2013-08-12 Uros Bizjak <ubizjak@gmail.com>
[official-gcc.git] / gcc / ChangeLog
blob85d0527fef9a6f4db7335df37b9c41404cd0eb0c
1 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
3         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
4         iterator instead of X87MODEF.
6 2013-08-12  Perez Read  <netfirewall@gmail.com>
8         PR target/58132
9         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
10         operand 0 for intel asm alternative.
11         (*movabs<mode>_2): Ditto for operand 1.
13 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
15         * config/aarch64/arm_none.h
16         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
18 2013-08-12  Nick Clifton  <nickc@redhat.com>
20         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
22 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
24         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
25         expand for QI/HImode operand to produce more effictive code for
26         unsigned char(short) --> float(double) conversion.
28 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
30         * doc/invoke.texi: Mention that -ftls-model does not force the final
31         model.
33 2013-08-12  Marek Polacek  <polacek@redhat.com>
34             Marc Glisse  <marc.glisse@inria.fr>
36         PR tree-optimization/57980
37         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
38         when creating -1 constant.
40 2013-08-10  Jan Hubicka  <jh@suse.cz>
42         Workaround binutils PR14342.
43         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
44         (gimple_init_edge_profiler): Likewise.
45         (gimple_gen_ic_func_profiler): Likewise.
47 2013-08-09  Jan Hubicka  <jh@suse.cz>
49         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
51 2013-08-09  Xinliang David Li  <davidxl@google.com>
53         * config/i386/stringop.def: New file.
54         * config/i386/stringop.opt: New file.
55         * config/i386/i386-opts.h: Include stringopt.def.
56         * config/i386/i386.opt: Include stringopt.opt.
57         * config/i386/i386.c (ix86_option_override_internal):
58         Override default size based stringop inline strategies with options.
59         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
60         New function.
62 2013-08-09  Jan Hubicka  <jh@suse.cz>
64         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
66 2013-08-09  Jan Hubicka  <jh@suse.cz>
68         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
69         CGRAPH_FREQ_MAX.
70         (dump_cgraph_node): Dump profile-id.
71         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
72         and common_target_probability.
73         * lto-cgraph.c (lto_output_edge): Stream common targets.
74         (lto_output_node): Stream profile ids.
75         (input_node): Stream profile ids.
76         (input_edge): Stream common targets.
77         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
78         * ipa.c: Include value-prof.h
79         (ipa_profile_generate_summary): Turn indirect call statement histograms
80         into common targets.
81         (ipa_profile): Turn common targets into speculative edges.
83 2013-08-09  Jan Hubicka  <jh@suse.cz>
85         * cgraph.h (cgraph_node): Add profile_id.
86         * value-prof.c (cgraph_node_map): Turn into pointer_map.
87         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
88         (del_node_map): Update.
89         (find_func_by_funcdef_no): Replace by ...
90         (find_func_by_profile_id): ... this one.
91         (gimple_ic_transform): Do not remove useful histograms when
92         speculation is not done; dump info when indirect call removal
93         can happen at LTO.
94         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
95         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
96         (__gcov_indirect_call_profiler_v2): .. this one.
97         * profile.h (init_node_map): Update.
98         * coverage.c (coverage_compute_profile_id): New function.
99         * coverage.h (coverage_compute_profile_id): Declare.
100         * tree-profile.c (init_ic_make_global_vars): Make
101         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
102         (gimple_init_edge_profiler): Update prototype of
103         __gcov_indirect_call_profiler.
104         (gimple_gen_ic_func_profiler): Simplify.
105         (tree_profiling): Use init_node_map
107 2013-08-09  Jan Hubicka  <jh@suse.cz>
109         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
110         non-speculative refs.
111         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
112         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
113         (cgraph_set_call_stmt): Likewise.
114         (cgraph_create_edge_1): Fix release checking compilatoin;
115         clear lto_stmt_uid.
116         (cgraph_free_edge): Free indirect info.
117         (cgraph_turn_edge_to_speculative): New function.
118         (cgraph_speculative_call_info): New function.
119         (cgraph_make_edge_direct): Return direct edge; handle speculation.
120         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
121         (dump_cgraph_node): Dump speculation.
122         (verify_edge_count_and_frequency): Accept speculative edges.
123         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
124         (verify_cgraph_node): Handle speculation.
125         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
126         (cgraph_set_call_stmt): Update prototype.
127         (cgraph_make_edge_direct): Update prototype.
128         (cgraph_speculative_call_info): Declare.
129         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
130         to change; update call of ipa_find_references.
131         * ipa-ref.c (ipa_record_reference): Fix return value; clear
132         lto_stmt_uid and speculative flags.
133         (ipa_dump_references): Dump speculation.
134         (ipa_clone_references): Clone speculative flag.
135         (ipa_clone_referring): Likewise.
136         (ipa_clone_ref): New function.
137         (ipa_find_reference): Look into lto_stmt_uids
138         (ipa_clear_stmts_in_references): Do not clear speculative calls.
139         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
140         (ipa_find_reference): Update declaration.
141         (ipa_clone_ref): Declare.
142         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
143         stream speculative flag.
144         (lto_output_ref): Stream statements uids and speculation.
145         (input_ref): Likewise.
146         (input_edge): Stream speuclation.
147         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
148         (cgraph_set_call_stmt_including_clones): Handle speculation.
149         * ipa-inline.c (heap_edge_removal_hook): New function.
150         (inline_small_functions): Register it.
151         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
152         also initialize refs.
153         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
154         edge to change.
155         (try_make_edge_direct_simple_call): Likewise.
156         (try_make_edge_direct_simple_call): Likewise.
157         (update_indirect_edges_after_inlining): Likewise.
158         (remove_described_reference): Look proper lto_stmt_uid.
159         (propagate_controlled_uses): Likewise.
160         (propagate_controlled_uses): Liekwise.
161         * tree-inline.c (copy_bb): Copy speculative edges.
162         (redirect_all_calls): New function.
163         (copy_cfg_body): Do redirection after loop info is updated.
164         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
166 2013-08-09  Jan Hubicka  <jh@suse.cz>
168         * lto-streamer-out.c (output_function): Renumber PHIs.
169         * lto-streamer-in.c (input_function): Likewise.
171 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
173         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
174         (get_lane_unsigned): Likewise.
175         (dup_lane_scalar): Likewise.
176         (get_lane): enable for VALL.
177         * config/aarch64/aarch64-simd.md
178         (aarch64_dup_lane_scalar<mode>): Remove.
179         (aarch64_get_lane_signed<mode>): Likewise.
180         (aarch64_get_lane_unsigned<mode>): Likewise.
181         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
182         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
183         (aarch64_get_lane<mode>): Enable for all vector modes.
184         (aarch64_get_lanedi): Remove misleading constraints.
185         * config/aarch64/arm_neon.h
186         (__aarch64_vget_lane_any): Define.
187         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
188         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
189         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
190         * config/aarch64/iterators.md (VDQQH): New.
191         (VDQQHS): Likewise.
192         (vwcore): Likewise.
194 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
196         * configure.ac: Add GAS check for LEON instructions on SPARC.
197         * configure: Regenerate.
198         * config.in: Likewise.
199         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
200         sparc*-*-* block.
201         * config/sparc/sparc.opt (LEON, LEON3): New masks.
202         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
203         for LEON or LEON3.
204         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
205         (AS_LEON_FLAG): New macro.
206         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
207         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
208         Deal with LEON and LEON3 for the memory model.
209         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
210         (atomic_compare_and_swap<mode>_1): Likewise.
211         (*atomic_compare_and_swap<mode>_1): Likewise.
213 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
215         * config/arm/neon.md (vcond): Fix floating-point vector
216         comparisons against 0.
218 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
220         * lra-constraints.c (emit_spill_move): Remove assert.
221         (process_alt_operands): Add more debugging
222         output.  Increase reject for spilling into memory.  Decrease
223         reject for reloading scratch.
224         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
226 2013-08-08  Steve Ellcey  <sellcey@mips.com>
228         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
229         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
230         micromips incompatible.  Add nan2008.
231         (MULTILIB_DIRNAMES): Add nan2008.
232         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
233         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
234         and micromips incompatible.  Add nan2008.
235         (MULTILIB_DIRNAMES): Add nan2008.
236         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
238 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
240         PR rtl-optimization/58079
241         * combine.c (combine_simplify_rtx): Avoid using SUBST if
242         simplify_comparison has widened a comparison with an integer.
244 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
246         * config/arm/neon.md (movmisalign<mode>): Disable when we
247         don't allow unaligned accesses.
248         (*movmisalign<mode>_neon_store): Likewise.
249         (*movmisalign<mode>_neon_load): Likewise.
250         (*movmisalign<mode>_neon_store): Likewise.
251         (*movmisalign<mode>_neon_load): Likewise.
253 2013-08-08  Jan Hubicka  <jh@suse.cz>
255         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
256         (make_pass_rebuild_cgraph_edges): Also clear references.
257         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
258         * ipa-inline-transform.c (inline_transform): Remove all references
259         after inlining.
260         * cgraphunit.c (expand_function): Remove all references after
261         expansion.
262         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
263         (ipa_find_reference): Rewrite to iterator.
264         (remove_stmt_references): Likewise.
265         (ipa_clear_stmts_in_references): New function.
266         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
267         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
268         clear references.
269         * ipa-split.c (split_function): Remove references in split function.
271 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
273         PR target/57431
274         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
275         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
277 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
279         PR target/56979
280         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
281         suggested mode for the assignment isn't compatible with the
282         registers required.
284 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
286         PR target/58065
287         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
289 2013-08-07  Xinliang David Li  <davidxl@google.com>
291         * config/i386/i386.opt: New option -mtune-ctrl=.
292         * config/i386/x86-tune.def: New file.
293         * config/i386/i386.h: include x86-tune.def.
294         * config/i386/i386.c (ix86_option_override_internal):
295         Parsing -mtune-ctrl= option and set tune features.
297 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
299         PR other/12081
300         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
301         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
302         to gen_2arg_fn_t.
304 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
306         * rtl.h (update_alignments): Declare.
307         * final.c (grow_label_align): New function extracted from...
308         (shorten_branches): ...here.  Call it.
309         (update_alignments): New function.
310         * reorg.c (sibling_labels): New variable.
311         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
312         the new label along with it onto the sibling_labels vector.
313         (fill_simple_delay_slots): Adjust call to get_label_before.
314         (fill_slots_from_thread): Likewise.
315         (relax_delay_slots): Likewise.
316         (make_return_insns): Likewise.
317         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
319 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
321         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
322         document its semantics.
323         (diagnostic_report_diagnostic): Adjust accordingly.
325 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
327         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
328         (sparc_option_override): ...and port to new C++ pass API.
329         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
331 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
333         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
335 2013-08-06  Caroline Tice  <cmtice@google.com>
337         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
338         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
339         * tree-pass.h: Add pass_vtable_verify.
340         * varasm.c (assemble_variable): Add code to properly set the comdat
341         section and name for the .vtable_map_vars section.
342         (assemble_vtyv_preinit_initializer): New function.
343         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
344         LINK_ONCE flag.
345         * output.h: Add function decl for assemble_vtv_preinit_initializer.
346         * vtable-verify.c: New file.
347         * vtable-verify.h: New file.
348         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
349         initialiation levels.
350         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
351         * passes.def: Insert pass_vtable_verify.
352         * aclocal.m4: Reorder includes.
353         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
354         -fvtv-counts options.
355         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
356         as appropriate, if -fvtable-verify=... is used.
357         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
358         -fvtable-verify=... is used.
359         * Makefile.in (OBJS):  Add vtable-verify.o to list.
360         (vtable-verify.o): Add new build rule.
361         (GTFILES): Add vtable-verify.c to list.
362         * common.opt (fvtable-verify=): New flag.
363         (vtv_priority): Values for fvtable-verify= flag.
364         (fvtv-counts): New flag.
365         (fvtv-debug): New flag.
366         * tree.h (save_vtable_map_decl): New extern function decl.
368 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
370         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
371         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
372         (pass_data_rl78_devirt): ...new pass_data instance and...
373         (make_pass_rl78_devirt): ...new function.
374         (rl78_asm_file_start): Port pass registration to new C++ API.
376 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
378         * coretypes.h (rtl_opt_pass): Add.
379         (gcc::context): Add.
380         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
381         (epiphany_init): Port to new C++ pass API.
382         (epiphany_optimize_mode_switching): Likewise.
383         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
384         (pass_manager::get_pass_mode_switching): New.
385         (pass_manager::get_pass_peephole2): New.
386         * mode-switching.c (pass_mode_switching): Add clone method.
387         * recog.c (pass_peephole2): Add clone method.
388         (pass_split_all_insns): Add clone method.
390 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
392         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
393         (mips_option_override): ...here, porting to new C++ API for passes.
395 2013-08-06  Jan Hubicka  <jh@suse.cz>
397         * cgraph.c (cgraph_get_body): New function based on lto.c
398         implementation.
399         * cgraph.h (cgraph_get_body): Declare.
400         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
401         LTO paths.
402         * cgraphunit.c (expand_function): Get body prior expanding.
403         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
404         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
405         really need.
406         * passes.c (do_per_function_toporder): Get body.
407         * tree-inline.c (expand_call_inline): Get body prior inlining it.
408         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
410 2013-08-06  Martin Jambor  <mjambor@suse.cz>
412         PR fortran/57987
413         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
414         is not re-finalized.  Rename second parameter to no_collect.
416 2013-08-06  Martin Jambor  <mjambor@suse.cz>
418         PR middle-end/58041
419         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
420         MEM_REF has proper alignment information.
422 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
424         PR other/12081
425         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
426         class insn_gen_fn.
427         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
428         rtx (*) (rtx, ...) with insn_gen_fn.
429         * genoutput.c (output_insn_data): Cast gen_? function pointers to
430         insn_gen_fn::stored_funcptr.  Add initializer braces.
432 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
434         Rewrite how instances of passes are cloned to remove assumptions
435         about their sizes (thus allowing pass subclasses to have
436         additional data fields, albeit non-GC-managed ones at this point).
438         * passes.c (make_pass_instance): Now that passes have clone
439         methods, rewrite this function to eliminate XNEW and memcpy
440         calls that used hardcoded sizes.  Since this function no longer
441         creates pass instances, rename it to...
442         (add_pass_instance): ...this.  Document the old way that passes were
443         numbered and flagged, and rework this function to continue using it.
444         (next_pass_1): Add an initial_pass argument for use by
445         add_pass_instance.
446         (position_pass): When adding multiple instances of a pass, use
447         the pass's clone method, rather than relying on the XNEW/memcpy
448         within the former make_pass_instance (now add_pass_instance).
449         (pass_manager::pass_manager): When invoking next_pass_1, also supply
450         the initial instance of the current pass within the pass manager.
452 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
454         This is the automated part of the conversion of passes from C
455         structs to C++ classes.
457         Patch autogenerated by refactor_passes.py from
458         https://github.com/davidmalcolm/gcc-refactoring-scripts
459         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
461         * asan.c (pass_asan): Convert from a global struct to a subclass of
462         gimple_opt_pass along with...
463         (pass_data_asan): ...new pass_data instance and...
464         (make_pass_asan): ...new function.
465         (pass_asan_O0): Convert from a global struct to a subclass of
466         gimple_opt_pass along with...
467         (pass_data_asan_O0): ...new pass_data instance and...
468         (make_pass_asan_O0): ...new function.
469         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
470         subclass of rtl_opt_pass along with...
471         (pass_data_inc_dec): ...new pass_data instance and...
472         (make_pass_inc_dec): ...new function.
473         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
474         a subclass of rtl_opt_pass along with...
475         (pass_data_reorder_blocks): ...new pass_data instance and...
476         (make_pass_reorder_blocks): ...new function.
477         (pass_duplicate_computed_gotos): Convert from a global struct to a
478         subclass of rtl_opt_pass along with...
479         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
480         (make_pass_duplicate_computed_gotos): ...new function.
481         (pass_partition_blocks): Convert from a global struct to a subclass of
482         rtl_opt_pass along with...
483         (pass_data_partition_blocks): ...new pass_data instance and...
484         (make_pass_partition_blocks): ...new function.
485         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
486         struct to a subclass of rtl_opt_pass along with...
487         (pass_data_branch_target_load_optimize1): ...new pass_data instance
488         and...
489         (make_pass_branch_target_load_optimize1): ...new function.
490         (pass_branch_target_load_optimize2): Convert from a global struct to a
491         subclass of rtl_opt_pass along with...
492         (pass_data_branch_target_load_optimize2): ...new pass_data instance
493         and...
494         (make_pass_branch_target_load_optimize2): ...new function.
495         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
496         of rtl_opt_pass along with...
497         (pass_data_jump): ...new pass_data instance and...
498         (make_pass_jump): ...new function.
499         (pass_jump2): Convert from a global struct to a subclass of
500         rtl_opt_pass along with...
501         (pass_data_jump2): ...new pass_data instance and...
502         (make_pass_jump2): ...new function.
503         * cfgexpand.c (pass_expand): Convert from a global struct to a
504         subclass of rtl_opt_pass along with...
505         (pass_data_expand): ...new pass_data instance and...
506         (make_pass_expand): ...new function.
507         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
508         of rtl_opt_pass along with...
509         (pass_data_free_cfg): ...new pass_data instance and...
510         (make_pass_free_cfg): ...new function.
511         (pass_into_cfg_layout_mode): Convert from a global struct to a
512         subclass of rtl_opt_pass along with...
513         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
514         (make_pass_into_cfg_layout_mode): ...new function.
515         (pass_outof_cfg_layout_mode): Convert from a global struct to a
516         subclass of rtl_opt_pass along with...
517         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
518         (make_pass_outof_cfg_layout_mode): ...new function.
519         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
520         struct to a subclass of gimple_opt_pass along with...
521         (pass_data_build_cgraph_edges): ...new pass_data instance and...
522         (make_pass_build_cgraph_edges): ...new function.
523         (pass_rebuild_cgraph_edges): Convert from a global struct to a
524         subclass of gimple_opt_pass along with...
525         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
526         (make_pass_rebuild_cgraph_edges): ...new function.
527         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
528         subclass of gimple_opt_pass along with...
529         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
530         and...
531         (make_pass_remove_cgraph_callee_edges): ...new function.
532         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
533         struct to a subclass of rtl_opt_pass along with...
534         (pass_data_stack_adjustments): ...new pass_data instance and...
535         (make_pass_stack_adjustments): ...new function.
536         * combine.c (pass_combine): Convert from a global struct to a subclass
537         of rtl_opt_pass along with...
538         (pass_data_combine): ...new pass_data instance and...
539         (make_pass_combine): ...new function.
540         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
541         global struct to a subclass of rtl_opt_pass along with...
542         (pass_data_compare_elim_after_reload): ...new pass_data instance
543         and...
544         (make_pass_compare_elim_after_reload): ...new function.
545         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
546         of rtl_opt_pass along with...
547         (pass_data_rtl_cprop): ...new pass_data instance and...
548         (make_pass_rtl_cprop): ...new function.
549         * cse.c (pass_cse): Convert from a global struct to a subclass of
550         rtl_opt_pass along with...
551         (pass_data_cse): ...new pass_data instance and...
552         (make_pass_cse): ...new function.
553         (pass_cse2): Convert from a global struct to a subclass of
554         rtl_opt_pass along with...
555         (pass_data_cse2): ...new pass_data instance and...
556         (make_pass_cse2): ...new function.
557         (pass_cse_after_global_opts): Convert from a global struct to a
558         subclass of rtl_opt_pass along with...
559         (pass_data_cse_after_global_opts): ...new pass_data instance and...
560         (make_pass_cse_after_global_opts): ...new function.
561         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
562         of rtl_opt_pass along with...
563         (pass_data_ud_rtl_dce): ...new pass_data instance and...
564         (make_pass_ud_rtl_dce): ...new function.
565         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
566         rtl_opt_pass along with...
567         (pass_data_fast_rtl_dce): ...new pass_data instance and...
568         (make_pass_fast_rtl_dce): ...new function.
569         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
570         a subclass of rtl_opt_pass along with...
571         (pass_data_df_initialize_opt): ...new pass_data instance and...
572         (make_pass_df_initialize_opt): ...new function.
573         (pass_df_initialize_no_opt): Convert from a global struct to a
574         subclass of rtl_opt_pass along with...
575         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
576         (make_pass_df_initialize_no_opt): ...new function.
577         (pass_df_finish): Convert from a global struct to a subclass of
578         rtl_opt_pass along with...
579         (pass_data_df_finish): ...new pass_data instance and...
580         (make_pass_df_finish): ...new function.
581         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
582         rtl_opt_pass along with...
583         (pass_data_rtl_dse1): ...new pass_data instance and...
584         (make_pass_rtl_dse1): ...new function.
585         (pass_rtl_dse2): Convert from a global struct to a subclass of
586         rtl_opt_pass along with...
587         (pass_data_rtl_dse2): ...new pass_data instance and...
588         (make_pass_rtl_dse2): ...new function.
589         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
590         subclass of rtl_opt_pass along with...
591         (pass_data_dwarf2_frame): ...new pass_data instance and...
592         (make_pass_dwarf2_frame): ...new function.
593         * except.c (pass_set_nothrow_function_flags): Convert from a global
594         struct to a subclass of rtl_opt_pass along with...
595         (pass_data_set_nothrow_function_flags): ...new pass_data instance
596         and...
597         (make_pass_set_nothrow_function_flags): ...new function.
598         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
599         subclass of rtl_opt_pass along with...
600         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
601         and...
602         (make_pass_convert_to_eh_region_ranges): ...new function.
603         * final.c (pass_compute_alignments): Convert from a global struct to a
604         subclass of rtl_opt_pass along with...
605         (pass_data_compute_alignments): ...new pass_data instance and...
606         (make_pass_compute_alignments): ...new function.
607         (pass_final): Convert from a global struct to a subclass of
608         rtl_opt_pass along with...
609         (pass_data_final): ...new pass_data instance and...
610         (make_pass_final): ...new function.
611         (pass_shorten_branches): Convert from a global struct to a subclass of
612         rtl_opt_pass along with...
613         (pass_data_shorten_branches): ...new pass_data instance and...
614         (make_pass_shorten_branches): ...new function.
615         (pass_clean_state): Convert from a global struct to a subclass of
616         rtl_opt_pass along with...
617         (pass_data_clean_state): ...new pass_data instance and...
618         (make_pass_clean_state): ...new function.
619         * function.c (pass_instantiate_virtual_regs): Convert from a global
620         struct to a subclass of rtl_opt_pass along with...
621         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
622         (make_pass_instantiate_virtual_regs): ...new function.
623         (pass_leaf_regs): Convert from a global struct to a subclass of
624         rtl_opt_pass along with...
625         (pass_data_leaf_regs): ...new pass_data instance and...
626         (make_pass_leaf_regs): ...new function.
627         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
628         subclass of rtl_opt_pass along with...
629         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
630         and...
631         (make_pass_thread_prologue_and_epilogue): ...new function.
632         (pass_match_asm_constraints): Convert from a global struct to a
633         subclass of rtl_opt_pass along with...
634         (pass_data_match_asm_constraints): ...new pass_data instance and...
635         (make_pass_match_asm_constraints): ...new function.
636         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
637         subclass of rtl_opt_pass along with...
638         (pass_data_rtl_fwprop): ...new pass_data instance and...
639         (make_pass_rtl_fwprop): ...new function.
640         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
641         rtl_opt_pass along with...
642         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
643         (make_pass_rtl_fwprop_addr): ...new function.
644         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
645         rtl_opt_pass along with...
646         (pass_data_rtl_pre): ...new pass_data instance and...
647         (make_pass_rtl_pre): ...new function.
648         (pass_rtl_hoist): Convert from a global struct to a subclass of
649         rtl_opt_pass along with...
650         (pass_data_rtl_hoist): ...new pass_data instance and...
651         (make_pass_rtl_hoist): ...new function.
652         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
653         subclass of gimple_opt_pass along with...
654         (pass_data_lower_cf): ...new pass_data instance and...
655         (make_pass_lower_cf): ...new function.
656         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
657         from a global struct to a subclass of gimple_opt_pass along with...
658         (pass_data_strength_reduction): ...new pass_data instance and...
659         (make_pass_strength_reduction): ...new function.
660         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
661         of rtl_opt_pass along with...
662         (pass_data_rtl_ifcvt): ...new pass_data instance and...
663         (make_pass_rtl_ifcvt): ...new function.
664         (pass_if_after_combine): Convert from a global struct to a subclass of
665         rtl_opt_pass along with...
666         (pass_data_if_after_combine): ...new pass_data instance and...
667         (make_pass_if_after_combine): ...new function.
668         (pass_if_after_reload): Convert from a global struct to a subclass of
669         rtl_opt_pass along with...
670         (pass_data_if_after_reload): ...new pass_data instance and...
671         (make_pass_if_after_reload): ...new function.
672         * init-regs.c (pass_initialize_regs): Convert from a global struct to
673         a subclass of rtl_opt_pass along with...
674         (pass_data_initialize_regs): ...new pass_data instance and...
675         (make_pass_initialize_regs): ...new function.
676         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
677         of ipa_opt_pass_d along with...
678         (pass_data_ipa_cp): ...new pass_data instance and...
679         (make_pass_ipa_cp): ...new function.
680         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
681         global struct to a subclass of gimple_opt_pass along with...
682         (pass_data_inline_parameters): ...new pass_data instance and...
683         (make_pass_inline_parameters): ...new function.
684         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
685         subclass of gimple_opt_pass along with...
686         (pass_data_early_inline): ...new pass_data instance and...
687         (make_pass_early_inline): ...new function.
688         (pass_ipa_inline): Convert from a global struct to a subclass of
689         ipa_opt_pass_d along with...
690         (pass_data_ipa_inline): ...new pass_data instance and...
691         (make_pass_ipa_inline): ...new function.
692         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
693         struct to a subclass of gimple_opt_pass along with...
694         (pass_data_local_pure_const): ...new pass_data instance and...
695         (make_pass_local_pure_const): ...new function.
696         (pass_ipa_pure_const): Convert from a global struct to a subclass of
697         ipa_opt_pass_d along with...
698         (pass_data_ipa_pure_const): ...new pass_data instance and...
699         (make_pass_ipa_pure_const): ...new function.
700         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
701         to a subclass of ipa_opt_pass_d along with...
702         (pass_data_ipa_reference): ...new pass_data instance and...
703         (make_pass_ipa_reference): ...new function.
704         * ipa-split.c (pass_split_functions): Convert from a global struct to
705         a subclass of gimple_opt_pass along with...
706         (pass_data_split_functions): ...new pass_data instance and...
707         (make_pass_split_functions): ...new function.
708         (pass_feedback_split_functions): Convert from a global struct to a
709         subclass of gimple_opt_pass along with...
710         (pass_data_feedback_split_functions): ...new pass_data instance and...
711         (make_pass_feedback_split_functions): ...new function.
712         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
713         global struct to a subclass of simple_ipa_opt_pass along with...
714         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
715         instance and...
716         (make_pass_ipa_function_and_variable_visibility): ...new function.
717         (pass_ipa_free_inline_summary): Convert from a global struct to a
718         subclass of simple_ipa_opt_pass along with...
719         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
720         (make_pass_ipa_free_inline_summary): ...new function.
721         (pass_ipa_whole_program_visibility): Convert from a global struct to a
722         subclass of ipa_opt_pass_d along with...
723         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
724         and...
725         (make_pass_ipa_whole_program_visibility): ...new function.
726         (pass_ipa_profile): Convert from a global struct to a subclass of
727         ipa_opt_pass_d along with...
728         (pass_data_ipa_profile): ...new pass_data instance and...
729         (make_pass_ipa_profile): ...new function.
730         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
731         ipa_opt_pass_d along with...
732         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
733         (make_pass_ipa_cdtor_merge): ...new function.
734         * ira.c (pass_ira): Convert from a global struct to a subclass of
735         rtl_opt_pass along with...
736         (pass_data_ira): ...new pass_data instance and...
737         (make_pass_ira): ...new function.
738         (pass_reload): Convert from a global struct to a subclass of
739         rtl_opt_pass along with...
740         (pass_data_reload): ...new pass_data instance and...
741         (make_pass_reload): ...new function.
742         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
743         subclass of rtl_opt_pass along with...
744         (pass_data_cleanup_barriers): ...new pass_data instance and...
745         (make_pass_cleanup_barriers): ...new function.
746         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
747         of rtl_opt_pass along with...
748         (pass_data_loop2): ...new pass_data instance and...
749         (make_pass_loop2): ...new function.
750         (pass_rtl_loop_init): Convert from a global struct to a subclass of
751         rtl_opt_pass along with...
752         (pass_data_rtl_loop_init): ...new pass_data instance and...
753         (make_pass_rtl_loop_init): ...new function.
754         (pass_rtl_loop_done): Convert from a global struct to a subclass of
755         rtl_opt_pass along with...
756         (pass_data_rtl_loop_done): ...new pass_data instance and...
757         (make_pass_rtl_loop_done): ...new function.
758         (pass_rtl_move_loop_invariants): Convert from a global struct to a
759         subclass of rtl_opt_pass along with...
760         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
761         (make_pass_rtl_move_loop_invariants): ...new function.
762         (pass_rtl_unswitch): Convert from a global struct to a subclass of
763         rtl_opt_pass along with...
764         (pass_data_rtl_unswitch): ...new pass_data instance and...
765         (make_pass_rtl_unswitch): ...new function.
766         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
767         subclass of rtl_opt_pass along with...
768         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
769         and...
770         (make_pass_rtl_unroll_and_peel_loops): ...new function.
771         (pass_rtl_doloop): Convert from a global struct to a subclass of
772         rtl_opt_pass along with...
773         (pass_data_rtl_doloop): ...new pass_data instance and...
774         (make_pass_rtl_doloop): ...new function.
775         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
776         a subclass of rtl_opt_pass along with...
777         (pass_data_lower_subreg): ...new pass_data instance and...
778         (make_pass_lower_subreg): ...new function.
779         (pass_lower_subreg2): Convert from a global struct to a subclass of
780         rtl_opt_pass along with...
781         (pass_data_lower_subreg2): ...new pass_data instance and...
782         (make_pass_lower_subreg2): ...new function.
783         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
784         struct to a subclass of ipa_opt_pass_d along with...
785         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
786         (make_pass_ipa_lto_gimple_out): ...new function.
787         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
788         of ipa_opt_pass_d along with...
789         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
790         (make_pass_ipa_lto_finish_out): ...new function.
791         * mode-switching.c (pass_mode_switching): Convert from a global struct
792         to a subclass of rtl_opt_pass along with...
793         (pass_data_mode_switching): ...new pass_data instance and...
794         (make_pass_mode_switching): ...new function.
795         * modulo-sched.c (pass_sms): Convert from a global struct to a
796         subclass of rtl_opt_pass along with...
797         (pass_data_sms): ...new pass_data instance and...
798         (make_pass_sms): ...new function.
799         * omp-low.c (pass_expand_omp): Convert from a global struct to a
800         subclass of gimple_opt_pass along with...
801         (pass_data_expand_omp): ...new pass_data instance and...
802         (make_pass_expand_omp): ...new function.
803         (pass_lower_omp): Convert from a global struct to a subclass of
804         gimple_opt_pass along with...
805         (pass_data_lower_omp): ...new pass_data instance and...
806         (make_pass_lower_omp): ...new function.
807         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
808         of gimple_opt_pass along with...
809         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
810         (make_pass_diagnose_omp_blocks): ...new function.
811         * passes.c (pass_early_local_passes): Convert from a global struct to
812         a subclass of simple_ipa_opt_pass along with...
813         (pass_data_early_local_passes): ...new pass_data instance and...
814         (make_pass_early_local_passes): ...new function.
815         (pass_all_early_optimizations): Convert from a global struct to a
816         subclass of gimple_opt_pass along with...
817         (pass_data_all_early_optimizations): ...new pass_data instance and...
818         (make_pass_all_early_optimizations): ...new function.
819         (pass_all_optimizations): Convert from a global struct to a subclass
820         of gimple_opt_pass along with...
821         (pass_data_all_optimizations): ...new pass_data instance and...
822         (make_pass_all_optimizations): ...new function.
823         (pass_all_optimizations_g): Convert from a global struct to a subclass
824         of gimple_opt_pass along with...
825         (pass_data_all_optimizations_g): ...new pass_data instance and...
826         (make_pass_all_optimizations_g): ...new function.
827         (pass_rest_of_compilation): Convert from a global struct to a subclass
828         of rtl_opt_pass along with...
829         (pass_data_rest_of_compilation): ...new pass_data instance and...
830         (make_pass_rest_of_compilation): ...new function.
831         (pass_postreload): Convert from a global struct to a subclass of
832         rtl_opt_pass along with...
833         (pass_data_postreload): ...new pass_data instance and...
834         (make_pass_postreload): ...new function.
835         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
836         subclass of rtl_opt_pass along with...
837         (pass_data_gcse2): ...new pass_data instance and...
838         (make_pass_gcse2): ...new function.
839         * postreload.c (pass_postreload_cse): Convert from a global struct to
840         a subclass of rtl_opt_pass along with...
841         (pass_data_postreload_cse): ...new pass_data instance and...
842         (make_pass_postreload_cse): ...new function.
843         * predict.c (pass_profile): Convert from a global struct to a subclass
844         of gimple_opt_pass along with...
845         (pass_data_profile): ...new pass_data instance and...
846         (make_pass_profile): ...new function.
847         (pass_strip_predict_hints): Convert from a global struct to a subclass
848         of gimple_opt_pass along with...
849         (pass_data_strip_predict_hints): ...new pass_data instance and...
850         (make_pass_strip_predict_hints): ...new function.
851         * recog.c (pass_peephole2): Convert from a global struct to a subclass
852         of rtl_opt_pass along with...
853         (pass_data_peephole2): ...new pass_data instance and...
854         (make_pass_peephole2): ...new function.
855         (pass_split_all_insns): Convert from a global struct to a subclass of
856         rtl_opt_pass along with...
857         (pass_data_split_all_insns): ...new pass_data instance and...
858         (make_pass_split_all_insns): ...new function.
859         (pass_split_after_reload): Convert from a global struct to a subclass
860         of rtl_opt_pass along with...
861         (pass_data_split_after_reload): ...new pass_data instance and...
862         (make_pass_split_after_reload): ...new function.
863         (pass_split_before_regstack): Convert from a global struct to a
864         subclass of rtl_opt_pass along with...
865         (pass_data_split_before_regstack): ...new pass_data instance and...
866         (make_pass_split_before_regstack): ...new function.
