise.
4664         (pointer_map_traverse): Likewise.
4665         * tree-streamer.h (struct streamer_tree_cache_d): Use a
4666         pointer_map<unsigned> instead of a pointer_map_t.
4667         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4668         (streamer_tree_cache_lookup): Likewise.
4669         (streamer_tree_cache_create): Likewise.
4670         (streamer_tree_cache_delete): Likewise.
4671         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4672         pointer_map<unsigned> instead of a pointer_map_t.
4673         (lto_init_tree_ref_encoder): Adjust.
4674         (lto_destroy_tree_ref_encoder): Likewise.
4675         * lto-section-out.c (lto_output_decl_index): Likewise.
4676         (lto_record_function_out_decl_state): Likewise.
4677         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4679 2013-06-24  Richard Biener  <rguenther@suse.de>
4681         PR tree-optimization/57488
4682         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4684 2013-06-24  Alan Modra  <amodra@gmail.com>
4686         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4687         (gen_easy_altivec_constant): Likewise.
4688         * config/rs6000/predicates.md (easy_vector_constant_add_self,
4689         easy_vector_constant_msb): Likewise.
4691 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
4693         PR target/57688
4694         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4695         add missing return true.
4697 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
4699         PR target/52483
4700         * config/sh/predicates.md (general_extend_operand): Invoke
4701         general_movsrc_operand for memory operands.
4702         (general_movsrc_operand): Allow reg+reg addressing, do not use
4703         general_operand for memory operands.
4705 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
4707         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
4708         when current target options does not apply.
4709         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4710         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4711         * config/i386/bmiintrin.h: Pass appropriate target
4712         attributes to header.
4713         * config/i386/mmintrin.h: Ditto.
4714         * config/i386/nmmintrin.h: Ditto.
4715         * config/i386/avx2intrin.h: Ditto.
4716         * config/i386/fxsrintrin.h: Ditto.
4717         * config/i386/tbmintrin.h: Ditto.
4718         * config/i386/xsaveintrin.h: Ditto.
4719         * config/i386/f16cintrin.h: Ditto.
4720         * config/i386/xtestintrin.h: Ditto.
4721         * config/i386/xsaveoptintrin.h: Ditto.
4722         * config/i386/bmi2intrin.h: Ditto.
4723         * config/i386/lzcntintrin.h: Ditto.
4724         * config/i386/smmintrin.h: Ditto.
4725         * config/i386/wmmintrin.h: Ditto.
4726         * config/i386/x86intrin.h: Remove all header include guards.
4727         * config/i386/prfchwintrin.h: Ditto.
4728         * config/i386/pmmintrin.h: Ditto.
4729         * config/i386/tmmintrin.h: Ditto.
4730         * config/i386/xmmintrin.h: Ditto.
4731         * config/i386/popcntintrin.h: Ditto.
4732         * config/i386/rdseedintrin.h: Ditto.
4733         * config/i386/ammintrin.h: Ditto.
4734         * config/i386/emmintrin.h: Ditto.
4735         * config/i386/immintrin.h: Remove all header include guards.
4736         * config/i386/fma4intrin.h: Ditto.
4737         * config/i386/lwpintrin.h: Ditto.
4738         * config/i386/xopintrin.h: Ditto.
4739         * config/i386/ia32intrin.h: Ditto.
4740         * config/i386/avxintrin.h: Ditto.
4741         * config/i386/rtmintrin.h: Ditto.
4742         * config/i386/fmaintrin.h: Ditto.
4743         * config/i386/mm3dnow.h: Ditto.
4745 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
4747         * common/config/i386/i386-common.c: Handle LZCNT.
4749 2013-06-22  Andi Kleen  <ak@linux.intel.com>
4751         * doc/extend.texi: Use __atomic_store_n instead of
4752         __atomic_store in HLE example.
4754 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
4756         * config/sh/sh.c: Remove <cstdlib> workaround.
4758 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4760         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
4762 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4764         * doc/extend.texi: Document that __atomic_clear and
4765           __atomic_test_and_set should only be used with bool.
4767 2013-06-20  Jan Hubicka  <jh@suse.cz>
4769         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4770         types_same_for_odr.
4771         * tree.c (decls_same_for_odr): New function.
4772         (same_for_edr): New function.
4773         (types_same_for_odr): New function.
4774         (get_binfo_at_offset): Use it.
4775         * tree.h (types_same_for_odr): Declare.
4777 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
4778             Jason Merrill  <jason@redhat.com>
4780         * system.h: Include <cstdlib> as well as <stdlib.h>.
4782 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
4784         PR target/57655
4785         * config/i386/i386.c (construct_container): Report error if
4786         long double is used with disabled x87 float returns.
4788 2013-06-20  Jan Hubicka  <jh@suse.cz>
4790         * lto-cgraph.c (input_symtab): Do not set cgraph state.
4792 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
4794         PR rtl-optimization/57425
4795         PR rtl-optimization/57569
4796         * alias.c (write_dependence_p): Remove parameters mem_mode and
4797         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
4798         Changed all callers.
4799         (canon_anti_dependence): Get comments and semantics in sync.
4800         Add parameter mem_canonicalized.  Changed all callers.
4801         * rtl.h (canon_anti_dependence): Update prototype.
4803 2013-06-20  Richard Biener  <rguenther@suse.de>
4805         * data-streamer-in.c (streamer_read_uhwi): Optimize single
4806         byte case, inline streamer_read_uchar and defer section
4807         overrun check.
4809 2013-06-20  Richard Biener  <rguenther@suse.de>
4811         PR tree-optimization/57584
4812         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4813         SSA names into the expanded expression that take part in
4814         abnormal coalescing.
4816 2013-06-19  Sharad Singhai  <singhai@google.com>
4818         * gcov.c (print_usage): Handle new option.
4819         (process_args): Ditto.
4820         (get_gcov_intermediate_filename): New function.
4821         (output_intermediate_file): New function.
4822         (output_gcov_file): New function
4823         (generate_results): Handle new option.
4824         (release_function): Relase demangled name.
4825         (read_graph_file): Handle demangled name.
4826         (output_lines): Ditto.
4827         * doc/gcov.texi: Document gcov intermediate format.
4829 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
4831         PR bootstrap/57604
4832         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4833         (lra_emit_add): Use the functions.  Add comment about Y as an
4834         address segment.
4836 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
4838         PR driver/57652
4839         * collect2.c (collect_atexit): New.
4840         (collect_exit): Delete.
4841         (main): Register collect_atexit with atexit.
4842         (collect_wait): Change collect_exit to exit.
4843         (do_wait): Same.
4844         * collect2.h (collect_exit): Delete.
4845         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4847 2013-06-19  Wei Mi  <wmi@google.com>
4849         PR rtl-optimization/57518
4850         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4851         if regno is used in paradoxical subreg.
4852         (update_equiv_regs): Check pdx_subregs[regno] before
4853         set a reg to be equivalent with a mem.
4855 2013-06-19  Matthias Klose  <doko@ubuntu.com>
4857         PR driver/57651
4858         * file-find.h (find_a_file): Add a mode parameter.
4859         * file-find.c (find_a_file): Likewise.
4860         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4861         with X_OK for the executables.
4862         * collect2.c (main): Call find_a_file with X_OK.
4864 2013-06-19  Steve Ellcey  <sellcey@mips.com>
4866         PR target/56942
4867         * config/mips/mips.md (casesi_internal_mips16_<mode>):
4868         Use NEXT_INSN instead of next_real_insn.
4870 2013-06-19  Jan Hubicka  <jh@suse.cz>
4872         * cgraph.h (const_value_known_p): Replace by ...
4873         (ctor_for_folding): .. this one.
4874         * cgraphunit.c (process_function_and_variable_attributes): Use it.
4875         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4876         * expr.c (expand_expr_real_1): Likewise.
4877         (string_constant): Likewise.
4878         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4879         * ipa.c (process_references): Likewise.
4880         (symtab_remove_unreachable_nodes): Likewise.
4881         * ipa-inline-analysis.c (param_change_prob): Likewise.
4882         * gimple-fold.c (canonicalize_constructor_val): Likewise.
4883         (get_base_constructor): Likwise.
4884         * varpool.c (varpool_remove_node): Likewise.
4885         (varpool_remove_initializer): LIkewise.
4886         (dump_varpool_node): LIkwise.
4887         (const_value_known_p): Rewrite to ...
4888         (ctor_for_folding): ... this one.
4890 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
4892         PR driver/57651
4893         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4894         PERSONALITY in $PATH derived prefixes.
4896 2013-06-19  Jeff Law  <law@redhat.com>
4898         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4899         in comment.
4901         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4902         (simplify_bitwise_binary): Use it to simpify certain binary ops on
4903         booleans.
4905 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
4907         * config/arm/vfp.md: Move VFP instruction classification documentation
4908         to ...
4909         * config/arm/arm.md: ... here.  Update instruction classification
4910         documentation.
4912 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
4914         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4915         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4916         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
4918 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
4920         PR target/57609
4921         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4922         with NEXT_INSN.  Use tablejump_p to check for jump table data
4923         insns.
4925 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
4927         PR c++/56544
4928         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4929         that now in C++ the value is correct per the C++ standards.
4931 2013-06-19  Richard Biener  <rguenther@suse.de>
4933         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4934         for global context.
4936 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4938         Revert:
4939         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4941         PR target/57609
4942         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4943         with next_active_insn.
4945 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
4947         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4948         functions are inlined during failures to flag an error.
4949         * tree-inline.c (expand_call_inline): Allow the error to be flagged
4950         in early inline pass.
4952 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
4954         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4955         in comments.
4957 2013-06-18  Julian Brown  <julian@codesourcery.com>
4959         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4960         Permit virtual register pre-reload if !strict.
4961         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4962         change.
4963         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4964         prototype.
4965         * config/arm/neon.md (movmisalign<mode>): Use
4966         neon_perm_struct_or_reg_operand instead of
4967         neon_struct_or_register_operand.
4968         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4969         neon_permissive_struct_operand instead of neon_struct_operand.
4970         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4971         neon_vector_mem_operand.
4972         * config/arm/predicates.md (neon_struct_operand): Adjust call to
4973         neon_vector_mem_operand.
4974         (neon_permissive_struct_operand): New.
4975         (neon_struct_or_register_operand): Rename to...
4976         (neon_perm_struct_or_reg_operand): This. Adjust call to
4977         neon_vector_mem_operand.
4979 2013-06-18  Richard Biener  <rguenther@suse.de>
4981         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4982         * lto-streamer.h: Include pointer-set.h.
4983         (struct lto_decl_slot): Remove.
4984         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4985         Remove next_index entry.
4986         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4987         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4988         (lto_init_tree_ref_encoder): Adjust.
4989         (lto_destroy_tree_ref_encoder): Likewise.
4990         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4991         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4992         (lto_output_decl_index): Adjust.
4993         (lto_new_out_decl_state): Likewise.
4994         (lto_record_function_out_decl_state): Likewise.
4995         * lto-streamer-out.c (copy_function): Likewise.
4997 2013-06-18  Richard Biener  <rguenther@suse.de>
4999         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5000         * cgraphunit.c: Include cfgloop.h.
5001         (init_lowered_empty_function): Initialize the loop tree.
5002         (assemble_thunk): Insert new BBs into loops.
5004 2013-06-18  Richard Biener  <rguenther@suse.de>
5006         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5007         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5008         the map from cache entry to cache index optional.
5009         (streamer_tree_cache_replace_tree): Adjust accordingly.
5010         (streamer_tree_cache_append): Likewise.
5011         (streamer_tree_cache_delete): Likewise.
5012         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5013         streamer cache map from cache entry to cache index.
5014         * lto-streamer-out.c (create_output_block): Adjust.
5016 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5018         * config/arm/arm.md (attribute "insn"): Move multiplication and
5019         division attributes to...
5020         (attribute "type"): ... here.  Remove mult.
5021         (attribute "mul32"): New attribute.
5022         (attribute "mul64"): Add umaal.
5023         (*arm_mulsi3): Update attributes.
5024         (*arm_mulsi3_v6): Likewise.
5025         (*thumb_mulsi3): Likewise.
5026         (*thumb_mulsi3_v6): Likewise.
5027         (*mulsi3_compare0): Likewise.
5028         (*mulsi3_compare0_v6): Likewise.
5029         (*mulsi_compare0_scratch): Likewise.
5030         (*mulsi_compare0_scratch_v6): Likewise.
5031         (*mulsi3addsi): Likewise.
5032         (*mulsi3addsi_v6): Likewise.
5033         (*mulsi3addsi_compare0): Likewise.
5034         (*mulsi3addsi_compare0_v6): Likewise.
5035         (*mulsi3addsi_compare0_scratch): Likewise.
5036         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5037         (*mulsi3subsi): Likewise.
5038         (*mulsidi3adddi): Likewise.
5039         (*mulsi3addsi_v6): Likewise.
5040         (*mulsidi3adddi_v6): Likewise.
5041         (*mulsidi3_nov6): Likewise.
5042         (*mulsidi3_v6): Likewise.
5043         (*umulsidi3_nov6): Likewise.
5044         (*umulsidi3_v6): Likewise.
5045         (*umulsidi3adddi): Likewise.
5046         (*umulsidi3adddi_v6): Likewise.
5047         (*smulsi3_highpart_nov6): Likewise.
5048         (*smulsi3_highpart_v6): Likewise.
5049         (*umulsi3_highpart_nov6): Likewise.
5050         (*umulsi3_highpart_v6): Likewise.
5051         (mulhisi3): Likewise.
5052         (*mulhisi3tb): Likewise.
5053         (*mulhisi3bt): Likewise.
5054         (*mulhisi3tt): Likewise.
5055         (maddhisi4): Likewise.
5056         (*maddhisi4tb): Likewise.
5057         (*maddhisi4tt): Likewise.
5058         (maddhidi4): Likewise.
5059         (*maddhidi4tb): Likewise.
5060         (*maddhidi4tt): Likewise.
5061         (divsi3): Likewise.
5062         (udivsi3): Likewise.
5063         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5064         (thumb2_mulsi_short_compare0): Likewise.
5065         (thumb2_mulsi_short_compare0_scratch): Likewise.
5066         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5067         (1020mult2): Likewise.
5068         (1020mult3): Likewise.
5069         (1020mult4): Likewise.
5070         (1020mult5): Likewise.
5071         (1020mult6): Likewise.
5072         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5073         change.
5074         (cortex_a15_mult64): Likewise.
5075         (cortex_a15_sdiv): Likewise.
5076         (cortex_a15_udiv): Likewise.
5077         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5078         (mult2): Likewise.
5079         (mult3): Likewise.
5080         (mult4): Likewise.
5081         (mult5): Likewise.
5082         (mult6): Likewise.
5083         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5084         (pj4_ir_div): Likewise.
5085         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5086         (11_mult2): Likewise.
5087         (11_mult3): Likewise.
5088         (11_mult4): Likewise.
5089         (11_mult5): Likewise.
5090         (11_mult6): Likewise.
5091         (11_mult7): Likewise.
5092         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5093         (cortex_a8_mla): Likewise.
5094         (cortex_a8_mull): Likewise.
5095         (cortex_a8_smulwy): Likewise.
5096         (cortex_a8_smlald): Likewise.
5097         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5098         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5099         (cortex_r4_mul_3): Likewise.
5100         (cortex_r4_mla_4): Likewise.
5101         (cortex_r4_mla_3): Likewise.
5102         (cortex_r4_smlald): Likewise.
5103         (cortex_r4_mull): Likewise.
5104         (cortex_r4_sdiv): Likewise.
5105         (cortex_r4_udiv): Likewise.
5106         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5107         (cortex_a7_idiv): Likewise.
5108         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5109         (9_mult2): Likewise.
5110         (9_mult3): Likewise.
5111         (9_mult4): Likewise.
5112         (9_mult5): Likewise.
5113         (9_mult6): Likewise.
5114         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5115         (cortex_a53_sdiv): Likewise.
5116         (cortex_a53_udiv): Likewise.
5117         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5118         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5119         (mp626_mult2): Likewise.
5120         (mp626_mult3): Likewise.
5121         (mp626_mult4): Likewise.
5122         * config/arm/fa526.md (526_mult1): Update attribute change.
5123         (526_mult2): Likewise.
5124         * config/arm/arm-generic.md (mult): Update attribute change.
5125         (mult_ldsched_strongarm): Likewise.
5126         (mult_ldsched): Likewise.
5127         (multi_cycle): Likewise.
5128         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5129         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5130         (606te_mult2): Likewise.
5131         (606te_mult3): Likewise.
5132         (606te_mult4): Likewise.
5133         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5134         (cortex_a9_mac16): Likewise.
5135         (cortex_a9_multiply): Likewise.
5136         (cortex_a9_mac): Likewise.
5137         (cortex_a9_multiply_long): Likewise.
5138         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5139         (626te_mult2): Likewise.
5140         (626te_mult3): Likewise.
5141         (626te_mult4): Likewise.
5143 2013-06-18  Richard Biener  <rguenther@suse.de>
5145         PR lto/57334
5146         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5148 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5150         PR target/57609
5151         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5152         with next_active_insn.
5154 2013-06-18  Alan Modra  <amodra@gmail.com>
5156         * config/rs6000/rs6000.h (enum data_align): New.
5157         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5158         (DATA_ABI_ALIGNMENT): Define.
5159         (CONSTANT_ALIGNMENT): Correct comment.
5160         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5161         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5163 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5165         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5166         ATTRIBUTE_UNUSED marking.
5168 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5170         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5171         alternative and update.
5172         (aarch64_dup_lanedi): Delete.
5173         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5174         * config/aarch64/aarch64-simd-builtins.def: Update.
5176 2013-06-17  Richard Biener  <rguenther@suse.de>
5178         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5179         (lto_input_scc): Declare.
5180         (lto_input_tree_1): Likewise.
5181         (struct lto_stats_d): Add num_tree_bodies_output and
5182         num_pickle_refs_output.
5183         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5184         (lto_read_tree_1): Split out from ...
5185         (lto_read_tree): ... this.
5186         (lto_input_scc): New function.
5187         (lto_input_tree_1): Split out from ...
5188         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5189         (lto_data_in_create): Create the streamer cache without hashes.
5190         * lto-streamer-out.c (create_output_block): Create the streamer
5191         cache with hashes when not doing WPA.
5192         (lto_write_tree_1): Split out from ...
5193         (lto_write_tree): ... this.
5194         (get_symbol_initial_value): New function.
5195         (lto_output_tree_1): Split out from ...
5196         (lto_output_tree): ... this.  Write trees as series of SCCs
5197         using a DFS walk via DFS_write_tree.
5198         (struct sccs, struct scc_entry): New types.
5199         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5200         (DFS_write_tree_body): New function.
5201         (DFS_write_tree): Likewise.
5202         (hash_tree): Likewise.
5203         (scc_entry_compare): Likewise.
5204         (hash_scc): Likewise.
5205         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5206         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5207         TREE_CHAIN as regular reference.
5208         (streamer_read_integer_cst): Remove.
5209         (streamer_get_pickled_tree): Adjust.
5210         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5211         of DECL_EXTERNALs in BLOCK_VARS for now.
5212         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5213         reference.
5214         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5215         Add hash value argument and record that if hashes are recorded
5216         in the cache.
5217         (streamer_tree_cache_insert_1): Adjust.
5218         (streamer_tree_cache_insert): Likewise.
5219         (streamer_tree_cache_insert_at): Rename to ...
5220         (streamer_tree_cache_replace_tree): ... this and adjust.
5221         (streamer_tree_cache_append): Adjust.
5222         (record_common_node): Likewise.
5223         (streamer_tree_cache_create): Add argument whether to
5224         record hash values together with trees.
5225         (streamer_tree_cache_delete): Adjust.
5226         * tree-streamer.h (struct streamer_tree_cache_d): Add
5227         vector of hashes.
5228         (streamer_read_integer_cst): Remove.
5229         (streamer_tree_cache_insert): Adjust.
5230         (streamer_tree_cache_append): Likewise.
5231         (streamer_tree_cache_insert_at): Rename to ...
5232         (streamer_tree_cache_replace_tree): ... this and adjust.
5233         (streamer_tree_cache_create): Add argument whether to record hashes.
5234         (streamer_tree_cache_get): Rename to ...
5235         (streamer_tree_cache_get_tree): ... this.
5236         (streamer_tree_cache_get_hash): New function.
5237         * tree.c (cache_integer_cst): New function.
5238         * tree.h (cache_integer_cst): Declare.
5239         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5240         * lto-symtab.c (lto_varpool_replace_node): Only release
5241         DECL_INITIAL of non-prevailing decls.
5242         * varpool.c (varpool_remove_initializer): Do not release
5243         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5245 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5247         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5248         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5249         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5250         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5251         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5252         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5253         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5254         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5255         instead of TARGET_64BIT.
5256         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5257         Require ISA_HAS_<D>DIV.
5259 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5261         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5262         (mips*-*-linux*): Move default with_llsc setting to where other
5263         defaults are set.
5264         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5265         with_arch block.
5266         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5267         Likewise.  Remove default with_tune setting.  Move default float
5268         setting to its own block.  Handle with_llsc in the same block as above.
5270 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5272         PR rtl-optimization/57425
5273         PR rtl-optimization/57569
5274         * alias.c (write_dependence_p): Add new parameters mem_mode,
5275         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5276         Changed all callers.
5277         (canon_anti_dependence): New function.
5278         * cse.c (check_dependence): Use canon_anti_dependence.
5279         * cselib.c (cselib_invalidate_mem): Likewise.
5280         * rtl.h (canon_anti_dependence): Declare.
5282 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5284         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5285         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5286         ".set mips3" for 64-bit targets.
5288 2013-06-15  Dehao Chen  <dehao@google.com>
5290         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5291         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5292         (gimple_check_call_args): Likewise.
5293         * value-prof.c (check_ic_target): Likewise.
5294         * ipa-inline.c (early_inliner): Likewise.
5295         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5296         * cgraph.c (cgraph_create_edge_1): Likewise.
5297         (cgraph_make_edge_direct): Likewise.
5299 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5301         PR target/57615
5302         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5303         rs6000_output_move_128bit to handle emitting quad memory
5304         operations.  Set attribute length to 8 bytes.
5306 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5308         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5309         New pattern.
5310         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5311         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5312         (aarch64_<su>mlsl<mode>): Likewise.
5314 2013-06-14  Mike Stump  <mikestump@comcast.net>
5316         * Makefile.in (TARGET_H): Add insn-codes.h.
5318 2013-06-14  Alan Modra  <amodra@gmail.com>
5320         PR middle-end/57134
5321         PR middle-end/57586
5322         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5323         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5324         bitfield expansion when EXPAND_MEMORY.
5325         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5327 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5329         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5330         test for clearing quad memory on 32-bit later.
5332 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5334         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5335         (fold_negate_expr): Likewise.
5336         (fold_real_zero_addition_p): Handle vectors.
5337         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5339 2013-06-14  Alan Modra  <amodra@gmail.com>
5341         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5343 2013-06-13  Jan Hubicka  <jh@suse.cz>
5345         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5346         Local comdats are not externally visible.
5347         * symtab.c (dump_symtab_base): Dump externally visible.
5348         (verify_symtab_base): Verify back links in the symtab hash.
5350 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5352         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5353         CONVERT_EXPR as equal nodes.
5355 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5357         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5359 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5361         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5362         Generalize to complex and vector.
5363         * tree.c (build_all_ones_cst): New function.
5364         * tree.h (build_all_ones_cst): Declare it.
5366 2013-06-13  Alan Modra  <amodra@gmail.com>
5368         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5369         * config/rs6000/rs6000.md (signbittf2): New insn.
5370         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5371         (abstf2_internal, cmptf_internal2): Likewise.
5372         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5374 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5375             Pat Haugen <pthaugen@us.ibm.com>
5376             Peter Bergner <bergner@vnet.ibm.com>
5378         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5379         power8 byte, half-word, and quad-word atomic instructions.
5380         (emit_store_conditional): Likewise.
5381         (rs6000_expand_atomic_compare_and_swap): Likewise.
5382         (rs6000_expand_atomic_op): Likewise.
5384         * config/rs6000/sync.md (larx): Add new modes for power8.
5385         (stcx): Likewise.
5386         (AINT): New mode iterator to include TImode as well as normal
5387         integer modes on power8.
5388         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5389         that VSX registers are not considered.  Use AINT mode iterator
5390         instead of INT1 to allow inclusion of quad word atomic operations
5391         on power8.
5392         (load_locked<mode>): Likewise.
5393         (store_conditional<mode>): Likewise.
5394         (atomic_compare_and_swap<mode>): Likewise.
5395         (atomic_exchange<mode>): Likewise.
5396         (atomic_nand<mode>): Likewise.
5397         (atomic_fetch_<fetchop_name><mode>): Likewise.
5398         (atomic_nand_fetch<mode>): Likewise.
5399         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5400         each type.
5401         (ATOMIC): On power8, add QImode, HImode modes.
5402         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5403         modes that promote to SImode.
5404         (load_lockedti): Convert TImode arguments to PTImode, so that we
5405         get a guaranteed even/odd register pair.
5406         (load_lockedpti): Likewise.
5407         (store_conditionalti): Likewise.
5408         (store_conditionalpti): Likewise.
5410         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5411         atomic load/store instructions.
5412         (HSI): Likewise.
5414 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5416         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5417         loads.
5418         (insn_count): New attribute, with most cases extracted from...
5419         (length): ...here.  Redefine most cases in terms of insn_count.
5420         (single_insn): Delete.
5421         (can_delay): Use insn_count to check for single instructions.
5422         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5423         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5424         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5425         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5426         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5427         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5428         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5429         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5430         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5431         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5432         rather than "length".
5433         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5434         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5435         Use "insn_count" rather than "length".
5436         * config/mips/mips-dsp.md
5437         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5438         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5439         length attributes.
5441 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5443         PR tree-optimization/57361
5444         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5446 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5448         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5449         to split.
5450         (aarch64_simd_combine<mode>): New instruction expansion.
5451         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5452         function prototype.
5453         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5454         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5456 2013-06-12  Jan Hubicka  <jh@suse.cz>
5458         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5459         decl has when in streaming stage.
5460         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5461         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5463 2013-06-12  Roland Stigge <stigge@antcom.de>
5465         PR target/57578
5466         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5468 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5470         PR tree-optimization/57537
5471         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5472         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5474 2013-06-12  Richard Biener  <rguenther@suse.de>
5476         * data-streamer.h (streamer_write_char_stream): CSE
5477         obs->current_pointer.
5478         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5479         streamer_write_char_stream manually and optimize the resulting loop.
5480         (streamer_write_hwi_stream): Likewise.
5482 2013-06-12  Jan Hubicka  <jh@suse.cz>
5484         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5485         * cgraph.h (varpool_create_empty_node): Declare.
5486         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5487         duplicated nodes.
5488         * symtab.c (symtab_unregister_node): Be lax about missin entries
5489         in node hash.
5490         (symtab_get_node): Update comment.
5491         * varpool.c (varpool_create_empty_node): Break out from ...
5492         (varpool_node_for_decl): ... here.
5493         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5495 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5497         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5498         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5499         part.  Use straight-line flow at the end.
5500         <COMPONENT_REF>: Remove superfluous else.
5501         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5503 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5505         PR target/56564
5506         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5507         target hook even for !TREE_PUBLIC decls.  If no resolution info
5508         is available, return false for common and external decls.
5510 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5512         * config/rl78/constraints.md (U): New constraint.
5513         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5514         valloc attribute.
5516 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5518         PR target/57589
5519         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5520         to allow returning address to AT_PLATFORM name.
5522 2013-06-11  Jan Hubicka  <jh@suse.cz>
5524         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5525         * cgraph.h (symtab_node_base): Add weakref flag.
5526         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5527         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5528         (output_weakrefs): Use weakref flag.
5529         * fold-const.c (simple_operand_p): Handle WEAK.
5530         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5531         * ipa.c (varpool_externally_visible_p): Drop weakref.
5532         (function_and_variable_visibility): Update comment; fix weakref
5533         sanity checks; do not clear DECL_WEAK on them.
5534         * lto-cgraph.c (lto_output_node): update.
5535         (lto_output_varpool_node): Update.
5536         (input_overwrite_node): Update.
5537         (input_node): Update.
5538         (input_varpool_node): Update.
5539         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5540         (lto_symtab_merge_symbols): Add sanity check.
5541         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5542         * passes.c (rest_of_decl_compilation): Set static flag, too.
5543         * symtab.c (dump_symtab_base): Dump weakref.
5544         (verify_symtab_base): Sanity check weakrefs.
5545         (symtab_make_decl_local): Remove duplicated code.
5546         (symtab_alias_ultimate_target): Simplify.
5547         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5549 2013-06-11  DJ Delorie  <dj@redhat.com>
5551         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5552         (rl78_unwind_word_mode): New.
5554 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5556         * final.c (debug_prefix_maps): Make static.
5558 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5560         * function.c (initial_trampoline): Remove stray copy.
5562 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5564         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5566 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5568         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5569         within bounds at the beginning of the function.
5571 2013-06-11  Alan Modra  <amodra@gmail.com>
5573         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5574         reporting.
5575         (get_named_section): Don't NULL !DECL_P decl.
5577 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5579         * doc/invoke.texi (core-avx2): Document.
5580         (slm): Likewise.
5581         (atom): Updated with MOVBE.
5583 2013-06-11  Richard Biener  <rguenther@suse.de>
5585         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5587 2013-06-11  Anton Blanchard  <anton@samba.org>
5589         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5590         correct shift value in little-endian mode.
5592 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5594         PR target/56564
5595         * varasm.c (get_variable_align): Move #endif to the right place.
5597 2013-06-10  Cary Coutant  <ccoutant@google.com>
5599         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5600         for hash so that hash table traversal order is deterministic.
5602 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5603             Pat Haugen <pthaugen@us.ibm.com>
5604             Peter Bergner <bergner@vnet.ibm.com>
5606         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5607         of vectors in GPRS if they are direct moves or quad word load or
5608         store moves.
5610         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5611         declaration.
5612         (direct_move_p): Likewise.
5613         (quad_load_store_p): Likewise.
5615         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5616         classes into bins based on the physical register type.
5617         (reg_class_to_reg_type): Likewise.
5618         (IS_STD_REG_TYPE): Likewise.
5619         (IS_FP_VECT_REG_TYPE): Likewise.
5620         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5621         use direct move instructions.
5622         (reload_gpr_vsx): Likewise.
5623         (reload_vsx_gpr): Likewise.
5624         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5625         information that is a simplification of register classes.  Also
5626         precalculate direct move reload helpers.
5627         (direct_move_p): New function to return true if the operation can
5628         be done as a direct move instruciton.
5629         (quad_load_store_p): New function to return true if the operation
5630         is a quad memory operation.
5631         (rs6000_legitimize_address): If quad memory, only allow register
5632         indirect for TImode addresses.
5633         (rs6000_legitimate_address_p): Likewise.
5634         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5635         (rs6000_reload_register_type): Likewise.
5636         (register_to_reg_type): Return register type.
5637         (rs6000_secondary_reload_simple_move): New helper function for
5638         secondary reload and secondary memory needed to identify anything
5639         that is a simple move, and does not need reloading.
5640         (rs6000_secondary_reload_direct_move): New helper function for
5641         secondary reload to identify cases that can be done with several
5642         instructions via the direct move instructions.
5643         (rs6000_secondary_reload_move): New helper function for secondary
5644         reload to identify moves between register types that can be done.
5645         (rs6000_secondary_reload): Add support for quad memory operations
5646         and for direct move.
5647         (rs6000_secondary_memory_needed): Likewise.
5648         (rs6000_debug_secondary_memory_needed): Change argument names.
5649         (rs6000_output_move_128bit): New function to return the move to
5650         use for 128-bit moves, including knowing about the various
5651         limitations of quad memory operations.
5653         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5654         memory operations.  call rs6000_output_move_128bit for the actual
5655         instruciton(s) to generate.
5656         (vsx_movti_64bit): Likewise.
5658         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5659         (UNSPEC_P8V_MTVSRWZ): Likewise.
5660         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5661         (UNSPEC_P8V_MTVSRD): Likewise.
5662         (UNSPEC_P8V_XXPERMDI): Likewise.
5663         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5664         (UNSPEC_FUSION_GPR): Likewise.
5665         (FMOVE128_GPR): New iterator for direct move.
5666         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5667         (f32_sv): Likewise.
5668         (f32_dm): Likewise.
5669         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5670         loads and direct move instructions.
5671         (zero_extendsidi2_lfiwzx): Likewise.
5672         (extendsidi2_lfiwax): Likewise.
5673         (extendsidi2_nocell): Likewise.
5674         (floatsi<mode>2_lfiwax): Likewise.
5675         (lfiwax): Likewise.
5676         (floatunssi<mode>2_lfiwzx): Likewise.
5677         (lfiwzx): Likewise.
5678         (fix_trunc<mode>_stfiwx): Likewise.
5679         (fixuns_trunc<mode>_stfiwx): Likewise.
5680         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5681         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5682         (parity<mode>2_cmpb): Set length/type attr.
5683         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5684         for 'mr.' to fast_compare.
5685         (bpermd_<mode>): Change type attr to popcnt.
5686         (p8_fmrgow_<mode>): New insns for power8 direct move support.
5687         (p8_mtvsrwz_1): Likewise.
5688         (p8_mtvsrwz_2): Likewise.
5689         (reload_fpr_from_gpr<mode>): Likewise.
5690         (p8_mtvsrd_1): Likewise.
5691         (p8_mtvsrd_2): Likewise.
5692         (p8_xxpermdi_<mode>): Likewise.
5693         (reload_vsx_from_gpr<mode>): Likewise.
5694         (reload_vsx_from_gprsf): Likewise.
5695         (p8_mfvsrd_3_<mode>): LIkewise.
5696         (reload_gpr_from_vsx<mode>): Likewise.
5697         (reload_gpr_from_vsxsf): Likewise.
5698         (p8_mfvsrd_4_disf): Likewise.
5699         (multi-word GPR splits): Do not split direct moves or quad memory
5700         operations.
5702 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
5704         * tree-into-ssa.c (interesting_blocks): Make static.
5706 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
5708         PR target/56564
5709         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5710         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5711         Use DATA_ABI_ALIGNMENT for that case instead if defined.
5712         (get_variable_align): New function.
5713         (get_variable_section, emit_bss, emit_common,
5714         assemble_variable_contents, place_block_symbol): Use
5715         get_variable_align instead of DECL_ALIGN.
5716         (assemble_noswitch_variable): Add align argument, use it
5717         instead of DECL_ALIGN.
5718         (assemble_variable): Adjust caller.  Use get_variable_align
5719         instead of DECL_ALIGN.
5720         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5721         caller.
5722         (DATA_ABI_ALIGNMENT): Define.
5723         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5724         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
5725         opt is false, only return the psABI mandated alignment increase.
5726         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5727         (DATA_ABI_ALIGNMENT): ... this.
5728         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5729         (DATA_ABI_ALIGNMENT): ... this.
5730         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5731         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5732         (DATA_ABI_ALIGNMENT): ... this.
5733         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5734         * doc/tm.texi: Regenerated.
5736 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
5738         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5739         cmp_code to construct REG_EQUAL note.
5741 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
5743         PR target/57568
5744         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5745         that operands[2] doesn't overlap with operands[0].
5747 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
5748             Jan Hubicka  <jh@suse.cz>
5750         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5751         hack to mark symbols as used.
5753 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
5755         PR rtl-optimization/57559
5756         * lra-constraints.c (process_alt_operands): Don't discourage
5757         memory with known offset for offsetable memory constraint.
5758         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5760 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
5762         * varasm.c (struct oc_local_state): Reorder fields.
5763         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5764         and adjust accordingly.
5765         (output_constructor): Reorder initialization code and adjust call to
5766         output_constructor_bitfield.
5768 2013-06-07  Jan Hubicka  <jh@suse.cz>
5770         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5772 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
5774         * tree-object-size.c (unknown): Make const.
5776 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5778         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5779         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5780         for last alternative in the cpu_facility attribute.
5782 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5784         PR target/56315
5785         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5786         (xordi3): Change operand 2 constraint to arm_xordi_operand.
5787         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5788         * config/arm/constraints.md (Dg): New constraint.
5789         * config/arm/neon.md (xordi3_neon): Remove.
5790         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5791         * config/arm/predicates.md (arm_xordi_operand): New predicate.
5793 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5795         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5796         Clean up alternatives.
5798 2013-06-07  Alan Modra  <amodra@gmail.com>
5800         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5801         va_list_gpr_size.
5803 2013-06-07  Alan Modra  <amodra@gmail.com>
5805         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5807 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5809         * config/arm/constraints.md (Df): New constraint.
5810         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5811         Correct length attribute for last two alternatives.
5813 2013-06-07  Alan Modra  <amodra@gmail.com>
5815         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5816         override user -mfp-in-toc.
5817         (offsettable_ok_by_alignment): Consider just the current access
5818         rather than the whole object, unless BLKmode.  Handle
5819         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5820         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5821         for -mcmodel=medium.
5822         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5823         override user -mfp-in-toc or -msum-in-toc.  Default to
5824         -mno-fp-in-toc for -mcmodel=medium.
5826 2013-06-06  DJ Delorie  <dj@redhat.com>
5828         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5829         TARGET_VALID_POINTER_MODE.
5831 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
5832             Pat Haugen <pthaugen@us.ibm.com>
5833             Peter Bergner <bergner@vnet.ibm.com>
5835         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5836         Document new power8 builtins.
5838         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5839         condition code register, to allow 128-bit logical operations to be
5840         done in the VSX or GPR registers.
5841         (nor<mode>3): Use the canonical form for nor.
5842         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5843         vclz*, and vpopcnt* vector instructions.
5844         (nand<mode>3): Likewise.
5845         (orc<mode>3): Likewise.
5846         (clz<mode>2): LIkewise.
5847         (popcount<mode>2): Likewise.
5849         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5850         that only the GPRs are recognized.
5852         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5853         support for new power8 builtins.
5855         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5856         builtin functions.
5857         (xscvdpspn): Likewise.
5858         (vclz): Likewise.
5859         (vclzb): Likewise.
5860         (vclzh): Likewise.
5861         (vclzw): Likewise.
5862         (vclzd): Likewise.
5863         (vpopcnt): Likewise.
5864         (vpopcntb): Likewise.
5865         (vpopcnth): Likewise.
5866         (vpopcntw): Likewise.
5867         (vpopcntd): Likewise.
5868         (vgbbd): Likewise.
5869         (vmrgew): Likewise.
5870         (vmrgow): Likewise.
5871         (eqv): Likewise.
5872         (eqv_v16qi3): Likewise.
5873         (eqv_v8hi3): Likewise.
5874         (eqv_v4si3): Likewise.
5875         (eqv_v2di3): Likewise.
5876         (eqv_v4sf3): Likewise.
5877         (eqv_v2df3): Likewise.
5878         (nand): Likewise.
5879         (nand_v16qi3): Likewise.
5880         (nand_v8hi3): Likewise.
5881         (nand_v4si3): Likewise.
5882         (nand_v2di3): Likewise.
5883         (nand_v4sf3): Likewise.
5884         (nand_v2df3): Likewise.
5885         (orc): Likewise.
5886         (orc_v16qi3): Likewise.
5887         (orc_v8hi3): Likewise.
5888         (orc_v4si3): Likewise.
5889         (orc_v2di3): Likewise.
5890         (orc_v4sf3): Likewise.
5891         (orc_v2df3): Likewise.
5893         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5894         allow power8 quad mode in 64-bit.
5895         (rs6000_builtin_vectorized_function): Add support to vectorize
5896         ISA 2.07 count leading zeros, population count builtins.
5897         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5898         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5899         (builtin_function_type): Add vgbbd builtin function which takes an
5900         unsigned argument.
5901         (altivec_expand_vec_perm_const): Add support for new power8 merge
5902         instructions.
5904         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5905         that does not include TImdoe for use with 32-bit.
5906         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5907         instructions.
5908         (UNSPEC_VSX_CVDPSPN): Likewise.
5909         (vsx_xscvdpspn): Likewise.
5910         (vsx_xscvspdpn): Likewise.
5911         (vsx_xscvdpspn_scalar): Likewise.
5912         (vsx_xscvspdpn_directmove): Likewise.
5913         (vsx_and<mode>3): Split logical operations into 32-bit and
5914         64-bit. Add support to do logical operations on TImode as well as
5915         VSX vector types.  Allow logical operations to be done in either
5916         VSX registers or in general purpose registers in 64-bit mode.  Add
5917         splitters if GPRs were used. For AND, add clobber of CCmode to
5918         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
5919         encoding.
5920         (vsx_and<mode>3_32bit): Likewise.
5921         (vsx_and<mode>3_64bit): Likewise.
5922         (vsx_ior<mode>3): Likewise.
5923         (vsx_ior<mode>3_32bit): Likewise.
5924         (vsx_ior<mode>3_64bit): Likewise.
5925         (vsx_xor<mode>3): Likewise.
5926         (vsx_xor<mode>3_32bit): Likewise.
5927         (vsx_xor<mode>3_64bit): Likewise.
5928         (vsx_one_cmpl<mode>2): Likewise.
5929         (vsx_one_cmpl<mode>2_32bit): Likewise.
5930         (vsx_one_cmpl<mode>2_64bit): Likewise.
5931         (vsx_nor<mode>3): Likewise.
5932         (vsx_nor<mode>3_32bit): Likewise.
5933         (vsx_nor<mode>3_64bit): Likewise.
5934         (vsx_andc<mode>3): Likewise.
5935         (vsx_andc<mode>3_32bit): Likewise.
5936         (vsx_andc<mode>3_64bit): Likewise.
5937         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5938         and xxlorc instructions.
5939         (vsx_eqv<mode>3_64bit): Likewise.
5940         (vsx_nand<mode>3_32bit): Likewise.
5941         (vsx_nand<mode>3_64bit): Likewise.
5942         (vsx_orc<mode>3_32bit): Likewise.
5943         (vsx_orc<mode>3_64bit): Likewise.
5945         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5947         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5948         instruction.
5949         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5950         (p8_vmrgow): Likewise.
5951         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5952         GPRs to be split under VSX.
5953         (p8v_clz<mode>2): Add power8 count leading zero support.
5954         (p8v_popcount<mode>2): Add power8 population count support.
5955         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5956         support.
5958         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5959         instruction.
5961         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5962         builtin functions.
5963         (vec_nand): Likewise.
5964         (vec_vclz): Likewise.
5965         (vec_vclzb): Likewise.
5966         (vec_vclzd): Likewise.
5967         (vec_vclzh): Likewise.
5968         (vec_vclzw): Likewise.
5969         (vec_vgbbd): Likewise.
5970         (vec_vmrgew): Likewise.
5971         (vec_vmrgow): Likewise.
5972         (vec_vpopcnt): Likewise.
5973         (vec_vpopcntb): Likewise.
5974         (vec_vpopcntd): Likewise.
5975         (vec_vpopcnth): Likewise.
5976         (vec_vpopcntw): Likewise.
5978 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5980         PR rtl-optimization/57468
5981         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5982         spilled pseudos.
5984 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5986         PR rtl-optimization/57459
5987         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5988         type when setting live regs.
5990 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5992         * config/s390/s390.opt (mlra): New option.
5993         * config/s390/s390.c (s390_decompose_address): Check displacement
5994         for all registers for LRA.
5995         (s390_secondary_reload): Don't used secondary reloads for LRA.
5996         (s390_lra_p): New function.
5997         (TARGET_LRA_P): Define.
5998         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5999         of attribute cpu_facility to zarch for the last alternative.
6000         (*cmpmem_short): Ditto.
6002 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6004         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6005         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6006         (arm_expand_prologue): Likewise.
6008 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6010         PR c++/53743
6011         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6012         as this is now done by redirect_edge_and_branch_force.
6013         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6014         barriers, and fix interaction with splitting.
6015         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6016         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6017         reflect changes made in the routine.
6018         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6019         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6020         since this is called in cfglayout mode, and replace partition fixup
6021         with assert as that is now done by force_nonfallthru_and_redirect.
6022         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6023         already be marked with region crossing note.
6024         (insert_section_boundary_note): Make non-static, gate on flag
6025         has_bb_partition, rewrite to also check for multiple partitions.
6026         (rest_of_handle_reorder_blocks): Remove call to
6027         insert_section_boundary_note, now done later during free_cfg.
6028         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6029         * bb-reorder.h (insert_section_boundary_note): Declare.
6030         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6031         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6032         invoke insert_section_boundary_note.
6033         (try_redirect_by_replacing_jump): Remove unnecessary
6034         check for region crossing note.
6035         (fixup_partition_crossing): New function.
6036         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6037         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6038         in non-cfglayout mode.
6039         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6040         remove old code that tried to do this. Emit barrier correctly
6041         when we are in cfglayout mode.
6042         (last_bb_in_partition): New function.
6043         (rtl_split_edge): Correctly fixup partition boundaries.
6044         (commit_one_edge_insertion): Remove old code that tried to
6045         fixup region crossing edge since this is now handled in
6046         split_block, and set up insertion point correctly since
6047         block may now end in a jump.
6048         (verify_hot_cold_block_grouping): Guard against checking when not in
6049         linearized RTL mode.
6050         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6051         notes.
6052         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6053         rtl_verify_flow_info, so not called in cfglayout mode.
6054         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6055         (fixup_reorder_chain): Remove old code that attempted to fixup region
6056         crossing note as this is now handled in force_nonfallthru_and_redirect.
6057         (duplicate_insn_chain): Don't duplicate switch section notes.
6058         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6059         note.
6060         * basic-block.h (emit_barrier_after_bb): Declare.
6062 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6064         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6065         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6066         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6068 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6070         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6071         where appropriate.
6072         * config/arm/ldmstm.md: Regenerate.
6074 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6076         * config/arm/sync.md (atomic_loaddi_1):
6077         Disable predication for arm_restrict_it.
6078         (arm_load_exclusive<mode>): Likewise.
6079         (arm_load_exclusivesi): Likewise.
6080         (arm_load_exclusivedi): Likewise.
6081         (arm_load_acquire_exclusive<mode>): Likewise.
6082         (arm_load_acquire_exclusivesi): Likewise.
6083         (arm_load_acquire_exclusivedi): Likewise.
6084         (arm_store_exclusive<mode>): Likewise.
6085         (arm_store_exclusive<mode>): Likewise.
6086         (arm_store_release_exclusivedi): Likewise.
6087         (arm_store_release_exclusive<mode>): Likewise.
6089 2013-06-06  Richard Biener  <rguenther@suse.de>
6091         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6092         after LTO_null.
6093         (lto_tag_is_tree_code_p): Adjust.
6094         (lto_tag_is_gimple_code_p): Likewise.
6095         (lto_gimple_code_to_tag): Likewise.
6096         (lto_tag_to_gimple_code): Likewise.
6097         (lto_tree_code_to_tag): Likewise.
6098         (lto_tag_to_tree_code): Likewise.
6099         * data-streamer.h (streamer_write_hwi_in_range): Use
6100         uhwi streaming to stream the normalized range.
6101         (streamer_read_hwi_in_range): Likewise.
6103 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6105         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6106         (predicable_short_it): Likewise.
6107         (predicated): Likewise.
6108         (enabled): Handle above.
6109         (define_cond_exec): Set predicated attribute to yes.
6111 2013-06-05  Mike Stump  <mikestump@comcast.net>
6113         * gdbinit.in (__FUNCTION__): Add.
6115 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6117         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6118         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6120 2013-06-05  Jan Hubicka  <jh@suse.cz>
6122         * varasm.c (mark_decl_referenced): Revert the removal until targets
6123         are fixed.
6125 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6127         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6128         instead of mark_decl_referenced.
6130 2013-06-05  Jan Hubicka  <jh@suse.cz>
6132         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6133         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6134         and symtab_used_from_object_file_p.
6135         (cgraph_make_node_local_1): Clear forced_by_abi.
6136         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6137         * cgraph.h (symtab_node_base): Add forced_by_abi.
6138         (decide_is_variable_needed): Remove.
6139         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6140         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6141         (decide_is_symbol_needed): ... this one; handle symbols in general;
6142         always analyze virtuals; honnor forced_by_abi.
6143         (cgraph_finalize_function): Update.
6144         (varpool_finalize_decl): Update.
6145         (symbol_defined_and_needed): Remove.
6146         (analyze_functions): Update.
6147         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6148         output_refs, input_overwrite_node): Handle forced_by_abi.
6149         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6150         (address_taken_from_non_vtable_p): ... this one.
6151         (comdat_can_be_unshared_p_1): New function.
6152         (cgraph_comdat_can_be_unshared_p): Rename to ...
6153         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6154         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6155         (function_and_variable_visibility): Clear forced_by_abi as needed.
6156         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6157         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6158         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6159         * varpool.c (decide_is_variable_needed): Remove.
6161 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6163         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6164         (arm_option_override): Override arm_restrict_it where appropriate.
6165         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6166         * config/arm/arm.opt (mrestrict-it): New command-line option.
6167         * doc/invoke.texi: Document -mrestrict-it.
6169 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6171         * tsan.c (tsan_atomic_table): Make const.
6173 2013-06-05  Richard Biener  <rguenther@suse.de>
6175         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6176         index associated with the tree we are supposed to replace.
6177         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6178         TREE_ASM_WRITTEN as zero for everything but SSA names.
6180 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6182         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6184 2013-06-04  Jan Hubicka  <jh@suse.cz>
6186         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6187         (input_node, input_varpool_node): Handle correctly external same
6188         body aliases.
6189         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6190         nodes at ltrans stage.
6192 2013-06-04  Jan Hubicka  <jh@suse.cz>
6194         * ipa-inline.c (update_caller_keys): Fix availability test.
6195         (update_callee_keys): Likewise.
6196         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6197         to follow ELF standard.
6199 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6201         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6202         (mips64r5900el-*-elf*): New configurations.
6203         * config/mips/mips-cpus.def (r5900): New processor.
6204         * config/mips/mips-tables.opt: Regenerate.
6205         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6206         (mips_issue_rate): Handle PROCESSOR_R5900.
6207         (mips_reorg_process_insns): Force reorder mode for the R5900.
6208         * config/mips/mips.h (TARGET_MIPS5900): Define.
6209         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6210         TARGET_MIPS5900.
6211         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6212         TARGET_MIPS5900.
6213         * config/mips/mips.md (processor): Add r5900.
6214         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6216 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6218         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6219         into function to generate MOVI instruction.
6220         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6221         (aarch64_preferred_simd_mode): Turn into wrapper.
6222         (aarch64_output_scalar_simd_mov_immediate): New function.
6223         * config/aarch64/aarch64-protos.h: Add prototype for above.
6225 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6227         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6228         element_char member.
6229         (sizetochar): Return signed char.
6230         (aarch64_simd_valid_immediate): Remove elchar and other
6231         unnecessary variables.
6232         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6233         Calculate element_char as required.
6234         * config/aarch64/aarch64-protos.h: Update and move prototype
6235         for aarch64_output_simd_mov_immediate.
6236         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6237         Update arguments.
6239 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6241         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6242         information completed by aarch64_simd_valid_immediate.
6243         (aarch64_legitimate_constant_p): Update arguments.
6244         (aarch64_simd_valid_immediate): Work with struct rather than many
6245         pointers.
6246         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6247         (aarch64_simd_make_constant): Update arguments.
6248         (aarch64_output_simd_mov_immediate): Work with struct rather than
6249         many pointers.  Output immediate directly rather than as operand.
6250         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6251         Update prototype.
6252         * config/aarch64/constraints.md (Dn): Update arguments.
6254 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6256         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6257         longer static.
6258         (aarch64_simd_immediate_valid_for_move): Remove.
6259         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6260         (aarch64_simd_make_constant): Update call.
6261         (aarch64_output_simd_mov_immediate): Update call.
6262         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6263         Add prototype.
6264         * config/aarch64/constraints.md (Dn): Update call.
6266 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6268         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6269         return type to bool for prototype.
6270         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6271         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6272         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6274 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6276         * config/mips/mips.opt (meva): New.
6277         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6278         (ASM_SPEC): Handle -meva.
6279         * doc/invoke.texi (meva):  Document.
6281 2013-06-04  Alan Modra  <amodra@gmail.com>
6283         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6284         constant output.
6286 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6288         * rtl.def: Add extra fourth optional field to define_cond_exec.
6289         * gensupport.c (process_one_cond_exec): Process attributes from
6290         define_cond_exec.
6291         * doc/md.texi: Document fourth field in define_cond_exec.
6293 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6295         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6296         out the processing order as in store_bit_field_1.
6298 2013-06-04  Jan Hubicka  <jh@suse.cz>
6300         PR middle-end/57500
6301         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6302         non-VAR_DECL node if it does not exist yet.
6304 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6306         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6307         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6308         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6309         target_cpu_default setting.
6311 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6313         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6314         default to -fopt-info=optimized instead of all.
6315         * doc/invoke.texi: Ditto.
6316         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6317         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6318         (execute_vect_slp): Emit BB vectorization success under
6319         MSG_OPTIMIZED_LOCATIONS.
6320         * tree-vect-slp.c (vect_slp_transform_bb): Change
6321         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6322         * tree-vect-loop.c (vect_transform_loop): Ditto.
6324 2013-06-03  Jason Merrill  <jason@redhat.com>
6326         PR c++/57415
6327         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6328         Use TARGET_EXPR for C++.
6330 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6332         PR rtl-optimization/57268
6333         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6334         if DEBUG_INSN_P (insn).
6336         Reapply
6337         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6339         PR rtl-optimization/57268
6340         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6341         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6343 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6345         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6346         (ix86_avoid_lea_for_addr): Likewise.
6347         (exact_dependency_1): Likewise.
6348         (ix86_adjust_cost): Likewise.
6349         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6350         removed.
6351         (do_reorder_for_imul): Fix typo, formatting and
6352         !reload_completed check removed.
6353         (ix86_sched_reorder): Fix typo and formatting.
6354         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6355         list.
6357 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6359         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6361 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6363         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6364         <STRING_CST>: Likewise.
6365         <VECTOR_CST>: Likewise.
6367 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6368             Mikael Morin  <mikael@gcc.gnu.org>
6370         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6371         * config.in: Regenerated.
6372         * configure: Regenerated.
6374 2013-06-01  Jan Hubicka  <jh@suse.cz>
6376         PR middle-end/57366
6377         * cgraphunit.c (compile): When weakref is not supported,
6378         set up transparent aliases before final output pass.
6379         * varasm.c (assemble_alias): Do not try to do it here.
6381 2013-06-01  Jan Hubicka  <jh@suse.cz>
6383         PR middle-end/57467
6384         * passes.c (for_per_function): Skip unanalyzed functions.
6386 2013-06-01  Jan Hubicka  <jh@suse.cz>
6388         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6389         (lto_symtab_merge_symbols_1): ... this one.
6390         (lto_symtab_merge_cgraph_nodes): Rename to ...
6391         (lto_symtab_merge_symbols): ... this one; simplify.
6392         * cgraph.c (same_body_aliases_done): Rename to ...
6393         (cpp_implicit_aliases_done): ... this one.
6394         (cgraph_create_function_alias): Update.
6395         (cgraph_same_body_alias): Update.
6396         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6397         (verify_edge_corresponds_to_fndecl): Simplify.
6398         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6399         (cgraph_node): Remove same_body_alias.
6400         (varpool_node): Remove alias_of and extra_name_alias.
6401         (same_body_aliases_done): Rename to ..
6402         (cpp_implicit_aliases_done): ... this one.
6403         (symtab_alias_ultimate_target): Add default parameter.
6404         (symtab_resolve_alias): New function.
6405         (fixup_same_cpp_alias_visibility): Declare.
6406         (cgraph_function_node): Add default parameter.
6407         (cgraph_node_asm_name): Likewise.
6408         (cgraph_function_or_thunk_node): Add default parameter; do
6409         not ICE when it is NULL.
6410         (varpool_variable_node): Likewise.
6411         * tree-emutls.c (create_emultls_var): Update.
6412         (ipa_lower_emutls): Update.
6413         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6414         (cgraph_reset_node): Reset alias info.
6415         (cgraph_finalize_function): Update.
6416         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6417         (analyze_function): Simplify.
6418         (cgraph_process_same_body_aliases): Simplify.
6419         (analyze_functions): Fixup same body aliases.
6420         (handle_alias_pairs): Simplify.
6421         (assemble_thunk): Update.
6422         (assemble_thunks_and_aliases): Update.
6423         (output_weakrefs): Rewrite.
6424         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6425         (lto_output_varpool_node): Likewise.
6426         (compute_ltrans_boundary): Remve assert.
6427         (get_alias_symbol): New functoin.
6428         (input_node): Rewrite alias handling.
6429         (input_varpool_node): Likewise.
6430         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6431         * ipa.c (process_references): Handle weakrefs correctly.
6432         (symtab_remove_unreachable_nodes): Likewise.
6433         * trans-mem.c (get_cg_data): Update.
6434         (ipa_tm_create_version_alias): Update.
6435         (ipa_tm_execute): Update.
6436         * symtab.c (dump_symtab_base): Dump aliases.
6437         (verify_symtab_base): Verify aliases.
6438         (symtab_node_availability): New function.
6439         (symtab_alias_ultimate_target): Simplify.
6440         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6441         handle all the fixup cases.
6442         (symtab_resolve_alias): New function.
6443         * passes.c (ipa_write_summaries): Handle weakrefs.
6444         * varpool.c (varpool_analyze_node): Simplify.
6445         (assemble_aliases): Update.
6446         (varpool_create_variable_alias): Simplify.
6447         (varpool_extra_name_alias): Simplify.
6448         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6449         (lto_symtab_merge_symbols): ... this one.
6451 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6453         Revert
6454         PR rtl-optimization/57268
6455         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6456         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6458 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6460         Partially reverted:
6461         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6463         PR middle-end/57073
6464         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6465         further up.
6467 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6469         PR rtl-optimization/57268
6470         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6471         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6473 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6475         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6476         unordered comparison operators when -fno-trapping-math is in effect
6477         on the e500.
6478         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6479         and implement unordered comparison operators properly on the e500.
6481 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6483         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6484         for constant scalar integers.
6485         (simplify_relational_operation_1): Likewise.
6487 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6489         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6490         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6491         Fix comment.
6493 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6494             Igor Zamyatin  <igor.zamyatin@intel.com>
6496         Silvermont (SLM) architecture performance tuning.
6497         * config/i386/i386.h (enum ix86_tune_indices): Add
6498         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6499         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6501         * config/i386/i386.c (initial_ix86_tune_features)
6502         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6503         (ix86_lea_outperforms): Handle Silvermont tuning.
6504         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6505         call.
6506         (ix86_use_lea_for_mov): Likewise.
6507         (ix86_avoid_lea_for_addr): Likewise.
6508         (ix86_lea_for_add_ok): Likewise.
6509         (exact_dependency_1): New function.
6510         (exact_store_load_dependency): Likewise.
6511         (ix86_adjust_cost): Handle Silvermont tuning.
6512         (do_reoder_for_imul): Likewise.
6513         (swap_top_of_ready_list): New function.
6514         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6516         * config/i386/i386.md (peepholes that split memory operand in fp
6517         converts): New.
6519 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6521         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6522         Remove un-necessary braces.
6524 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6526         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6527         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6529 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6531         PR middle-end/57073
6532         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6534 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6536         PR target/56315
6537         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6538         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6539         * config/arm/neon.md (iordi3_neon): Remove.
6540         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6541         * config/arm/predicates.md (imm_for_neon_logic_operand):
6542         Move to earlier in the file.
6543         (neon_logic_op2): Likewise.
6544         (arm_iordi_operand_neon): New predicate.
6546 2013-05-31  Richard Biener  <rguenther@suse.de>
6548         PR tree-optimization/57478
6549         PR tree-optimization/57453
6550         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6551         are life as well.
6553 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6555         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6556         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6558 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6559             Thomas Koenig  <tkoenig@gcc.gnu.org>
6561         PR middle-end/57073
6562         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6563         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6565 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6567         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6569 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6571         * target.def (register_usage_leveling_p): New hook.
6572         * targhooks.c (default_register_usage_leveling_p): New.
6573         * targhooks.h (default_register_usage_leveling_p): New prototype.
6574         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6575         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6576         * doc/tm.texi: Update.
6577         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6579 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6581         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6582         (*insv_reg<mode>): New define_insn.
6584 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6586         PR rtl-optimization/57439
6587         * postreload.c (move2add_valid_value_p): Check that we have
6588         a zero subreg_regno_offset when accessing the register in
6589         the requested mode.
6591 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6592             Igor Zamyatin  <igor.zamyatin@intel.com>
6594         Silvermont (SLM) architecture pipeline model, tuning and
6595         insn selection.
6596         * config.gcc: Add slm config options and target.
6598         * config/i386/slm.md: New.
6600         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6602         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6603         PROCESSOR_SLM.
6604         (ix86_target_macros_internal): Likewise.
6606         * gcc/config/i386/i386.c (slm_cost): New cost.
6607         (m_SLM): New macro flag.
6608         (initial_ix86_tune_features): Set m_SLM.
6609         (x86_accumulate_outgoing_args): Likewise.
6610         (x86_arch_always_fancy_math_387): Likewise.
6611         (processor_target_table): Add slm cost.
6612         (cpu_names): Add slm cpu name.
6613         (x86_option_override_internal): Set SLM ISA.
6614         (ix86_issue_rate): New case PROCESSOR_SLM.
6615         (ia32_multipass_dfa_lookahead): Likewise.
6616         (fold_builtin_cpu): Add slm.
6618         * config/i386/i386.h (TARGET_SLM): New target macro.
6619         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6620         (processor_type): Add PROCESSOR_SLM.
6622         * config/i386/i386.md (cpu): Add new value "slm".
6623         (slm.md): Include slm.md.
6625 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6626             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6628         * config/arm/arm-protos.h: Add and update function protos.
6629         * config/arm/arm.c (use_simple_return_p): New added.
6630         (thumb2_expand_return): Check simple_return flag.
6631         * config/arm/arm.md: Add simple_return and conditional simple_return.
6632         * config/arm/iterators.md: Add iterator for return and simple_return.
6634 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6636         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6637         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6638         (arm_emit_vfp_multi_reg_pop): Likewise.
6639         (thumb2_emit_ldrd_pop): Likewise.
6640         (arm_expand_epilogue): Add misc REG_CFA notes.
6641         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6643 2013-05-29  Lawrence Crowl  <crowl@google.com>
6645         * config/arm/t-arm: Update for below.
6647         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6648         Change type to hash_table.  Update dependent calls and types.
6650         * config/i386/t-cygming: Update for below.
6652         * config/i386/t-interix: Update for below.
6654         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6655         Change type to hash_table.  Update dependent calls and types.
6656         (i386_find_on_wrapper_list::wrappers): Likewise.
6658         * config/ia64/t-ia64: Update for below.
6660         * config/ia64/ia64.c (bundle_state_table):
6661         Change type to hash_table.  Update dependent calls and types.
6663         * config/mips/mips.c (mips_reorg_process_insns::htab):
6664         Change type to hash_table.  Update dependent calls and types.
6666         * config/sol2.c (solaris_comdat_htab):
6667         Change type to hash_table.  Update dependent calls and types.
6669         * config/t-sol2: Update for above.
6671 2013-05-29  Teresa Johnson  <tejohnson@google.com>
6673         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6674         functions are not yet marked as defined.
6676 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
6677             Pat Haugen <pthaugen@us.ibm.com>
6678             Peter Bergner <bergner@vnet.ibm.com>
6680         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6681         instructions.
6682         (VEC_A): Likewise.
6683         (VEC_C): Likewise.
6684         (vrotl<mode>3): Likewise.
6685         (vashl<mode>3): Likewise.
6686         (vlshr<mode>3): Likewise.
6687         (vashr<mode>3): Likewise.
6689         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6690         support for power8 V2DI builtins.
6692         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6693         power8 V2DI builtins.
6694         (vupkhsw): Likewise.
6695         (vupklsw): Likewise.
6696         (vaddudm): Likewise.
6697         (vminsd): Likewise.
6698         (vmaxsd): Likewise.
6699         (vminud): Likewise.
6700         (vmaxud): Likewise.
6701         (vpkudum): Likewise.
6702         (vpksdss): Likewise.
6703         (vpkudus): Likewise.
6704         (vpksdus): Likewise.
6705         (vrld): Likewise.
6706         (vsld): Likewise.
6707         (vsrd): Likewise.
6708         (vsrad): Likewise.
6709         (vsubudm): Likewise.
6710         (vcmpequd): Likewise.
6711         (vcmpgtsd): Likewise.
6712         (vcmpgtud): Likewise.
6713         (vcmpequd_p): Likewise.
6714         (vcmpgtsd_p): Likewise.
6715         (vcmpgtud_p): Likewise.
6716         (vupkhsw): Likewise.
6717         (vupklsw): Likewise.
6718         (vaddudm): Likewise.
6719         (vmaxsd): Likewise.
6720         (vmaxud): Likewise.
6721         (vminsd): Likewise.
6722         (vminud): Likewise.
6723         (vpksdss): Likewise.
6724         (vpksdus): Likewise.
6725         (vpkudum): Likewise.
6726         (vpkudus): Likewise.
6727         (vrld): Likewise.
6728         (vsld): Likewise.
6729         (vsrad): Likewise.
6730         (vsrd): Likewise.
6731         (vsubudm): Likewise.
6733         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6734         support for power8 V2DI instructions.
6736         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6737         power8 V2DI instructions.  Combine pack and unpack insns to use an
6738         iterator for each mode.  Check whether a particular mode supports
6739         Altivec instructions instead of just checking TARGET_ALTIVEC.
6740         (UNSPEC_VPKUWUM): Likewise.
6741         (UNSPEC_VPKSHSS): Likewise.
6742         (UNSPEC_VPKSWSS): Likewise.
6743         (UNSPEC_VPKUHUS): Likewise.
6744         (UNSPEC_VPKSHUS): Likewise.
6745         (UNSPEC_VPKUWUS): Likewise.
6746         (UNSPEC_VPKSWUS): Likewise.
6747         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6748         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6749         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6750         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6751         (UNSPEC_VUPKHSB): Likewise.
6752         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6753         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6754         (UNSPEC_VUPKHSH): Likewise.
6755         (UNSPEC_VUPKLSB): Likewise.
6756         (UNSPEC_VUPKLSH): Likewise.
6757         (VI2): Likewise.
6758         (VI_char): Likewise.
6759         (VI_scalar): Likewise.
6760         (VI_unit): Likewise.
6761         (VP): Likewise.
6762         (VP_small): Likewise.
6763         (VP_small_lc): Likewise.
6764         (VU_char): Likewise.
6765         (add<mode>3): Likewise.
6766         (altivec_vaddcuw): Likewise.
6767         (altivec_vaddu<VI_char>s): Likewise.
6768         (altivec_vadds<VI_char>s): Likewise.
6769         (sub<mode>3): Likewise.
6770         (altivec_vsubcuw): Likewise.
6771         (altivec_vsubu<VI_char>s): Likewise.
6772         (altivec_vsubs<VI_char>s): Likewise.
6773         (altivec_vavgs<VI_char>): Likewise.
6774         (altivec_vcmpbfp): Likewise.
6775         (altivec_eq<mode>): Likewise.
6776         (altivec_gt<mode>): Likewise.
6777         (altivec_gtu<mode>): Likewise.
6778         (umax<mode>3): Likewise.
6779         (smax<mode>3): Likewise.
6780         (umin<mode>3): Likewise.
6781         (smin<mode>3): Likewise.
6782         (altivec_vpkuhum): Likewise.
6783         (altivec_vpkuwum): Likewise.
6784         (altivec_vpkshss): Likewise.
6785         (altivec_vpkswss): Likewise.
6786         (altivec_vpkuhus): Likewise.
6787         (altivec_vpkshus): Likewise.
6788         (altivec_vpkuwus): Likewise.
6789         (altivec_vpkswus): Likewise.
6790         (altivec_vpks<VI_char>ss): Likewise.
6791         (altivec_vpks<VI_char>us): Likewise.
6792         (altivec_vpku<VI_char>us): Likewise.
6793         (altivec_vpku<VI_char>um): Likewise.
6794         (altivec_vrl<VI_char>): Likewise.
6795         (altivec_vsl<VI_char>): Likewise.
6796         (altivec_vsr<VI_char>): Likewise.
6797         (altivec_vsra<VI_char>): Likewise.
6798         (altivec_vsldoi_<mode>): Likewise.
6799         (altivec_vupkhsb): Likewise.
6800         (altivec_vupkhs<VU_char>): Likewise.
6801         (altivec_vupkls<VU_char>): Likewise.
6802         (altivec_vupkhsh): Likewise.
6803         (altivec_vupklsb): Likewise.
6804         (altivec_vupklsh): Likewise.
6805         (altivec_vcmpequ<VI_char>_p): Likewise.
6806         (altivec_vcmpgts<VI_char>_p): Likewise.
6807         (altivec_vcmpgtu<VI_char>_p): Likewise.
6808         (abs<mode>2): Likewise.
6809         (vec_unpacks_hi_v16qi): Likewise.
6810         (vec_unpacks_hi_v8hi): Likewise.
6811         (vec_unpacks_lo_v16qi): Likewise.
6812         (vec_unpacks_hi_<VP_small_lc>): Likewise.
6813         (vec_unpacks_lo_v8hi): Likewise.
6814         (vec_unpacks_lo_<VP_small_lc>): Likewise.
6815         (vec_pack_trunc_v8h): Likewise.
6816         (vec_pack_trunc_v4si): Likewise.
6817         (vec_pack_trunc_<mode>): Likewise.
6819         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6820         V2DI builtins.
6821         (vec_vmaxsd): Likewise.
6822         (vec_vmaxud): Likewise.
6823         (vec_vminsd): Likewise.
6824         (vec_vminud): Likewise.
6825         (vec_vpksdss): Likewise.
6826         (vec_vpksdus): Likewise.
6827         (vec_vpkudum): Likewise.
6828         (vec_vpkudus): Likewise.
6829         (vec_vrld): Likewise.
6830         (vec_vsld): Likewise.
6831         (vec_vsrad): Likewise.
6832         (vec_vsrd): Likewise.
6833         (vec_vsubudm): Likewise.
6834         (vec_vupkhsw): Likewise.
6835         (vec_vupklsw): Likewise.
6837 2013-05-29  Jan Hubicka  <jh@suse.cz>
6839         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6840         flags; reorder rest of fields in more consistent way.
6841         (varpool_node): Remove analyzed, finalized and alias.
6842         (cgraph_ndoe): Likewise.
6843         (symtab_alias_ultimate_target): New function.
6844         (cgraph_function_node): Move offline.
6845         (cgraph_reset_node): Declare.
6846         (cgraph_comdat_can_be_unshared_p): Remove.
6847         (varpool_remove_initializer): Declare.
6848         (varpool_first_defined_variable, varpool_next_defined_variable
6849         cgraph_first_defined_function, cgraph_next_defined_function): Update.
6850         (cgraph_function_with_gimple_body_p): Update.
6851         (varpool_all_refs_explicit_p): Update.
6852         (symtab_alias_target): New function.
6853         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6854         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
6855         (cgraph_function_or_thunk_node): Simplify using
6856         symtab_alias_ultimate_target.
6857         (varpool_variable_node): Likewise.
6858         * cgraph.c (cgraph_create_function_alias): Update.
6859         (cgraph_add_thunk): Update.
6860         (cgraph_remove_node): Update.
6861         (dump_cgraph_node): Do not dump removed flags.
6862         (cgraph_function_body_availability): Update.
6863         (cgraph_propagate_frequency): Update.
6864         (verify_cgraph_node): Check sanity of local flag.
6865         (cgraph_function_node): Move here from cgraph.h; revamp for
6866         cgraph_function_or_thunk_node.
6867         * lto-symtab.c (lto_varpool_replace_node): Update.
6868         (lto_symtab_resolve_can_prevail_p): Update.
6869         (lto_symtab_merge_cgraph_nodes): Update.
6870         * ipa-cp.c (determine_versionability, initialize_node_lattices,
6871         propagate_constants_accross_call, devirtualization_time_bonus,
6872         ipcp_propagate_stage): Update.
6873         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
6874         * ipa-inline-transform.c (clone_inlined_nodes,
6875         preserve_function_body_p): Update.
6876         * ipa-reference.c (propagate): Update.
6877         (write_node_summary_p): Update.
6878         * toplev.c (wrapup_global_declaration_2): Update.
6879         * cgraphunit.c (cgraph_analyze_function): Rename to ...
6880         (analyze_function) ... this one.
6881         (cgraph_process_new_functions): Update.
6882         (cgraph_reset_node): Export.
6883         (cgraph_finalize_function): Update.
6884         (cgraph_add_new_function): Update.
6885         (process_function_and_variable_attributes): Update.
6886         (varpool_finalize_decl): Update.
6887         (symbol_finalized): Remove.
6888         (symbol_finalized_and_needed): Rename to ...
6889         (symbol_defined_and_needed): ... update.
6890         (cgraph_analyze_functions): Update.
6891         (handle_alias_pairs): Update.
6892         (mark_functions_to_output): Update.
6893         (assemble_thunk): Update.
6894         (output_in_order): Update.
6895         (output_weakrefs): Update.
6896         (finalize_compilation_unit): Update.
6897         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6898         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6899         input_node, input_varpool_node): Update.
6900         * dbxout.c (dbxout_expand_expr): Update.
6901         * cgraphclones.c (cgraph_clone_node): Update.
6902         (cgraph_copy_node_for_versioning): Update.
6903         (cgraph_materialize_clone): Update.
6904         (cgraph_materialize_all_clones): Update.
6905         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6906         propagate_pure_const, propagate_nothrow): Update.
6907         * lto-streamer-out.c (lto_output, write_symbol): Update.
6908         * ipa-utils.c (ipa_reverse_postorder): Update.
6909         * ipa-inline.c (can_inline_edge_p): Update.
6910         (update_caller_keys, ipa_inline): Update.
6911         * dwarf2out.c (reference_to_unused,
6912         premark_types_used_by_global_vars_helper): Update.
6913         * tree-eh.c (tree_could_trap_p): Update.
6914         * ipa-split.c (consider_split, execute_split_functions): Update.
6915         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
6916         has_addr_references_p): Update; move ahead in file for better
6917         readability.
6918         (process_references): Simplify.
6919         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6920         bodies are removed.
6921         (cgraph_comdat_can_be_unshared_p): Make static.
6922         (cgraph_externally_visible_p): Update.
6923         (varpool_externally_visible_p): Update.
6924         (function_and_variable_visibility): Update.
6925         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6926         ipa_tm_mark_force_output_node): Update.
6927         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6928         estimate_edge_devirt_benefit, inline_generate_summary,
6929         inline_write_summary): Update.
6930         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6931         * ipa-prop.c (ipa_compute_jump_functions): Update.
6932         (ipa_print_node_params, ipa_prop_read_section,
6933         ipa_update_after_lto_read, read_replacements_section): Update.
6934         * varasm.c (mark_decl_referenced): Update.
6935         (assemble_alias, dump_tm_clone_pairs): Update.
6936         * tree-inline.c (copy_bb): Update.
6937         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6938         Update.
6939         * symtab.c (dump_symtab_base): Print new flags.
6940         (verify_symtab_base): Verify new flags.
6941         (symtab_alias_ultimate_target): New function.
6942         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6943         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6944         Update.
6945         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6946         Update.
6947         * i386.c (ix86_get_function_versions_dispatcher,
6948         ix86_generate_version_dispatcher_body): Update.
6949         (fold_builtin_cpu): Use varpool_add_new_variable.
6950         * varpool.c (varpool_remove_initializer): Break out from ...
6951         (varpool_remove_node): ... this one.
6952         (dump_varpool_node, varpool_node_for_asm,
6953         cgraph_variable_initializer_availability, varpool_analyze_node,
6954         varpool_assemble_decl, varpool_remove_unreferenced_decls,
6955         varpool_finalize_named_section_flags, varpool_create_variable_alias):
6956         Update.
6958 2013-05-29  Jan Hubicka  <jh@suse.cz>
6960         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6962 2013-05-29  Easwaran Raman  <eraman@google.com>
6964         PR tree-optimization/57442
6965         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6966         when control exits the main loop.
6968 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
6970         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6971         and RX600.
6972         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
6973         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6974         * rx/t-rx: Add rx100 under multi library matches option for nofpu
6975         option.
6977 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6979         PR tree-optimization/57441
6980         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6981         Don't limit size of incr_vec to number of candidates.
6983 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
6985         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6986         and mips16 directories.
6987         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
6988         (MULTILIB_DIRNAMES): Ditto.
6989         (MULTILIB_EXCEPTIONS): Add new exceptions.
6990         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6991         (MULTILIB_DIRNAMES): Ditto.
6992         (MULTILIB_EXCEPTIONS): Add new exceptions.
6994 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6995             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6997         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6998         SYMBOL_TINY_ABSOLUTE.
6999         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7000         SYMBOL_TINY_ABSOLUTE.
7001         (aarch64_expand_mov_immediate): Likewise.
7002         (aarch64_classify_symbol): Likewise.
7003         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7004         Permit SYMBOL_TINY_ABSOLUTE.
7005         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7007 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7008             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7010         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7011         Refactor if/switch.  Replace gcc_assert with if.
7013 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7015         * config/i386/i386.c (initial_ix86_tune_features): Enable
7016         FP Reassociation for AMD bdver1 and bdver2.
7018 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7020         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7021         and IMAGPART_EXPR do not occur within other handled_components.
7023 2013-05-29  Richard Biener  <rguenther@suse.de>
7025         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7026         access on whether the use is in the BB we currently try to
7027         vectorize.
7028         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7029         vectorize to vect_bb_slp_scalar_cost.
7031 2013-05-29  Richard Biener  <rguenther@suse.de>
7033         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7034         computing scalar cost offsetted by stmts that are kept live
7035         by scalar uses.
7036         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7037         for computation of scalar cost.
7039 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7041         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7043 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7045         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7046         array notation built-in reduction functions.
7047         * doc/passes.texi (Passes): Added documentation about changes done
7048         for Cilk Plus.
7049         * doc/invoke.texi (C Dialect Options): Added documentation about
7050         the -fcilkplus flag.
7051         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7052         (BUILTINS_DEF): Depend on cilkplus.def.
7053         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7054         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7055         * cilkplus.def: New file.
7057 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7059         PR rtl-optimization/57439
7060         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7062 2013-05-28  Easwaran Raman  <eraman@google.com>
7064         PR tree-optimization/57337
7065         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7066         (find_insert_point): Correctly identify the insertion point
7067         when two statements with the same UID is compared.
7069 2013-05-28  Richard Biener  <rguenther@suse.de>
7071         PR tree-optimization/56787
7072         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7073         from the list of data references.
7074         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7075         clobbers.
7076         (vect_analyze_loop_operations): Likewise.
7077         (vect_transform_loop): Remove clobbers.
7079 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7081         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7082         and REALPART_EXPRs have scalar type.
7084 2013-05-28  Richard Biener  <rguenther@suse.de>
7086         PR tree-optimization/57411
7087         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7088         virtual operands.
7089         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7090         virtual operand propagation.
7092 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7094         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7095         destination register for bmasksi_vis.
7096         (vector_init_bshuffle): Likewise.
7097         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7099 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7101         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7102         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7103         mode if the instruction isn't available in the original mode.
7104         * config/sparc/sparc.opt (mfix-ut699): New option.
7105         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7106         (divdf3): Turn into expander.
7107         (divdf3_nofix): New insn.
7108         (divdf3_fix): Likewise.
7109         (divsf3): Disable if -mfix-ut699.
7110         (sqrtdf2): Turn into expander.
7111         (sqrtdf2_nofix): New insn.
7112         (sqrtdf2_fix): Likewise.
7113         (sqrtsf2): Disable if -mfix-ut699.
7115 2013-05-27  Richard Biener  <rguenther@suse.de>
7117         PR middle-end/57412
7118         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7119         block for the new loop.
7121 2013-05-27  Richard Biener  <rguenther@suse.de>
7123         PR tree-optimization/57343
7124         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7125         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7126         (number_of_iterations_cond): Do not build the folded tree.
7128 2013-05-27  Richard Biener  <rguenther@suse.de>
7130         Revert
7131         PR middle-end/57381
7132         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7133         OEP_CONSTANT_ADDRESS_OF retained.
7135         PR tree-optimization/57417
7136         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7137         for unchanged base.
7138         (set_ssa_val_to): Compare addresses using
7139         get_addr_base_and_unit_offset.
7141 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7143         PR rtl-optimization/56833
7144         * postreload.c (move2add_record_mode): New function.
7145         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7146         (move2add_use_add2_insn): Use move2add_record_sym_value.
7147         (move2add_use_add3_insn): Likewise.
7148         (reload_cse_move2add): Use move2add_valid_value_p and
7149         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7150         affected regs by setting reg_mode to VOIDmode.
7151         (move2add_note_store): Don't pretend the inside of a SUBREG is
7152         the actual destination.  Invalidate single/leading registers by
7153         setting reg_mode to VOIDmode.
7154         Use move2add_record_sym_value, move2add_valid_value_p and
7155         move2add_record_mode.
7157 2013-05-27  Richard Biener  <rguenther@suse.de>
7159         PR tree-optimization/57396
7160         * tree-affine.c (double_int_constant_multiple_p): Properly
7161         return false for val == 0 and div != 0.
7163 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7165         * config/mips/mips.h: Use #elif in preprocessor conditions.
7167 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7169         PR target/53916
7170         * config/mips/constraints.md (kl): New constraint.
7171         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7172         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7173         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7174         from LO for MIPS16.
7175         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7177 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7179         PR target/55777
7180         * config/mips/mips.c (mips_can_inline_p): New function.
7181         (TARGET_CAN_INLINE_P): Define.
7183 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7185         * sched-int.h (ds_t, dw_t): Make unsigned int.
7186         Fix documentation that describes how all the ds_t bits are used.
7187         Reserve the last bit for delayed-branch scheduling.
7188         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7189         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7190         (gen_dep_weak_1): Remove prototype.
7191         * sched-deps.c (get_dep_weak_1): Make static.
7192         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7193         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7194         * doc/tm.texi: Regenerate.
7195         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7197 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7199         PR debug/56950
7200         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7202 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7203             Sandra Loosemore  <sandra@codesourcery.com>
7205         * config.gcc (powerpc-*): Allow native for with-cpu.
7207 2013-05-24  Jeff Law  <law@redhat.com>
7209         PR tree-optimization/57124
7210         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7211         conversion feeding a condition if the range has an overflow
7212         if -fstrict-overflow.  Add warnings for when we do make the
7213         transformation.
7215 2013-05-24  Dehao Chen  <dehao@google.com>
7217         * tree-cfg.c (locus_discrim_map): Fix the typo.
7218         (locus_discrim_hasher): Likewise.
7219         (locus_discrim_hasher::hash): Likewise.
7220         (locus_discrim_hasher::equal): Likewise.
7222 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7224         PR tree-optimization/57294
7225         * cgraph.h (ipa_record_stmt_references): Declare.
7226         * cgraphbuild.c (ipa_record_stmt_references): New function.
7227         (build_cgraph_edges): Use ipa_record_stmt_references.
7228         (rebuild_cgraph_edges): Likewise.
7229         (cgraph_rebuild_references): Likewise.
7230         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7231         associated with the old statement and build references from the
7232         newly built statements.
7233         * ipa-ref.c (ipa_remove_stmt_references): New function.
7234         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7236 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7238         * lra-constraints.c (emit_spill_move): Use smaller mode for
7239         mem-mem moves.
7240         (check_and_process_move): Consider mem-reg moves for secondary
7241         too.
7242         (curr_insn_transform): Don't lose insns emitted before for
7243         secondary memory moves.
7244         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7245         reg set up in the current insn.
7247 2013-05-24  Dehao Chen  <dehao@google.com>
7249         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7250         hash function.
7251         (locus_descrim_hasher::equal): Likewise.
7252         (build_gimple_cfg): New discriminator assignment algorithm.
7253         (make_edges): Likewise.
7254         (next_discriminator_for_locus): Likewise.
7255         (same_line_p): Likewise.
7256         (assign_discriminators): Likewise.
7257         (make_cond_expr_edges): Likewise.
7258         (make_gimple_switch_edges): Likewise.
7259         (make_goto_expr_edges): Likewise.
7260         (make_gimple_asm_edges): Likewise.
7262 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7264         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7265         X format specifier to only display bottom 16 bits.
7266         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7267         immediate to match for operand 2, since it will be masked.
7269 2013-05-24  Richard Biener  <rguenther@suse.de>
7271         PR tree-optimization/57287
7272         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7273         all SSA names that occur in abnormal PHIs.
7275 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7277         PR tree-ssa/57385
7278         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7279         that index is not negative.
7281 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7283         PR rtl-optimization/55177
7284         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7285         (simplify_byte_swapping_operation): New.
7286         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7287         (simplify_relational_operation_1): Deal with BSWAP.
7289 2013-05-23  Richard Henderson  <rth@redhat.com>
7291         PR target/56742
7292         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7293         (ix86_reorg): Call it.
7295 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7297         PR target/57379
7298         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7299         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7300         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7302 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7304         PR debug/57351
7305         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7307 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7308             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7310         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7311         * config/aarch64/constraints.md (Usa): Remove.
7312         * doc/md.texi (AArch64 Usa): Remove.
7314 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7315             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7317         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7318         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7319         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7320         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7322 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7324         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7325         instruction (AdvSIMD).
7326         * config/aarch64/aarch64-builtins.c
7327         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7328         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7330 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7332         PR middle-end/57347
7333         * tree.h (contains_bitfld_component_ref_p): Declare.
7334         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7335         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7336         caller.
7337         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7338         not access a bit-field.  Assert all final offsets are byte-aligned.
7340 2013-05-23  Richard Biener  <rguenther@suse.de>
7342         PR tree-optimization/57380
7343         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7344         least one invariant or re-used load.
7345         * passes.c (init_optimization_passes): Move pass_phiprop before
7346         pass_forwprop.
7348 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7350         * config/aarch64/aarch64-simd.md
7351         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7353 2013-05-23  Richard Biener  <rguenther@suse.de>
7355         PR middle-end/57381
7356         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7357         OEP_CONSTANT_ADDRESS_OF retained.
7359 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7361         PR middle-end/57344
7362         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7363         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7365 2013-05-23  Richard Biener  <rguenther@suse.de>
7367         PR rtl-optimization/57341
7368         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7369         instead of true_dependence.
7371 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7373         * bb-reorder.c (branch_threshold): Make const.
7374         (exec_threshold): Ditto.
7376 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7377             Pat Haugen <pthaugen@us.ibm.com>
7378             Peter Bergner <bergner@vnet.ibm.com>
7380         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7381         documentation for the power8 crypto builtins.
7383         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7385         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7386         macros for defining power8 builtin functions.
7387         (BU_P8V_AV_2): Likewise.
7388         (BU_P8V_AV_P): Likewise.
7389         (BU_P8V_VSX_1): Likewise.
7390         (BU_P8V_OVERLOAD_1): Likewise.
7391         (BU_P8V_OVERLOAD_2): Likewise.
7392         (BU_CRYPTO_1): Likewise.
7393         (BU_CRYPTO_2): Likewise.
7394         (BU_CRYPTO_3): Likewise.
7395         (BU_CRYPTO_OVERLOAD_1): Likewise.
7396         (BU_CRYPTO_OVERLOAD_2): Likewise.
7397         (XSCVSPDP): Fix typo, point to the correct instruction.
7398         (VCIPHER): Add power8 crypto builtins.
7399         (VCIPHERLAST): Likewise.
7400         (VNCIPHER): Likewise.
7401         (VNCIPHERLAST): Likewise.
7402         (VPMSUMB): Likewise.
7403         (VPMSUMH): Likewise.
7404         (VPMSUMW): Likewise.
7405         (VPERMXOR_V2DI): Likewise.
7406         (VPERMXOR_V4SI: Likewise.
7407         (VPERMXOR_V8HI: Likewise.
7408         (VPERMXOR_V16QI: Likewise.
7409         (VSHASIGMAW): Likewise.
7410         (VSHASIGMAD): Likewise.
7411         (VPMSUM): Likewise.
7412         (VPERMXOR): Likewise.
7413         (VSHASIGMA): Likewise.
7415         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7416         __CRYPTO__ if the crypto instructions are available.
7417         (altivec_overloaded_builtins): Add support for overloaded power8
7418         builtins.
7420         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7421         support for power8 crypto builtins.
7422         (builtin_function_type): Likewise.
7423         (altivec_init_builtins): Add support for builtins that take vector
7424         long long (V2DI) arguments.
7426         * config/rs6000/crypto.md: New file, define power8 crypto
7427         instructions.
7429 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7430             Pat Haugen <pthaugen@us.ibm.com>
7431             Peter Bergner <bergner@vnet.ibm.com>
7433         * doc/invoke.texi (Option Summary): Add power8 options.
7434         (RS/6000 and PowerPC Options): Likewise.
7436         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7437         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7438         wm, wn, wr documentation.
7440         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7441         registers if direct move instructions are enabled.
7442         (wn): New constraint for no registers.
7443         (wq): New constraint for quad word even GPR registers.
7444         (wr): New constraint if 64-bit instructions are enabled.
7445         (wv): New constraint if power8 vector instructions are enabled.
7446         (wQ): New constraint for quad word memory locations.
7448         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7449         constraint for 0..15 for crypto instructions.
7450         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7451         as GPR and floating point registers.
7452         (int_reg_operand): New predicate to match only GPR registers.
7453         (base_reg_operand): New predicate to match base registers.
7454         (quad_int_reg_operand): New predicate to match even GPR registers
7455         for quad memory operations.
7456         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7457         operations in both GPR and VSX registers.
7458         (quad_memory_operand): New predicate for quad memory operations.
7459         (reg_or_indexed_operand): New predicate for direct move support.
7461         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7462         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7463         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7464         (POWERPC_MASKS): Add power8 options.
7465         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7466         various options.
7468         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7469         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7471         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7472         (-mpower8-fusion): New power8 options.
7473         (-mpower8-fusion-sign): Likewise.
7474         (-mpower8-vector): Likewise.
7475         (-mcrypto): Likewise.
7476         (-mdirect-move): Likewise.
7477         (-mquad-memory): Likewise.
7479         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7480         power8.
7481         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7482         registers.
7483         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7484         (rs6000_debug_vector_unit): Add p8_vector.
7485         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7486         definitions.  Also print fusion state.
7487         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7488         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7489         (rs6000_option_override_internal): Add support for power8.
7490         (rs6000_common_init_builtins): Add debugging for skipped builtins
7491         if -mdebug=builtin.
7492         (rs6000_adjust_cost): Add power8 support.
7493         (rs6000_issue_rate): Likewise.
7494         (insn_must_be_first_in_group): Likewise.
7495         (insn_must_be_last_in_group): Likewise.
7496         (force_new_group): Likewise.
7497         (rs6000_register_move_cost): Likewise.
7498         (rs6000_opt_masks): Likewise.
7500         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7501         power8 capable assembler, default to power7 options.
7502         (TARGET_DIRECT_MOVE): Likewise.
7503         (TARGET_CRYPTO): Likewise.
7504         (TARGET_P8_VECTOR): Likewise.
7505         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7506         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7507         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7508         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7509         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7510         (TARGET_XSCVDPSPN): Likewise.
7511         (TARGET_XSCVSPDPN): Likewsie.
7512         (TARGET_SYNC_HI_QI): Likewise.
7513         (TARGET_SYNC_TI): Likewise.
7514         (MASK_CRYPTO): Likewise.
7515         (MASK_DIRECT_MOVE): Likewise.
7516         (MASK_P8_FUSION): Likewise.
7517         (MASK_P8_VECTOR): Likewise.
7518         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7519         temporary used by some of the direct move instructions to get two FP
7520         temporary registers does not force creation of a stack frame.
7521         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7522         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7523         that any VSX registers are tieable, even if they are also an
7524         Altivec vector mode.
7525         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7526         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7527         (RS6000_BTM_CRYPTO): Likewise.
7528         (RS6000_BTM_COMMON): Likewise.
7530         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7531         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7532         (enum rs6000_vector): Add power8 vector support.
7534 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7536         PR target/19599
7537         PR target/57340
7538         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7539         (any_sibcall_could_use_r3): this and handle indirect calls.
7540         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7542 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7544         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7546 2013-05-22  Richard Biener  <rguenther@suse.de>
7548         PR middle-end/57349
7549         * profile.c (branch_prob): Do not split blocks that are
7550         abnormally receiving from ECF_RETURNS_TWICE functions.
7552 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7554         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7555         address mode.  Move pointer mode initialization to the same place.
7557 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7559         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7560         while it has any effect.
7562 2013-05-21  Easwaran Raman  <eraman@google.com>
7564         PR tree-optimization/57322
7565         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7566         UID of the statement added to the BB to be 1.
7568 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7570         PR tree-optimization/57331
7571         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7572         of conversion from pointer type to integral type with integer.
7574 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7576         PR lto/57289
7577         * ipa-prop.c (ipa_read_node_info): Process param_used and
7578         controlled_uses in the same order as when writing.
7580 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7582         PR plugins/56754
7583         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7585 2013-05-21  Richard Biener  <rguenther@suse.de>
7587         PR tree-optimization/57318
7588         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7589         estimate stmts with side-effects as likely eliminated.
7591 2013-05-21  Richard Biener  <rguenther@suse.de>
7593         PR tree-optimization/57330
7594         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7595         preserve the call stmts fntype.
7597 2013-05-21  Richard Biener  <rguenther@suse.de>
7599         PR tree-optimization/57303
7600         * tree-ssa-sink.c (statement_sink_location): Improve killing
7601         stmt detection and properly handle self-assignments.
7603 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7605         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7606         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7607         registers. Set register size out of the PARALLEL loop.
7609 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7611         PR target/56547
7612         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7613         (*fmasf4, *fmasf4_media): New insns.
7615 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7617         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7618         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7619         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7620         (mips_idiv_insns): Update the comments to say that the returned
7621         instruction counts are in units of BASE_INSN_LENGTH.
7622         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7623         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7624         using 2 rather than 4 as the length of indirect MIPS16 and
7625         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7626         length of a NOP.  Don't divide MIPS16 lengths by 2.
7627         (mips16_split_long_branches): Assume a branch is long if the
7628         length is greater than 4 rather than 8.
7629         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7630         rather than multiplying them by 2.  Multiply instruction counts
7631         by BASE_INSN_LENGTH rather than 4.
7632         (*jump_mips16, tls_get_tp_mips16_<mode>)
7633         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7635 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7637         * config/mips/mips.md (extended_mips16): Remove branch case.
7638         (length): Remove duplicated extended_mips16 test.
7640 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7642         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7644 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7646         * recog.h (Recog_data): Rename to...
7647         (recog_data_d): ...this.
7648         (recog_data): Update accordingly.
7649         * recog.c (recog_data): Likewise.
7650         * reload.c (save_recog_data): Likewise.
7651         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7652         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7654 2013-05-17  Julian Brown  <julian@codesourcery.com>
7656         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7657         found in a REG_EQUAL note, invalidate it.
7659 2013-05-17  Easwaran Raman  <eraman@google.com>
7661         * tree-ssa-reassoc.c (find_insert_point): New function.
7662         (insert_stmt_after): Likewise.
7663         (get_def_stmt): Likewise.
7664         (ensure_ops_are_available): Likewise.
7665         (not_dominated_by): Likewise.
7666         (rewrite_expr_tree): Do not move statements beyond what is
7667         necessary. Remove call to swap_ops_for_binary_stmt...
7668         (reassociate_bb): ... and move it here.
7669         (build_and_add_sum): Assign UIDs for new statements.
7670         (linearize_expr): Likewise.
7671         (do_reassoc): Renumber gimple statement UIDs.
7673 2013-05-17  Jan Hubicka  <jh@suse.cz>
7675         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7676         weakrefs.
7677         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7678         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7679         weakrefs.
7680         (output_weakrefs): Update.
7682 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
7683             Martin Jambor  <mjambor@suse.cz>
7685         PR middle-end/57276
7686         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7687         value that corresponds to the given aggval is found in values vector.
7689 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
7691         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7692         sse, sse2, sse3, ssse3 and sse4a flags to options.
7694 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
7696         * gengtype-state.c: (s_expr_writer): New class, to handle
7697         prettifying of output layout of s-expressions.
7698         (state_writer): New class, to write out gtype.state.
7699         (state_written_type_count): Move this variable into member data of
7700         state_writer.
7701         (s_expr_writer::s_expr_writer): New code: constructor for new class
7702         (state_writer::state_writer(): ditto
7703         (s_expr_writer::write_new_line): New function
7704         (s_expr_writer::write_any_indent): ditto
7705         (s_expr_writer::begin_s_expr): ditto
7706         (s_expr_writer::end_s_expr): ditto
7707         (write_state_fileloc): convert to method of state_writer...
7708         (state_writer:: write_state_fileloc): ...and use methods of
7709         s_expr_writer to write indentation into the gtype.state output file
7710         to visually represent the hierarchical structure of the list
7711         structures
7712         (write_state_fields): ditto, renaming to...
7713         (state_writer::write_state_fields)
7714         (write_state_a_string): ditto, renaming to...
7715         (state_writer::write_state_a_string)
7716         (write_state_string_option): ditto, renaming to...
7717         (state_writer::write_state_string_option)
7718         (write_state_type_option): ditto, renaming to...
7719         (state_writer::write_state_type_option)
7720         (write_state_nested_option): ditto, renaming to...
7721         (state_writer::write_state_nested_option)
7722         (write_state_option): ditto, renaming to...
7723         (state_writer::write_state_option)
7724         (write_state_options): ditto, renaming to...
7725         (state_writer::write_state_options)
7726         (write_state_lang_bitmap): ditto, renaming to...
7727         (state_writer::write_state_lang_bitmap)
7728         (write_state_version): ditto, renaming to...
7729         (state_writer::write_state_version)
7730         (write_state_scalar_type): ditto, renaming to...
7731         (state_writer::write_state_scalar_type)
7732         (write_state_string_type): ditto, renaming to...
7733         (state_writer::write_state_string_type)
7734         (write_state_undefined_type): ditto, renaming to...
7735         (state_writer::write_state_undefined_type)
7736         (write_state_struct_union_type): ditto, renaming to...
7737         (state_writer::write_state_struct_union_type)
7738         (write_state_struct_type): ditto, renaming to...
7739         (state_writer::write_state_struct_type)
7740         (write_state_user_struct_type): ditto, renaming to...
7741         (state_writer::write_state_user_struct_type)
7742         (write_state_lang_struct_type): ditto, renaming to...
7743         (state_writer::write_state_lang_struct_type)
7744         (write_state_param_struct_type): ditto, renaming to...
7745         (state_writer::write_state_param_struct_type)
7746         (write_state_pointer_type): ditto, renaming to...
7747         (state_writer::write_state_pointer_type)
7748         (write_state_array_type): ditto, renaming to...
7749         (state_writer::write_state_array_type)
7750         (write_state_gc_used): ditto, renaming to...
7751         (state_writer::write_state_gc_used)
7752         (write_state_common_type_content): ditto, renaming to...
7753         (state_writer::write_state_common_type_content)
7754         (write_state_type): ditto, renaming to...
7755         (state_writer::write_state_type)
7756         (write_state_pair_list): ditto, renaming to...
7757         (state_writer::write_state_pair_list)
7758         (write_state_pair): ditto, renaming to...
7759         (state_writer::write_state_pair)
7760         (write_state_typedefs): ditto, renaming to...
7761         (state_writer::write_state_typedefs)
7762         (write_state_structures): ditto, renaming to...
7763         (state_writer::write_state_structures)
7764         (write_state_param_structs): ditto, renaming to...
7765         (state_writer::write_state_param_structs)
7766         (write_state_variables): ditto, renaming to...
7767         (state_writer::write_state_variables)
7768         (write_state_srcdir): ditto, renaming to...
7769         (state_writer::write_state_srcdir)
7770         (write_state_files_list): ditto, renaming to...
7771         (state_writer::write_state_files_list)
7772         (write_state_languages): ditto, renaming to...
7773         (state_writer::write_state_languages)
7774         (write_state): create a state_writer instance and use it when
7775         writing out the state file
7777 2013-05-17  Mike Stump  <mikestump@comcast.net>
7779         PR rtl-optimization/57304
7780         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7781         accessing DF_REF_REAL_LOC.
7783 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7785         PR rtl-optimization/57281
7786         PR rtl-optimization/57300
7787         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7788         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7789         what the other splitter did if the registers are dead.
7791 2013-05-17  Richard Biener  <rguenther@suse.de>
7793         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7794         MEM_REF offsets.
7796 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7798         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7799         linking.
7801 2013-05-17  Marek Polacek  <polacek@redhat.com>
7803         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7804         length when doing non-zero store of storing '\0' to '\0'.
7806 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7808         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7809         vect_external_def oprnd1 with loop_vinfo, try to emit
7810         optional cast, negation and and stmts on the loop preheader
7811         edge instead of into the pattern def seq.
7813         PR tree-optimization/57051
7814         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7815         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7817 2013-05-16  Nick Clifton  <nickc@redhat.com>
7819         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7820         (rl78_is_naked_func): New function.