867         (pass_split_before_sched2): Convert from a global struct to a subclass
868         of rtl_opt_pass along with...
869         (pass_data_split_before_sched2): ...new pass_data instance and...
870         (make_pass_split_before_sched2): ...new function.
871         (pass_split_for_shorten_branches): Convert from a global struct to a
872         subclass of rtl_opt_pass along with...
873         (pass_data_split_for_shorten_branches): ...new pass_data instance
874         and...
875         (make_pass_split_for_shorten_branches): ...new function.
876         * ree.c (pass_ree): Convert from a global struct to a subclass of
877         rtl_opt_pass along with...
878         (pass_data_ree): ...new pass_data instance and...
879         (make_pass_ree): ...new function.
880         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
881         subclass of rtl_opt_pass along with...
882         (pass_data_stack_regs): ...new pass_data instance and...
883         (make_pass_stack_regs): ...new function.
884         (pass_stack_regs_run): Convert from a global struct to a subclass of
885         rtl_opt_pass along with...
886         (pass_data_stack_regs_run): ...new pass_data instance and...
887         (make_pass_stack_regs_run): ...new function.
888         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
889         subclass of rtl_opt_pass along with...
890         (pass_data_cprop_hardreg): ...new pass_data instance and...
891         (make_pass_cprop_hardreg): ...new function.
892         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
893         subclass of rtl_opt_pass along with...
894         (pass_data_reginfo_init): ...new pass_data instance and...
895         (make_pass_reginfo_init): ...new function.
896         * regmove.c (pass_regmove): Convert from a global struct to a subclass
897         of rtl_opt_pass along with...
898         (pass_data_regmove): ...new pass_data instance and...
899         (make_pass_regmove): ...new function.
900         * regrename.c (pass_regrename): Convert from a global struct to a
901         subclass of rtl_opt_pass along with...
902         (pass_data_regrename): ...new pass_data instance and...
903         (make_pass_regrename): ...new function.
904         * reorg.c (pass_delay_slots): Convert from a global struct to a
905         subclass of rtl_opt_pass along with...
906         (pass_data_delay_slots): ...new pass_data instance and...
907         (make_pass_delay_slots): ...new function.
908         (pass_machine_reorg): Convert from a global struct to a subclass of
909         rtl_opt_pass along with...
910         (pass_data_machine_reorg): ...new pass_data instance and...
911         (make_pass_machine_reorg): ...new function.
912         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
913         of rtl_opt_pass along with...
914         (pass_data_sched): ...new pass_data instance and...
915         (make_pass_sched): ...new function.
916         (pass_sched2): Convert from a global struct to a subclass of
917         rtl_opt_pass along with...
918         (pass_data_sched2): ...new pass_data instance and...
919         (make_pass_sched2): ...new function.
920         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
921         to a subclass of rtl_opt_pass along with...
922         (pass_data_stack_ptr_mod): ...new pass_data instance and...
923         (make_pass_stack_ptr_mod): ...new function.
924         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
925         to a subclass of rtl_opt_pass along with...
926         (pass_data_rtl_store_motion): ...new pass_data instance and...
927         (make_pass_rtl_store_motion): ...new function.
928         * tracer.c (pass_tracer): Convert from a global struct to a subclass
929         of gimple_opt_pass along with...
930         (pass_data_tracer): ...new pass_data instance and...
931         (make_pass_tracer): ...new function.
932         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
933         to a subclass of gimple_opt_pass along with...
934         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
935         (make_pass_diagnose_tm_blocks): ...new function.
936         (pass_lower_tm): Convert from a global struct to a subclass of
937         gimple_opt_pass along with...
938         (pass_data_lower_tm): ...new pass_data instance and...
939         (make_pass_lower_tm): ...new function.
940         (pass_tm_init): Convert from a global struct to a subclass of
941         gimple_opt_pass along with...
942         (pass_data_tm_init): ...new pass_data instance and...
943         (make_pass_tm_init): ...new function.
944         (pass_tm_mark): Convert from a global struct to a subclass of
945         gimple_opt_pass along with...
946         (pass_data_tm_mark): ...new pass_data instance and...
947         (make_pass_tm_mark): ...new function.
948         (pass_tm_edges): Convert from a global struct to a subclass of
949         gimple_opt_pass along with...
950         (pass_data_tm_edges): ...new pass_data instance and...
951         (make_pass_tm_edges): ...new function.
952         (pass_tm_memopt): Convert from a global struct to a subclass of
953         gimple_opt_pass along with...
954         (pass_data_tm_memopt): ...new pass_data instance and...
955         (make_pass_tm_memopt): ...new function.
956         (pass_ipa_tm): Convert from a global struct to a subclass of
957         simple_ipa_opt_pass along with...
958         (pass_data_ipa_tm): ...new pass_data instance and...
959         (make_pass_ipa_tm): ...new function.
960         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
961         subclass of gimple_opt_pass along with...
962         (pass_data_call_cdce): ...new pass_data instance and...
963         (make_pass_call_cdce): ...new function.
964         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
965         subclass of gimple_opt_pass along with...
966         (pass_data_build_cfg): ...new pass_data instance and...
967         (make_pass_build_cfg): ...new function.
968         (pass_split_crit_edges): Convert from a global struct to a subclass of
969         gimple_opt_pass along with...
970         (pass_data_split_crit_edges): ...new pass_data instance and...
971         (make_pass_split_crit_edges): ...new function.
972         (pass_warn_function_return): Convert from a global struct to a
973         subclass of gimple_opt_pass along with...
974         (pass_data_warn_function_return): ...new pass_data instance and...
975         (make_pass_warn_function_return): ...new function.
976         (pass_warn_function_noreturn): Convert from a global struct to a
977         subclass of gimple_opt_pass along with...
978         (pass_data_warn_function_noreturn): ...new pass_data instance and...
979         (make_pass_warn_function_noreturn): ...new function.
980         (pass_warn_unused_result): Convert from a global struct to a subclass
981         of gimple_opt_pass along with...
982         (pass_data_warn_unused_result): ...new pass_data instance and...
983         (make_pass_warn_unused_result): ...new function.
984         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
985         a subclass of gimple_opt_pass along with...
986         (pass_data_merge_phi): ...new pass_data instance and...
987         (make_pass_merge_phi): ...new function.
988         * tree-complex.c (pass_lower_complex): Convert from a global struct to
989         a subclass of gimple_opt_pass along with...
990         (pass_data_lower_complex): ...new pass_data instance and...
991         (make_pass_lower_complex): ...new function.
992         (pass_lower_complex_O0): Convert from a global struct to a subclass of
993         gimple_opt_pass along with...
994         (pass_data_lower_complex_O0): ...new pass_data instance and...
995         (make_pass_lower_complex_O0): ...new function.
996         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
997         subclass of gimple_opt_pass along with...
998         (pass_data_lower_eh): ...new pass_data instance and...
999         (make_pass_lower_eh): ...new function.
1000         (pass_refactor_eh): Convert from a global struct to a subclass of
1001         gimple_opt_pass along with...
1002         (pass_data_refactor_eh): ...new pass_data instance and...
1003         (make_pass_refactor_eh): ...new function.
1004         (pass_lower_resx): Convert from a global struct to a subclass of
1005         gimple_opt_pass along with...
1006         (pass_data_lower_resx): ...new pass_data instance and...
1007         (make_pass_lower_resx): ...new function.
1008         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1009         of gimple_opt_pass along with...
1010         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1011         (make_pass_lower_eh_dispatch): ...new function.
1012         (pass_cleanup_eh): Convert from a global struct to a subclass of
1013         gimple_opt_pass along with...
1014         (pass_data_cleanup_eh): ...new pass_data instance and...
1015         (make_pass_cleanup_eh): ...new function.
1016         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1017         to a subclass of simple_ipa_opt_pass along with...
1018         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1019         (make_pass_ipa_lower_emutls): ...new function.
1020         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1021         a subclass of gimple_opt_pass along with...
1022         (pass_data_if_conversion): ...new pass_data instance and...
1023         (make_pass_if_conversion): ...new function.
1024         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1025         subclass of gimple_opt_pass along with...
1026         (pass_data_build_ssa): ...new pass_data instance and...
1027         (make_pass_build_ssa): ...new function.
1028         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1029         global struct to a subclass of gimple_opt_pass along with...
1030         (pass_data_loop_distribution): ...new pass_data instance and...
1031         (make_pass_loop_distribution): ...new function.
1032         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1033         subclass of gimple_opt_pass along with...
1034         (pass_data_mudflap_1): ...new pass_data instance and...
1035         (make_pass_mudflap_1): ...new function.
1036         (pass_mudflap_2): Convert from a global struct to a subclass of
1037         gimple_opt_pass along with...
1038         (pass_data_mudflap_2): ...new pass_data instance and...
1039         (make_pass_mudflap_2): ...new function.
1040         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1041         subclass of gimple_opt_pass along with...
1042         (pass_data_mudflap_1): ...new pass_data instance and...
1043         (make_pass_mudflap_1): ...new function.
1044         (pass_mudflap_2): Convert from a global struct to a subclass of
1045         gimple_opt_pass along with...
1046         (pass_data_mudflap_2): ...new pass_data instance and...
1047         (make_pass_mudflap_2): ...new function.
1048         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1049         gimple_opt_pass along with...
1050         (pass_data_nrv): ...new pass_data instance and...
1051         (make_pass_nrv): ...new function.
1052         (pass_return_slot): Convert from a global struct to a subclass of
1053         gimple_opt_pass along with...
1054         (pass_data_return_slot): ...new pass_data instance and...
1055         (make_pass_return_slot): ...new function.
1056         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1057         to a subclass of gimple_opt_pass along with...
1058         (pass_data_object_sizes): ...new pass_data instance and...
1059         (make_pass_object_sizes): ...new function.
1060         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1061         global struct to a subclass of gimple_opt_pass along with...
1062         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1063         and...
1064         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1065         (pass_fixup_cfg): Convert from a global struct to a subclass of
1066         gimple_opt_pass along with...
1067         (pass_data_fixup_cfg): ...new pass_data instance and...
1068         (make_pass_fixup_cfg): ...new function.
1069         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1070         (make_pass_mudflap_1): ...new function.
1071         (pass_mudflap_2): Replace declaration with that of...
1072         (make_pass_mudflap_2): ...new function.
1073         (pass_asan): Replace declaration with that of...
1074         (make_pass_asan): ...new function.
1075         (pass_asan_O0): Replace declaration with that of...
1076         (make_pass_asan_O0): ...new function.
1077         (pass_tsan): Replace declaration with that of...
1078         (make_pass_tsan): ...new function.
1079         (pass_tsan_O0): Replace declaration with that of...
1080         (make_pass_tsan_O0): ...new function.
1081         (pass_lower_cf): Replace declaration with that of...
1082         (make_pass_lower_cf): ...new function.
1083         (pass_refactor_eh): Replace declaration with that of...
1084         (make_pass_refactor_eh): ...new function.
1085         (pass_lower_eh): Replace declaration with that of...
1086         (make_pass_lower_eh): ...new function.
1087         (pass_lower_eh_dispatch): Replace declaration with that of...
1088         (make_pass_lower_eh_dispatch): ...new function.
1089         (pass_lower_resx): Replace declaration with that of...
1090         (make_pass_lower_resx): ...new function.
1091         (pass_build_cfg): Replace declaration with that of...
1092         (make_pass_build_cfg): ...new function.
1093         (pass_early_tree_profile): Replace declaration with that of...
1094         (make_pass_early_tree_profile): ...new function.
1095         (pass_cleanup_eh): Replace declaration with that of...
1096         (make_pass_cleanup_eh): ...new function.
1097         (pass_sra): Replace declaration with that of...
1098         (make_pass_sra): ...new function.
1099         (pass_sra_early): Replace declaration with that of...
1100         (make_pass_sra_early): ...new function.
1101         (pass_early_ipa_sra): Replace declaration with that of...
1102         (make_pass_early_ipa_sra): ...new function.
1103         (pass_tail_recursion): Replace declaration with that of...
1104         (make_pass_tail_recursion): ...new function.
1105         (pass_tail_calls): Replace declaration with that of...
1106         (make_pass_tail_calls): ...new function.
1107         (pass_tree_loop): Replace declaration with that of...
1108         (make_pass_tree_loop): ...new function.
1109         (pass_tree_loop_init): Replace declaration with that of...
1110         (make_pass_tree_loop_init): ...new function.
1111         (pass_lim): Replace declaration with that of...
1112         (make_pass_lim): ...new function.
1113         (pass_tree_unswitch): Replace declaration with that of...
1114         (make_pass_tree_unswitch): ...new function.
1115         (pass_predcom): Replace declaration with that of...
1116         (make_pass_predcom): ...new function.
1117         (pass_iv_canon): Replace declaration with that of...
1118         (make_pass_iv_canon): ...new function.
1119         (pass_scev_cprop): Replace declaration with that of...
1120         (make_pass_scev_cprop): ...new function.
1121         (pass_empty_loop): Replace declaration with that of...
1122         (make_pass_empty_loop): ...new function.
1123         (pass_record_bounds): Replace declaration with that of...
1124         (make_pass_record_bounds): ...new function.
1125         (pass_graphite): Replace declaration with that of...
1126         (make_pass_graphite): ...new function.
1127         (pass_graphite_transforms): Replace declaration with that of...
1128         (make_pass_graphite_transforms): ...new function.
1129         (pass_if_conversion): Replace declaration with that of...
1130         (make_pass_if_conversion): ...new function.
1131         (pass_loop_distribution): Replace declaration with that of...
1132         (make_pass_loop_distribution): ...new function.
1133         (pass_vectorize): Replace declaration with that of...
1134         (make_pass_vectorize): ...new function.
1135         (pass_slp_vectorize): Replace declaration with that of...
1136         (make_pass_slp_vectorize): ...new function.
1137         (pass_complete_unroll): Replace declaration with that of...
1138         (make_pass_complete_unroll): ...new function.
1139         (pass_complete_unrolli): Replace declaration with that of...
1140         (make_pass_complete_unrolli): ...new function.
1141         (pass_parallelize_loops): Replace declaration with that of...
1142         (make_pass_parallelize_loops): ...new function.
1143         (pass_loop_prefetch): Replace declaration with that of...
1144         (make_pass_loop_prefetch): ...new function.
1145         (pass_iv_optimize): Replace declaration with that of...
1146         (make_pass_iv_optimize): ...new function.
1147         (pass_tree_loop_done): Replace declaration with that of...
1148         (make_pass_tree_loop_done): ...new function.
1149         (pass_ch): Replace declaration with that of...
1150         (make_pass_ch): ...new function.
1151         (pass_ccp): Replace declaration with that of...
1152         (make_pass_ccp): ...new function.
1153         (pass_phi_only_cprop): Replace declaration with that of...
1154         (make_pass_phi_only_cprop): ...new function.
1155         (pass_build_ssa): Replace declaration with that of...
1156         (make_pass_build_ssa): ...new function.
1157         (pass_build_alias): Replace declaration with that of...
1158         (make_pass_build_alias): ...new function.
1159         (pass_build_ealias): Replace declaration with that of...
1160         (make_pass_build_ealias): ...new function.
1161         (pass_dominator): Replace declaration with that of...
1162         (make_pass_dominator): ...new function.
1163         (pass_dce): Replace declaration with that of...
1164         (make_pass_dce): ...new function.
1165         (pass_dce_loop): Replace declaration with that of...
1166         (make_pass_dce_loop): ...new function.
1167         (pass_cd_dce): Replace declaration with that of...
1168         (make_pass_cd_dce): ...new function.
1169         (pass_call_cdce): Replace declaration with that of...
1170         (make_pass_call_cdce): ...new function.
1171         (pass_merge_phi): Replace declaration with that of...
1172         (make_pass_merge_phi): ...new function.
1173         (pass_split_crit_edges): Replace declaration with that of...
1174         (make_pass_split_crit_edges): ...new function.
1175         (pass_pre): Replace declaration with that of...
1176         (make_pass_pre): ...new function.
1177         (pass_profile): Replace declaration with that of...
1178         (make_pass_profile): ...new function.
1179         (pass_strip_predict_hints): Replace declaration with that of...
1180         (make_pass_strip_predict_hints): ...new function.
1181         (pass_lower_complex_O0): Replace declaration with that of...
1182         (make_pass_lower_complex_O0): ...new function.
1183         (pass_lower_complex): Replace declaration with that of...
1184         (make_pass_lower_complex): ...new function.
1185         (pass_lower_vector): Replace declaration with that of...
1186         (make_pass_lower_vector): ...new function.
1187         (pass_lower_vector_ssa): Replace declaration with that of...
1188         (make_pass_lower_vector_ssa): ...new function.
1189         (pass_lower_omp): Replace declaration with that of...
1190         (make_pass_lower_omp): ...new function.
1191         (pass_diagnose_omp_blocks): Replace declaration with that of...
1192         (make_pass_diagnose_omp_blocks): ...new function.
1193         (pass_expand_omp): Replace declaration with that of...
1194         (make_pass_expand_omp): ...new function.
1195         (pass_expand_omp_ssa): Replace declaration with that of...
1196         (make_pass_expand_omp_ssa): ...new function.
1197         (pass_object_sizes): Replace declaration with that of...
1198         (make_pass_object_sizes): ...new function.
1199         (pass_strlen): Replace declaration with that of...
1200         (make_pass_strlen): ...new function.
1201         (pass_fold_builtins): Replace declaration with that of...
1202         (make_pass_fold_builtins): ...new function.
1203         (pass_stdarg): Replace declaration with that of...
1204         (make_pass_stdarg): ...new function.
1205         (pass_early_warn_uninitialized): Replace declaration with that of...
1206         (make_pass_early_warn_uninitialized): ...new function.
1207         (pass_late_warn_uninitialized): Replace declaration with that of...
1208         (make_pass_late_warn_uninitialized): ...new function.
1209         (pass_cse_reciprocals): Replace declaration with that of...
1210         (make_pass_cse_reciprocals): ...new function.
1211         (pass_cse_sincos): Replace declaration with that of...
1212         (make_pass_cse_sincos): ...new function.
1213         (pass_optimize_bswap): Replace declaration with that of...
1214         (make_pass_optimize_bswap): ...new function.
1215         (pass_optimize_widening_mul): Replace declaration with that of...
1216         (make_pass_optimize_widening_mul): ...new function.
1217         (pass_warn_function_return): Replace declaration with that of...
1218         (make_pass_warn_function_return): ...new function.
1219         (pass_warn_function_noreturn): Replace declaration with that of...
1220         (make_pass_warn_function_noreturn): ...new function.
1221         (pass_cselim): Replace declaration with that of...
1222         (make_pass_cselim): ...new function.
1223         (pass_phiopt): Replace declaration with that of...
1224         (make_pass_phiopt): ...new function.
1225         (pass_forwprop): Replace declaration with that of...
1226         (make_pass_forwprop): ...new function.
1227         (pass_phiprop): Replace declaration with that of...
1228         (make_pass_phiprop): ...new function.
1229         (pass_tree_ifcombine): Replace declaration with that of...
1230         (make_pass_tree_ifcombine): ...new function.
1231         (pass_dse): Replace declaration with that of...
1232         (make_pass_dse): ...new function.
1233         (pass_nrv): Replace declaration with that of...
1234         (make_pass_nrv): ...new function.
1235         (pass_rename_ssa_copies): Replace declaration with that of...
1236         (make_pass_rename_ssa_copies): ...new function.
1237         (pass_sink_code): Replace declaration with that of...
1238         (make_pass_sink_code): ...new function.
1239         (pass_fre): Replace declaration with that of...
1240         (make_pass_fre): ...new function.
1241         (pass_check_data_deps): Replace declaration with that of...
1242         (make_pass_check_data_deps): ...new function.
1243         (pass_copy_prop): Replace declaration with that of...
1244         (make_pass_copy_prop): ...new function.
1245         (pass_vrp): Replace declaration with that of...
1246         (make_pass_vrp): ...new function.
1247         (pass_uncprop): Replace declaration with that of...
1248         (make_pass_uncprop): ...new function.
1249         (pass_return_slot): Replace declaration with that of...
1250         (make_pass_return_slot): ...new function.
1251         (pass_reassoc): Replace declaration with that of...
1252         (make_pass_reassoc): ...new function.
1253         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1254         (make_pass_rebuild_cgraph_edges): ...new function.
1255         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1256         (make_pass_remove_cgraph_callee_edges): ...new function.
1257         (pass_build_cgraph_edges): Replace declaration with that of...
1258         (make_pass_build_cgraph_edges): ...new function.
1259         (pass_local_pure_const): Replace declaration with that of...
1260         (make_pass_local_pure_const): ...new function.
1261         (pass_tracer): Replace declaration with that of...
1262         (make_pass_tracer): ...new function.
1263         (pass_warn_unused_result): Replace declaration with that of...
1264         (make_pass_warn_unused_result): ...new function.
1265         (pass_diagnose_tm_blocks): Replace declaration with that of...
1266         (make_pass_diagnose_tm_blocks): ...new function.
1267         (pass_lower_tm): Replace declaration with that of...
1268         (make_pass_lower_tm): ...new function.
1269         (pass_tm_init): Replace declaration with that of...
1270         (make_pass_tm_init): ...new function.
1271         (pass_tm_mark): Replace declaration with that of...
1272         (make_pass_tm_mark): ...new function.
1273         (pass_tm_memopt): Replace declaration with that of...
1274         (make_pass_tm_memopt): ...new function.
1275         (pass_tm_edges): Replace declaration with that of...
1276         (make_pass_tm_edges): ...new function.
1277         (pass_split_functions): Replace declaration with that of...
1278         (make_pass_split_functions): ...new function.
1279         (pass_feedback_split_functions): Replace declaration with that of...
1280         (make_pass_feedback_split_functions): ...new function.
1281         (pass_strength_reduction): Replace declaration with that of...
1282         (make_pass_strength_reduction): ...new function.
1283         (pass_ipa_lower_emutls): Replace declaration with that of...
1284         (make_pass_ipa_lower_emutls): ...new function.
1285         (pass_ipa_function_and_variable_visibility): Replace declaration with
1286         that of...
1287         (make_pass_ipa_function_and_variable_visibility): ...new function.
1288         (pass_ipa_tree_profile): Replace declaration with that of...
1289         (make_pass_ipa_tree_profile): ...new function.
1290         (pass_early_local_passes): Replace declaration with that of...
1291         (make_pass_early_local_passes): ...new function.
1292         (pass_ipa_whole_program_visibility): Replace declaration with that
1293         of...
1294         (make_pass_ipa_whole_program_visibility): ...new function.
1295         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1296         (make_pass_ipa_lto_gimple_out): ...new function.
1297         (pass_ipa_increase_alignment): Replace declaration with that of...
1298         (make_pass_ipa_increase_alignment): ...new function.
1299         (pass_ipa_inline): Replace declaration with that of...
1300         (make_pass_ipa_inline): ...new function.
1301         (pass_ipa_free_lang_data): Replace declaration with that of...
1302         (make_pass_ipa_free_lang_data): ...new function.
1303         (pass_ipa_free_inline_summary): Replace declaration with that of...
1304         (make_pass_ipa_free_inline_summary): ...new function.
1305         (pass_ipa_cp): Replace declaration with that of...
1306         (make_pass_ipa_cp): ...new function.
1307         (pass_ipa_reference): Replace declaration with that of...
1308         (make_pass_ipa_reference): ...new function.
1309         (pass_ipa_pure_const): Replace declaration with that of...
1310         (make_pass_ipa_pure_const): ...new function.
1311         (pass_ipa_pta): Replace declaration with that of...
1312         (make_pass_ipa_pta): ...new function.
1313         (pass_ipa_lto_finish_out): Replace declaration with that of...
1314         (make_pass_ipa_lto_finish_out): ...new function.
1315         (pass_ipa_tm): Replace declaration with that of...
1316         (make_pass_ipa_tm): ...new function.
1317         (pass_ipa_profile): Replace declaration with that of...
1318         (make_pass_ipa_profile): ...new function.
1319         (pass_ipa_cdtor_merge): Replace declaration with that of...
1320         (make_pass_ipa_cdtor_merge): ...new function.
1321         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1322         of...
1323         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1324         (pass_init_datastructures): Replace declaration with that of...
1325         (make_pass_init_datastructures): ...new function.
1326         (pass_fixup_cfg): Replace declaration with that of...
1327         (make_pass_fixup_cfg): ...new function.
1328         (pass_expand): Replace declaration with that of...
1329         (make_pass_expand): ...new function.
1330         (pass_instantiate_virtual_regs): Replace declaration with that of...
1331         (make_pass_instantiate_virtual_regs): ...new function.
1332         (pass_rtl_fwprop): Replace declaration with that of...
1333         (make_pass_rtl_fwprop): ...new function.
1334         (pass_rtl_fwprop_addr): Replace declaration with that of...
1335         (make_pass_rtl_fwprop_addr): ...new function.
1336         (pass_jump): Replace declaration with that of...
1337         (make_pass_jump): ...new function.
1338         (pass_jump2): Replace declaration with that of...
1339         (make_pass_jump2): ...new function.
1340         (pass_lower_subreg): Replace declaration with that of...
1341         (make_pass_lower_subreg): ...new function.
1342         (pass_cse): Replace declaration with that of...
1343         (make_pass_cse): ...new function.
1344         (pass_fast_rtl_dce): Replace declaration with that of...
1345         (make_pass_fast_rtl_dce): ...new function.
1346         (pass_ud_rtl_dce): Replace declaration with that of...
1347         (make_pass_ud_rtl_dce): ...new function.
1348         (pass_rtl_dce): Replace declaration with that of...
1349         (make_pass_rtl_dce): ...new function.
1350         (pass_rtl_dse1): Replace declaration with that of...
1351         (make_pass_rtl_dse1): ...new function.
1352         (pass_rtl_dse2): Replace declaration with that of...
1353         (make_pass_rtl_dse2): ...new function.
1354         (pass_rtl_dse3): Replace declaration with that of...
1355         (make_pass_rtl_dse3): ...new function.
1356         (pass_rtl_cprop): Replace declaration with that of...
1357         (make_pass_rtl_cprop): ...new function.
1358         (pass_rtl_pre): Replace declaration with that of...
1359         (make_pass_rtl_pre): ...new function.
1360         (pass_rtl_hoist): Replace declaration with that of...
1361         (make_pass_rtl_hoist): ...new function.
1362         (pass_rtl_store_motion): Replace declaration with that of...
1363         (make_pass_rtl_store_motion): ...new function.
1364         (pass_cse_after_global_opts): Replace declaration with that of...
1365         (make_pass_cse_after_global_opts): ...new function.
1366         (pass_rtl_ifcvt): Replace declaration with that of...
1367         (make_pass_rtl_ifcvt): ...new function.
1368         (pass_into_cfg_layout_mode): Replace declaration with that of...
1369         (make_pass_into_cfg_layout_mode): ...new function.
1370         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1371         (make_pass_outof_cfg_layout_mode): ...new function.
1372         (pass_loop2): Replace declaration with that of...
1373         (make_pass_loop2): ...new function.
1374         (pass_rtl_loop_init): Replace declaration with that of...
1375         (make_pass_rtl_loop_init): ...new function.
1376         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1377         (make_pass_rtl_move_loop_invariants): ...new function.
1378         (pass_rtl_unswitch): Replace declaration with that of...
1379         (make_pass_rtl_unswitch): ...new function.
1380         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1381         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1382         (pass_rtl_doloop): Replace declaration with that of...
1383         (make_pass_rtl_doloop): ...new function.
1384         (pass_rtl_loop_done): Replace declaration with that of...
1385         (make_pass_rtl_loop_done): ...new function.
1386         (pass_web): Replace declaration with that of...
1387         (make_pass_web): ...new function.
1388         (pass_cse2): Replace declaration with that of...
1389         (make_pass_cse2): ...new function.
1390         (pass_df_initialize_opt): Replace declaration with that of...
1391         (make_pass_df_initialize_opt): ...new function.
1392         (pass_df_initialize_no_opt): Replace declaration with that of...
1393         (make_pass_df_initialize_no_opt): ...new function.
1394         (pass_reginfo_init): Replace declaration with that of...
1395         (make_pass_reginfo_init): ...new function.
1396         (pass_inc_dec): Replace declaration with that of...
1397         (make_pass_inc_dec): ...new function.
1398         (pass_stack_ptr_mod): Replace declaration with that of...
1399         (make_pass_stack_ptr_mod): ...new function.
1400         (pass_initialize_regs): Replace declaration with that of...
1401         (make_pass_initialize_regs): ...new function.
1402         (pass_combine): Replace declaration with that of...
1403         (make_pass_combine): ...new function.
1404         (pass_if_after_combine): Replace declaration with that of...
1405         (make_pass_if_after_combine): ...new function.
1406         (pass_ree): Replace declaration with that of...
1407         (make_pass_ree): ...new function.
1408         (pass_partition_blocks): Replace declaration with that of...
1409         (make_pass_partition_blocks): ...new function.
1410         (pass_match_asm_constraints): Replace declaration with that of...
1411         (make_pass_match_asm_constraints): ...new function.
1412         (pass_regmove): Replace declaration with that of...
1413         (make_pass_regmove): ...new function.
1414         (pass_split_all_insns): Replace declaration with that of...
1415         (make_pass_split_all_insns): ...new function.
1416         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1417         (make_pass_fast_rtl_byte_dce): ...new function.
1418         (pass_lower_subreg2): Replace declaration with that of...
1419         (make_pass_lower_subreg2): ...new function.
1420         (pass_mode_switching): Replace declaration with that of...
1421         (make_pass_mode_switching): ...new function.
1422         (pass_sms): Replace declaration with that of...
1423         (make_pass_sms): ...new function.
1424         (pass_sched): Replace declaration with that of...
1425         (make_pass_sched): ...new function.
1426         (pass_ira): Replace declaration with that of...
1427         (make_pass_ira): ...new function.
1428         (pass_reload): Replace declaration with that of...
1429         (make_pass_reload): ...new function.
1430         (pass_clean_state): Replace declaration with that of...
1431         (make_pass_clean_state): ...new function.
1432         (pass_branch_prob): Replace declaration with that of...
1433         (make_pass_branch_prob): ...new function.
1434         (pass_value_profile_transformations): Replace declaration with that
1435         of...
1436         (make_pass_value_profile_transformations): ...new function.
1437         (pass_postreload_cse): Replace declaration with that of...
1438         (make_pass_postreload_cse): ...new function.
1439         (pass_gcse2): Replace declaration with that of...
1440         (make_pass_gcse2): ...new function.
1441         (pass_split_after_reload): Replace declaration with that of...
1442         (make_pass_split_after_reload): ...new function.
1443         (pass_branch_target_load_optimize1): Replace declaration with that
1444         of...
1445         (make_pass_branch_target_load_optimize1): ...new function.
1446         (pass_thread_prologue_and_epilogue): Replace declaration with that
1447         of...
1448         (make_pass_thread_prologue_and_epilogue): ...new function.
1449         (pass_stack_adjustments): Replace declaration with that of...
1450         (make_pass_stack_adjustments): ...new function.
1451         (pass_peephole2): Replace declaration with that of...
1452         (make_pass_peephole2): ...new function.
1453         (pass_if_after_reload): Replace declaration with that of...
1454         (make_pass_if_after_reload): ...new function.
1455         (pass_regrename): Replace declaration with that of...
1456         (make_pass_regrename): ...new function.
1457         (pass_cprop_hardreg): Replace declaration with that of...
1458         (make_pass_cprop_hardreg): ...new function.
1459         (pass_reorder_blocks): Replace declaration with that of...
1460         (make_pass_reorder_blocks): ...new function.
1461         (pass_branch_target_load_optimize2): Replace declaration with that
1462         of...
1463         (make_pass_branch_target_load_optimize2): ...new function.
1464         (pass_leaf_regs): Replace declaration with that of...
1465         (make_pass_leaf_regs): ...new function.
1466         (pass_split_before_sched2): Replace declaration with that of...
1467         (make_pass_split_before_sched2): ...new function.
1468         (pass_compare_elim_after_reload): Replace declaration with that of...
1469         (make_pass_compare_elim_after_reload): ...new function.
1470         (pass_sched2): Replace declaration with that of...
1471         (make_pass_sched2): ...new function.
1472         (pass_stack_regs): Replace declaration with that of...
1473         (make_pass_stack_regs): ...new function.
1474         (pass_stack_regs_run): Replace declaration with that of...
1475         (make_pass_stack_regs_run): ...new function.
1476         (pass_df_finish): Replace declaration with that of...
1477         (make_pass_df_finish): ...new function.
1478         (pass_compute_alignments): Replace declaration with that of...
1479         (make_pass_compute_alignments): ...new function.
1480         (pass_duplicate_computed_gotos): Replace declaration with that of...
1481         (make_pass_duplicate_computed_gotos): ...new function.
1482         (pass_variable_tracking): Replace declaration with that of...
1483         (make_pass_variable_tracking): ...new function.
1484         (pass_free_cfg): Replace declaration with that of...
1485         (make_pass_free_cfg): ...new function.
1486         (pass_machine_reorg): Replace declaration with that of...
1487         (make_pass_machine_reorg): ...new function.
1488         (pass_cleanup_barriers): Replace declaration with that of...
1489         (make_pass_cleanup_barriers): ...new function.
1490         (pass_delay_slots): Replace declaration with that of...
1491         (make_pass_delay_slots): ...new function.
1492         (pass_split_for_shorten_branches): Replace declaration with that of...
1493         (make_pass_split_for_shorten_branches): ...new function.
1494         (pass_split_before_regstack): Replace declaration with that of...
1495         (make_pass_split_before_regstack): ...new function.
1496         (pass_convert_to_eh_region_ranges): Replace declaration with that
1497         of...
1498         (make_pass_convert_to_eh_region_ranges): ...new function.
1499         (pass_shorten_branches): Replace declaration with that of...
1500         (make_pass_shorten_branches): ...new function.
1501         (pass_set_nothrow_function_flags): Replace declaration with that of...
1502         (make_pass_set_nothrow_function_flags): ...new function.
1503         (pass_dwarf2_frame): Replace declaration with that of...
1504         (make_pass_dwarf2_frame): ...new function.
1505         (pass_final): Replace declaration with that of...
1506         (make_pass_final): ...new function.
1507         (pass_rtl_seqabstr): Replace declaration with that of...
1508         (make_pass_rtl_seqabstr): ...new function.
1509         (pass_release_ssa_names): Replace declaration with that of...
1510         (make_pass_release_ssa_names): ...new function.
1511         (pass_early_inline): Replace declaration with that of...