7821         (rl78_expand_prologue): Skip prologue generation for naked functions.
7822         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
7823         * doc/extend.texi (naked): Add RL78 to the list of processors
7824         that supports this attribute.
7826 2013-05-16  Jeff Law  <law@redhat.com>
7828         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7830 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7832         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7833         cache parameters using detect_caches_amd also for CYRIX,
7834         NSC and TM2 signatures.
7836 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7837             Dzianis Kahanovich  <mahatma@eu.by>
7839         PR target/45359
7840         PR target/46396
7841         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7842         VIA/Centaur processors and determine their cache parameters
7843         using detect_caches_amd.
7845 2013-05-16  Teresa Johnson  <tejohnson@google.com>
7847         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7848         (rtl_verify_edges): New function.
7849         (rtl_verify_bb_insns): Ditto.
7850         (rtl_verify_bb_pointers): Ditto.
7851         (rtl_verify_bb_insn_chain): Ditto.
7852         (rtl_verify_fallthru): Ditto.
7853         (rtl_verify_bb_layout): Ditto.
7854         (rtl_verify_flow_info_1): Outline checks into new functions.
7855         (rtl_verify_flow_info): Ditto.
7857 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
7859         * cfghooks.c (copy_bbs): Add update_dominance argument.
7860         * cfghooks.h (copy_bbs): Update prototype.
7861         * tree-cfg.c (gimple_duplicate_sese_region):
7862         Add update_dominance argument.
7863         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7864         * tree-ssa-loop-ch.c (copy_loop_headers): Update
7865         gimple_duplicate_sese_region call.
7866         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7867         Update copy_bbs call.
7868         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7869         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7871 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7873         * tree-vectorizer.h (NUM_PATTERNS): Increment.
7874         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7875         vect_recog_rotate_pattern.
7876         (vect_recog_rotate_pattern): New function.
7878 2013-05-16  Jason Merrill  <jason@redhat.com>
7880         * Makefile.in (LLINKER): New variable.
7881         (mostlyclean): Remove link mutex.
7882         * configure.ac: Handle --enable-link-mutex.
7883         * lock-and-run.sh: New script.
7885 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7887         PR target/19599
7888         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7889         for NULL decl.
7891 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7893         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7895 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
7897         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7898         * config/arm/arm.c (next_consecutive_mem): New function.
7899         (gen_movmem_ldrd_strd): Likewise.
7900         * config/arm/arm.md (movmemqi): Update condition and code.
7901         (unaligned_loaddi, unaligned_storedi): New patterns.
7903 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7905         * config.gcc: Obsolete *-*-solaris2.9*.
7906         * doc/install.texi (Specific, *-*-solaris2*): Document it.
7908 2013-05-16  Richard Biener  <rguenther@suse.de>
7910         * passes.c (init_optimization_passes): Move pass_parallelize_loops
7911         earlier, after GRAPHITE transforms and IV canonicalization.
7913 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7915         * omp-low.c (extract_omp_for_data): For collapsed loops,
7916         if at least one of the loops is known at compile time to
7917         iterate zero times, set count to 0.
7918         (expand_omp_regimplify_p): New function.
7919         (expand_omp_for_generic): For collapsed loops, if at least
7920         one of the loops isn't known to iterate at least once,
7921         add runtime check with setting count to 0.
7922         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7923         For unsigned types if it isn't known at compile time that
7924         the loop will iterate at least once, add runtime check to bypass
7925         the whole loop if initial condition isn't true.
7927 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
7929         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7931 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
7933         PR middle-end/57286
7934         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7935         transformations to avoid an infinite loop.
7937 2013-05-16  Marek Polacek  <polacek@redhat.com>
7939         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7941 2013-05-15  Leif Ekblad  <leif@rdos.net>
7943         * config/i386/i386.c (ix86_decompose_address): Use
7944         DEFAULT_TLS_SEG_REG to access TLS segment register.
7945         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7946         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7947         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7949 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
7951         PR target/57260
7952         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7953         sibling calls to functions that would normally be lazily bound,
7954         unless $gp is call-clobbered.
7956 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
7958         * config/i386/i386.c (ix86_option_override_internal): Update
7959         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
7960         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7961         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
7962         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7963         of TARGET_3DNOW.
7964         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7966 2013-05-15  Andreas Schwab  <schwab@suse.de>
7968         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7969         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
7970         third operand.
7972 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7974         * loop-unroll.c (report_unroll_peel): Check decision before
7975         emitting unroll/peel message.
7977 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7979         * function.h (has_bb_partition): New rtl_data flag.
7980         (bb_reorder_complete): Ditto.
7981         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7982         instead of flag_reorder_blocks_and_partition.
7983         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7984         with some enhancements.
7985         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7986         * bb-reorder.c (connect_traces): Check for has_bb_partition
7987         instead of flag_reorder_blocks_and_partition.
7988         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7989         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7990         verify_hot_cold_block_grouping.
7991         (partition_hot_cold_basic_blocks): Set has_bb_partition.
7993 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7995         PR target/19599
7996         * config/arm/predicates.md (call_insn_operand): New predicate.
7997         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
7998         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7999         if insn is not a tail call.
8000         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8001         registers.
8002         * config/arm/arm.h (enum reg_class): New caller save register class.
8003         (REG_CLASS_NAMES): Likewise.
8004         (REG_CLASS_CONTENTS): Likewise.
8005         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8006         without decls.
8008 2013-05-15  Richard Biener  <rguenther@suse.de>
8010         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8011         of MSG_OPTIMIZED_LOCATIONS.
8012         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8013         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8014         message.
8015         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8016         of MSG_OPTIMIZED_LOCATIONS.
8017         (execute_vect_slp): Likewise.
8018         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8019         (vect_create_cond_for_alias_checks): Likewise.
8020         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8021         (vect_recog_widen_mult_pattern): Likewise.
8022         (vect_recog_widen_sum_pattern): Likewise.
8023         (vect_recog_over_widening_pattern): Likewise.
8024         (vect_recog_widen_shift_pattern): Likewise.
8025         (vect_recog_vector_vector_shift_pattern): Likewise.
8026         (vect_recog_divmod_pattern): Likewise.
8027         (vect_recog_mixed_size_cond_pattern): Likewise.
8028         (vect_recog_bool_pattern): Likewise.
8029         (vect_pattern_recog_1): Likewise.
8031 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8033         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8034         non-functions to builtin_unreachable.
8035         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8036         correct when new direct edges were discovered.
8038 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8040         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8041         header, print symbol order instead of node uid, print more information
8042         about indirect edge targets.
8043         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8044         uids.
8045         (ipa_make_edge_direct_to_target): Likewise.
8046         (remove_described_reference): Likewise.
8047         (propagate_controlled_uses): Likewise.
8048         (ipa_print_node_params): Also print symbol order.
8049         (ipcp_transform_function): Print symbol order instead of node uids.
8050         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8051         (cgraph_get_create_real_symbol_node): Likewise.
8052         * ipa-cp.c (print_lattice): Likewise.
8053         (print_all_lattices): Likewise.
8054         (determine_versionability): Likewise.
8055         (initialize_node_lattices): Likewise.
8056         (estimate_local_effects): Likewise.
8057         (update_profiling_info): Likewise.
8058         (create_specialized_node): Likewise.
8059         (perhaps_add_new_callers): Likewise.
8060         (decide_about_value): Likewise.
8061         (decide_whether_version_node): Likewise.
8062         (identify_dead_nodes): Likewise.
8063         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8064         (dump_inline_summary): Likewise.
8065         (estimate_node_size_and_time): Likewise.
8066         (inline_analyze_function): Likewise.
8067         * ipa-inline.c (report_inline_failed_reason): Likewise.
8068         (want_early_inline_function_p): Likewise.
8069         (edge_badness): Likewise.
8070         (update_edge_key): Likewise.
8071         (inline_small_functions): Likewise.  Add dumping of order to two other
8072         dumps.
8073         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8074         instead of node uids.
8075         (propagate_pure_const): Likewise.
8076         (propagate_pure_const): Likewise.
8077         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8078         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8079         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8080         of node uids.
8081         * tree-pretty-print.c (dump_function_header): Likewise.
8082         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8083         Print symbol order instead of node uids.
8085 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8087         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8088         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8090 2013-05-15  Richard Biener  <rguenther@suse.de>
8092         PR tree-optimization/57275
8093         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8094         return value for fail to do runtime alias checks for gather loads.
8096 2013-05-15  Jan Hubicka  <jh@suse.cz>
8098         PR lto/57038
8099         PR lto/47375
8100         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8101         weakrefs are not external.
8102         (lto_symtab_merge_decls): Fix thinko when dealing with
8103         non-lto_symtab decls.
8104         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8105         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8106         * varpool.c (dump_varpool_node): Dump more flags.
8108 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8110         * config/i386/i386.c (processor_alias_table): Add instruction
8111         FSGSBASE for AMD bdver3 architecture.
8113 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8115         * tree.c (warn_deprecated_use): Print file:line using locus color.
8116         * diagnostic.c (diagnostic_report_current_module): Print file:line
8117         and file:line:column using locus color.
8119 2013-05-14  Mike Stump  <mikestump@comcast.net>
8121         * gdbinit.in: Add __null.
8123 2013-05-14  Mike Stump  <mikestump@comcast.net>
8125         * recog.h: Rename struct recog_data to Recog_data.
8126         * recog.c: Likewise.
8127         * reload.c (can_reload_into): Likewise.
8128         * config/picochip/picochip.c: Likewise.
8130 2013-05-14  Mike Stump  <mikestump@comcast.net>
8132         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8134 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8136         * resource.h (struct resources): Remove unch_memory member.
8137         (CLEAR_RESOURCE): Don't clear unch_memory.
8138         * resource.c (mark_referenced_resources): Don't set it.
8139         (mark_set_resources): Likewise.
8140         (mark_target_live_regs): Don't clear it.
8141         (init_resource_info): Likewise.
8142         * reorg.c (resource_conflicts_p): Don't compare it.
8143         (redundant_insn): Don't set it.
8145         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8146         Remove prototypes.
8147         * emit-rtl.c (next_label): Remove unused function.
8148         (skip_consecutive_labels, link_cc0_insns): Move to ...
8149         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8150         only place where these functions are used, and make them static.
8152 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8154         * fold-const.c (fold_negate_expr): Handle vectors.
8155         (fold_truth_not_expr): Make it static.
8156         (fold_invert_truthvalue): New static function.
8157         (invert_truthvalue_loc): Handle vectors. Do not call
8158         fold_truth_not_expr directly.
8159         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8160         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8161         (fold_comparison): Handle vector constants.
8162         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8163         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8164         * tree.h (fold_truth_not_expr): Remove declaration.
8166 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8168         * config/aarch64/aarch64-simd.md
8169         (aarch64_vcond_internal<mode>): Rename to...
8170         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8171         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8172         float modes. Clarify all iterator modes.
8173         (vcond<mode><mode>): Use new name for vcond expanders.
8174         (vcond<v_cmp_result><mode>): Likewise.
8175         (vcondu<mode><mode>: Likewise.
8176         * config/aarch64/iterators.md (VDQF_COND): New.
8178 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8180         PR bootstrap/57266
8181         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8182         variable for the shift amount. Check that we shift by non-negative
8183         amounts.
8185 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8187         PR target/42017
8188         * config/arm/arm.h (EPILOGUE_USES): Only return true
8189         for LR_REGNUM after epilogue_completed.
8191 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8193         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8194         is error_mark_node.
8196 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8198         PR target/57261
8199         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8200         and Solaris 11+/x86 with gld.
8201         * configure: Regenerate.
8203 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8205         * expmed.c (expand_shift_1): Canonicalize rotates by
8206         constant bitsize / 2 to bitsize - 1.
8207         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8208         case ROTATERT>: Likewise.
8210         Revert:
8211         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8213         * config/i386/i386.md (rotateinv): New code attr.
8214         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8215         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8216         roll $31, %eax, etc.
8218 2013-05-14  Richard Biener  <rguenther@suse.de>
8220         PR middle-end/57235
8221         * tree-eh.c (sink_clobbers): Give up for successors with
8222         multiple predecessors and no virtual uses.
8224 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8226         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8227         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8229 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8231         PR middle-end/57251
8232         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8233         the case when both op0 and op1 have VOIDmode.
8235 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8237         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8238         in multiply-accumulate mode.
8240 2013-05-13  Guozhi Wei  <carrot@google.com>
8242         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8244 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8246         PR target/56975
8247         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8248         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8249         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8250         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8251         (ix86_expand_prologue): Likewise.
8252         (ix86_expand_split_stack_prologue): Likewise.
8253         (legitimate_pic_address_disp_p): Likewise.
8254         (legitimize_pic_address): Likewise.
8255         (legitimize_tls_address): Likewise.
8256         (legitimize_pe_coff_symbol): Likewise.
8257         (output_pic_addr_const): Likewise.
8258         (construct_plt_address): Likewise.
8259         (ix86_expand_call): Likewise.
8260         (x86_output_mi_thunk): Likewise.
8261         (x86_function_profiler): Likewise.
8263 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8265         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8266         similar switch cases.
8267         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8268         (aarch64_simd_mov_to_<mode>low): Delete.
8269         (aarch64_simd_mov_to_<mode>high): Delete.
8270         (move_lo_quad_<mode>): Add w<-r alternative.
8271         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8272         (aarch64_simd_mov_from_*): Update type attribute.
8273         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8274         statement.
8276 2013-05-13  Jan Hubicka  <jh@suse.cz>
8278         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8279         * config/i386/i386.c (ix86_compute_frame_layout,
8280         ix86_expand_epilogue, emit_i387_cw_initialization,
8281         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8282         ix86_local_alignment): Fix use of size/speed predicates.
8284 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8286         PR tree-optimization/45216
8287         PR tree-optimization/57157
8288         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8289         the (-Y) & (B - 1) variant if OP is |.
8290         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8291         return shifted.  Use (-op1) & (prec - 1) as other_amount
8292         instead of prec - op1.
8294 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8296         PR middle-end/42371
8297         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8298         (ipa_constant_data): New type.
8299         (ipa_jump_func): Use ipa_constant_data to hold information about
8300         constant jump functions.
8301         (ipa_get_jf_constant): Adjust to jump function type changes.
8302         (ipa_get_jf_constant_rdesc): New function.
8303         (ipa_param_descriptor): New field controlled_uses.
8304         (ipa_get_controlled_uses): New function.
8305         (ipa_set_controlled_uses): Likewise.
8306         * ipa-ref.h (ipa_find_reference): Declare.
8307         * ipa-prop.c (ipa_cst_ref_desc): New type.
8308         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8309         changes.
8310         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8311         New parameter cs.  Adjust all callers.
8312         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8313         (remove_described_reference): New function.
8314         (jfunc_rdesc_usable): Likewise.
8315         (try_make_edge_direct_simple_call): Decrement controlled use count,
8316         attempt to remove reference if it hits zero.
8317         (combine_controlled_uses_counters): New function.
8318         (propagate_controlled_uses): Likewise.
8319         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8320         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8321         (ipa_print_node_params): Print described use counter.
8322         (ipa_write_jump_function): Adjust to jump function type changes.
8323         (ipa_read_jump_function): New parameter CS, pass it to
8324         ipa_set_jf_constant.  Adjust caller.
8325         (ipa_write_node_info): Stream controlled use count
8326         (ipa_read_node_info): Likewise.
8327         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8328         asserting.
8329         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8330         count.  Remove cloning-added reference if it reaches zero.
8331         * ipa-ref.c (ipa_find_reference): New function.
8333 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8335         * config/i386/i386.c (processor_target_table): Modified default
8336         alignment values for AMD BD and BT architectures.
8338 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8340         * tree-vect-generic.c (uniform_vector_p): Move ...
8341         * tree.c (uniform_vector_p): ... here.
8342         * tree.h (uniform_vector_p): Declare it.
8343         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8344         into a scalar.
8346 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8348         PR tree-optimization/57230
8349         * tree-ssa-strlen.c (handle_char_store): Record length for
8350         array store from STRING_CST.
8352         PR tree-optimization/57230
8353         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8354         check.
8356 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8358         * config/epiphany/epiphany.c (epiphany_init): Check size of
8359         NUM_MODES_FOR_MODE_SWITCHING.
8360         (epiphany_expand_prologue):
8361         Remove CONFIG_REGNUM initial value handling code.
8362         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8363         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8364         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8365         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8366         Don't return 1 for FP_MODE_NONE.
8367         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8368         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8369         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8370         * config/epiphany/epiphany.md (save_config): New pattern.
8372 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8374         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8376 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8378         * config/i386/i386.md (memory): Handle sseishft1.
8379         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8380         (*vec_extractv2di_1): Ditto.
8382 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8384         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8385         saved registers.
8387 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8389         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8390         Add mthumb/march=armv7-a multilib.
8391         Add mthumb/march=armv7-r multilib.
8392         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8394 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8396         * config/v850/t-rtems: Add more multilibs.
8398 2013-05-10  Richard Biener  <rguenther@suse.de>
8400         PR tree-optimization/57214
8401         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8402         not propagate from SSA names that occur in abnormal PHI nodes.
8404 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8406         * stor-layout.c (element_precision): New function.
8407         * machmode.h (element_precision): Declare it.
8408         * tree.c (build_minus_one_cst): New function.
8409         (element_precision): Likewise.
8410         * tree.h (build_minus_one_cst): Declare new function.
8411         (element_precision): Likewise.
8412         * fold-const.c (operand_equal_p): Use element_precision.
8413         (fold_binary_loc): Handle vector types.
8414         * convert.c (convert_to_integer): Use element_precision.
8415         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8416         separately.
8418 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8420         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8421         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8422         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8423         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8424         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8425         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8426         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8427         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8428         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8429         (Uuw8): New constraints.
8430         (Usb4): Move into alphabetical order.
8431         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8432         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8433         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8434         previously unnamed patterns.
8435         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8436         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8437         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8438         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8439         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8440         of specific lengths.
8442 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8444         * config/i386/i386.md (rotateinv): New code attr.
8445         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8446         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8447         roll $31, %eax, etc.
8449         PR tree-optimization/45216
8450         PR tree-optimization/57157
8451         * tree-ssa-forwprop.c (simplify_rotate): New function.
8452         (ssa_forward_propagate_and_combine): Call it.
8454 2013-05-10  Richard Biener  <rguenther@suse.de>
8456         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8457         disable peeling when we version for aliasing.
8458         (vector_alignment_reachable_p): Honor explicit user alignment.
8459         (vect_supportable_dr_alignment): Likewise.
8460         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8461         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8462         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8463         then peeling to arrange for the cost-model check to come first.
8465 2013-05-10  Alan Modra  <amodra@gmail.com>
8467         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8468         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8469         * configure: Regenerate.
8471 2013-05-10  Alan Modra  <amodra@gmail.com>
8473         PR target/55033
8474         * varasm.c (default_elf_select_section): Move !DECL_P check..
8475         (get_named_section): ..to here before calling get_section_name.
8476         Adjust assertion.
8477         (default_section_type_flags): Add DECL_P check.
8478         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8479         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8481 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8483         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8484         When using gen_stack_adjust_str with a register offset, add a
8485         REG_FRAME_RELATED_EXPR note.
8487 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8489         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8490         (*vec_extractv4si_zext_mem): Ditto.
8491         (*vec_extractv2di): Add 0->x and x->x alternatives.
8492         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8493         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8495 2013-05-09  Jason Merrill  <jason@redhat.com>
8497         N3639 C++1y VLA support
8498         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8499         DECL_VALUE_EXPR.
8501         * tree.c (build_constructor_va): New.
8502         * tree.h: Declare it.
8504 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8506         PR lto/57084
8507         * gimple-fold.c (canonicalize_constructor_val): Call
8508         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8510 2013-05-09  Jan Hubicka  <jh@suse.cz>
8511             Richard Biener  <rguenther@suse.de>
8513         PR lto/54095
8514         * symtab.c (symtab_make_decl_local): Do not add private names.
8516 2013-05-09  Jan Hubicka  <jh@suse.cz>
8518         PR lto/54095
8519         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8520         (unlink_from_assembler_name_hash): Likewise.
8521         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8522         symtab_unregister_node, symtab_initialize_asm_name_hash,
8523         change_decl_assembler_name): Update.
8525 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8527         * config/aarch64/aarch64.md: New movtf split.
8528         (*movtf_aarch64): Update.
8529         (aarch64_movdi_tilow): Handle TF modes and rename to
8530         aarch64_movdi_<mode>low.
8531         (aarch64_movdi_tihigh): Handle TF modes and rename to
8532         aarch64_movdi_<mode>high
8533         (aarch64_movtihigh_di): Handle TF modes and rename to
8534         aarch64_mov<mode>high_di
8535         (aarch64_movtilow_di): Handle TF modes and rename to
8536         aarch64_mov<mode>low_di
8537         (aarch64_movtilow_tilow): Remove spurious whitespace.
8538         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8539         splits.
8540         (aarch64_print_operand): Update.
8542 2013-05-09  Alan Modra  <amodra@gmail.com>
8544         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8545         powerpc64le.
8546         * configure: Regenerate.
8548 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8550         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8551         splitter preparation statements.
8552         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8553         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8554         adjust_address_nv.
8556 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8558         * gimple-ssa-strength-reduction.c (count_candidates): Change
8559         return value to int.
8560         (analyze_candidates_and_replace): Change type of length to int.
8562 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8564         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8565         (*vec_extract<mode>): Use VI12_128 mode iterator.
8566         (*vec_extract<mode>_mem): Ditto.
8567         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8568         attribute.
8570 2013-05-08  Diego Novillo  <dnovillo@google.com>
8572         PR bootstrap/54659
8574         Revert:
8575         2012-08-17  Diego Novillo  <dnovillo@google.com>
8577         PR bootstrap/54281
8578         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8579         * config.in: Regenerate.
8580         * configure: Regenerate.
8581         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8583 2013-05-08  Jan Hubicka  <jh@suse.cz>
8585         PR lto/54095
8586         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8587         * cgraph.h (symtab_node_base): Add unique_name.
8588         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8589         input_overwrite_node, input_varpool_node): Stream unique_name.
8590         * cgraphclones.c (cgraph_create_virtual_clone,
8591         cgraph_function_versioning): Set unique_name.
8592         * ipa.c (function_and_variable_visibility): Set unique_name.
8594 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8596         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8597         (alloc_cand_and_find_basis): Restrict conditional candidate
8598         processing to CAND_MULTs.
8600 2013-05-08  Jan Hubicka  <jh@suse.cz>
8602         PR lto/54095
8603         lto-symtab.c (lto_symtab_symbol_p): New function.
8604         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8605         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8606         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8607         Skip static symbols.
8609 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8611         PR tree-optimization/57200
8612         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8613         Only call inform if the preceding warning_at returns true.
8615 2013-05-07  Han Shen  <shenhan@google.com>
8617         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8618         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8619         * common.opt (fstack-protector-strong): New option.
8620         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8621         * doc/invoke.texi (Optimization Options): Document
8622         "-fstack-protector-strong".
8623         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8625 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8627         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8629 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8631         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8632         it is not lhs of the insn.
8633         (setup_reg_equiv): Remove insn having equiv memory which it is not
8634         lhs of the insn.
8635         * lra-constraints.c (process_address): Try to improve generation
8636         code for address base + disp.
8637         (lra_constraints): Make correct the code for checking insn setting
8638         up backward equivalence.  Remove insn only if it is in the init
8639         insn list.
8640         * lra-eliminations.c (update_reg_eliminate): Change return value.
8641         (lra_eliminate): Use the result.
8643 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8645         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8646         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8647         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8648         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8649         register target operands.
8650         (*vec_extractv8hi_sse2): New pattern.
8651         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8652         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8653         (*vec_extract<mode>_mem): New insn and split pattern.
8655 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8657         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8658         (TARGET_ASAN_SHADOW_OFFSET): Define.
8659         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8660         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8662 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8664         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8665         (incr_vec_index): Return -1 if increment not found.
8666         (create_add_on_incoming_edge): Assert if increment not found.
8667         (record_increment): Limit number of increments recorded.
8668         (all_phi_incrs_profitable): Return false if an increment not found.
8669         (replace_profitable_candidates): Don't process increments that were
8670         not recorded.
8671         (analyze_candidates_and_replace): Limit size of incr_vec.
8673 2013-05-07  Richard Biener  <rguenther@suse.de>
8675         * calls.c (special_function_p): setjmp-like functions are leaf.
8676         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8677         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8679 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
8681         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8682         (aarch64_simd_mov<mode>): New expander.
8683         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8684         (aarch64_simd_mov_to_<mode>high): Likewise.
8685         (aarch64_simd_mov_from_<mode>low): Likewise.
8686         (aarch64_simd_mov_from_<mode>high): Likewise.
8687         (aarch64_dup_lane<mode>): Update.
8688         (aarch64_dup_lanedi): New instruction pattern.
8689         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8690         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8692 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8694         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8695         (replace_mult_candidate): Remove unnecessary argument; remove
8696         unnecessary parameter from call to introduce_cast_before_cand.
8697         (replace_unconditional_candidate): Remove unnecessary parameter
8698         from call to replace_mult_candidate.
8699         (replace_conditional_candidate): Likewise.
8700         (insert_initializers): Use make_temp_ssa_name.
8701         (introduce_cast_before_cand): Remove unnecessary argument; use
8702         make_temp_ssa_name.
8703         (replace_one_candidate): Remove unnecessary argument; remove
8704         unnecessary parameter from calls to introduce_cast_before_cand.
8705         (replace_profitable_candidates): Remove unnecessary parameters
8706         from calls to replace_one_candidate.
8708 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8710         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8711         phi def as possibly hiding a basis for a CAND_ADD whose operands
8712         have been commuted in the analysis.
8713         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8715 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
8717         * config/aarch64/aarch64.md
8718         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8719         shift value between 0-4.
8721 2013-05-07  Richard Biener  <rguenther@suse.de>
8723         * double-int.h (rshift): New overload.
8724         * double-int.c (rshift): New function.
8725         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8726         (create_reference_ops_from_ref): Remove.
8727         (vn_reference_insert): Use shared ops for constructing the
8728         reference and copy it.
8730 2013-05-07  Richard Biener  <rguenther@suse.de>
8732         PR middle-end/57190
8733         * tree-eh.c (sink_clobbers): Properly propagate
8734         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8736 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
8738         PR tree-optimization/57149
8739         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8740         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8741         collect_phi_def_edges, execute_late_warn_uninitialized): Use
8742         uninit_undefined_value_p instead of ssa_undefined_value_p.
8744         PR debug/57184
8745         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8746         for modifier == EXPAND_INITIALIZER.
8748 2013-05-07  Anton Blanchard  <anton@samba.org>
8750         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8751         for powerpc64 little endian.
8752         * configure: Regenerate.
8754 2013-05-06  Graham Stott  <grahams@btinternet.com>
8756         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8757         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8758         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8759         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8761 2013-05-06  Graham Stott  <grahams@btinternet.com>
8763         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8764         codes which allow non-lvalues.
8766 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
8768         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8769         components are all 1s.
8770         (integer_minus_onep): New function.
8771         * tree.h (integer_minus_onep): Declare it.
8772         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8773         integer_minus_onep instead of integer_all_onesp.
8775 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8777         PR target/52933
8778         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8779         variations of these patterns.
8781 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8783         * config/i386/i386.md (isa): Add x64_sse4 member.
8784         (enabled): Handle x64_sse4.
8785         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8786         instruction for 64bit SSE4_1 targets.  Update insn attributes.
8787         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8788         instruction for SSE4_1 targets.  Update insn attributes.
8789         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8790         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8791         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8792         const_1 selector.
8793         (*vec_extractv4si): Rename from *sse4_1_pextrd.
8794         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8795         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8797 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8799         PR target/57108
8800         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8802 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
8804         * final.c (do_assembler_dialects): Don't handle curly braces and
8805         vertical bar escaped by % as dialect delimiters.
8806         (output_asm_insn): Print curly braces and vertical bar if escaped
8807         by % and ASSEMBLER_DIALECT defined.
8808         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8809         * doc/tm.texi: Regenerated.
8811 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8813         * config/mips/mips.c: Include tree-pass.h.
8814         (mips_reorg): Split in pre- and post-dbr_schedule parts.
8815         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8816         (pass_mips_machine_reorg2): New machine specific pass.
8817         (insert_pass_mips_machine_reorg2): New pass plugin definition.
8818         (mips_option_override): Register the new pass.
8819         * rtl.h (cleanup_barriers): Remove prototype.
8820         (dbr_schedule): Likewise.
8821         * jump.c (cleanup_barriers): Make static.
8822         * reorg.c (dbr_schedule): Likewise.
8824 2013-05-06  Richard Biener  <rguenther@suse.de>
8826         PR tree-optimization/57185
8827         * tree-parloops.c (add_field_for_reduction): Handle anonymous
8828         SSA names properly.
8830 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8832         PR target/57106
8833         * config/i386/i386.c (add_parameter_dependencies): Add dependence
8834         between "first_arg" and "insn", not "last" and "insn".
8836 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8838         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8839         (find_candidates_in_block): Re-enable slsr_process_phi.
8840         (create_phi_basis): Fix double counting of candidate adjustment.
8842 2013-05-06  Richard Biener  <rguenther@suse.de>
8844         PR middle-end/57147
8845         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8846         the edge is also fallthru, preserve it and just clear the
8847         abnormal flag.
8848         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8849         also complex, preserve that and just clear the fallthru flag.
8850         * tree-inline.c (update_ssa_across_abnormal_edges): Also
8851         update virtual operands.
8853 2013-05-06  Alan Modra  <amodra@gmail.com>
8855         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8856         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8857         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8858         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8859         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8860         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8862 2013-05-06  Alan Modra  <amodra@gmail.com>
8864         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8865         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
8866         (DEFAULT_ASM_ENDIAN): Define.
8867         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8868         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8869         Update -K PIC clause from sysv4.h.
8870         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8871         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8873 2013-05-06  Alan Modra  <amodra@gmail.com>
8875         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8876         twice for little-endian.
8877         (ashrdi3_no_power, ashrdi3): Support little-endian.
8879 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8881         PR target/55303
8882         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8883         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8884         related expanders.
8885         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8886         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8887         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8888         New predicates.
8890 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
8891             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
8893         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8894         * config/pa/pa.opt: Make mbig-switch a no-op.
8895         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8896         (CASE_VECTOR_MODE): Always return SImode.
8897         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8898         for the !TARGET_BIG_SWITCH case.
8899         * config/pa/pa-linux.h: Likewise.
8900         * config/pa/pa-openbsd.h: Likewise.
8901         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8902         * config/pa/pa.md (short_jump): Remove define_insn.
8903         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8904         (casesi0): Remove define_insn.
8905         (type): Remove btable_branch.
8906         (pa_combine_type): Likewise.
8907         (in_nullified_branch_delay): Likewise.
8908         (in_call_delay): Likewise.
8909         (define_delay): Likewise.
8910         (define_insn_reservation "Z3"): Likewise.
8911         (define_insn_reservation "Z4"): Likewise.
8912         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8913         (pa_adjust_insn_length): Remove adjustment for btable branches.
8914         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8915         and mno-big-switch
8917 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
8919         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8920         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8921         Add m->r,x alternatives.
8922         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8923         splitters using SWI48x mode iterator.
8924         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
8925         TARGET_64BIT.  Add m->x alternative.
8926         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8927         Add o->x alternative.  Enable for TARGET_SSE.
8928         (sse_storeq): Remove expander.
8929         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8930         with memory input operand.
8931         (*vec_extractv2di_1 splitter): New.
8932         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8933         * config/i386/i386.md (ssevecmodelower): New mode attribute.
8935 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8937         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
8938         (INT_LOWPART): Delete.
8939         (extract_MB): Adjust.
8940         (extract_ME): Adjust.
8941         (print_operand): Adjust.
8943 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8945         * config/rs6000/predicates.md (reg_or_add_cint_operand,
8946         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8947         (reg_or_logical_cint_operand, easy_fp_constant,
8948         logical_const_operand): Delete "CONST_DOUBLE" case.
8949         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8950         "HOST_BITS_PER_WIDE_INT == 64" test.
8951         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
8952         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8953         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8954         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8955         test.
8956         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8957         CONST_DOUBLE DImode/VOIDmode case.
8958         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8959         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8960         CONST_DOUBLE VOIDmode case.
8961         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8962         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8963         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8964         Delete CONST_DOUBLE case.
8965         (splitters for mov FMOVE64 const_double): Delete
8966         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8967         "HOST_BITS_PER_WIDE_INT >= 64" test.
8968         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8969         case.
8970         (mov DI const_double): Delete.
8972 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
8974         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8975         on op shows all bits zero in mode of a lowpart subreg, return zero.
8977 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
8979         PR target/57150
8980         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8981         to save TFmode registers and DImode to save TImode registers for
8982         caller save operations.
8983         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8984         mark being partially clobbered since they only use the first
8985         double word.
8987         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8988         and TDmode only use the upper 64-bits of each VSX register.
8990 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8992         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8993         (find_candidates_in_block): Disable slsr_process_phi.
8995 2013-05-03  Guozhi Wei  <carrot@google.com>
8997         * coverage.c (coverage_obj_init): Move the construction of gcov
8998         constructor to ...
8999         (build_init_ctor): ... here.
9001 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9003         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9004         (slsr_cand_d): Redefine def_phi.
9005         (stride_status, phi_adjust_status, count_phis_status): New enums.
9006         (find_phi_def): New.
9007         (find_basis_for_base_expr): New.
9008         (find_basis_for_candidate): Handle hidden bases.
9009         (alloc_cand_and_find_basis): Handle phi candidates.
9010         (slsr_process_phi): New.
9011         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9012         (create_mul_imm_cand): Likewise.
9013         (create_add_ssa_cand): Exclude phi base candidates.
9014         (create_add_imm_cand): Likewise.
9015         (slsr_process_cast): Likewise.
9016         (slsr_process_copy): Likewise.
9017         (find_candidates_in_block): Handle phi candidates.
9018         (dump_candidate): Likewise.
9019         (unconditional_cands): Delete.
9020         (unconditional_cands_with_known_stride_p): Delete.
9021         (phi_dependent_cand_p): New.
9022         (cand_increment): Handle phi-dependent candidates.
9023         (replace_dependent): Delete.
9024         (replace_mult_candidate): New.
9025         (replace_unconditional_candidate): New.
9026         (incr_vec_index): Move to avoid forward reference.
9027         (create_add_on_incoming_edge): New.
9028         (create_phi_basis): New.
9029         (replace_dependents): Delete.
9030         (replace_conditional_candidate): New.
9031         (phi_add_costs): New.
9032         (replace_uncond_cands_and_profitable_phis): New.
9033         (record_increment): Handle phi adjustments.
9034         (record_phi_increments): New.
9035         (record_increments): Handle phi adjustments.
9036         (phi_incr_cost): New.
9037         (lowest_cost_path): Handle phis.
9038         (total_savings): Likewise.
9039         (analyze_increments): Likewise.
9040         (ncd_with_phi): New.
9041         (ncd_of_cand_and_phis): New.
9042         (nearest_common_dominator_for_cands): Handle phi increments.
9043         (all_phi_incrs_profitable): New.
9044         (replace_profitable_candidates): Handle phi-dependent candidates.
9045         (analyze_candidates_and_replace): Likewise.
9047 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9049         PR bootstrap/57154
9050         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9051         do not exceed REG_BR_PROB_BASE.
9053 2013-05-03  Jeff Law  <law@redhat.com>
9055         PR tree-optimization/57144
9056         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9057         operand of the condition will bit into the new type when eliminating
9058         a cast feeding a condition.
9060 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9062         PR rtl-optimization/57130
9063         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9064         of COMPARE as in_code to the recursive call if needed.
9066 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9068         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9069         (enabled): Handle new members.
9070         * config/i386/sse.md (*vec_concatv2si): Merge from
9071         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9072         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9074 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9076         PR tree-optimization/57027
9077         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9078         for fnms opportunity, check we got the prerequisite kind
9079         of tree / gimple before using accessor functions.
9081 2013-05-03  Richard Biener  <rguenther@suse.de>
9083         * double-int.h (lshift): New overload without precision
9084         and arith argument.
9085         (operator *=, operator +=, operator -=): Move ...
9086         * double-int.c (operator *=, operator +=, operator -=): ... here
9087         and implement more efficiently.
9088         (mul_double_with_sign): Remove.
9089         (lshift_double): Adjust to take unsinged shift argument, push
9090         dispatching code to callers.
9091         (mul_double_wide_with_sign): Add early out for callers that
9092         are not interested in high parts or overflow.
9093         (lshift): New function.
9094         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9095         dispatch code here.
9096         (lrotate, rrotate): Use logical shifts.
9097         * expr.c (get_inner_reference): Use lshift.
9098         * fixed-value.c (do_fixed_divide): Likewise.
9099         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9100         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9101         (indirect_refs_may_alias_p): Likewise.
9102         (stmt_kills_ref_p_1): Likewise.
9104 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9106         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9108 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9110         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9111         scalar form of FABD instruction.
9113 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9115         * lra-constraints.c (process_alt_operands): Add checking alt
9116         number to choose the best alternative.
9118 2013-05-02  Richard Biener  <rguenther@suse.de>
9120         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9121         bitmap and its handling.
9122         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9124 2013-05-02  Richard Biener  <rguenther@suse.de>
9126         PR middle-end/57140
9127         * tree-inline.c (copy_loops): Properly handle removed loops.
9128         (copy_cfg_body): Mark destination loops for fixup if source
9129         loops needed fixup.
9131 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9133         PR target/56732
9134         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9135         generating simple_return for naked functions.
9137 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9139         PR middle-end/56988
9140         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9141         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9142         flags match.
9143         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9144         ipa_agg_replacement_value structures.
9145         (known_aggs_to_agg_replacement_list): Likewise.
9146         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9147         (read_agg_replacement_chain): Likewise.
9148         (ipcp_transform_function): Also check that by_ref flags match.
9150 2013-05-02  Richard Biener  <rguenther@suse.de>
9152         * graphds.h (struct graph): Add obstack member.
9153         * graphds.c (new_graph): Initialize obstack and allocate
9154         vertices from it.
9155         (add_edge): Allocate edge from the obstack.
9156         (free_graph): Free the obstack instead of all edges and vertices.
9158 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9160         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9161         divides.
9162         * cfg.c (update_bb_profile_for_threading): Ditto.
9163         * tree-inline.c (copy_bb): Ditto.
9164         (copy_edges_for_bb): Ditto.
9165         (initialize_cfun): Ditto.
9166         (copy_cfg_body): Ditto.
9167         (expand_call_inline): Ditto.
9168         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9169         (estimate_node_size_and_time): Ditto.
9170         (inline_merge_summary): Ditto.
9171         * cgraphclones.c (cgraph_clone_edge): Ditto.
9172         (cgraph_clone_node): Ditto.
9173         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9174         (compute_trg_info): Ditto.
9176 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9178         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9179         S reg when fp attribute set.
9180         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9182 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9184         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9185         New pattern.
9186         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9187         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9188         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9190 2013-05-02  Richard Biener  <rguenther@suse.de>
9192         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9193         (struct instantiate_cache_entry): New type.
9194         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9195         (struct instantiate_cache_type): New type.
9196         (set_instantiated_value, get_instantiated_value): Remove.
9197         (get_instantiated_value_entry): New function.
9198         (instantiate_scev_name): Use the new cache and adjust.
9199         (instantiate_scev_poly): Adjust.
9200         (instantiate_scev_binary): Likewise.
9201         (instantiate_array_ref): Likewise.
9202         (instantiate_scev_convert): Likewise.
9203         (instantiate_scev_not): Likewise.
9204         (instantiate_scev_3): Likewise.
9205         (instantiate_scev_2): Likewise.
9206         (instantiate_scev_r): Likewise.
9207         (instantiate_scev): Likewise.
9208         (resolve_mixers): Likewise.
9210 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9212         PR target/57091
9213         * lra-constraints.c (best_small_class_operands_num): Remove.
9214         (process_alt_operands): Remove small_class_operands_num.  Take
9215         small classes operands into losers and only if the operand is not
9216         matched.  Modify debugging output.
9217         (curr_insn_transform): Remove best_small_class_operands_num.
9218         Print insn name.
9220 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9222         * config/aarch64/aarch64-builtins.c
9223         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9224         * config/aarch64/aarch64-simd-builtins.def
9225         (reduc_splus_): Add new modes.
9226         (reduc_uplus_): New.
9227         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9228         (reduc_uplus_v4sf): Likewise.
9229         (reduc_splus_v4sf): Likewise.
9230         (aarch64_addv<mode>): Likewise.
9231         (reduc_uplus_<mode>): Likewise.
9232         (reduc_splus_<mode>): Likewise.
9233         (aarch64_addvv2di): Likewise.
9234         (reduc_uplus_v2di): Likewise.
9235         (reduc_splus_v2di): Likewise.
9236         (aarch64_addvv2si): Likewise.
9237         (reduc_uplus_v2si): Likewise.
9238         (reduc_splus_v2si): Likewise.
9239         (reduc_<sur>plus_<mode>): New.
9240         (reduc_<sur>plus_v2di): Likewise.
9241         (reduc_<sur>plus_v2si): Likewise.
9242         (reduc_<sur>plus_v4sf): Likewise.
9243         (aarch64_addpv4sf): Likewise.
9244         * config/aarch64/arm_neon.h
9245         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9246         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9247         add UNSPEC_SADDV, UNSPEC_UADDV.
9248         (SUADDV): New.
9249         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9251 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9253         * config/aarch64/arm_neon.h
9254         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9256 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9258         * config/aarch64/aarch64-builtins
9259         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9261 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9263         * config/aarch64/aarch64-simd-builtins.def
9264         (reduc_smax_): New.
9265         (reduc_smin_): Likewise.
9266         (reduc_umax_): Likewise.
9267         (reduc_umin_): Likewise.
9268         (reduc_smax_nan_): Likewise.
9269         (reduc_smin_nan_): Likewise.
9270         (fmax): Remove.
9271         (fmin): Likewise.
9272         (smax): Update for V2SF, V4SF and V2DF modes.
9273         (smin): Likewise.
9274         (smax_nan): New.
9275         (smin_nan): Likewise.
9276         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9277         (<su><maxmin><mode>3): ...This, refactor.
9278         (s<maxmin><mode>3): New.
9279         (<maxmin_uns><mode>3): Likewise.
9280         (reduc_<maxmin_uns>_<mode>): Refactor.
9281         (reduc_<maxmin_uns>_v4sf): Likewise.
9282         (reduc_<maxmin_uns>_v2si): Likewise.
9283         (aarch64_<fmaxmin><mode>: Remove.
9284         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9285         new builtin names.
9286         (vmin<q>_f<32,64>): Likewise.
9287         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9288         (FMAXMIN): New.
9289         (su): Add mappings for smax, smin, umax, umin.
9290         (maxmin): New.
9291         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9292         (FMAXMIN): Rename as...
9293         (FMAXMIN_UNS): ...This.
9294         (maxminv): Remove.
9295         (fmaxminv): Likewise.
9296         (fmaxmin): Likewise.
9297         (maxmin_uns): New.
9298         (maxmin_uns_op): Likewise.
9300 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9302         * config/aarch64/arm_neon.h
9303         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9304         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9305         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9307 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9309         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9310         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9312 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9314         * config/aarch64/aarch64-simd.md
9315         (vcond<mode>_internal): Handle special cases for constant masks.
9316         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9317         (vcondu<mode><mode>): Likewise.
9318         (vcond<v_cmp_result><mode>): New.
9320 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9322         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9323         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9324         * config/aarch64/aarch64-simd-builtins.def
9325         (cmeq): Update to BUILTIN_VALLDI.
9326         (cmgt): Likewise.
9327         (cmge): Likewise.
9328         (cmle): Likewise.
9329         (cmlt): Likewise.
9330         * config/aarch64/arm_neon.h
9331         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9332         to builtins or C as appropriate.
9334 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9336         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9337         (cmgeu): ...This.
9338         (cmhi): Rename to...
9339         (cmgtu): ...This.
9340         * config/aarch64/aarch64-simd.md
9341         (simd_mode): Add SF.
9342         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9343         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9344         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9345         (cstore<mode>_neg): ...This.
9346         * config/aarch64/iterators.md
9347         (VALLF): new.
9348         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9349         (COMPARISONS): New.
9350         (UCOMPARISONS): Likewise.
9351         (optab): Add missing comparisons.
9352         (n_optab): New.
9353         (cmp_1): Likewise.
9354         (cmp_2): Likewise.
9355         (CMP): Likewise.
9356         (cmp): Remove.
9357         (VCMP_S): Likewise.
9358         (VCMP_U): Likewise.
9359         (V_cmp_result): Add DF, SF modes.
9360         (v_cmp_result): Likewise.
9361         (v): Likewise.
9362         (vmtype): Likewise.
9363         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9365 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9367         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9368         define_insn to define_insn_and_split.
9369         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9370         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9371         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9372         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9373         (thumb2_negscc): Likewise.
9375 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9377         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9379 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9381         * config/arm/thumb2.md: Remove trailing whitespaces.
9383 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9385         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9386         Use gen_int_mode rather than GEN_INT.
9388 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9390         * value-prof.c (stream_in_histogram_value): Remove the strayed
9391         debug_gimple_stmt.
9393 2013-04-30  Richard Biener  <rguenther@suse.de>
9395         PR middle-end/57122
9396         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9398 2013-04-30  Richard Biener  <rguenther@suse.de>
9400         PR middle-end/57107
9401         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9403 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9405         PR rtl-optimization/56957
9406         PR rtl-optimization/57105
9407         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9408         variable.  Use just INSN_UID for determining whether an insn
9409         should be only disconnected from the insn stream.
9410         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9412 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9414         PR tree-optimization/57104
9415         * tsan.c (instrument_expr): Don't instrument accesses to
9416         DECL_HARD_REGISTER VAR_DECLs.
9418 2013-04-30  Richard Biener  <rguenther@suse.de>
9420         * function.h (loops_for_fn): New inline function.
9421         (set_loops_for_fn): Likewise.
9422         * cfgloop.h (place_new_loop): Add struct function parameter.
9423         (get_loop): Likewise.
9424         (get_loops): Likewise.
9425         (number_of_loops): Likewise.
9426         (fel_next): Adjust.
9427         (fel_init): Likewise.
9428         * cfg.c (get_loop_copy): Adjust.
9429         * cfgloop.c (flow_loops_dump): Likewise.
9430         (record_loop_exits): Likewise.
9431         (verify_loop_structure): Likewise.
9432         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9433         (estimate_reg_pressure_cost): Likewise.
9434         (mark_loop_exit_edges): Likewise.
9435         * cfgloopmanip.c (place_new_loop): Likewise.
9436         (add_loop): Likewise.
9437         (duplicate_loop): Likewise.
9438         * graph.c (draw_cfg_nodes): Likewise.
9439         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9440         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9441         (extract_affine_chrec): Likewise.
9442         (build_scop_iteration_domain): Likewise.
9443         * graphite.c (graphite_initialize): Likewise.
9444         * ira-build.c (create_loop_tree_nodes): Likewise.
9445         (more_one_region_p): Likewise.
9446         (rebuild_regno_allocno_maps): Likewise.
9447         (mark_loops_for_removal): Likewise.
9448         (mark_all_loops_for_removal): Likewise.
9449         (remove_unnecessary_regions): Likewise.
9450         (ira_build): Likewise.
9451         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9452         * loop-init.c (fix_loop_structure): Likewise.
9453         (gate_rtl_move_loop_invariants): Likewise.
9454         (gate_rtl_unswitch): Likewise.
9455         (gate_rtl_unroll_and_peel_loops): Likewise.
9456         (rtl_doloop): Likewise.
9457         * lto-streamer-in.c (input_cfg): Likewise.
9458         * lto-streamer-out.c (output_cfg): Likewise.
9459         * modulo-sched.c (sms_schedule): Likewise.
9460         * predict.c (tree_estimate_probability): Likewise.
9461         (tree_estimate_probability_driver): Likewise.
9462         (estimate_loops): Likewise.
9463         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9464         (move_sese_region_to_fn): Likewise.
9465         (debug_loop_num): Likewise.
9466         * tree-chrec.c (chrec_evaluate): Likewise.
9467         (hide_evolution_in_other_loops_than_loop): Likewise.
9468         (chrec_component_in_loop_num): Likewise.
9469         (reset_evolution_in_loop): Likewise.
9470         (evolution_function_is_invariant_rec_p): Likewise.
9471         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9472         * tree-inline.c (copy_loops): Likewise.
9473         (copy_cfg_body): Likewise.
9474         (tree_function_versioning): Likewise.
9475         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9476         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9477         Likewise.
9478         (add_to_evolution_1): Likewise.
9479         (scev_const_prop): Likewise.
9480         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9481         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9482         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9483         (tree_ssa_lim_initialize): Likewise.
9484         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9485         (verify_loop_closed_ssa): Likewise.
9486         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9487         (tree_ssa_loop_im): Likewise.
9488         (tree_ssa_loop_unswitch): Likewise.
9489         (tree_vectorize): Likewise.
9490         (check_data_deps): Likewise.
9491         (tree_ssa_loop_ivcanon): Likewise.
9492         (tree_ssa_loop_bounds): Likewise.
9493         (tree_complete_unroll): Likewise.
9494         (tree_complete_unroll_inner): Likewise.
9495         (tree_parallelize_loops): Likewise.
9496         (tree_ssa_loop_prefetch): Likewise.
9497         (tree_ssa_loop_ivopts): Likewise.
9498         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9499         * tree-vectorizer.c (vectorize_loops): Likewise.
9501 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9503         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9504         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9505         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9506         with EABI_LINK_SPEC.
9508 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9510         PR target/44578
9511         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9512         alternative.
9514 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9516         PR target/57097
9517         * lra-constraints.c (process_alt_operands): Discourage a bit more
9518         using memory for pseudos.  Print cost dump for alternatives.
9519         Modify cost values for conflicts with early clobbers.
9520         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9522 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9524         PR target/57098
9525         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9527 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9529         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9530         from/to S register.
9531         (movdi_aarch64): Support LDR/STR from/to D register.
9533 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9535         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9536         or higher by default.
9538 2013-04-29  Richard Biener  <rguenther@suse.de>
9540         PR middle-end/57075
9541         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9542         even if not adding abnormal edges for calls that can make
9543         abnormal gotos.
9545 2013-04-29  Richard Biener  <rguenther@suse.de>
9547         PR middle-end/57103
9548         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9549         TREE_BLOCK.
9550         (move_stmt_r): Remove redundant checking.
9552 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9554         PR bootstrap/57077
9555         * basic-block.h (apply_scale): New function.
9556         (apply_probability): Use apply_scale.
9557         * gimple-streamer-in.c (input_bb): Ditto.
9558         * lto-streamer-in.c (input_cfg): Ditto.
9559         * lto-cgraph.c (merge_profile_summaries): Ditto.
9560         * tree-optimize.c (execute_fixup_cfg): Ditto.
9561         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9562         (copy_edges_for_bb): Ditto.
9563         (copy_cfg_body): Ditto.
9565 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9567         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9568         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9569         (tail_merge_optimize): Handle current_loops == NULL.
9571 2013-04-26  Jeff Law  <law@redhat.com>
9573         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9574         (simplify_cond_using_ranges): Generalize code to simplify
9575         COND_EXPRs where one argument is a constant and the other
9576         is an SSA_NAME created by an integral type conversion.
9578 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9580         * config/arm/arm.md (store_minmaxsi): Use only when
9581         optimize_insn_for_size_p.
9583 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9585         PR target/57108
9586         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9588 2013-04-29  Richard Biener  <rguenther@suse.de>
9590         PR middle-end/57089
9591         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9592         loop tree make sure to schedule a fixup for the child as well.
9593         (expand_omp_for_generic): Properly add loops.
9594         (expand_omp_for_static_nochunk): Likewise.
9595         (expand_omp_for_static_chunk): Likewise.
9596         (expand_omp_for): For the degenerate case fixup loops.
9597         (expand_omp_sections): Fix default bb placement in loops.
9598         (expand_omp_atomic_pipeline): Properly add loops.
9600 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9602         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9604 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9606         * tree-ssa-tail-merge.c: Update header comment.
9608 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9610         * config/aarch64/arm_neon.h
9611         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9612         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9613         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9614         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9615         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9616         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9617         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9618         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9620 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9622         * config/aarch64/aarch64-simd.md
9623         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9624         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9625         fix_trunc, fixuns_trunc.
9626         (ftrunc<VDQF:mode>2): New.
9627         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9628         (fix_trunc_optab): New.
9630 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9632         * config/aarch64/aarch64-builtins.c
9633         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9634         iceilf, lround, iroundf.
9636 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9638         PR target/54349
9639         * config/i386/i386.h (enum ix86_tune_indices)
9640         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9641         New, split from X86_TUNE_INTER_UNIT_MOVES.
9642         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9643         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9644         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9645         (TARGET_INTER_UNIT_MOVES): Remove.
9646         * config/i386/i386.c (initial_ix86_tune_features): Update.
9647         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9648         (ix86_expand_convert_uns_didf_sse): Use
9649         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9650         (ix86_expand_vector_init_one_nonzero): Ditto.
9651         (ix86_expand_vector_init_interleave): Ditto.
9652         (inline_secondary_memory_needed): Return true for moves from SSE class
9653         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9654         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9655         * config/i386/constraints.md (Yi, Ym): Depend on
9656         TARGET_INTER_UNIT_MOVES_TO_VEC.
9657         (Yj, Yn): New constraints.
9658         * config/i386/i386.md (*movdi_internal): Change constraints of
9659         operand 1 from Yi to Yj and from Ym to Yn.
9660         (*movsi_internal): Ditto.
9661         (*movdf_internal): Ditto.
9662         (*movsf_internal): Ditto.
9663         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9664         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9665         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9666         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9667         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9668         * config/i386/sse.md (movdi_to_sse): Ditto.
9669         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9670         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9671         TARGET_INTER_UNIT_MOVES.
9672         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9673         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9674         instead of TARGET_INTER_UNIT_MOVES.
9675         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9676         operand 1 from Yi to Yj and from Ym to Yn.
9678 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9680         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9681         (float_truncate_hi_): Likewise.
9682         (float_extend_lo_): Likewise.
9683         (float_truncate_lo_): Likewise.
9684         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9685         (aarch64_float_extend_lo_v2df): Likewise.
9686         (vec_unpacks_hi_v4sf): Likewise.
9687         (aarch64_float_truncate_lo_v2sf): Likewise.
9688         (aarch64_float_truncate_hi_v4sf): Likewise.
9689         (vec_pack_trunc_v2df): Likewise.
9690         (vec_pack_trunc_df): Likewise.
9692 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9694         * config/aarch64/aarch64-builtins.c
9695         (aarch64_fold_builtin): Fold float conversions.
9696         * config/aarch64/aarch64-simd-builtins.def
9697         (floatv2si, floatv4si, floatv2di): New.
9698         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9699         * config/aarch64/aarch64-simd.md
9700         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9701         * config/aarch64/iterators.md (FLOATUORS): New.
9702         (optab): Add float, floatuns.
9703         (su_optab): Likewise.
9705 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9707         * config/aarch64/aarch64-builtins.c
9708         (aarch64_builtin_vectorized_function): Use new names for
9709         fcvt builtins.
9710         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9711         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9712         (fcvtzu): Split as...
9713         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9714         (fcvtas): Split as...
9715         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9716         (fcvtau): Split as...
9717         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9718         (fcvtps): Split as...
9719         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9720         (fcvtpu): Split as...
9721         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9722         (fcvtms): Split as...
9723         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9724         (fcvtmu): Split as...
9725         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9726         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9727         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9728         (lfrintnusf, lfrintnudf): Likewise.
9729         * config/aarch64/aarch64-simd.md
9730         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9731         define_insn.
9732         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9733         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9734         (fcvt_pattern): Likewise.
9736 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9738         * config/aarch64/aarch64-simd.md
9739         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9740         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9742 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9744         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9745         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9746         (vrnd<a,m,n,p>_f32): Implement using builtins.
9747         (vrnd<i,x><q>_f<32, 64>): New.
9749 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9751         * config/aarch64/aarch64-builtins.c
9752         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9753         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9754         (frintz): Rename to...
9755         (btrunc): ...this.
9756         (frintp): Rename to...
9757         (ceil): ...this.
9758         (frintm): Rename to...
9759         (floor): ...this.
9760         (frinti): Rename to...
9761         (nearbyint): ...this.
9762         (frintx): Rename to...
9763         (rint): ...this.
9764         (frinta): Rename to...
9765         (round): ...this.
9766         * config/aarch64/aarch64-simd.md
9767         (aarch64_frint<frint_suffix><mode>): Delete.
9768         (<frint_pattern><mode>2): Convert to insn.
9769         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9770         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9771         (frint_pattern): Likewise.
9772         (frint_suffix): Likewise.
9774 2013-04-29  Richard Biener  <rguenther@suse.de>
9776         PR tree-optimization/57081
9777         * loop-init.c: Include tree-flow.h.
9778         (loop_optimizer_finalize): Free number of iteration estimates.
9779         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9781 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
9783         PR tree-optimization/57083
9784         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9785         non-singleton shift count range, zero extend low_bound for uns case.
9787         * config/i386/predicates.md (general_vector_operand): New predicate.
9788         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9789         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9790         if they aren't nonimmediate operands.  If their original values
9791         satisfy const_vector_equal_evenodd_p, don't shift them.
9792         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9793         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
9794         if not nonimmediate_operand.
9795         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9796         instead of register_operand.
9797         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9799 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
9801         * stor-layout.c (finalize_size_functions): Allocate a structure and
9802         reset cfun before dumping the functions.
9804 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
9806         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9808         PR target/56866
9809         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9810         use xop_pmacsdqh if uns_p.
9811         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9812         the immediate rotate count.
9814 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
9816         * rtl.h (struct rtx_def): Add comment for field jump.
9817         (LRA_SUBREG_P): New macro.
9818         * recog.c (register_operand): Check LRA_SUBREG_P.
9819         * lra.c (lra): Add note at the end of RTL code. Align non-empty
9820         stack frame.
9821         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9822         (lra_final_code_change): Skip subreg change for operators.
9823         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9824         if there are no operand changes.
9825         * lra-constraints.c (curr_insn_set): New.
9826         (match_reload): Set LRA_SUBREG_P.
9827         (emit_spill_move): Ditto.
9828         (check_and_process_move): Use curr_insn_set. Process only single
9829         set insns.  Don't initialize sec_mem_p and change_p.
9830         (simplify_operand_subreg): Use LRA_SUBREG_P.
9831         (reg_in_class_p): New function.
9832         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
9833         of #ifdef.  Add code to remove cycling.
9834         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9835         non-null disp.  Reload inner instead of disp when base and index
9836         are null.  Try to put lo_sum into register.
9837         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9838         (check_and_process_move): Move code for move cost check to
9839         simple_move_p.  Remove equiv_substitution.
9840         (simple_move_p): New function.
9841         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
9842         curr_insn_set.  Call check_and_process_move only for single set
9843         insns.  Use the new function.  Move call of check_and_process_move
9844         after operand equiv substitution and address process.
9846 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9848         PR go/57045
9849         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9850         with nonlocal goto receivers or returns twice calls, ignore
9851         unininitialized values from abnormal edges to nl goto receiver
9852         or returns twice call.
9854 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9856         PR tree-optimization/57051
9857         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9858         and VEC_RSHIFT_EXPR if shift count is a multiple of element
9859         bitsize.
9861 2013-04-26  Richard Biener  <rguenther@suse.de>
9863         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9864         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
9865         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9866         (fixup_loop_arrays_after_move): New function.
9867         (move_sese_region_to_fn): Properly outline the loop tree parts
9868         of the SESE region.
9870 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
9872         * config/i386/i386.md (type, unit): Fix long lines.
9874 2013-04-26  Richard Biener  <rguenther@suse.de>
9876         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9877         (lto-streamer-out.o): Likewise.
9878         * cfgloop.c (init_loops_structure): Export, add struct function
9879         argument and adjust.
9880         (flow_loops_find): Adjust.
9881         * cfgloop.h (enum loop_estimation): Add EST_LAST.
9882         (init_loops_structure): Declare.
9883         * lto-streamer-in.c: Include cfgloop.h.
9884         (input_cfg): Input the loop tree.
9885         * lto-streamer-out.c: Include cfgloop.h.
9886         (output_cfg): Output the loop tree.
9887         (output_struct_function_base): Do not drop PROP_loops.
9889 2013-03-26  Richard Biener  <rguenther@suse.de>
9891         * tree-cfg.c (execute_build_cfg): Build the loop tree.
9892         (pass_build_cfg): Provide PROP_loops.
9893         (move_sese_region_to_fn): Remove loops that are outlined into fn
9894         for now.
9895         * tree-inline.c: Include cfgloop.h.
9896         (initialize_cfun): Do not drop PROP_loops.
9897         (copy_loops): New function.
9898         (copy_cfg_body): Copy loop structure.
9899         (tree_function_versioning): Initialize destination loop tree.
9900         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9901         (pass_parallelize_loops): Do IL verification.
9902         * loop-init.c (loop_optimizer_init): Fixup loops if required.
9903         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9904         the CFG make sure we fixup loops as well.
9905         * tree-ssa-tail-merge.c: Include cfgloop.h.
9906         (replace_block_by): When merging loop latches mark loops for fixup.
9907         * lto-streamer-out.c (output_struct_function_base): Drop
9908         PROP_loops for now.
9909         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9910         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9911         * ipa-split.c: Include cfgloop.h.
9912         (split_function): Add the new return block to the loop tree root.
9913         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9914         whether we have removed the forwarder block.
9915         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9916         * cfgloop.h (place_new_loop): Declare.
9917         * cfgloopmanip.c (place_new_loop): Export.
9918         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9919         (tree-switch-conversion.o): Likewise.
9920         (tree-complex.o): Likewise.
9921         (tree-inline.o): Likewise.
9922         (tree-ssa-tailmerge.o): Likewise.
9923         (ipa-split.o): Likewise.
9924         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9925         (tree-ssa-copy.o): Likewise.
9926         * tree-switch-conversion.c: Include cfgloop.h
9927         (process_switch): If we emit a bit-test cascade, schedule loops
9928         for fixup.
9929         * tree-complex.c: Include cfgloop.h.
9930         (expand_complex_div_wide): Properly add new basic-blocks to loops.
9931         * asan.c: Include cfgloop.h.
9932         (create_cond_insert_point): Properly add new basic-blocks to
9933         loops, schedule loop fixup.
9934         * cfgloop.c (verify_loop_structure): Check that looks are not
9935         marked for fixup.
9936         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9937         to loops.
9938         (expand_omp_for_generic): Likewise.
9939         (expand_omp_sections): Likewise.
9940         (expand_omp_atomic_pipeline): Schedule loops for fixup.
9941         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9942         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9943         is initialized, not when loops are present.
9944         * tree-parloops.c (parallelize_loops): Remove checking here.
9945         * passes.c (init_optimization_passes): Schedule a copy-propagation
9946         pass before complete unrolling of inner loops.
9948 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9950         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9951         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9952         (colorize_init): Add argument to _WIN32 version.
9953         * toplev.c: Include diagnostic-color.h.
9954         (process_options): Default to -fdiagnostics-color=auto if
9955         GCC_COLORS env var is in the environment.
9956         * common.opt (fdiagnostics-color=): Add Var and Init.
9957         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9958         env var is in the environment, the default is auto rather than never.
9960         * diagnostic.h (file_name_as_prefix): Add context argument.
9961         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
9962         the string as locus.
9963         * langhooks.c (lhd_print_error_function): Adjust caller.
9965 2013-04-25  Lawrence Crowl  <crowl@google.com>
9967         * var-tracking.c (shared_hash_def::htab):
9968         Change type to hash_table.  Update dependent calls and types.
9970 2013-04-25  Lawrence Crowl  <crowl@google.com>
9972         * Makefile.in: Update as needed below.
9974         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9975         Move declaration to after the type's method definitons.
9977         * attribs.c (htab_t scoped_attributes::attribute_hash):
9978         Change type to hash_table.  Update dependent calls and types.
9980         * bitmap.c (htab_t bitmap_desc_hash):
9981         Change type to hash_table.  Update dependent calls and types.
9983         * cselib.c (htab_t cselib_hash_table):
9984         Change type to hash_table.  Update dependent calls and types.
9986         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9987         (hash_string_slot_node): Move implementation into lto-streamer.h
9988         struct string_slot_hasher.
9989         (eq_string_slot_node): Likewise.
9991         * data-streamer-out.c: Update output_block::string_hash_table
9992         dependent calls and types.
9994         * dwarf2cfi.c (htab_t trace_index):
9995         Change type to hash_table.  Update dependent calls and types.
9997         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9998         Change type to hash_table.  Update dependent calls and types.
9999         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10000         (htab_t optimize_external_refs::map): Likewise.
10001         (htab_t output_comp_unit::extern_map): Likewise.
10002         (htab_t output_comdat_type_unit::extern_map): Likewise.
10003         (htab_t output_macinfo::macinfo_htab): Likewise.
10004         (htab_t optimize_location_lists::htab): Likewise.
10005         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10007         * except.c (htab_t ehspec_hash_type):
10008         Change type to hash_table.  Update dependent calls and types.
10009         (assign_filter_values::ttypes): Likewise.
10010         (assign_filter_values::ehspec): Likewise.
10011         (sjlj_assign_call_site_values::ar_hash): Likewise.
10012         (convert_to_eh_region_ranges::ar_hash): Likewise.
10014         * gcse.c (htab_t pre_ldst_table):
10015         Change type to hash_table.  Update dependent calls and types.
10017         * ggc-common.c (htab_t saving_htab):
10018         Change type to hash_table.  Update dependent calls and types.
10019         (htab_t loc_hash): Likewise.
10020         (htab_t ptr_hash): Likewise.
10021         (call_count): Rename ggc_call_count.
10022         (call_alloc): Rename ggc_call_alloc.
10023         (loc_descriptor): Rename make_loc_descriptor.
10024         (add_statistics): Rename ggc_add_statistics.
10026         * ggc-common.c (saving_htab):
10027         Change type to hash_table.  Update dependent calls and types.
10029         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10030         (push_gimplify_context): Likewise.
10031         (pop_gimplify_context): Likewise.
10032         (struct gimple_temp_hash_elt): Added.
10033         (struct gimplify_hasher): Likewise.
10034         (struct gimplify_ctx.temp_htab):
10035         Change type to hash_table.  Update dependent calls and types.
10037         * gimple-fold.c: Include gimplify-ctx.h.
10039         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10040         Change type to hash_table.  Update dependent calls and types.
10041         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10042         avoid potential global name collision.
10044         * gimplify.c: Include gimplify-ctx.h.
10045         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10046         (htab_t gimplify_ctx::temp_htab):
10047         Update dependent calls and types for new type hash_table.
10048         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10049         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10051         * gimplify-ctx.h: New.
10052         (struct gimple_temp_hash_elt): Move from gimplify.c.
10053         (class gimplify_hasher): New.
10054         (struct gimplify_ctx): Move from gimple.h.
10055         (htab_t gimplify_ctx::temp_htab):
10056         Change type to hash_table.  Update dependent calls and types.
10058         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10059         (htab_t ivs_params::newivs_index):
10060         Change type to hash_table.  Update dependent calls and types.
10061         (htab_t ivs_params::params_index): Likewise.
10062         (htab_t print_generated_program::params_index): Likewise.
10063         (htab_t gloog::newivs_index): Likewise.
10064         (htab_t gloog::params_index): Likewise.
10066         * graphite.c: Include graphite-htab.h.
10067         4htab_t graphite_transform_loops::bb_pbb_mapping):
10068         Change type to hash_table.  Update dependent calls and types.
10070         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10071         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10072         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10074         * graphite-dependences.c: Include graphite-htab.h.
10075         (loop_is_parallel_p): Change hash table type of parameter.
10077         * graphite-htab.h: New.
10078         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10079         (extern find_pbb_via_hash): Move from graphite-poly.h.
10080         (extern loop_is_parallel_p): Move from graphite-poly.h.
10081         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10083         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10084         (extern loop_is_parallel_p): Move to graphite-htab.h.
10085         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10087         * haifa-sched.c (htab_t delay_htab):
10088         Change type to hash_table.  Update dependent calls and types.
10089         (htab_t delay_htab_i2): Likewise.
10091         * ira-color.c (htab_t allocno_hard_regs_htab):
10092         Change type to hash_table.  Update dependent calls and types.
10094         * ira-costs.c (htab_t cost_classes_htab):
10095         Change type to hash_table.  Update dependent calls and types.
10097         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10098         Change type to hash_table.  Update dependent calls and types.
10100         * loop-iv.c (htab_t bivs):
10101         Change type to hash_table.  Update dependent calls and types.
10103         * loop-unroll.c (htab_t opt_info::insns_to_split):
10104         Change type to hash_table.  Update dependent calls and types.
10105         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10107         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10108         (struct string_slot_hasher): New.
10109         (htab_t output_block::string_hash_table):
10110         Change type to hash_table.  Update dependent calls and types.
10112         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10113         (htab_t file_name_hash_table):
10114         Change type to hash_table.  Update dependent calls and types.
10116         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10117         calls and types.
10119         * lto-streamer.c (htab_t tree_htab):
10120         Change type to hash_table.  Update dependent calls and types.
10122         * omp-low.c: Include gimplify-ctx.h.
10124         * passes.c (htab_t name_to_pass_map):
10125         Change type to hash_table.  Update dependent calls and types.
10126         (pass_traverse): Rename to passes_pass_traverse.
10128         * plugin.c (htab_t event_tab):
10129         Change type to hash_table.  Update dependent calls and types.
10131         * postreload-gcse.c (htab_t expr_table):
10132         Change type to hash_table.  Update dependent calls and types.
10133         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10135         * sese.c (debug_rename_map_1): Make extern.
10136         (htab_t copy_bb_and_scalar_dependences::rename_map):
10137         Change type to hash_table.  Update dependent calls and types.
10139         * sese.h (extern debug_rename_map): Move to .c file.
10141         * store-motion.c (htab_t store_motion_mems_table):
10142         Change type to hash_table.  Update dependent calls and types.
10144         * trans-mem.c (htab_t tm_new_mem_hash):
10145         Change type to hash_table.  Update dependent calls and types.
10147         * tree-browser.c (htab_t TB_up_ht):
10148         Change type to hash_table.  Update dependent calls and types.
10150         * tree-cfg.c (htab_t discriminator_per_locus):
10151         Change type to hash_table.  Update dependent calls and types.
10153         * tree-complex.c: Include tree-hasher.h
10154         (htab_t complex_variable_components):
10155         Change type to hash_table.  Update dependent calls and types.
10157         * tree-eh.c (htab_t finally_tree):
10158         Change type to hash_table.  Update dependent calls and types.
10160         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10161         struct int_tree_hasher.
10162         (extern int_tree_map_eq): Likewise.
10163         (uid_decl_map_hash): Removed.
10164         (extern decl_tree_map_eq): Likewise.
10166         * tree-hasher.h: New.
10167         (struct int_tree_hasher): New.
10168         (typedef int_tree_htab_type): New.
10170         * tree-inline.c: Include gimplify-ctx.h.
10172         * tree-mudflap.c: Include gimplify-ctx.h.
10174         * tree-parloops.c: Include tree-hasher.h.
10175         (htab_t eliminate_local_variables_stmt::decl_address):
10176         Change type to hash_table.  Update dependent calls and types.
10177         (htab_t separate_decls_in_region::decl_copies): Likewise.
10179         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10180         Change type to hash_table.  Update dependent calls and types.
10182         * tree-sra.c (candidates):
10183         Change type to hash_table.  Update dependent calls and types.
10185         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10186         in tree-flow.h.
10187         (int_tree_map_hash): Likewise.
10189         * tree-ssa-dom.c (htab_t avail_exprs):
10190         Change type to hash_table.  Update dependent calls and types.
10192         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10193         Change type to hash_table.  Update dependent calls and types.
10195         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10196         Change type to hash_table.  Update dependent calls and types.
10198         * tree-ssa-phiopt.c (seen_ssa_names):
10199         Change type to hash_table.  Update dependent calls and types.
10201         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10202         Change type to hash_table.  Update dependent calls and types.
10204         * tree-ssa-uncprop.c (equiv):
10205         Change type to hash_table.  Update dependent calls and types.
10207 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10209         PR rtl-optimization/57003
10210         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10211         call note_stores with kill_clobbered_value callback again after
10212         killing regs_invalidated_by_call.
10214 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10216         * config/aarch64/aarch64-simd.md
10217         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10218         (aarch64_simd_bsl<mode>): Likewise.
10219         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10221 2013-04-25  Marek Polacek  <polacek@redhat.com>
10223         PR tree-optimization/57066
10224         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10226 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10228         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10230 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10232         * config/aarch64/aarch64-builtins.c
10233         (aarch64_fold_builtin): New.
10234         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10235         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10236         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10237         * config/aarch64/arm_neon.h
10238         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10240 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10241             Tejas Belagod  <tejas.belagod@arm.com>
10243         * config/aarch64/aarch64-builtins.c
10244         (aarch64_gimple_fold_builtin): New.
10245         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10246         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10247         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10248         (addvv4sf): Update.
10249         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10251 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10253         * config/aarch64/aarch64.md
10254         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10256 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10258         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10259         (*ngcsi_uxtw): New pattern.
10261 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10262             Julian Brown  <julian@codesourcery.com>
10264         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10265         (TB_DREG): Add T_V4HF.
10266         (v4hf_UP): New macro.
10267         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10268         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10269         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10270         built-ins.
10271         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10272         (arm_vector_mode_supported_p): Handle V4HF.
10273         (arm_mangle_map): Handle V4HFmode.
10274         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10275         * config/arm/arm_neon_builtins.def: Add entries for
10276         vcvtv4hfv4sf, vcvtv4sfv4hf.
10277         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10278         (neon_vcvtv4hfv4sf): Likewise.
10279         * config/arm/neon-gen.ml: Handle half-precision floating point
10280         features.
10281         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10282         * config/arm/arm_neon.h: Regenerate.
10283         * config/arm/neon.ml (type elts): Add F16.
10284         (type vectype): Add T_float16x4, T_floatHF.
10285         (type vecmode): Add V4HF.
10286         (type features): Add Requires_FP_bit feature.
10287         (elt_width): Handle F16.
10288         (elt_class): Likewise.
10289         (elt_of_class_width): Likewise.
10290         (mode_of_elt): Refactor.
10291         (type_for_elt): Handle F16, fix error messages.
10292         (vectype_size): Handle T_float16x4.
10293         (vcvt_sh): New function.
10294         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10295         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10296         (string_of_mode): Handle V4HF.
10297         * doc/arm-neon-intrinsics.texi: Regenerate.
10299 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10301         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10302         format specifier in 'X' case.
10304 2013-04-25  Alan Modra  <amodra@gmail.com>
10306         PR target/57052
10307         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10308         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10309         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10310         Repeat for many other rotate/shift and mask patterns using subregs.
10311         Name lshiftrt insns.
10312         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10313         on WORDS_BIG_ENDIAN.
10315 2013-04-25  Alan Modra  <amodra@gmail.com>
10317         * config.gcc: Support little-endian powerpc-linux targets.
10318         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10319         (LINK_OS_LINUX_SPEC): Define.
10320         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10321         Preserve MASK_LITTLE_ENDIAN.
10322         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10323         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10324         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10325         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10326         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10327         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10328         Correct fp word order for little-endian.  Don't shift toc entries
10329         smaller than a word for little-endian.
10330         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10331         (bswapdi2 splits): Correct low-part subreg for little-endian.
10332         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10333         low/high where such is correct only for be.
10334         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10335         little-endian for -mcall-aixdesc.
10337 2013-04-25  Alan Modra  <amodra@gmail.com>
10339         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10340         replace_equiv_address_nv.
10342 2013-04-25  Alan Modra  <amodra@gmail.com>
10344         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10346 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10348         Revert:
10349         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10350         * rtl.h (struct rtx_def): ...
10352 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10354         PR rtl-optimizations/57046
10355         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10356         for multi-reg splits.
10358 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10360         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10362 2013-04-24  Sterling Augustine  <saugustine@google.com>
10364         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10365         (comp_dir_string, debug_str_dwo_section): New.
10366         (DEBUG_STR_DWO_SECTION): Rename to ...
10367         (DEBUG_DWO_STR_SECTION): ... this.
10368         (DEBUG_NORM_STR_SECTION): Delete.
10369         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10370         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10371         (find_AT_string): Move most logic to ...
10372         (find_AT_string_in_table): ... here.  New.
10373         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10374         add_skeleton_AT_string.  Delete logic.
10375         (output_skeleton_debug_sections): Remove call to
10376         add_top_level_skeleton_die_attrs.
10377         (add_comp_dir_attribute): Move logic to comp_dir_string.
10378         (dwarf2out_init): Initialize debug_str_dwo_section.
10379         (output_indirect_string): Call find_string_form.
10380         (output_indirect_strings): Rewrite.
10381         (prune_unused_types): Empty skeleton_debug_str_hash.
10382         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10383         (dwarf2out_finish):  Call output_indirect_strings.
10385 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10387         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10389 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10391         * rtl.h (struct rtx_def): Add comment for field jump.
10392         (LRA_SUBREG_P): New macro.
10393         * recog.c (register_operand): Check LRA_SUBREG_P.
10394         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10395         stack frame.
10396         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10397         (lra_final_code_change): Skip subreg change for operators.
10398         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10399         if there are no operand changes.
10400         * lra-constraints.c (curr_insn_set): New.
10401         (match_reload): Set LRA_SUBREG_P.
10402         (emit_spill_move): Ditto.
10403         (check_and_process_move): Use curr_insn_set. Process only single
10404         set insns.  Don't initialize sec_mem_p and change_p.
10405         (simplify_operand_subreg): Use LRA_SUBREG_P.
10406         (reg_in_class_p): New function.
10407         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10408         of #ifdef.  Add code to remove cycling.
10409         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10410         non-null disp.  Reload inner instead of disp when base and index
10411         are null.  Try to put lo_sum into register.
10412         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10413         (check_and_process_move): Move code for move cost check to
10414         simple_move_p.  Remove equiv_substitution.
10415         (simple_move_p): New function.
10416         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10417         curr_insn_set.  Call check_and_process_move only for single set
10418         insns.  Use the new function.  Move call of check_and_process_move
10419         after operand equiv substitution and address process.
10421 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10423         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10424         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10425         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10427 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10429         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10431 2013-04-24  Marek Polacek  <polacek@redhat.com>
10433         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10434         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10435         (select_loops_exit_conditions): Likewise.
10436         (number_of_iterations_for_all_loops): Likewise.
10437         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10438         (scev_analysis): Likewise.
10440 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10441             Chao-ying Fu <fu@mips.com>
10443         * config/mips/micromips.md (jraddiusp): New pattern.
10444         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10445         instruction if possible.
10447 2013-04-24  Alan Modra  <amodra@gmail.com>
10449         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10451 2013-04-24  Julian Brown  <julian@codesourcery.com>
10452             Chung-Lin Tang  <cltang@codesourcery.com>
10454         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10455         dependency behavior in enumeration type DIE generation. Add TODO note
10456         to comments about future DW_FORM_sdata/udata re-work of related code.
10458 2013-04-23  Lawrence Crowl  <crowl@google.com>
10460         * Makefile.in: Update as needed below.
10462         * hash-table.h (class hash_table):
10463         Correct many methods with parameter types compare_type to the correct
10464         value_type.  (Correct code was unlikely to notice the change.)
10465         (hash_table::elements_with_deleted) New.
10466         (class hashtable::iterator): New.
10467         (hashtable::begin()): New.
10468         (hashtable::end()): New.
10469         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10471         * statistics.c (statistics_hashes):
10472         Change type to hash_table.  Update dependent calls and types.
10474         * tree-into-ssa.c (var_infos):
10475         Change type to hash_table.  Update dependent calls and types.
10477         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10478         Change type to hash_table.  Update dependent calls and types.
10480         * tree-ssa-loop-im.c (struct mem_ref.refs):
10481         Change type to hash_table.  Update dependent calls and types.
10483         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10484         Change type to hash_table.  Update dependent calls and types.
10486         * tree-ssa-sccvn.c (vn_tables_s::nary):
10487         Change type to hash_table.  Update dependent calls and types.
10488         (vn_tables_s::phis): Likewise.
10489         (vn_tables_s::references): Likewise.
10491         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10492         (vn_reference_eq): Update parameter and return types.
10494         * tree-ssa-structalias.c (pointer_equiv_class_table):
10495         Change type to hash_table.  Update dependent calls and types.
10496         (location_equiv_class_table): Likewise.
10498         * tree-vect-data-refs.c: Consequential changes for making
10499         peeling a hash_table.
10501         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10502         (destroy_loop_vec_info): Dependent hash_table update.
10504         * tree-vectorizer.h (peeling_htab):
10505         Change type to hash_table.  Update dependent calls and types.
10507 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10509         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10510         to check the register content is equal or not.
10511         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10512         to assign register content record.
10513         * lra-eliminations.c (update_reg_eliminate): Use
10514         lra_update_reg_val_offset to update register content offset.
10515         * lra-int.h (struct lra_reg): Add offset member.
10516         (lra_reg_val_equal_p): New static inline function.
10517         (lra_update_reg_val_offset): New static inline function.
10518         (lra_assign_reg_val): New static inline function.
10519         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10520         to assign register content record.
10521         (initialize_lra_reg_info_element): Initial offset to zero.
10523 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10525         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10526         operands.  Record compression.
10528 2013-04-23  Xinliang David Li  <davidxl@google.com>
10530         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10532 2013-04-23  Richard Biener  <rguenther@suse.de>
10534         PR middle-end/57036
10535         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10536         parameter, only add abnormal goto edges from the copied body
10537         if the call could perform abnormal gotos.
10538         (copy_cfg_body): Adjust.
10540 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10542         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10544 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10546         * coretypes.h (gimple_stmt_iterator): Add struct to make
10547         compatible with C.
10549 2013-04-23  Richard Biener  <rguenther@suse.de>
10551         PR tree-optimization/57026
10552         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10553         from SSA names occuring in abnormal PHI nodes.
10555 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10557         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10559 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10561         * lto/lto.c (print_lto_report_1): Declare early.
10562         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10564 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10566         * common.opt (-flto-report-wpa): Add.
10567         * doc/invoke.texi (-flto-report-wpa): Add.
10568         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10569         (lto_main): dito.
10571 2013-04-22  Xinliang David Li  <davidxl@google.com>
10573         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10574         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10575         * Makefile.in: New dependency
10577             David Daney <ddaney.cavm@gmail.com>
10579         * configure.ac (gcc_cv_as_micromips_support):  Use the
10580         --fatal-warnings option.
10581         * configure: Regenerate.
10583 2013-04-22  Marek Polacek  <polacek@redhat.com>
10585         PR sanitizer/56990
10586         * tsan.c (instrument_expr): Don't instrument expression
10587         in case its size is zero.
10589 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10591         PR target/57032
10592         Revert:
10593         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10595         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10597 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10599         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10600         (gimple_stmt_iterator): New typedef.
10601         * gimple.h (gimple_stmt_iterator): Rename to...
10602         (gimple_stmt_iterator_d): ... This.
10603         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10604         trees be valid for GIMPLE and GENERIC.
10605         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10606         * gimple-fold.c (gimple_fold_call): Call target hook
10607         gimple_fold_builtin.
10608         * hooks.c (hook_bool_gsiptr_false): New.
10609         * hooks.h (hook_bool_gsiptr_false): New.
10610         * target.def (fold_stmt): New.
10611         * doc/tm.texi: Regenerate.
10613 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10615         PR target/57018
10616         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10617         a set sp if no stack realignment.
10619 2013-04-22  Nick Clifton  <nickc@redhat.com>
10621         * config.gcc (tilegx-linux): Extend extra_objs rather than
10622         overwriting it.
10623         (tilepro-linux): Likewise.
10625 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10627         * config/aarch64/aarch64-builtins.c
10628         (CF): Remove.
10629         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10630         (VAR<1-12>): Add MAP parameter.
10631         (BUILTIN_*): Likewise.
10632         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10633         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10634         (aarch64_ushl_n<mode>): Likewise.
10635         (aarch64_sshr_n<mode>): Likewise.
10636         (aarch64_ushr_n<mode>): Likewise.
10637         (aarch64_<maxmin><mode>): Likewise.
10638         (aarch64_sqrt<mode>): Likewise.
10639         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10640         (vshr<q>_n_*): Likewise.
10642 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10644         * config/aarch64/aarch64-builtins.c
10645         (aarch64_simd_builtin_type_mode): Handle SF types.
10646         (sf_UP): Define.
10647         (BUILTIN_GPF): Define.
10648         (aarch64_init_simd_builtins): Handle SF types.
10649         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10650         (frecps): Likewise.
10651         (frecpx): Likewise.
10652         * config/aarch64/aarch64-simd.md
10653         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10654         (aarch64_frecpe<mode>): New.
10655         (aarch64_frecps<mode>): Likewise.
10656         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10657         (v8type): Add frecp<esx>.
10658         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10659         (aarch64_frecps<mode>): Likewise.
10660         * config/aarch64/iterators.md (FRECP): New.
10661         (frecp_suffix): Likewise.
10662         * config/aarch64/arm_neon.h
10663         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10665 2013-04-22  Christian Bruel  <christian.bruel@st.com>
10667         PR target/56995
10668         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10669         (REG_CLASS_NAMES): Idem.
10670         (REG_CLASS_CONTENTS): Idem.
10671         (REGCLASS_HAS_FP_REG): Idem.
10672         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10673         (sh_conditional_register_usage): Idem.
10675 2013-04-21  Jeff Law  <law@redhat.com>
10677         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10678         (ssa_forward_propagate_and_combine): Use it.
10680 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10682         * lra.c: Update the flow chart diagram.
10684 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10686         PR rtl-optimization/56847
10687         * lra-constraints.c (process_alt_operands): Discourage alternative
10688         with non-matche doffsettable memory constraint fro memory with
10689         known offset.
10691 2013-04-19  Richard Biener  <rguenther@suse.de>
10693         PR tree-optimization/56982
10694         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10695         function.
10696         * gimplify.c (gimplify_call_expr): Notice special calls.
10697         (gimplify_modify_expr): Likewise.
10698         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10699         abnormal control flow receivers.
10700         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10701         in the same way as cfun->has_nonlocal_labels.
10702         (gimple_purge_dead_abnormal_call_edges): Likewise.
10703         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10704         receivers start a basic-block.
10706 2013-04-19  Richard Biener  <rguenther@suse.de>
10708         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10709         member ...
10710         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
10711         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10712         (SLP_TREE_LOAD_PERMUTATION): Add.
10713         (vect_transform_slp_perm_load): Adjust prototype.
10714         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10715         (vect_free_slp_instance): Likewise.
10716         (vect_create_new_slp_node): Likewise.
10717         (vect_supported_slp_permutation_p): Remove.
10718         (vect_slp_rearrange_stmts): Adjust.
10719         (vect_supported_load_permutation_p): Likewise.  Inline
10720         vect_supported_slp_permutation_p here.
10721         (vect_analyze_slp_instance): Compute load permutations per
10722         slp node instead of per instance.
10723         (vect_get_slp_defs): Adjust.
10724         (vect_transform_slp_perm_load): Likewise.
10725         (vect_schedule_slp_instance): Remove redundant code.
10726         (vect_schedule_slp): Remove hack for PR56270, add it ...
10727         * tree-vect-stmts.c (vectorizable_load): ... here, do not
10728         CSE loads for SLP.  Adjust.
10730 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10732         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10733         spelling in two comments.
10735 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10737         PR target/56797
10738         * config/arm/arm.c (load_multiple_sequence): Require SP
10739         as base register for loads if SP is in the register list.
10741 2013-04-19  Martin Jambor  <mjambor@suse.cz>
10743         PR tree-optimization/56718
10744         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10745         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10746         and made public.  Adjusted all callers.
10747         (ipa_intraprocedural_devirtualization): New function.
10748         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10749         (ipa_intraprocedural_devirtualization): Likewise.
10750         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10752 2013-04-19  Richard Biener  <rguenther@suse.de>
10754         PR tree-optimization/57000
10755         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10757 2013-04-19  Terry Guo  <terry.guo@arm.com>
10759         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10760         Replace with ...
10761         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
10762         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10763         (cortex_m4_fmacs): Use new reservations.
10764         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10766 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
10768         PR rtl-optimization/56999
10769         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10770         related code.
10771         (lra_coalesce): Remove split_origin_bitmap and related code.
10772         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
10773         ranges if necessary.
10775 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
10777         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10778         New array.
10779         (ix86_expand_call): Remove clobbered_registers array and use
10780         x86_64_ms_sysv_extra_clobbered_registers instead.
10781         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10782         Declare here.
10783         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10784         predicate.
10785         * config/i386/i386.md (*call_rex64_ms_sysv): Use
10786         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
10787         (*call_value_rex64_ms_sysv): Ditto.
10789 2013-04-18  Cary Coutant  <ccoutant@google.com>
10791         * dwarf2out.c (output_pubnames): Check die_perennial_p of
10792         parent instead of die_mark.
10794 2013-04-18  Diego Novillo  <dnovillo@google.com>
10796         * gimple.c (create_gimple_tmp): New.
10797         (get_expr_type): New.
10798         (build_assign): New.
10799         (build_type_cast): New.
10800         * gimple.h (enum ssa_mode): Define.
10801         (gimple_seq_set_location): New.
10802         * asan.c (build_check_stmt): Change some gimple_build_* calls
10803         to use build_assign and build_type_cast.
10805 2013-04-18  Richard Biener  <rguenther@suse.de>
10807         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10808         handle negative step.  Remove redundant checks.
10809         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10810         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10811         for negative step and grouped loads fail to vectorize.
10813 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10815         * emit-rtl.c (reset_insn_used_flags): New function.
10816         (reset_all_used_flags): Use it.
10817         (verify_insn_sharing): New function.
10818         (verify_rtl_sharing): Fix verification for SEQUENCEs.
10820 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
10822         PR tree-optimization/56984
10823         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10824         and (x >> M) >= N don't register any assertion if N << M is the
10825         minimum value.
10827 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10829         * lower-subreg.c (resolve_simple_move): If called self-recursive,
10830         do not delete_insn insns that have not yet been emitted, only
10831         unlink them with remove_insn.
10832         * df-scan.c (df_insn_delete): Revert r197492.
10834 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
10836         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10837         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10839 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10841         * config/arm/arm.md (movsicc_insn): Convert define_insn into
10842         define_insn_and_split.
10843         (and_scc,ior_scc,negscc): Likewise.
10844         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10846 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10848         * config/arm/arm.c (use_return_insn): Return 0 for targets that
10849         can benefit from using a sequence of LDRD instructions in epilogue
10850         instead of a single LDM instruction.
10852 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
10854         PR 45688
10855         * doc/extend.texi: Fix typo.
10857 2013-04-17  Richard Biener  <rguenther@suse.de>
10859         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10860         (vect_build_slp_tree): ... here.
10861         (vect_build_slp_tree_1): Compute which stmts of the SLP group
10862         match.  Remove special-casing of mismatched complex loads.
10863         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10864         re-try the match with swapped commutative operands.
10865         (vect_supported_load_permutation_p): Remove special-casing of
10866         mismatched complex loads.
10867         (vect_analyze_slp_instance): Adjust.
10869 2013-04-17  Richard Biener  <rguenther@suse.de>
10871         PR rtl-optimization/56921
10872         * cfgloop.h (struct loop): Add simple_loop_desc member.
10873         (struct niter_desc): Mark with GTY(()).
10874         (simple_loop_desc): Do not use aux field but simple_loop_desc.
10875         * loop-iv.c (get_simple_loop_desc): Likewise.
10876         (free_simple_loop_desc): Likewise.
10878         Revert
10879         2013-04-16  Richard Biener  <rguenther@suse.de>
10881         PR rtl-optimization/56921
10882         * loop-init.c (pass_rtl_move_loop_invariants): Add
10883         TODO_do_not_ggc_collect to todo_flags_finish.
10884         (pass_rtl_unswitch): Same.
10885         (pass_rtl_unroll_and_peel_loops): Same.
10886         (pass_rtl_doloop): Same.
10888 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
10890         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10891         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10892         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10893         references.
10894         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10895         * tree-streamer.c (record_common_node): Adjust reference in comment.
10897 2013-04-17  Terry Guo  <terry.guo@arm.com>
10899         * config/arm/cortex-m4.md: Add a new bypass.
10901 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10903         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10904         New pattern.
10905         (*subs_<optab><mode>_multp2): New pattern.
10906         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10907         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10909 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10911         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10912         (*subs_mul_imm_<mode>): New pattern.
10914 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
10916         PR target/56948
10917         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10918         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
10919         (vsx_movti_32bit): Change j->wa to O->wa.
10921 2013-04-16  Richard Biener  <rguenther@suse.de>
10923         PR rtl-optimization/56921
10924         * loop-init.c (pass_rtl_move_loop_invariants): Add
10925         TODO_do_not_ggc_collect to todo_flags_finish.
10926         (pass_rtl_unswitch): Same.
10927         (pass_rtl_unroll_and_peel_loops): Same.
10928         (pass_rtl_doloop): Same.
10930 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
10932         * config/arm/arm.c (emit_multi_reg_push): New declaration
10933         for an existing function.
10934         (arm_emit_strd_push): New function.
10935         (arm_expand_prologue): Used here.
10936         (arm_emit_ldrd_pop): New function.
10937         (arm_expand_epilogue): Used here.
10938         (arm_get_frame_offsets): Update condition.
10939         (arm_emit_multi_reg_pop): Add a special case for load of a single
10940         register with writeback.
10942 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
10944         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10945         description.
10947 2013-04-16  Richard Biener  <rguenther@suse.de>
10949         PR tree-optimization/56756
10950         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10951         (first_mem_ref_loc): New.
10952         (execute_sm): Place the load temporarily before a previous
10953         access instead of in the latch edge to ensure its SSA dependencies
10954         are defined at points dominating the load.
10956 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
10958         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10959         correct fix by moving header and footer insn to the footer of
10960         the merged basic block.  Clear BB_END of the merged-away block.
10962         PR middle-end/43631
10963         * emit-rtl.c (make_note_raw): New function.
10964         (link_insn_into_chain): New static inline function.
10965         (add_insn): Use it.
10966         (add_insn_before, add_insn_after): Factor insn chain linking code...
10967         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10968         using link_insn_into_chain.
10969         (note_outside_basic_block_p): New helper function for emit_note_after
10970         and emit_note_before.
10971         (emit_note_after): Use nobb variant of add_insn_after if the note
10972         should not be contained in a basic block.
10973         (emit_note_before): Use nobb variant of add_insn_before if the note
10974         should not be contained in a basic block.
10975         (emit_note_copy): Use make_note_raw.
10976         (emit_note): Likewise.
10977         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10978         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10979         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10980         the moved barrier the tail of the basic block it follows.
10981         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10983 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
10985         PR tree-optimization/56962
10986         * gimple-ssa-strength-reduction.c (record_increment): Only set
10987         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10988         either rhs1 or rhs2 is equal to c->base_expr.
10990 2013-04-15  Richard Biener  <rguenther@suse.de>
10992         PR tree-optimization/56933
10993         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10994         member.
10995         (GROUP_READ_WRITE_DEPENDENCE): Remove.
10996         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10997         * tree-vect-data-refs.c (vect_analyze_group_access): Move
10998         dependence check ...
10999         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11000         ... here.
11001         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11002         GROUP_READ_WRITE_DEPENDENCE.
11004 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11006         * emit-rtl.c (reset_all_used_flags): New function.
11007         (verify_rtl_sharing): Call reset_all_used_flags before and after
11008         performing the checks.
11010 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11012         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11013         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11014         * config/arm/constraints.md (De): New constraint.
11015         * config/arm/neon.md (anddi3_neon): Delete.
11016         (neon_vand<mode>): Expand to standard anddi3 pattern.
11017         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11018         Move earlier in the file.
11019         (neon_inv_logic_op2): Likewise.
11020         (arm_anddi_operand_neon): New predicate.
11022 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11024         * configure.ac (gcc_cv_ld_as_needed): Set
11025         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11026         Use -z ignore, -z record on *-*-solaris2*.
11027         (HAVE_LD_AS_NEEDED): Update comment.
11028         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11029         * configure: Regenerate.
11030         * config.in: Regenerate.
11031         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11032         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11033         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11034         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11035         equivalents.  Fix markup.
11036         * doc/tm.texi: Regenerate.
11038 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11040         * config/i386/i386.opt: New option mstack-protector-guard=.
11041         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11042         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11043         TARGET_SSP_TLS_GUARD.
11044         * config/i386/i386.c (ix86_option_override_internal): Set
11045         ix86_stack_protector_guard.
11046         * config/i386/i386.md (stack_protect_set): Enable for
11047         TARGET_SSP_TLS_GUARD only.
11048         (stack_protect_set_<mode>): Ditto.
11049         (stack_protect_test): Ditto.
11050         (stack_protect_test_<mode>): Ditto.
11051         * doc/invoke.texi (i386 Option): Document.
11053 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11055         PR target/56890
11056         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11057         (S_MODES): Set H_MODE bit.
11058         (SF_MODES): Set only S_MODE and SF_MODE bits.
11059         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11060         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11061         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11062         <MODE_FLOAT>: Likewise.
11064 2013-04-15  Joey Ye  <joey.ye@arm.com>
11066         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11068 2013-04-15  Joey Ye  <joey.ye@arm.com>
11070         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11071         for real far jump.
11072         (thumb_far_jump_used_p): Count instruction size and set
11073         far_jump_used.
11075 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11077         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11078         * resource.c (mark_target_live_regs): Reformat conditional block.
11080 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11082         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11083         notes, they are emitted only just before final.
11084         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11086 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11088         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11089         * cfgrtl.c (delete_insn): Call it here instead.
11090         * lra-spills.c (lra_final_code_change): Use delete_insn.
11091         * haifa-sched.c (sched_remove_insn): Likewise.
11092         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11093         returning to the nop pool.
11094         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11095         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11097 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11099         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11100         * doc/tm.texi: Regenerated.
11102 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11104         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11105         QImode checks.
11107 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11109         * df-core.c (df_find_def): Compare register numbers.
11110         (df_find_use): Likewise.
11112 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11114         PR target/56903
11115         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11116         lra_in_progress for return.
11118 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11120         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11121         define_insn into define_insn_and_split and emit movsicc patterns.
11123 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11125         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11127 2013-04-12  Richard Biener  <rguenther@suse.de>
11129         * tree-pass.h (TODO_do_not_ggc_collect): New.
11130         * passes.c (execute_one_ipa_transform_pass): Honor
11131         TODO_do_not_ggc_collect.
11132         (execute_one_pass): Likewise.
11134         Revert
11135         2013-04-10  Richard Biener  <rguenther@suse.de>
11137         * passes.c (init_optimization_passes): Remove reload pass.
11138         * ira.c (do_reload): Merge into ...
11139         (ira): ... this.
11140         (rest_of_handle_reload): Remove.
11141         (pass_reload): Likewise.
11142         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11143         of reload for vzeroupper pass placement.
11145 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11147         PR tree-optimization/56918
11148         PR tree-optimization/56920
11149         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11150         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11151         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11152         use wide_mul_with_sign method.
11154 2013-04-12  Richard Biener  <rguenther@suse.de>
11156         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11157         not be considered a gimple constant.
11159 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11161         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11162         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11163         operations.
11165 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11166             Jakub Jelinek  <jakub@redhat.com>
11168         * opts.c: Include diagnostic-color.h.
11169         (common_handle_option): Handle OPT_fdiagnostics_color_.
11170         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11171         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11172         (diagnostic-color.o): New.
11173         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11174         (diagnostic_color_rule): New enum.
11175         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11176         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11177         the location string.
11178         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11179         either NULL, or color kind.
11180         * diagnostic-color.c: New file.
11181         * diagnostic-color.h: New file.
11182         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11183         arguments.
11184         * doc/invoke.texi (-fdiagnostics-color): Document.
11185         * pretty-print.h (pp_show_color): Define.
11186         (struct pretty_print_info): Add show_color field.
11187         * diagnostic.c: Include diagnostic-color.h.
11188         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11189         macros.  Colorize error:, warning: etc. strings and also the location
11190         string.
11191         (diagnostic_show_locus): Colorize the caret line.
11192         * pretty-print.c: Include diagnostic-color.h.
11193         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11194         inside of %< %> quotes or quoted through q format modifier.
11196 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11198         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11200 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11202         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11203         code in CC_NZ mode.
11204         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11205         pattern.
11207 2013-04-11  Marek Polacek  <polacek@redhat.com>
11209         PR tree-optimization/48184
11210         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11212 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11214         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11215         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11216         (skip_simple_arithmetic): Tidy up.
11217         * tree.h (skip_simple_constant_arithmetic): Declare.
11219 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11221         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11223 2013-04-11  Richard Biener  <rguenther@suse.de>
11225         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11226         generate vector constants.
11228 2013-04-11  Richard Biener  <rguenther@suse.de>
11230         PR tree-optimization/56878
11231         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11232         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11233         New function.