1512         (make_pass_early_inline): ...new function.
1513         (pass_inline_parameters): Replace declaration with that of...
1514         (make_pass_inline_parameters): ...new function.
1515         (pass_update_address_taken): Replace declaration with that of...
1516         (make_pass_update_address_taken): ...new function.
1517         (pass_convert_switch): Replace declaration with that of...
1518         (make_pass_convert_switch): ...new function.
1519         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1520         to a subclass of simple_ipa_opt_pass along with...
1521         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1522         (make_pass_ipa_tree_profile): ...new function.
1523         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1524         subclass of gimple_opt_pass along with...
1525         (pass_data_sra_early): ...new pass_data instance and...
1526         (make_pass_sra_early): ...new function.
1527         (pass_sra): Convert from a global struct to a subclass of
1528         gimple_opt_pass along with...
1529         (pass_data_sra): ...new pass_data instance and...
1530         (make_pass_sra): ...new function.
1531         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1532         gimple_opt_pass along with...
1533         (pass_data_early_ipa_sra): ...new pass_data instance and...
1534         (make_pass_early_ipa_sra): ...new function.
1535         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1536         subclass of gimple_opt_pass along with...
1537         (pass_data_ccp): ...new pass_data instance and...
1538         (make_pass_ccp): ...new function.
1539         (pass_fold_builtins): Convert from a global struct to a subclass of
1540         gimple_opt_pass along with...
1541         (pass_data_fold_builtins): ...new pass_data instance and...
1542         (make_pass_fold_builtins): ...new function.
1543         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1544         subclass of gimple_opt_pass along with...
1545         (pass_data_copy_prop): ...new pass_data instance and...
1546         (make_pass_copy_prop): ...new function.
1547         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1548         global struct to a subclass of gimple_opt_pass along with...
1549         (pass_data_rename_ssa_copies): ...new pass_data instance and...
1550         (make_pass_rename_ssa_copies): ...new function.
1551         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1552         subclass of gimple_opt_pass along with...
1553         (pass_data_dce): ...new pass_data instance and...
1554         (make_pass_dce): ...new function.
1555         (pass_dce_loop): Convert from a global struct to a subclass of
1556         gimple_opt_pass along with...
1557         (pass_data_dce_loop): ...new pass_data instance and...
1558         (make_pass_dce_loop): ...new function.
1559         (pass_cd_dce): Convert from a global struct to a subclass of
1560         gimple_opt_pass along with...
1561         (pass_data_cd_dce): ...new pass_data instance and...
1562         (make_pass_cd_dce): ...new function.
1563         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1564         subclass of gimple_opt_pass along with...
1565         (pass_data_dominator): ...new pass_data instance and...
1566         (make_pass_dominator): ...new function.
1567         (pass_phi_only_cprop): Convert from a global struct to a subclass of
1568         gimple_opt_pass along with...
1569         (pass_data_phi_only_cprop): ...new pass_data instance and...
1570         (make_pass_phi_only_cprop): ...new function.
1571         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1572         subclass of gimple_opt_pass along with...
1573         (pass_data_dse): ...new pass_data instance and...
1574         (make_pass_dse): ...new function.
1575         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1576         a subclass of gimple_opt_pass along with...
1577         (pass_data_forwprop): ...new pass_data instance and...
1578         (make_pass_forwprop): ...new function.
1579         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1580         struct to a subclass of gimple_opt_pass along with...
1581         (pass_data_tree_ifcombine): ...new pass_data instance and...
1582         (make_pass_tree_ifcombine): ...new function.
1583         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1584         subclass of gimple_opt_pass along with...
1585         (pass_data_ch): ...new pass_data instance and...
1586         (make_pass_ch): ...new function.
1587         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1588         subclass of gimple_opt_pass along with...
1589         (pass_data_tree_loop): ...new pass_data instance and...
1590         (make_pass_tree_loop): ...new function.
1591         (pass_tree_loop_init): Convert from a global struct to a subclass of
1592         gimple_opt_pass along with...
1593         (pass_data_tree_loop_init): ...new pass_data instance and...
1594         (make_pass_tree_loop_init): ...new function.
1595         (pass_lim): Convert from a global struct to a subclass of
1596         gimple_opt_pass along with...
1597         (pass_data_lim): ...new pass_data instance and...
1598         (make_pass_lim): ...new function.
1599         (pass_tree_unswitch): Convert from a global struct to a subclass of
1600         gimple_opt_pass along with...
1601         (pass_data_tree_unswitch): ...new pass_data instance and...
1602         (make_pass_tree_unswitch): ...new function.
1603         (pass_predcom): Convert from a global struct to a subclass of
1604         gimple_opt_pass along with...
1605         (pass_data_predcom): ...new pass_data instance and...
1606         (make_pass_predcom): ...new function.
1607         (pass_vectorize): Convert from a global struct to a subclass of
1608         gimple_opt_pass along with...
1609         (pass_data_vectorize): ...new pass_data instance and...
1610         (make_pass_vectorize): ...new function.
1611         (pass_graphite): Convert from a global struct to a subclass of
1612         gimple_opt_pass along with...
1613         (pass_data_graphite): ...new pass_data instance and...
1614         (make_pass_graphite): ...new function.
1615         (pass_graphite_transforms): Convert from a global struct to a subclass
1616         of gimple_opt_pass along with...
1617         (pass_data_graphite_transforms): ...new pass_data instance and...
1618         (make_pass_graphite_transforms): ...new function.
1619         (pass_check_data_deps): Convert from a global struct to a subclass of
1620         gimple_opt_pass along with...
1621         (pass_data_check_data_deps): ...new pass_data instance and...
1622         (make_pass_check_data_deps): ...new function.
1623         (pass_iv_canon): Convert from a global struct to a subclass of
1624         gimple_opt_pass along with...
1625         (pass_data_iv_canon): ...new pass_data instance and...
1626         (make_pass_iv_canon): ...new function.
1627         (pass_scev_cprop): Convert from a global struct to a subclass of
1628         gimple_opt_pass along with...
1629         (pass_data_scev_cprop): ...new pass_data instance and...
1630         (make_pass_scev_cprop): ...new function.
1631         (pass_record_bounds): Convert from a global struct to a subclass of
1632         gimple_opt_pass along with...
1633         (pass_data_record_bounds): ...new pass_data instance and...
1634         (make_pass_record_bounds): ...new function.
1635         (pass_complete_unroll): Convert from a global struct to a subclass of
1636         gimple_opt_pass along with...
1637         (pass_data_complete_unroll): ...new pass_data instance and...
1638         (make_pass_complete_unroll): ...new function.
1639         (pass_complete_unrolli): Convert from a global struct to a subclass of
1640         gimple_opt_pass along with...
1641         (pass_data_complete_unrolli): ...new pass_data instance and...
1642         (make_pass_complete_unrolli): ...new function.
1643         (pass_parallelize_loops): Convert from a global struct to a subclass
1644         of gimple_opt_pass along with...
1645         (pass_data_parallelize_loops): ...new pass_data instance and...
1646         (make_pass_parallelize_loops): ...new function.
1647         (pass_loop_prefetch): Convert from a global struct to a subclass of
1648         gimple_opt_pass along with...
1649         (pass_data_loop_prefetch): ...new pass_data instance and...
1650         (make_pass_loop_prefetch): ...new function.
1651         (pass_iv_optimize): Convert from a global struct to a subclass of
1652         gimple_opt_pass along with...
1653         (pass_data_iv_optimize): ...new pass_data instance and...
1654         (make_pass_iv_optimize): ...new function.
1655         (pass_tree_loop_done): Convert from a global struct to a subclass of
1656         gimple_opt_pass along with...
1657         (pass_data_tree_loop_done): ...new pass_data instance and...
1658         (make_pass_tree_loop_done): ...new function.
1659         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1660         struct to a subclass of gimple_opt_pass along with...
1661         (pass_data_cse_reciprocals): ...new pass_data instance and...
1662         (make_pass_cse_reciprocals): ...new function.
1663         (pass_cse_sincos): Convert from a global struct to a subclass of
1664         gimple_opt_pass along with...
1665         (pass_data_cse_sincos): ...new pass_data instance and...
1666         (make_pass_cse_sincos): ...new function.
1667         (pass_optimize_bswap): Convert from a global struct to a subclass of
1668         gimple_opt_pass along with...
1669         (pass_data_optimize_bswap): ...new pass_data instance and...
1670         (make_pass_optimize_bswap): ...new function.
1671         (pass_optimize_widening_mul): Convert from a global struct to a
1672         subclass of gimple_opt_pass along with...
1673         (pass_data_optimize_widening_mul): ...new pass_data instance and...
1674         (make_pass_optimize_widening_mul): ...new function.
1675         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1676         subclass of gimple_opt_pass along with...
1677         (pass_data_phiopt): ...new pass_data instance and...
1678         (make_pass_phiopt): ...new function.
1679         (pass_cselim): Convert from a global struct to a subclass of
1680         gimple_opt_pass along with...
1681         (pass_data_cselim): ...new pass_data instance and...
1682         (make_pass_cselim): ...new function.
1683         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1684         subclass of gimple_opt_pass along with...
1685         (pass_data_phiprop): ...new pass_data instance and...
1686         (make_pass_phiprop): ...new function.
1687         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1688         subclass of gimple_opt_pass along with...
1689         (pass_data_pre): ...new pass_data instance and...
1690         (make_pass_pre): ...new function.
1691         (pass_fre): Convert from a global struct to a subclass of
1692         gimple_opt_pass along with...
1693         (pass_data_fre): ...new pass_data instance and...
1694         (make_pass_fre): ...new function.
1695         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1696         subclass of gimple_opt_pass along with...
1697         (pass_data_reassoc): ...new pass_data instance and...
1698         (make_pass_reassoc): ...new function.
1699         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1700         subclass of gimple_opt_pass along with...
1701         (pass_data_sink_code): ...new pass_data instance and...
1702         (make_pass_sink_code): ...new function.
1703         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1704         subclass of gimple_opt_pass along with...
1705         (pass_data_strlen): ...new pass_data instance and...
1706         (make_pass_strlen): ...new function.
1707         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1708         struct to a subclass of gimple_opt_pass along with...
1709         (pass_data_build_alias): ...new pass_data instance and...
1710         (make_pass_build_alias): ...new function.
1711         (pass_build_ealias): Convert from a global struct to a subclass of
1712         gimple_opt_pass along with...
1713         (pass_data_build_ealias): ...new pass_data instance and...
1714         (make_pass_build_ealias): ...new function.
1715         (pass_ipa_pta): Convert from a global struct to a subclass of
1716         simple_ipa_opt_pass along with...
1717         (pass_data_ipa_pta): ...new pass_data instance and...
1718         (make_pass_ipa_pta): ...new function.
1719         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1720         subclass of gimple_opt_pass along with...
1721         (pass_data_uncprop): ...new pass_data instance and...
1722         (make_pass_uncprop): ...new function.
1723         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1724         global struct to a subclass of gimple_opt_pass along with...
1725         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1726         (make_pass_late_warn_uninitialized): ...new function.
1727         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1728         to a subclass of gimple_opt_pass along with...
1729         (pass_data_init_datastructures): ...new pass_data instance and...
1730         (make_pass_init_datastructures): ...new function.
1731         (pass_early_warn_uninitialized): Convert from a global struct to a
1732         subclass of gimple_opt_pass along with...
1733         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1734         (make_pass_early_warn_uninitialized): ...new function.
1735         (pass_update_address_taken): Convert from a global struct to a
1736         subclass of gimple_opt_pass along with...
1737         (pass_data_update_address_taken): ...new pass_data instance and...
1738         (make_pass_update_address_taken): ...new function.
1739         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1740         struct to a subclass of gimple_opt_pass along with...
1741         (pass_data_release_ssa_names): ...new pass_data instance and...
1742         (make_pass_release_ssa_names): ...new function.
1743         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1744         subclass of gimple_opt_pass along with...
1745         (pass_data_stdarg): ...new pass_data instance and...
1746         (make_pass_stdarg): ...new function.
1747         * tree-switch-conversion.c (pass_convert_switch): Convert from a
1748         global struct to a subclass of gimple_opt_pass along with...
1749         (pass_data_convert_switch): ...new pass_data instance and...
1750         (make_pass_convert_switch): ...new function.
1751         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1752         to a subclass of gimple_opt_pass along with...
1753         (pass_data_tail_recursion): ...new pass_data instance and...
1754         (make_pass_tail_recursion): ...new function.
1755         (pass_tail_calls): Convert from a global struct to a subclass of
1756         gimple_opt_pass along with...
1757         (pass_data_tail_calls): ...new pass_data instance and...
1758         (make_pass_tail_calls): ...new function.
1759         * tree-vect-generic.c (pass_lower_vector): Convert from a global
1760         struct to a subclass of gimple_opt_pass along with...
1761         (pass_data_lower_vector): ...new pass_data instance and...
1762         (make_pass_lower_vector): ...new function.
1763         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1764         gimple_opt_pass along with...
1765         (pass_data_lower_vector_ssa): ...new pass_data instance and...
1766         (make_pass_lower_vector_ssa): ...new function.
1767         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1768         to a subclass of gimple_opt_pass along with...
1769         (pass_data_slp_vectorize): ...new pass_data instance and...
1770         (make_pass_slp_vectorize): ...new function.
1771         (pass_ipa_increase_alignment): Convert from a global struct to a
1772         subclass of simple_ipa_opt_pass along with...
1773         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1774         (make_pass_ipa_increase_alignment): ...new function.
1775         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1776         gimple_opt_pass along with...
1777         (pass_data_vrp): ...new pass_data instance and...
1778         (make_pass_vrp): ...new function.
1779         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1780         subclass of simple_ipa_opt_pass along with...
1781         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1782         (make_pass_ipa_free_lang_data): ...new function.
1783         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1784         gimple_opt_pass along with...
1785         (pass_data_tsan): ...new pass_data instance and...
1786         (make_pass_tsan): ...new function.
1787         (pass_tsan_O0): Convert from a global struct to a subclass of
1788         gimple_opt_pass along with...
1789         (pass_data_tsan_O0): ...new pass_data instance and...
1790         (make_pass_tsan_O0): ...new function.
1791         * var-tracking.c (pass_variable_tracking): Convert from a global
1792         struct to a subclass of rtl_opt_pass along with...
1793         (pass_data_variable_tracking): ...new pass_data instance and...
1794         (make_pass_variable_tracking): ...new function.
1795         * web.c (pass_web): Convert from a global struct to a subclass of
1796         rtl_opt_pass along with...
1797         (pass_data_web): ...new pass_data instance and...
1798         (make_pass_web): ...new function.
1799         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1800         declaration with that of...
1801         (make_pass_mode_switch_use): ...new function.
1802         (pass_resolve_sw_modes): Replace declaration with that of...
1803         (make_pass_resolve_sw_modes): ...new function.
1804         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1805         from a global struct to a subclass of rtl_opt_pass along with...
1806         (pass_data_mode_switch_use): ...new pass_data instance and...
1807         (make_pass_mode_switch_use): ...new function.
1808         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1809         from a global struct to a subclass of rtl_opt_pass along with...
1810         (pass_data_resolve_sw_modes): ...new pass_data instance and...
1811         (make_pass_resolve_sw_modes): ...new function.
1812         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1813         struct to a subclass of rtl_opt_pass along with...
1814         (pass_data_insert_vzeroupper): ...new pass_data instance and...
1815         (make_pass_insert_vzeroupper): ...new function.
1816         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1817         global struct to a subclass of rtl_opt_pass along with...
1818         (pass_data_work_around_errata): ...new pass_data instance and...
1819         (make_pass_work_around_errata): ...new function.
1820         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1821         struct to a subclass of rtl_opt_pass along with...
1822         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1823         (make_pass_mips_machine_reorg2): ...new function.
1825 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1827         * passes.c (pass_manager::operator new): New.
1829 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1831         Handwritten part of conversion of passes to C++ classes.
1833         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1834         (toplev.o): Add dep on PASS_MANAGER_H.
1835         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1836         of early local pases to reflect this moving from a global to a
1837         member of gcc::pass_manager.
1838         (cgraph_add_new_function): Likewise.
1839         * lto-cgraph.c (lto_output_node): Update for conversion of
1840         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1841         * passes.c (opt_pass::clone): New.
1842         (opt_pass::gate): New.
1843         (opt_pass::execute): New.
1844         (opt_pass::opt_pass): New.
1845         (pass_manager::execute_early_local_passes): New.
1846         (pass_manager::execute_pass_mode_switching): new.
1847         (finish_optimization_passes): Convert to...
1848         (pass_manager::finish_optimization_passes): ...this.
1849         (finish_optimization_passes): Update for conversion of passes to
1850         C++ classes.
1851         (register_dump_files_1): Use has_gate since we cannot portably
1852         check a vtable entry against NULL.
1853         (dump_one_pass): Likewise.
1854         (ipa_write_summaries_2): Likewise.
1855         (ipa_write_optimization_summaries_1): Likewise.
1856         (ipa_read_summaries_1): Likewise.
1857         (ipa_read_optimization_summaries_1): Likewise.
1858         (execute_ipa_stmt_fixups): Likewise.
1859         (pass_manager::pass_manager): Rewrite pass-creation, invoking
1860         pass-creation functions rather than wiring up globals, and
1861         storing the results in fields of pass_manager generated using
1862         pass-instances.def.
1863         (pass_manager::dump_profile_report): Update for conversion of
1864         passes to C++ classes.
1865         (pass_manager::execute_ipa_summary_passes): Likewise.
1866         (execute_one_ipa_transform_pass): Likewise.
1867         (execute_one_pass): Use has_gate and has_execute since we cannot
1868         portably check a vtable entry against NULL.
1869         * pass_manager.h (pass_manager::finish_optimization_passes): New.
1870         (pass_manager): Use pass-instances.def to add fields for the
1871         various pass instances.
1872         * toplev.c (finalize): Update for move of
1873         finish_optimization_passes to a method of gcc::pass_manager.
1874         * toplev.h (finish_optimization_passes): Move to method of class
1875         pass_manager.
1876         * tree-pass.h (struct pass_data): New.
1877         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
1878         (opt_pass::gate): Convert to virtual function.
1879         (opt_pass::~opt_pass): New.
1880         (opt_pass::clone): New.
1881         (opt_pass::execute): Convert to virtual function.
1882         (opt_pass::opt_pass): New.
1883         (opt_pass::ctxt_): new.
1884         (gimple_opt_pass): Convert to subclass of opt_pass.
1885         (gimple_opt_pass::gimple_opt_pass): New.
1886         (rtl_opt_pass): Convert to subclass of opt_pass.
1887         (rtl_opt_pass::rtl_opt_pass): New.
1888         (ipa_opt_pass_d): Convert to subclass of opt_pass.
1889         (ipa_opt_pass_d::ipa_opt_pass_d): New.
1890         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
1891         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
1892         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
1893         invocation of pass_mode_switching to reflect this moving from a
1894         global to a member of gcc::pass_manager.
1895         (ix86_option_override): Rework how pass_insert_vzeroupper is
1896         added to the pass_manager to reflect autogenerated changes.
1897         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
1899 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
1901         PR rtl-optimization/57708
1902         * recog.c (peep2_find_free_register): Validate all regs in a
1903         multi-reg mode.
1905 2013-08-05  Jan Hubicka  <jh@suse.cz>
1907         PR lto/57602
1908         * cgraph.c (verify_cgraph_node): Accept local flags from other
1909         partitions.
1910         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
1911         (function_and_variable_visibility): Likewise.
1912         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
1914 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1916         * graph.c (init_graph_slim_pretty_print): Remove.
1917         (print_graph_cfg): Do not call it.  Use local pretty printer.
1918         (start_graph_dump): Likewise.
1920 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1922         * gimple-pretty-print.c (buffer): Remove.
1923         (initialized): Likewise.
1924         (maybe_init_pretty_print): Likewise.
1925         (print_gimple_stmt): Do not call it.  Use non-static local
1926         pretty_printer variable.
1927         (print_gimple_expr): Likewise.
1928         (print_gimple_seq): Likewise.
1929         (gimple_dump_bb): Likewise.
1931 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1933         * asan.c (asan_pp): Remove.
1934         (asan_pp_initialized): Likewise.
1935         (asan_pp_initialize): Likewise.
1936         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
1937         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
1938         (asan_add_global): Likewise.
1940 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1942         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
1943         * diagnostic.h (diagnostic_flush_buffer): Adjust.
1944         * pretty-print.c (pp_formatted_text_data): Likewise.
1945         (pp_indent): Rename from pp_base_indent.
1946         (pp_format): Rename from pp_base_format.
1947         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
1948         (pp_format_verbatim): Rename from pp_base_format_verbatim.
1949         (pp_flush): Rename from pp_base_flush.
1950         (pp_set_line_maximum_length): Rename from
1951         pp_base_set_line_maximum_length.
1952         (pp_clear_output_area): Rename from pp_base_clear_output_area.
1953         (pp_set_prefix): Rename from pp_base_set_prefix.
1954         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
1955         (pp_emit_prefix): Rename from pp_base_emit_prefix.
1956         (pp_append_text): Rename from pp_base_append_text.
1957         (pp_formatted_text): Rename from pp_base_formatted_text.
1958         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
1959         (pp_remaining_character_count_for_line): Rename from
1960         pp_base_remaining_character_count_for_line.
1961         (pp_newline): Rename from pp_base_newline.
1962         (pp_character): Rename from pp_base_character.
1963         (pp_string): Rename from pp_base_string.
1964         (pp_maybe_space): Rename from pp_base_maybe_space.
1965         * asan.c (asan_pp_string): Adjust.
1966         (asan_emit_stack_protection): Likewise.
1967         (asan_add_global): Likewise.
1968         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
1969         * tree-mudflap.c (mf_varname_tree): Likewise.
1970         * tree-pretty-print.c (pp_tree_identifier): Rename from
1971         pp_base_tree_identifier.
1972         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
1973         Declare as function.
1975 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1977         * pretty-print.h (pp_bar_bar): New.
1978         (pp_ampersand_ampersand): Likewise.
1979         (pp_less_equal): Likewise.
1980         (pp_greater_equal): Likewise.
1981         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
1982         printer functions instead of pp_string or operators and punctuators.
1983         (dump_gimple_call): Likewise.
1984         (dump_gimple_omp_for): Likewise.
1985         (dump_gimple_transaction): Likewise.
1986         (dump_gimple_phi): Likewise.
1987         (pp_gimple_stmt_1): Likewise.
1988         * sched-vis.c (print_insn): Likewise.
1989         * tree-mudflap.c (mf_varname_tree): Likewise.
1990         * tree-pretty-print.c (dump_block_node): Likewise.
1991         (dump_generic_node): Likewise.
1993 2013-08-02  Jan Hubicka  <jh@suse.cz>
1995         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
1996         boundaries.
1997         * lto-streamer-out.c (tree_is_indexable): Results decls and
1998         parm decls are not indexable.
1999         (DFS_write_tree_body): Do not follow args and results.
2000         (hash_tree): Likewise.
2001         (output_functions): Rearrange so struct function is needed
2002         only when real body is output; be able to also ouptut abstract
2003         functions; output DECL_ARGUMENTS and DECL_RESULT.
2004         (lto_output): When not in WPA, ale store abstract functions.
2005         (write_symbol): Do not care about RESULT_DECL.
2006         (output_symbol_p): Handle correctly sbtract decls.
2007         * lto-streamer-in.c (input_function): Rearrange so struct
2008         function can be NULL at entry; allow streaming of
2009         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2010         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2011         sanity check during LTO.
2012         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2013         RESULT_DECl and DECL_ARGUMENTS.
2014         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2015         Likewise.
2017 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2019         * pretty-print.h (pp_underscore): New.
2020         (pp_comma): Tidy.
2021         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2022         printer functions instead of pp_character.
2023         (dump_binary_rhs): Likewise.
2024         (dump_ternary_rhs): Likewise.
2025         (dump_gimple_call_args): Likewise.
2026         (pp_points_to_solution): Likewise.
2027         (dump_gimple_call): Likewise.
2028         (dump_gimple_switch): Likewise.
2029         (dump_gimple_cond): Likewise.
2030         (dump_gimple_bind): Likewise.
2031         (dump_gimple_try): Likewise.
2032         (dump_gimple_omp_for): Likewise.
2033         (dump_gimple_omp_continue): Likewise.
2034         (dump_gimple_omp_single): Likewise.
2035         (dump_gimple_omp_sections): Likewise.
2036         (dump_gimple_omp_block): Likewise.
2037         (dump_gimple_omp_critical): Likewise.
2038         (dump_gimple_transaction): Likewise.
2039         (dump_gimple_asm): Likewise.
2040         (dump_gimple_phi): Likewise.
2041         (dump_gimple_omp_parallel): Likewise.
2042         (dump_gimple_omp_task): Likewise.
2043         (dump_gimple_omp_atomic_load): Likewise.
2044         (dump_gimple_omp_atomic_store): Likewise.
2045         (dump_gimple_mem_ops): Likewise.
2046         (pp_gimple_stmt_1): Likewise.
2047         (pp_cfg_jump): Likewise.
2048         (dump_implicit_edges): Likewise.
2049         (gimple_dump_bb_for_graph): Likewise.
2050         * graph.c (draw_cfg_node): Likewise.
2051         * langhooks.c (lhd_print_error_function): Likewise.
2052         * sched-vis.c (print_exp): Likewise.
2053         (print_value): Likewise.
2054         (print_pattern): Likewise.
2055         (print_insn): Likewise.
2056         (rtl_dump_bb_for_graph): Likewise.
2057         * tree-pretty-print.c (dump_function_declaration): Likewise.
2058         (dump_array_domain): Likewise.
2059         (dump_omp_clause): Likewise.
2060         (dump_location): Likewise.
2061         (dump_generic_node): Likewise.
2062         (print_struct_decl): Likewise.
2063         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2065 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2067         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2068         candidate table when replacing a candidate statement.
2069         (replace_rhs_if_not_dup): Likewise.
2070         (replace_one_candidate): Likewise.
2072 2013-08-02  Jan Hubicka  <jh@suse.cz>
2073             Martin Liska  <marxin.liska@gmail.com>
2075         * cgraphunit.c (add_new_function): Fix logic when adding from
2076         late IPA pass.
2077         (assemble_thunk): Rename to ...
2078         (expand_thunk); .. this one; export; get it working with
2079         general functions; make produced gimple valid.
2080         * cgraph.h (expand_thunk): Declare.
2082 2013-08-02  Jan Hubicka  <jh@suse.cz>
2084         * ipa-cp.c (gather_context_independent_values): Use
2085         ipa_get_param_move_cost.
2086         (get_replacement_map): Remove PARAM; move parameter folding
2087         into tree-inline.c
2088         (create_specialized_node): Update.
2089         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2090         assert that we have gimple body; update move_cost.
2091         (count_formal_params): Assert that we have gimple body.
2092         (ipa_dump_param): New function.
2093         (ipa_alloc_node_params): Break out from ...
2094         (ipa_initialize_node_params): ... here.
2095         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2096         (ipa_write_node_info): Stream move costs.
2097         (ipa_read_node_info): Read move costs.
2098         (ipa_update_after_lto_read): Do not recompute node params.
2099         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2100         (ipa_get_param): Check we are not in WPA.
2101         (ipa_get_param_move_cost): New.
2102         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2103         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2104         parm numbers to be present.
2106 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2108         PR rtl-optimization/58048
2109         * lra-constraints.c (process_alt_operands): Don't check asm
2110         operand on register.
2112 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2114         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2115         the implied StoreLoad barrier for atomic operations if before.
2117 2013-08-02  Jan Hubicka  <jh@suse.cz>
2118             Martin Liska  <marxin.liska@gmail.com>
2120         * cgraph.c (cgraph_function_body_availability): Do not check
2121         cgraph flags.
2122         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2123         symtab_node_availability): Declare.
2124         * ipa.c (can_replace_by_local_alias): New.
2125         (function_and_variable_visibility): Use it.
2126         * symtab.c (symtab_for_node_and_aliases,
2127         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2129 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2131         PR rtl-optimization/57963
2132         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2133         (lra_constraints): Use them.
2135 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2137         * config/arm/types.md (define_attr "type"): Add "load_acq"
2138         and "store_rel".
2139         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2140         changes.
2141         (cortex_a53_store1): Likewise.
2143 2013-08-01  Jan Hubicka  <jh@suse.cz>
2145         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2146         partitions are not needed.
2148 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2150         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2151         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2152         MAYBE_NON_Q_CLASS_P where appropriate.
2154 2013-08-01  Jan Hubicka  <jh@suse.cz>
2156         * cgraph.h (release_function_body): Declare.
2157         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2158         of unused delcarations.
2160 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2162         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2163         RTL form when subtracting a constant.
2165 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2167         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2168         Generate canonical plus rtx with negated immediate instead of minus
2169         where appropriate.
2170         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2172 2013-08-01  Jan Hubicka  <jh@suse.cz>
2174         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2175         (cgraph_release_function_body): Likewise.
2176         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2177         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2178         to used_as_abstract_origin.
2179         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2180         symbols used as abstract origins.
2181         * cgraphunit.c (analyze_functions): Update.
2182         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2183         used_as_abstract_origin.
2184         * tree-inline.c (tree_function_versioning): Update
2185         used_as_abstract_origin; be ready for DECL_RESULT and
2186         DECL_ARGUMENTS to be NULL.
2188         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2189         for abstract functions.
2190         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2191         real symbols.
2193 2013-08-01  Jan Hubicka  <jh@suse.cz>
2195         * profile.c (compute_value_histograms): Fix thinko.
2197 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2199         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2200         aarch-common-protos.h to extra_headers.
2201         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2202         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2203         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2205 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2207         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2208         Include "../arm/types.md".  Define "type" attribute for all patterns.
2209         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2210         attribute changes.
2212 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2214         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2215         to support power8 load fusion.
2216         (fusion_gpr_mem_load): Likewise.
2218         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2220         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2221         declarations for power8 load fusion.
2222         (emit_fusion_gpr_load): Likewise.
2224         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2225         tuning for power8, turn on fusion mode by default.  Turn on sign
2226         extending fusion mode if normal fusion mode is on, and we are at
2227         -O2 or -O3.
2228         (fusion_gpr_load_p): New function, return true if we can fuse an
2229         addis instruction with a dependent load to a GPR.
2230         (emit_fusion_gpr_load): Emit the instructions for power8 load
2231         fusion to GPRs.
2233         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2234         (VSX load fusion peepholes): New peepholes to fuse together an
2235         addi instruction with a VSX load instruction.
2237         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2238         peepholes to fuse an addis instruction with a load to a GPR base
2239         register.  If we are supporting sign extending fusions, convert
2240         sign extending loads to zero extending loads and add an explicit
2241         sign extension.
2243 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2245         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2246         aarch-common-protos.h to extra_headers.
2247         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2248         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2249         (arm_early_store_addr_dep): Likewise.
2250         (arm_no_early_alu_shift_dep): Likewise.
2251         (arm_no_early_alu_shift_value_dep): Likewise.
2252         (arm_no_early_mul_dep): Likewise.
2253         (arm_no_early_store_addr_dep): Likewise.
2254         (arm_mac_accumulator_is_mul_result): Likewise.
2255         (arm_mac_accumulator_is_result): Likewise.
2256         * config/arm/aarch-common.c: ... here.  New file.
2257         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2258         here to ...
2259         (arm_early_store_addr_dep): Likewise.
2260         (arm_no_early_alu_shift_dep): Likewise.
2261         (arm_no_early_alu_shift_value_dep): Likewise.
2262         (arm_no_early_mul_dep): Likewise.
2263         (arm_no_early_store_addr_dep): Likewise.
2264         (arm_mac_accumulator_is_mul_result): Likewise.
2265         (arm_mac_accumulator_is_result): Likewise.
2266         * config/arm/aarch-common-protos.h: ... here.  New file.
2267         * config/arm/t-arm (aarch-common.o): Define.
2269 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2271         * config/arm/arm.md: Include new file "types.md".
2272         (define_attr "type"): Move from here to ...
2273         (define_attr "mul32"): Likewise.
2274         (define_attr "mul64"): Likewise.
2275         * config/arm/types.md: ... here.  New file.
2277 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2279         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2280         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2282 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2284         * gen-pass-instances.awk: Fix offset of substr().
2286 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2288         * Makefile.in (pass-instances.def): New.
2289         (passes.o): Replace dependency on passes.def with one on
2290         pass-instances.def
2292         * gen-pass-instances.awk: New.
2294         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2295         rather than passes.def, updating local definition of NEXT_PASS
2296         macro to add an extra NUM parameter (currently unused).
2298 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2300         * Makefile.in (PASS_MANAGER_H): New.
2301         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2302         (passes.o): Likewise.
2303         (statistics.o): Likewise.
2304         (cgraphunit.o): Likewise.
2305         (context.o): Depend on PASS_MANAGER_H.
2307         * pass_manager.h: New.
2309         * cgraphunit.c (cgraph_add_new_function): Update for moves
2310         of globals to fields of pass_manager.
2311         (analyze_function): Likewise.
2312         (expand_function): Likewise.
2313         (ipa_passes): Likewise.
2314         (compile): Likewise.
2316         * context.c (context::context): New.
2317         * context.h  (context::context): New.
2318         (context::get_passes): New.
2319         (context::passes_): New.
2321         * lto-cgraph.c (input_node): Update for moves of globals to
2322         fields of pass_manager.
2324         * passes.c (all_passes): Remove, in favor of a field of the
2325         same name within the new class pass_manager.
2326         (all_small_ipa_passes): Likewise.
2327         (all_lowering_passes): Likewise.
2328         (all_regular_ipa_passes): Likewise.
2329         (all_late_ipa_passes): Likewise.
2330         (all_lto_gen_passes): Likewise.
2331         (passes_by_id): Likewise.
2332         (passes_by_id_size): Likewise.
2333         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2334         the new class pass_manager.
2335         (set_pass_for_id): Convert to...
2336         (pass_manager::set_pass_for_id): ...method.
2337         (get_pass_for_id): Convert to...
2338         (pass_manager::get_pass_for_id): ...method.
2339         (register_one_dump_file): Move body of implementation into...
2340         (pass_manager::register_one_dump_file): ...here.
2341         (register_dump_files_1): Convert to...
2342         (pass_manager::register_dump_files_1): ...method.
2343         (register_dump_files): Convert to...
2344         (pass_manager::register_dump_files): ...method.
2345         (create_pass_tab): Update for moves of globals to fields of
2346         pass_manager.
2347         (dump_passes): Move body of implementation into...