11234         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11235         Prefer to align the DR with the most invariant base address.
11237 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11239         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11240         comment.
11242 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11244         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11245         floating-point vector comparisons against 0.
11247 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11249         PR tree-optimization/56899
11250         * fold-const.c (extract_muldiv_1): Apply distributive law
11251         only if TYPE_OVERFLOW_WRAPS (ctype).
11253 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11255         PR target/56124
11256         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11257         loading has side effect.
11259 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11261         * config/sparc/sparc.c: Include tree-pass.h.
11262         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11263         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11264         head of file.  Change return type.  Split off gate function.
11265         (sparc_gate_work_around_errata): New function.
11266         (pass_work_around_errata): New pass definition.
11267         (insert_pass_work_around_errata) New pass insert definition to
11268         insert pass_work_around_errata just after delayed-branch scheduling.
11269         (sparc_option_override): Insert the pass.
11270         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11272 2013-04-10  David S. Miller  <davem@davemloft.net>
11274         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11275         or -mcpu=hypersparc.
11277         * target.def (cstore_mode): New hook.
11278         * target.h: Include insn-codes.h
11279         * targhooks.c: Likewise.
11280         (default_cstore_mode): New function.
11281         * targhooks.h: Declare it.
11282         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11283         * doc/tm.texi: Rebuild.
11284         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11285         target hook, rather than inspecting the insn_data.
11286         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11287         (TARGET_CSTORE_MODE): Redefine.
11288         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11289         result patterns.
11290         * config/sparc/predicates.md (cstore_result_operand): New special
11291         predicate.
11292         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11293         Use it for operand 0.
11294         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11295         (*snesi_special): Likewise.
11296         (*snesi_zero): Likewise.
11297         (*seqsi_zero): Likewise.
11298         (*sltu_insn): Likewise.
11299         (*sgeu_insn): Likewise.
11300         (*seqdi_special): Make operand 0 and comparison operation be of
11301         DImode.
11302         (*snedi_special): Likewise.
11303         (*snedi_special_vis3): Likewise.
11304         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11305         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11306         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11307         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11308         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11309         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11310         (*sltu_extend_sp64): Likewise.
11311         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11312         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11313         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11314         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11315         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11317 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11319         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11320         (aarch64_start_file): Use the new function.
11322 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11323             Jason Merrill  <jason@redhat.com>
11325         * common.opt: Add -gdwarf.
11326         * opts.c (common_handle_option): Handle it.
11327         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11329 2013-04-10  Richard Biener  <rguenther@suse.de>
11331         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11332         (execute_one_ipa_transform_pass): But unconditionally here.
11333         (execute_one_pass): And here.
11334         (init_optimization_passes): Remove reload pass.
11335         * tree-pass.h (TODO_ggc_collect): Remove.
11336         (pass_reload): Likewise.
11337         * ira.c (do_reload): Merge into ...
11338         (ira): ... this.
11339         (rest_of_handle_reload): Remove.
11340         (pass_reload): Likewise.
11341         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11342         of reload for vzeroupper pass placement.
11343         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11344         and todo_flags_finish of all passes.
11346 2013-04-10  Richard Biener  <rguenther@suse.de>
11348         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11349         first_const_oprnd field, rename first_def_type to first_op_type.
11350         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11351         (vect_get_and_check_slp_defs): Always use the type of the
11352         operand.  Allow mixed vect_external_def, vect_constant_def types.
11353         (vect_get_constant_vectors): Handle mixed vect_external_def,
11354         vect_constant_def types.
11356 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11358         PR tree-optimization/55524
11359         * tree-ssa-math-opts.c
11360         (convert_mult_to_fma): Don't use an fms construct
11361         when we don't have an fms operation, but fnma, and it looks
11362         likely that we'll be able to use the latter.
11364 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11366         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11367         function.
11368         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11369         inline fail caused by overwritable functions.
11371 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11373         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11374         unnecessary bits in the constant power of two case.
11376 2013-04-10  Richard Biener  <rguenther@suse.de>
11378         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11379         broken code swapping operands.
11380         (vect_build_slp_tree): Do not compute load permutations here.
11381         (vect_analyze_slp_instance): Compute load permutations here,
11382         after building the SLP tree.
11384 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11386         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11387         of next/prev_real_insn.
11389 2013-04-09  Jan Hubicka  <jh@suse.cz>
11391         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11392         Drop aliased parameter.
11393         (function_and_variable_visibility): Do not handle alias pairs.
11394         * cgraph.c (varpool_externally_visible_p): Update prototype.
11395         * varpool.c (varpool_add_new_variable): Update.
11397 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11399         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11401 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11403         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11405         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11407 2013-04-09  Marek Polacek  <polacek@redhat.com>
11409         PR tree-optimization/48762
11410         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11412 2013-04-09  Richard Biener  <rguenther@suse.de>
11414         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11415         dealing with cost.
11416         (vect_build_slp_tree): Likewise.
11417         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11418         calculating the cost of a SLP instance.
11419         (vect_analyze_slp_instance): Use it from here, after building
11420         the SLP tree.
11422 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11424         PR middle-end/56883
11425         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11426         expand_omp_for_static_chunk): Use simple_p = true in
11427         force_gimple_operand_gsi calls when assigning to addressable decls.
11429 2013-04-09  Jeff Law  <law@redhat.com>
11431         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11432         when the boolean was created by converting a wider object which
11433         had a boolean range.
11435 2013-04-09  Richard Biener  <rguenther@suse.de>
11437         * tree-vectorizer.h (slp_void_p): Remove.
11438         (slp_tree): Typedef before _slp_tree declaration.
11439         (struct _slp_tree): Use a vector of slp_tree as children.
11440         (vect_get_place_in_interleaving_chain): Remove.
11441         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11442         Move ...
11443         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11444         and make static.
11445         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11446         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11447         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11448         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11449         Use slp_node instead of slp_void_p and adjust.
11451 2013-04-09  Richard Biener  <rguenther@suse.de>
11453         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11454         work that is not necessary.
11456 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11458         PR tree-optimization/56854
11459         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11460         forward into clobber stmts if it would change MEM_REF lhs into
11461         non-MEM_REF.
11463 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11465         * tree.c (type_hash_lookup, type_hash_add): Make static.
11466         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11468 2013-04-09  Richard Biener  <rguenther@suse.de>
11470         * tree.h (unsave_expr_now): Remove.
11471         * tree-inline.c (mark_local_for_remap_r): Remove.
11472         (unsave_expr_1): Likewise.
11473         (unsave_r): Likewise.
11474         (unsave_expr_now): Likewise.
11475         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11476         (propagate_tree_value): Likewise.
11478 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11480         * doc/rtl.texi (sequence): Rewrite documentation to match the
11481         current use of SEQUENCE rtl objects.
11482         * rtl.def (SEQUENCE): Likewise.
11484         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11485         Update documentation.
11486         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11487         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11489         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11491 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11493         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11494         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11495         methods.
11496         (estimate_edge_size_and_time): Add comment to suggest using rounding
11497         methods.
11498         (estimate_node_size_and_time): Ditto.
11499         (remap_edge_change_prob): Use helper rounding divide methods.
11500         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11501         (gimple_mod_pow2_value_transform): Ditto.
11502         (gimple_mod_subtract_transform): Ditto.
11503         (gimple_ic_transform): Ditto.
11504         (gimple_stringops_transform): Ditto.
11505         * stmt.c (conditional_probability): Ditto.
11506         (emit_case_dispatch_table): Ditto.
11507         * lto-cgraph.c (merge_profile_summaries): Ditto.
11508         * tree-optimize.c (execute_fixup_cfg): Ditto.
11509         * cfgcleanup.c (try_forward_edges): Ditto.
11510         * cfgloopmanip.c (scale_loop_profile): Ditto.
11511         (loopify): Ditto.
11512         (duplicate_loop_to_header_edge): Ditto.
11513         (lv_adjust_loop_entry_edge): Ditto.
11514         * tree-vect-loop.c (vect_transform_loop): Ditto.
11515         * profile.c (compute_branch_probabilities): Ditto.
11516         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11517         * lto-streamer-in.c (input_cfg): Ditto.
11518         * gimple-streamer-in.c (input_bb): Ditto.
11519         * ipa-cp.c (update_profiling_info): Ditto.
11520         (update_specialized_profile): Ditto.
11521         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11522         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11523         rounding methods.
11524         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11525         (compute_trg_info): Ditto.
11526         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11527         (purge_dead_edges): Ditto.
11528         * loop-unswitch.c (unswitch_loop): Ditto.
11529         * cgraphclones.c (cgraph_clone_edge): Ditto.
11530         (cgraph_clone_node): Ditto.
11531         * tree-inline.c (copy_bb): Ditto.
11532         (copy_edges_for_bb): Ditto.
11533         (initialize_cfun): Ditto.
11534         (copy_cfg_body): Ditto.
11535         (expand_call_inline): Ditto.
11537 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11539         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11540         TARGET_CYGWIN64 by TARGET_64BIT.
11542 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11544         * config/epiphany/epiphany.md (GPR_1): New constant.
11545         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11546         * config/epiphany/epiphany.c (gen_compare_reg):
11547         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11548         is already in place.
11549         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11550         Don't require being called during rtl expansion; If y operlaps r0,
11551         return 0.
11552         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11553         (epiphany_expand_epilogue): Likewise.
11555         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11556         Don't use CC_FPmode for ORDERED / UNORDERED.
11557         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11559         * config/epiphany/constraints.md (CnL): New constraint.
11560         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11561         * config/epiphany/predicates.md (add_operand): Allow 1024.
11563         * config/epiphany/epiphany.md (logical_op): New code iterator.
11564         (op_mnc): New code attribute.
11565         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11566         (mov_f+1, mov_f+2): New peephole2 patterns.
11568         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11569         (cstoresi4): Also allow re-use of zero result when doing a NE
11570         comparison to a non-zero operand.
11571         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11573         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11574         Use gen_addsi3_i / gen_subsi3_i.
11576 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11578         PR c++/34949
11579         PR c++/50243
11580         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11581         contain anything but clobbers, at most one __builtin_stack_restore,
11582         optionally debug stmts and final resx, and if it has at least one
11583         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11584         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11585         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11586         which isn't defaut definition, remove them.
11587         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11588         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11589         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11590         with MEM_REF LHS with SSA_NAME address.
11592 2013-04-08  Jeff Law  <law@redhat.com>
11594         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11596 2013-04-08  Richard Biener  <rguenther@suse.de>
11598         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11599         extra newline.
11600         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11601         determined vector type.
11602         (vect_analyze_data_refs): Likewise.
11603         (vect_get_new_vect_var): Adjust.
11604         (vect_create_destination_var): Preserve SSA name versions.
11605         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11606         not dump anything here.
11608 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11610         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11611         Add member lr_slot_known.
11612         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11613         if necessary.
11614         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11615         Remove code that sets lr_slot_offset according to what a previous
11616         version of epiphany_emit_save_restore used to do.
11617         (epiphany_emit_save_restore): When doing an lr save or restore,
11618         set/verify lr_slot_known and lr_slot_offset.
11620 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11622         PR target/54338
11623         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11624         in ALL_REGS.
11626 2013-04-08  Richard Biener  <rguenther@suse.de>
11628         * alias.c (find_base_term): Fix thinko in previous change.
11630 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11632         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11633         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11634         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11635         if possible to compute val.
11636         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11637         For QImode integers don't require anything about precision.  Use
11638         const_with_all_bytes_same to find out if the constant doesn't have
11639         repeated bytes in it.
11641 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11643         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11644         within mode size.
11646 2013-04-08  Marek Polacek  <polacek@redhat.com>
11648         PR rtl-optimization/48182
11649         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11650         value to 1.
11652 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11654         PR target/55487
11655         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11656         nuses, make sure we have a label.
11658 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11660         PR target/56843
11661         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11662         (rs6000_emit_swdiv_low_precision): Remove.
11663         (rs6000_emit_swdiv): Rewrite to handle between one and four
11664         iterations of Newton-Raphson generally; modify required number of
11665         iterations for some cases.
11666         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11668 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
11670         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11671         set-but-unused variable.
11673         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11674         basic blocks of released function bodies garbage-collectable.
11676         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11677         (struct rtl_opt_pass): Add TODO_df_finish.
11679         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11681 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11683         * config/arm/constraints.md (q): New constraint.
11684         * config/arm/ldrdstrd.md: New file.
11685         * config/arm/arm.md (ldrdstrd.md) New include.
11686         (arm_movdi): Use "q" instead of "r" constraint
11687         for double-word memory access.
11688         (movdf_soft_insn): Likewise.
11689         * config/arm/vfp.md (movdi_vfp): Likewise.
11690         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11691         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11692         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11693         (mem_ok_for_ldrd_strd): Likewise.
11694         (output_move_double): Update assertion.
11696 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11698         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11700 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11702         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11703         define_insn_and_split.
11704         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11706 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11708         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11709         define_insn_and_split.
11710         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11711         (shiftsi3_compare): New pattern.
11712         (rrx): New pattern.
11713         * config/arm/unspecs.md (UNSPEC_RRX): New.
11715 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11717         * config/arm/arm.md (negdi_extendsidi): New pattern.
11718         (negdi_zero_extendsidi): Likewise.
11720 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11722         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11723         define_insn_and_split.
11724         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11725         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11727 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11729         * config/arm/arm.md (arm_subdi3): Convert define_insn into
11730         define_insn_and_split.
11731         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11732         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11734 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11736         * config/arm/arm.md (subsi3_carryin): New pattern.
11737         (subsi3_carryin_const): Likewise.
11738         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11739         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11741 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11743         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11745 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11747         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
11748         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
11750 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11752         * config/arm/arm.c (arm_expand_builtin): Change fcode
11753         type to unsigned int.
11755 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11757         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11759 2013-04-04  Ian Lance Taylor  <iant@google.com>
11761         * doc/standards.texi (Standards): The Go frontend supports the Go 1
11762         language standard.
11764 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
11766         PR middle-end/56729
11767         * df-scan.c (df_insn_delete): Disable failing assert.
11769 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11771         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11772         New function prototype.
11773         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11774         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11775         (arm_builtin_vectorized_function): New function.
11777 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11779         * config/arm/arm_neon_builtins.def: New file.
11780         * config/arm/arm.c (neon_builtin_data): Move contents to
11781         arm_neon_builtins.def.
11782         (enum arm_builtins): Include neon builtin definitions.
11783         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
11784         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
11786 2013-04-04  Marek Polacek  <polacek@redhat.com>
11788         PR tree-optimization/48186
11789         * predict.c (maybe_hot_frequency_p): Return false if
11790         HOT_BB_FREQUENCY_FRACTION is 0.
11791         (cgraph_maybe_hot_edge_p): Likewise.
11793 2013-04-04  Richard Biener  <rguenther@suse.de>
11795         PR tree-optimization/56826
11796         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11797         more accurately.
11799 2013-04-04  Richard Biener  <rguenther@suse.de>
11801         PR tree-optimization/56213
11802         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
11803         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
11805 2013-04-04  Richard Biener  <rguenther@suse.de>
11807         PR tree-optimization/56837
11808         * tree-loop-distribution.c (classify_partition): For non-zero
11809         values require that the value has the same precision as its
11810         mode to be useful as memset value.
11812 2013-04-03  Nick Clifton  <nickc@redhat.com>
11814         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
11815         (fmssf4): Use fmsf.s on E3V5 architectures.
11816         (fnmasf4): Use fnmaf.s on E3V5 architectures.
11817         (fnmssf4): Use fnmsf.s on E3V5 architectures.
11819 2013-04-03  Jeff Law  <law@redhat.com>
11821         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11822         (lra-eliminations.o): Likewise.
11824 2013-04-03  Teresa Johnson  <tejohnson@google.com>
11826         * gcov-io.c (compute_working_sets): Moved most of body of old
11827         compute_working_sets here from profile.c.
11828         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
11829         (gcov_working_set_t): Moved typedef here from basic-block.h
11830         (compute_working_set): Declare.
11831         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11832         (get_working_sets): Renamed from compute_working_set,
11833         replace most of body with call to new compute_working_sets.
11834         (get_exec_counts): Replace call to compute_working_sets
11835         to get_working_sets.
11836         * profile.h (get_working_sets): Renamed from compute_working_set.
11837         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
11838         to get_working_sets.
11839         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11840         * gcov-dump.c (dump_working_sets): New function.
11842 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
11844         * hwint.c (sext_hwi, zext_hwi): New functions.
11845         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11846         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11847         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11848         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11849         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11850         (sext_hwi, zext_hwi): New functions.
11852 2013-04-03  Jeff Law  <law@redhat.com>
11854         PR tree-optimization/56799
11855         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11856         back test for widening conversion erroneously dropped in prior change.
11858 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11860         PR target/56809
11861         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11862         instead of next_real_insn.
11864 2013-04-03  Marek Polacek  <polacek@redhat.com>
11866         PR sanitizer/55702
11867         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
11869 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11871         PR target/56809
11872         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11873         next_real_insn.
11874         (thumb1_output_casesi): Likewise.
11875         (thumb2_output_casesi): Likewise.
11877 2013-04-03  Richard Biener  <rguenther@suse.de>
11879         PR tree-optimization/56817
11880         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11881         Split out ...
11882         (tree_unroll_loops_completely_1): ... new function to manually
11883         walk the loop tree, properly defering outer loops of unrolled
11884         loops to later iterations.
11886 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11888         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11889         (vectorizable_load): Likewise.
11890         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11891         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11893 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11895         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11896         BIT_FIELD_REF.
11898 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
11900         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11902 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
11904         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11906 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11908         PR tree-optimization/56790
11909         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11910         folding.
11912 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11914         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11915         Handle VEC_MERGE.
11916         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11917         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11918         equal arguments.
11920 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
11922         PR c/19449
11923         * tree.h (force_folding_builtin_constant_p): New decl.
11924         * builtins.c (force_folding_builtin_constant_p): New variable.
11925         (fold_builtin_constant_p): Fold immediately also if
11926         force_folding_builtin_constant_p.
11928 2013-04-03  Richard Biener  <rguenther@suse.de>
11930         PR tree-optimization/56812
11931         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11932         DRs of the same interleaving chain are independent.
11934 2013-04-02  Jason Merrill  <jason@redhat.com>
11936         * gdbinit.in (pbb): Use debug fn.
11938 2013-04-02  Lawrence Crowl  <crowl@google.com>
11940         * sese.h (struct ivtype_map_elt_s): Remove unused.
11941         (extern debug_ivtype_map): Remove unused.
11942         (extern eq_ivtype_map_elts): Remove unused.
11943         * sese.c (debug_ivtype_map): Removed unused.
11944         (debug_ivtype_map_1): Removed unused.
11945         (debug_ivtype_elt): Remove unused.
11946         (eq_ivtype_map_elts): Remove unused.
11949 2013-04-02  Kai Tietz  <ktietz@redhat.com>
11951         PR target/52790
11952         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11953         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
11954         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11955         function.
11956         (legitimize_pe_coff_symbol): Likewise.
11957         (is_imported_p): New helper-function.
11958         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11959         for Windows x64 targets.
11960         (ix86_expand_prologue): Optimize for pe-coff targets.
11961         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11962         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11963         medium/large code-model.
11964         (legitimize_pic_address): Likewise.
11965         (legitimize_tls_address): Likewise.
11966         (ix86_expand_call): Likewise.
11967         (x86_output_mi_thunk): Likewise.
11968         (get_dllimport_decl): Add new beimport argument.
11969         (construct_plt_address): Don't assert for x64 pe-coff targets.
11970         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11971         targets.
11972         (SYMBOL_FLAG_STUBVAR): New macro.
11973         (SYMBOL_REF_STUBVAR_P): Likewise.
11974         * config/i386/winnt.c (stub_list): New structure.
11975         (stub_head): New local variable.
11976         (i386_pe_record_stub): New function.
11977         (i386_pe_file_end): Emit refptr-stubs.
11979 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
11981         PR rtl-optimization/56745
11982         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11983         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11985         PR c++/34949
11986         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11987         and both of them are MEM_REFs, just compare first argument for
11988         equality and attempt to deal even with differing offsets.
11990         PR c++/34949
11991         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11992         of gimple_clobber_p to be MEM_REF.
11993         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11994         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
11995         after gimplification.
11996         * asan.c (get_mem_ref_of_assignment): Don't instrument
11997         gimple_clobber_p stmts.
11998         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11999         gimple_clobber_p stmt if they have MEM_REF lhs and
12000         are dead because of another gimple_clobber_p stmt.
12001         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12002         gimple_clobber_p stmts like debug stmts.
12003         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12004         that refer to unused VAR_DECLs or uninitialized values.
12005         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12006         gimple_clobber_p stmts if they refer to removed parameters.
12007         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12008         formatting.
12010 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12012         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12013         using SWI48 mode attribute.
12015 2013-04-02  Wei Mi  <wmi@google.com>
12017         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12018         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12019         *<rotate_insn><mode>3_mask in i386.md.
12021 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12023         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12025 2013-04-02  Richard Biener  <rguenther@suse.de>
12027         PR tree-optimization/56778
12028         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12029         Runtime alias tests are not supported for gather loads.
12030         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12031         stmts referenced from SSA operands before updating SSA form.
12033 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12034             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12036         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12037         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12038         * config/arm/cortex-a53.md: New file.
12039         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12040         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12041         * config/arm/arm.c (arm_issue_rate): Likewise.
12042         * config/arm/arm-tune.md: Regenerate
12043         * config/arm/arm-tables.opt: Regenerate.
12044         * config/arm/arm-cores.def: Add cortex-a53.
12046 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12048         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12049         non-static link.
12051 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12053         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12054         scalar load/store operations using B/H registers.
12055         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12057 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12059         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12060         scalar move.
12061         * config/aarch64/aarch64.c
12062         (aarch64_simd_scalar_immediate_valid_for_move): New.
12063         * config/aarch64/aarch64-protos.h
12064         (aarch64_simd_scalar_immediate_valid_for_move): New.
12065         * config/aarch64/constraints.md (Dh, Dq): New.
12066         * config/aarch64/iterators.md (hq): New.
12068 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12070         * reorg.c (get_branch_condition): Deal with conditional returns.
12071         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12073 2013-04-01  Wei Mi  <wmi@google.com>
12075         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12076         Truncate operand 2 using %b asm operand modifier.
12077         (*<shift_insn><mode>3_mask): Ditto.
12078         (*<rotate_insn><mode>3_mask): Ditto.
12080 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12082         PR middle-end/56798
12083         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12085 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12087         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12088         of next_real_insn.
12089         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12091 2013-03-30  Lawrence Crowl  <crowl@google.com>
12093         * dse.c (clear_alias_sets): Remove never set.
12094         (disqualified_clear_alias_sets): Remove never set.
12095         (clear_alias_mode_pool): Remove never set.
12096         (dse_step0): Remove condition that is never true.
12097         (canon_address): Remove condition that is never true.
12098         (dse_step7): Remove condition that is never true.
12099         (rest_of_handle_dse): Remove condition that is never true.
12100         (rest_of_handle_dse::did_global): Remove never read from above.
12101         (dse_step2_spill): Remove never called from above.
12102         (dse_step5_spill): Remove never called from above.
12104 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12106         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12107         JUMP_TABLE_DATA changes.
12108         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12109         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12110         (Insns) <jump_table_data>: New entry.
12111         * doc/tm.texi: Regenerate.
12113         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12115         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12116         for table jump at the end of a basic block using tablejump_p.
12117         * targhooks.c (default_invalid_within_doloop): Likewise.
12118         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12119         target hook implementation that is identical to the default hook.
12120         (rs6000_invalid_within_doloop): Remove.
12122         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12123         unused variable from tablejump_p call.
12125         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12126         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12127         (INSN_DELETED_P): Likewise.
12128         (emit_jump_table_data): New prototype.
12129         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12130         after 4th as unused.
12131         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12132         * sched-vis.c (print_insn): Likewise.
12133         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12134         insn for compatibility with back ends that use next_active_insn to
12135         identify jump table data.
12136         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12137         (remove_insn): Likewise.
12138         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12139         to be emitted.
12140         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12141         (emit_jump_table_data): New function.
12143         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12144         basic block, a JUMP_TABLE_DATA never is.
12145         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12146         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12147         off from code handling real insns.
12148         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12149         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12150         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12151         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12152         is not a NONDEBUG_INSN_P.
12153         * ira-costs.c (scan_one_insn): Likewise.
12154         * jump.c (mark_all_labels): Likewise.
12155         (mark_jump_label_1): Likewise.
12156         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12157         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12158         a basic block.
12159         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12160         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12161         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12162         JUMP_TABLE_DATA_P insns.
12163         (calculate_elim_costs_all_insns): Likewise.
12164         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12165         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12166         (delete_output_reload): Code style fixups.
12167         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12168         insn flags on this non-insn.
12169         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12170         as scheduling barriers, for pre-change compatibility.
12171         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12172         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12174         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12175         redundant JUMP_TABLE_DATA_P test.
12176         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12177         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12178         (frv_for_each_packet): Likewise.
12179         * config/i386/i386.c (min_insn_size): Likewise.
12180         (ix86_avoid_jump_mispredicts): Likewise.
12181         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12182         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12183         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12184         (mips16_insn_length): Robustify.
12185         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12186         (mips16_split_long_branches): Likewise.
12187         * config/pa/pa.c (pa_combine_instructions): Likewise.
12188         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12189         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12190         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12191         as contributing to pool range lengths.
12192         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12193         Remove redundant JUMP_TABLE_DATA_P test.
12194         (sh_loop_align): Likewise.
12195         (split_branches): Likewise.
12196         (sh_insn_length_adjustment): Likewise.
12197         * config/spu/spu.c (get_branch_target): Likewise.
12199 2013-03-29  Jan Hubicka  <jh@suse.cz>
12201         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12202         gcov streaming; stream hot bb threshold to ltrans.
12203         * predict.c (get_hot_bb_threshold): Break out from ....
12204         (maybe_hot_count_p): ... here.
12205         (set_hot_bb_threshold): New function.
12206         * lto-section-in.c (lto_section_name): Add profile.
12207         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12208         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12209         and data-streamer.h
12210         (histogram_entry): New structure.
12211         (histogram, histogram_pool): New global vars.
12212         (histogram_hash): New structure.
12213         (histogram_hash::hash): New method.
12214         (histogram_hash::equal): Likewise.
12215         (account_time_size): New function.
12216         (cmp_counts): New function.
12217         (dump_histogram): New function.
12218         (ipa_profile_generate_summary): New function.
12219         (ipa_profile_write_summary): New function.
12220         (ipa_profile_read_summary): New function.
12221         (ipa_profile): Decide on threshold.
12222         (pass_ipa_profile): Add ipa_profile_write_summary and
12223         ipa_profile_read_summary.
12224         * Makefile.in (ipa.o): Update dependencies.
12225         * lto-streamer.h (LTO_section_ipa_profile): New section.
12227 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12229         * tree.h (VAR_P): New.
12231 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12233         PR lto/56777
12234         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12236 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12238         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12239         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12240         (control_flow_insn_p): Likewise.
12241         * cfgrtl.c (duplicate_insn_chain): Likewise.
12242         * final.c (get_attr_length_1): Likewise.
12243         (shorten_branches): Likewise.
12244         (final_scan_insn): Likewise.
12245         * function.c (instantiate_virtual_regs): Likewise.
12246         * gcse.c (insert_insn_end_basic_block): Likewise.
12247         * ira-costs.c (scan_one_insn): Likewise.
12248         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12249         * lra.c (check_rtl): Likewise.
12250         * reload1.c (elimination_costs_in_insn): Likewise.
12251         * reorg.c (follow_jumps): Likewise.
12253         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12254         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12255         (thumb_far_jump_used_p): Likewise.
12256         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12257         (workaround_speculation): Likewise.
12258         (add_sched_insns_for_speculation): Likewise.
12259         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12260         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12261         (frv_for_each_packet): Likewise.
12262         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12263         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12264         (final_emit_insn_group_barriers): Likewise.
12265         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12266         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12267         (mips16_insn_length): Likewise.
12268         * config/pa/pa.c (pa_reorg): Likewise.
12269         (pa_combine_instructions): Likewise.
12270         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12271         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12272         (sh_reorg): Likewise.
12273         (split_branches): Likewise.
12274         * config/spu/spu.c (get_branch_target): Likewise.
12276         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12277         JUMP_TABLE_DATA_P.
12279 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12281         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12282         Fix declaration name.
12284 2013-03-28  Lawrence Crowl  <crowl@google.com>
12286         * graphds.h (struct graph.indicies): Remove unused.
12287         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12288         (SCOP_ORIGINAL_PDDRS): Remove unused.
12289         * sese.h (extern insert_loop_close_phis): Removed unused.
12290         (extern insert_guard_phis): Removed unused.
12291         (extern ivtype_map_elt_info): Removed unused.
12292         (new_ivtype_map_elt): Removed unused.
12293         * sese.c (ivtype_map_elt_info): Removed unused.
12295 2013-03-28  Lawrence Crowl  <crowl@google.com>
12297         * Makefile.in: Add several missing include dependences.
12298         (DUMPFILE_H): New.
12299         (test-dump.o): New.  This object is not added to any executable,
12300         but is present for ad-hoc testing.
12301         * bitmap.c
12302         (debug (const bitmap_head_def &)): New.
12303         (debug (const bitmap_head_def *)): New.
12304         * bitmap.h
12305         (extern debug (const bitmap_head_def &)): New.
12306         (extern debug (const bitmap_head_def *)): New.
12307         * cfg.c
12308         (debug (edge_def &)): New.
12309         (debug (edge_def *)): New.
12310         * cfghooks.c
12311         (debug (basic_block_def &)): New.
12312         (debug (basic_block_def *)): New.
12313         * dumpfile.h
12314         (dump_node (const_tree, int, FILE *)): Correct source file.
12315         * dwarf2out.c
12316         (debug (die_struct &)): New.
12317         (debug (die_struct *)): New.
12318         * dwarf2out.h
12319         (extern debug (die_struct &)): New.
12320         (extern debug (die_struct *)): New.
12321         * gimple-pretty-print.c
12322         (debug (gimple_statement_d &)): New.
12323         (debug (gimple_statement_d *)): New.
12324         * gimple-pretty-print.h
12325         (extern debug (gimple_statement_d &)): New.
12326         (extern debug (gimple_statement_d *)): New.
12327         * ira-build.c
12328         (debug (ira_allocno_copy &)): New.
12329         (debug (ira_allocno_copy *)): New.
12330         (debug (ira_allocno &)): New.
12331         (debug (ira_allocno *)): New.
12332         * ira-int.h
12333         (extern debug (ira_allocno_copy &)): New.
12334         (extern debug (ira_allocno_copy *)): New.
12335         (extern debug (ira_allocno &)): New.
12336         (extern debug (ira_allocno *)): New.
12337         * ira-lives.c
12338         (debug (live_range &)): New.
12339         (debug (live_range *)): New.
12340         * lra-int.h
12341         (debug (lra_live_range &)): New.
12342         (debug (lra_live_range *)): New.
12343         * lra-lives.c
12344         (debug (lra_live_range &)): New.
12345         (debug (lra_live_range *)): New.
12346         * omega.c
12347         (debug (omega_pb_d &)): New.
12348         (debug (omega_pb_d *)): New.
12349         * omega.h
12350         (extern debug (omega_pb_d &)): New.
12351         (extern debug (omega_pb_d *)): New.
12352         * print-rtl.c
12353         (debug (const rtx_def &)): New.
12354         (debug (const rtx_def *)): New.
12355         * print-tree.c
12356         (debug_tree (tree): Move within file.
12357         (debug_raw (const tree_node &)): New.
12358         (debug_raw (const tree_node *)): New.
12359         (dump_tree_via_hooks (const tree_node *, int)): New.
12360         (debug (const tree_node &)): New.
12361         (debug (const tree_node *)): New.
12362         (debug_verbose (const tree_node &)): New.
12363         (debug_verbose (const tree_node *)): New.
12364         (debug_head (const tree_node &)): New.
12365         (debug_head (const tree_node *)): New.
12366         (debug_body (const tree_node &)): New.
12367         (debug_body (const tree_node *)): New.
12368         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12369         (debug (vec<tree, va_gc> &)): New.
12370         (debug (vec<tree, va_gc> *)): New.
12371         * rtl.h
12372         (extern debug (const rtx_def &)): New.
12373         (extern debug (const rtx_def *)): New.
12374         * sbitmap.c
12375         (debug_raw (simple_bitmap_def &)): New.
12376         (debug_raw (simple_bitmap_def *)): New.
12377         (debug (simple_bitmap_def &)): New.
12378         (debug (simple_bitmap_def *)): New.
12379         * sbitmap.h
12380         (extern debug (simple_bitmap_def &)): New.
12381         (extern debug (simple_bitmap_def *)): New.
12382         (extern debug_raw (simple_bitmap_def &)): New.
12383         (extern debug_raw (simple_bitmap_def *)): New.
12384         * sel-sched-dump.c
12385         (debug (vinsn_def &)): New.
12386         (debug (vinsn_def *)): New.
12387         (debug_verbose (vinsn_def &)): New.
12388         (debug_verbose (vinsn_def *)): New.
12389         (debug (expr_def &)): New.
12390         (debug (expr_def *)): New.
12391         (debug_verbose (expr_def &)): New.
12392         (debug_verbose (expr_def *)): New.
12393         (debug (vec<rtx> &)): New.
12394         (debug (vec<rtx> *)): New.
12395         * sel-sched-dump.h
12396         (extern debug (vinsn_def &)): New.
12397         (extern debug (vinsn_def *)): New.
12398         (extern debug_verbose (vinsn_def &)): New.
12399         (extern debug_verbose (vinsn_def *)): New.
12400         (extern debug (expr_def &)): New.
12401         (extern debug (expr_def *)): New.
12402         (extern debug_verbose (expr_def &)): New.
12403         (extern debug_verbose (expr_def *)): New.
12404         (extern debug (vec<rtx> &)): New.
12405         (extern debug (vec<rtx> *)): New.
12406         * sel-sched-ir.h
12407         (_list_iter_cond_expr): Make inline instead of static.
12408         * sreal.c
12409         (debug (sreal &)): New.
12410         (debug (sreal *)): New.
12411         * sreal.h
12412         (extern debug (sreal &)): New.
12413         (extern debug (sreal *)): New.
12414         * tree.h
12415         (extern debug_raw (const tree_node &)): New.
12416         (extern debug_raw (const tree_node *)): New.
12417         (extern debug (const tree_node &)): New.
12418         (extern debug (const tree_node *)): New.
12419         (extern debug_verbose (const tree_node &)): New.
12420         (extern debug_verbose (const tree_node *)): New.
12421         (extern debug_head (const tree_node &)): New.
12422         (extern debug_head (const tree_node *)): New.
12423         (extern debug_body (const tree_node &)): New.
12424         (extern debug_body (const tree_node *)): New.
12425         (extern debug (vec<tree, va_gc> &)): New.
12426         (extern debug (vec<tree, va_gc> *)): New.
12427         * tree-cfg.c
12428         (debug (struct loop &)): New.
12429         (debug (struct loop *)): New.
12430         (debug_verbose (struct loop &)): New.
12431         (debug_verbose (struct loop *)): New.
12432         * tree-dump.c: Add header dependence.
12433         * tree-flow.h
12434         (extern debug (struct loop &)): New.
12435         (extern debug (struct loop *)): New.
12436         (extern debug_verbose (struct loop &)): New.
12437         (extern debug_verbose (struct loop *)): New.
12438         * tree-data-ref.c
12439         (debug (data_reference &)): New.
12440         (debug (data_reference *)): New.
12441         (debug (vec<data_reference_p> &)): New.
12442         (debug (vec<data_reference_p> *)): New.
12443         (debug (vec<ddr_p> &)): New.
12444         (debug (vec<ddr_p> *)): New.
12445         * tree-data-ref.h
12446         (extern debug (data_reference &)): New.
12447         (extern debug (data_reference *)): New.
12448         (extern debug (vec<data_reference_p> &)): New.
12449         (extern debug (vec<data_reference_p> *)): New.
12450         (extern debug (vec<ddr_p> &)): New.
12451         (extern debug (vec<ddr_p> *)): New.
12452         * tree-ssa-alias.c
12453         (debug (pt_solution &)): New.
12454         (debug (pt_solution *)): New.
12455         * tree-ssa-alias.h
12456         (extern debug (pt_solution &)): New.
12457         (extern debug (pt_solution *)): New.
12458         * tree-ssa-alias.c
12459         (debug (_var_map &)): New.
12460         (debug (_var_map *)): New.
12461         (debug (tree_live_info_d &)): New.
12462         (debug (tree_live_info_d *)): New.
12463         * tree-ssa-alias.h
12464         (extern debug (_var_map &)): New.
12465         (extern debug (_var_map *)): New.
12466         (extern debug (tree_live_info_d &)): New.
12467         (extern debug (tree_live_info_d *)): New.
12469 2013-03-28  Jan Hubicka  <jh@suse.cz>
12471         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12473 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12475         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12476         record only when desired or required.
12478 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12480         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12481         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12483 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12485         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12486         (*andsi3_compare0_uxtw): New pattern.
12487         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12488         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12490 2013-03-28  Jan Hubicka  <jh@suse.cz>
12492         * data-streamer-in.c (streamer_read_gcov_count): New function.
12493         * gimple-streamer-out.c: Include value-prof.h.
12494         (output_gimple_stmt): Output histogram.
12495         (output_bb): Use streamer_write_gcov_count.
12496         * value-prof.c: Include data-streamer.h
12497         (dump_histogram_value): Add HIST_TYPE_MAX.
12498         (stream_out_histogram_value): New function.
12499         (stream_in_histogram_value): New function.
12500         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12501         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12502         * data-streamer-out.c (streamer_write_gcov_count): New function.
12503         (streamer_write_gcov_count_stream): New function.
12504         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12505         (lto_output_node): Likewise.
12506         (input_node, input_edge): Likewise.
12507         * lto-streamer-out.c (output_cfg): Update streaming.
12508         * lto-streamer-in.c (input_cfg): Likewise.
12509         * data-streamer.h (streamer_write_gcov_count,
12510         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12511         * gimple-streamer-in.c: Include value-prof.h
12512         (input_gimple_stmt): Input histograms.
12513         (input_bb): Update profile streaming.
12515 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12517         * genmodes.c (emit_max_int): New function.
12518         (emit_insn_modes_h): Added call to emit_max_function.
12519         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12520         Added doc.
12521         * machmode.def: Fixed comment.
12523 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12525         * combine.c (try_combine): Removed useless assert.
12526         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12528 2013-03-28  Marek Polacek  <polacek@redhat.com>
12529             Richard Biener  <rguenther@suse.de>
12531         PR tree-optimization/56695
12532         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12533         build signed result of a vector comparison.
12534         * tree-cfg.c (verify_gimple_comparison): Check that a result
12535         of a vector comparison has signed type.
12537 2013-03-28  Richard Biener  <rguenther@suse.de>
12539         PR tree-optimization/37021
12540         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12541         do not restrict gaps between groups.
12542         * tree-vect-stmts.c (vectorizable_load): Properly account for
12543         a gap between groups.
12545 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12547         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12548         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12549         is not enabled.
12551 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12553         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12554         * doc/extend.texi (Named Address Spaces): Ditto.
12555         (Variable Attributes): Ditto.
12557 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12559         * config.build: Add support for cygwin x64 target.
12560         * config.gcc: Likewise.
12561         * config.host: Likewise.
12562         * configure.ac: Likewise
12563         * configure: Regenerated.
12565 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12567         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12568         * config/i386/t-cygwin-w64: New file.
12569         * config/i386/cygwin-w64.h: New file.
12570         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12571         and add support for x64-cygwin target.
12572         (CPP_SPEC): Likewise.
12573         (CXX_WRAP_SPEC_LIST): Undefine before define.
12574         (LIBGCJ_SONAME): Use 15 as version.
12576 2013-03-27  Richard Biener  <rguenther@suse.de>
12578         PR tree-optimization/56716
12579         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12580         dumping for ref nodes.
12582 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12584         PR tree-optimization/55334
12585         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12586         restricted pointers to arrays.
12588 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12590         * Makefile.in (.SUFFIXES): Add .cc.
12591         (.c.o): Apply same recipe for implicit rule .cc.o.
12593 2013-03-27  Richard Biener  <rguenther@suse.de>
12595         PR tree-optimization/37021
12596         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12597         REALPART/IMAGPART_EXPRs around the supported refs.
12598         * tree-ssa-structalias.c (find_func_aliases): Assume that
12599         floating-point values are not used to transfer pointers.
12601 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12603         * target.def (TARGET_HAS_IFUNC_P): New target hook.
12604         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12605         * doc/tm.texi: Regenerate.
12606         * targhooks.h (default_has_ifunc_p): New.
12607         * targhooks.c (default_has_ifunc_p): Ditto.
12608         * config/linux-protos.h: New file.
12609         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12610         this hook for linux which disables support of indirect functions in
12611         android.
12612         * config/linux-android.c: New file.
12613         * config/t-linux-android.c: Ditto.
12614         * config.gcc: Added new object file linux-android.o.
12615         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
12616         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12617         * varasm.c (do_assemble_alias): Likewise.
12618         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12619         doesn't support indirect functions.
12620         * configure: Regenerate.
12622 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12624         PR target/56102
12625         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12626         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12627         mult-word mode.
12629 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12631         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12633 2013-03-27  Terry Guo  <terry.guo@arm.com>
12635         * config/arm/arm-cores.def: Added core cortex-r7.
12636         * config/arm/arm-tune.md: Regenerated.
12637         * config/arm/arm-tables.opt: Regenerated.
12638         * doc/invoke.texi: Added entry for core cortex-r7.
12640 2013-03-27  Walter Lee  <walt@tilera.com>
12642         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12643         double-decrement of next_scratch_regno.
12645 2013-03-27  Walter Lee  <walt@tilera.com>
12647         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12648         input operands.
12649         (insn_v1mulus): Ditto.
12650         (insn_v2muls): Ditto.
12652 2013-03-27  Walter Lee  <walt@tilera.com>
12654         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12655         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12657 2013-03-27  Walter Lee  <walt@tilera.com>
12659         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12660         (*sibcall_value): Ditto.
12662 2013-03-27  Walter Lee  <walt@tilera.com>
12664         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12665         (insn_mnz_v8qi): ... this ...
12666         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
12667         vector equivalent.
12668         (insn_v<n>mnz): Replaced by ...
12669         (insn_v1mnz): ... this ...
12670         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
12671         equivalent.
12672         (insn_mz_<mode>): Replaced by ...
12673         (insn_mz_v8qi): ... this ...
12674         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
12675         vector equivalent.
12676         (insn_v<n>mz): Replaced by ...
12677         (insn_v1mz): ... this ...
12678         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
12679         equivalent.
12681 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
12683         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12685 2013-03-26  Roland McGrath  <mcgrathr@google.com>
12687         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12688         than fprintf with a non-constant, non-format string.
12690 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12692         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12693         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
12694         operand 0 predicate.
12695         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12696         attribute.  Use general_x64nomem_operand as operand 1 predicate.
12697         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12698         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12699         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12700         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12701         (mov<mode>_insv_1): Remove expander.  Merge insn with
12702         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12703         Use general_x64nomem_operand as operand 1 predicate.
12704         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12705         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12706         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12707         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12708         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
12709         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12710         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12711         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12712         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
12713         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12714         (general_x64nomem_operand): Ditto.
12716 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12718         * config/rtems.opt: Add -pthread option.
12720 2013-03-26  Richard Biener  <rguenther@suse.de>
12722         * alias.c (find_base_term): Avoid redundant and not used recursion.
12723         (base_alias_check): Get the initial base term from the caller.
12724         (true_dependence_1): Compute and pass base terms to base_alias_check.
12725         (write_dependence_p): Likewise.
12726         (may_alias_p): Likewise.
12728 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
12730         * config/aarch64/aarch64.c (aarch64_classify_address): Support
12731         PC-relative load in SI modes and above only.
12733 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
12735         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12736         * config/arm/iwmmxt.md (WCGR0): Update.
12737         (WCGR1, WCGR2, WCGR3): Likewise.
12739 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12741         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12742         Use x64 and nox64 isa attributes.
12744 2013-03-26  Richard Biener  <rguenther@suse.de>
12746         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12747         alignment computations and rely on get_object_alignment_1
12748         for the !TYPE_P case.
12749         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12751 2013-03-26  Walter Lee  <walt@tilera.com>
12753         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12754         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12756 2013-03-25  Jeff Law  <law@redhat.com>
12758         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12759         check for INTEGRAL_TYPE_P that was missing due to checking in
12760         wrong version of prior patch.
12762 2013-03-25  Walter Lee  <walt@tilera.com>
12764         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12765         TILEGX_INSN_SHUFFLEBYTES1.
12766         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12767         shufflebytes1.
12768         (tilegx_builtins): Ditto.
12769         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12771 2013-03-25  Walter Lee  <walt@tilera.com>
12773         * config/tilegx/tilegx.md (floatsisf2): New pattern.
12774         (floatunssisf2): New pattern.
12775         (floatsidf2): New pattern.
12776         (floatunssidf2): New pattern.
12778 2013-03-25  Walter Lee  <walt@tilera.com>
12780         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12781         tests for constraint J, K, N, P.
12783 2013-03-25  Walter Lee  <walt@tilera.com>
12785         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12786         Use indirect/pcrel encoding.
12787         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12788         Ditto.
12790 2013-03-25  Steve Ellcey  <sellcey@mips.com>
12792         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12793         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12794         * config/mips/mips.c (mips_option_override): Set IMADD default.
12795         * config/mips/mips.h (PTF_AVOID_IMADD): New.
12796         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12797         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12798         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12799         * doc/invoke.texi (-mimadd/-mno-imadd): New.
12801 2013-03-25  Jeff Law  <law@redhat.com>
12803         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12804         slightly to avoid creating and folding useless trees.  Simplify
12805         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12807 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
12809         * config/i386/i386.md (*zero_extendsidi2): Merge with
12810         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
12811         * config/i386/predicates.md (x86_64_zext_operand): Rename from
12812         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
12813         targets.  Clarify comment.
12815 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12817         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12818         pass-through jump functions differently.
12819         (ipa_read_jump_function): Likewise.  Also use setter functions to set
12820         up jump functions.
12822 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12824         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12825         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12826         process it.
12827         (ipa_get_indirect_edge_target): New function.
12828         (devirtualization_time_bonus): New parameter known_aggs, pass it to
12829         ipa_get_indirect_edge_target.  Update all callers.
12830         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
12831         ipa_get_indirect_edge_target_1 instead of calling
12832         ipa_get_indirect_edge_target.
12833         (create_specialized_node): Pass aggvlas to
12834         ipcp_discover_new_direct_edges.
12836 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12838         * config/arm/arm.md (f_sels, f_seld): New types.
12839         (*cmov<mode>): New pattern.
12840         * config/arm/predicates.md (arm_vsel_comparison_operator): New
12841         predicate.
12843 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12845         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12846         POSIX-printf for mingw-hosted  builds.
12848 2013-03-25  Richard Biener  <rguenther@suse.de>
12850         PR middle-end/56694
12851         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12852         must-not-throw stmt location.
12854 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12856         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12857         Emit load-acquire versions when acq is true.
12858         (arm_emit_store_exclusive): Add rel parameter.
12859         Emit store-release versions when rel is true.
12860         (arm_split_compare_and_swap): Use acquire-release instructions
12861         instead.
12862         of barriers when appropriate.
12863         (arm_split_atomic_op): Likewise.
12864         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12865         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12866         (VUNSPEC_SLX): Likewise.
12867         (VUNSPEC_LDA): Likewise.
12868         (VUNSPEC_STL): Likewise.
12869         * config/arm/sync.md (atomic_load<mode>): New pattern.
12870         (atomic_store<mode>): Likewise.
12871         (arm_load_acquire_exclusive<mode>): Likewise.
12872         (arm_load_acquire_exclusivesi): Likewise.
12873         (arm_load_acquire_exclusivedi): Likewise.
12874         (arm_store_release_exclusive<mode>): Likewise.
12876 2013-03-25  Catherine Moore  <clm@codesourcery.com>
12878         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12879         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12880         * config/mip/predicates.md (lwsp_swsp_operand,
12881         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12882         sb16_operand, db4_operand, db7_operand, ib3_operand,
12883         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12884         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12885         andi16_operand): New predicates.
12886         * config/mips/mips.md (compression): New attribute.
12887         (enabled): New attribute.
12888         (length): Consider compression in computing length.
12889         (shift_compression): New code attribute.
12890         (*add<mode>3): New operands. Record compression.
12891         (sub<mode>3): Likewise.
12892         (one_cmpl<mode>2): Likewise.
12893         (*and<mode>3): Likewise.
12894         (*ior<mode>3): Likewise.
12895         (unnamed pattern for xor): Likewise.
12896         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12897         (*<optab><mode>3): Likewise.
12898         (*mov<mode>_internal: Likewise.
12899         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12900         (mips_unsigned_immediate_p): New.
12901         (umips_lwsp_swsp_address_p): New.
12902         (m16_based_address_p): New.
12903         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12904         (mips_unsigned_immediate_p): New prototype.
12905         (lwsp_swsp_address_p): New prototype.
12906         (m16_based_address_p): New prototype.
12907         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12908         (mips_signed_immediate_p): New function.
12909         (m16_based_address_p): New function.
12910         (lwsp_swsp_address_p): New function.
12911         (mips_print_operand_punctuation): Recognize short delay slot insns
12912         for microMIPS.add<mode>3"
12914 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12916         PR target/56720
12917         * config/arm/iterators.md (v_cmp_result): New mode attribute.
12918         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12920 2013-03-25  Richard Biener  <rguenther@suse.de>
12922         PR tree-optimization/56689
12923         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12924         any edge.
12926 2013-03-25  Richard Biener  <rguenther@suse.de>
12928         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12929         of bitmap.
12930         (memory_references): Likewise.
12931         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12932         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12933         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12934         (gather_mem_refs_in_loops): Fold into ...
12935         (analyze_memory_references): ... this.  Move initialization
12936         to tree_ssa_lim_initialize.
12937         (fill_always_executed_in): Rename to ...
12938         (fill_always_executed_in_1): ... this.
12939         (fill_always_executed_in): Move contains_call computation to
12940         this new function from ...
12941         (tree_ssa_lim_initialize): ... here.
12942         (tree_ssa_lim): Call fill_always_executed_in.
12944 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12946         * postreload.c (reload_combine): Fix code detecting returns.
12948 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12950         * function.c (emit_use_return_register_into_block): On cc0 targets,
12951         do not emit the sequence between cc0 setter and user.
12953 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12955         * config/i386/predicates.md (local_symbolic_operand): Interpret
12956         dll-imported symbols as none-local.