2348         (pass_manager::dump_passes): ...here.
2349         (register_pass): Move body of implementation into...
2350         (pass_manager::register_pass): ...here.
2351         (init_optimization_passes): Convert into...
2352         (pass_manager::pass_manager): ...constructor for new
2353         pass_manager class, and initialize the pass_lists array.
2354         (check_profile_consistency): Update for moves of globals to
2355         fields of pass_manager.
2356         (dump_profile_report): Move body of implementation into...
2357         (pass_manager::dump_profile_report): ...here.
2358         (ipa_write_summaries_1): Update for moves of pass lists from
2359         being globals to fields of pass_manager.
2360         (ipa_write_optimization_summaries): Likewise.
2361         (ipa_read_summaries):  Likewise.
2362         (ipa_read_optimization_summaries): Likewise.
2363         (execute_all_ipa_stmt_fixups): Likewise.
2365         * statistics.c (statistics_fini): Update for moves of globals to
2366         fields of pass_manager.
2368         * toplev.c (general_init): Replace call to
2369         init_optimization_passes with construction of the pass_manager
2370         instance.
2372         * tree-pass.h (all_passes): Remove, in favor of a field of the
2373         same name within the new class pass_manager.
2374         (all_small_ipa_passes): Likewise.
2375         (all_lowering_passes): Likewise.
2376         (all_regular_ipa_passes): Likewise.
2377         (all_lto_gen_passes): Likewise.
2378         (all_late_ipa_passes): Likewise.
2379         (passes_by_id): Likewise.
2380         (passes_by_id_size): Likewise.
2381         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2382         the new class pass_manager.
2383         (get_pass_for_id): Remove.
2385 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2387         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2388         configs.
2390 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2392         * arm.md (mulhi3): New expand pattern.
2394 2013-07-30  Jan Hubicka  <jh@suse.cz>
2395             Martin Liska  <marxin.liska@gmail.com>
2397         * profile.c (compute_value_histograms): Do not ICE when
2398         there is mismatch only on some counters.
2400 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2402         PR rtl-optimization/57637
2403         * function.c (move_insn_for_shrink_wrap): Also check the
2404         GEN set of the LIVE problem for the liveness analysis
2405         if it exists, otherwise give up.
2407 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2409         PR tree-optimization/57993
2410         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2411         replaced statement in the candidate table.
2412         (phi_add_costs): Return infinite cost when the hidden basis does
2413         not dominate all phis on which the candidate is dependent.
2414         (replace_one_candidate): Record replaced statement in the
2415         candidate table.
2417 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2419         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2420         (ashlv2si3): New expander.
2421         (*ashlv2si3_i): New define_insn_and_split.
2422         * predicates.md (float_operation): Allow patterns with three
2423         basic sub-patterns.
2425         PR rtl-optimization/58021
2426         * mode-switching.c (create_pre_exit): Always split off preceding
2427         insns if we are not at the basic block head.
2429 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2431         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2432         (UCLIBC_DYNAMIC_LINKER): New macro.
2433         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2434         `-mnan=2008'.
2435         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2436         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2437         `-mnan=2008'.
2438         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2439         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2440         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2441         for SF and DF modes.  Use ieee_quad_format for TF mode.
2442         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2443         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2444         (mips_option_override): Handle `-mnan=legacy'.
2445         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2446         `-mabs=2008' and `-mnan=2008'.
2447         (OPTION_DEFAULT_SPECS): Add "nan" default.
2448         (ASM_SPEC): Handle `-mnan='.
2449         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2450         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2451         comment accordingly.
2452         (neg<mode>2): Likewise.
2453         * config/mips/mips.opt (mabs, mnan): New options.
2454         * doc/install.texi (Configuration): Document `--with-nan=' option.
2455         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2456         `-mnan=' options.
2457         (MIPS Options): Document them.
2458         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2459         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2460         * configure: Regenerate.
2461         * config.in: Regenerate.
2463 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2465         * config/i386/i386.md (float post-reload splitters): Do not check
2466         for subregs of SSE registers.
2468 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2469             H.J. Lu  <hongjiu.lu@intel.com>
2471         PR target/57954
2472         PR target/57988
2473         * config/i386/i386.md (post-reload splitter
2474         to avoid partial SSE reg dependency stalls): New pattern.
2476 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2478         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2480 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2482         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2483         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2485 2013-07-26  Cary Coutant  <ccoutant@google.com>
2487         * dwarf2out.c (die_checksum_ordered): Don't include template
2488         instantiations in signature.
2489         (is_template_parameter): New function.
2490         (is_template_instantiation): New function.
2491         (generate_skeleton_bottom_up): Don't include template instantiations
2492         in type unit DIE.
2493         (generate_skeleton): Likewise.
2494         (break_out_comdat_types): Move recursive call to break out nested
2495         types earlier.
2496         (prune_unused_types_mark_generic_parms_dies): Call
2497         is_template_parameter.
2499 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2501         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2502         uses vector registers.
2503         * config/aarch64/iterators.md: Add attributes rtn and vas.
2505 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2506             Richard Earnshaw  <richard.earnshaw@arm.com>
2508         * combine.c (simplify_comparison): Re-canonicalize operands
2509         where appropriate.
2510         * config/arm/arm.md (movcond_addsi): New splitter.
2512 2013-07-25  Sterling Augustine  <saugustine@google.com>
2514         * dwarf2out.c (size_of_pubnames): Move code to...
2515         (include_pubname_in_output): ...here.  New.
2516         (want_pubnames): Rearrange.
2517         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2519 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2521         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2523 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2525         PR target/38836
2526         * doc/extend.texi: Remove obsolete builtins. Fix
2527         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2529 2013-07-25  Jan Hubicka  <jh@suse.cz>
2531         * cgraph.c (release_function_body): Break out from ...
2532         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2533         and DECL_ARGUMENTS.
2534         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2535         old_tree in the map.
2536         (create_specialized_node): Update.
2537         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2538         into index.
2539         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2540         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2541         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2542         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2543         DECL_RESULT.
2545 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2547         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2548         addsi3_carryin_alt2_<optab>): Correct output template.
2550 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2552         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2553         Adjust for arm_restrict_it.
2554         Remove trailing whitespace.
2556 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
2558         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2559         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2561         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2563 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
2565         PR rtl-optimization/57960
2566         * lra-constraints.c (process_alt_operands): Use the right mode
2567         when checking strict_low.
2569 2013-07-25  Jan Hubicka  <jh@suse.cz>
2571         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2572         * cgraph.c (cgraph_remove_node): Do not release function body
2573         when in cgraph streaming.
2574         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2575         in other partitions are not considered reachable; fix handling of
2576         clones.
2578 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2580         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2582 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2584         PR target/19599
2585         PR target/57731
2586         PR target/57837
2587         * config/arm/arm.md ("*sibcall_insn): Replace use of
2588         Ss with US. Adjust output for v5 and v4t.
2589         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2591         * config/arm/constraints.md ("Ss"): Rename to US.
2593 2013-07-25  Terry Guo  <terry.guo@arm.com>
2595         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2596         shift_add/shift_sub0/shift_sub1 RTXs.
2598 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
2599             Anton Blanchard <anton@au1.ibm.com>
2601         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2602         (altivec_vpks<VI_char>ss): Likewise.
2603         (altivec_vpks<VI_char>us): Likewise.
2604         (altivec_vpku<VI_char>us): Likewise.
2605         (altivec_vpku<VI_char>um): Likewise.
2607 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
2609         Introduce context class.
2611         * Makefile.in (CONTEXT_H): New.
2612         (OBJS): Add context.o.
2613         (toplev.o): Add CONTEXT_H to dependencies.
2614         (context.o): New.
2616         * toplev.c (general_init): Create the singleton gcc::context instance.
2618         * context.c: New.
2620         * context.h: New.
2622 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2624         PR rtl-optimization/57968
2625         * mode-switching.c (create_pre_exit): Allow instructions that
2626         don't set a return register to need a non-exit mode.
2628 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2629             Anton Blanchard <anton@au1.ibm.com>
2631         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2632         operands to vperm for little endian.
2633         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2634         of lvsl to create the control mask for a vperm for little endian.
2636 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2637             Anton Blanchard <anton@au1.ibm.com>
2639         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2640         two operands for little-endian.
2642 2013-07-23  Steve Ellcey  <sellcey@mips.com>
2644         * config/mips/mips.c (mips_case_values_threshold): New.
2645         (TARGET_CASE_VALUES_THRESHOLD): Define.
2647 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2648             Anton Blanchard <anton@au1.ibm.com>
2650         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2651         selection of field for vector splat in little endian mode.
2653 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
2655         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2656         expanders to rs6000.md.
2657         (ior<mode>3): Likewise.
2658         (and<mode>3): Likewise.
2659         (one_cmpl<mode>2): Likewise.
2660         (nor<mode>3): Likewise.
2661         (andc<mode>3): Likewise.
2662         (eqv<mode>3): Likewise.
2663         (nand<mode>3): Likewise.
2664         (orc<mode>3): Likewise.
2666         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2667         declaration.
2669         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2670         to split multi-word logical operations.
2671         (rs6000_split_logical_di): Likewise.
2672         (rs6000_split_logical): Likewise.
2674         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2675         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2676         and allow TImode operations in 32-bit.
2677         (vsx_and<mode>3_64bit): Likewise.
2678         (vsx_ior<mode>3_32bit): Likewise.
2679         (vsx_ior<mode>3_64bit): Likewise.
2680         (vsx_xor<mode>3_32bit): Likewise.
2681         (vsx_xor<mode>3_64bit): Likewise.
2682         (vsx_one_cmpl<mode>2_32bit): Likewise.
2683         (vsx_one_cmpl<mode>2_64bit): Likewise.
2684         (vsx_nor<mode>3_32bit): Likewise.
2685         (vsx_nor<mode>3_64bit): Likewise.
2686         (vsx_andc<mode>3_32bit): Likewise.
2687         (vsx_andc<mode>3_64bit): Likewise.
2688         (vsx_eqv<mode>3_32bit): Likewise.
2689         (vsx_eqv<mode>3_64bit): Likewise.
2690         (vsx_nand<mode>3_32bit): Likewise.
2691         (vsx_nand<mode>3_64bit): Likewise.
2692         (vsx_orc<mode>3_32bit): Likewise.
2693         (vsx_orc<mode>3_64bit): Likewise.
2695         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2696         logical types in GPRs.
2698         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2699         logical insns to rs6000.md, and allow TImode operations in
2700         32-bit.
2701         (altivec_ior<mode>3): Likewise.
2702         (altivec_xor<mode>3): Likewise.
2703         (altivec_one_cmpl<mode>2): Likewise.
2704         (altivec_nor<mode>3): Likewise.
2705         (altivec_andc<mode>3): Likewise.
2707         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2708         attributes for moving the 128-bit logical operations into
2709         rs6000.md.
2710         (BOOL_REGS_OUTPUT): Likewise.
2711         (BOOL_REGS_OP1): Likewise.
2712         (BOOL_REGS_OP2): Likewise.
2713         (BOOL_REGS_UNARY): Likewise.
2714         (BOOL_REGS_AND_CR0): Likewise.
2715         (one_cmpl<mode>2): Add support for DI logical operations on
2716         32-bit, splitting the operations to 32-bit.
2717         (anddi3): Likewise.
2718         (iordi3): Likewise.
2719         (xordi3): Likewise.
2720         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2721         changes to combine the 32/64-bit code, allow logical operations on
2722         TI mode in 32-bit, and to use similar match_operator patterns like
2723         scalar mode uses.  Combine the Altivec and VSX code for logical
2724         operations, and move it here.
2725         (ior<mode>3, 128-bit types): Likewise.
2726         (xor<mode>3, 128-bit types): Likewise.
2727         (one_cmpl<mode>3, 128-bit types): Likewise.
2728         (nor<mode>3, 128-bit types): Likewise.
2729         (andc<mode>3, 128-bit types): Likewise.
2730         (eqv<mode>3, 128-bit types): Likewise.
2731         (nand<mode>3, 128-bit types): Likewise.
2732         (orc<mode>3, 128-bit types): Likewise.
2733         (and<mode>3_internal): Likewise.
2734         (bool<mode>3_internal): Likewise.
2735         (boolc<mode>3_internal1): Likewise.
2736         (boolc<mode>3_internal2): Likewise.
2737         (boolcc<mode>3_internal1): Likewise.
2738         (boolcc<mode>3_internal2): Likewise.
2739         (eqv<mode>3_internal1): Likewise.
2740         (eqv<mode>3_internal2): Likewise.
2741         (one_cmpl1<mode>3_internal): Likewise.
2743 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2745         * config/microblaze/microblaze.c (microblaze_expand_prologue):
2746         Rename flag_stack_usage to flag_stack_usage_info.
2748 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2750         * config/microblaze/sync.md: New file.
2751         * config/microblaze/microblaze.md: Include sync.md
2752         * config/microblaze/microblaze.c: Add print_operand 'y'.
2753         * config/microblaze/constraints.md: Add memory_contraint
2754         'Q' which is a single register.
2756 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
2758         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2760 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2762         * reload.c (find_reloads): Exit loop once we find this operand
2763         cannot be reloaded somehow for this alternative.
2765         * reload.c (find_reloads): Exit loop once we find a hard register.
2767         * rtlanal.c (computed_jump_p): Exit loop once we find label
2768         reference is used.
2770         * i386.c (ix86_pad_returns): Exit loop after setting replace.
2772         * cfgloopmanip.c (remove_path): Exit loop after setting
2773         irred_invalidated.
2775         * gensupport.c (subst_dup): Avoid loop if code is not
2776         MATCH_DUP nor MATCH_OP_DUP.
2778 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
2780         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2782 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2784         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2785         true for SP_REGNUM if mode == ptr_mode.
2786         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2787         with value R0_REGNUM + 31.
2789 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2791         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2792         pad pointer-typed argument downward.
2794 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2796         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2797         and __ILP32__ when the ILP32 model is in use.
2799 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2801         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2802         (aarch64_load_symref_appropriately): In the case of
2803         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2804         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2805         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2806         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2807         if 'mode' doesn't equal to 'ptr_mode'.
2808         (aarch64_output_mi_thunk): Add an assertion on the alignment of
2809         'vcall_offset'; change to call aarch64_emit_move differently depending
2810         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2811         to calculate the upper bound of 'vcall_offset'.
2812         (aarch64_cannot_force_const_mem): Change to also return true if
2813         mode != ptr_mode.
2814         (aarch64_legitimize_reload_address): In the case of large
2815         displacements, add new local variable 'xmode' and an assertion
2816         based on it; change to use 'xmode' to generate the new rtx and
2817         reload.
2818         (aarch64_asm_trampoline_template): Change to generate the template
2819         differently depending on TARGET_ILP32 or not; change to use
2820         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2821         (aarch64_trampoline_size): Removed.
2822         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2823         and replace immediate literals with it.  Change to use 'ptr_mode'
2824         instead of 'DImode' and call convert_memory_address if the mode
2825         of 'fnaddr' doesn't equal to 'ptr_mode'.
2826         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2827         to output symbol.
2828         (aarch64_elf_asm_destructor): Likewise.
2829         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2830         on TARGET_ILP32 instead of aarch64_trampoline_size.
2831         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2832         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2833         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2834         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
2835         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2836         (storewb_pair<GPI:mode>_<P:mode>): ... this.
2837         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2838         depending on the value of 'mode'.
2839         (add_losym_<mode>): New.
2840         (ldr_got_small_<mode>): New, based on ldr_got_small.
2841         (ldr_got_small): Remove.
2842         (ldr_got_small_sidi): New.
2843         * config/aarch64/iterators.md (P): New.
2844         (PTR): Change to 'ptr_mode' in the condition.
2846 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2848         * config.gcc (aarch64*-*-*): Support --with-abi.
2849         (aarch64*-*-elf): Support --with-multilib-list.
2850         (aarch64*-*-linux*): Likewise.
2851         (supported_defaults): Add abi to aarch64*-*-*.
2852         * configure.ac: Mention AArch64 for --with-multilib-list.
2853         * configure: Re-generated.
2854         * config/aarch64/biarchilp32.h: New file.
2855         * config/aarch64/biarchlp64.h: New file.
2856         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2857         (ABI_SPEC): Ditto.
2858         (MULTILIB_DEFAULTS): Ditto.
2859         (DRIVER_SELF_SPECS): Ditto.
2860         (ASM_SPEC): Update to also substitute -mabi.
2861         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2862         file whose name depends on -mabi= and -mbig-endian.
2863         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2864         TARGET_ILP32.
2865         (POINTER_SIZE): New define.
2866         (POINTERS_EXTEND_UNSIGNED): Ditto.
2867         (enum aarch64_abi_type): New enumeration tag.
2868         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2869         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2870         (TARGET_ILP32): New define.
2871         * config/aarch64/aarch64.opt (mabi): New.
2872         (aarch64_abi): New.
2873         (ilp32, lp64): New values for -mabi.
2874         * config/aarch64/t-aarch64 (comma): New define.
2875         (MULTILIB_OPTIONS): Ditto.
2876         (MULTILIB_DIRNAMES): Ditto.
2877         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2878         * doc/invoke.texi: Document -mabi for AArch64.
2880 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
2882         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
2884 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2885             Anton Blanchard <anton@au1.ibm.com>
2887         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
2888         endianness when selecting field to splat.
2890 2013-07-22  Eric Christopher  <echristo@gmail.com>
2892         * dwarf2out.c (die_odr_checksum): New function to use
2893         CHECKSUM_ macros and ULEB128 for DIE tag.
2894         (generate_type_signature): Use.
2896 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
2898         * config.gcc (sparc*-*-*): Accept leon3 processor.
2899         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
2900         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
2901         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
2902         * config/sparc/sparc.opt (enum processor_type): Add leon3.
2903         (mfix-ut699): Adjust comment.
2904         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
2905         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
2906         (CPP_CPU_SPEC): Likewise.
2907         (ASM_CPU_SPEC): Likewise.
2908         * config/sparc/sparc.c (leon3_cost): New constant.
2909         (sparc_option_override): Add leon3 support.
2910         (mem_ref): New function.
2911         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
2912         (sparc_do_work_around_errata): Look into the instruction in the delay
2913         slot and adjust accordingly.  Add fix for the data cache nullify issues
2914         of the UT699.  Change insertion position for the NOP.
2915         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
2916         (leon3_load): New reservation.
2917         (leon_store): Bump latency to 2.
2918         (grfpu): New automaton.
2919         (grfpu_alu): New unit.
2920         (grfpu_ds): Likewise.
2921         (leon_fp_alu): Adjust.
2922         (leon_fp_mult): Delete.
2923         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
2924         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
2925         * config/sparc/sparc.md (cpu): Add leon3.
2926         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
2927         (swapsi): Likewise.
2928         (atomic_test_and_set): Likewise.
2929         (ldstub): Likewise.
2931 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
2933         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
2934         default for R5900 targets.
2935         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
2936         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
2937         * config/mips/mips.c (mips_option_override): Report an error for
2938         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
2939         for -march=r5900 -mhard-float.
2941 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2943         * df-problems.c (can_move_insns_across): Exit loop once we
2944         find a non-fixed, non-global register.
2946         * ipa-pure-const.c (propagate_nothrow): Exit loop after
2947         setting can_throw.
2949         * omega.c (omega_eliminate_red): Break after setting red_found.
2950         (omega_problem_has_red_equations): Similarly after setting found.
2951         (omega_query_variable): Similarly after setting coupled.
2953 2013-07-22  Marek Polacek  <polacek@redhat.com>
2955         * gimplify.c: Don't include gimple.h twice.
2957 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2959         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
2960         instead of TARGET_THUMB1.
2961         (Pz): New constraint.
2962         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
2963         encodings.
2964         (compare_negsi_si): Likewise.
2965         (compare_addsi2_op0): Likewise.
2966         (compare_addsi2_op1): Likewise.
2967         (addsi3_carryin_<optab>): Likewise.
2968         (addsi3_carryin_alt2_<optab>): Likewise.
2969         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
2970         for arm_restrict_it.
2971         (subsi3_carryin): Likewise.
2972         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
2973         (minmax_arithsi): Disable for arm_restrict_it.
2974         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
2975         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
2976         (satsi_<SAT:code>_shift): Likewise.
2977         (arm_shiftsi3): Add alternative for 16-bit encoding.
2978         (arm32_movhf): Disable for arm_restrict_it.
2979         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
2980         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
2982 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
2984         * config/arm/arm.md (attribute "insn"): Delete.
2985         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
2986         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
2987         (not_shiftsi): Update for attribute change.
2988         (not_shiftsi_compare0): Likewise.
2989         (not_shiftsi_compare0_scratch): Likewise.
2990         (arm_one_cmplsi2): Likewise.
2991         (thumb1_one_cmplsi2): Likewise.
2992         (notsi_compare0): Likewise.
2993         (notsi_compare0_scratch): Likewise.
2994         (thumb1_movdi_insn): Likewise.
2995         (arm_movsi_insn): Likewise.
2996         (movhi_insn_arch4): Likewise.
2997         (movhi_bytes): Likewise.
2998         (arm_movqi_insn): Likewise.
2999         (thumb1_movqi_insn): Likewise.
3000         (arm32_movhf): Likewise.
3001         (thumb1_movhf): Likewise.
3002         (arm_movsf_soft_insn): Likewise.
3003         (thumb1_movsf_insn): Likewise.
3004         (thumb_movdf_insn): Likewise.
3005         (movsicc_insn): Likewise.
3006         (movsfcc_soft_insn): Likewise.
3007         (and_scc): Likewise.
3008         (cond_move): Likewise.
3009         (if_move_not): Likewise.
3010         (if_not_move): Likewise.
3011         (if_shift_move): Likewise.
3012         (if_move_shift): Likewise.
3013         (if_shift_shift): Likewise.
3014         (if_not_arith): Likewise.
3015         (if_arith_not): Likewise.
3016         (cond_move_not): Likewise.
3017         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3018         (neon_mov<mode>): Likewise.
3019         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3020         (thumb2_movsi_vfp): Likewise.
3021         (movsf_vfp): Likewise.
3022         (thumb2_movsf_vfp): Likewise.
3023         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3024         change.
3025         (cortexa7_older_only): Likewise.
3026         (cortexa7_younger): Likewise.
3027         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3028         (1020alu_shift_op): Likewise.
3029         (1020alu_shift_reg_op): Likewise.
3030         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3031         (alu_shift_op): Likewise.
3032         (alu_shift_reg_op): Likewise.
3033         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3034         (11_alu_shift_op): Likewise.
3035         (11_alu_shift_reg_op): Likewise.
3036         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3037         (9_alu_shift_reg_op): Likewise.
3038         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3039         change.
3040         (cortex_a15_alu_shift): Likewise.
3041         (cortex_a15_alu_shift_reg): Likewise.
3042         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3043         (cortex_a5_alu_shift): Likewise.
3044         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3045         change.
3046         (cortex_a53_alu_shift): Likewise.
3047         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3048         change.
3049         (cortex_a7_alu_reg): Likewise.
3050         (cortex_a7_alu_shift): Likewise.
3051         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3052         (cortex_a8_alu_shift): Likewise.
3053         (cortex_a8_alu_shift_reg): Likewise.
3054         (cortex_a8_mov): Likewise.
3055         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3056         (cortex_a9_dp_shift): Likewise.
3057         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3058         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3059         (cortex_r4_mov): Likewise.
3060         (cortex_r4_alu_shift): Likewise.
3061         (cortex_r4_alu_shift_reg): Likewise.
3062         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3063         (526_alu_shift_op): Likewise.
3064         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3065         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3066         (626te_alu_shift_op): Likewise.
3067         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3068         (726te_alu_op): Likewise.
3069         (726te_alu_shift_op): Likewise.
3070         (726te_alu_shift_reg_op): Likewise.
3071         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3072         (mp626_alu_shift_op): Likewise.
3073         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3074         (pj4_alu_e1_conds): Likewise.
3075         (pj4_alu): Likewise.
3076         (pj4_alu_conds): Likewise.
3077         (pj4_shift): Likewise.
3078         (pj4_shift_conds): Likewise.
3079         (pj4_alu_shift): Likewise.
3080         (pj4_alu_shift_conds): Likewise.
3082 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3084         * config/arm/predicates.md (shiftable_operator_strict_it):
3085         New predicate.
3086         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3087         Disable cond_exec version for arm_restrict_it.
3088         (thumb2_smaxsi3): Convert to generate cond_exec.
3089         (thumb2_sminsi3): Likewise.
3090         (thumb32_umaxsi3): Likewise.
3091         (thumb2_uminsi3): Likewise.
3092         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3093         (thumb2_neg_abssi2): Likewise.
3094         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3095         (thumb2_movsicc_insn): Adjust alternatives.
3096         (thumb2_mov_negscc): Disable for arm_restrict_it.
3097         (thumb2_mov_negscc_strict_it): New pattern.
3098         (thumb2_mov_notscc_strict_it): New pattern.
3099         (thumb2_mov_notscc): Disable for arm_restrict_it.
3100         (thumb2_ior_scc): Likewise.
3101         (thumb2_ior_scc_strict_it): New pattern.
3102         (thumb2_cond_move): Adjust for arm_restrict_it.
3103         (thumb2_cond_arith): Disable for arm_restrict_it.
3104         (thumb2_cond_arith_strict_it): New pattern.
3105         (thumb2_cond_sub): Adjust for arm_restrict_it.
3106         (thumb2_movcond): Likewise.
3107         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3108         (thumb2_zero_extendhisi2_v6): Likewise.
3109         (thumb2_zero_extendqisi2_v6): Likewise.
3110         (orsi_notsi_si): Likewise.
3111         (orsi_not_shiftsi_si): Likewise.
3113 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3115         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3116         instruction sequence is 1 byte shorter.
3118 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3120         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3121         it is not needed after split.
3123 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3125         PR target/51784
3126         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3127         second label for nonlocal goto receivers. Don't output pic base labels
3128         unless we're producing PIC; mark that action unreachable().
3129         (ix86_save_reg): If the function contains a nonlocal label, save the
3130         PIC base reg.
3131         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3132         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3133         (update_pic_label_number_if_needed): New.
3134         (machopic_output_function_base_name): Adjust for nonlocal receiver
3135         case.
3136         (machopic_should_output_picbase_label): New.
3137         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3138         (nonlocal_goto_receiver): New insn and split.
3140 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3142         * config/aarch64/aarch64-builtins.c
3143         (aarch64_fold_builtin): Fold abs in all modes.
3144         * config/aarch64/aarch64-simd-builtins.def
3145         (abs): Enable for all modes.
3146         * config/aarch64/arm_neon.h
3147         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3148         (vabs_f64): Add missing intrinsic.
3150 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3152         * config/aarch64/arm_neon.h (vabs_s64): New function
3154 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3156         PR target/57516
3157         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3158         * config/avr/avr.md (adjust_len): Add `round'.
3159         * config/avr/avr-protos.h (avr_out_round): New prototype.
3160         (avr_out_plus): Add `out_label' argument.
3161         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3162         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3163         Handle the case where `insn' is just a pattern.
3164         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3165         (avr_out_round): New function.
3166         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3168 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3170         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3171         Add check for flag_stack_usage to handle -fstack-usage support
3173 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3175         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3176         interaction for new Power8 flags and VSX.
3178 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3180         PR middle-end/57698
3181         * tree-inline.c (expand_call_inline): Emit errors during
3182         early_inlining only if optimization is not turned on.
3184 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3186         * passes.def: New.
3188         * passes.c (init_optimization_passes): Move the construction of
3189         the pass hierarchy into a new passes.def file.
3191         * Makefile.in (passes.o): Add dependency on passes.def.
3193 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3195         * passes.c (init_optimization_passes): Introduce macros for
3196         constructing the tree of passes (INSERT_PASSES_AFTER,
3197         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3198         TERMINATE_PASS_LIST).
3200 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3201             Wei Mi  <wmi@google.com>
3203         PR rtl-optimization/57878
3204         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3205         top.
3206         (reload_pseudo_compare_func): Check nregs first for reload
3207         pseudos.
3209 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3211         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3213 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3215         * read-rtl.c (validate_const_int): Once an invalid character is
3216         seen, quit the loop.
3218         * gengtype.c (write_roots): Similarly once we find the "deletable"
3219         or "if_marked" option.
3221 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3223         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3224         "xtab" and "sat".  Move value "clz" from here to ...
3225         (attriubte "type"): ... here.
3226         (satsi_<SAT:code>): Delete "insn" attribute.
3227         (satsi_<SAT:code>_shift): Likewise.
3228         (arm_zero_extendqisi2addsi): Likewise.
3229         (arm_extendqisi2addsi): Likewise.
3230         (clzsi2): Update for attribute changes.
3231         (rbitsi2): Likewise.
3232         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3233         attribute.
3234         (arm_usatsihi): Likewise.
3235         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3237 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3239         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3240         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3241         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3242         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3243         in alphabetical order.
3244         (attribute "core_cycles"): Update for attribute changes.
3245         (arm_addsi3): Likewise.
3246         (addsi3_compare0): Likewise.
3247         (addsi3_compare0_scratch): Likewise.
3248         (addsi3_compare_op1): Likewise.
3249         (addsi3_compare_op2): Likewise.
3250         (compare_addsi2_op0): Likewise.
3251         (compare_addsi2_op1): Likewise.
3252         (addsi3_carryin_shift_<optab>): Likewise.
3253         (subsi3_carryin_shift): Likewise.
3254         (rsbsi3_carryin_shift): Likewise.
3255         (arm_subsi3_insn): Likewise.
3256         (subsi3_compare0): Likewise.
3257         (subsi3_compare): Likewise.
3258         (arm_andsi3_insn): Likewise.
3259         (thumb1_andsi3_insn): Likewise.
3260         (andsi3_compare0): Likewise.
3261         (andsi3_compare0_scratch): Likewise.
3262         (zeroextractsi_compare0_scratch
3263         (andsi_not_shiftsi_si): Likewise.
3264         (iorsi3_insn): Likewise.
3265         (iorsi3_compare0): Likewise.
3266         (iorsi3_compare0_scratch): Likewise.
3267         (arm_xorsi3): Likewise.
3268         (thumb1_xorsi3_insn): Likewise.
3269         (xorsi3_compare0): Likewise.
3270         (xorsi3_compare0_scratch): Likewise.
3271         (satsi_<SAT:code>_shift): Likewise.
3272         (rrx): Likewise.
3273         (arm_shiftsi3): Likewise.
3274         (shiftsi3_compare0): Likewise.
3275         (not_shiftsi): Likewise.
3276         (not_shiftsi_compare0): Likewise.
3277         (not_shiftsi_compare0_scratch): Likewise.
3278         (arm_one_cmplsi2): Likewise.
3279         (thumb_one_complsi2): Likewise.
3280         (notsi_compare0): Likewise.
3281         (notsi_compare0_scratch): Likewise.
3282         (thumb1_zero_extendhisi2): Likewise.
3283         (arm_zero_extendhisi2): Likewise.
3284         (arm_zero_extendhisi2_v6): Likewise.
3285         (arm_zero_extendhisi2addsi): Likewise.
3286         (thumb1_zero_extendqisi2): Likewise.
3287         (thumb1_zero_extendqisi2_v6): Likewise.
3288         (arm_zero_extendqisi2): Likewise.
3289         (arm_zero_extendqisi2_v6): Likewise.
3290         (arm_zero_extendqisi2addsi): Likewise.
3291         (thumb1_extendhisi2): Likewise.
3292         (arm_extendhisi2): Likewise.
3293         (arm_extendhisi2_v6): Likewise.
3294         (arm_extendqisi): Likewise.
3295         (arm_extendqisi_v6): Likewise.
3296         (arm_extendqisi2addsi): Likewise.
3297         (thumb1_extendqisi2): Likewise.
3298         (thumb1_movdi_insn): Likewise.
3299         (arm_movsi_insn): Likewise.
3300         (movsi_compare0): Likewise.
3301         (movhi_insn_arch4): Likewise.
3302         (movhi_bytes): Likewise.
3303         (arm_movqi_insn): Likewise.
3304         (thumb1_movqi_insn): Likewise.
3305         (arm32_movhf): Likewise.
3306         (thumb1_movhf): Likewise.
3307         (arm_movsf_soft_insn): Likewise.
3308         (thumb1_movsf_insn): Likewise.
3309         (movdf_soft_insn): Likewise.
3310         (thumb_movdf_insn): Likewise.
3311         (arm_cmpsi_insn): Likewise.
3312         (cmpsi_shiftsi): Likewise.
3313         (cmpsi_shiftsi_swp): Likewise.
3314         (arm_cmpsi_negshiftsi_si): Likewise.
3315         (movsicc_insn): Likewise.
3316         (movsfcc_soft_insn): Likewise.
3317         (arith_shiftsi): Likewise.
3318         (arith_shiftsi_compare0
3319         (arith_shiftsi_compare0_scratch
3320         (sub_shiftsi): Likewise.
3321         (sub_shiftsi_compare0
3322         (sub_shiftsi_compare0_scratch
3323         (and_scc): Likewise.
3324         (cond_move): Likewise.
3325         (if_plus_move): Likewise.
3326         (if_move_plus): Likewise.
3327         (if_move_not): Likewise.
3328         (if_not_move): Likewise.
3329         (if_shift_move): Likewise.
3330         (if_move_shift): Likewise.
3331         (if_shift_shift): Likewise.
3332         (if_not_arith): Likewise.
3333         (if_arith_not): Likewise.
3334         (cond_move_not): Likewise.
3335         (thumb1_ashlsi3): Set type attribute.
3336         (thumb1_ashrsi3): Likewise.
3337         (thumb1_lshrsi3): Likewise.
3338         (thumb1_rotrsi3): Likewise.
3339         (shiftsi3_compare0_scratch): Likewise.
3340         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3341         (neon_mov<mode>): Likewise.
3342         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3343         attribute changes.
3344         (thumb2_movsi_insn): Likewise.
3345         (thumb2_cmpsi_neg_shiftsi): Likewise.
3346         (thumb2_extendqisi_v6): Likewise.
3347         (thumb2_zero_extendhisi2_v6): Likewise.
3348         (thumb2_zero_extendqisi2_v6): Likewise.
3349         (thumb2_shiftsi3_short): Likewise.
3350         (thumb2_addsi3_compare0_scratch): Likewise.
3351         (orsi_not_shiftsi_si): Likewise.
3352         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3353         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3354         changes.
3355         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3356         (1020alu_shift_op): Likewise.
3357         (1020alu_shift_reg_op): Likewise.
3358         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3359         (alu_shift_op): Likewise.