12958 2013-03-25  Richard Biener  <rguenther@suse.de>
12960         * tree-ssa-loop-im.c (struct depend): Remove.
12961         (struct lim_aux_data): Make depends a vec of gimples.
12962         (free_lim_aux_data): Adjust.
12963         (add_dependency): Likewise.
12964         (set_level): Likewise.
12966 2013-03-25  Richard Biener  <rguenther@suse.de>
12968         PR middle-end/56434
12969         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12970         the pointer returned by calls with ECF_MALLOC set.
12972 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12974         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12976 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12978         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
12979         using MMXMODE mode iterator.
12980         (*move<mode>_internal): Merge with *movv2sf_internal and
12981         *movv2sf_internal_rex64 using MMXMODE mode iterator.
12983 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12985         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12986         (record_last_mem_set_info): Likewise.
12988         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12989         of XNEWVEC followed by memset.
12990         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12992 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12994         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12995         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12996         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12997         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12998         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12999         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13000         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13001         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13002         BARRIER_P instead of GET_CODE.
13004 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13006         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13007         inaccuracy in the probing code.
13009         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13010         (ctrapdi4): Likewise.
13012 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13014         * calls.c (expand_call): Add missing guard to code handling return
13015         of non-BLKmode structures in MSB.
13016         * function.c (expand_function_end): Likewise.
13018 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13020         * combine.c (try_combine): Adjust comment.  Do not add the set of
13021         insn #0 if the destination indirectly is set or dies in insn #2.
13022         Tidy up code to distribute a new note.
13024 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13026         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13027         also for alternatives 16 and 17.
13029 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13031         * config/i386/sse.md (*mov<mode>_internal): Merge with
13032         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13033         Emit insn template depending on type attribute.  Use
13034         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13035         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13036         attribute calculation.  Remove unit attribute calculation.
13037         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13038         Set prefix_data16 attribute for DImode ssemov types.
13039         Use Ym instead of y for SSE-MMX conversion alternatives.
13040         Reorder operand constraints.
13042 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13044         * df.h (df_insn_delete): Adjust prototype.
13045         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13046         and let it decide whether mark the basic block dirty.
13047         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13048         * df-scan.c (df_insn_info_delete): New helper function, split
13049         off from df_insn_delete.
13050         (df_scan_free_bb_info): Use it.
13051         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13052         Likewise.
13053         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13054         that the insn is actually an insn and it has a non-NULL basic block.
13055         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13057 2013-03-22  Richard Biener  <rguenther@suse.de>
13059         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13060         dep_ref members.
13061         (mem_ref_alloc): Do not allocate them.
13062         (refs_independent_p): Do not query or maintain a cache.
13064 2013-03-22  Richard Biener  <rguenther@suse.de>
13066         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13067         (gather_mem_refs_in_loops): Do not compute it.
13068         (analyze_memory_references): Do not allocate it.
13069         (tree_ssa_lim_finalize): Do not free it.
13070         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13072 2013-03-22  Richard Biener  <rguenther@suse.de>
13074         * is-a.h (as_a): Use gcc_checking_assert.
13076 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13078         * config/aarch64/aarch64.c (aarch64_print_operand): New
13079         format specifier for printing a constant in hex.
13080         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13081         format specifier for printing second operand.
13083 2013-03-22  Richard Biener  <rguenther@suse.de>
13085         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13086         bitmaps.
13087         (gather_mem_refs_in_loops): Perform store accumulation here.
13088         (create_vop_ref_mapping_loop): Remove.
13089         (create_vop_ref_mapping): Likewise.
13090         (analyze_memory_references): Initialize refs_stored_in_loop.
13091         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13092         (record_indep_loop): Remove.
13093         (record_dep_loop): New function.
13094         (ref_indep_loop_p_1): Adjust to only walk over references
13095         in the loop, not its subloops.
13096         (ref_indep_loop_p): Rename to ...
13097         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13098         maintaining a more fine-grained cache.
13099         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13100         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13102 2013-03-22  Richard Biener  <rguenther@suse.de>
13104         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13105         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13106         aggregate mem_ref_loc.
13107         (free_mem_ref_locs): Inline into ...
13108         (memref_free): ... this and adjust.
13109         (mem_ref_alloc): Adjust.
13110         (mem_ref_locs_alloc): Remove.
13111         (record_mem_ref_loc): Adjust.
13112         (get_all_locs_in_loop): Rewrite into ...
13113         (for_all_locs_in_loop): ... this iterator.
13114         (rewrite_mem_ref_loc): New functor.
13115         (rewrite_mem_refs): Use for_all_locs_in_loop.
13116         (sm_set_flag_if_changed): New functor.
13117         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13118         (ref_always_accessed): New functor.
13119         (ref_always_accessed_p): Use for_all_locs_in_loop.
13121 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13123         * tree-pass.h (PROP_gimple_lvec): New.
13124         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13125         (init_optimization_passes): Move pass_lower_vector.
13126         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13127         PROP_gimple_lvec.
13128         (pass_lower_vector): Provide PROP_gimple_lvec.
13129         (pass_lower_vector_ssa): Likewise.
13130         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13132 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13134         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13136 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13138         * config/i386/i386.md (*movdi_internal): Disparage slightly
13139         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13140         conversion alternatives.
13142 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13144         PR middle-end/48087
13145         * diagnostic.def (DK_WERROR): New kind.
13146         * diagnostic.h (werrorcount): Define.
13147         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13148         promoted to DK_ERROR, increment DK_WERROR counter instead of
13149         DK_ERROR counter.
13150         * toplev.c (toplev_main): Call print_ignored_options even if
13151         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13152         even if just werrorcount is non-zero.
13154         PR debug/55608
13155         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13156         on failure.
13157         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13158         (string_cst_pool_decl): New function.
13159         (optimize_one_addr_into_implicit_ptr): New function.
13160         (resolve_addr_in_expr): Optimize DWARF location expression
13161         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13162         which doesn't live in memory, but has DW_AT_location or
13163         DW_AT_const_value, or refers to a string literal, into
13164         DW_OP_GNU_implicit_pointer.
13165         (optimize_location_into_implicit_ptr): New function.
13166         (resolve_addr): If removing DW_AT_location of a variable because
13167         it was DW_OP_addr of address of the variable, but the variable doesn't
13168         live in memory, try to emit const value attribute for the initializer.
13170 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13172         * tree.h (VECTOR_TYPE_P): New macro.
13173         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13174         TYPE_MODE): Use it.
13175         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13176         VEC_COND_EXPR cannot be lvalues.
13177         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13179 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13181         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13182         Restrict the transformation to equal modes.
13184 2013-03-21  Richard Biener  <rguenther@suse.de>
13186         PR tree-optimization/39326
13187         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13188         (MEM_ANALYZABLE): Adjust.
13189         (record_mem_ref_loc): Move bitmap ops ...
13190         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13191         unanalyzable refs, do not record locations for it.
13192         (analyze_memory_references): Allocate ref zero as shared
13193         unanalyzable ref.
13194         (refs_independent_p): Do not test for unanalyzed mems here.
13195         (ref_indep_loop_p_1): Special-case disambiguation against
13196         the unanalyzed ref.
13197         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13199 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13201         * config/arm/arm-protos.h (tune_params): Add
13202         prefer_neon_for_64bits field.
13203         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13204         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13205         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13206         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13207         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13208         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13209         (arm_option_override): Handle -mneon-for-64bits new option.
13210         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13211         (prefer_neon_for_64bits): Declare new variable.
13212         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13213         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13214         (arch_enabled): Handle new arch types. Remove support for onlya8
13215         and nota8.
13216         (one_cmpldi2): Use new arch names.
13217         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13218         * config/arm/arm.opt (mneon-for-64bits): Add option.
13219         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13220         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13221         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13222         of onlya8.
13223         * doc/invoke.texi (-mneon-for-64bits): Document.
13225 2013-03-21  Richard Biener  <rguenther@suse.de>
13227         PR tree-optimization/39326
13228         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13229         (sort_bbs_in_loop_postorder_cmp): New function.
13230         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13232 2013-03-21  Richard Biener  <rguenther@suse.de>
13234         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13235         (vect_insert_into_interleaving_chain): Likewise.
13236         (vect_drs_dependent_in_basic_block): Inline ...
13237         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13238         split out from ...
13239         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13240         (vect_check_interleaving): Simplify.
13241         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13242         (vect_slp_analyze_data_ref_dependences): ... this new function.
13243         (dr_group_sort_cmp): New function.
13244         (vect_analyze_data_ref_accesses): Compute data-reference groups
13245         here instead of in vect_analyze_data_ref_dependence.  Use
13246         a more efficient algorithm.
13247         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13248         vect_slp_analyze_data_ref_dependences.  Call
13249         vect_analyze_data_ref_accesses earlier.
13250         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13251         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13252         (vect_slp_analyze_data_ref_dependences): New prototype.
13254 2013-03-21  Richard Biener  <rguenther@suse.de>
13256         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13257         ref is stored in the loop.
13258         (find_refs_for_sm): Walk only over all stores.
13259         (store_motion_loop): Allocate from lim_bitmap_obstack.
13260         (store_motion): Likewise.
13262 2013-03-21  Richard Biener  <rguenther@suse.de>
13264         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13265         Update virtual SSA form.
13267 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13269         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13270         * configure: Regenerate.
13271         * config.in: Regenerate.
13272         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13273         if !HAVE_LD_EH_FRAME_CIEV3.
13275 2013-03-21  Richard Biener  <rguenther@suse.de>
13277         * tree-cfg.c (verify_expr_no_block): New function.
13278         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13279         nor DECL_VALUE_EXPR have locations with associated blocks.
13280         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13281         (clear_unused_block_pointer): Remove code dealing with
13282         blocks in DECL_DEBUG_EXPR locations.
13284 2013-03-21  Richard Biener  <rguenther@suse.de>
13286         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13287         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13288         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13289         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13290         instead of DECL_DEBUG_EXPR_IS_FROM.
13291         * gimplify.c (gimplify_modify_expr): Likewise.
13292         * tree-cfg.c (verify_expr_location_1): Likewise.
13293         * tree-complex.c (create_one_component_var): Likewise.
13294         * tree-sra.c (create_access_replacement): Likewise.
13295         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13296         (clear_unused_block_pointer): Likewise.
13297         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13298         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13299         * var-tracking.c (var_debug_decl): Likewise.
13300         (track_expr_p): Likewise.
13301         * tree-inline.c (add_local_variables): Likewise.  Set
13302         DECL_HAS_DEBUG_EXPR_P after copying it.
13303         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13304         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13306 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13308         PR bootstrap/56656
13309         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13310         * configure: Regenerate.
13311         * config.in: Regenerate.
13312         * config/i386/i386.md (*movdf_internal): Use
13313         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13314         movd instead of movq mnemonic for interunit moves.
13315         (*movdi_internal): Ditto.
13317 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13319         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13320         (abd<mode>_3): New pattern.
13321         (aba<mode>_3): New pattern.
13322         (fabd<mode>_3): New pattern.
13324 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13326         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13327         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13328         occurrence of REGISTER_PREFIX as its empty string.
13330 2013-03-20  Jeff Law  <law@redhat.com>
13332         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13333         addititional equivalences for equality comparisons between an SSA_NAME
13334         and a constant where the SSA_NAME was set from a widening conversion.
13336 2013-03-20  Walter Lee  <walt@tilera.com>
13338         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13340 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13342         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13343         depending on type attribute.
13344         (*movti_internal): Ditto.
13345         (*movtf_internal): Ditto.
13346         (*movxf_internal): Ditto.
13347         (*movdf_internal): Ditto.
13348         (*movsf_internal): Ditto.
13350 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13352         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13353         maybe_vex for sselog1 and ssemov types.
13354         (*movdi_internal): Reorder operand constraints.
13355         (*movsi_internal): Ditto.  Set prefix attribute to
13356         maybe_vex for sselog1 and ssemov types.
13357         (*movtf_internal): Set prefix attribute to maybe_vex
13358         for sselog1 and ssemov types.
13359         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13360         DImode ssemov types.  Reorder operand constraints.
13361         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13362         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13363         attribute for SImode ssemov types.  Reorder operand constraints.
13365 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13367         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13368         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13370 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13372         * config/rs6000/predicates.md (indexed_address, update_address_mem
13373         update_indexed_address_mem): New predicates.
13374         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13375         attribute for load/store instructions.
13376         * config/rs6000/dfp.md (movsd_store): Likewise.
13377         (movsd_load): Likewise.
13378         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13379         (unnamed HI->DI extend define_insn): Likewise.
13380         (unnamed SI->DI extend define_insn): Likewise.
13381         (unnamed QI->SI extend define_insn): Likewise.
13382         (unnamed QI->HI extend define_insn): Likewise.
13383         (unnamed HI->SI extend define_insn): Likewise.
13384         (unnamed HI->SI extend define_insn): Likewise.
13385         (extendsfdf2_fpr): Likewise.
13386         (movsi_internal1): Likewise.
13387         (movsi_internal1_single): Likewise.
13388         (movhi_internal): Likewise.
13389         (movqi_internal): Likewise.
13390         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13391         attribute for load/store instructions.
13392         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13393         instructions.
13394         (mov<mode>_softfloat): Likewise.
13395         (mov<mode>_hardfloat32): Likewise.
13396         (mov<mode>_hardfloat64): Likewise.
13397         (mov<mode>_softfloat64): Likewise.
13398         (movdi_internal32): Likewise.
13399         (movdi_internal64): Likewise.
13400         (probe_stack_<mode>): Likewise.
13402 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13404         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13405         floating point, and decimal floating point to reload iterator.
13407         * config/rs6000/constraints.md (wl constraint): New constraints to
13408         return FLOAT_REGS if certain options are used to reduce the number
13409         of separate patterns that exist in the file.
13410         (wx constraint): Likewise.
13411         (wz constraint): Likewise.
13413         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13414         -mdebug=reg, print wg, wl, wx, and wz constraints.
13415         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13416         Initialize the reload functions for 64-bit binary/decimal floating
13417         point types.
13418         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13419         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13420         create the buffer on the stack to overcome not having a 32-bit
13421         load and store.
13422         (rs6000_emit_move): Likewise.
13423         (rs6000_secondary_memory_needed_rtx): Likewise.
13424         (rs6000_alloc_sdmode_stack_slot): Likewise.
13425         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13426         via xxlxor, just like DFmode 0.0.
13428         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13429         define as 1 if we are running on a power7 or newer.
13430         (enum r6000_reg_class_enum): Add new constraints.
13432         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13433         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13434         with other moves by using conditional constraits (wg).  Use LFIWZX
13435         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13436         (movsd splitter): Likewise.
13437         (movsd_hardfloat): Likewise.
13438         (movsd_softfloat): Likewise.
13440         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13441         binary and decimal floating point moves.
13442         (fmove_ok): New attributes to combine binary and decimal floating
13443         point moves, and to combine power6x (mfpgpr) moves along normal
13444         floating moves.
13445         (real_value_to_target): Likewise.
13446         (f32_lr): Likewise.
13447         (f32_lm): Likewise.
13448         (f32_li): Likewise.
13449         (f32_sr): Likewise.
13450         (f32_sm): Likewise.
13451         (f32_si): Likewise.
13452         (movsf): Combine binary and decimal floating point moves.  Combine
13453         power6x (mfpgpr) moves with other moves by using conditional
13454         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13455         (mov<mode> for SFmode/SDmode); Likewise.
13456         (SFmode/SDmode splitters): Likewise.
13457         (movsf_hardfloat): Likewise.
13458         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13459         (movsf_softfloat): Likewise.
13460         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13462         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13463         wx and wz constraints.
13465         * config/rs6000/constraints.md (wg constraint): New constraint to
13466         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13468         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13469         constraint.
13471         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13472         -mdebug=reg, print wg, wl, wx, and wz constraints.
13473         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13474         Initialize the reload functions for 64-bit binary/decimal floating
13475         point types.
13476         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13477         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13478         create the buffer on the stack to overcome not having a 32-bit
13479         load and store.
13480         (rs6000_emit_move): Likewise.
13481         (rs6000_secondary_memory_needed_rtx): Likewise.
13482         (rs6000_alloc_sdmode_stack_slot): Likewise.
13483         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13484         via xxlxor, just like DFmode 0.0.
13486         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13487         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13488         with other moves by using conditional constraits (wg).  Use LFIWZX
13489         and STFIWX for loading SDmode on power7.
13490         (movdd splitters): Likewise.
13491         (movdd_hardfloat32): Likewise.
13492         (movdd_softfloat32): Likewise.
13493         (movdd_hardfloat64_mfpgpr): Likewise.
13494         (movdd_hardfloat64): Likewise.
13495         (movdd_softfloat64): Likewise.
13497         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13498         64-bit binary and decimal floating point moves.
13499         (FMOVE64X): Likewise.
13500         (movdf): Combine 64-bit binary and decimal floating point moves.
13501         Combine power6x (mfpgpr) moves with other moves by using
13502         conditional constraits (wg).
13503         (mov<mode> for DFmode/DDmode): Likewise.
13504         (DFmode/DDmode splitters): Likewise.
13505         (movdf_hardfloat32): Likewise.
13506         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13507         (movdf_softfloat32): Likewise.
13508         (movdf_hardfloat64_mfpgpr): Likewise.
13509         (movdf_hardfloat64): Likewise.
13510         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13511         (movdf_softfloat64): Likewise.
13512         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13513         (reload_<mode>_load): Move to later in the file so they aren't in
13514         the middle of the floating point move insns.
13515         (reload_<mode>_store): Likewise.
13517         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13518         constraint.
13520         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13521         constraint if -mdebug=reg.
13522         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13523         Enable using dd reload support if needed.
13525         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13526         binary and decimal floating point moves in rs6000.md.
13527         (movtd_internal): Likewise.
13529         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13530         decimal floating point moves.
13531         (movtf): Likewise.
13532         (movtf_internal): Likewise.
13533         (mov<mode>_internal, TDmode/TFmode): Likewise.
13534         (movtf_softfloat): Likewise.
13535         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13537         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13538         movdi_internal64, using wg constraint for move direct operations.
13539         (movdi_internal64): Likewise.
13541         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13542         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13543         the various virtual registers. Use GPR/FPR first/last values,
13544         instead of hard coding the register numbers.  Print which modes
13545         have reload functions registered.
13546         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13547         settings before/after setting cpu, target and subtarget settings.
13548         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13549         and for secondary reload failures in rs6000_secondary_reload_inner.
13550         (rs6000_secondary_reload_fail): Likewise.
13551         (rs6000_secondary_reload_inner): Likewise.
13553         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13554         macros for first/last GPR and FPR registers.
13555         (LAST_GPR_REGNO): Likewise.
13556         (FIRST_FPR_REGNO): Likewise.
13557         (LAST_FPR_REGNO): Likewise.
13559         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13560         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13561         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13562         (vcond<mode><mode>): Likewise.
13563         (vcondu<mode><mode>): Likewise.
13564         (vector_gtu<mode>): Likewise.
13565         (vector_gte<mode>): Likewise.
13566         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13567         to prevent the compiler from converting DImode operations to TImode.
13568         (ior<mode>3): Likewise.
13569         (and<mode>3): Likewise.
13570         (one_cmpl<mode>2): Likewise.
13571         (nor<mode>3): Likewise.
13572         (andc<mode>3): Likewise.
13574         * config/rs6000/constraints.md (wt constraint): New constraint
13575         that returns VSX_REGS if TImode is allowed in VSX registers.
13577         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13578         constant under VSX.
13580         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13581         similar to TImode, but it is restricted to being in the GPRs.
13583         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13584         TImode to occupy a single VSX register.
13586         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13587         -mvsx-timode for power7/power8.
13588         (power7 cpu): Likewise.
13589         (power8 cpu): Likewise.
13591         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13592         sure that TFmode/TDmode take up two registers if they are ever
13593         allowed in the upper VSX registers.
13594         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13595         registers.
13596         (rs6000_init_hard_regno_mode_ok): Likewise.
13597         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13598         constraint.  Print if LRA is turned on.
13599         (rs6000_option_override_internal): Give an error if -mvsx-timode
13600         and VSX is not enabled.
13601         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13602         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13603         to reg+offset addressing.  Use PTImode when checking offset
13604         addresses for validity.
13605         (reg_offset_addressing_ok_p): Likewise.
13606         (rs6000_legitimate_offset_address_p): Likewise.
13607         (rs6000_legitimize_address): Likewise.
13608         (rs6000_legitimize_reload_address): Likewise.
13609         (rs6000_legitimate_address_p): Likewise.
13610         (rs6000_eliminate_indexed_memrefs): Likewise.
13611         (rs6000_emit_move): Likewise.
13612         (rs6000_secondary_reload): Likewise.
13613         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13614         reloads to fpr registers to continue to use reg+offset addressing,
13615         but 64-bit reloads to altivec registers need reg+reg addressing.
13616         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13617         it.  Treat LO_SUM like a PLUS operation.
13618         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13619         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13620         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13621         registers to share a register with a smaller sized type, since VSX
13622         puts scalars in the upper 64-bits.
13623         (print_operand): Add support for PTImode.
13624         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13625         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13626         registers, but don't have arithmetic support.
13627         (rs6000_memory_move_cost): Add test for VSX.
13628         (rs6000_opt_masks): Add -mvsx-timode.
13630         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13631         for TImode.
13632         (VSs): Likewise.
13633         (VSr): Use wt constraint for TImode.
13634         (VSv): Drop TImode support.
13635         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13636         (vsx_movti_64bit): Likewise.
13637         (vsx_movti_32bit): Likewise.
13638         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13639         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13640         one '?' on the appropriate output constraint.  Do not allow TImode
13641         logical operations on 32-bit systems.
13642         (vsx_ior<mode>3): Likewise.
13643         (vsx_xor<mode>3): Likewise.
13644         (vsx_one_cmpl<mode>2): Likewise.
13645         (vsx_nor<mode>3): Likewise.
13646         (vsx_andc<mode>3): Likewise.
13647         (vsx_concat_<mode>): Likewise.
13648         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13650         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13651         OPTION_MASK_VSX_TIMODE.
13652         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13653         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13655         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13656         (TI2 iterator): New iterator for TImode, PTImode.
13657         (wd mode attribute): Add values for vector types.
13658         (movti_string): Replace TI move operations with operations for TImode
13659         and PTImode.  Add support for TImode being allowed in VSX registers.
13660         (mov<mode>_string, TImode/PTImode): Likewise.
13661         (movti_ppc64): Likewise.
13662         (mov<mode>_ppc64, TImode/PTImode): Likewise.
13663         (TI mode splitters): Likewise.
13665         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13666         constraint.
13668 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
13670         PR tree-optimization/56355
13671         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13672         Also handle integers with undefined overflow.
13674 2013-03-20  Catherine Moore  <clm@codesourcery.com>
13675             Maciej W. Rozycki  <macro@codesourcery.com>
13676             Tom de Vries  <tom@codesourcery.com>
13677             Nathan Sidwell <nathan@codesourcery.com>
13678             Iain Sandoe  <iain@codesourcery.com>
13679             Nathan Froyd  <froydnj@codesourcery.com>
13680             Chao-ying Fu <fu@mips.com>
13682         * doc/extend.texi: (micromips, nomicromips, nocompression):
13683         Document new function attributes.
13684         * doc/invoke.texi (minterlink-compressed, mmicromips,
13685         m14k, m14ke, m14kec): Document new options.
13686         (minterlink-mips16): Update documentation.
13687         * doc/md.texi (ZC, ZD): Document new constraints.
13688         * configure.ac (gcc_cv_as_micromips): Check if linker
13689         supports the .set micromips directive.
13690         * configure: Regenerate.
13691         * config.in: Regenerate.
13692         * config/mips/mips-tables.opt: Regenerate.
13693         * config/mips/micromips.md: New file.
13694         * constraints.md (ZC, ZD): New constraints.
13695         * config/mips/predicates.md (movep_src_register): New predicate.
13696         (movep_src_operand): New predicate.
13697         (non_volatile_mem_operand): New predicate.
13698         * config/mips/mips.md (multimem): New type.
13699         (length): Differentiate between 17-bit and 18-bit branch offsets.
13700         (MOVEP1, MOVEP2): New mode iterator.
13701         (mov_<load>l): Use ZC constraint.
13702         (mov_<load>r): Likewise.
13703         (mov_<store>l): Likewise.
13704         (mov_<store>r): Likewise.
13705         (*branch_equality<mode>_inverted): Add microMIPS support.
13706         (*branch_equality<mode>): Likewise.
13707         (*jump_absolute): Likewise.
13708         (indirect_jump_<mode>): Likewise.
13709         (tablejump_<mode>): Likewise.
13710         (<optab>_internal): Likewise.
13711         (sibcall_internal): Likewise.
13712         (sibcall_value_internal): Likewise.
13713         (prefetch): Use constraint ZD.
13714         * config/mips/mips.opt (minterlink-compressed): New option.
13715         (minterlink-mips16): Now an alias for minterlink-compressed.
13716         (mmicromips): New option.
13717         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13718         (compare_and_swap_12): Likewise.
13719         (sync_add<mode>): Likewise.
13720         (sync_<optab>_12): Likewise.
13721         (sync_old_<optab>_12): Likewise.
13722         (sync_new_<optab>_12): Likewise.
13723         (sync_nand_12): Likewise.
13724         (sync_old_nand_12): Likewise.
13725         (sync_new_nand_12): Likewise.
13726         (sync_sub<mode>): Likewise.
13727         (sync_old_add<mode>): Likewise.
13728         (sync_old_sub<mode>): Likewise.
13729         (sync_new_add<mode>): Likewise.
13730         (sync_new_sub<mode>): Likewise.
13731         (sync_<optab><mode>): Likewise.
13732         (sync_old_<optab><mode>): Likewise.
13733         (sync_new_<optab><mode>): Likewise.
13734         (sync_nand<mode>): Likewise.
13735         (sync_old_nand<mode>): Likewise.
13736         (sync_new_nand<mode>): Likewise.
13737         (sync_lock_test_and_set<mode>): Likewise.
13738         (test_and_set_12): Likewise.
13739         (atomic_compare_and_swap<mode>): Likewise.
13740         (atomic_exchange<mode>_llsc): Likewise.
13741         (atomic_fetch_add<mode>_llsc): Likewise.
13742         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13743         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13744         (umips_save_restore_pattern_p): Likewise.
13745         (umips_load_store_pair_p): Likewise.
13746         (umips_output_load_store_pair): Likewise.
13747         (umips_movep_target_p): Likewise.
13748         (umips_12bit_offset_address_p): Likewise.
13749         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13750         (mips_base_mips16): Rename this...
13751         (mips_base_compression_flags): ...to this. Update all uses.
13752         (mips_attribute_table): Add micromips, nomicromips and nocompression.
13753         (mips_mips16_decl_p): Delete.
13754         (mips_nomips16_decl_p): Delete.
13755         (mips_get_compress_on_flags): New function.
13756         (mips_get_compress_off_flags): New function.
13757         (mips_get_compress_mode): New function.
13758         (mips_get_compress_on_name): New function.
13759         (mips_get_compress_off_name): New function.
13760         (mips_insert_attributes): Support multiple compression types.
13761         (mips_merge_decl_attributes): Likewise.
13762         (umips_12bit_offset_address_p): New function.
13763         (mips_start_function_definition): Emit .set micromips directive.
13764         (mips_call_may_need_jalx_p): New function.
13765         (mips_function_ok_for_sibcall): Add microMIPS support.
13766         (mips_print_operand_punctuation): Support short delay slots and
13767         compact jumps.
13768         (umips_swm_mask, umips_swm_encoding): New.
13769         (umips_build_save_restore): New function.
13770         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13771         (was_mips16_p): Remove.
13772         (old_compression_mode): New.
13773         (mips_set_compression_mode): New function.
13774         (mips_set_current_function): Add microMIPS support.
13775         (mips_option_override): Likewise.
13776         (umips_save_restore_pattern_p): New function.
13777         (umips_output_save_restore): New function.
13778         (umips_load_store_pair_p_1): New function.
13779         (umips_load_store_pair_p): New function.
13780         (umips_output_load_store_pair_1): New function.
13781         (umips_output_load_store_pair): New function.
13782         (umips_movep_target_p) New function.
13783         (mips_prepare_pch_save): Add microMIPS support.
13784         * config/mips/mips.h (TARGET_COMPRESSION): New.
13785         (TARGET_CPU_CPP_BUILTINS): Update macro
13786         to use new compression flags and to support microMIPS.
13787         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13788         (MIPS_ARCH_FLOAT_SPEC): Likewise.
13789         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13790         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13791         (ASM_SPEC): Support mmicromips and mno-micromips.
13792         (M16STORE_REG_P): New macro.
13793         (MIPS_CALL): Support TARGET_MICROMIPS.
13794         (MICROMIPS_J): New macro.
13795         (mips_base_mips16): Rename this...
13796         (mips_base_compression_flags): ...to this.
13797         (UMIPS_12BIT_OFFSET_P): New macro.
13798         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13799         (MULTILIB_DIRNAMES): Likewise.
13800 2013-03-20  Richard Biener  <rguenther@suse.de>
13802         PR tree-optimization/56661
13803         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13804         the result does not have to be distinct.
13806 2013-03-20  Richard Biener  <rguenther@suse.de>
13808         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13809         remap_gimple_op_r.
13811 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13812             Steven Bosscher <steven@gcc.gnu.org>
13814         PR rtl-optimization/56605
13815         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13817 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13819         PR bootstrap/56656
13820         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13821         that require movd instead of movq.
13823 2013-03-20  Richard Biener  <rguenther@suse.de>
13825         * tree-ssa-structalias.c (struct variable_info): Add pointer
13826         to the first field of an aggregate with sub-vars.  Make
13827         this and the pointer to the next subfield its ID.
13828         (vi_next): New function.
13829         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13830         storedanything_id, integer_id): Increment by one.
13831         (new_var_info, get_call_vi, lookup_call_clobber_vi,
13832         get_call_clobber_vi): Adjust.
13833         (solution_set_expand): Simplify and speedup.
13834         (solution_set_add): Inline into ...
13835         (set_union_with_increment): ... this.  Adjust accordingly.
13836         (do_sd_constraint): Likewise.
13837         (do_ds_constraint): Likewise.
13838         (do_complex_constraint): Simplify.
13839         (build_pred_graph): Adjust.
13840         (solve_graph): Likewise.  Simplify and speedup.
13841         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13842         get_constraint_for_component_ref, get_constraint_for_1,
13843         first_vi_for_offset, first_or_preceding_vi_for_offset,
13844         create_function_info_for, create_variable_info_for_1,
13845         create_variable_info_for, intra_create_variable_infos): Adjust.
13846         (init_base_vars): Push NULL for ID zero.
13847         (compute_points_to_sets): Adjust.
13849 2013-03-20  Richard Biener  <rguenther@suse.de>
13851         * cfgloop.c (verify_loop_structure): Streamline and avoid
13852         ICEing on corrupt loop tree.
13853         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13854         loop tree.
13856 2013-03-20  Richard Biener  <rguenther@suse.de>
13858         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13859         check whether an SSA update is needed.
13861 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
13863         * config/mips/constraints.md (T): Rename to...
13864         (Yf): ...this.
13865         (U): Rename to...
13866         (Yd): ...this.
13867         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13868         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13870 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13872         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13873         (*subsi3_carryin_uxtw): Likewise.
13875 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13877         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13878         (*rorsi3_insn_uxtw): Likewise.
13880 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13882         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13883         (*extrsi5_insn_uxtw): Likewise.
13885 2013-03-19  Richard Biener  <rguenther@suse.de>
13887         PR tree-optimization/56273
13888         * passes.c (init_optimization_passes): Move second VRP after DOM.
13890 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
13892         * config/i386/i386.md (*movti_internal): Merge from
13893         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
13894         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
13895         nox64 isa attributes.
13897 2013-03-18  Richard Biener  <rguenther@suse.de>
13899         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13900         (unite): Likewise.
13901         (merge_node_constraints): Likewise.
13902         (build_succ_graph): Likewise.
13903         (valid_graph_edge): Inline into single caller.
13904         (unify_nodes): Likewise.  Use bitmap_set_bit return value
13905         and cache varinfo.
13906         (scc_visit): Fix formatting and variable use.
13907         (do_sd_constraint): Use gcc_checking_assert.
13908         (do_ds_constraint): Likewise.
13909         (do_complex_constraint): Likewise.
13910         (condense_visit): Likewise.  Cleanup.
13911         (dump_pred_graph): New function.
13912         (perform_var_substitution): Dump the pred-graph before
13913         variable substitution.
13914         (find_equivalent_node): Use gcc_checking_assert.
13915         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13917 2013-03-18  Richard Biener  <rguenther@suse.de>
13919         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13920         Remove cond_expr_stmt_list argument and do not gimplify the
13921         built expression.
13922         (vect_loop_versioning): Adjust.
13923         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13924         Cleanup to use less temporaries.
13925         (vect_create_data_ref_ptr): Cleanup.
13927 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13929         PR tree-optimization/56635
13930         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13931         require types_compatible_p types.
13933 2013-03-18  Nick Clifton  <nickc@redhat.com>
13935         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13936         spurious backslash.
13938         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13939         Add missing line to comment describing function.
13941 2013-03-18  Richard Biener  <rguenther@suse.de>
13943         PR tree-optimization/56210
13944         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13945         Handle string / character search functions.
13946         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13948 2013-03-18  Richard Biener  <rguenther@suse.de>
13950         PR middle-end/56483
13951         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13952         and implement properly.
13953         * gimple.h (gimple_cond_single_var_p): Remove.
13955 2013-03-18  Richard Biener  <rguenther@suse.de>
13957         * tree-data-ref.h (find_data_references_in_loop): Declare.
13958         * tree-data-ref.c (get_references_in_stmt): Use a stack
13959         vector pre-allocated in the callers.
13960         (find_data_references_in_stmt): Adjust.
13961         (graphite_find_data_references_in_stmt): Likewise.
13962         (create_rdg_vertices): Likewise.
13963         (find_data_references_in_loop): Export.
13964         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13965         Compute dependences here...
13966         (vect_analyze_data_refs): ...not here.  When we encounter
13967         a non-vectorizable data reference in basic-block vectorization
13968         truncate the data reference vector.  Do not bother to
13969         fixup data-dependence information for gather loads.
13970         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13971         of data references, as reported.
13973 2013-03-18  Richard Biener  <rguenther@suse.de>
13975         PR tree-optimization/3713
13976         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
13977         has_constants and expr.
13978         (stmt_has_constants): Properly valueize SSA names when deciding
13979         whether the stmt has constants.
13981 2013-03-18  Richard Biener  <rguenther@suse.de>
13983         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13984         whole function when there is nothing to do.
13985         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13986         * tree-vectorizer.c (vectorize_loops): Update virtual and
13987         loop-closed SSA once.
13988         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13990 2013-03-18  Richard Biener  <rguenther@suse.de>
13992         PR middle-end/56113
13993         * domwalk.c (bb_postorder): New global static.
13994         (cmp_bb_postorder): New function.
13995         (walk_dominator_tree): Replace scheme imposing an order for
13996         visiting dominator sons by one sorting them at the time they
13997         are pushed on the stack.
13999 2013-03-18  Richard Biener  <rguenther@suse.de>
14001         PR tree-optimization/39326
14002         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14003         (struct mem_ref): Replace mem member with ao_ref typed member.
14004         (MEM_ANALYZABLE): Adjust.
14005         (memref_eq): Likewise.
14006         (mem_ref_alloc): Likewise.
14007         (gather_mem_refs_stmt): Likewise.
14008         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14009         (execute_sm_if_changed_flag_set): Adjust.
14010         (execute_sm): Likewise.
14011         (ref_always_accessed_p): Likewise.
14012         (refs_independent_p): Likewise.
14013         (can_sm_ref_p): Likewise.
14015 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14017         PR c/56566
14018         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14019         return 1 even for !unsignedp.
14021 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14023         * config/i386/i386.md (isa): Add x64 and nox64.
14024         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14025         (*pushtf): Enable *roF alternative for x64 isa only.
14026         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14027         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14028         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14029         (*movtf_internal): Merge from *movtf_internal_rex64 and
14030         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14031         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14032         nox64 isa attributes.
14033         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14034         nox64 isa attributes.
14035         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14037 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14039         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14041 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14043         PR target/56640
14044         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14045         class names.  Remove trailing comma after "ALL_REGS".
14047 2013-03-16  Jan Hubicka  <jh@suse.cz>
14049         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14050         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14051         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14052         of cgraph_get_create_node.
14053         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14055 2013-03-16  Jason Merrill  <jason@redhat.com>
14057         PR debug/49090
14058         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14059         with DW_AT_default_value.
14061 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14063         * BASE-VER: Set to 4.9.0.
14065 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14067         PR target/56619
14068         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14069         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14070         Document _x* TSX intrinsics.
14072 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14073             David Holsgrove  <david.holsgrove@xilinx.com>
14075         * configure.ac: Add MicroBlaze TLS support detection.
14076         * configure: Regenerate.
14077         * config/microblaze/microblaze-protos.h
14078         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14079         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14080         * config/microblaze/microblaze.c (microblaze_address_type): Add
14081         ADDRESS_TLS and tls_reloc address types.
14082         (microblaze_address_info): Add tls_reloc.
14083         (TARGET_HAVE_TLS): Define.
14084         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14085          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14086          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14087         load_tls_operand,  microblaze_call_tls_get_addr,
14088         microblaze_legitimize_tls_address): New functions.
14089         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14090         (get_base_reg): Use microblaze_tls_symbol_p.
14091         (microblaze_classify_address): Handle TLS.
14092         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14093         label_mentioned_p and microblaze_tls_referenced_p.
14094         (microblaze_legitimize_address): Handle TLS.
14095         (microblaze_address_insns): Handle ADDRESS_TLS.
14096         (pic_address_needs_scratch): Handle TLS.
14097         (print_operand_address): Handle TLS.
14098         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14099         (microblaze_expand_move): Handle TLS.
14100         (microblaze_legitimate_constant_p): Check
14101         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14102         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14103         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14104         (PIC_OFFSET_TABLE_REGNUM): Set.
14105         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14106         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14107         (addsi3, movsi_internal2, movdf_internal): Update constraints
14108         * config/microblaze/predicates.md (arith_plus_operand): Define
14109         (move_operand): Redefine as move_src_operand,
14110         check microblaze_tls_referenced_p.
14112 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14114         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14115         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14117 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14119         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14120         CC mode for AND.
14122 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14124         PR tree-optimization/53265
14125         * common.opt (Waggressive-loop-optimizations): New option.
14126         * tree-ssa-loop-niter.c: Include tree-pass.h.
14127         (do_warn_aggressive_loop_optimizations): New function.
14128         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14129         if number_of_latch_executions returned constant.
14130         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14131         early.  If number_of_latch_executions returned constant, set
14132         nb_iterations_upper_bound back to it.
14133         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14134         field.
14135         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14136         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14138         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14139         (MULTILIB_OSDIRNAMES): Set.
14140         * genmultilib: If defaultosdirname doesn't start with :: , set
14141         defaultosdirname2 instead, clear it and emit two . multilib_raw
14142         entries instead of just one.
14144 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14146         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14147         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14148         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14149         (SUBTARGET_OVERRIDE_OPTIONS): New.
14151 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14153         PR target/49880
14154         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14155         (musermode): Convert to Var(TARGET_USERMODE).
14156         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14157         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14158         * config/sh/sh.c (sh_option_override): Use
14159         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14160         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14161         condition.
14162         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14163         TARGET_SH4.
14164         (udivsi3_i4_single, divsi3_i4_single): Use
14165         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14167 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14169         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14170         default setting.
14172 2013-03-13  Richard Biener  <rguenther@suse.de>
14174         PR tree-optimization/56608
14175         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14176         calls when vectorizing basic-blocks.
14178 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14180         PR plugins/45078
14181         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14182         tm_file.
14184 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14186         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14188 2013-03-11  Jan Hubicka  <jh@suse.cz>
14190         PR lto/56557
14191         * lto-streamer-out.c (output_symbol_p): Skip references from
14192         constructors of external variables.
14194 2013-03-11  Jan Hubicka  <jh@suse.cz>
14196         PR middle-end/56571
14197         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14198         from pseudos.
14199         * emit-rtl.c (verify_rtx_sharing): Likewise.
14200         (copy_insn_1): Likewise.
14201         * rtl.c (copy_rtx): Likewise.
14203 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14205         PR target/56591
14206         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14207         output_operand_lossage message.
14209 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14211         PR target/56470
14212         * arm.c (shift_op): Validate RTL pattern on the fly.
14213         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14214         the RTL.
14216 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14218         PR target/56347
14219         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14220         new call patterns that clobber %fr12.
14221         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14222         split and postreload patterns.
14223         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14224         registers %fr12 and %fr12R as call used.
14226 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14228         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14229         (canon_address, record_store, replace_read, check_mem_read_rtx,
14230         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14231         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14232         rest_of_handle_dse): Likewise.
14234 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14236         PR middle-end/56524
14237         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14238         Add base_optabs.
14239         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14240         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14241         (save_optabs_if_changed): Replace with...
14242         (init_tree_optimization_optabs): ...this.
14243         * optabs.c (save_optabs_if_changed): Rename to...
14244         (init_tree_optimization_optabs): ...this.  Take the optimization node
14245         as argument.  Do nothing if the base optabs are already correct.
14246         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14247         to recompute optabs.
14248         * function.h (function): Remove optabs field.
14249         * function.c (invoke_set_current_function_hook): Call
14250         init_tree_optimization_optabs.  Use the result to initialize
14251         this_fn_optabs.
14253 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14255         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14256         if GTMA_HAS_NO_INSTRUMENTATION.
14257         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14258         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14259         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14260         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14261         GTMA_HAS_NO_INSTRUMENTATION.
14263 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14265         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14266         libasan_preinit.o.
14268 2013-03-08  Marek Polacek  <polacek@redhat.com>
14269             Jakub Jelinek  <jakub@redhat.com>
14271         PR tree-optimization/56478
14272         * predict.c (is_comparison_with_loop_invariant_p): Change the
14273         type of loop_step to tree.
14274         (predict_loops): Adjust.
14275         (predict_iv_comparison): Perform the computations on double_ints.
14277 2013-03-08  Richard Biener  <rguenther@suse.de>
14279         PR tree-optimization/56570
14280         * tree-cfg.c (verify_expr_location_1): Verify locations for
14281         DECL_DEBUG_EXPR.
14282         * tree-sra.c (create_access_replacement): Strip locations
14283         from DECL_DEBUG_EXPRs.
14285 2013-03-08  Richard Biener  <rguenther@suse.de>
14287         * tree-inline.c (expand_call_inline): Do not associate
14288         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14289         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14291 2013-03-08  Richard Biener  <rguenther@suse.de>
14293         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14294         or block changes with -Og.  Fix for location / block encoding
14295         changes and PHI arguments with locations.
14297 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14299         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14300         for all counters.
14301         (struct output_info): Likewise.
14302         (register_overhead): Remove bad gcc_assert.
14303         (bitmap_find_bit): If there is only a single bitmap element, do not
14304         count a miss as a search.
14305         (print_statistics): Update for counter type changes.
14306         (dump_bitmap_statistics): Likewise.  Print headers such that they
14307         are properly lined up with the printed counters.
14309 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14311         PR tree-optimization/56559
14312         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14313         check that it has only a single use.
14315 2013-03-07  Richard Biener  <rguenther@suse.de>
14317         * doc/invoke.texi (fwhole-program): Discourage use in combination
14318         with -flto.
14320 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14322         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14324         PR tree-optimization/56539
14325         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14326         instead of GSI_CONTINUE_LINKING as last argument to
14327         force_gimple_operand_gsi.  Adjust function comment.
14329         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14330         aarch64-cores.def.
14332         PR middle-end/56548
14333         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14334         promoted mode, convert the result back to the original mode.
14336 2013-03-06  Richard Biener  <rguenther@suse.de>
14338         PR middle-end/56294
14339         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14340         (insert_updated_phi_nodes_compare_uids): New function.
14341         (update_ssa): Sort symbols_to_rename after UID before
14342         traversing it to insert PHI nodes.
14344 2013-03-06  Richard Biener  <rguenther@suse.de>
14346         PR middle-end/50494
14347         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14348         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14350         Revert
14351         2013-02-13  Richard Biener  <rguenther@suse.de>
14353         PR lto/50494
14354         * varasm.c (output_constant_def_1): Get the decl representing
14355         the constant as argument.
14356         (output_constant_def): Wrap output_constant_def_1.
14357         (make_decl_rtl): Use output_constant_def_1 with the decl
14358         representing the constant.
14359         (build_constant_desc): Optionally re-use a decl already
14360         representing the constant.
14361         (tree_output_constant_def): Adjust.
14363 2013-03-06  Joey Ye  <joey.ye@arm.com>
14365         PR lto/50293
14366         * gcc.c (convert_white_space): New function.
14367         (main): Handles white space in function name.
14369 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14371         PR target/56529
14372         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14373         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14374         to SH_DIV_CALL_TABLE for TARGET_SH2.
14375         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14376         list.
14377         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14378         call-table options.
14380 2013-03-05  Sterling Augustine  <saugustine@google.com>
14381             Cary Coutant  <ccoutant@google.com>
14383         PR debug/55364
14384         * dwarf2out.c (resolve_addr): Don't call
14385         remove_loc_list_addr_table_entries a second time for the same
14386         expression.
14388 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14390         PR debug/56510
14391         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14392         (avoid_complex_debug_insns): New function.
14393         (expand_debug_locations): Call it.
14395         PR rtl-optimization/56484
14396         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14397         lifetimes of hard registers on small register class machines.
14399 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14401         * config/microblaze/microblaze-protos.h: Rename
14402         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14403         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14404         fast_interrupt.
14405         (microblaze_fast_interrupt_function_p): New function.
14406         (microblaze_is_interrupt_handler): Rename to
14407         microblaze_is_interrupt_variant and add fast_interrupt check.
14408         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14409         (save_restore_insns): Likewise.
14410         (compute_frame_size): Likewise.
14411         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14412         (microblaze_globalize_label): Likewise.
14413         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14414         * config/microblaze/microblaze.md: Use wrapper
14415         microblaze_is_interrupt_variant.
14417 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14419         * sdbout.c (sdbout_one_type): Switch to current function's section
14420         supporting cold/hot.
14422 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14424         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14425         -mxl-reorder.
14427 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14429         PR middle-end/56461
14430         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14431         if VALGRIND_GET_VBITS is defined, temporarily make object
14432         memory all defined, and restore previous valgrind addressability
14433         and definability afterwards.  Free this_object at the end.
14435         PR middle-end/56461
14436         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14437         right before calling lra_create_live_ranges, also call it
14438         when clearing live_p.  Only call lra_clear_live_ranges
14439         at the end if live_p.
14441         PR middle-end/56461
14442         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14444 2013-03-05  Richard Biener  <rguenther@suse.de>
14446         PR tree-optimization/56521
14447         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14448         value-id.
14450 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14452         PR c++/55135
14453         * except.h (remove_unreachable_eh_regions): New prototype.
14454         * except.c (remove_eh_handler_splicer): New function, split out
14455         of remove_eh_handler.
14456         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14457         warning about running it on many EH regions one at a time.
14458         (remove_unreachable_eh_regions_worker): New function, walk the
14459         EH tree in depth-first order and remove non-marked regions.
14460         (remove_unreachable_eh_regions): New function.
14461         * tree-eh.c (mark_reachable_handlers): New function, split out
14462         from remove_unreachable_handlers.
14463         (remove_unreachable_handlers): Use mark_reachable_handlers and
14464         remove_unreachable_eh_regions.
14465         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14466         and remove_unreachable_eh_regions.
14468 2013-03-05  Richard Biener  <rguenther@suse.de>
14470         PR middle-end/56525
14471         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14472         not freeing them until the end.
14474 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14476         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14478 2013-03-05  Richard Biener  <rguenther@suse.de>
14480         PR tree-optimization/56270
14481         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14482         of loads after scheduling an SLP instance.
14484 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14486         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14487         tic6x.exp.
14488         (check_gcc_parallelize): Run guality.exp as a separate job from
14489         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14490         struct-layout-1.exp with stackalign.exp.
14492         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14494         PR middle-end/56461
14495         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14496         load_index sbitmap even if some bit in it isn't set.
14498         PR middle-end/56461
14499         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14500         (discover_iteration_bound_by_body_walk): Change queues to
14501         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14502         spelling in comment.  Call safe_push on queues[bound_index] directly.
14503         Release queues[queue_index] in every iteration unconditionally.
14504         Release bounds vector.
14506         PR middle-end/56461
14507         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14508         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14509         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14510         inner_phis vector.
14512 2013-03-05  Richard Biener  <rguenther@suse.de>
14514         PR lto/56515
14515         * tree-inline.c (remap_blocks_to_null): New function.
14516         (expand_call_inline): When expanding a call stmt without
14517         an associated block inline remap all callee blocks to NULL.
14519 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14521         PR rtl-optimization/56494
14522         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14523         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14524         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14526         PR middle-end/56461
14527         * sel-sched-ir.c (free_sched_pools): Release
14528         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14529         if succs_info_pool.max_top isn't -1.
14531         PR bootstrap/56509
14532         * opts.c (opts_obstack, opts_concat): Moved to...
14533         * opts-common.c (opts_obstack, opts_concat): ... here.
14535 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14537         PR middle-end/56461
14538         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14540 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14542         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14543         all appropriate places.
14545 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14547         PR tree-optimization/56424
14548         * ipa-split.c (split_function): Do not set the RSO flag if result is
14549         not by reference and its type is a register type.
14551 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14553         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14554         (microblaze_legitimate_pic_operand): Likewise
14555         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14556         new function microblaze_legitimate_pic_operand
14557         * config/microblaze/microblaze-protos.h
14558         (microblaze_legitimate_pic_operand): Declare.
14560 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14562         * config/microblaze/predicates.md (call_insn_simple_operand):
14563         New predicate for supported rtx code types.
14564         * config/microblaze/microblaze.md (call_internal1): Use
14565         call_insn_simple_operand predicate.
14567 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14569         PR middle-end/56461
14570         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14571         partitions.ordered_remove.
14573         PR middle-end/56461
14574         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14575         vec_oprnds0.create (1) for modifier == NONE.
14577         PR middle-end/56461
14578         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14579         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14580         vec_oprnds1 right before pushing anything to it for
14581         scalar_shift_arg.
14583         PR middle-end/56461
14584         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14585         set nbbs to 0 instead of having separate code path.
14586         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14587         instead of false as last argument if returning NULL.
14589 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14591         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14592         the attribute is now called "target" instead of "option".
14593         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14594         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14595         attribute/pragma name for TARGET_OPTION_VALID_P and
14596         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14597         * doc/tm.texi: Regenerated.
14599 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14601         * config/microblaze/microblaze.c:
14602         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14603         * config/microblaze/microblaze.h: Add -mxl-reorder to
14604         DRIVER_SELF_SPECS.
14605         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14606         instructions emitted if TARGET_REORDER.
14607         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14608         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14609         separately.
14611 2013-03-01  Xinliang David Li  <davidxl@google.com>
14613         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14614         walk length.
14616 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14618         PR middle-end/56461
14619         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14620         vector even when returning true.  Fix up function comment formatting.
14622         PR middle-end/56461
14623         * ira-build.c (ira_loop_nodes_count): New variable.
14624         (create_loop_tree_nodes): Initialize it.
14625         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14627         PR middle-end/56461
14628         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14629         method on dr_chain and result_chain.
14630         * tree-vect-stmts.c (vectorizable_store): Only call
14631         result_chain.create if j == 0.
14633         PR middle-end/56461
14634         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14635         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14636         before overwriting it.
14638 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14640         * doc/extended.texi (C Extensions): Change order in @menu
14641         to match @node.
14642         (Other MIPS Built-in Functions): Move last MIPS entry before
14643         "picoChip Built-in Functions".
14644         (SH Built-in Functions): Move after RX Built-in Functions.
14645         * doc/gcc.texi (Introduction): Change order in @menu
14646         to match @node.
14647         * doc/md.texi (Constraints): Ditto.
14648         * gty.texi (Type Information): Ditto.
14649         (User-provided marking routines for template types): Make
14650         subsection.
14651         * doc/invoke.texi (AArch64 Options): Move before
14652         "Adapteva Epiphany Options".
14654 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14655             Jakub Jelinek  <jakub@redhat.com>
14657         PR sanitizer/56454
14658         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14659         no_address_safety_analysis attribute.
14660         * doc/extend.texi (no_address_safety_attribute): Rename to
14661         no_sanitize_address attribute, mention no_address_safety_analysis
14662         attribute as deprecated alias.
14664 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
14666         PR middle-end/56461
14667         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14668         type to vec<vec<tree> > *.
14669         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
14670         to be vec<tree> instead of vec<tree> *, set vec_defs
14671         to vNULL and call vec_defs.create (number_of_vects), adjust other
14672         uses of vec_defs.
14673         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14674         vectorizable_condition): Adjust vect_get_slp_defs callers.
14676 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14678         * config/aarch64/aarch64.c
14679         (aarch64_float_const_representable): Remove unused variable.
14681 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14683         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14685 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14687         * config/aarch64/aarch64-builtins.c
14688         (aarch64_init_simd_builtins): Make static.
14690 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14692         * config/aarch64/aarch64.c
14693         (aarch64_simd_make_constant): Make static.
14695 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14697         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14698         with no initialization to the RHS of debug statements.
14700 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14702         PR tree-optimization/56294
14703         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14704         Adjust dumping.
14705         (get_access_replacement): Do not call create_access_replacement.
14706         Assert a replacement exists.
14707         (get_repl_default_def_ssa_name): Create the replacement declaration
14708         itself.
14710 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14712         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14713         final_end_function.
14715 2013-02-28  Marek Polacek  <polacek@redhat.com>
14717         PR rtl-optimization/56466
14718         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14719         if we're changing a loop.
14720         (peel_loops_completely): Likewise.
14722 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
14724         PR c++/55813
14725         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14727 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14729         PR target/56445
14730         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14731         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14732         INTX_FTYPE_FX, FX_FTYPE_INTX.
14733         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14735 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14737         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14738         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14739         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14740         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14741         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14742         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14743         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14744         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14745         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14746         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14747         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14748         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14749         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14750         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14751         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14752         (avrxmega6): Increase max flash segments from 5 to 6.
14753         * config/avr/t-multilib: Regenerate.
14754         * config/avr/avr-tables.opt: Regenerate.
14755         * doc/avr-mmcu.texi: Regenerate.
14757 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14759         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14760         (avr_device_to_arch): Rename to avr_device_to_ld.
14761         (avr_device_to_as): New prototype.
14762         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14763         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14764         * config/avr/driver-avr.c (avr_device_to_as): New.
14765         (avr_device_to_arch): Rename to avr_device_to_ld.
14767 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14769         PR middle-end/56461
14770         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14771         method on dr_chain and result_chain.
14773         PR middle-end/56461
14774         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14775         pointer_set_destroy on not_executed_last_iteration.
14777         PR middle-end/56461
14778         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
14780         PR middle-end/56461
14781         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14782         FOR_EACH_DEFINED_FUNCTION when freeing state.
14784         PR middle-end/56461
14785         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14786         pool_free.
14787         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14788         overwriting it.
14790         PR middle-end/56461
14791         * ipa-cp.c (decide_whether_version_node): Call vec_free on
14792         known_aggs[i].items and release known_aggs vector.
14794         PR middle-end/56461
14795         * ipa-reference.c (propagate): Free node_info even for alias nodes.
14797 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14799         * config/microblaze/microblaze.c (microblaze_emit_compare):
14800         Use xor for EQ/NE comparisions.
14801         * config/microblaze/microblaze.md (cstoresf4): Add constraints
14802         (cbranchsf4): Adjust operator to comparison_operator.
14804 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14806         PR middle-end/56461
14807         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14808         vector.
14809         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14810         vec_safe_push, always update *slot.
14811         (redirect_edge_var_map_clear): Use vec_free.
14812         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14813         (free_var_map_entry): Use vec_free.
14814         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14815         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14817 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
14819         PR middle-end/45472
14820         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14821         when the may_trap_p bit of the exprs being merged differs.
14822         Reorder tests for speculativeness in the logical and operator.
14824 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14826         * incpath.c (add_standard_paths): Use reconcat instead of concat
14827         where appropriate and avoid leaking memory.
14829         * opts.h: Include obstack.h.
14830         (opts_concat): New prototype.
14831         (opts_obstack): New declaration.
14832         * opts.c (opts_concat): New function.
14833         (opts_obstack): New variable.
14834         (init_options_struct): Call gcc_init_obstack on opts_obstack.
14835         (finish_options): Use opts_concat instead of concat
14836         and XOBNEWVEC instead of XNEWVEC.
14837         * opts-common.c (generate_canonical_option, decode_cmdline_option,
14838         generate_option): Likewise.
14839         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14840         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14842         PR target/56455
14843         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14844         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14846 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14848         PR middle-end/56461
14849         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14851 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14853         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14854         (arm_block_move_unaligned_straight): Likewise.
14855         (arm_adjust_block_mem): Likewise.
14857 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14859         PR target/48901
14860         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14861         temp, cond and label.
14862         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14864         PR target/52500
14865         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14866         * config/c6x/c6x.h (dbx_register_map): Update declaration.
14868         PR target/52501
14869         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14870         of prologue/epilogue functions.
14872         PR target/52550
14873         * config/tilegx/tilegx.c (tilegx_expand_prologue):
14874         Remove unused variable cfa_offset.
14875         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14877         PR target/54639
14878         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14879         type promotion to unsigned.
14881         PR target/54640
14882         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14883         for HOST_WIDE_INT of 32 bit / same size as int.
14884         (arm_block_move_unaligned_straight): Likewise.
14885         (arm_adjust_block_mem): Likewise.
14887         PR target/54662
14888         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14889         ALL_CFLAGS.
14891 2013-02-26  Marek Polacek  <polacek@redhat.com>
14893         PR tree-optimization/56426
14894         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
14896 2013-02-26  Richard Biener  <rguenther@suse.de>
14898         PR target/56444
14899         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14900         unused variable loops.
14902 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14904         PR tree-optimization/56448
14905         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14906         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14907         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14908         later operands of the references, or even first operand for
14909         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14911         PR tree-optimization/56443
14912         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14913         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14914         to type_for_mode langhook.
14916 2013-02-25  Matt Turner  <mattst88@gmail.com>
14918         * doc/invoke.texi: Document r4700.
14920 2013-02-25  Richard Biener  <rguenther@suse.de>
14922         PR tree-optimization/56175
14923         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14924         split out from ...
14925         (simplify_bitwise_binary): ... here.  Also guard the conversion
14926         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14928 2013-02-25  Catherine Moore  <clm@codesourcery.com>
14930         Revert:
14931         2013-02-24  Catherine Moore  <clm@codesourcery.com>
14932             Maciej W. Rozycki  <macro@codesourcery.com>
14933             Tom de Vries  <tom@codesourcery.com>
14934             Nathan Sidwell  <nathan@codesourcery.com>
14935             Iain Sandoe  <iain@codesourcery.com>
14936             Nathan Froyd  <froydnj@codesourcery.com>
14937             Chao-ying Fu  <fu@mips.com>
14939         * doc/extend.texi: (micromips, nomicromips, nocompression):
14940         Document new function attributes.
14941         * doc/invoke.texi (minterlink-compressed, mmicromips,
14942         m14k, m14ke, m14kec): Document new options.
14943         (minterlink-mips16): Update documentation.
14944         * doc/md.texi (ZC, ZD): Document new constraints.
14945         * configure.ac (gcc_cv_as_micromips): Check if linker
14946         supports the .set micromips directive.
14947         * configure: Regenerate.
14948         * config.in: Regenerate.
14949         * config/mips/mips-tables.opt: Regenerate.
14950         * config/mips/micromips.md: New file.
14951         * constraints.md (ZC, AD): New constraints.
14952         * config/mips/predicates.md (movep_src_register): New predicate.
14953         (movep_src_operand): New predicate.
14954         (non_volatile_mem_operand): New predicate.
14955         * config/mips/mips.md (multimem): New type.
14956         (length): Differentiate between 17-bit and 18-bit branch offsets.
14957         (MOVEP1, MOVEP2): New mode iterator.
14958         (mov_<load>l): Use ZC constraint.
14959         (mov_<load>r): Likewise.
14960         (mov_<store>l): Likewise.
14961         (mov_<store>r): Likewise.
14962         (*branch_equality<mode>_inverted): Add microMIPS support.
14963         (*branch_equality<mode>): Likewise.
14964         (*jump_absolute): Likewise.
14965         (indirect_jump_<mode>): Likewise.
14966         (tablejump_<mode>): Likewise.
14967         (<optab>_internal): Likewise.
14968         (sibcall_internal): Likewise.
14969         (sibcall_value_internal): Likewise.
14970         (prefetch): Use constraint ZD.
14971         * config/mips/mips.opt (minterlink-compressed): New option.
14972         (minterlink-mips16): Now an alias for minterlink-compressed.
14973         (mmicromips): New option.
14974         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14975         (compare_and_swap_12): Likewise.
14976         (sync_add<mode>): Likewise.
14977         (sync_<optab>_12): Likewise.
14978         (sync_old_<optab>_12): Likewise.
14979         (sync_new_<optab>_12): Likewise.
14980         (sync_nand_12): Likewise.
14981         (sync_old_nand_12): Likewise.
14982         (sync_new_nand_12): Likewise.
14983         (sync_sub<mode>): Likewise.
14984         (sync_old_add<mode>): Likewise.
14985         (sync_old_sub<mode>): Likewise.
14986         (sync_new_add<mode>): Likewise.
14987         (sync_new_sub<mode>): Likewise.
14988         (sync_<optab><mode>): Likewise.
14989         (sync_old_<optab><mode>): Likewise.
14990         (sync_new_<optab><mode>): Likewise.
14991         (sync_nand<mode>): Likewise.
14992         (sync_old_nand<mode>): Likewise.
14993         (sync_new_nand<mode>): Likewise.
14994         (sync_lock_test_and_set<mode>): Likewise.
14995         (test_and_set_12): Likewise.
14996         (atomic_compare_and_swap<mode>): Likewise.
14997         (atomic_exchange<mode>_llsc): Likewise.
14998         (atomic_fetch_add<mode>_llsc): Likewise.
14999         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15000         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15001         (umips_save_restore_pattern_p): Likewise.
15002         (umips_load_store_pair_p): Likewise.
15003         (umips_output_load_store_pair): Likewise.
15004         (umips_movep_target_p): Likewise.
15005         (umips_12bit_offset_address_p): Likewise.
15006         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15007         (mips_base_mips16): Rename this...
15008         (mips_base_compression_flags): ...to this. Update all uses.
15009         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15010         (mips_mips16_decl_p): Delete.
15011         (mips_nomips16_decl_p): Delete.
15012         (mips_get_compress_on_flags): New function.
15013         (mips_get_compress_off_flags): New function.
15014         (mips_get_compress_mode): New function.
15015         (mips_get_compress_on_name): New function.
15016         (mips_get_compress_off_name): New function.
15017         (mips_insert_attributes): Support multiple compression types.
15018         (mips_merge_decl_attributes): Likewise.
15019         (umips_12bit_offset_address_p): New function.
15020         (mips_start_function_definition): Emit .set micromips directive.
15021         (mips_call_may_need_jalx_p): New function.
15022         (mips_function_ok_for_sibcall): Add microMIPS support.
15023         (mips_print_operand_punctuation): Support short delay slots and
15024         compact jumps.
15025         (umips_swm_mask, umips_swm_encoding): New.
15026         (umips_build_save_restore): New function.
15027         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15028         (was_mips16_p): Remove.
15029         (old_compression_mode): New.
15030         (mips_set_compression_mode): New function.
15031         (mips_set_current_function): Add microMIPS support.
15032         (mips_option_override): Likewise.
15033         (umips_save_restore_pattern_p): New function.
15034         (umips_output_save_restore): New function.
15035         (umips_load_store_pair_p_1): New function.
15036         (umips_load_store_pair_p): New function.
15037         (umips_output_load_store_pair_1): New function.
15038         (umips_output_load_store_pair): New function.
15039         (umips_movep_target_p) New function.
15040         (mips_prepare_pch_save): Add microMIPS support.
15041         * config/mips/mips.h (TARGET_COMPRESSION): New.
15042         (TARGET_CPU_CPP_BUILTINS): Update macro
15043         to use new compression flags and to support microMIPS.
15044         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15045         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15046         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15047         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15048         (ASM_SPEC): Support mmicromips and mno-micromips.
15049         (M16STORE_REG_P): New macro.
15050         (MIPS_CALL): Support TARGET_MICROMIPS.
15051         (MICROMIPS_J): New macro.
15052         (mips_base_mips16): Rename this...
15053         (mips_base_compression_flags): ...to this.
15054         (UMIPS_12BIT_OFFSET_P): New macro.
15055         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15056         (MULTILIB_DIRNAMES): Likewise.
15058 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15060         PR rtl-optimization/56131
15061         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15062         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15063         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15065 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15067         * doc/invoke.texi (-fsanitize=): Move from optimization
15068         to debugging options.
15070 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15072         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15074 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15075             Alexander Monakov  <amonakov@ispras.ru>
15077         PR middle-end/56077
15078         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15079         flush pending lists also on non-jumps.  Adjust comment.
15081 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15082             Maciej W. Rozycki  <macro@codesourcery.com>
15083             Tom de Vries  <tom@codesourcery.com>
15084             Nathan Sidwell  <nathan@codesourcery.com>
15085             Iain Sandoe  <iain@codesourcery.com>
15086             Nathan Froyd  <froydnj@codesourcery.com>
15087             Chao-ying Fu  <fu@mips.com>
15089         * doc/extend.texi: (micromips, nomicromips, nocompression):
15090         Document new function attributes.
15091         * doc/invoke.texi (minterlink-compressed, mmicromips,
15092         m14k, m14ke, m14kec): Document new options.
15093         (minterlink-mips16): Update documentation.
15094         * doc/md.texi (ZC, ZD): Document new constraints.
15095         * configure.ac (gcc_cv_as_micromips): Check if linker
15096         supports the .set micromips directive.
15097         * configure: Regenerate.
15098         * config.in: Regenerate.
15099         * config/mips/mips-tables.opt: Regenerate.
15100         * config/mips/micromips.md: New file.
15101         * constraints.md (ZC, AD): New constraints.
15102         * config/mips/predicates.md (movep_src_register): New predicate.
15103         (movep_src_operand): New predicate.
15104         (non_volatile_mem_operand): New predicate.
15105         * config/mips/mips.md (multimem): New type.
15106         (length): Differentiate between 17-bit and 18-bit branch offsets.
15107         (MOVEP1, MOVEP2): New mode iterator.
15108         (mov_<load>l): Use ZC constraint.
15109         (mov_<load>r): Likewise.
15110         (mov_<store>l): Likewise.
15111         (mov_<store>r): Likewise.
15112         (*branch_equality<mode>_inverted): Add microMIPS support.
15113         (*branch_equality<mode>): Likewise.
15114         (*jump_absolute): Likewise.
15115         (indirect_jump_<mode>): Likewise.
15116         (tablejump_<mode>): Likewise.
15117         (<optab>_internal): Likewise.
15118         (sibcall_internal): Likewise.
15119         (sibcall_value_internal): Likewise.
15120         (prefetch): Use constraint ZD.
15121         * config/mips/mips.opt (minterlink-compressed): New option.
15122         (minterlink-mips16): Now an alias for minterlink-compressed.
15123         (mmicromips): New option.
15124         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15125         (compare_and_swap_12): Likewise.
15126         (sync_add<mode>): Likewise.
15127         (sync_<optab>_12): Likewise.
15128         (sync_old_<optab>_12): Likewise.
15129         (sync_new_<optab>_12): Likewise.
15130         (sync_nand_12): Likewise.
15131         (sync_old_nand_12): Likewise.
15132         (sync_new_nand_12): Likewise.
15133         (sync_sub<mode>): Likewise.
15134         (sync_old_add<mode>): Likewise.
15135         (sync_old_sub<mode>): Likewise.
15136         (sync_new_add<mode>): Likewise.
15137         (sync_new_sub<mode>): Likewise.
15138         (sync_<optab><mode>): Likewise.
15139         (sync_old_<optab><mode>): Likewise.
15140         (sync_new_<optab><mode>): Likewise.
15141         (sync_nand<mode>): Likewise.
15142         (sync_old_nand<mode>): Likewise.
15143         (sync_new_nand<mode>): Likewise.
15144         (sync_lock_test_and_set<mode>): Likewise.
15145         (test_and_set_12): Likewise.
15146         (atomic_compare_and_swap<mode>): Likewise.
15147         (atomic_exchange<mode>_llsc): Likewise.
15148         (atomic_fetch_add<mode>_llsc): Likewise.
15149         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15150         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15151         (umips_save_restore_pattern_p): Likewise.
15152         (umips_load_store_pair_p): Likewise.
15153         (umips_output_load_store_pair): Likewise.
15154         (umips_movep_target_p): Likewise.
15155         (umips_12bit_offset_address_p): Likewise.
15156         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15157         (mips_base_mips16): Rename this...
15158         (mips_base_compression_flags): ...to this. Update all uses.
15159         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15160         (mips_mips16_decl_p): Delete.
15161         (mips_nomips16_decl_p): Delete.
15162         (mips_get_compress_on_flags): New function.
15163         (mips_get_compress_off_flags): New function.
15164         (mips_get_compress_mode): New function.
15165         (mips_get_compress_on_name): New function.
15166         (mips_get_compress_off_name): New function.
15167         (mips_insert_attributes): Support multiple compression types.
15168         (mips_merge_decl_attributes): Likewise.
15169         (umips_12bit_offset_address_p): New function.
15170         (mips_start_function_definition): Emit .set micromips directive.
15171         (mips_call_may_need_jalx_p): New function.
15172         (mips_function_ok_for_sibcall): Add microMIPS support.
15173         (mips_print_operand_punctuation): Support short delay slots and
15174         compact jumps.
15175         (umips_swm_mask, umips_swm_encoding): New.
15176         (umips_build_save_restore): New function.
15177         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15178         (was_mips16_p): Remove.
15179         (old_compression_mode): New.
15180         (mips_set_compression_mode): New function.
15181         (mips_set_current_function): Add microMIPS support.
15182         (mips_option_override): Likewise.
15183         (umips_save_restore_pattern_p): New function.
15184         (umips_output_save_restore): New function.
15185         (umips_load_store_pair_p_1): New function.
15186         (umips_load_store_pair_p): New function.
15187         (umips_output_load_store_pair_1): New function.
15188         (umips_output_load_store_pair): New function.
15189         (umips_movep_target_p) New function.
15190         (mips_prepare_pch_save): Add microMIPS support.
15191         * config/mips/mips.h (TARGET_COMPRESSION): New.
15192         (TARGET_CPU_CPP_BUILTINS): Update macro
15193         to use new compression flags and to support microMIPS.
15194         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15195         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15196         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15197         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15198         (ASM_SPEC): Support mmicromips and mno-micromips.
15199         (M16STORE_REG_P): New macro.
15200         (MIPS_CALL): Support TARGET_MICROMIPS.
15201         (MICROMIPS_J): New macro.
15202         (mips_base_mips16): Rename this...
15203         (mips_base_compression_flags): ...to this.
15204         (UMIPS_12BIT_OFFSET_P): New macro.
15205         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15206         (MULTILIB_DIRNAMES): Likewise.
15208 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15210         PR target/52555
15211         * target-globals.c (save_target_globals): For init_reg_sets and
15212         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15214 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15216         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15217         * config/aarch64/t-aarch64
15218         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15220 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15222         PR inline-asm/56148
15223         * lra-constraints.c (process_alt_operands): Reload operand
15224         conflicting with earlier clobber only if no more other conflicting
15225         operands.
15227 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15229         PR sanitizer/56393
15230         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15231         if not linking a shared library.
15233 2013-02-22  Seth LaForge  <sethml@google.com>
15235         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15237 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15239         * config/arm/arm.md (split for extendsidi): Update condition.
15240         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15241         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15242         (qhs_zextenddi_cstr): Likewise.
15244 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15246         PR middle-end/56420
15247         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15248         avoid signed wrapping.
15249         (expand_mult): Handle properly multiplication by
15250         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15251         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15252         in the compiler if coeff is HOST_WIDE_INT_MIN.
15253         (expand_divmod): Don't make ext_op1 static, change it's type to
15254         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15256         PR rtl-optimization/50339
15257         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15258         field.
15259         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15260         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15261         into splitting_ashiftrt field.
15262         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15263         ASHIFTRT.
15264         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15265         choices.
15267 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15269         PR middle-end/56108
15270         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15271         are sure to go irrevocable.
15273 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15275         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15276         scalars are valid operands.
15278 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15280         PR tree-optimization/56310
15281         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15282         only matching indices and non-negative final offsets.
15283         (intersect_aggregates_with_edge): Pass src_idx to
15284         agg_replacements_to_vector.  Pass src_idx insstead of index to
15285         intersect_with_agg_replacements.
15287 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15289         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15290         instead of hard-wired defaults.
15292 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15294         * doc/invoke.texi (MIPS Options): Update documentation of the
15295         floating-point multiply-accumulate instruction restrictions.
15297 2013-02-21  Kostya Serebryany  <kcc@google.com>
15299         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15300         asan_shadow_offset on x86_64 linux.
15302 2013-02-21  Richard Biener  <rguenther@suse.de>
15304         PR tree-optimization/56415
15305         Revert
15306         2013-02-11  Richard Biener  <rguenther@suse.de>
15308         PR tree-optimization/56273
15309         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15310         first VRP run.
15312 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15314         PR bootstrap/56258
15315         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15316         instead of @itemx.
15318         PR inline-asm/56405
15319         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15320         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15322 2013-02-20  Jan Hubicka  <jh@suse.cz>
15324         PR tree-optimization/56265
15325         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15326         when target is referenced for first time.
15328 2013-02-20  Richard Biener  <rguenther@suse.de>
15330         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15331         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15332         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15333         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15334         not return anything.
15335         (rename_ssa_copies): Do not remove unused locals.
15336         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15337         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15338         * passes.c (execute_function_todo): Do not schedule unused locals
15339         removal if cleanup_tree_cfg did something.
15340         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15341         about the number of removed locals.
15343 2013-02-20  Richard Biener  <rguenther@suse.de>
15345         PR tree-optimization/56398
15346         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15348 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15350         PR tree-optimization/55334
15351         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15352         restricted pointers to arrays.
15354 2013-02-20  Richard Biener  <rguenther@suse.de>
15355             Jakub Jelinek  <jakub@redhat.com>
15357         PR tree-optimization/56396
15358         * tree-ssa-ccp.c (n_const_val): New static variable.
15359         (get_value): Return NULL for SSA names we don't have a lattice
15360         entry for.
15361         (ccp_initialize): Initialize n_const_val.
15362         * tree-ssa-copy.c (n_copy_of): New static variable.
15363         (init_copy_prop): Initialize n_copy_of.
15364         (get_value): Return NULL_TREE for SSA names we don't have a
15365         lattice entry for.
15367 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15369         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15371 2013-02-20  Richard Biener  <rguenther@suse.de>
15373         * genpreds.c (write_lookup_constraint): Do not compare first
15374         letter of the constraint again.
15376 2013-02-20  Richard Biener  <rguenther@suse.de>
15378         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15379         and ceil_log2.
15380         (get_use_iv_cost): Terminate hashtable walk when coming across
15381         an empty entry.
15383 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15385         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15386         reassociation for avx2 targets.
15388 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15390         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15391         Add version check for v8.10.a to enable microblaze_has_clz
15392         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15393         version and TARGET_PATTERN_COMPARE check
15394         * config/microblaze/microblaze.md: New clzsi2 instruction
15396 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15398         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15399         function before branching.
15401 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15403         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15404         DUMP_INSN_RTX_UID.
15405         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15407 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15409         PR middle-end/55889
15410         * sel-sched.c: Include ira.h.
15411         (implicit_clobber_conflict_p): New function.
15412         (moveup_expr): Use it.
15413         * Makefile.in (sel-sched.o): Depend on ira.h.
15415 2013-02-19  Richard Biener  <rguenther@suse.de>
15417         PR tree-optimization/56384
15418         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15419         (vn_hash_type): Split out from ...
15420         (vn_hash_constant_with_type): ... here.
15421         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15422         (vn_phi_eq): Compare types from vn_phi_s structure.
15423         (vn_phi_lookup): Populate vn_phi_s type.
15424         (vn_phi_insert): Likewise.
15426 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15428         PR tree-optimization/56350
15429         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15430         if haven't found reduction or nested cycle operand, rather than
15431         asserting we must find it.
15433         PR tree-optimization/56381
15434         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15435         to fold_build3.
15437 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15438             Jakub Jelinek  <jakub@redhat.com>
15440         PR target/52555
15441         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15442         (swap_optab_enable): Same.
15443         (init_all_optabs): Use argument instead of global.
15444         * tree.h (struct tree_optimization_option): New field target_optabs.
15445         * expr.h (init_all_optabs): Add argument to prototype.
15446         (TREE_OPTIMIZATION_OPTABS): New.
15447         (save_optabs_if_changed): Protoize.
15448         * optabs.h: Declare this_fn_optabs.
15449         * optabs.c (save_optabs_if_changed): New.
15450         Declare this_fn_optabs.
15451         (init_optabs): Add argument to init_all_optabs() call.
15452         * function.c (invoke_set_current_function_hook): Handle per
15453         function optabs.
15454         * function.h (struct function): New field optabs.
15455         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15456         optimization_current_node has changed.
15457         * target-globals.h (save_target_globals_default_opts): Protoize.
15458         * target-globals.c (save_target_globals_default_opts): New.
15460 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15462         PR target/56347
15463         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15464         registers %fr12 and %fr12R as call used.
15466         PR target/56214
15467         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15468         and HImode, require all displacements to be an integer multiple of
15469         their mode size.
15470         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15471         only allow QImode and HImode when reload is in progress and strict is
15472         true.  Likewise for symbolic addresses.  Use base14_operand to check
15473         displacements in REG+BASE addresses.
15475 2013-02-18  Richard Biener  <rguenther@suse.de>
15477         PR tree-optimization/56366
15478         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15479         handle sign-conversion of outer-loop initial induction value.
15481 2013-02-18  Richard Biener  <rguenther@suse.de>
15483         PR middle-end/56349
15484         * cfghooks.c (merge_blocks): If we merge a latch into another
15485         block adjust references to it.
15486         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15487         (verify_loop_structure): Verify that a recorded latch is in fact
15488         a latch.
15490 2013-02-18  Richard Biener  <rguenther@suse.de>
15492         PR tree-optimization/56321
15493         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15494         order SSA name release and virtual operand unlinking.
15496 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15498         * config/microblaze/microblaze.md (save_stack_block): Define.
15499         (restore_stack_block): Likewise.
15501 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15503         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15504         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15505         * config/microblaze/microblaze.c (microblaze_option_override):
15506         Bail out early for PIC modes when target does not support PIC.
15508 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15510         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15511         Replace with a microblaze version.
15512         (microblaze_trampoline_init): Adapt for microblaze.
15513         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15514         microblaze.
15516 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15517             Dodji Seketeli  <dodji@redhat.com>
15519         PR asan/56330
15520         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15521         (instrument_mem_region_access): Do not forget to always put
15522         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15523         0) statement, even for cases where either 'base' or 'base + len'
15524         are not instrumented -- because they have been previously
15525         instrumented.  Simplify the logic by putting all the statements
15526         instrument 'base + len' inside a sequence, and then insert that
15527         sequence right before the current insertion point.  Then, to
15528         instrument 'base + len', just get an iterator on that statement.
15529         And do not forget to update the pointer to iterator the function
15530         received as argument.
15532 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15534         PR rtl-optimization/56348
15535         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15537 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15539         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15540         (clean_graph_dump_file): Pass base to start_graph_dump.
15542 2013-02-14  Richard Henderson  <rth@redhat.com>
15544         PR target/55941
15545         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15547 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15549         * collect2-aix.h: Define F_LOADONLY.
15551 2013-02-14  Richard Biener  <rguenther@suse.de>
15553         PR lto/50494
15554         * varasm.c (output_constant_def_1): Get the decl representing
15555         the constant as argument.
15556         (output_constant_def): Wrap output_constant_def_1.
15557         (make_decl_rtl): Use output_constant_def_1 with the decl
15558         representing the constant.
15559         (build_constant_desc): Optionally re-use a decl already
15560         representing the constant.
15561         (tree_output_constant_def): Adjust.
15563 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15565         Fix an asan crash
15566         * asan.c (instrument_builtin_call):  Really put the length of the
15567         second source argument into src1_len.
15569 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15571         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15572         argument.  If it is false, don't create edge from then_bb to
15573         fallthru_bb.
15574         (insert_if_then_before_iter): Pass true to it.
15575         (build_check_stmt): Pass false to it.
15576         (transform_statements): Flush hash table only on extended basic
15577         block boundaries, rather than at the beginning of every bb.
15578         Don't flush hash table on nonfreeing_call_p calls.
15579         * tree-flow.h (nonfreeing_call_p): New prototype.
15580         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15582 2013-02-13  David S. Miller  <davem@davemloft.net>
15584         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15586 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15588         PR target/56184
15589         * ira.c (max_regno_before_ira): Move from ...
15590         (ira): ... here.
15591         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15592         vec_safe_length.
15594 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15596         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15598 2013-02-13  Richard Biener  <rguenther@suse.de>
15600         PR lto/56295
15601         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15602         globals in MEM_REFs.
15604 2013-02-13  Richard Biener  <rguenther@suse.de>
15606         * loop-init.c (loop_optimizer_init): Clear loop state when
15607         re-initializing preserved loops.
15608         * loop-unswitch.c (unswitch_single_loop): Return whether
15609         we unswitched the loop.  Do not verify loop state here.
15610         (unswitch_loops): When we unswitched a loop discover new loops.
15612 2013-02-13  Kostya Serebryany  <kcc@google.com>
15614         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15615         on x86_64 linux.
15616         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15618 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15620         Avoid instrumenting duplicated memory access in the same basic block
15621         * Makefile.in (asan.o): Add new dependency on hash-table.h
15622         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15623         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15624         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15625         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15626         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15627         (get_mem_ref_of_assignment): New functions.
15628         (get_mem_refs_of_builtin_call): Extract from
15629         instrument_builtin_call and tweak a little bit to make it fit with
15630         the new signature.
15631         (instrument_builtin_call): Use the new
15632         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15633         of is_gimple_builtin_call.
15634         (instrument_derefs, instrument_mem_region_access): Insert the
15635         instrumented memory reference into the hash table.
15636         (maybe_instrument_assignment): Renamed instrument_assignment into
15637         this, and change it to advance the iterator when instrumentation
15638         actually happened and return true in that case.  This makes it
15639         homogeneous with maybe_instrument_assignment, and thus give a
15640         chance to callers to be more 'regular'.
15641         (transform_statements): Clear the memory reference hash table
15642         whenever we enter a new BB, when we cross a function call, or when
15643         we are done transforming statements.  Use
15644         maybe_instrument_assignment instead of instrumentation.  No more
15645         need to special case maybe_instrument_assignment and advance the
15646         iterator after calling it; it's now handled just like
15647         maybe_instrument_call.  Update comment.
15649 2013-02-13  Richard Biener  <rguenther@suse.de>
15651         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15652         Fix loop discovery code.
15654 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15656         PR inline-asm/56148
15657         * lra-constraints.c (process_alt_operands): Match early clobber
15658         operand with itself.  Check conflicts with earlyclobber only if
15659         the operand is not reloaded.  Prefer to reload conflicting operand
15660         if earlyclobber and matching operands are the same.
15662 2013-02-12  Richard Biener  <rguenther@suse.de>
15664         PR lto/56297
15665         * lto-streamer-out.c (write_symbol): Do not output symbols
15666         for hard register variables.
15668 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
15670         PR target/54222
15671         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15672         (umulsidi3_insn, mulsidi3_insn): New insns.
15674 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
15676         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15677         (struct tune_params): Add vec_costs field.
15678         * config/arm/arm.c (arm_builtin_vectorization_cost)
15679         (arm_add_stmt_cost): New functions.
15680         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15681         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15682         (arm_default_vec_cost): New struct of type cpu_vec_costs.
15683         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15684         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15685         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15686         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15688 2013-02-12  Richard Biener  <rguenther@suse.de>
15690         PR lto/56295
15691         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15692         decls again if possible.
15694 2013-02-12  Richard Biener  <rguenther@suse.de>
15696         PR middle-end/56288
15697         * tree-ssa.c (verify_ssa_name): Fix check, move
15698         SSA_NAME_IN_FREE_LIST check up.
15700 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
15701             Steven Bosscher   <steven@gcc.gnu.org>
15703         PR rtl-optimization/56151
15704         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15705         equal to op0 or op1, and last_insn pattern is CODE operation
15706         with MEM dest and one of the operands matches that MEM.
15708 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
15710         * doc/extend.texi: Document Function Multiversioning and "default"
15711         parameter string to target attribute.
15712         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15713         target attribute parameter is "default".
15714         (ix86_compare_version_priority): Remove checks for target attribute.
15715         (ix86_mangle_function_version_assembler_name): Change error to sorry.
15716         Remove check for target attribute equal to NULL. Add assert.
15717         (ix86_generate_version_dispatcher_body): Change error to sorry.
15719 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
15720             Jack Howarth  <howarth@bromo.med.uc.edu>
15721             Patrick Marlier  <patrick.marlier@gmail.com>
15723         PR libitm/55693
15724         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15725         define ENDFILE_SPEC as TM_DESTRUCTOR.
15726         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15728 2013-02-11  Alexander Potapenko  <glider@google.com>
15729             Jack Howarth  <howarth@bromo.med.uc.edu>
15730             Jakub Jelinek  <jakub@redhat.com>
15732         PR sanitizer/55617
15733         * config/darwin.c (cdtor_record): Rename ctor_record.
15734         (sort_cdtor_records): Rename sort_ctor_records.
15735         (finalize_dtors): New routine to sort destructors by
15736         priority before use in assemble_integer.
15737         (machopic_asm_out_destructor): Use finalize_dtors if needed.
15739 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
15741         PR rtl-optimization/56275
15742         * simplify-rtx.c (avoid_constant_pool_reference): Check that
15743         offset is non-negative and less than cmode size before
15744         calling simplify_subreg.
15746 2013-02-11  Richard Biener  <rguenther@suse.de>
15748         PR tree-optimization/56264
15749         * cfgloop.h (fix_loop_structure): Adjust prototype.
15750         * loop-init.c (fix_loop_structure): Return the number of
15751         newly discovered loops.
15752         * tree-cfgcleanup.c (repair_loop_structures): When new loops
15753         are discovered, do a full loop-closed SSA rewrite.
15755 2013-02-11  Richard Biener  <rguenther@suse.de>
15757         PR tree-optimization/56273
15758         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15759         first VRP run.
15760         (check_array_ref): Fix missing newline in dumps.
15761         (search_for_addr_array): Likewise.
15763 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
15765         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15767 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
15769         PR target/56256
15770         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15772 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
15774         PR rtl-optimization/56246
15775         * lra-constraints.c (simplify_operand_subreg): Try to reuse
15776         reload pseudo.
15777         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15778         constraints are satisfied.
15780 2013-02-08  Jeff Law  <law@redhat.com>
15782         PR debug/53948
15783         * emit-rtl.c (reg_is_parm_p): New function.
15784         * regs.h (reg_is_parm_p): New prototype.
15785         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15786         callee-clobbered registers.
15788 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
15790         PR target/56043
15791         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15792         If there is no implicit builtin declaration, just return NULL.
15794 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15796         * config/i386/sse.md (FMAMODEM): New mode iterator.
15797         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15798         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15800 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15802         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15803         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15804         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15806 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15808         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15809         (microblaze*-*-elf): Likewise.
15810         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15811         LINK_SPEC.
15812         * config/microblaze/microblaze-c.c: Add builtin defines for
15813         _LITTLE_ENDIAN and _BIG_ENDIAN.
15814         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15815         add to TARGET_DEFAULT flags.
15816         Expand ASM_SPEC and LINK_SPEC.
15817         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15818         * config/microblaze/microblaze.md: Update extendsidi2 and
15819         movdi_internal instructions to use low-order / high-order reg
15820         print_operands.
15821         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15822         options and inversemask / mask of LITTLE_ENDIAN.
15823         * config/microblaze/t-microblaze: Expand multilib options to
15824         include mlittle-endian (le) and update exceptions patterns.
15826 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
15828         PR rtl-optimization/56195
15829         * lra-constraints.c (get_reload_reg): Don't reuse regs
15830         if they have smaller mode than requested, if they have
15831         wider mode than requested, try to return a SUBREG.
15833         PR tree-optimization/56250
15834         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15835         if type is unsigned and code isn't MULT_EXPR.
15837 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15839         PR tree-optimization/56064
15840         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15841         bits according to mode.
15842         * fixed-value.h (fixed_from_double_int)
15843         (const_fixed_from_double_int): Adjust comments.
15845 2013-02-08  Richard Biener  <rguenther@suse.de>
15847         PR lto/56231
15848         * lto-streamer.h (struct data_in): Remove current_file, current_line
15849         and current_col members.
15850         * lto-streamer-out.c (lto_output_location): Stream changed bits
15851         en-block for efficiency.
15852         * lto-streamer-in.c (clear_line_info): Remove.
15853         (lto_input_location): Cache current file, line and column
15854         globally via local statics.  Read changed bits en-block.
15855         (input_function): Do not call clear_line_info.
15856         (lto_read_body): Likewise.
15857         (lto_input_toplevel_asms): Likewise.
15859 2013-02-08  Michael Matz  <matz@suse.de>
15861         PR tree-optimization/52448
15862         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15863         (nt_call_phase): New static.
15864         (add_or_mark_expr): Only mark accesses with newer phase than any
15865         call seen.
15866         (nonfreeing_call_p): New.
15867         (nt_init_block): Update nt_call_phase, mark blocks as visited.
15868         (nt_fini_block): Keep blocks marked as visited.
15869         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15871 2013-02-08  Richard Biener  <rguenther@suse.de>
15873         * ira.c (ira): Free broken dominator information.
15875 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15877         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15879 2013-02-08  Marek Polacek  <polacek@redhat.com>
15881         * cfgloop.c (verify_loop_structure): Add more checking of headers.
15883 2013-02-08  Richard Biener  <rguenther@suse.de>
15885         PR middle-end/56181
15886         * cfgloop.h (flow_loops_find): Adjust.
15887         (bb_loop_header_p): Declare.
15888         * cfgloop.c (bb_loop_header_p): New function split out from ...
15889         (flow_loops_find): ... here.  Adjust function signature,
15890         support incremental loop structure update.
15891         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
15892         * cfgloopmanip.c (fix_loop_structure): Move ...
15893         * loop-init.c (fix_loop_structure): ... here.
15894         (apply_loop_flags): Split out from ...
15895         (loop_optimizer_init): ... here.
15896         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
15897         in incremental mode, only remove dead loops here.
15899 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15901         PR target/54222
15902         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15903         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15904         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15905         (*round<mode>3.libgcc): New insns for fixed-modes.
15906         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15907         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15908         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15909         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15910         implementations.  Define to __builtin_avr_absFX,
15911         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15912         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15913         __builtin_avr_countlsFX, respectively.
15914         * config/avr/avr-c.c (target.h): Include it.
15915         (enum avr_builtin_id): New enum.
15916         (avr_resolve_overloaded_builtin): New static function.
15917         (avr_register_target_pragmas): Use it to set
15918         targetm.resolve_overloaded_builtin.
15919         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15920         tree nodes used by DEF_BUILTIN.
15921         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15922         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15923         <AVR_BUILTIN_xxBITS>: Same.
15925 2013-02-08  Richard Biener  <rguenther@suse.de>
15927         * cfgloop.c (verify_loop_structure): Properly handle
15928         a loop exiting to another loop header.
15929         * ira-int.h (ira_loops): Remove.
15930         * ira.c (ira_loops): Remove.
15931         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15932         (do_reload): Use loop_optimizer_finalize.
15933         * ira-build.c (create_loop_tree_nodes): Use get_loops and
15934         number_of_loops to access the loop tree.
15935         (more_one_region_p): Likewise.
15936         (finish_loop_tree_nodes): Likewise.
15937         (rebuild_regno_allocno_maps): Likewise.
15938         (mark_loops_for_removal): Likewise.
15939         (mark_all_loops_for_removal): Likewise.
15940         (remove_unnecessary_regions): Likewise.
15941         (ira_build): Likewise.
15942         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15944 2013-02-08  Richard Biener  <rguenther@suse.de>
15946         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15947         * ipa-pure-const.c (analyze_function): Avoid calling
15948         mark_irreducible_loops twice.
15949         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
15951 2013-02-07  David S. Miller  <davem@davemloft.net>
15953         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15954         on 'reg'.
15955         * var-tracking.c (vt_add_function_parameter): Test the presence of
15956         HAVE_window_save properly and do not remap argument registers when
15957         we have a leaf function.
15959 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
15961         PR bootstrap/56227
15962         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15963         instead of "ll".
15964         * config/i386/i386.c (ix86_print_operand): Ditto.
15966 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15968         * lra-constraints.c (process_alt_operands): Fix recently added comment.
15970 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15972         PR rtl-optimization/56225
15973         * lra-constraints.c (process_alt_operands): Check that reload hard
15974         reg can hold value for strict_low_part.
15976 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15978         PR debug/56154
15979         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15980         dwarf2out_end_function.
15981         (in_first_function_p, maybe_at_text_label_p,
15982         first_loclabel_num_not_at_text_label): New variables.
15983         (dwarf2out_var_location): In the first function find out
15984         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
15985         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15986         functions.
15988 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
15990         PR rtl-optimization/56178
15991         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15992         SUBREG of a register.  Tidy up related block of code.
15993         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15994         note if the source is a register or a SUBREG of a register.
15996 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15998         PR target/56228
15999         * config/rs6000/rs6000.md (ptrm): New mode attr.
16000         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16001         call_value_indirect_aix<pttrsize>,
16002         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16003         m in constraints.
16005 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16007         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16008         if -bnortl. Convert to strcmp and strncmp.
16010 2013-02-07  Alan Modra  <amodra@gmail.com>
16012         PR target/54009
16013         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16014         addresses won't wrap when offsetting.
16015         (rs6000_secondary_reload): Provide secondary reloads needed for
16016         wrapping LO_SUM addresses.
16018 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16020         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16021         MACH, just __MACH__.
16023 2013-02-06  Richard Biener  <rguenther@suse.de>
16025         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16026         instead of calling fix_loop_structure.
16028 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16030         PR middle-end/56217
16031         * omp-low.c (use_pointer_for_field): Return false if
16032         lower_send_shared_vars doesn't generate any copy-out code.
16034 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16036         PR rtl-optimization/56131
16037         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16038         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16039         of the label is NULL.  Add comment.
16041 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16043         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16045         PR sanitizer/55374
16046         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16047         (STATIC_LIBTSAN_LIBS): Likewise.
16048         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16049         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16050         is defined, don't add anything else beyond that.
16051         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16052         (LINK_COMMAND_SPEC): Use them.
16054         PR tree-optimization/56205
16055         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16056         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16057         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16059 2013-02-05  Richard Biener  <rguenther@suse.de>
16061         PR tree-optimization/53342
16062         PR tree-optimization/53185
16063         * tree-vectorizer.h (vect_check_strided_load): Remove.
16064         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16065         not disallow peeling for vectorized strided loads.
16066         (vect_check_strided_load): Make static and simplify.
16067         (vect_analyze_data_refs): Adjust.
16068         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16069         correctly when vectorizing strided loads.
16071 2013-02-05  Richard Biener  <rguenther@suse.de>
16073         * doc/install.texi: Refer to ISL, not PPL.
16075 2013-02-05  Jan Hubicka  <jh@suse.cz>
16077         PR tree-optimization/55789
16078         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16080 2013-02-05  Jan Hubicka  <jh@suse.cz>
16082         PR tree-optimization/55789
16083         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16084         the dead call anyway.
16086 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16088         PR sanitizer/55374
16089         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16091 2013-02-04  Alexander Potapenko  <glider@google.com>
16092             Jack Howarth  <howarth@bromo.med.uc.edu>
16093             Jakub Jelinek  <jakub@redhat.com>
16095         PR sanitizer/55617
16096         * config/darwin.c (sort_ctor_records): Stabilized qsort
16097         on constructor priority by using original position.
16098         (finalize_ctors): New routine to sort constructors by
16099         priority before use in assemble_integer.
16100         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16102 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16104         PR libstdc++/54314
16105         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16106         about visibility on artificial decls.
16107         * config/sol2.c (solaris_assemble_visibility): Likewise.
16109 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16111         PR target/56186
16112         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16113         argument and improve checking of return-argument types for 16-byte
16114         modes.
16115         (ix86_function_value_1): Add additional valtype argument on call
16116         of function_value_64.
16117         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16118         handling infunction_value_64 function.
16120 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16122         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16124 2013-02-04  Richard Biener  <rguenther@suse.de>
16126         PR tree-optimization/56188
16127         * tree-ssa-structalias.c (label_visit): Consider case with
16128         initially non-empty points-to set.
16129         (perform_var_substitution): Dump node mapping and clean up.
16131 2013-02-04  Richard Guenther  <rguenther@suse.de>
16133         PR lto/56168
16134         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16135         node prevail as last resort.
16136         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16137         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16139 2013-02-04  Richard Biener  <rguenther@suse.de>
16141         PR tree-optimization/56113
16142         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16143         Merge into ...
16144         (equiv_class_lookup_or_add): ... this.
16145         (label_visit): Adjust and fix error in previous patch.
16146         (perform_var_substitution): Adjust.
16148 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16150         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16151         * config/sh/sh4-300.md: Likewise.
16152         * config/sh/sh4a.md: Likewise.
16153         * config/sh/constraints.md: Likewise.
16154         * config/sh/sh.md: Likewise.
16155         * config/sh/netbsd-elf.h: Likewise.
16156         * config/sh/predicates.md: Likewise.
16157         * config/sh/sh-protos.h: Likewise.
16158         * config/sh/ushmedia.h: Likewise.
16159         * config/sh/linux.h: Likewise.
16160         * config/sh/sh.c: Likewise.
16161         * config/sh/superh.h: Likewise.
16162         * config/sh/elf.h: Likewise.
16163         * config/sh/sh4.md: Likewise.
16164         * config/sh/sh.h: Likewise.
16166 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16168         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16169         constraint to match_test floating_point_store_memory_operand().
16170         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16171         (base14_operand): New.
16172         (floating_point_store_memory_operand): New.
16173         (integer_store_memory_operand): Revise to use base14_operand and
16174         reg_plus_base_memory_operand.
16175         (move_dest_operand): Allow symbolic_memory_operands.
16176         (symbolic_memory_operand): Check for LO_SOM.
16177         (symbolic_operand): Change default case to break.
16178         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16179         CONST_DOUBLE values to be reloaded by putting them into memory when
16180         the destination is a floating point register.
16181         (movdf): Remove code to handle CONST_DOUBLE.
16182         (movsf): Likewise.
16183         (reload_indf_r1): New.
16184         (reload_insf_r1): New.
16185         Consistently use "Q" and "T" constraints with integer and floating
16186         point move instructions, respectively.
16187         (movdi): Remove FAIL.
16188         Change predicate for source operand unamed DImode move from
16189         general_operand to move_src_operand.
16190         (umulsidi3): Change predicate for destination operand to
16191         register_operand.
16192         Likewise for similar unamed patterns.
16193         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16194         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16195         (hppa_legitimize_address): Simplify mask calculation.
16196         (pa_emit_move_sequence): Revised handling of secondary reloads from
16197         REG+D addresses for floating point loads and stores.  Directly handle
16198         loading CONST0_RTX (mode) to a floating point register.
16199         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16200         to floating point registers.  Don't restrict secondary reloads to
16201         floating point registers to integer modes.  Revise some comments and
16202         cleanup some code.
16203         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16204         (pa_legitimate_address_p): New.
16205         (pa_legitimize_reload_address): New.
16206         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16207         (STRICT_REG_OK_FOR_BASE_P): New.
16208         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16209         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16211 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16212             Andrew Dixie  <andrewd@gentrack.com>
16214         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16215         flag set.
16217 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16219         * expmed.c (extract_bit_field_1): Pass the full width of the
16220         structure to get_best_reg_extraction_insn.
16222 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16224         PR target/54601
16225         * configure.ac (use_cxa_atexit): Add AIX.
16226         * configure: Regenerate.
16228         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16230 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16232         PR debug/54793
16233         * final.c (need_profile_function): New variable.
16234         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16235         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16236         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16237         notes, targetm.asm_out.function_prologue doesn't emit anything,
16238         HAVE_prologue and profiler should be emitted before prologue,
16239         set need_profile_function instead of emitting it.
16240         (final_scan_insn): If need_profile_function, emit
16241         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16242         NOTE_INSN_FUNCTION_BEG note.
16244 2013-02-01  Richard Henderson  <rth@redhat.com>
16246         * config/rs6000/rs6000.md (smulditi3): New.