3360         (alu_shift_reg_op): Likewise.
3361         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3362         (11_alu_shift_op): Likewise.
3363         (11_alu_shift_reg_op): Likewise.
3364         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3365         (9_alu_shift_reg_op): Likewise.
3366         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3367         changes.
3368         (cortex_a15_alu_shift): Likewise.
3369         (cortex_a15_alu_shift_reg): Likewise.
3370         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3371         changes.
3372         (cortex_a5_alu_shift): Likewise.
3373         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3374         changes.
3375         (cortex_a53_alu_shift): Likewise.
3376         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3377         changes.
3378         (cortex_a7_alu_reg): Likewise.
3379         (cortex_a7_alu_shift): Likewise.
3380         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3381         changes.
3382         (cortex_a8_alu_shift): Likewise.
3383         (cortex_a8_alu_shift_reg): Likewise.
3384         (cortex_a8_mov): Likewise.
3385         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3386         (cortex_a9_dp_shift): Likewise.
3387         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3388         changes.
3389         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3390         changes.
3391         (cortex_r4_mov): Likewise.
3392         (cortex_r4_alu_shift): Likewise.
3393         (cortex_r4_alu_shift_reg): Likewise.
3394         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3395         (526_alu_shift_op): Likewise.
3396         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3397         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3398         (626te_alu_shift_op): Likewise.
3399         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3400         (726te_alu_op): Likewise.
3401         (726te_alu_shift_op): Likewise.
3402         (726te_alu_shift_reg_op): Likewise.
3403         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3404         (mp626_alu_shift_op): Likewise.
3405         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3406         (pj4_alu_e1_conds): Likewise.
3407         (pj4_alu): Likewise.
3408         (pj4_alu_conds): Likewise.
3409         (pj4_shift): Likewise.
3410         (pj4_shift_conds): Likewise.
3411         (pj4_alu_shift): Likewise.
3412         (pj4_alu_shift_conds): Likewise.
3413         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3414         changes.
3415         (cortexa7_older_only): Likewise.
3416         (cortexa7_younger): Likewise.
3418 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3420         * ipa-pure-const.c (generate_summary): Rename to...
3421         (pure_const_generate_summary): ... this.
3423 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3425         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3427 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3429         PR target/57909
3430         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3431         usage in HI mode.
3433 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3435         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3436         enabled without -march=zEC12.
3437         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3438         flags to be set.
3440 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3442         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3443         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3444         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3445         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3446         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3447         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3448         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3449         ISA_HAS_FP4.
3450         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3451         and ISA_HAS_NMADD3_NMSUB3.
3452         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3453         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3454         (nmsub4<mode>, nmsub3<mode>): Likewise.
3455         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3457 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3459         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3460         TARGET_MIPS5400 checking.
3462 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3463             Peter Bergner  <bergner@vnet.ibm.com>
3465         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3466         registers in the comment.
3467         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3468         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3469         rather than FIRST_PSEUDO_REGISTERS.
3471 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3473         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3474         enable extra ISA flags with TARGET_HTM.
3476 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3478         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3479         Fix comment typos.
3481 2013-07-15  Cong Hou  <congh@google.com>
3483         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3484         in compare function for sorting.
3486 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3488         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3489         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3490         * config/rs6000/rs6000.opt: Add -mhtm option.
3491         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3492         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3493         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3494         __HTM__ if the HTM instructions are available.
3495         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3496         htm_spr_reg_operand): New define_predicates.
3497         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3498         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3499         Include htm.md.
3500         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3501         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3502         HTM builtin functions.
3503         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3504         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3505         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3506         (rs6000_builtin_mask_calculate): Likewise.
3507         (rs6000_option_override_internal): Likewise.
3508         (bdesc_htm): Add new HTM builtin support.
3509         (htm_spr_num): New function.
3510         (htm_spr_regno): Likewise.
3511         (rs6000_htm_spr_icode): Likewise.
3512         (htm_expand_builtin): Likewise.
3513         (htm_init_builtins): Likewise.
3514         (rs6000_expand_builtin): Add support for HTM builtin functions.
3515         (rs6000_init_builtins): Likewise.
3516         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3517         option.
3518         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3519         (TARGET_HTM, MASK_HTM): Define macros.
3520         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3521         (FIXED_REGISTERS): Likewise.
3522         (CALL_USED_REGISTERS): Likewise.
3523         (CALL_REALLY_USED_REGISTERS): Likewise.
3524         (REG_ALLOC_ORDER): Likewise.
3525         (enum reg_class): Likewise.
3526         (REG_CLASS_NAMES): Likewise.
3527         (REG_CLASS_CONTENTS): Likewise.
3528         (REGISTER_NAMES): Likewise.
3529         (ADDITIONAL_REGISTER_NAMES): Likewise.
3530         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3531         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3532         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3533         * config/rs6000/htm.md: New file.
3534         * config/rs6000/htmintrin.h: New file.
3535         * config/rs6000/htmxlintrin.h: New file.
3537 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3539         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3540         Define SYMBOL_TINY_GOT, update comment.
3541         * config/aarch64/aarch64.c
3542         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3543         (aarch64_expand_mov_immediate): Likewise.
3544         (aarch64_print_operand): Likewise.
3545         (aarch64_classify_symbol): Likewise.
3546         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3547         (ldr_got_tiny): Define.
3549 2013-07-13  Tobias Grosser  <tobias@grosser.es>
3551         PR tree-optimization/54094
3552         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3553           scheduling dimension for the parallelism check from the polyhedral
3554           information in the AST.
3555         * graphite-dependences.c (carries_deps): Do not assume the schedule is
3556           in 2D + 1 form.
3558 2013-07-13  Jason Merrill  <jason@redhat.com>
3560         * print-tree.c (debug_vec_tree): Use debug_raw.
3561         (debug_raw (vec<tree, va_gc> &)): New.
3562         (debug_raw (vec<tree, va_gc> *)): New.
3563         * tree.h: Declare them.
3565 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
3567         * ifcvt.c (ifcvt_after_combine): New static variable.
3568         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3569         for size.
3570         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
3571         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3572         rest_of_handle_if_after_reload): Pass new argument for if_convert.
3574 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
3576         * config/mips/mips.c (mips_expand_call): Remove empty statement.
3578 2013-07-12  Michael Matz  <matz@suse.de>
3580         PR middle-end/55771
3581         * convert.c (convert_to_real): Reject non-float inner types.
3583 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
3585         * config/aarch64/aarch64-protos.h
3586         (aarch64_simd_immediate_valid_for_move): Remove.
3587         * config/aarch64/aarch64.c (simd_immediate_info): New member.
3588         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3589         cases.
3590         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3592 2013-07-11  Steve Ellcey  <sellcey@mips.com>
3594         * config/mips/mips.c (mips_conditional_register_usage): Do not
3595         use t[0-7] registers in MIPS16 mode when optimizing for size.
3597 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
3599         * config/i386/i386.c (dispatch_function_versions): Fix array
3600         indexing of function_version_info to match actual_versions.
3602 2013-07-11  Teresa Johnson  <tejohnson@google.com>
3604         * vec.h (struct va_gc): Move release out-of-line.
3605         (va_gc::release): Call ggc_free on released vec.
3607 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3609         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3610         Require GOT register as additional operand in UNSPEC.
3611         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3612         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3613         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3614         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3615         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3616         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3617         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3619 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
3621         PR target/57631
3622         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3623         name seen by assembler/linker rather if available.
3625 2013-07-11  Andreas Schwab  <schwab@suse.de>
3627         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3629 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
3631         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
3633 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
3635         * doc/tm.texi.in: Move hook documentation to ....
3636         * target.def: ... here.
3638         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3639         text on @hook line.
3640         * doc/tm.texi: Regenerate.
3642 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
3644         PR c++/57869
3645         * doc/invoke.texi: Document -Wconditionally-supported.
3647 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3649         PR target/57844
3650         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3651         of my_fp.
3653 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3655         PR target/57506
3656         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3657         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3658         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3659         Remove duplicate devices.
3660         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3661         * config/avr/t-multilib: Regenerate.
3662         * config/avr/avr-tables.opt: Regenerate.
3663         * doc/avr-mmcu.texi: Regenerate.
3665 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3667         PR target/56987
3668         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3670 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
3672         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3673         the cost of MULT when optimizing for size.
3675 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3677         * config/cr16/cr16-protos.h: Don't include target.h.
3679 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
3681         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3682         adjust register size for TDmode and TFmode for VSX registers.
3684 2013-07-08  Kai Tietz  <ktietz@redhat.com>
3686         PR target/56892
3687         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3688         hook_bool_const_tree_true.
3690 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3692         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3693         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3694         * config/s390/s390.md: Define FPR*_REGNUM constants.
3695         Fix FPR2_REGNUM constant (18 -> 17).
3696         ("*trunc<BFP:mode><DFP_ALL:mode>2")
3697         ("*trunc<DFP_ALL:mode><BFP:mode>2")
3698         ("trunc<BFP:mode><DFP_ALL:mode>2")
3699         ("trunc<DFP_ALL:mode><BFP:mode>2")
3700         ("*extend<BFP:mode><DFP_ALL:mode>2")
3701         ("*extend<DFP_ALL:mode><BFP:mode>2")
3702         ("extend<BFP:mode><DFP_ALL:mode>2")
3703         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3704         FPR4_REGNUM.
3706 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
3708         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
3710 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3712         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3713         and cfun_fpr_bit_p to cfun_fpr_save_p.
3714         (s390_frame_area, s390_register_info, s390_frame_info)
3715         (s390_emit_prologue, s390_emit_epilogue)
3716         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3717         register numbers.
3718         * config/s390/s390.h: Define *_REGNUM macros for floating point
3719         register numbers.
3721 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
3723         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3725 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
3727         PR rtl-optimization/57786
3728         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3729         and break out of the loop when it is set to false.
3731 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
3733         PR target/57819
3734         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3735         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3736         (const_int 63)) 0)).
3737         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3738         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3739         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3741         PR rtl-optimization/57829
3742         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3743         mask bits outside of mode are just sign-extension from mode to HWI.
3745 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
3747         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3748         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3749         adjust_address instead of change_address to keep info about alignment.
3750         (emit_strmov): Remove.
3751         (emit_memmov): New function.
3752         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3753         (expand_movmem_epilogue): Likewise and return updated rtx for
3754         destination.
3755         (expand_constant_movmem_prologue): Likewise and return updated rtx for
3756         destination and source.
3757         (decide_alignment): Refactor, handle vector_loop.
3758         (ix86_expand_movmem): Likewise.
3759         (ix86_expand_setmem): Likewise.
3760         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3762 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
3764         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3765         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3767 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3769         * config/i386/sse.md (sse_movlhps): Change alternative 3
3770         of operand 2 to "m".
3772 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3774         PR target/57807
3775         * config/i386/sse.md (iptr): New mode attribute.
3776         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3777         (<sse>_vm<plusminus_insn><mode>3): Ditto.
3778         (<sse>_vmmul<mode>3): Ditto.
3779         (<sse>_vmdiv<mode>3): Ditto.
3780         (sse_vmrcpv4sf2): Ditto.
3781         (<sse>_vmsqrt<mode>2): Ditto.
3782         (sse_vmrsqrtv4sf2): Ditto.
3783         (<sse>_vm<code><mode>3): Ditto.
3784         (avx_vmcmp<mode>3): Ditto.
3785         (<sse>_vmmaskcmp<mode>3): Ditto.
3786         (<sse>_comi): Ditto.
3787         (<sse>_ucomi): Ditto.
3788         (*xop_vmfrcz_<mode>): Ditto.
3789         (*fmai_fmadd_<mode>): Ditto.
3790         (*fmai_fmsub_<mode>): Ditto.
3791         (*fmai_fnmadd_<mode>): Ditto.
3792         (*fmai_fnmsub_<mode>): Ditto.
3793         (*fma4i_vmfmadd_<mode>): Ditto.
3794         (*fma4i_vmfmsub_<mode>): Ditto.
3795         (*fma4i_vmfnmadd_<mode>): Ditto.
3796         (*fma4i_vmfnmsub_<mode>): Ditto.
3797         (*xop_vmfrcz_<mode>): Ditto.
3798         (sse_cvtps2pi): Ditto.
3799         (sse_cvttps2pi): Ditto.
3800         (sse_cvtss2si): Ditto.
3801         (sse_cvtss2si_2): Ditto.
3802         (sse_cvtss2siq_2): Ditto.
3803         (sse_cvttss2si): Ditto.
3804         (sse_cvttss2siq): Ditto.
3805         (sse_cvtsd2si): Ditto.
3806         (sse_cvtsd2si_2): Ditto.
3807         (sse_cvtsd2siq_2): Ditto.
3808         (sse_cvttsd2si): Ditto.
3809         (sse_cvttsd2siq): Ditto.
3810         (sse_cvtsd2ss): Ditto.
3811         (sse_cvtss2sd): Ditto.
3812         (avx2_pbroadcast<mode>): Ditto.
3813         (avx2_pbroadcast<mode>_1): Ditto.
3814         (*avx_vperm_broadcast_v4sf): Ditto.
3816         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3817         (sse_movlhps): Ditto.
3818         (sse_storehps): Ditto.
3819         (sse_loadhps): Ditto.
3820         (sse_storelps): Ditto.
3821         (sse_loadlps): Ditto.
3822         (*vec_concatv4sf): Ditto.
3823         (*vec_interleave_highv2df): Ditto.
3824         (*vec_interleave_lowv2df): Ditto.
3825         (*vec_extractv2df_1_sse): Ditto.
3826         (*vec_extractv2df_0_sse): Ditto.
3827         (sse2_storelpd): Ditto.
3828         (sse2_loadlpd): Ditto.
3829         (sse2_movsd): Ditto.
3830         (*vec_concatv4si): Ditto.
3831         (vec_concatv2di): Ditto.
3833         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3834         for Intel asm dialect.
3835         (mmx_punpcklwd): Ditto.
3836         (mmx_punpckldq): Ditto.
3838         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3839         for intel assembler dialect.
3841 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
3843         PR target/29776
3844         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3845         for BUILT_IN_C{LZ,LRSB}*.
3846         * tree.h (CASE_INT_FN): Add FN##IMAX case.
3847         * tree-vrp.c (extract_range_basic): Handle
3848         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
3849         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3850         fall thru to code calling set_value*.
3851         * builtins.c (expand_builtin): Remove *IMAX cases.
3852         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3853         if width is bigger than 2*HWI.
3855 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
3857         PR rtl-optimization/55342
3858         * lra-int.h (lra_subreg_reload_pseudos): New.
3859         * lra.c: Add undoing optional reloads to the block diagram.
3860         (lra_subreg_reload_pseudos): New.
3861         (lra_optional_reload_pseudos): Change comments.
3862         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
3863         lra_optional_reload_pseudos after undo transformations.
3864         * lra-assigns.c (pseudo_prefix_title): New.
3865         (lra_setup_reg_renumber): Use it.
3866         (spill_for): Ditto.  Check subreg reload pseudos too.
3867         (assign_by_spills): Consider subreg reload pseudos too.
3868         * lra-constraints.c (simplify_operand_subreg): Use
3869         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3870         (curr_insn_transform): Recognize and do optional reloads.
3871         (undo_optional_reloads): New.
3872         (lra_undo_inheritance): Call undo_optional_reloads.
3874 2013-07-05  Thomas Quinot  <quinot@adacore.com>
3876         * tree-complex.c (expand_complex_operations_1): Fix typo.
3878 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
3880         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3881         (tune_params): New member 'const vec_costs'.
3882         * config/aarch64/aarch64.c (generic_vector_cost): New.
3883         (generic_tunings): New member 'generic_vector_cost'.
3884         (aarch64_builtin_vectorization_cost): New.
3885         (aarch64_add_stmt_cost): New.
3886         (TARGET_VECTORIZE_ADD_STMT_COST): New.
3887         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
3889 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
3891         PR target/57777
3892         * config/i386/predicates.md (vsib_address_operand): Disallow
3893         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
3895 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
3897         PR middle-end/55030
3898         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
3899         expand_builtin_setjmp_receiver.
3900         (expand_label): Adjust, call expand_builtin_setjmp_receiver
3901         with NULL for the label parameter.
3902         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
3903         the frame-pointer.  Adjust comments.
3904         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
3905         only if LABEL is non-NULL.
3907 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
3909         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
3910         (ARM_ABI_AAPCS64): Ditto.
3911         (arm_abi): Ditto.
3912         (ARM_DEFAULT_ABI): Ditto.
3914 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
3916         * config/aarch64/aarch64-builtins.c
3917         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
3918         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
3919         (st1): Likewise.
3920         * config/aarch64/aarch64-simd.md
3921         (aarch64_ld1<VALL:mode>): New.
3922         (aarch64_st1<VALL:mode>): Likewise.
3923         * config/aarch64/arm_neon.h
3924         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
3926 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
3928         * config/i386/i386.c (gate_insert_vzeroupper): Check if
3929         target ISA is AVX.
3930         (ix86_option_override_internal):Turn on all -mavx target flags by
3931         default as they are dependent on AVX anyway.
3933 2013-07-02  Cary Coutant  <ccoutant@google.com>
3935         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
3936         deterministic hash.
3937         (loc_checksum_ordered): Likewise.
3938         (hash_loc_operands): Remove inline keyword.
3940 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
3942         PR tree-optimization/57741
3943         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
3944         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
3945         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
3946         Allow REAL_CST step_exprs if flag_associative_math.
3947         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
3949 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3951         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
3953 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3955         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
3957 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3959         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
3960         encoding.
3961         (iorsi3_insn): Likewise.
3962         (arm_xorsi3): Likewise.
3964 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
3966         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
3967         to ...
3968         (attribute "type"): ... here, and prefix with "wmmx_".
3969         (attribute "core_cycles"): Update for attribute changes.
3970         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
3971         (tbcstv4hi): Likewise.
3972         (tbcstv2si): Likewise.
3973         (iwmmxt_iordi3): Likewise.
3974         (iwmmxt_xordi3): Likewise.
3975         (iwmmxt_anddi3): Likewise.
3976         (iwmmxt_nanddi3): Likewise.
3977         (iwmmxt_arm_movdi): Likewise.
3978         (iwmmxt_movsi_insn): Likewise.
3979         (mov<mode>_internal): Likewise.
3980         (and<mode>3_iwmmxt): Likewise.
3981         (ior<mode>3_iwmmxt): Likewise.
3982         (xor<mode>3_iwmmxt): Likewise.
3983         (add<mode>3_iwmmxt): Likewise.
3984         (ssaddv8qi3): Likewise.
3985         (ssaddv4hi3): Likewise.
3986         (ssaddv2si3): Likewise.
3987         (usaddv8qi3): Likewise.
3988         (usaddv4hi3): Likewise.
3989         (usaddv2si3): Likewise.
3990         (sub<mode>3_iwmmxt): Likewise.
3991         (sssubv8qi3): Likewise.
3992         (sssubv4hi3): Likewise.
3993         (sssubv2si3): Likewise.
3994         (ussubv8qi3): Likewise.
3995         (ussubv4hi3): Likewise.
3996         (ussubv2si3): Likewise.
3997         (mulv4hi3_iwmmxt): Likewise.
3998         (smulv4hi3_highpart): Likewise.
3999         (umulv4hi3_highpart): Likewise.
4000         (iwmmxt_wmacs): Likewise.
4001         (iwmmxt_wmacsz): Likewise.
4002         (iwmmxt_wmacu): Likewise.
4003         (iwmmxt_wmacuz): Likewise.
4004         (iwmmxt_clrdi): Likewise.
4005         (iwmmxt_clrv8qi): Likewise.
4006         (iwmmxt_clr4hi): Likewise.
4007         (iwmmxt_clr2si): Likewise.
4008         (iwmmxt_uavgrndv8qi3): Likewise.
4009         (iwmmxt_uavgrndv4hi3): Likewise.
4010         (iwmmxt_uavgv8qi3): Likewise.
4011         (iwmmxt_uavgv4hi3): Likewise.
4012         (iwmmxt_tinsrb): Likewise.
4013         (iwmmxt_tinsrh): Likewise.
4014         (iwmmxt_tinsrw): Likewise.
4015         (iwmmxt_textrmub): Likewise.
4016         (iwmmxt_textrmsb): Likewise.
4017         (iwmmxt_textrmuh): Likewise.
4018         (iwmmxt_textrmsh): Likewise.
4019         (iwmmxt_textrmw): Likewise.
4020         (iwmxxt_wshufh): Likewise.
4021         (eqv8qi3): Likewise.
4022         (eqv4hi3): Likewise.
4023         (eqv2si3): Likewise.
4024         (gtuv8qi3): Likewise.
4025         (gtuv4hi3): Likewise.
4026         (gtuv2si3): Likewise.
4027         (gtv8qi3): Likewise.
4028         (gtv4hi3): Likewise.
4029         (gtv2si3): Likewise.
4030         (smax<mode>3_iwmmxt): Likewise.
4031         (umax<mode>3_iwmmxt): Likewise.
4032         (smin<mode>3_iwmmxt): Likewise.
4033         (umin<mode>3_iwmmxt): Likewise.
4034         (iwmmxt_wpackhss): Likewise.
4035         (iwmmxt_wpackwss): Likewise.
4036         (iwmmxt_wpackdss): Likewise.
4037         (iwmmxt_wpackhus): Likewise.
4038         (iwmmxt_wpackwus): Likewise.
4039         (iwmmxt_wpackdus): Likewise.
4040         (iwmmxt_wunpckihb): Likewise.
4041         (iwmmxt_wunpckihh): Likewise.
4042         (iwmmxt_wunpckihw): Likewise.
4043         (iwmmxt_wunpckilb): Likewise.
4044         (iwmmxt_wunpckilh): Likewise.
4045         (iwmmxt_wunpckilw): Likewise.
4046         (iwmmxt_wunpckehub): Likewise.
4047         (iwmmxt_wunpckehuh): Likewise.
4048         (iwmmxt_wunpckehuw): Likewise.
4049         (iwmmxt_wunpckehsb): Likewise.
4050         (iwmmxt_wunpckehsh): Likewise.
4051         (iwmmxt_wunpckehsw): Likewise.
4052         (iwmmxt_wunpckelub): Likewise.
4053         (iwmmxt_wunpckeluh): Likewise.
4054         (iwmmxt_wunpckeluw): Likewise.
4055         (iwmmxt_wunpckelsb): Likewise.
4056         (iwmmxt_wunpckelsh): Likewise.
4057         (iwmmxt_wunpckelsw): Likewise.
4058         (ror<mode>3): Likewise.
4059         (ashr<mode>3_iwmmxt): Likewise.
4060         (lshr<mode>3_iwmmxt): Likewise.
4061         (ashl<mode>3_iwmmxt): Likewise.
4062         (ror<mode>3_di): Likewise.
4063         (ashr<mode>3_di): Likewise.
4064         (lshr<mode>3_di): Likewise.
4065         (ashl<mode>3_di): Likewise.
4066         (iwmmxt_wmadds): Likewise.
4067         (iwmmxt_wmaddu): Likewise.
4068         (iwmmxt_tmia): Likewise.
4069         (iwmmxt_tmiaph): Likewise.
4070         (iwmmxt_tmiabb): Likewise.
4071         (iwmmxt_tmiatb): Likewise.
4072         (iwmmxt_tmiabt): Likewise.
4073         (iwmmxt_tmiatt): Likewise.
4074         (iwmmxt_tmovmskb): Likewise.
4075         (iwmmxt_tmovmskh): Likewise.
4076         (iwmmxt_tmovmskw): Likewise.
4077         (iwmmxt_waccb): Likewise.
4078         (iwmmxt_wacch): Likewise.
4079         (iwmmxt_waccw): Likewise.
4080         (iwmmxt_waligni): Likewise.
4081         (iwmmxt_walignr): Likewise.
4082         (iwmmxt_walignr0): Likewise.
4083         (iwmmxt_walignr1): Likewise.
4084         (iwmmxt_walignr2): Likewise.
4085         (iwmmxt_walignr3): Likewise.
4086         (iwmmxt_wsadb): Likewise.
4087         (iwmmxt_wsadh): Likewise.
4088         (iwmmxt_wsadbz): Likewise.
4089         (iwmmxt_wsadhz): Likewise.
4090         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4091         (iwmmxt_wabsdiffb): Likewise.
4092         (iwmmxt_wabsdiffh): Likewise.
4093         (iwmmxt_wabsdiffw): Likewise.
4094         (iwmmxt_waddsubhx): Likewise
4095         (iwmmxt_wsubaddhx): Likewise.
4096         (addc<mode>3): Likewise.
4097         (iwmmxt_avg4): Likewise.
4098         (iwmmxt_avg4r): Likewise.
4099         (iwmmxt_wmaddsx): Likewise.
4100         (iwmmxt_wmaddux): Likewise.
4101         (iwmmxt_wmaddsn): Likewise.
4102         (iwmmxt_wmaddun): Likewise.
4103         (iwmmxt_wmulwsm): Likewise.
4104         (iwmmxt_wmulwum): Likewise.
4105         (iwmmxt_wmulsmr): Likewise.
4106         (iwmmxt_wmulumr): Likewise.
4107         (iwmmxt_wmulwsmr): Likewise.
4108         (iwmmxt_wmulwumr): Likewise.
4109         (iwmmxt_wmulwl): Likewise.
4110         (iwmmxt_wqmulm): Likewise.
4111         (iwmmxt_wqmulwm): Likewise.
4112         (iwmmxt_wqmulmr): Likewise.
4113         (iwmmxt_wqmulwmr): Likewise.
4114         (iwmmxt_waddbhusm): Likewise.
4115         (iwmmxt_waddbhusl): Likewise.
4116         (iwmmxt_wqmiabb): Likewise.
4117         (iwmmxt_wqmiabt): Likewise.
4118         (iwmmxt_wqmiatb): Likewise.
4119         (iwmmxt_wqmiatt): Likewise.
4120         (iwmmxt_wqmiabbn): Likewise.
4121         (iwmmxt_wqmiabtn): Likewise.
4122         (iwmmxt_wqmiatbn): Likewise.
4123         (iwmmxt_wqmiattn): Likewise.
4124         (iwmmxt_wmiabb): Likewise.
4125         (iwmmxt_wmiabt): Likewise.
4126         (iwmmxt_wmiatb): Likewise.
4127         (iwmmxt_wmiatt): Likewise.
4128         (iwmmxt_wmiabbn): Likewise.
4129         (iwmmxt_wmiabtn): Likewise.
4130         (iwmmxt_wmiatbn): Likewise.
4131         (iwmmxt_wmiattn): Likewise.
4132         (iwmmxt_wmiawbb): Likewise.
4133         (iwmmxt_wmiawbt): Likewise.
4134         (iwmmxt_wmiawtb): Likewise.
4135         (iwmmxt_wmiawtt): Likewise.
4136         (iwmmxt_wmiawbbn): Likewise.
4137         (iwmmxt_wmiawbtn): Likewise.
4138         (iwmmxt_wmiawtbn): Likewise.
4139         (iwmmxt_wmiawttn): Likewise.
4140         (iwmmxt_wmerge): Likewise.
4141         (iwmmxt_tandc<mode>3): Likewise.
4142         (iwmmxt_torc<mode>3): Likewise.
4143         (iwmmxt_torvsc<mode>3): Likewise.
4144         (iwmmxt_textrc<mode>3): Likewise.
4145         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4146         (wmmxt_pack): Likewise.
4147         (wmmxt_mult_c1): Likewise.
4148         (wmmxt_mult_c2): Likewise.
4149         (wmmxt_alu_c1): Likewise.
4150         (wmmxt_alu_c2): Likewise.
4151         (wmmxt_alu_c3): Likewise.
4152         (wmmxt_transfer_c1): Likewise.
4153         (wmmxt_transfer_c2): Likewise.
4154         (wmmxt_transfer_c3): Likewise.
4155         (marvell_f_iwmmxt_wstr): Likewise.
4156         (marvell_f_iwmmxt_wldr): Likewise.
4158 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4160         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4162 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4164         Revert:
4165         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4166         * lra-constraints.c (need_for_split_p): Check call used hard regs
4167         living through calls.
4169         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4170         call used regs for call insn.
4172 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4174         PR target/57736
4175         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4176         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4178 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4180         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4182 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4184         * lra-constraints.c (need_for_split_p): Check call used hard regs
4185         living through calls.
4187 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4189         PR target/57744
4190         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4191         to tie with any other modes.  Eliminate Altivec vector mode tests,
4192         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4193         code, to return 0 if testing MODE2 for a condition, if we've
4194         already tested MODE1 for the same condition.
4196 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4198         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4199         layout.
4201 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4203         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4204         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4206 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4208         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4209         Define.
4210         (aarch64_symbolic_constant_p): Remove.
4211         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4212         static.  Fix line length and white space.
4213         (aarch64_symbolic_constant_p): Remove.
4214         * config/aarch64/predicates.md (aarch64_valid_symref):
4215         Use aarch64_classify_symbol_expression.
4217 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4219         * config/arm/constraints.md (Ts): New constraint.
4220         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4221         16-bit encodings.
4222         (compare_scc): Use "Ts" constraint for operand 0.
4223         (ior_scc_scc): Likewise.
4224         (and_scc_scc): Likewise.
4225         (and_scc_scc_nodom): Likewise.
4226         (ior_scc_scc_cmp): Likewise for operand 7.
4227         (and_scc_scc_cmp): Likewise.
4228         * config/arm/thumb2.md (thumb2_movsi_insn):
4229         Add alternatives for 16-bit encodings.
4230         (thumb2_movhi_insn): Likewise.
4231         (thumb2_movsicc_insn): Likewise.
4232         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4233         (thumb2_negscc): Use "Ts" constraint.
4234         Move mvn instruction outside cond_exec block.
4235         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4236         for 16-bit encodings.
4238 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4240         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4241         encoding.
4242         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4243         (mulsi3subsi): Likewise.
4244         (mulsidi3adddi): Likewise.
4245         (mulsidi3_v6): Likewise.
4246         (umulsidi3_v6): Likewise.
4247         (umulsidi3adddi_v6): Likewise.
4248         (smulsi3_highpart_v6): Likewise.
4249         (umulsi3_highpart_v6): Likewise.
4250         (mulhisi3tb): Likewise.
4251         (mulhisi3bt): Likewise.
4252         (mulhisi3tt): Likewise.
4253         (maddhisi4): Likewise.
4254         (maddhisi4tb): Likewise.
4255         (maddhisi4tt): Likewise.
4256         (maddhidi4): Likewise.
4257         (maddhidi4tb): Likewise.
4258         (maddhidi4tt): Likewise.
4259         (zeroextractsi_compare0_scratch): Likewise.
4260         (insv_zero): Likewise.
4261         (insv_t2): Likewise.
4262         (anddi_notzesidi_di): Likewise.
4263         (anddi_notsesidi_di): Likewise.
4264         (andsi_notsi_si): Likewise.
4265         (iordi_zesidi_di): Likewise.
4266         (xordi_zesidi_di): Likewise.
4267         (andsi_iorsi3_notsi): Likewise.
4268         (smax_0): Likewise.
4269         (smax_m1): Likewise.
4270         (smin_0): Likewise.
4271         (not_shiftsi): Likewise.
4272         (unaligned_loadsi): Likewise.
4273         (unaligned_loadhis): Likewise.
4274         (unaligned_loadhiu): Likewise.
4275         (unaligned_storesi): Likewise.
4276         (unaligned_storehi): Likewise.
4277         (extv_reg): Likewise.
4278         (extzv_t2): Likewise.
4279         (divsi3): Likewise.
4280         (udivsi3): Likewise.
4281         (arm_zero_extendhisi2addsi): Likewise.
4282         (arm_zero_extendqisi2addsi): Likewise.
4283         (compareqi_eq0): Likewise.
4284         (arm_extendhisi2_v6): Likewise.
4285         (arm_extendqisi2addsi): Likewise.
4286         (arm_movt): Likewise.
4287         (thumb2_ldrd): Likewise.
4288         (thumb2_ldrd_base): Likewise.
4289         (thumb2_ldrd_base_neg): Likewise.
4290         (thumb2_strd): Likewise.
4291         (thumb2_strd_base): Likewise.
4292         (thumb2_strd_base_neg): Likewise.
4293         (arm_negsi2): Add alternative for 16-bit encoding.
4294         (arm_one_cmplsi2): Likewise.
4296 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4298         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4299         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4300         (movdfcc): Likewise.
4301         * config/arm/vfp.md (*thumb2_movsf_vfp):
4302         Disable predication for arm_restrict_it.
4303         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4304         (*thumb2_movdfcc_vfp): Likewise.
4305         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4306         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4307         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4308         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4309         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4310         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4311         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4312         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4313         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4314         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4315         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4316         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4317         Disable predication for arm_restrict_it.
4319 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4321         * config/i386/bmiintrin.h (_bextr_u32): New.
4322         (_bextr_u64): Ditto.
4324 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4326         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4327         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4328         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4329         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4330         * config/mips/n32-elf.h: ...this new file.
4332 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4334         PR target/57224
4335         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4336         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4338 2013-06-27  Catherine Moore <clm@codesourcery.com>
4340         * config/mips/mips-tables.opt: Regenerate.
4341         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4342         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4343         * doc/invoke.texi: Add -m14kc.
4345 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4347         PR target/57623
4348         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4349         constraints of operand 1 and 2.
4351         PR target/57623
4352         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4353         to match RTL canonicalization.  Swap predicates and
4354         constraints of operand 1 and 2.
4356 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4358         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4359         Process OP_INOUT regs for splitting too.
4361 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4363         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4364         decl before the loop, initialize to NULL.
4365         (vectorizable_load): Initialize ptr_incr to NULL.
4367 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4369         PR lto/57208
4370         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4371         * ipa-ref.c (ipa_maybe_record_reference): New function.
4372         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4373         * ipa-cp.c (create_specialized_node): Record potential references from
4374         aggvals.
4375         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4377 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4379         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4380         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4381         'mode' to force_reg.
4382         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4383         (aarch64_expand_mov_immediate): Likewise.
4385 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4387         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4388         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4390 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4392         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4393         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4394         (struct machine_function): Add tbegin_p.
4395         (s390_canonicalize_comparison): Fold CC mode compares to
4396         conditional jump if possible.
4397         (s390_emit_jump): Return the emitted jump.
4398         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4399         Handle CCRAWmode compares.
4400         (s390_option_override): Default to -mhtm if available.
4401         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4402         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4403         FPRs instead of df_regs_ever_live_p.
4404         (s390_optimize_nonescaping_tx): New function.
4405         (s390_init_frame_layout): Extend clobbered_regs array to cover
4406         FPRs as well.
4407         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4408         (s390_expand_tbegin): New function.
4409         (enum s390_builtin): New enum definition.
4410         (code_for_builtin): New array definition.
4411         (s390_init_builtins): New function.
4412         (s390_expand_builtin): New function.
4413         (TARGET_INIT_BUILTINS): Define.
4414         (TARGET_EXPAND_BUILTIN): Define.
4415         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4416         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4417         (s390_alc_comparison): Likewise.
4418         * config/s390/s390-modes.def: Add CCRAWmode.
4419         * config/s390/s390.h (processor_flags): Add PF_TX.
4420         (TARGET_CPU_HTM): Define macro.
4421         (TARGET_HTM): Define macro.
4422         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4423         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4424         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4425         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4426         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4427         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4428         ("*cc_to_int"): Move up.
4429         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4430         constants other than 0.
4431         ("*ccraw_to_int"): New insn and splitter definition.
4432         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4433         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4434         ("tx_assist"): New expander.
4435         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4436         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4437         * config/s390/s390.opt: Add -mhtm option.
4438         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4439         * config/s390/htmxlintrin.h: New file.
4440         * config/s390/htmintrin.h: New file.
4441         * config/s390/s390intrin.h: New file.
4442         * doc/extend.texi: Document htm builtins.
4443         * config.gcc: Add the new header files to extra_headers.
4445 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4447         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4448         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4450 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4451             Pat Haugen <pthaugen@us.ibm.com>
4452             Peter Bergner <bergner@vnet.ibm.com>
4454         * config/rs6000/power8.md: New.
4455         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4456         setting for power8 entry.
4457         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4458         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4459         test for Power4/Power5 only.
4460         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4461         support.
4462         (force_new_group): Adjust comment.
4463         * config/rs6000/rs6000.md: Include power8.md.
4465 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4467         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4468         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4469         declaration.
4470         (tune_params): Update comment.
4471         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4472         (arm_max_conditional_execute): New function.
4473         (thumb2_final_prescan_insn): Use max_insn_skipped and
4474         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4476 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4478         PR tree-optimization/57705
4479         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4480         SSA_NAME step, provided that it is not defined inside the loop.
4481         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4482         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4484 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4486         PR middle-end/57670
4487         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4488         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4489         calls in the dump.
4490         (ipa_note_param_call): Initialize member_ptr flag.
4491         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4492         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4493         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4494         (ipa_read_indirect_edge_info): Likewise.
4496 2013-06-25  Richard Biener  <rguenther@suse.de>
4498         PR middle-end/56977
4499         * passes.c (init_optimization_passes): Move pass_fold_builtins
4500         and pass_dce earlier with -Og.
4502 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4504         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4505         <BIT_FIELD_REF>: Remove trailing TAB.
4506         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4507         remove blank line.
4509 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4511         PR tree-optimization/57358
4512         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4513         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4514         (ipa_analyze_params_uses): Generate pessimistic info when true.
4516 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4518         PR tree-optimization/57539
4519         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4520         global.inlined_to of the new node to it.  All callers changed.
4521         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4522         inlining_into, pass it to cgraph_clone_node.
4523         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4524         ipa_free_edge_args_substructures.
4525         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4526         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4527         Assert we have found an rdesc in the rdesc list.
4529 2013-06-24  Richard Biener  <rguenther@suse.de>
4531         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4532         (pointer_set_lookup): Declare.
4533         (class pointer_map): New template class implementing a
4534         generic pointer to T map.
4535         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4536         pointer_map<T>::contains, pointer_map<T>::insert,
4537         pointer_map<T>::traverse): New functions.
4538         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4539         (pointer_set_lookup): New function.
4540         (pointer_set_contains): Use pointer_set_lookup.
4541         (pointer_set_insert): Likewise.
4542         (insert_aux): Remove.
4543         (struct pointer_map_t): Embed a pointer_set_t.
4544         (pointer_map_create): Adjust.
4545         (pointer_map_destroy): Likewise.
4546         (pointer_map_contains): Likewise.
4547         (pointer_map_insert): Likewise.
4548         (pointer_map_traverse): Likewise.
4549         * tree-streamer.h (struct streamer_tree_cache_d): Use a
4550         pointer_map<unsigned> instead of a pointer_map_t.
4551         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4552         (streamer_tree_cache_lookup): Likewise.
4553         (streamer_tree_cache_create): Likewise.
4554         (streamer_tree_cache_delete): Likewise.
4555         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4556         pointer_map<unsigned> instead of a pointer_map_t.
4557         (lto_init_tree_ref_encoder): Adjust.
4558         (lto_destroy_tree_ref_encoder): Likewise.
4559         * lto-section-out.c (lto_output_decl_index): Likewise.
4560         (lto_record_function_out_decl_state): Likewise.
4561         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4563 2013-06-24  Richard Biener  <rguenther@suse.de>
4565         PR tree-optimization/57488
4566         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4568 2013-06-24  Alan Modra  <amodra@gmail.com>
4570         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4571         (gen_easy_altivec_constant): Likewise.
4572         * config/rs6000/predicates.md (easy_vector_constant_add_self,
4573         easy_vector_constant_msb): Likewise.
4575 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
4577         PR target/57688
4578         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4579         add missing return true.
4581 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
4583         PR target/52483
4584         * config/sh/predicates.md (general_extend_operand): Invoke
4585         general_movsrc_operand for memory operands.
4586         (general_movsrc_operand): Allow reg+reg addressing, do not use
4587         general_operand for memory operands.
4589 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
4591         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
4592         when current target options does not apply.
4593         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4594         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4595         * config/i386/bmiintrin.h: Pass appropriate target
4596         attributes to header.
4597         * config/i386/mmintrin.h: Ditto.
4598         * config/i386/nmmintrin.h: Ditto.
4599         * config/i386/avx2intrin.h: Ditto.
4600         * config/i386/fxsrintrin.h: Ditto.
4601         * config/i386/tbmintrin.h: Ditto.
4602         * config/i386/xsaveintrin.h: Ditto.
4603         * config/i386/f16cintrin.h: Ditto.
4604         * config/i386/xtestintrin.h: Ditto.
4605         * config/i386/xsaveoptintrin.h: Ditto.
4606         * config/i386/bmi2intrin.h: Ditto.
4607         * config/i386/lzcntintrin.h: Ditto.
4608         * config/i386/smmintrin.h: Ditto.
4609         * config/i386/wmmintrin.h: Ditto.
4610         * config/i386/x86intrin.h: Remove all header include guards.
4611         * config/i386/prfchwintrin.h: Ditto.
4612         * config/i386/pmmintrin.h: Ditto.
4613         * config/i386/tmmintrin.h: Ditto.
4614         * config/i386/xmmintrin.h: Ditto.
4615         * config/i386/popcntintrin.h: Ditto.
4616         * config/i386/rdseedintrin.h: Ditto.
4617         * config/i386/ammintrin.h: Ditto.
4618         * config/i386/emmintrin.h: Ditto.
4619         * config/i386/immintrin.h: Remove all header include guards.
4620         * config/i386/fma4intrin.h: Ditto.
4621         * config/i386/lwpintrin.h: Ditto.
4622         * config/i386/xopintrin.h: Ditto.
4623         * config/i386/ia32intrin.h: Ditto.
4624         * config/i386/avxintrin.h: Ditto.
4625         * config/i386/rtmintrin.h: Ditto.
4626         * config/i386/fmaintrin.h: Ditto.
4627         * config/i386/mm3dnow.h: Ditto.
4629 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
4631         * common/config/i386/i386-common.c: Handle LZCNT.
4633 2013-06-22  Andi Kleen  <ak@linux.intel.com>
4635         * doc/extend.texi: Use __atomic_store_n instead of
4636         __atomic_store in HLE example.
4638 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
4640         * config/sh/sh.c: Remove <cstdlib> workaround.
4642 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4644         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
4646 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4648         * doc/extend.texi: Document that __atomic_clear and
4649           __atomic_test_and_set should only be used with bool.
4651 2013-06-20  Jan Hubicka  <jh@suse.cz>
4653         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4654         types_same_for_odr.
4655         * tree.c (decls_same_for_odr): New function.
4656         (same_for_edr): New function.
4657         (types_same_for_odr): New function.
4658         (get_binfo_at_offset): Use it.
4659         * tree.h (types_same_for_odr): Declare.
4661 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
4662             Jason Merrill  <jason@redhat.com>
4664         * system.h: Include <cstdlib> as well as <stdlib.h>.
4666 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
4668         PR target/57655
4669         * config/i386/i386.c (construct_container): Report error if
4670         long double is used with disabled x87 float returns.
4672 2013-06-20  Jan Hubicka  <jh@suse.cz>
4674         * lto-cgraph.c (input_symtab): Do not set cgraph state.
4676 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
4678         PR rtl-optimization/57425
4679         PR rtl-optimization/57569
4680         * alias.c (write_dependence_p): Remove parameters mem_mode and
4681         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
4682         Changed all callers.
4683         (canon_anti_dependence): Get comments and semantics in sync.
4684         Add parameter mem_canonicalized.  Changed all callers.
4685         * rtl.h (canon_anti_dependence): Update prototype.
4687 2013-06-20  Richard Biener  <rguenther@suse.de>
4689         * data-streamer-in.c (streamer_read_uhwi): Optimize single
4690         byte case, inline streamer_read_uchar and defer section
4691         overrun check.
4693 2013-06-20  Richard Biener  <rguenther@suse.de>
4695         PR tree-optimization/57584
4696         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4697         SSA names into the expanded expression that take part in
4698         abnormal coalescing.
4700 2013-06-19  Sharad Singhai  <singhai@google.com>
4702         * gcov.c (print_usage): Handle new option.
4703         (process_args): Ditto.
4704         (get_gcov_intermediate_filename): New function.
4705         (output_intermediate_file): New function.
4706         (output_gcov_file): New function
4707         (generate_results): Handle new option.
4708         (release_function): Relase demangled name.
4709         (read_graph_file): Handle demangled name.
4710         (output_lines): Ditto.
4711         * doc/gcov.texi: Document gcov intermediate format.
4713 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
4715         PR bootstrap/57604
4716         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4717         (lra_emit_add): Use the functions.  Add comment about Y as an
4718         address segment.
4720 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
4722         PR driver/57652
4723         * collect2.c (collect_atexit): New.
4724         (collect_exit): Delete.
4725         (main): Register collect_atexit with atexit.
4726         (collect_wait): Change collect_exit to exit.
4727         (do_wait): Same.
4728         * collect2.h (collect_exit): Delete.
4729         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4731 2013-06-19  Wei Mi  <wmi@google.com>
4733         PR rtl-optimization/57518
4734         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4735         if regno is used in paradoxical subreg.
4736         (update_equiv_regs): Check pdx_subregs[regno] before
4737         set a reg to be equivalent with a mem.
4739 2013-06-19  Matthias Klose  <doko@ubuntu.com>
4741         PR driver/57651
4742         * file-find.h (find_a_file): Add a mode parameter.
4743         * file-find.c (find_a_file): Likewise.
4744         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4745         with X_OK for the executables.
4746         * collect2.c (main): Call find_a_file with X_OK.
4748 2013-06-19  Steve Ellcey  <sellcey@mips.com>
4750         PR target/56942
4751         * config/mips/mips.md (casesi_internal_mips16_<mode>):
4752         Use NEXT_INSN instead of next_real_insn.
4754 2013-06-19  Jan Hubicka  <jh@suse.cz>
4756         * cgraph.h (const_value_known_p): Replace by ...
4757         (ctor_for_folding): .. this one.
4758         * cgraphunit.c (process_function_and_variable_attributes): Use it.
4759         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4760         * expr.c (expand_expr_real_1): Likewise.
4761         (string_constant): Likewise.
4762         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4763         * ipa.c (process_references): Likewise.
4764         (symtab_remove_unreachable_nodes): Likewise.
4765         * ipa-inline-analysis.c (param_change_prob): Likewise.
4766         * gimple-fold.c (canonicalize_constructor_val): Likewise.
4767         (get_base_constructor): Likwise.
4768         * varpool.c (varpool_remove_node): Likewise.
4769         (varpool_remove_initializer): LIkewise.
4770         (dump_varpool_node): LIkwise.
4771         (const_value_known_p): Rewrite to ...
4772         (ctor_for_folding): ... this one.
4774 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
4776         PR driver/57651
4777         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4778         PERSONALITY in $PATH derived prefixes.
4780 2013-06-19  Jeff Law  <law@redhat.com>
4782         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4783         in comment.
4785         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4786         (simplify_bitwise_binary): Use it to simpify certain binary ops on
4787         booleans.
4789 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
4791         * config/arm/vfp.md: Move VFP instruction classification documentation
4792         to ...
4793         * config/arm/arm.md: ... here.  Update instruction classification
4794         documentation.
4796 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
4798         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4799         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4800         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
4802 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
4804         PR target/57609
4805         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4806         with NEXT_INSN.  Use tablejump_p to check for jump table data
4807         insns.
4809 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
4811         PR c++/56544
4812         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4813         that now in C++ the value is correct per the C++ standards.
4815 2013-06-19  Richard Biener  <rguenther@suse.de>
4817         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4818         for global context.
4820 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4822         Revert:
4823         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4825         PR target/57609
4826         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4827         with next_active_insn.
4829 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
4831         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4832         functions are inlined during failures to flag an error.
4833         * tree-inline.c (expand_call_inline): Allow the error to be flagged
4834         in early inline pass.
4836 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
4838         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4839         in comments.
4841 2013-06-18  Julian Brown  <julian@codesourcery.com>
4843         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4844         Permit virtual register pre-reload if !strict.
4845         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4846         change.
4847         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4848         prototype.
4849         * config/arm/neon.md (movmisalign<mode>): Use
4850         neon_perm_struct_or_reg_operand instead of
4851         neon_struct_or_register_operand.
4852         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4853         neon_permissive_struct_operand instead of neon_struct_operand.
4854         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4855         neon_vector_mem_operand.
4856         * config/arm/predicates.md (neon_struct_operand): Adjust call to
4857         neon_vector_mem_operand.
4858         (neon_permissive_struct_operand): New.
4859         (neon_struct_or_register_operand): Rename to...
4860         (neon_perm_struct_or_reg_operand): This. Adjust call to
4861         neon_vector_mem_operand.
4863 2013-06-18  Richard Biener  <rguenther@suse.de>
4865         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4866         * lto-streamer.h: Include pointer-set.h.
4867         (struct lto_decl_slot): Remove.
4868         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4869         Remove next_index entry.
4870         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4871         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4872         (lto_init_tree_ref_encoder): Adjust.
4873         (lto_destroy_tree_ref_encoder): Likewise.
4874         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4875         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4876         (lto_output_decl_index): Adjust.
4877         (lto_new_out_decl_state): Likewise.
4878         (lto_record_function_out_decl_state): Likewise.
4879         * lto-streamer-out.c (copy_function): Likewise.
4881 2013-06-18  Richard Biener  <rguenther@suse.de>
4883         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
4884         * cgraphunit.c: Include cfgloop.h.
4885         (init_lowered_empty_function): Initialize the loop tree.
4886         (assemble_thunk): Insert new BBs into loops.
4888 2013-06-18  Richard Biener  <rguenther@suse.de>
4890         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
4891         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
4892         the map from cache entry to cache index optional.
4893         (streamer_tree_cache_replace_tree): Adjust accordingly.
4894         (streamer_tree_cache_append): Likewise.
4895         (streamer_tree_cache_delete): Likewise.
4896         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
4897         streamer cache map from cache entry to cache index.
4898         * lto-streamer-out.c (create_output_block): Adjust.
4900 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
4902         * config/arm/arm.md (attribute "insn"): Move multiplication and
4903         division attributes to...
4904         (attribute "type"): ... here.  Remove mult.
4905         (attribute "mul32"): New attribute.
4906         (attribute "mul64"): Add umaal.
4907         (*arm_mulsi3): Update attributes.
4908         (*arm_mulsi3_v6): Likewise.
4909         (*thumb_mulsi3): Likewise.
4910         (*thumb_mulsi3_v6): Likewise.
4911         (*mulsi3_compare0): Likewise.
4912         (*mulsi3_compare0_v6): Likewise.
4913         (*mulsi_compare0_scratch): Likewise.
4914         (*mulsi_compare0_scratch_v6): Likewise.
4915         (*mulsi3addsi): Likewise.
4916         (*mulsi3addsi_v6): Likewise.
4917         (*mulsi3addsi_compare0): Likewise.
4918         (*mulsi3addsi_compare0_v6): Likewise.
4919         (*mulsi3addsi_compare0_scratch): Likewise.
4920         (*mulsi3addsi_compare0_scratch_v6): Likewise.
4921         (*mulsi3subsi): Likewise.
4922         (*mulsidi3adddi): Likewise.
4923         (*mulsi3addsi_v6): Likewise.
4924         (*mulsidi3adddi_v6): Likewise.
4925         (*mulsidi3_nov6): Likewise.
4926         (*mulsidi3_v6): Likewise.
4927         (*umulsidi3_nov6): Likewise.
4928         (*umulsidi3_v6): Likewise.
4929         (*umulsidi3adddi): Likewise.
4930         (*umulsidi3adddi_v6): Likewise.
4931         (*smulsi3_highpart_nov6): Likewise.
4932         (*smulsi3_highpart_v6): Likewise.
4933         (*umulsi3_highpart_nov6): Likewise.
4934         (*umulsi3_highpart_v6): Likewise.
4935         (mulhisi3): Likewise.
4936         (*mulhisi3tb): Likewise.
4937         (*mulhisi3bt): Likewise.
4938         (*mulhisi3tt): Likewise.
4939         (maddhisi4): Likewise.
4940         (*maddhisi4tb): Likewise.
4941         (*maddhisi4tt): Likewise.
4942         (maddhidi4): Likewise.
4943         (*maddhidi4tb): Likewise.
4944         (*maddhidi4tt): Likewise.
4945         (divsi3): Likewise.
4946         (udivsi3): Likewise.
4947         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
4948         (thumb2_mulsi_short_compare0): Likewise.
4949         (thumb2_mulsi_short_compare0_scratch): Likewise.
4950         * config/arm/arm1020e.md (1020mult1): Update attribute change.
4951         (1020mult2): Likewise.
4952         (1020mult3): Likewise.
4953         (1020mult4): Likewise.
4954         (1020mult5): Likewise.
4955         (1020mult6): Likewise.
4956         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
4957         change.
4958         (cortex_a15_mult64): Likewise.
4959         (cortex_a15_sdiv): Likewise.
4960         (cortex_a15_udiv): Likewise.
4961         * config/arm/arm1026ejs.md (mult1): Update attribute change.
4962         (mult2): Likewise.
4963         (mult3): Likewise.
4964         (mult4): Likewise.
4965         (mult5): Likewise.
4966         (mult6): Likewise.
4967         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
4968         (pj4_ir_div): Likewise.
4969         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
4970         (11_mult2): Likewise.
4971         (11_mult3): Likewise.
4972         (11_mult4): Likewise.
4973         (11_mult5): Likewise.
4974         (11_mult6): Likewise.
4975         (11_mult7): Likewise.
4976         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
4977         (cortex_a8_mla): Likewise.
4978         (cortex_a8_mull): Likewise.
4979         (cortex_a8_smulwy): Likewise.
4980         (cortex_a8_smlald): Likewise.
4981         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
4982         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
4983         (cortex_r4_mul_3): Likewise.
4984         (cortex_r4_mla_4): Likewise.
4985         (cortex_r4_mla_3): Likewise.
4986         (cortex_r4_smlald): Likewise.
4987         (cortex_r4_mull): Likewise.
4988         (cortex_r4_sdiv): Likewise.
4989         (cortex_r4_udiv): Likewise.
4990         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
4991         (cortex_a7_idiv): Likewise.
4992         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
4993         (9_mult2): Likewise.
4994         (9_mult3): Likewise.
4995         (9_mult4): Likewise.
4996         (9_mult5): Likewise.
4997         (9_mult6): Likewise.
4998         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
4999         (cortex_a53_sdiv): Likewise.
5000         (cortex_a53_udiv): Likewise.
5001         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5002         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5003         (mp626_mult2): Likewise.
5004         (mp626_mult3): Likewise.
5005         (mp626_mult4): Likewise.
5006         * config/arm/fa526.md (526_mult1): Update attribute change.
5007         (526_mult2): Likewise.
5008         * config/arm/arm-generic.md (mult): Update attribute change.
5009         (mult_ldsched_strongarm): Likewise.
5010         (mult_ldsched): Likewise.
5011         (multi_cycle): Likewise.
5012         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5013         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5014         (606te_mult2): Likewise.
5015         (606te_mult3): Likewise.
5016         (606te_mult4): Likewise.
5017         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5018         (cortex_a9_mac16): Likewise.
5019         (cortex_a9_multiply): Likewise.
5020         (cortex_a9_mac): Likewise.
5021         (cortex_a9_multiply_long): Likewise.
5022         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5023         (626te_mult2): Likewise.
5024         (626te_mult3): Likewise.
5025         (626te_mult4): Likewise.
5027 2013-06-18  Richard Biener  <rguenther@suse.de>
5029         PR lto/57334
5030         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5032 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5034         PR target/57609
5035         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5036         with next_active_insn.
5038 2013-06-18  Alan Modra  <amodra@gmail.com>
5040         * config/rs6000/rs6000.h (enum data_align): New.
5041         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5042         (DATA_ABI_ALIGNMENT): Define.
5043         (CONSTANT_ALIGNMENT): Correct comment.
5044         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5045         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5047 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5049         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5050         ATTRIBUTE_UNUSED marking.
5052 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5054         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5055         alternative and update.
5056         (aarch64_dup_lanedi): Delete.
5057         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5058         * config/aarch64/aarch64-simd-builtins.def: Update.
5060 2013-06-17  Richard Biener  <rguenther@suse.de>
5062         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5063         (lto_input_scc): Declare.
5064         (lto_input_tree_1): Likewise.
5065         (struct lto_stats_d): Add num_tree_bodies_output and
5066         num_pickle_refs_output.
5067         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5068         (lto_read_tree_1): Split out from ...
5069         (lto_read_tree): ... this.
5070         (lto_input_scc): New function.
5071         (lto_input_tree_1): Split out from ...
5072         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5073         (lto_data_in_create): Create the streamer cache without hashes.
5074         * lto-streamer-out.c (create_output_block): Create the streamer
5075         cache with hashes when not doing WPA.
5076         (lto_write_tree_1): Split out from ...
5077         (lto_write_tree): ... this.
5078         (get_symbol_initial_value): New function.
5079         (lto_output_tree_1): Split out from ...
5080         (lto_output_tree): ... this.  Write trees as series of SCCs
5081         using a DFS walk via DFS_write_tree.
5082         (struct sccs, struct scc_entry): New types.
5083         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5084         (DFS_write_tree_body): New function.
5085         (DFS_write_tree): Likewise.
5086         (hash_tree): Likewise.
5087         (scc_entry_compare): Likewise.
5088         (hash_scc): Likewise.
5089         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5090         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5091         TREE_CHAIN as regular reference.
5092         (streamer_read_integer_cst): Remove.
5093         (streamer_get_pickled_tree): Adjust.
5094         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5095         of DECL_EXTERNALs in BLOCK_VARS for now.
5096         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5097         reference.
5098         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5099         Add hash value argument and record that if hashes are recorded
5100         in the cache.
5101         (streamer_tree_cache_insert_1): Adjust.
5102         (streamer_tree_cache_insert): Likewise.
5103         (streamer_tree_cache_insert_at): Rename to ...
5104         (streamer_tree_cache_replace_tree): ... this and adjust.
5105         (streamer_tree_cache_append): Adjust.
5106         (record_common_node): Likewise.
5107         (streamer_tree_cache_create): Add argument whether to
5108         record hash values together with trees.
5109         (streamer_tree_cache_delete): Adjust.
5110         * tree-streamer.h (struct streamer_tree_cache_d): Add
5111         vector of hashes.
5112         (streamer_read_integer_cst): Remove.
5113         (streamer_tree_cache_insert): Adjust.
5114         (streamer_tree_cache_append): Likewise.
5115         (streamer_tree_cache_insert_at): Rename to ...
5116         (streamer_tree_cache_replace_tree): ... this and adjust.
5117         (streamer_tree_cache_create): Add argument whether to record hashes.
5118         (streamer_tree_cache_get): Rename to ...
5119         (streamer_tree_cache_get_tree): ... this.
5120         (streamer_tree_cache_get_hash): New function.
5121         * tree.c (cache_integer_cst): New function.
5122         * tree.h (cache_integer_cst): Declare.
5123         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5124         * lto-symtab.c (lto_varpool_replace_node): Only release
5125         DECL_INITIAL of non-prevailing decls.
5126         * varpool.c (varpool_remove_initializer): Do not release
5127         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5129 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5131         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5132         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5133         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5134         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5135         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5136         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5137         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5138         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5139         instead of TARGET_64BIT.
5140         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5141         Require ISA_HAS_<D>DIV.
5143 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5145         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5146         (mips*-*-linux*): Move default with_llsc setting to where other
5147         defaults are set.
5148         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5149         with_arch block.
5150         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5151         Likewise.  Remove default with_tune setting.  Move default float
5152         setting to its own block.  Handle with_llsc in the same block as above.
5154 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5156         PR rtl-optimization/57425
5157         PR rtl-optimization/57569
5158         * alias.c (write_dependence_p): Add new parameters mem_mode,
5159         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5160         Changed all callers.
5161         (canon_anti_dependence): New function.
5162         * cse.c (check_dependence): Use canon_anti_dependence.
5163         * cselib.c (cselib_invalidate_mem): Likewise.
5164         * rtl.h (canon_anti_dependence): Declare.
5166 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5168         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5169         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5170         ".set mips3" for 64-bit targets.
5172 2013-06-15  Dehao Chen  <dehao@google.com>
5174         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5175         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5176         (gimple_check_call_args): Likewise.
5177         * value-prof.c (check_ic_target): Likewise.
5178         * ipa-inline.c (early_inliner): Likewise.
5179         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5180         * cgraph.c (cgraph_create_edge_1): Likewise.
5181         (cgraph_make_edge_direct): Likewise.
5183 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5185         PR target/57615
5186         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5187         rs6000_output_move_128bit to handle emitting quad memory
5188         operations.  Set attribute length to 8 bytes.
5190 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5192         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5193         New pattern.
5194         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5195         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5196         (aarch64_<su>mlsl<mode>): Likewise.
5198 2013-06-14  Mike Stump  <mikestump@comcast.net>
5200         * Makefile.in (TARGET_H): Add insn-codes.h.
5202 2013-06-14  Alan Modra  <amodra@gmail.com>
5204         PR middle-end/57134
5205         PR middle-end/57586
5206         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5207         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5208         bitfield expansion when EXPAND_MEMORY.
5209         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5211 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5213         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5214         test for clearing quad memory on 32-bit later.
5216 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5218         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5219         (fold_negate_expr): Likewise.
5220         (fold_real_zero_addition_p): Handle vectors.
5221         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5223 2013-06-14  Alan Modra  <amodra@gmail.com>
5225         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5227 2013-06-13  Jan Hubicka  <jh@suse.cz>
5229         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5230         Local comdats are not externally visible.
5231         * symtab.c (dump_symtab_base): Dump externally visible.
5232         (verify_symtab_base): Verify back links in the symtab hash.
5234 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5236         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5237         CONVERT_EXPR as equal nodes.
5239 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5241         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5243 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5245         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5246         Generalize to complex and vector.
5247         * tree.c (build_all_ones_cst): New function.
5248         * tree.h (build_all_ones_cst): Declare it.
5250 2013-06-13  Alan Modra  <amodra@gmail.com>
5252         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5253         * config/rs6000/rs6000.md (signbittf2): New insn.
5254         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5255         (abstf2_internal, cmptf_internal2): Likewise.
5256         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5258 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5259             Pat Haugen <pthaugen@us.ibm.com>
5260             Peter Bergner <bergner@vnet.ibm.com>
5262         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5263         power8 byte, half-word, and quad-word atomic instructions.
5264         (emit_store_conditional): Likewise.
5265         (rs6000_expand_atomic_compare_and_swap): Likewise.
5266         (rs6000_expand_atomic_op): Likewise.
5268         * config/rs6000/sync.md (larx): Add new modes for power8.
5269         (stcx): Likewise.
5270         (AINT): New mode iterator to include TImode as well as normal
5271         integer modes on power8.
5272         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5273         that VSX registers are not considered.  Use AINT mode iterator
5274         instead of INT1 to allow inclusion of quad word atomic operations
5275         on power8.
5276         (load_locked<mode>): Likewise.
5277         (store_conditional<mode>): Likewise.
5278         (atomic_compare_and_swap<mode>): Likewise.
5279         (atomic_exchange<mode>): Likewise.
5280         (atomic_nand<mode>): Likewise.
5281         (atomic_fetch_<fetchop_name><mode>): Likewise.
5282         (atomic_nand_fetch<mode>): Likewise.
5283         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5284         each type.
5285         (ATOMIC): On power8, add QImode, HImode modes.
5286         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5287         modes that promote to SImode.
5288         (load_lockedti): Convert TImode arguments to PTImode, so that we
5289         get a guaranteed even/odd register pair.
5290         (load_lockedpti): Likewise.
5291         (store_conditionalti): Likewise.
5292         (store_conditionalpti): Likewise.
5294         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5295         atomic load/store instructions.
5296         (HSI): Likewise.
5298 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5300         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5301         loads.
5302         (insn_count): New attribute, with most cases extracted from...
5303         (length): ...here.  Redefine most cases in terms of insn_count.
5304         (single_insn): Delete.
5305         (can_delay): Use insn_count to check for single instructions.
5306         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5307         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5308         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5309         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5310         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5311         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5312         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5313         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5314         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5315         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5316         rather than "length".
5317         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5318         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5319         Use "insn_count" rather than "length".
5320         * config/mips/mips-dsp.md
5321         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5322         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5323         length attributes.
5325 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5327         PR tree-optimization/57361
5328         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5330 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5332         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5333         to split.
5334         (aarch64_simd_combine<mode>): New instruction expansion.
5335         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5336         function prototype.
5337         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5338         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5340 2013-06-12  Jan Hubicka  <jh@suse.cz>
5342         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5343         decl has when in streaming stage.
5344         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5345         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5347 2013-06-12  Roland Stigge <stigge@antcom.de>
5349         PR target/57578
5350         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5352 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5354         PR tree-optimization/57537
5355         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5356         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5358 2013-06-12  Richard Biener  <rguenther@suse.de>
5360         * data-streamer.h (streamer_write_char_stream): CSE
5361         obs->current_pointer.
5362         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5363         streamer_write_char_stream manually and optimize the resulting loop.
5364         (streamer_write_hwi_stream): Likewise.
5366 2013-06-12  Jan Hubicka  <jh@suse.cz>
5368         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5369         * cgraph.h (varpool_create_empty_node): Declare.
5370         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5371         duplicated nodes.
5372         * symtab.c (symtab_unregister_node): Be lax about missin entries
5373         in node hash.
5374         (symtab_get_node): Update comment.
5375         * varpool.c (varpool_create_empty_node): Break out from ...
5376         (varpool_node_for_decl): ... here.
5377         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5379 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5381         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5382         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5383         part.  Use straight-line flow at the end.
5384         <COMPONENT_REF>: Remove superfluous else.
5385         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5387 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5389         PR target/56564
5390         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5391         target hook even for !TREE_PUBLIC decls.  If no resolution info
5392         is available, return false for common and external decls.
5394 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5396         * config/rl78/constraints.md (U): New constraint.
5397         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5398         valloc attribute.
5400 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5402         PR target/57589
5403         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5404         to allow returning address to AT_PLATFORM name.
5406 2013-06-11  Jan Hubicka  <jh@suse.cz>
5408         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5409         * cgraph.h (symtab_node_base): Add weakref flag.
5410         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5411         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5412         (output_weakrefs): Use weakref flag.
5413         * fold-const.c (simple_operand_p): Handle WEAK.
5414         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5415         * ipa.c (varpool_externally_visible_p): Drop weakref.
5416         (function_and_variable_visibility): Update comment; fix weakref
5417         sanity checks; do not clear DECL_WEAK on them.
5418         * lto-cgraph.c (lto_output_node): update.
5419         (lto_output_varpool_node): Update.
5420         (input_overwrite_node): Update.
5421         (input_node): Update.
5422         (input_varpool_node): Update.
5423         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5424         (lto_symtab_merge_symbols): Add sanity check.
5425         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5426         * passes.c (rest_of_decl_compilation): Set static flag, too.
5427         * symtab.c (dump_symtab_base): Dump weakref.
5428         (verify_symtab_base): Sanity check weakrefs.
5429         (symtab_make_decl_local): Remove duplicated code.
5430         (symtab_alias_ultimate_target): Simplify.
5431         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5433 2013-06-11  DJ Delorie  <dj@redhat.com>
5435         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5436         (rl78_unwind_word_mode): New.
5438 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5440         * final.c (debug_prefix_maps): Make static.
5442 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5444         * function.c (initial_trampoline): Remove stray copy.
5446 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5448         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5450 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5452         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5453         within bounds at the beginning of the function.
5455 2013-06-11  Alan Modra  <amodra@gmail.com>
5457         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5458         reporting.
5459         (get_named_section): Don't NULL !DECL_P decl.
5461 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5463         * doc/invoke.texi (core-avx2): Document.
5464         (slm): Likewise.
5465         (atom): Updated with MOVBE.
5467 2013-06-11  Richard Biener  <rguenther@suse.de>
5469         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5471 2013-06-11  Anton Blanchard  <anton@samba.org>
5473         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5474         correct shift value in little-endian mode.
5476 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5478         PR target/56564
5479         * varasm.c (get_variable_align): Move #endif to the right place.
5481 2013-06-10  Cary Coutant  <ccoutant@google.com>
5483         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5484         for hash so that hash table traversal order is deterministic.
5486 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5487             Pat Haugen <pthaugen@us.ibm.com>
5488             Peter Bergner <bergner@vnet.ibm.com>
5490         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5491         of vectors in GPRS if they are direct moves or quad word load or
5492         store moves.
5494         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5495         declaration.
5496         (direct_move_p): Likewise.
5497         (quad_load_store_p): Likewise.
5499         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5500         classes into bins based on the physical register type.
5501         (reg_class_to_reg_type): Likewise.
5502         (IS_STD_REG_TYPE): Likewise.
5503         (IS_FP_VECT_REG_TYPE): Likewise.