16247         (umulditi3): New.
16249         * config/alpha/alpha.md (umulditi3): New.
16251 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16253         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16254         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16256 2013-02-01  Richard Biener  <rguenther@suse.de>
16258         PR tree-optimization/56113
16259         * tree-ssa-structalias.c (label_visit): Reduce work for
16260         single-predecessor nodes.
16262 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16264         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16265         range isn't testing for zero.
16267 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16269         PR middle-end/56113
16270         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16272 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16273             Nick Clifton  <nickc@redhat.com>
16275         * config/v850/constraints.md (Q): Define as a memory constraint.
16276         * config/v850/predicates.md (label_ref_operand): New predicate.
16277         (e3v5_shift_operand): New predicate.
16278         (ior_operator): New predicate.
16279         * config/v850/t-v850: Add e3v5 multilib.
16280         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16281         (v850_gen_movdi): Prototype.
16282         * config/v850/v850.c: Add support for e3v5 architecture.
16283         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16284         TARGET_V850E_UP.
16285         (construct_save_jarl): Add e3v5 long JARL support.
16286         (v850_adjust_insn_length): New function.  Adjust length of call
16287         insns when using e3v5 instructions.
16288         (v850_gen_movdi): New function: Generate instructions to move a
16289         DImode value.
16290         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16291         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16292         (TARGET_USE_FPU): Enable for e3v5.
16293         (CONST_OK_FOR_W): New macro.
16294         (ADJUST_INSN_LENGTH): Define.
16295         * config/v850/v850.md (UNSPEC_LOOP): Define.
16296         (attr cpu): Add v850e3v5.
16297         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16298         (movdi): New pattern.
16299         (movdi_internal): New pattern.
16300         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16301         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16302         (cstoresf4): Likewise.
16303         (cstoredf4): Likewise.
16304         (insv): New pattern.
16305         (rotlso3_a): New pattern.
16306         (rotlsi3_b): New pattern
16307         (rotlsi3_v850e3v5): New pattern.
16308         (doloop_begin): New pattern.
16309         (fix_loop_counter): New pattern.
16310         (doloop_end): New pattern.
16311         (branch_normal): Add e3v5 long branch support.
16312         (branch_invert): Likewise.
16313         (branch_z_normal): Likewise.
16314         (branch_z_invert): Likewise.
16315         (branch_nz_normal): Likewise.
16316         (branch_nz_invert): Likewise.
16317         (call_internal_short): Add e3v5 register-indirect JARL support.
16318         (call_internal_long): Likewise.
16319         (call_value_internal_short): Likewise.
16320         (call_value_internal_long): Likewise.
16321         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16322         (mloop): New option.
16323         * config.gcc: Add support for configuring v840e3v5 target.
16324         * doc/invoke.texi: Document new v850 specific command line options.
16326 2013-01-31  Paul Koning  <ni1d@arrl.net>
16328         PR debug/55059
16329         PR debug/54508
16330         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16331         children if parent is a class.
16332         (prune_unused_types_prune): Don't add DW_AT_declaration.
16334 2013-01-31  Richard Biener  <rguenther@suse.de>
16336         PR tree-optimization/56157
16337         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16338         match up operand with SLP child.
16340 2013-01-31  Jason Merrill  <jason@redhat.com>
16342         PR debug/54410
16343         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16344         parameters the first time.
16345         (gen_scheduled_generic_parms_dies): Check completeness here.
16347 2013-01-31  Richard Biener  <rguenther@suse.de>
16349         PR middle-end/53073
16350         * common.opt (faggressive-loop-optimizations): New flag,
16351         enabled by default.
16352         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16353         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16354         infer_loop_bounds_from_undefined by it.
16356 2013-01-31  Richard Biener  <rguenther@suse.de>
16358         PR tree-optimization/56150
16359         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16360         visit virtual operands.
16361         (find_uses_to_rename_bb): Likewise.
16363 2013-01-31  Richard Biener  <rguenther@suse.de>
16365         PR tree-optimization/56150
16366         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16367         mixed store non-store stmts.
16369 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16371         PR sanitizer/55374
16372         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16373         LIBASAN_EARLY_SPEC is defined.
16374         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16375         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16376         before %o.
16377         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16379         PR c++/55742
16380         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16381         invalid args instead of ICEing on it.
16382         (ix86_valid_target_attribute_tree): Return error_mark_node if
16383         ix86_valid_target_attribute_inner_p failed.
16384         (ix86_valid_target_attribute_p): Return false only if
16385         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16386         target("default") attribute.
16387         (sorted_attr_string): Change argument from const char * to tree,
16388         merge in all target attribute arguments rather than just one.
16389         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16390         instead of free.  Avoid using strcat.
16391         (ix86_mangle_function_version_assembler_name): Mangle
16392         target("default") as if no target attribute is present.  Adjust
16393         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16394         instead of xmalloc and XDELETEVEC instead of free.
16395         (ix86_function_versions): Don't return true if one of the decls
16396         doesn't have target attribute.  If they don't and one of the decls
16397         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16398         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16399         (ix86_supports_function_versions): Remove.
16400         (make_name): Fix up formatting.
16401         (make_dispatcher_decl): Remove resolver_name and its initialization.
16402         Avoid leaking memory.
16403         (is_function_default_version): Return true if there is
16404         target("default") attribute rather than no target attribute at all.
16405         (make_resolver_func): Avoid leaking memory.
16406         (ix86_generate_version_dispatcher_body): Likewise.
16407         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16408         * target.def (supports_function_versions): Remove.
16409         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16410         * doc/tm.texi: Regenerated.
16412 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16414         PR rtl-optimization/56144
16415         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16416         for values with side effects.
16418 2013-01-30  Richard Biener  <rguenther@suse.de>
16420         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16421         (sparseset_pop): Likewise.
16422         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16423         to be able to use quick_push in the worker loop.
16425 2013-01-30  Marek Polacek  <polacek@redhat.com>
16427         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16429 2013-01-30  Richard Biener  <rguenther@suse.de>
16431         PR lto/56147
16432         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16434 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16436         PR tree-optimization/56064
16437         * fixed-value.c (fixed_from_double_int): New function.
16438         * fixed-value.h (fixed_from_double_int): New prototype.
16439         (const_fixed_from_double_int): New static inline function.
16440         * fold-const.c (native_interpret_fixed): New static function.
16441         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16442         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16443         (native_encode_fixed): New static function.
16444         (native_encode_expr) <FIXED_CST>: Use it.
16445         (native_interpret_int): Move double_int worker code to...
16446         * double-int.c (double_int::from_buffer): ...this new static method.
16447         * double-int.h (double_int::from_buffer): Prototype it.
16449 2013-01-30  Richard Biener  <rguenther@suse.de>
16451         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16452         New pointer-map and obstack.
16453         (init_alias_vars): Allocate pointer-map and obstack.
16454         (delete_points_to_sets): Free them.
16455         (find_what_var_points_to): Cache result.
16456         (find_what_p_points_to): Adjust for changed interface of
16457         find_what_var_points_to.
16458         (compute_points_to_sets): Likewise.
16459         (ipa_pta_execute): Likewise.
16461 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16463         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16464         * configure: Regenerate.
16465         * config.in: Regenerate.
16466         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16467         #nobits/#progbits if supported.
16469 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16471         PR target/56121
16472         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16473         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16474         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16476 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16478         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16479         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16481 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16483         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16484         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16486 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16488         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16489         declaration.
16490         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16491         * config/arm/cortex-a7.md: New bypasses using
16492         arm_mac_accumulator_is_result.
16494 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16496         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16497         (cortex_a7_neon_mla): Likewise.
16498         (cortex_a7_fpfmad): New reservation.
16499         (cortex_a7_fpmacs): Use ffmas and update required units.
16500         (cortex_a7_fpmuld): Update required units and latency.
16501         (cortex_a7_fpmacd): Likewise.
16502         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16503         (cortex_a7_neon). Likewise.
16504         (bypass) Update participating units.
16506 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16508         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16509         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16510         from fmac to ffma.
16511         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16512         (vfp_fmul): Use ffmad.
16513         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16514         (cortex_r4_fmacd): Use ffmad.
16515         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16516         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16517         (cortex_a9_fmacd): Use ffmad.
16518         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16519         (cortex_a8_vfp_macd): Use ffmad.
16520         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16521         (cortex_a5_fpmacd): Use ffmad.
16522         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16523         (cortex_a15_vfp_macd): Use ffmad.
16524         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16526 2013-01-29  Jason Merrill  <jason@redhat.com>
16528         PR libstdc++/54314
16529         * varasm.c (default_assemble_visibility): Don't warn about
16530         visibility on artificial decls.
16532 2013-01-29  Richard Biener  <rguenther@suse.de>
16534         PR tree-optimization/56113
16535         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16536         the bitmap leader.
16537         (label_visit): Free duplicate bitmaps and record the leader instead.
16538         (perform_var_substitution): Adjust.
16540 2013-01-29  Richard Biener  <rguenther@suse.de>
16542         PR tree-optimization/55270
16543         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16544         the CFG, schedule loops for fixup.
16546 2013-01-29  Nick Clifton  <nickc@redhat.com>
16548         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16549         SP_REG.
16551 2013-01-28  Leif Ekblad  <leif@rdos.net>
16553         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16554         * config/i386/i386.h (TARGET_RDOS): New macro.
16555         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16556         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16557         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16558         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16559         DEFAULT_LARGE_SECTION_THRESHOLD.
16560         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16561         * config/i386/rdos.h: New file.
16562         * config/i386/rdos64.h: New file.
16564 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16566         PR other/54814
16567         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16568         TEST_HARD_REG_BIT.
16570 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16572         PR rtl-optimization/56117
16573         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16574         call cselib_lookup_from_insn on the MEM before calling
16575         add_insn_mem_dependence.
16577 2013-01-28  Richard Biener  <rguenther@suse.de>
16579         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16580         to a stmt that didn't have one.
16581         (copy_phis_for_bb): Likewise for PHI arguments.
16582         (copy_debug_stmt): Likewise for debug stmts.
16584 2013-01-28  Richard Biener  <rguenther@suse.de>
16586         PR tree-optimization/56034
16587         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16588         (partition_builtin_p): Adjust.
16589         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16590         it is the last partition.
16591         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16592         up the vertex for the definition.
16593         (classify_partition): Classify whether a partition is a
16594         PKIND_REDUCTION, thus has uses outside of the loop.
16595         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16596         Merge all PKIND_REDUCTION partitions into the last partition.
16597         (tree_loop_distribution): Seed partitions from reductions as well.
16599 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16601         PR tree-optimization/56125
16602         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16603         pow(x,c) into sqrt(x) * powi(x, n/2) or
16604         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16605         optimizing for size.
16606         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16607         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16608         integer.
16610         PR tree-optimization/56094
16611         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16612         to UNKNOWN_LOCATION while gimplifying expr.
16614 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16616         PR target/56114
16617         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16618         operand 0 in movabs insn template for -masm=intel asm alternative.
16619         (*movabs<mode>_2): Ditto for operand 1.
16621 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16623         PR target/54663
16624         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16625         of microblaze-c.o
16627 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16629         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16630         tm_file.
16632 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16634         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16635         Undef to avoid warning.
16637 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16639         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16640         * configure: Regenerate.
16642 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16644         PR tree-optimization/56098
16645         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16646         for stmts with volatile ops.
16647         (cond_store_replacement): Don't optimize if assign has volatile ops.
16648         (cond_if_else_store_replacement_1): Don't optimize if either
16649         then_assign or else_assign have volatile ops.
16650         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16651         volatile ops.
16653 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16655         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16657 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16659         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16660         missing ':' in asm example.
16662 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
16664         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16665         entries into lane and laneq entries.
16666         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16667         Remove AdvSIMD scalar modes.
16668         (aarch64_sq<r>dmulh_laneq<mode>): New.
16669         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16670         modes.
16671         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16672         builtin implementations to relfect changes in RTL in aarch64-simd.md.
16673         * config/aarch64/iterators.md (VCOND): New.
16674         (VCONQ): New.
16676 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16678         PR target/54222
16679         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16680         Add NULL LIBNAME argument to existing definitions.
16681         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16682         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16683         * config/avr/avr.c (DEF_BUILTIN): Same.
16684         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16685         (avr_expand_builtin): Expand to a vanilla call if a libgcc
16686         implementation is available (DECL_ASSEMBLER_NAME is set).
16687         (avr_fold_absfx): New static function.
16688         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16689         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16690         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16691         AVR_BUILTIN_ABSLLK.
16692         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16693         (abshk, absk, abslk, absllk): Provide as static inline functions.
16695 2013-01-25  Marek Polacek  <polacek@redhat.com>
16697         PR tree-optimization/56035
16698         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16700 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
16702         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16703         (*movtf_internal_rex64): Add (!o,C) alternative
16704         (*movxf_internal_rex64): Ditto.
16705         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16707 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
16709         * doc/invoke.texi: fix typo.
16710         * doc/objc.texi: fix typo.
16712 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
16714         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16715         for the first two alternatives.
16717 2013-01-24  Diego Novillo  <dnovillo@google.com>
16719         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
16720         (ggc-zone.o): Remove.
16721         * configure.ac: Remove option --with-gc.
16722         * configure: Re-generate.
16723         * doc/install.texi: Remove documentation for --with-gc.
16724         * gengtype.c (write_enum_defn): Remove.  Update all users.
16725         (write_Types_process_field): Remove generation of gt_e_* argument.
16726         (output_type_enum): Remove.  Update all users.
16727         (write_enum_defn): Remove.  Update all users.
16728         (enum alloc_zone): Remove.  Update all users.
16729         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16730         * ggc-common.c (ggc_splay_alloc): Remove first argument.
16731         Update all callers.
16732         (struct ptr_data): Remove field TYPE.  Update all users.
16733         (gt_pch_note_object): Remove argument TYPE.  Update all users.
16734         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16735         Update all users.
16736         * ggc-none.c (ggc_alloc_typed_stat): Remove.
16737         (struct alloc_zone): Remove.
16738         (ggc_internal_alloc_zone_stat): Remove.
16739         (ggc_internal_cleared_alloc_zone_stat): Remove.
16740         * ggc-page.c (ggc_alloc_typed_stat): Remove.
16741         (ggc_pch_count_object): Remove last argument.  Update all users.
16742         (ggc_pch_alloc_object): Remove last argument.  Update all users.
16743         (struct alloc_zone): Remove.
16744         * ggc-zone.c: Remove.
16745         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
16746         (struct alloc_zone): Remove.
16747         (ggc_alloc_typed_stat): Remove.
16748         (ggc_alloc_typed): Remove.
16749         (ggc_splay_alloc): Remove first argument.
16750         (rtl_zone): Remove.  Update all users.
16751         (tree_zone): Remove.  Update all users.
16752         (tree_id_zone): Remove.  Update all users.
16753         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
16754         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
16755         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
16756         * tree-ssanames.c: Remove references to zone allocator in comments.
16758 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
16760         * config/avr/avr.c (avr_out_fract): Make register numbers that
16761         might be outside of source operand signed.
16763 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
16765         * config/i386/constraints.md (Yf): New constraint.
16766         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16767         of f constraint to conditionaly disable x87 register preferences.
16768         (*movdf_internal): Ditto.
16769         (*movsf_internal): Ditto.
16771 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
16773         PR inline-asm/55934
16774         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16775         that have operands with impossible constraints.
16776         Add a FIXME for a speed-up opportunity.
16777         * lra-constraints.c (process_alt_operands): Verify that a class
16778         selected from constraints on asms is valid for the operand mode.
16779         (curr_insn_transform): Remove incorrect comment.
16781 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
16783         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16784         TOC operand is a valid symbol ref in the constant pool.
16786 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16788         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
16790 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
16792         PR target/54222
16793         * config/avr/stdfix.h: New file.
16794         * t-avr (stdfix-gcc.h): New rule to build it.
16795         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16797 2013-01-23  Kostya Serebryany  <kcc@google.com>
16799         * config/darwin.h: remove dependency on
16800         CoreFoundation (asan on Mac OS).
16802 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
16804         PR target/49069
16805         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16806         instead of cmpdi_operand for first comparison operand.
16807         Don't assert that comparison operands aren't both constants.
16809 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16811         * doc/install.texi (Downloading the Source): Update references to
16812         downloading separate components.
16814 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16816         * doc/extend.texi (__int128): Improve grammar.
16818 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
16820         PR target/56028
16821         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16822         alternative to (o,r).
16823         (*movdi_internal_rex64): Remove (!o,n) alternative.
16824         (DImode immediate->memory splitter): Remove.
16825         (DImode immediate->memory peephole2): Remove.
16826         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16827         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16828         alternative to (!o,*r).
16829         (*movtf_internal_sse): New pattern.
16830         (*movxf_internal_rex64): New pattern.
16831         (*movxf_internal): Disable for TARGET_64BIT.
16832         (*movdf_internal_rex64): Remove (!o,F) alternative.
16834 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
16836         PR middle-end/56074
16837         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16838         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16839         * tree-vect-loop-manip.c (find_loop_location): Also ignore
16840         stmt locations where LOCATION_LOCUS of the stmt location is
16841         UNKNOWN_LOCATION or BUILTINS_LOCATION.
16843         PR target/55686
16844         * config/i386/i386.md (UNSPEC_STOS): New.
16845         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16846         *strsetqi_1): Add UNSPEC_STOS.
16848 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
16850         PR c++/56067
16851         * doc/invoke.texi: Remove left over -Wsynth example.
16853 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16855         PR tree-optimization/56051
16856         * fold-const.c (fold_binary_loc): Don't fold
16857         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16858         a narrowing conversion, or widening conversion from signed
16859         to unsigned.
16861 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16863         PR rtl-optimization/56023
16864         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16865         dependent on debug instruction.
16867 2013-01-21  Martin Jambor  <mjambor@suse.cz>
16869         PR middle-end/56022
16870         * function.c (allocate_struct_function): Call
16871         invoke_set_current_function_hook earlier.
16873 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16875         * reload1.c (init_reload): Only initialize reload_obstack
16876         during the first call.
16878 2013-01-21  Marek Polacek  <polacek@redhat.com>
16880         * cfgloop.c (verify_loop_structure): Fix up grammar.
16882 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
16884         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16885         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16887 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16889         PR target/56058
16890         * config/arm/marvell-pj4.md: Update copyright year.
16891         Fix up use of alu to alu_reg and simple_alu_imm.
16893 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16895         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16897 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
16899         PR target/55433
16900         * lra-constraints.c (curr_insn_transform): Don't reuse original
16901         insn for secondary memory move when memory mode should be different.
16903 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16905         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16906         atomic_storedi_1): New patterns.
16908 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
16910         btver2 pipeline descriptions.
16911         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16912         descriptions.
16913         * config/i386/i386.md (btver2_decode): New type attributes.
16914         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
16915         type attributes.
16916         * config/i386/btver2.md: New file describing btver2 pipelines.
16918 2013-01-19  Andrew Pinski  <apinski@cavium.com>
16920         PR tree-optimization/52631
16921         * tree-ssa-sccvn (visit_use): Before looking up the original
16922         statement, try looking up the simplified expression.
16924 2013-01-19  Anthony Green  <green@moxielogic.com>
16926         * config/moxie/moxie.c (moxie_expand_prologue): Set
16927         current_function_static_stack_size.
16929 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16931         PR tree-optimization/56029
16932         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16933         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16935 2013-01-18  Sharad Singhai  <singhai@google.com>
16937         PR tree-optimization/55995
16938         * dumpfile.c (dump_loc): Print location only if available.
16939         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16941 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
16943         PR target/55433
16944         * lra-constraints.c (curr_insn_transform): Reuse original insn for
16945         secondary memory move.
16946         (inherit_reload_reg): Use rclass instead of cl for
16947         check_secondary_memory_needed_p.
16949 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16951         PR middle-end/56015
16952         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
16953         the case where writing real complex part of target modifies op1.
16955 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
16957         * config/aarch64/aarch64-simd.md
16958         (aarch64_vcond_internal<mode>): Handle unordered cases.
16959         * config/aarch64/iterators.md (v_cmp_result): New.
16961 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
16962             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16964         * config/arm/marvell-pj4.md: New file.
16965         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16966         * config/arm/arm.md (generic_sched): Add marvell_pj4.
16967         (generic_vfp): Likewise.
16968         * config/arm/arm-cores.def: Add marvell-pj4.
16969         * config/arm/arm-tune.md: Regenerate.
16970         * config/arm/arm-tables.opt: Regenerate.
16971         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16972         * doc/invoke.texi: Document marvell-pj4.
16974 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
16976         * config/aarch64/arm_neon.h: Map scalar types to standard types.
16978 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16980         PR debug/54114
16981         PR debug/54402
16982         PR debug/49888
16983         * var-tracking.c (negative_power_of_two_p): New.
16984         (global_get_addr_cache, local_get_addr_cache): New.
16985         (get_addr_from_global_cache, get_addr_from_local_cache): New.
16986         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
16987         heading comment.
16988         (vt_stack_offset_p): Remove.
16989         (vt_canon_true_dep): Always canonicalize loc's address.
16990         (clobber_overlapping_mems): Make sure we have a MEM.
16991         (local_get_addr_clear_given_value): New.
16992         (val_reset): Clear local cached entries.
16993         (compute_bb_dataflow): Create and release the local cache.
16994         Disable duplicate MEMs clobbering.
16995         (emit_notes_in_bb): Clobber MEMs likewise.
16996         (vt_emit_notes): Create and release the local cache.
16997         (vt_initialize, vt_finalize): Create and release the global
16998         cache, respectively.
16999         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17001 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17003         PR libmudflap/53359
17004         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17005         not found in the symtab.
17007 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17009         PR debug/56006
17010         PR rtl-optimization/55547
17011         PR rtl-optimization/53827
17012         PR debug/53671
17013         PR debug/49888
17014         * alias.c (offset_overlap_p): New, factored out of...
17015         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17016         the conservative special case for symbolic constants.  Don't
17017         adjust zero sizes on alignment.
17019 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17021         PR rtl-optimization/52573
17022         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17023         REG_UNUSED for the same register.
17025 2013-01-17  Richard Biener  <rguenther@suse.de>
17026             Marek Polacek  <polacek@redhat.com>
17028         PR rtl-optimization/55833
17029         * loop-unswitch.c (unswitch_loops): Move loop verification...
17030         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17031         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17032         Set it to true when we're removing a loop from hierarchy tree in
17033         an irreducible region.
17034         (fix_bb_placements): Adjust caller.
17035         (fix_loop_placements): Likewise.
17037 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17039         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17040         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17041         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17042         Remove ID.  Adjust comments.
17043         * config/avr/avr-c.c (avr_builtin_name): Remove.
17044         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17045         * config/avr/avr.c (avr_tolower): New static function.
17046         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17047         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17048         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17049         default expansion.
17051 2013-01-17  Jan Hubicka  <jh@suse.cz>
17053         PR tree-optimization/55273
17054         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17056 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17058         PR target/55981
17059         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17060         store through atomic_store<mode>_1.
17061         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17063 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17065         PR tree-optimizations/55264
17066         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17067         for virtual methods.
17068         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17069         virtual methods before inlining is over.
17070         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17071         virtual functions.
17072         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17073         non-virtual.
17075 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17077         PR rtl-optimization/56005
17078         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17079         pending reads for prefetch.
17081 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17083         * config/aarch64/aarch64.md
17084         (*cstoresi_neg_uxtw): New pattern.
17085         (*cmovsi_insn_uxtw): New pattern.
17086         (*<optab>si3_uxtw): New pattern.
17087         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17088         (*<optab>si3_insn_uxtw): New pattern.
17089         (*bswapsi2_uxtw): New pattern.
17091 2013-01-16  Richard Biener  <rguenther@suse.de>
17093         * tree-inline.c (tree_function_versioning): Remove set but
17094         never used variable.
17096 2013-01-16  Richard Biener  <rguenther@suse.de>
17098         PR tree-optimization/55964
17099         * tree-flow.h (rename_variables_in_loop): Remove.
17100         (rename_variables_in_bb): Likewise.
17101         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17102         (copy_loop_before): Adjust and delete update-ssa status.
17103         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17104         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17105         (rename_variables_in_loop): Remove.
17106         (slpeel_update_phis_for_duplicate_loop): Likewise.
17107         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17108         use available cfg machinery instead of duplicating it.
17109         Update PHI nodes and perform poor-mans SSA update here.
17110         (slpeel_tree_peel_loop_to_edge): Adjust.
17112 2013-01-16  Richard Biener  <rguenther@suse.de>
17114         PR tree-optimization/54767
17115         PR tree-optimization/53465
17116         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17117         (vrp_visit_phi_node): For PHI arguments coming via backedges
17118         drop all symbolical range information.
17119         (execute_vrp): Compute backedges.
17121 2013-01-16  Richard Biener  <rguenther@suse.de>
17123         * doc/install.texi: Update CLooG and ISL requirements to
17124         0.18.0 and 0.11.1.
17126 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17128         PR target/55301
17129         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17130         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17131         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17133 2013-01-16  DJ Delorie  <dj@redhat.com>
17135         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17136         (UNSPECV_SP_SWITCH_E): New.
17137         (sp_switch_1): Change to an unspec.
17138         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17139         replace $r15.
17141 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17143         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17144         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17145         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17146         (expand_mem_thread_fence): Ditto.
17147         (expand_mem_signal_fence): Ditto.
17148         (expand_atomic_load): Ditto.
17149         (expand_atomic_store): Ditto.
17151 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17153         PR rtl-optimization/55547
17154         PR rtl-optimization/53827
17155         PR debug/53671
17156         PR debug/49888
17157         * alias.c (memrefs_conflict_p): Set sizes to negative after
17158         AND adjustments.
17160 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17162         PR target/55940
17163         * function.c (thread_prologue_and_epilogue_insns): Always
17164         add crtl->drap_reg to set_up_by_prologue.set, even if
17165         stack_realign_drap is false.
17167 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17169         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17170         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17171         *call): Fix indention.
17173 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17175         PR target/55876
17176         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17177         Update comment.
17179 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17181         PR rtl-optimization/55153
17182         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17184 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17186         PR tree-optimization/55920
17187         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17188         accesses as grp_to_be_debug_replaced.
17190 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17192         PR tree-optimization/55920
17193         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17194         there is non-useless type conversion needed from debug rhs to lhs,
17195         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17197 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17198             Mikael Pettersson  <mikpe@it.uu.se>
17200         PR target/43961
17201         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17202         Thumb.
17203         (ASM_OUTPUT_CASE_LABEL): Remove.
17204         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17205         * final.c (shorten_branches): Update alignment of labels before
17206         jump tables if CASE_VECTOR_SHORTEN_MODE.
17208 2013-01-15  Richard Biener  <rguenther@suse.de>
17210         PR bootstrap/55961
17211         * system.h: Do not include gmp.h for building host tools.
17213 2013-01-15  Richard Biener  <rguenther@suse.de>
17215         PR middle-end/55882
17216         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17217         account for bitpos when computing alignment.
17219 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17221         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17222         (ix86_target_macros_internal): Likewise.
17224         * config/i386/i386.c (m_CORE2I7): Removed.
17225         (m_CORE_HASWELL): New macro.
17226         (m_CORE_ALL): Likewise.
17227         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17228         (initial_ix86_arch_features): Likewise.
17229         (processor_target_table): Initializations for Core avx2.
17230         (cpu_names): New names "core-avx2".
17231         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17232         PROCESSOR_CORE_HASWELL.
17233         (ix86_issue_rate): New case.
17234         (ia32_multipass_dfa_lookahead): Likewise.
17235         (ix86_sched_init_global): Likewise.
17237         * config/i386/i386.h (TARGET_HASWELL): New macro.
17238         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17239         (processor_type): New PROCESSOR_HASWELL.
17241 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17243         PR tree-optimization/55955
17244         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17245         *SHIFT_EXPR and *ROTATE_EXPR codes.
17247         PR tree-optimization/48766
17248         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17249         -ftrapv disable -fwrapv.
17251 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17253         PR target/55974
17254         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17255         etc. to 1 and not to __flash.
17256         Use LL suffix for __INT24_MAX__ with -mint8.
17257         Use ULL suffix for __UINT24_MAX__ with -mint8.
17259 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17261         * config/avr/avr-arch.h
17262         (struct base_arch_s): Use typedef avr_arch_t instead.
17263         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17264         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17265         * config/avr/avr.c: Same.
17266         * config/avr/avr-devices.c: Same.
17267         * config/avr/driver-avr.c: Same.
17268         * config/avr/gen-avr-mmcu-texi.c: Same.
17269         * config/avr/avr-mcus.def: Adjust comment.
17271 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17273         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17274         * config/aarch64/iterators.md (VALLDI): New.
17276 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17277             Andi Kleen  <ak@linux.intel.com>
17279         PR target/55948
17280         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17281         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17282         memmodel flag.
17284 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17286         * config/avr/avr-stdint.h: Remove trailing blanks.
17287         * config/avr/avr-log.h: Same.
17288         * config/avr/avr-arch.h: Same.
17289         * config/avr/avr-devices.c: Same.
17290         * config/avr/avr-dimode.md: Same.
17291         * config/avr/predicates.md: Same.
17292         * config/avr/avr-c.c: Same.  And fix typo.
17294         * config/avr/avr-protos.h: Same.  And:
17295         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17296         (init_cumulative_args): Rename to avr_init_cumulative_args.
17297         (expand_prologue): Rename to avr_expand_prologue.
17298         (expand_epilogue): Rename to avr_expand_epilogue.
17299         (adjust_insn_length): Rename to avr_adjust_insn_length.
17300         (notice_update_cc): Rename to avr_notice_update_cc.
17301         (final_prescan_insn): Rename to avr_final_prescan_insn.
17302         * config/avr/avr.c: Same.
17303         * config/avr/avr.h: Same.
17304         * config/avr/avr.md: Remove trailing blanks.
17305         (prologue): Use avr_expand_prologue.
17306         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17308 2013-01-14  Richard Biener  <rguenther@suse.de>
17310         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17311         verify_location, collect_subblocks): New functions.
17312         (verify_gimple_in_cfg): Verify that locations only reference
17313         BLOCKs in the functions BLOCK tree.
17315 2013-01-14  Richard Biener  <rguenther@suse.de>
17317         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17318         PHI argument.
17319         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17320         unshare reference.
17321         (insert_out_of_ssa_copy_on_edge): Likewise.
17322         (rewrite_close_phi_out_of_ssa): Likewise.
17323         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17324         debug expressions.
17325         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17326         propagated constants.
17327         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17328         can not be shared.
17330 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17332         * config/avr/avr-modes.def: Add GPL copyright notice.
17334 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17336         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17337         MEMMODEL_MASK to determine memory model.
17338         (atomic_store<mode>): Ditto from operands[2].
17339         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17341 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17343         PR fortran/55935
17344         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17345         (fold_gimple_assign): Don't call unshare_expr here.
17346         (fold_ctor_reference): Call unshare_expr.
17348 2013-01-13  Terry Guo  <terry.guo@arm.com>
17350         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17351         * doc/fragments.texi: Document MULTILIB_REUSE.
17352         * gcc.c (multilib_reuse): New internal spec.
17353         (set_multilib_dir): Also search multilib from multilib_reuse.
17354         * genmultilib (tmpmultilib3): Refactor code.
17355         (tmpmultilib4): Ditto.
17356         (multilib_reuse): New multilib argument.
17358 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17360         * Makefile.in: Update copyright.
17362 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17364         PR middle-end/55890
17365         * calls.c (expand_call): Check if arg_nr is valid.
17367 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17369         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17370         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17371         documentation.  Add missing '__' in front of
17372         __builtin_ia32_packssdw256.
17374 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17376         PR target/55719
17377         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17378         NO_REGS for larl operands.
17379         (s390_reload_larl_operand): Use s390_load_address instead of
17380         emit_move_insn.
17382 2013-01-11  Richard Biener  <rguenther@suse.de>
17384         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17385         (verify_node_sharing): ... here.
17386         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17388 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17390         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17391         Substitute TREECHECKING.
17392         * configure: Regenerate.
17393         * Makefile.in (TREECHECKING): New.
17395 2013-01-11  Richard Guenther  <rguenther@suse.de>
17397         PR tree-optimization/44061
17398         * tree-vrp.c (extract_range_basic): Compute zero as
17399         value-range for __builtin_constant_p of function parameters.
17401 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17403         Update copyright years.
17405 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17407         PR rtl-optimization/55672
17408         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17409         const to be eliminable.
17411 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17413         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17414         * configure: Regenerate.
17416 2013-01-10  Richard Biener  <rguenther@suse.de>
17418         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17420 2013-01-10  Richard Biener  <rguenther@suse.de>
17422         PR bootstrap/55792
17423         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17424         locations for virtual PHI arguments.
17425         (rewrite_update_phi_arguments): Likewise.
17427 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17429         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17430         on to assembler.
17432 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17434         PR tree-optimization/55921
17435         * tree-complex.c (expand_complex_asm): New function.
17436         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17438 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17440         PR target/55718
17441         * config/s390/s390.c (s390_symref_operand_p)
17442         (s390_loadrelative_operand_p): Merge the two functions.
17443         (s390_check_qrst_address, print_operand_address): Add parameters
17444         to s390_loadrelative_operand_p invokation.
17445         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17446         (s390_reload_larl_operand, s390_secondary_reload): Use
17447         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17448         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17450 2013-01-09  Mike Stump  <mikestump@comcast.net>
17452         * dse.c (record_store): Remove unnecessary assert.
17454 2013-01-09  Jan Hubicka  <jh@suse.cz>
17456         PR tree-optimization/55569
17457         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17458         * cfgloop.h (scale_loop_profile): Likewise.
17460 2013-01-09  Jan Hubicka  <jh@suse.cz>
17462         PR lto/45375
17463         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17464         functions.
17465         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17467 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17469         PR middle-end/55114
17470         * expr.h (maybe_emit_group_store): Declare.
17471         * expr.c (maybe_emit_group_store): New function.
17472         * builtins.c (expand_builtin_int_roundingfn): Call it.
17473         (expand_builtin_int_roundingfn_2): Likewise.
17475 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17477         PR rtl-optimization/55829
17478         * lra-constraints.c (match_reload): Add code for absent output.
17479         (curr_insn_transform): Add code for reloads of matched inputs
17480         without output.
17482 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17484         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17485         attribute of movddup insn to DF.
17486         (*vec_interleave_lowv2df): Ditto.
17487         (vec_dupv2df): Ditto.
17489 2013-01-09  Jan Hubicka  <jh@suse.cz>
17491         PR tree-optimiation/55875
17492         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17493         EVERY_ITERATION parameter.
17494         (number_of_iterations_exit): Check if exit is executed every iteration.
17495         (idx_infer_loop_bounds): Similarly here.
17496         (n_of_executions_at_most): Simplify
17497         to only test for cases where statement is dominated by the
17498         particular bound; handle correctly the "postdominance" test.
17499         (scev_probably_wraps_p): Use max loop iterations info
17500         as a global bound first.
17502 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17503             Nick Clifton  <nickc@redhat.com>
17505         * config/v850/v850.md (cbranchsf4): New pattern.
17506         (cstoresf4): New pattern.
17507         (cbranchdf4): New pattern.
17508         (cstoredf4): New pattern.
17509         (movsicc): Disallow floating point comparisons.
17510         (cmpsf_le_insn): Fix order of operators.
17511         (cmpsf_lt_insn): Likewise.
17512         (cmpsf_eq_insn): Likewise.
17513         (cmpdf_le_insn): Likewise.
17514         (cmpdf_lt_insn): Likewise.
17515         (cmpdf_eq_insn): Likewise.
17516         (cmpsf_ge_insn): Use LE comparison.
17517         (cmpdf_ge_insn): Likewise.
17518         (cmpsf_gt_insn): Use LT comparison.
17519         (cmpdf_gt_insn): Likewise.
17520         (cmpsf_ne_insn): Delete pattern.
17521         (cmpdf_ne_insn): Delete pattern.
17522         * config/v850/v850.c (v850_gen_float_compare): Use
17523         gen_cmpdf_eq_insn for NE comparison.
17524         (v850_float_z_comparison_operator)
17525         (v850_float_nz_comparison_operator): Move from here ...
17526         * config/v850/predicates.md: ... to here.  Move GT and GE
17527         comparisons into v850_float_z_comparison_operator.
17528         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17529         Delete prototype.
17530         (v850_float_nz_comparison_operator): Likewise.
17532 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17534         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17535         with calls to gen_insvsi/gen_insvdi.
17537 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17539         * config/i386/i386.c (initial_ix86_tune_features): Set up
17540         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17542 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17543             Jakub Jelinek  <jakub@redhat.com>
17545         PR tree-optimization/48189
17546         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17547         If nitercst is 0, don't predict the exit edge.
17549 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17551         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17552         in asm_fprintf with reg_names.
17553         (aarch64_print_operand_address): Likewise.
17554         (aarch64_return_addr): Likewise.
17555         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17557 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17559         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17560         (INT_U6_BITS): Likewise.
17561         * config/pa/predicates.md (uint6_operand): New predicate.
17562         (shift5_operand, shift6_operand): Likewise.
17563         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17564         arith32_operand.
17565         (lshrdi3): Use shift6_operand.
17566         (shrpsi4, shrpdi4): New insn patterns.
17567         (extzv): Delete expander.
17568         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17569         predicates in unamed zero extract patterns.  Tighten common constraint.
17570         (extv): Delete expander.
17571         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17572         predicates in unamed sign extract patterns.  Tighten common constraint.
17573         (insv): Delete expander.
17574         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17575         predicates in unamed insert patterns.  Tighten common constraint.
17576         Change uint32_operand predicate to uint6_operand predicate in unamed
17577         DImode pattern to insert constant values of type 1...1xxxx.
17579 2013-01-04  Jan Hubicka  <jh@suse.cz>
17581         PR tree-optimization/55823
17582         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17583         issue.
17585 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17586             Uros Bizjak  <ubizjak@gmail.com>
17588         PR rtl-optimization/55845
17589         * df-problems.c (can_move_insns_across): Stop scanning at
17590         volatile_insn_p source instruction or give up if
17591         across_from .. across_to range contains any volatile_insn_p
17592         instructions.
17594 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17596         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17597         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17598         Declare.
17599         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17600         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17602 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17604         PR fortran/55341
17605         * asan.c (asan_clear_shadow): New function.
17606         (asan_emit_stack_protection): Use it.
17608 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17610         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17611         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17612         with tab instead of space.
17614 2013-01-08  Nick Clifton  <nickc@redhat.com>
17616         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17617         register bank 0 at the start of an interrupt handler.
17618         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17619         MDBH registers.
17621 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17623         * config/aarch64/aarch64-simd.md
17624         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17625         (aarch64_simd_bsl): Likewise.
17626         (aarch64_vcond_internal<mode>): Likewise.
17627         (vcond<mode><mode>): Likewise.
17628         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17629         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17631 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17633         * config/aarch64/aarch64-builtins.c
17634         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17636 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17638         PR debug/55579
17639         * tree-sra.c (analyze_access_subtree): Return true also after
17640         potentially creating a debug-only replacement.
17642 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17644         PR middle-end/55890
17645         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17647         PR tree-optimization/54120
17648         * tree-vrp.c (range_fits_type_p): Don't allow
17649         src_precision < precision from signed vr to unsigned_p
17650         if vr->min or vr->max is negative.
17651         (simplify_float_conversion_using_ranges): Test can_float_p
17652         against CODE_FOR_nothing.
17654 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17655             Richard Biener  <rguenther@suse.de>
17657         PR middle-end/55851
17658         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17659         types instead of just INTEGER_TYPE types.
17661 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
17663         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17664         TF_SIZE): Define.
17666 2013-01-07  Steve Ellcey  <sellcey@mips.com>
17668         PR target/42661
17669         * config/mips/mips.opt: Change mad to mmad to match documentation.
17671 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17673         PR target/55897
17674         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17675         .progmemx.data now.
17677 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17679         PR target/55897
17680         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17681         (avr_addrspace_t): Add .section_name field.
17682         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17683         array size.
17684         (avr_addrspace): Same.  Initialize .section_name.  Remove last
17685         NULL entry.  Put __memx into .progmemx.data.
17686         (progmem_section_prefix): Remove.
17687         (avr_asm_init_sections): No need to initialize progmem_section.
17688         (avr_asm_named_section): Use avr_addrspace[].section_name to get
17689         section name prefix.
17690         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
17691         retrieve the progmem section.
17692         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17693         boundary to run over avr_addrspace[].
17694         (avr_register_target_pragmas): Ditto.
17696 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
17698         * varasm.c (output_constant_def_contents): For asan_protect_global
17699         protected strings, adjust DECL_ALIGN if needed, before testing for
17700         anchored symbols.
17701         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17702         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
17703         normal decls.
17704         (output_object_block): For asan protected decls, emit asan padding
17705         after their contents.
17706         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17707         (asan_finish_file): Test it here instead.
17709 2013-01-07  Nick Clifton  <nickc@redhat.com>
17710             Matthias Klose  <doko@debian.org>
17711             Doug Kwan  <dougkwan@google.com>
17712             H.J. Lu  <hongjiu.lu@intel.com>
17714         PR driver/55470
17715         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17717         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17719         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17721         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17723         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17725 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17727         PR target/54461
17728         * doc/install.texi (Cross-Compiler-Specific Options): Document
17729         --with-avrlibc.
17731 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
17733         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17734         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17735         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17736         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17737         vqmovun_high_s64): Fix source operand number and update copyright.
17739 2013-01-07  Richard Biener  <rguenther@suse.de>
17741         PR middle-end/55890
17742         * gimple.h (gimple_call_builtin_p): New overload.
17743         * gimple.c (validate_call): New function.
17744         (gimple_call_builtin_p): Likewise.
17745         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17746         Use gimple_call_builtin_p.
17747         (find_func_clobbers): Likewise.
17748         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17749         (strlen_optimize_stmt): Likewise.
17751 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17753         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17754         (vld1q_dup_*): Likewise.
17755         (vld1_*): Likewise.
17756         (vld1q_*): Likewise.
17757         (vld1_lane_*): Likewise.
17758         (vld1q_lane_*): Likewise.
17760 2013-01-07  Richard Biener  <rguenther@suse.de>
17762         * lto-streamer.h (LTO_minor_version): Bump to 2.
17764 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17766         * config/aarch64/aarch64-protos.h
17767         (aarch64_const_double_zero_rtx_p): Rename to...
17768         (aarch64_float_const_zero_rtx_p): ...this.
17769         (aarch64_float_const_representable_p): New.
17770         (aarch64_output_simd_mov_immediate): Likewise.
17771         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17772         move immediate case.
17773         * config/aarch64/aarch64.c
17774         (aarch64_const_double_zero_rtx_p): Rename to...
17775         (aarch64_float_const_zero_rtx_p): ...this.
17776         (aarch64_print_operand): Allow printing of new constants.
17777         (aarch64_valid_floating_const): New.
17778         (aarch64_legitimate_constant_p): Check for valid floating-point
17779         constants.
17780         (aarch64_simd_valid_immediate): Likewise.
17781         (aarch64_vect_float_const_representable_p): New.
17782         (aarch64_float_const_representable_p): Likewise.
17783         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17784         (aarch64_output_simd_mov_immediate): New.
17785         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17786         (*movdf_aarch64): Likewise.
17787         * config/aarch64/constraints.md (Ufc): New.
17788         (Y): call aarch64_float_const_zero_rtx.
17789         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17791 2013-01-07  Richard Biener  <rguenther@suse.de>
17793         PR tree-optimization/55888
17794         PR tree-optimization/55862
17795         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17796         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17797         not if it is contained therein.
17799 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17801         * config/avr/t-avr: Typo.
17803 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17805         PR55243
17806         * config/avr/t-avr: Don't automatically rebuild
17807         $(srcdir)/config/avr/t-multilib
17808         $(srcdir)/config/avr/avr-tables.opt
17809         $(srcdir)/doc/avr-mmcu.texi
17810         (avr-mcus): New phony target to build them on request.
17811         (s-avr-mlib, s-avr-mmcu-texi): Remove.
17812         * avr/avr-mcus.def: Adjust comments.
17814 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
17816         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17818 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17820         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17822 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17824         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17826 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
17828         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17829         to generate profiling.
17830         * config/rs6000/aix64.h (LIB_SPEC): Same.
17832 2013-01-04  Andrew Pinski  <apinski@cavium.com>
17834         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17835         New function.
17836         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17838 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
17840         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17841         unconditionally.
17842         (ix86_expand_move): Ditto.
17843         (ix86_zero_extend_to_Pmode): Ditto.
17844         (ix86_expand_call): Ditto.
17845         (ix86_expand_special_args_builtin): Ditto.
17846         (ix86_expand_builtin): Ditto.
17848 2013-01-04  Richard Biener  <rguenther@suse.de>
17850         PR tree-optimization/55862
17851         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17852         translating them through PHI nodes.
17854 2013-01-04  Martin Jambor  <mjambor@suse.cz>
17856         PR tree-optimization/55755
17857         * tree-sra.c (sra_modify_assign): Do not check that an access has no
17858         children when trying to avoid producing a VIEW_CONVERT_EXPR.
17860 2013-01-04  Marek Polacek  <polacek@redhat.com>
17862         PR middle-end/55859
17863         * opts.c (default_options_optimization): Clarify error message.
17865 2013-01-04  Richard Biener  <rguenther@suse.de>
17867         PR middle-end/55863
17868         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17869         reassociation.
17871 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17873         PR target/53789
17874         * config/pa/pa.md (movsi): Revert previous change.
17875         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17876         references.
17878 2013-01-03  Richard Henderson  <rth@redhat.com>
17880         * config/i386/i386.c (ix86_expand_move): Always assign to op1
17881         after eliminating TLS symbols.
17883 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
17885         PR bootstrap/50167
17886         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17887         * graphite-poly.c (debug_gmp_value): Likewise.
17889 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
17891         PR target/55712
17892         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17893         selected code model, define __code_mode_small__, __code_model_medium__,
17894         __code_model_large__, __code_model_32__ or __code_model_kernel__.
17895         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17896         xchg temporary register with %k.  Declare temporary register as
17897         early clobbered.
17898         [__x86_64__]: For medium and large code models, preserve %rbx register.
17900 2013-01-03  Richard Biener  <rguenther@suse.de>
17902         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
17903         (dump_subscript): Adjust.
17904         (finalize_ddr_dependent): Do not dump redundant info.
17905         (analyze_siv_subscript): Adjust.
17906         (subscript_dependence_tester): Likewise.
17907         (compute_affine_dependence): Likewise.
17909 2013-01-03  Richard Biener  <rguenther@suse.de>
17911         Revert
17912         2013-01-03  Richard Biener  <rguenther@suse.de>
17914         PR tree-optimization/55857
17915         * tree-vect-stmts.c (vectorizable_load): Do not setup
17916         re-alignment for invariant loads.
17918         2013-01-02  Richard Biener  <rguenther@suse.de>
17920         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17921         invariant load do not generate a vector load from the scalar location.
17923 2013-01-03  Richard Biener  <rguenther@suse.de>
17925         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17926         for not vectorizing.
17927         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17928         not build INDIRECT_REFs, call get_name once only.
17929         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
17930         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17932 2013-01-03  Richard Biener  <rguenther@suse.de>
17934         PR tree-optimization/55857
17935         * tree-vect-stmts.c (vectorizable_load): Do not setup
17936         re-alignment for invariant loads.
17938 2013-01-03  Richard Biener  <rguenther@suse.de>
17940         PR lto/55848
17941         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17942         prefer a built-in decl.
17944 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17946         * gcc.c (process_command): Update copyright notice dates.
17947         * gcov.c (print_version): Likewise.
17948         * gcov-dump.c (print_version): Likewise.
17950         PR rtl-optimization/55838
17951         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17952         iv0.step, iv1.step and step.
17954 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17955             Marc Glisse  <marc.glisse@inria.fr>
17957         PR tree-optimization/55832
17958         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17959         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17960         integer_{one,zero}_node.
17962 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17964         PR debug/54402
17965         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17966         * var-tracking.c (reverse_op): Don't add reverse ops to
17967         VALUEs that have already
17968         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
17970 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
17972         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17974 2013-01-02  Teresa Johnson  <tejohnson@google.com>
17976         * dumpfile.c (dump_loc): Print filename with location.
17977         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17978         new location_t parameter to emit complete unroll message with
17979         new dump framework.
17980         (canonicalize_loop_induction_variables): Compute loops location
17981         and pass to try_unroll_loop_completely.
17982         * loop-unroll.c (report_unroll_peel): New function.
17983         (peel_loops_completely): Use new dump format with location
17984         for main dumpfile message, and invoke report_unroll_peel on success.
17985         (decide_unrolling_and_peeling): Ditto.
17986         (decide_peel_once_rolling): Remove old dumpfile message subsumed
17987         by report_unroll_peel.
17988         (decide_peel_completely): Ditto.
17989         (decide_unroll_constant_iterations): Ditto.
17990         (decide_unroll_runtime_iterations): Ditto.
17991         (decide_peel_simple): Ditto.
17992         (decide_unroll_stupid): Ditto.
17993         * cfgloop.c (get_loop_location): New function.
17994         * cfgloop.h (get_loop_location): Declare.
17996 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17998         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
17999         NULL.
18001 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18003         PR middle-end/55198
18004         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18005         BLKmode objects when EXPAND_MEMORY is specified.
18007 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18009         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18010         in loop predicate.
18011         (fold_builtin_cpu): Do not share cpu model decls across statements.
18013 2013-01-02  Jason Merrill  <jason@redhat.com>
18015         PR c++/55804
18016         * tree.c (build_array_type_1): Revert earlier change.
18018 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18020         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18021         "cortex-a57".
18022         * config/aarch64/aarch64-tune.md: Re-generate.
18024 2013-01-02  Richard Biener  <rguenther@suse.de>
18026         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18027         invariant load do not generate a vector load from the scalar location.
18029 2013-01-02  Richard Biener  <rguenther@suse.de>
18031         PR bootstrap/55784
18032         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18033         * configure: Regenerate.
18035 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18037         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18038         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18039         (expand_builtin_int_roundingfn_2): Keep the original target around
18040         for the fallback case.
18042 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18044         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18045         to be clear for sign changes.
18047 2013-01-01  Jan Hubicka  <jh@suse.cz>
18049         * ipa-inline-analysis.c: Fix formatting.
18051 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18053         PR tree-optimization/55831
18054         * tree-vect-loop.c (get_initial_def_for_induction): Use
18055         gsi_after_labels instead of gsi_start_bb.
18057 Copyright (C) 2013 Free Software Foundation, Inc.
18059 Copying and distribution of this file, with or without modification,
18060 are permitted in any medium without royalty provided the copyright
18061 notice and this notice are preserved.