5504         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5505         use direct move instructions.
5506         (reload_gpr_vsx): Likewise.
5507         (reload_vsx_gpr): Likewise.
5508         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5509         information that is a simplification of register classes.  Also
5510         precalculate direct move reload helpers.
5511         (direct_move_p): New function to return true if the operation can
5512         be done as a direct move instruciton.
5513         (quad_load_store_p): New function to return true if the operation
5514         is a quad memory operation.
5515         (rs6000_legitimize_address): If quad memory, only allow register
5516         indirect for TImode addresses.
5517         (rs6000_legitimate_address_p): Likewise.
5518         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5519         (rs6000_reload_register_type): Likewise.
5520         (register_to_reg_type): Return register type.
5521         (rs6000_secondary_reload_simple_move): New helper function for
5522         secondary reload and secondary memory needed to identify anything
5523         that is a simple move, and does not need reloading.
5524         (rs6000_secondary_reload_direct_move): New helper function for
5525         secondary reload to identify cases that can be done with several
5526         instructions via the direct move instructions.
5527         (rs6000_secondary_reload_move): New helper function for secondary
5528         reload to identify moves between register types that can be done.
5529         (rs6000_secondary_reload): Add support for quad memory operations
5530         and for direct move.
5531         (rs6000_secondary_memory_needed): Likewise.
5532         (rs6000_debug_secondary_memory_needed): Change argument names.
5533         (rs6000_output_move_128bit): New function to return the move to
5534         use for 128-bit moves, including knowing about the various
5535         limitations of quad memory operations.
5537         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5538         memory operations.  call rs6000_output_move_128bit for the actual
5539         instruciton(s) to generate.
5540         (vsx_movti_64bit): Likewise.
5542         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5543         (UNSPEC_P8V_MTVSRWZ): Likewise.
5544         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5545         (UNSPEC_P8V_MTVSRD): Likewise.
5546         (UNSPEC_P8V_XXPERMDI): Likewise.
5547         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5548         (UNSPEC_FUSION_GPR): Likewise.
5549         (FMOVE128_GPR): New iterator for direct move.
5550         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5551         (f32_sv): Likewise.
5552         (f32_dm): Likewise.
5553         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5554         loads and direct move instructions.
5555         (zero_extendsidi2_lfiwzx): Likewise.
5556         (extendsidi2_lfiwax): Likewise.
5557         (extendsidi2_nocell): Likewise.
5558         (floatsi<mode>2_lfiwax): Likewise.
5559         (lfiwax): Likewise.
5560         (floatunssi<mode>2_lfiwzx): Likewise.
5561         (lfiwzx): Likewise.
5562         (fix_trunc<mode>_stfiwx): Likewise.
5563         (fixuns_trunc<mode>_stfiwx): Likewise.
5564         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5565         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5566         (parity<mode>2_cmpb): Set length/type attr.
5567         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5568         for 'mr.' to fast_compare.
5569         (bpermd_<mode>): Change type attr to popcnt.
5570         (p8_fmrgow_<mode>): New insns for power8 direct move support.
5571         (p8_mtvsrwz_1): Likewise.
5572         (p8_mtvsrwz_2): Likewise.
5573         (reload_fpr_from_gpr<mode>): Likewise.
5574         (p8_mtvsrd_1): Likewise.
5575         (p8_mtvsrd_2): Likewise.
5576         (p8_xxpermdi_<mode>): Likewise.
5577         (reload_vsx_from_gpr<mode>): Likewise.
5578         (reload_vsx_from_gprsf): Likewise.
5579         (p8_mfvsrd_3_<mode>): LIkewise.
5580         (reload_gpr_from_vsx<mode>): Likewise.
5581         (reload_gpr_from_vsxsf): Likewise.
5582         (p8_mfvsrd_4_disf): Likewise.
5583         (multi-word GPR splits): Do not split direct moves or quad memory
5584         operations.
5586 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
5588         * tree-into-ssa.c (interesting_blocks): Make static.
5590 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
5592         PR target/56564
5593         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5594         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5595         Use DATA_ABI_ALIGNMENT for that case instead if defined.
5596         (get_variable_align): New function.
5597         (get_variable_section, emit_bss, emit_common,
5598         assemble_variable_contents, place_block_symbol): Use
5599         get_variable_align instead of DECL_ALIGN.
5600         (assemble_noswitch_variable): Add align argument, use it
5601         instead of DECL_ALIGN.
5602         (assemble_variable): Adjust caller.  Use get_variable_align
5603         instead of DECL_ALIGN.
5604         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5605         caller.
5606         (DATA_ABI_ALIGNMENT): Define.
5607         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5608         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
5609         opt is false, only return the psABI mandated alignment increase.
5610         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5611         (DATA_ABI_ALIGNMENT): ... this.
5612         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5613         (DATA_ABI_ALIGNMENT): ... this.
5614         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5615         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5616         (DATA_ABI_ALIGNMENT): ... this.
5617         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5618         * doc/tm.texi: Regenerated.
5620 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
5622         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5623         cmp_code to construct REG_EQUAL note.
5625 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
5627         PR target/57568
5628         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5629         that operands[2] doesn't overlap with operands[0].
5631 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
5632             Jan Hubicka  <jh@suse.cz>
5634         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5635         hack to mark symbols as used.
5637 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
5639         PR rtl-optimization/57559
5640         * lra-constraints.c (process_alt_operands): Don't discourage
5641         memory with known offset for offsetable memory constraint.
5642         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5644 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
5646         * varasm.c (struct oc_local_state): Reorder fields.
5647         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5648         and adjust accordingly.
5649         (output_constructor): Reorder initialization code and adjust call to
5650         output_constructor_bitfield.
5652 2013-06-07  Jan Hubicka  <jh@suse.cz>
5654         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5656 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
5658         * tree-object-size.c (unknown): Make const.
5660 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5662         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5663         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5664         for last alternative in the cpu_facility attribute.
5666 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5668         PR target/56315
5669         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5670         (xordi3): Change operand 2 constraint to arm_xordi_operand.
5671         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5672         * config/arm/constraints.md (Dg): New constraint.
5673         * config/arm/neon.md (xordi3_neon): Remove.
5674         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5675         * config/arm/predicates.md (arm_xordi_operand): New predicate.
5677 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5679         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5680         Clean up alternatives.
5682 2013-06-07  Alan Modra  <amodra@gmail.com>
5684         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5685         va_list_gpr_size.
5687 2013-06-07  Alan Modra  <amodra@gmail.com>
5689         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5691 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5693         * config/arm/constraints.md (Df): New constraint.
5694         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5695         Correct length attribute for last two alternatives.
5697 2013-06-07  Alan Modra  <amodra@gmail.com>
5699         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5700         override user -mfp-in-toc.
5701         (offsettable_ok_by_alignment): Consider just the current access
5702         rather than the whole object, unless BLKmode.  Handle
5703         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5704         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5705         for -mcmodel=medium.
5706         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5707         override user -mfp-in-toc or -msum-in-toc.  Default to
5708         -mno-fp-in-toc for -mcmodel=medium.
5710 2013-06-06  DJ Delorie  <dj@redhat.com>
5712         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5713         TARGET_VALID_POINTER_MODE.
5715 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
5716             Pat Haugen <pthaugen@us.ibm.com>
5717             Peter Bergner <bergner@vnet.ibm.com>
5719         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5720         Document new power8 builtins.
5722         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5723         condition code register, to allow 128-bit logical operations to be
5724         done in the VSX or GPR registers.
5725         (nor<mode>3): Use the canonical form for nor.
5726         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5727         vclz*, and vpopcnt* vector instructions.
5728         (nand<mode>3): Likewise.
5729         (orc<mode>3): Likewise.
5730         (clz<mode>2): LIkewise.
5731         (popcount<mode>2): Likewise.
5733         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5734         that only the GPRs are recognized.
5736         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5737         support for new power8 builtins.
5739         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5740         builtin functions.
5741         (xscvdpspn): Likewise.
5742         (vclz): Likewise.
5743         (vclzb): Likewise.
5744         (vclzh): Likewise.
5745         (vclzw): Likewise.
5746         (vclzd): Likewise.
5747         (vpopcnt): Likewise.
5748         (vpopcntb): Likewise.
5749         (vpopcnth): Likewise.
5750         (vpopcntw): Likewise.
5751         (vpopcntd): Likewise.
5752         (vgbbd): Likewise.
5753         (vmrgew): Likewise.
5754         (vmrgow): Likewise.
5755         (eqv): Likewise.
5756         (eqv_v16qi3): Likewise.
5757         (eqv_v8hi3): Likewise.
5758         (eqv_v4si3): Likewise.
5759         (eqv_v2di3): Likewise.
5760         (eqv_v4sf3): Likewise.
5761         (eqv_v2df3): Likewise.
5762         (nand): Likewise.
5763         (nand_v16qi3): Likewise.
5764         (nand_v8hi3): Likewise.
5765         (nand_v4si3): Likewise.
5766         (nand_v2di3): Likewise.
5767         (nand_v4sf3): Likewise.
5768         (nand_v2df3): Likewise.
5769         (orc): Likewise.
5770         (orc_v16qi3): Likewise.
5771         (orc_v8hi3): Likewise.
5772         (orc_v4si3): Likewise.
5773         (orc_v2di3): Likewise.
5774         (orc_v4sf3): Likewise.
5775         (orc_v2df3): Likewise.
5777         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5778         allow power8 quad mode in 64-bit.
5779         (rs6000_builtin_vectorized_function): Add support to vectorize
5780         ISA 2.07 count leading zeros, population count builtins.
5781         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5782         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5783         (builtin_function_type): Add vgbbd builtin function which takes an
5784         unsigned argument.
5785         (altivec_expand_vec_perm_const): Add support for new power8 merge
5786         instructions.
5788         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5789         that does not include TImdoe for use with 32-bit.
5790         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5791         instructions.
5792         (UNSPEC_VSX_CVDPSPN): Likewise.
5793         (vsx_xscvdpspn): Likewise.
5794         (vsx_xscvspdpn): Likewise.
5795         (vsx_xscvdpspn_scalar): Likewise.
5796         (vsx_xscvspdpn_directmove): Likewise.
5797         (vsx_and<mode>3): Split logical operations into 32-bit and
5798         64-bit. Add support to do logical operations on TImode as well as
5799         VSX vector types.  Allow logical operations to be done in either
5800         VSX registers or in general purpose registers in 64-bit mode.  Add
5801         splitters if GPRs were used. For AND, add clobber of CCmode to
5802         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
5803         encoding.
5804         (vsx_and<mode>3_32bit): Likewise.
5805         (vsx_and<mode>3_64bit): Likewise.
5806         (vsx_ior<mode>3): Likewise.
5807         (vsx_ior<mode>3_32bit): Likewise.
5808         (vsx_ior<mode>3_64bit): Likewise.
5809         (vsx_xor<mode>3): Likewise.
5810         (vsx_xor<mode>3_32bit): Likewise.
5811         (vsx_xor<mode>3_64bit): Likewise.
5812         (vsx_one_cmpl<mode>2): Likewise.
5813         (vsx_one_cmpl<mode>2_32bit): Likewise.
5814         (vsx_one_cmpl<mode>2_64bit): Likewise.
5815         (vsx_nor<mode>3): Likewise.
5816         (vsx_nor<mode>3_32bit): Likewise.
5817         (vsx_nor<mode>3_64bit): Likewise.
5818         (vsx_andc<mode>3): Likewise.
5819         (vsx_andc<mode>3_32bit): Likewise.
5820         (vsx_andc<mode>3_64bit): Likewise.
5821         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5822         and xxlorc instructions.
5823         (vsx_eqv<mode>3_64bit): Likewise.
5824         (vsx_nand<mode>3_32bit): Likewise.
5825         (vsx_nand<mode>3_64bit): Likewise.
5826         (vsx_orc<mode>3_32bit): Likewise.
5827         (vsx_orc<mode>3_64bit): Likewise.
5829         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5831         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5832         instruction.
5833         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5834         (p8_vmrgow): Likewise.
5835         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5836         GPRs to be split under VSX.
5837         (p8v_clz<mode>2): Add power8 count leading zero support.
5838         (p8v_popcount<mode>2): Add power8 population count support.
5839         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5840         support.
5842         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5843         instruction.
5845         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5846         builtin functions.
5847         (vec_nand): Likewise.
5848         (vec_vclz): Likewise.
5849         (vec_vclzb): Likewise.
5850         (vec_vclzd): Likewise.
5851         (vec_vclzh): Likewise.
5852         (vec_vclzw): Likewise.
5853         (vec_vgbbd): Likewise.
5854         (vec_vmrgew): Likewise.
5855         (vec_vmrgow): Likewise.
5856         (vec_vpopcnt): Likewise.
5857         (vec_vpopcntb): Likewise.
5858         (vec_vpopcntd): Likewise.
5859         (vec_vpopcnth): Likewise.
5860         (vec_vpopcntw): Likewise.
5862 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5864         PR rtl-optimization/57468
5865         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5866         spilled pseudos.
5868 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5870         PR rtl-optimization/57459
5871         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5872         type when setting live regs.
5874 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5876         * config/s390/s390.opt (mlra): New option.
5877         * config/s390/s390.c (s390_decompose_address): Check displacement
5878         for all registers for LRA.
5879         (s390_secondary_reload): Don't used secondary reloads for LRA.
5880         (s390_lra_p): New function.
5881         (TARGET_LRA_P): Define.
5882         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5883         of attribute cpu_facility to zarch for the last alternative.
5884         (*cmpmem_short): Ditto.
5886 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
5888         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
5889         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
5890         (arm_expand_prologue): Likewise.
5892 2013-06-06  Teresa Johnson  <tejohnson@google.com>
5894         PR c++/53743
5895         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
5896         as this is now done by redirect_edge_and_branch_force.
5897         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
5898         barriers, and fix interaction with splitting.
5899         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
5900         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
5901         reflect changes made in the routine.
5902         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
5903         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
5904         since this is called in cfglayout mode, and replace partition fixup
5905         with assert as that is now done by force_nonfallthru_and_redirect.
5906         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
5907         already be marked with region crossing note.
5908         (insert_section_boundary_note): Make non-static, gate on flag
5909         has_bb_partition, rewrite to also check for multiple partitions.
5910         (rest_of_handle_reorder_blocks): Remove call to
5911         insert_section_boundary_note, now done later during free_cfg.
5912         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
5913         * bb-reorder.h (insert_section_boundary_note): Declare.
5914         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
5915         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
5916         invoke insert_section_boundary_note.
5917         (try_redirect_by_replacing_jump): Remove unnecessary
5918         check for region crossing note.
5919         (fixup_partition_crossing): New function.
5920         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
5921         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
5922         in non-cfglayout mode.
5923         (force_nonfallthru_and_redirect): Fixup partition boundaries,
5924         remove old code that tried to do this. Emit barrier correctly
5925         when we are in cfglayout mode.
5926         (last_bb_in_partition): New function.
5927         (rtl_split_edge): Correctly fixup partition boundaries.
5928         (commit_one_edge_insertion): Remove old code that tried to
5929         fixup region crossing edge since this is now handled in
5930         split_block, and set up insertion point correctly since
5931         block may now end in a jump.
5932         (verify_hot_cold_block_grouping): Guard against checking when not in
5933         linearized RTL mode.
5934         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
5935         notes.
5936         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
5937         rtl_verify_flow_info, so not called in cfglayout mode.
5938         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
5939         (fixup_reorder_chain): Remove old code that attempted to fixup region
5940         crossing note as this is now handled in force_nonfallthru_and_redirect.
5941         (duplicate_insn_chain): Don't duplicate switch section notes.
5942         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
5943         note.
5944         * basic-block.h (emit_barrier_after_bb): Declare.
5946 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5948         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
5949         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
5950         arm_usatsihi): Adjust alternatives for arm_restrict_it.
5952 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5954         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
5955         where appropriate.
5956         * config/arm/ldmstm.md: Regenerate.
5958 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5960         * config/arm/sync.md (atomic_loaddi_1):
5961         Disable predication for arm_restrict_it.
5962         (arm_load_exclusive<mode>): Likewise.
5963         (arm_load_exclusivesi): Likewise.
5964         (arm_load_exclusivedi): Likewise.
5965         (arm_load_acquire_exclusive<mode>): Likewise.
5966         (arm_load_acquire_exclusivesi): Likewise.
5967         (arm_load_acquire_exclusivedi): Likewise.
5968         (arm_store_exclusive<mode>): Likewise.
5969         (arm_store_exclusive<mode>): Likewise.
5970         (arm_store_release_exclusivedi): Likewise.
5971         (arm_store_release_exclusive<mode>): Likewise.
5973 2013-06-06  Richard Biener  <rguenther@suse.de>
5975         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
5976         after LTO_null.
5977         (lto_tag_is_tree_code_p): Adjust.
5978         (lto_tag_is_gimple_code_p): Likewise.
5979         (lto_gimple_code_to_tag): Likewise.
5980         (lto_tag_to_gimple_code): Likewise.
5981         (lto_tree_code_to_tag): Likewise.
5982         (lto_tag_to_tree_code): Likewise.
5983         * data-streamer.h (streamer_write_hwi_in_range): Use
5984         uhwi streaming to stream the normalized range.
5985         (streamer_read_hwi_in_range): Likewise.
5987 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5989         * config/arm/arm.md (enabled_for_depr_it): New attribute.
5990         (predicable_short_it): Likewise.
5991         (predicated): Likewise.
5992         (enabled): Handle above.
5993         (define_cond_exec): Set predicated attribute to yes.
5995 2013-06-05  Mike Stump  <mikestump@comcast.net>
5997         * gdbinit.in (__FUNCTION__): Add.
5999 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6001         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6002         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6004 2013-06-05  Jan Hubicka  <jh@suse.cz>
6006         * varasm.c (mark_decl_referenced): Revert the removal until targets
6007         are fixed.
6009 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6011         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6012         instead of mark_decl_referenced.
6014 2013-06-05  Jan Hubicka  <jh@suse.cz>
6016         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6017         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6018         and symtab_used_from_object_file_p.
6019         (cgraph_make_node_local_1): Clear forced_by_abi.
6020         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6021         * cgraph.h (symtab_node_base): Add forced_by_abi.
6022         (decide_is_variable_needed): Remove.
6023         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6024         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6025         (decide_is_symbol_needed): ... this one; handle symbols in general;
6026         always analyze virtuals; honnor forced_by_abi.
6027         (cgraph_finalize_function): Update.
6028         (varpool_finalize_decl): Update.
6029         (symbol_defined_and_needed): Remove.
6030         (analyze_functions): Update.
6031         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6032         output_refs, input_overwrite_node): Handle forced_by_abi.
6033         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6034         (address_taken_from_non_vtable_p): ... this one.
6035         (comdat_can_be_unshared_p_1): New function.
6036         (cgraph_comdat_can_be_unshared_p): Rename to ...
6037         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6038         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6039         (function_and_variable_visibility): Clear forced_by_abi as needed.
6040         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6041         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6042         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6043         * varpool.c (decide_is_variable_needed): Remove.
6045 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6047         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6048         (arm_option_override): Override arm_restrict_it where appropriate.
6049         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6050         * config/arm/arm.opt (mrestrict-it): New command-line option.
6051         * doc/invoke.texi: Document -mrestrict-it.
6053 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6055         * tsan.c (tsan_atomic_table): Make const.
6057 2013-06-05  Richard Biener  <rguenther@suse.de>
6059         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6060         index associated with the tree we are supposed to replace.
6061         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6062         TREE_ASM_WRITTEN as zero for everything but SSA names.
6064 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6066         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6068 2013-06-04  Jan Hubicka  <jh@suse.cz>
6070         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6071         (input_node, input_varpool_node): Handle correctly external same
6072         body aliases.
6073         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6074         nodes at ltrans stage.
6076 2013-06-04  Jan Hubicka  <jh@suse.cz>
6078         * ipa-inline.c (update_caller_keys): Fix availability test.
6079         (update_callee_keys): Likewise.
6080         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6081         to follow ELF standard.
6083 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6085         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6086         (mips64r5900el-*-elf*): New configurations.
6087         * config/mips/mips-cpus.def (r5900): New processor.
6088         * config/mips/mips-tables.opt: Regenerate.
6089         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6090         (mips_issue_rate): Handle PROCESSOR_R5900.
6091         (mips_reorg_process_insns): Force reorder mode for the R5900.
6092         * config/mips/mips.h (TARGET_MIPS5900): Define.
6093         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6094         TARGET_MIPS5900.
6095         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6096         TARGET_MIPS5900.
6097         * config/mips/mips.md (processor): Add r5900.
6098         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6100 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6102         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6103         into function to generate MOVI instruction.
6104         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6105         (aarch64_preferred_simd_mode): Turn into wrapper.
6106         (aarch64_output_scalar_simd_mov_immediate): New function.
6107         * config/aarch64/aarch64-protos.h: Add prototype for above.
6109 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6111         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6112         element_char member.
6113         (sizetochar): Return signed char.
6114         (aarch64_simd_valid_immediate): Remove elchar and other
6115         unnecessary variables.
6116         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6117         Calculate element_char as required.
6118         * config/aarch64/aarch64-protos.h: Update and move prototype
6119         for aarch64_output_simd_mov_immediate.
6120         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6121         Update arguments.
6123 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6125         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6126         information completed by aarch64_simd_valid_immediate.
6127         (aarch64_legitimate_constant_p): Update arguments.
6128         (aarch64_simd_valid_immediate): Work with struct rather than many
6129         pointers.
6130         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6131         (aarch64_simd_make_constant): Update arguments.
6132         (aarch64_output_simd_mov_immediate): Work with struct rather than
6133         many pointers.  Output immediate directly rather than as operand.
6134         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6135         Update prototype.
6136         * config/aarch64/constraints.md (Dn): Update arguments.
6138 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6140         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6141         longer static.
6142         (aarch64_simd_immediate_valid_for_move): Remove.
6143         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6144         (aarch64_simd_make_constant): Update call.
6145         (aarch64_output_simd_mov_immediate): Update call.
6146         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6147         Add prototype.
6148         * config/aarch64/constraints.md (Dn): Update call.
6150 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6152         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6153         return type to bool for prototype.
6154         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6155         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6156         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6158 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6160         * config/mips/mips.opt (meva): New.
6161         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6162         (ASM_SPEC): Handle -meva.
6163         * doc/invoke.texi (meva):  Document.
6165 2013-06-04  Alan Modra  <amodra@gmail.com>
6167         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6168         constant output.
6170 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6172         * rtl.def: Add extra fourth optional field to define_cond_exec.
6173         * gensupport.c (process_one_cond_exec): Process attributes from
6174         define_cond_exec.
6175         * doc/md.texi: Document fourth field in define_cond_exec.
6177 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6179         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6180         out the processing order as in store_bit_field_1.
6182 2013-06-04  Jan Hubicka  <jh@suse.cz>
6184         PR middle-end/57500
6185         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6186         non-VAR_DECL node if it does not exist yet.
6188 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6190         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6191         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6192         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6193         target_cpu_default setting.
6195 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6197         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6198         default to -fopt-info=optimized instead of all.
6199         * doc/invoke.texi: Ditto.
6200         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6201         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6202         (execute_vect_slp): Emit BB vectorization success under
6203         MSG_OPTIMIZED_LOCATIONS.
6204         * tree-vect-slp.c (vect_slp_transform_bb): Change
6205         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6206         * tree-vect-loop.c (vect_transform_loop): Ditto.
6208 2013-06-03  Jason Merrill  <jason@redhat.com>
6210         PR c++/57415
6211         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6212         Use TARGET_EXPR for C++.
6214 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6216         PR rtl-optimization/57268
6217         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6218         if DEBUG_INSN_P (insn).
6220         Reapply
6221         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6223         PR rtl-optimization/57268
6224         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6225         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6227 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6229         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6230         (ix86_avoid_lea_for_addr): Likewise.
6231         (exact_dependency_1): Likewise.
6232         (ix86_adjust_cost): Likewise.
6233         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6234         removed.
6235         (do_reorder_for_imul): Fix typo, formatting and
6236         !reload_completed check removed.
6237         (ix86_sched_reorder): Fix typo and formatting.
6238         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6239         list.
6241 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6243         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6245 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6247         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6248         <STRING_CST>: Likewise.
6249         <VECTOR_CST>: Likewise.
6251 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6252             Mikael Morin  <mikael@gcc.gnu.org>
6254         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6255         * config.in: Regenerated.
6256         * configure: Regenerated.
6258 2013-06-01  Jan Hubicka  <jh@suse.cz>
6260         PR middle-end/57366
6261         * cgraphunit.c (compile): When weakref is not supported,
6262         set up transparent aliases before final output pass.
6263         * varasm.c (assemble_alias): Do not try to do it here.
6265 2013-06-01  Jan Hubicka  <jh@suse.cz>
6267         PR middle-end/57467
6268         * passes.c (for_per_function): Skip unanalyzed functions.
6270 2013-06-01  Jan Hubicka  <jh@suse.cz>
6272         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6273         (lto_symtab_merge_symbols_1): ... this one.
6274         (lto_symtab_merge_cgraph_nodes): Rename to ...
6275         (lto_symtab_merge_symbols): ... this one; simplify.
6276         * cgraph.c (same_body_aliases_done): Rename to ...
6277         (cpp_implicit_aliases_done): ... this one.
6278         (cgraph_create_function_alias): Update.
6279         (cgraph_same_body_alias): Update.
6280         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6281         (verify_edge_corresponds_to_fndecl): Simplify.
6282         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6283         (cgraph_node): Remove same_body_alias.
6284         (varpool_node): Remove alias_of and extra_name_alias.
6285         (same_body_aliases_done): Rename to ..
6286         (cpp_implicit_aliases_done): ... this one.
6287         (symtab_alias_ultimate_target): Add default parameter.
6288         (symtab_resolve_alias): New function.
6289         (fixup_same_cpp_alias_visibility): Declare.
6290         (cgraph_function_node): Add default parameter.
6291         (cgraph_node_asm_name): Likewise.
6292         (cgraph_function_or_thunk_node): Add default parameter; do
6293         not ICE when it is NULL.
6294         (varpool_variable_node): Likewise.
6295         * tree-emutls.c (create_emultls_var): Update.
6296         (ipa_lower_emutls): Update.
6297         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6298         (cgraph_reset_node): Reset alias info.
6299         (cgraph_finalize_function): Update.
6300         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6301         (analyze_function): Simplify.
6302         (cgraph_process_same_body_aliases): Simplify.
6303         (analyze_functions): Fixup same body aliases.
6304         (handle_alias_pairs): Simplify.
6305         (assemble_thunk): Update.
6306         (assemble_thunks_and_aliases): Update.
6307         (output_weakrefs): Rewrite.
6308         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6309         (lto_output_varpool_node): Likewise.
6310         (compute_ltrans_boundary): Remve assert.
6311         (get_alias_symbol): New functoin.
6312         (input_node): Rewrite alias handling.
6313         (input_varpool_node): Likewise.
6314         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6315         * ipa.c (process_references): Handle weakrefs correctly.
6316         (symtab_remove_unreachable_nodes): Likewise.
6317         * trans-mem.c (get_cg_data): Update.
6318         (ipa_tm_create_version_alias): Update.
6319         (ipa_tm_execute): Update.
6320         * symtab.c (dump_symtab_base): Dump aliases.
6321         (verify_symtab_base): Verify aliases.
6322         (symtab_node_availability): New function.
6323         (symtab_alias_ultimate_target): Simplify.
6324         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6325         handle all the fixup cases.
6326         (symtab_resolve_alias): New function.
6327         * passes.c (ipa_write_summaries): Handle weakrefs.
6328         * varpool.c (varpool_analyze_node): Simplify.
6329         (assemble_aliases): Update.
6330         (varpool_create_variable_alias): Simplify.
6331         (varpool_extra_name_alias): Simplify.
6332         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6333         (lto_symtab_merge_symbols): ... this one.
6335 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6337         Revert
6338         PR rtl-optimization/57268
6339         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6340         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6342 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6344         Partially reverted:
6345         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6347         PR middle-end/57073
6348         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6349         further up.
6351 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6353         PR rtl-optimization/57268
6354         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6355         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6357 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6359         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6360         unordered comparison operators when -fno-trapping-math is in effect
6361         on the e500.
6362         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6363         and implement unordered comparison operators properly on the e500.
6365 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6367         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6368         for constant scalar integers.
6369         (simplify_relational_operation_1): Likewise.
6371 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6373         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6374         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6375         Fix comment.
6377 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6378             Igor Zamyatin  <igor.zamyatin@intel.com>
6380         Silvermont (SLM) architecture performance tuning.
6381         * config/i386/i386.h (enum ix86_tune_indices): Add
6382         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6383         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6385         * config/i386/i386.c (initial_ix86_tune_features)
6386         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6387         (ix86_lea_outperforms): Handle Silvermont tuning.
6388         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6389         call.
6390         (ix86_use_lea_for_mov): Likewise.
6391         (ix86_avoid_lea_for_addr): Likewise.
6392         (ix86_lea_for_add_ok): Likewise.
6393         (exact_dependency_1): New function.
6394         (exact_store_load_dependency): Likewise.
6395         (ix86_adjust_cost): Handle Silvermont tuning.
6396         (do_reoder_for_imul): Likewise.
6397         (swap_top_of_ready_list): New function.
6398         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6400         * config/i386/i386.md (peepholes that split memory operand in fp
6401         converts): New.
6403 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6405         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6406         Remove un-necessary braces.
6408 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6410         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6411         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6413 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6415         PR middle-end/57073
6416         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6418 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6420         PR target/56315
6421         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6422         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6423         * config/arm/neon.md (iordi3_neon): Remove.
6424         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6425         * config/arm/predicates.md (imm_for_neon_logic_operand):
6426         Move to earlier in the file.
6427         (neon_logic_op2): Likewise.
6428         (arm_iordi_operand_neon): New predicate.
6430 2013-05-31  Richard Biener  <rguenther@suse.de>
6432         PR tree-optimization/57478
6433         PR tree-optimization/57453
6434         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6435         are life as well.
6437 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6439         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6440         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6442 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6443             Thomas Koenig  <tkoenig@gcc.gnu.org>
6445         PR middle-end/57073
6446         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6447         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6449 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6451         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6453 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6455         * target.def (register_usage_leveling_p): New hook.
6456         * targhooks.c (default_register_usage_leveling_p): New.
6457         * targhooks.h (default_register_usage_leveling_p): New prototype.
6458         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6459         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6460         * doc/tm.texi: Update.
6461         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6463 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6465         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6466         (*insv_reg<mode>): New define_insn.
6468 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6470         PR rtl-optimization/57439
6471         * postreload.c (move2add_valid_value_p): Check that we have
6472         a zero subreg_regno_offset when accessing the register in
6473         the requested mode.
6475 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6476             Igor Zamyatin  <igor.zamyatin@intel.com>
6478         Silvermont (SLM) architecture pipeline model, tuning and
6479         insn selection.
6480         * config.gcc: Add slm config options and target.
6482         * config/i386/slm.md: New.
6484         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6486         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6487         PROCESSOR_SLM.
6488         (ix86_target_macros_internal): Likewise.
6490         * gcc/config/i386/i386.c (slm_cost): New cost.
6491         (m_SLM): New macro flag.
6492         (initial_ix86_tune_features): Set m_SLM.
6493         (x86_accumulate_outgoing_args): Likewise.
6494         (x86_arch_always_fancy_math_387): Likewise.
6495         (processor_target_table): Add slm cost.
6496         (cpu_names): Add slm cpu name.
6497         (x86_option_override_internal): Set SLM ISA.
6498         (ix86_issue_rate): New case PROCESSOR_SLM.
6499         (ia32_multipass_dfa_lookahead): Likewise.
6500         (fold_builtin_cpu): Add slm.
6502         * config/i386/i386.h (TARGET_SLM): New target macro.
6503         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6504         (processor_type): Add PROCESSOR_SLM.
6506         * config/i386/i386.md (cpu): Add new value "slm".
6507         (slm.md): Include slm.md.
6509 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6510             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6512         * config/arm/arm-protos.h: Add and update function protos.
6513         * config/arm/arm.c (use_simple_return_p): New added.
6514         (thumb2_expand_return): Check simple_return flag.
6515         * config/arm/arm.md: Add simple_return and conditional simple_return.
6516         * config/arm/iterators.md: Add iterator for return and simple_return.
6518 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6520         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6521         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6522         (arm_emit_vfp_multi_reg_pop): Likewise.
6523         (thumb2_emit_ldrd_pop): Likewise.
6524         (arm_expand_epilogue): Add misc REG_CFA notes.
6525         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6527 2013-05-29  Lawrence Crowl  <crowl@google.com>
6529         * config/arm/t-arm: Update for below.
6531         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6532         Change type to hash_table.  Update dependent calls and types.
6534         * config/i386/t-cygming: Update for below.
6536         * config/i386/t-interix: Update for below.
6538         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6539         Change type to hash_table.  Update dependent calls and types.
6540         (i386_find_on_wrapper_list::wrappers): Likewise.
6542         * config/ia64/t-ia64: Update for below.
6544         * config/ia64/ia64.c (bundle_state_table):
6545         Change type to hash_table.  Update dependent calls and types.
6547         * config/mips/mips.c (mips_reorg_process_insns::htab):
6548         Change type to hash_table.  Update dependent calls and types.
6550         * config/sol2.c (solaris_comdat_htab):
6551         Change type to hash_table.  Update dependent calls and types.
6553         * config/t-sol2: Update for above.
6555 2013-05-29  Teresa Johnson  <tejohnson@google.com>
6557         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6558         functions are not yet marked as defined.
6560 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
6561             Pat Haugen <pthaugen@us.ibm.com>
6562             Peter Bergner <bergner@vnet.ibm.com>
6564         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6565         instructions.
6566         (VEC_A): Likewise.
6567         (VEC_C): Likewise.
6568         (vrotl<mode>3): Likewise.
6569         (vashl<mode>3): Likewise.
6570         (vlshr<mode>3): Likewise.
6571         (vashr<mode>3): Likewise.
6573         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6574         support for power8 V2DI builtins.
6576         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6577         power8 V2DI builtins.
6578         (vupkhsw): Likewise.
6579         (vupklsw): Likewise.
6580         (vaddudm): Likewise.
6581         (vminsd): Likewise.
6582         (vmaxsd): Likewise.
6583         (vminud): Likewise.
6584         (vmaxud): Likewise.
6585         (vpkudum): Likewise.
6586         (vpksdss): Likewise.
6587         (vpkudus): Likewise.
6588         (vpksdus): Likewise.
6589         (vrld): Likewise.
6590         (vsld): Likewise.
6591         (vsrd): Likewise.
6592         (vsrad): Likewise.
6593         (vsubudm): Likewise.
6594         (vcmpequd): Likewise.
6595         (vcmpgtsd): Likewise.
6596         (vcmpgtud): Likewise.
6597         (vcmpequd_p): Likewise.
6598         (vcmpgtsd_p): Likewise.
6599         (vcmpgtud_p): Likewise.
6600         (vupkhsw): Likewise.
6601         (vupklsw): Likewise.
6602         (vaddudm): Likewise.
6603         (vmaxsd): Likewise.
6604         (vmaxud): Likewise.
6605         (vminsd): Likewise.
6606         (vminud): Likewise.
6607         (vpksdss): Likewise.
6608         (vpksdus): Likewise.
6609         (vpkudum): Likewise.
6610         (vpkudus): Likewise.
6611         (vrld): Likewise.
6612         (vsld): Likewise.
6613         (vsrad): Likewise.
6614         (vsrd): Likewise.
6615         (vsubudm): Likewise.
6617         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6618         support for power8 V2DI instructions.
6620         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6621         power8 V2DI instructions.  Combine pack and unpack insns to use an
6622         iterator for each mode.  Check whether a particular mode supports
6623         Altivec instructions instead of just checking TARGET_ALTIVEC.
6624         (UNSPEC_VPKUWUM): Likewise.
6625         (UNSPEC_VPKSHSS): Likewise.
6626         (UNSPEC_VPKSWSS): Likewise.
6627         (UNSPEC_VPKUHUS): Likewise.
6628         (UNSPEC_VPKSHUS): Likewise.
6629         (UNSPEC_VPKUWUS): Likewise.
6630         (UNSPEC_VPKSWUS): Likewise.
6631         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6632         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6633         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6634         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6635         (UNSPEC_VUPKHSB): Likewise.
6636         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6637         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6638         (UNSPEC_VUPKHSH): Likewise.
6639         (UNSPEC_VUPKLSB): Likewise.
6640         (UNSPEC_VUPKLSH): Likewise.
6641         (VI2): Likewise.
6642         (VI_char): Likewise.
6643         (VI_scalar): Likewise.
6644         (VI_unit): Likewise.
6645         (VP): Likewise.
6646         (VP_small): Likewise.
6647         (VP_small_lc): Likewise.
6648         (VU_char): Likewise.
6649         (add<mode>3): Likewise.
6650         (altivec_vaddcuw): Likewise.
6651         (altivec_vaddu<VI_char>s): Likewise.
6652         (altivec_vadds<VI_char>s): Likewise.
6653         (sub<mode>3): Likewise.
6654         (altivec_vsubcuw): Likewise.
6655         (altivec_vsubu<VI_char>s): Likewise.
6656         (altivec_vsubs<VI_char>s): Likewise.
6657         (altivec_vavgs<VI_char>): Likewise.
6658         (altivec_vcmpbfp): Likewise.
6659         (altivec_eq<mode>): Likewise.
6660         (altivec_gt<mode>): Likewise.
6661         (altivec_gtu<mode>): Likewise.
6662         (umax<mode>3): Likewise.
6663         (smax<mode>3): Likewise.
6664         (umin<mode>3): Likewise.
6665         (smin<mode>3): Likewise.
6666         (altivec_vpkuhum): Likewise.
6667         (altivec_vpkuwum): Likewise.
6668         (altivec_vpkshss): Likewise.
6669         (altivec_vpkswss): Likewise.
6670         (altivec_vpkuhus): Likewise.
6671         (altivec_vpkshus): Likewise.
6672         (altivec_vpkuwus): Likewise.
6673         (altivec_vpkswus): Likewise.
6674         (altivec_vpks<VI_char>ss): Likewise.
6675         (altivec_vpks<VI_char>us): Likewise.
6676         (altivec_vpku<VI_char>us): Likewise.
6677         (altivec_vpku<VI_char>um): Likewise.
6678         (altivec_vrl<VI_char>): Likewise.
6679         (altivec_vsl<VI_char>): Likewise.
6680         (altivec_vsr<VI_char>): Likewise.
6681         (altivec_vsra<VI_char>): Likewise.
6682         (altivec_vsldoi_<mode>): Likewise.
6683         (altivec_vupkhsb): Likewise.
6684         (altivec_vupkhs<VU_char>): Likewise.
6685         (altivec_vupkls<VU_char>): Likewise.
6686         (altivec_vupkhsh): Likewise.
6687         (altivec_vupklsb): Likewise.
6688         (altivec_vupklsh): Likewise.
6689         (altivec_vcmpequ<VI_char>_p): Likewise.
6690         (altivec_vcmpgts<VI_char>_p): Likewise.
6691         (altivec_vcmpgtu<VI_char>_p): Likewise.
6692         (abs<mode>2): Likewise.
6693         (vec_unpacks_hi_v16qi): Likewise.
6694         (vec_unpacks_hi_v8hi): Likewise.
6695         (vec_unpacks_lo_v16qi): Likewise.
6696         (vec_unpacks_hi_<VP_small_lc>): Likewise.
6697         (vec_unpacks_lo_v8hi): Likewise.
6698         (vec_unpacks_lo_<VP_small_lc>): Likewise.
6699         (vec_pack_trunc_v8h): Likewise.
6700         (vec_pack_trunc_v4si): Likewise.
6701         (vec_pack_trunc_<mode>): Likewise.
6703         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6704         V2DI builtins.
6705         (vec_vmaxsd): Likewise.
6706         (vec_vmaxud): Likewise.
6707         (vec_vminsd): Likewise.
6708         (vec_vminud): Likewise.
6709         (vec_vpksdss): Likewise.
6710         (vec_vpksdus): Likewise.
6711         (vec_vpkudum): Likewise.
6712         (vec_vpkudus): Likewise.
6713         (vec_vrld): Likewise.
6714         (vec_vsld): Likewise.
6715         (vec_vsrad): Likewise.
6716         (vec_vsrd): Likewise.
6717         (vec_vsubudm): Likewise.
6718         (vec_vupkhsw): Likewise.
6719         (vec_vupklsw): Likewise.
6721 2013-05-29  Jan Hubicka  <jh@suse.cz>
6723         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6724         flags; reorder rest of fields in more consistent way.
6725         (varpool_node): Remove analyzed, finalized and alias.
6726         (cgraph_ndoe): Likewise.
6727         (symtab_alias_ultimate_target): New function.
6728         (cgraph_function_node): Move offline.
6729         (cgraph_reset_node): Declare.
6730         (cgraph_comdat_can_be_unshared_p): Remove.
6731         (varpool_remove_initializer): Declare.
6732         (varpool_first_defined_variable, varpool_next_defined_variable
6733         cgraph_first_defined_function, cgraph_next_defined_function): Update.
6734         (cgraph_function_with_gimple_body_p): Update.
6735         (varpool_all_refs_explicit_p): Update.
6736         (symtab_alias_target): New function.
6737         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6738         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
6739         (cgraph_function_or_thunk_node): Simplify using
6740         symtab_alias_ultimate_target.
6741         (varpool_variable_node): Likewise.
6742         * cgraph.c (cgraph_create_function_alias): Update.
6743         (cgraph_add_thunk): Update.
6744         (cgraph_remove_node): Update.
6745         (dump_cgraph_node): Do not dump removed flags.
6746         (cgraph_function_body_availability): Update.
6747         (cgraph_propagate_frequency): Update.
6748         (verify_cgraph_node): Check sanity of local flag.
6749         (cgraph_function_node): Move here from cgraph.h; revamp for
6750         cgraph_function_or_thunk_node.
6751         * lto-symtab.c (lto_varpool_replace_node): Update.
6752         (lto_symtab_resolve_can_prevail_p): Update.
6753         (lto_symtab_merge_cgraph_nodes): Update.
6754         * ipa-cp.c (determine_versionability, initialize_node_lattices,
6755         propagate_constants_accross_call, devirtualization_time_bonus,
6756         ipcp_propagate_stage): Update.
6757         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
6758         * ipa-inline-transform.c (clone_inlined_nodes,
6759         preserve_function_body_p): Update.
6760         * ipa-reference.c (propagate): Update.
6761         (write_node_summary_p): Update.
6762         * toplev.c (wrapup_global_declaration_2): Update.
6763         * cgraphunit.c (cgraph_analyze_function): Rename to ...
6764         (analyze_function) ... this one.
6765         (cgraph_process_new_functions): Update.
6766         (cgraph_reset_node): Export.
6767         (cgraph_finalize_function): Update.
6768         (cgraph_add_new_function): Update.
6769         (process_function_and_variable_attributes): Update.
6770         (varpool_finalize_decl): Update.
6771         (symbol_finalized): Remove.
6772         (symbol_finalized_and_needed): Rename to ...
6773         (symbol_defined_and_needed): ... update.
6774         (cgraph_analyze_functions): Update.
6775         (handle_alias_pairs): Update.
6776         (mark_functions_to_output): Update.
6777         (assemble_thunk): Update.
6778         (output_in_order): Update.
6779         (output_weakrefs): Update.
6780         (finalize_compilation_unit): Update.
6781         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6782         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6783         input_node, input_varpool_node): Update.
6784         * dbxout.c (dbxout_expand_expr): Update.
6785         * cgraphclones.c (cgraph_clone_node): Update.
6786         (cgraph_copy_node_for_versioning): Update.
6787         (cgraph_materialize_clone): Update.
6788         (cgraph_materialize_all_clones): Update.
6789         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6790         propagate_pure_const, propagate_nothrow): Update.
6791         * lto-streamer-out.c (lto_output, write_symbol): Update.
6792         * ipa-utils.c (ipa_reverse_postorder): Update.
6793         * ipa-inline.c (can_inline_edge_p): Update.
6794         (update_caller_keys, ipa_inline): Update.
6795         * dwarf2out.c (reference_to_unused,
6796         premark_types_used_by_global_vars_helper): Update.
6797         * tree-eh.c (tree_could_trap_p): Update.
6798         * ipa-split.c (consider_split, execute_split_functions): Update.
6799         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
6800         has_addr_references_p): Update; move ahead in file for better
6801         readability.
6802         (process_references): Simplify.
6803         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6804         bodies are removed.
6805         (cgraph_comdat_can_be_unshared_p): Make static.
6806         (cgraph_externally_visible_p): Update.
6807         (varpool_externally_visible_p): Update.
6808         (function_and_variable_visibility): Update.
6809         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6810         ipa_tm_mark_force_output_node): Update.
6811         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6812         estimate_edge_devirt_benefit, inline_generate_summary,
6813         inline_write_summary): Update.
6814         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6815         * ipa-prop.c (ipa_compute_jump_functions): Update.
6816         (ipa_print_node_params, ipa_prop_read_section,
6817         ipa_update_after_lto_read, read_replacements_section): Update.
6818         * varasm.c (mark_decl_referenced): Update.
6819         (assemble_alias, dump_tm_clone_pairs): Update.
6820         * tree-inline.c (copy_bb): Update.
6821         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6822         Update.
6823         * symtab.c (dump_symtab_base): Print new flags.
6824         (verify_symtab_base): Verify new flags.
6825         (symtab_alias_ultimate_target): New function.
6826         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6827         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6828         Update.
6829         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6830         Update.
6831         * i386.c (ix86_get_function_versions_dispatcher,
6832         ix86_generate_version_dispatcher_body): Update.
6833         (fold_builtin_cpu): Use varpool_add_new_variable.
6834         * varpool.c (varpool_remove_initializer): Break out from ...
6835         (varpool_remove_node): ... this one.
6836         (dump_varpool_node, varpool_node_for_asm,
6837         cgraph_variable_initializer_availability, varpool_analyze_node,
6838         varpool_assemble_decl, varpool_remove_unreferenced_decls,
6839         varpool_finalize_named_section_flags, varpool_create_variable_alias):
6840         Update.
6842 2013-05-29  Jan Hubicka  <jh@suse.cz>
6844         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6846 2013-05-29  Easwaran Raman  <eraman@google.com>
6848         PR tree-optimization/57442
6849         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6850         when control exits the main loop.
6852 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
6854         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6855         and RX600.
6856         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
6857         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6858         * rx/t-rx: Add rx100 under multi library matches option for nofpu
6859         option.
6861 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6863         PR tree-optimization/57441
6864         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6865         Don't limit size of incr_vec to number of candidates.
6867 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
6869         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6870         and mips16 directories.
6871         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
6872         (MULTILIB_DIRNAMES): Ditto.
6873         (MULTILIB_EXCEPTIONS): Add new exceptions.
6874         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6875         (MULTILIB_DIRNAMES): Ditto.
6876         (MULTILIB_EXCEPTIONS): Add new exceptions.
6878 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6879             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6881         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6882         SYMBOL_TINY_ABSOLUTE.
6883         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
6884         SYMBOL_TINY_ABSOLUTE.
6885         (aarch64_expand_mov_immediate): Likewise.
6886         (aarch64_classify_symbol): Likewise.
6887         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
6888         Permit SYMBOL_TINY_ABSOLUTE.
6889         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
6891 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6892             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6894         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
6895         Refactor if/switch.  Replace gcc_assert with if.
6897 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
6899         * config/i386/i386.c (initial_ix86_tune_features): Enable
6900         FP Reassociation for AMD bdver1 and bdver2.
6902 2013-05-29  Martin Jambor  <mjambor@suse.cz>
6904         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
6905         and IMAGPART_EXPR do not occur within other handled_components.
6907 2013-05-29  Richard Biener  <rguenther@suse.de>
6909         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
6910         access on whether the use is in the BB we currently try to
6911         vectorize.
6912         (vect_bb_vectorization_profitable_p): Pass the BB we currently
6913         vectorize to vect_bb_slp_scalar_cost.
6915 2013-05-29  Richard Biener  <rguenther@suse.de>
6917         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
6918         computing scalar cost offsetted by stmts that are kept live
6919         by scalar uses.
6920         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
6921         for computation of scalar cost.
6923 2013-05-28  Steve Ellcey  <sellcey@mips.com>
6925         * config/mips/mips-cpus.def (mips32r2): Change processor type.
6927 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6929         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
6930         array notation built-in reduction functions.
6931         * doc/passes.texi (Passes): Added documentation about changes done
6932         for Cilk Plus.
6933         * doc/invoke.texi (C Dialect Options): Added documentation about
6934         the -fcilkplus flag.
6935         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
6936         (BUILTINS_DEF): Depend on cilkplus.def.
6937         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
6938         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
6939         * cilkplus.def: New file.
6941 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
6943         PR rtl-optimization/57439
6944         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
6946 2013-05-28  Easwaran Raman  <eraman@google.com>
6948         PR tree-optimization/57337
6949         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
6950         (find_insert_point): Correctly identify the insertion point
6951         when two statements with the same UID is compared.
6953 2013-05-28  Richard Biener  <rguenther@suse.de>
6955         PR tree-optimization/56787
6956         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
6957         from the list of data references.
6958         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
6959         clobbers.
6960         (vect_analyze_loop_operations): Likewise.
6961         (vect_transform_loop): Remove clobbers.
6963 2013-05-28  Martin Jambor  <mjambor@suse.cz>
6965         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
6966         and REALPART_EXPRs have scalar type.
6968 2013-05-28  Richard Biener  <rguenther@suse.de>
6970         PR tree-optimization/57411
6971         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
6972         virtual operands.
6973         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
6974         virtual operand propagation.
6976 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6978         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
6979         destination register for bmasksi_vis.
6980         (vector_init_bshuffle): Likewise.
6981         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
6983 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6985         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
6986         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
6987         mode if the instruction isn't available in the original mode.
6988         * config/sparc/sparc.opt (mfix-ut699): New option.
6989         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
6990         (divdf3): Turn into expander.
6991         (divdf3_nofix): New insn.
6992         (divdf3_fix): Likewise.
6993         (divsf3): Disable if -mfix-ut699.
6994         (sqrtdf2): Turn into expander.
6995         (sqrtdf2_nofix): New insn.
6996         (sqrtdf2_fix): Likewise.
6997         (sqrtsf2): Disable if -mfix-ut699.
6999 2013-05-27  Richard Biener  <rguenther@suse.de>
7001         PR middle-end/57412
7002         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7003         block for the new loop.
7005 2013-05-27  Richard Biener  <rguenther@suse.de>
7007         PR tree-optimization/57343
7008         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7009         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7010         (number_of_iterations_cond): Do not build the folded tree.
7012 2013-05-27  Richard Biener  <rguenther@suse.de>
7014         Revert
7015         PR middle-end/57381
7016         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7017         OEP_CONSTANT_ADDRESS_OF retained.
7019         PR tree-optimization/57417
7020         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7021         for unchanged base.
7022         (set_ssa_val_to): Compare addresses using
7023         get_addr_base_and_unit_offset.
7025 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7027         PR rtl-optimization/56833
7028         * postreload.c (move2add_record_mode): New function.
7029         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7030         (move2add_use_add2_insn): Use move2add_record_sym_value.
7031         (move2add_use_add3_insn): Likewise.
7032         (reload_cse_move2add): Use move2add_valid_value_p and
7033         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7034         affected regs by setting reg_mode to VOIDmode.
7035         (move2add_note_store): Don't pretend the inside of a SUBREG is
7036         the actual destination.  Invalidate single/leading registers by
7037         setting reg_mode to VOIDmode.
7038         Use move2add_record_sym_value, move2add_valid_value_p and
7039         move2add_record_mode.
7041 2013-05-27  Richard Biener  <rguenther@suse.de>
7043         PR tree-optimization/57396
7044         * tree-affine.c (double_int_constant_multiple_p): Properly
7045         return false for val == 0 and div != 0.
7047 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7049         * config/mips/mips.h: Use #elif in preprocessor conditions.
7051 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7053         PR target/53916
7054         * config/mips/constraints.md (kl): New constraint.
7055         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7056         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7057         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7058         from LO for MIPS16.
7059         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7061 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7063         PR target/55777
7064         * config/mips/mips.c (mips_can_inline_p): New function.
7065         (TARGET_CAN_INLINE_P): Define.
7067 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7069         * sched-int.h (ds_t, dw_t): Make unsigned int.
7070         Fix documentation that describes how all the ds_t bits are used.
7071         Reserve the last bit for delayed-branch scheduling.
7072         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7073         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7074         (gen_dep_weak_1): Remove prototype.
7075         * sched-deps.c (get_dep_weak_1): Make static.
7076         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7077         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7078         * doc/tm.texi: Regenerate.
7079         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7081 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7083         PR debug/56950
7084         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7086 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7087             Sandra Loosemore  <sandra@codesourcery.com>
7089         * config.gcc (powerpc-*): Allow native for with-cpu.
7091 2013-05-24  Jeff Law  <law@redhat.com>
7093         PR tree-optimization/57124
7094         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7095         conversion feeding a condition if the range has an overflow
7096         if -fstrict-overflow.  Add warnings for when we do make the
7097         transformation.
7099 2013-05-24  Dehao Chen  <dehao@google.com>
7101         * tree-cfg.c (locus_discrim_map): Fix the typo.
7102         (locus_discrim_hasher): Likewise.
7103         (locus_discrim_hasher::hash): Likewise.
7104         (locus_discrim_hasher::equal): Likewise.
7106 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7108         PR tree-optimization/57294
7109         * cgraph.h (ipa_record_stmt_references): Declare.
7110         * cgraphbuild.c (ipa_record_stmt_references): New function.
7111         (build_cgraph_edges): Use ipa_record_stmt_references.
7112         (rebuild_cgraph_edges): Likewise.
7113         (cgraph_rebuild_references): Likewise.
7114         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7115         associated with the old statement and build references from the
7116         newly built statements.
7117         * ipa-ref.c (ipa_remove_stmt_references): New function.
7118         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7120 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7122         * lra-constraints.c (emit_spill_move): Use smaller mode for
7123         mem-mem moves.
7124         (check_and_process_move): Consider mem-reg moves for secondary
7125         too.
7126         (curr_insn_transform): Don't lose insns emitted before for
7127         secondary memory moves.
7128         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7129         reg set up in the current insn.
7131 2013-05-24  Dehao Chen  <dehao@google.com>
7133         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7134         hash function.
7135         (locus_descrim_hasher::equal): Likewise.
7136         (build_gimple_cfg): New discriminator assignment algorithm.
7137         (make_edges): Likewise.
7138         (next_discriminator_for_locus): Likewise.
7139         (same_line_p): Likewise.
7140         (assign_discriminators): Likewise.
7141         (make_cond_expr_edges): Likewise.
7142         (make_gimple_switch_edges): Likewise.
7143         (make_goto_expr_edges): Likewise.
7144         (make_gimple_asm_edges): Likewise.
7146 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7148         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7149         X format specifier to only display bottom 16 bits.
7150         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7151         immediate to match for operand 2, since it will be masked.
7153 2013-05-24  Richard Biener  <rguenther@suse.de>
7155         PR tree-optimization/57287
7156         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7157         all SSA names that occur in abnormal PHIs.
7159 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7161         PR tree-ssa/57385
7162         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7163         that index is not negative.
7165 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7167         PR rtl-optimization/55177
7168         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7169         (simplify_byte_swapping_operation): New.
7170         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7171         (simplify_relational_operation_1): Deal with BSWAP.
7173 2013-05-23  Richard Henderson  <rth@redhat.com>
7175         PR target/56742
7176         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7177         (ix86_reorg): Call it.
7179 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7181         PR target/57379
7182         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7183         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7184         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7186 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7188         PR debug/57351
7189         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7191 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7192             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7194         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7195         * config/aarch64/constraints.md (Usa): Remove.
7196         * doc/md.texi (AArch64 Usa): Remove.
7198 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7199             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7201         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7202         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7203         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7204         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7206 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7208         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7209         instruction (AdvSIMD).
7210         * config/aarch64/aarch64-builtins.c
7211         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7212         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7214 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7216         PR middle-end/57347
7217         * tree.h (contains_bitfld_component_ref_p): Declare.
7218         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7219         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7220         caller.
7221         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7222         not access a bit-field.  Assert all final offsets are byte-aligned.
7224 2013-05-23  Richard Biener  <rguenther@suse.de>
7226         PR tree-optimization/57380
7227         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7228         least one invariant or re-used load.
7229         * passes.c (init_optimization_passes): Move pass_phiprop before
7230         pass_forwprop.
7232 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7234         * config/aarch64/aarch64-simd.md
7235         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7237 2013-05-23  Richard Biener  <rguenther@suse.de>
7239         PR middle-end/57381
7240         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7241         OEP_CONSTANT_ADDRESS_OF retained.
7243 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7245         PR middle-end/57344
7246         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7247         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7249 2013-05-23  Richard Biener  <rguenther@suse.de>
7251         PR rtl-optimization/57341
7252         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7253         instead of true_dependence.
7255 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7257         * bb-reorder.c (branch_threshold): Make const.
7258         (exec_threshold): Ditto.
7260 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7261             Pat Haugen <pthaugen@us.ibm.com>
7262             Peter Bergner <bergner@vnet.ibm.com>
7264         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7265         documentation for the power8 crypto builtins.
7267         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7269         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7270         macros for defining power8 builtin functions.
7271         (BU_P8V_AV_2): Likewise.
7272         (BU_P8V_AV_P): Likewise.
7273         (BU_P8V_VSX_1): Likewise.
7274         (BU_P8V_OVERLOAD_1): Likewise.
7275         (BU_P8V_OVERLOAD_2): Likewise.
7276         (BU_CRYPTO_1): Likewise.
7277         (BU_CRYPTO_2): Likewise.
7278         (BU_CRYPTO_3): Likewise.
7279         (BU_CRYPTO_OVERLOAD_1): Likewise.
7280         (BU_CRYPTO_OVERLOAD_2): Likewise.
7281         (XSCVSPDP): Fix typo, point to the correct instruction.
7282         (VCIPHER): Add power8 crypto builtins.
7283         (VCIPHERLAST): Likewise.
7284         (VNCIPHER): Likewise.
7285         (VNCIPHERLAST): Likewise.
7286         (VPMSUMB): Likewise.
7287         (VPMSUMH): Likewise.
7288         (VPMSUMW): Likewise.
7289         (VPERMXOR_V2DI): Likewise.
7290         (VPERMXOR_V4SI: Likewise.
7291         (VPERMXOR_V8HI: Likewise.
7292         (VPERMXOR_V16QI: Likewise.
7293         (VSHASIGMAW): Likewise.
7294         (VSHASIGMAD): Likewise.
7295         (VPMSUM): Likewise.
7296         (VPERMXOR): Likewise.
7297         (VSHASIGMA): Likewise.
7299         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7300         __CRYPTO__ if the crypto instructions are available.
7301         (altivec_overloaded_builtins): Add support for overloaded power8
7302         builtins.
7304         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7305         support for power8 crypto builtins.
7306         (builtin_function_type): Likewise.
7307         (altivec_init_builtins): Add support for builtins that take vector
7308         long long (V2DI) arguments.
7310         * config/rs6000/crypto.md: New file, define power8 crypto
7311         instructions.
7313 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7314             Pat Haugen <pthaugen@us.ibm.com>
7315             Peter Bergner <bergner@vnet.ibm.com>
7317         * doc/invoke.texi (Option Summary): Add power8 options.
7318         (RS/6000 and PowerPC Options): Likewise.
7320         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7321         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7322         wm, wn, wr documentation.
7324         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7325         registers if direct move instructions are enabled.
7326         (wn): New constraint for no registers.
7327         (wq): New constraint for quad word even GPR registers.
7328         (wr): New constraint if 64-bit instructions are enabled.
7329         (wv): New constraint if power8 vector instructions are enabled.
7330         (wQ): New constraint for quad word memory locations.
7332         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7333         constraint for 0..15 for crypto instructions.
7334         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7335         as GPR and floating point registers.
7336         (int_reg_operand): New predicate to match only GPR registers.
7337         (base_reg_operand): New predicate to match base registers.
7338         (quad_int_reg_operand): New predicate to match even GPR registers
7339         for quad memory operations.
7340         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7341         operations in both GPR and VSX registers.
7342         (quad_memory_operand): New predicate for quad memory operations.
7343         (reg_or_indexed_operand): New predicate for direct move support.
7345         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7346         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7347         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7348         (POWERPC_MASKS): Add power8 options.
7349         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7350         various options.
7352         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7353         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7355         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7356         (-mpower8-fusion): New power8 options.
7357         (-mpower8-fusion-sign): Likewise.
7358         (-mpower8-vector): Likewise.
7359         (-mcrypto): Likewise.
7360         (-mdirect-move): Likewise.
7361         (-mquad-memory): Likewise.
7363         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7364         power8.
7365         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7366         registers.
7367         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7368         (rs6000_debug_vector_unit): Add p8_vector.
7369         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7370         definitions.  Also print fusion state.
7371         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7372         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7373         (rs6000_option_override_internal): Add support for power8.
7374         (rs6000_common_init_builtins): Add debugging for skipped builtins
7375         if -mdebug=builtin.
7376         (rs6000_adjust_cost): Add power8 support.
7377         (rs6000_issue_rate): Likewise.
7378         (insn_must_be_first_in_group): Likewise.
7379         (insn_must_be_last_in_group): Likewise.
7380         (force_new_group): Likewise.
7381         (rs6000_register_move_cost): Likewise.
7382         (rs6000_opt_masks): Likewise.
7384         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7385         power8 capable assembler, default to power7 options.
7386         (TARGET_DIRECT_MOVE): Likewise.
7387         (TARGET_CRYPTO): Likewise.
7388         (TARGET_P8_VECTOR): Likewise.
7389         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7390         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7391         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7392         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7393         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7394         (TARGET_XSCVDPSPN): Likewise.
7395         (TARGET_XSCVSPDPN): Likewsie.
7396         (TARGET_SYNC_HI_QI): Likewise.
7397         (TARGET_SYNC_TI): Likewise.
7398         (MASK_CRYPTO): Likewise.
7399         (MASK_DIRECT_MOVE): Likewise.
7400         (MASK_P8_FUSION): Likewise.
7401         (MASK_P8_VECTOR): Likewise.
7402         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7403         temporary used by some of the direct move instructions to get two FP
7404         temporary registers does not force creation of a stack frame.
7405         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7406         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7407         that any VSX registers are tieable, even if they are also an
7408         Altivec vector mode.
7409         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7410         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7411         (RS6000_BTM_CRYPTO): Likewise.
7412         (RS6000_BTM_COMMON): Likewise.
7414         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7415         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7416         (enum rs6000_vector): Add power8 vector support.
7418 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7420         PR target/19599
7421         PR target/57340
7422         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7423         (any_sibcall_could_use_r3): this and handle indirect calls.
7424         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7426 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7428         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7430 2013-05-22  Richard Biener  <rguenther@suse.de>
7432         PR middle-end/57349
7433         * profile.c (branch_prob): Do not split blocks that are
7434         abnormally receiving from ECF_RETURNS_TWICE functions.
7436 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7438         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7439         address mode.  Move pointer mode initialization to the same place.
7441 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7443         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7444         while it has any effect.
7446 2013-05-21  Easwaran Raman  <eraman@google.com>
7448         PR tree-optimization/57322
7449         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7450         UID of the statement added to the BB to be 1.
7452 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7454         PR tree-optimization/57331
7455         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7456         of conversion from pointer type to integral type with integer.
7458 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7460         PR lto/57289
7461         * ipa-prop.c (ipa_read_node_info): Process param_used and
7462         controlled_uses in the same order as when writing.
7464 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7466         PR plugins/56754
7467         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7469 2013-05-21  Richard Biener  <rguenther@suse.de>
7471         PR tree-optimization/57318
7472         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7473         estimate stmts with side-effects as likely eliminated.
7475 2013-05-21  Richard Biener  <rguenther@suse.de>
7477         PR tree-optimization/57330
7478         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7479         preserve the call stmts fntype.
7481 2013-05-21  Richard Biener  <rguenther@suse.de>
7483         PR tree-optimization/57303
7484         * tree-ssa-sink.c (statement_sink_location): Improve killing
7485         stmt detection and properly handle self-assignments.
7487 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7489         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7490         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7491         registers. Set register size out of the PARALLEL loop.
7493 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7495         PR target/56547
7496         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7497         (*fmasf4, *fmasf4_media): New insns.
7499 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7501         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7502         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7503         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7504         (mips_idiv_insns): Update the comments to say that the returned
7505         instruction counts are in units of BASE_INSN_LENGTH.
7506         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7507         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7508         using 2 rather than 4 as the length of indirect MIPS16 and
7509         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7510         length of a NOP.  Don't divide MIPS16 lengths by 2.
7511         (mips16_split_long_branches): Assume a branch is long if the
7512         length is greater than 4 rather than 8.
7513         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7514         rather than multiplying them by 2.  Multiply instruction counts
7515         by BASE_INSN_LENGTH rather than 4.
7516         (*jump_mips16, tls_get_tp_mips16_<mode>)
7517         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7519 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7521         * config/mips/mips.md (extended_mips16): Remove branch case.
7522         (length): Remove duplicated extended_mips16 test.
7524 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7526         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7528 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7530         * recog.h (Recog_data): Rename to...
7531         (recog_data_d): ...this.
7532         (recog_data): Update accordingly.
7533         * recog.c (recog_data): Likewise.
7534         * reload.c (save_recog_data): Likewise.
7535         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7536         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7538 2013-05-17  Julian Brown  <julian@codesourcery.com>
7540         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7541         found in a REG_EQUAL note, invalidate it.
7543 2013-05-17  Easwaran Raman  <eraman@google.com>
7545         * tree-ssa-reassoc.c (find_insert_point): New function.
7546         (insert_stmt_after): Likewise.
7547         (get_def_stmt): Likewise.
7548         (ensure_ops_are_available): Likewise.
7549         (not_dominated_by): Likewise.
7550         (rewrite_expr_tree): Do not move statements beyond what is
7551         necessary. Remove call to swap_ops_for_binary_stmt...
7552         (reassociate_bb): ... and move it here.
7553         (build_and_add_sum): Assign UIDs for new statements.
7554         (linearize_expr): Likewise.
7555         (do_reassoc): Renumber gimple statement UIDs.
7557 2013-05-17  Jan Hubicka  <jh@suse.cz>
7559         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7560         weakrefs.
7561         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7562         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7563         weakrefs.
7564         (output_weakrefs): Update.
7566 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
7567             Martin Jambor  <mjambor@suse.cz>
7569         PR middle-end/57276
7570         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7571         value that corresponds to the given aggval is found in values vector.
7573 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
7575         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7576         sse, sse2, sse3, ssse3 and sse4a flags to options.
7578 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
7580         * gengtype-state.c: (s_expr_writer): New class, to handle
7581         prettifying of output layout of s-expressions.
7582         (state_writer): New class, to write out gtype.state.
7583         (state_written_type_count): Move this variable into member data of
7584         state_writer.
7585         (s_expr_writer::s_expr_writer): New code: constructor for new class
7586         (state_writer::state_writer(): ditto
7587         (s_expr_writer::write_new_line): New function
7588         (s_expr_writer::write_any_indent): ditto
7589         (s_expr_writer::begin_s_expr): ditto
7590         (s_expr_writer::end_s_expr): ditto
7591         (write_state_fileloc): convert to method of state_writer...
7592         (state_writer:: write_state_fileloc): ...and use methods of
7593         s_expr_writer to write indentation into the gtype.state output file
7594         to visually represent the hierarchical structure of the list
7595         structures
7596         (write_state_fields): ditto, renaming to...
7597         (state_writer::write_state_fields)
7598         (write_state_a_string): ditto, renaming to...
7599         (state_writer::write_state_a_string)
7600         (write_state_string_option): ditto, renaming to...
7601         (state_writer::write_state_string_option)
7602         (write_state_type_option): ditto, renaming to...
7603         (state_writer::write_state_type_option)
7604         (write_state_nested_option): ditto, renaming to...
7605         (state_writer::write_state_nested_option)
7606         (write_state_option): ditto, renaming to...
7607         (state_writer::write_state_option)
7608         (write_state_options): ditto, renaming to...
7609         (state_writer::write_state_options)
7610         (write_state_lang_bitmap): ditto, renaming to...
7611         (state_writer::write_state_lang_bitmap)
7612         (write_state_version): ditto, renaming to...
7613         (state_writer::write_state_version)
7614         (write_state_scalar_type): ditto, renaming to...
7615         (state_writer::write_state_scalar_type)
7616         (write_state_string_type): ditto, renaming to...
7617         (state_writer::write_state_string_type)
7618         (write_state_undefined_type): ditto, renaming to...
7619         (state_writer::write_state_undefined_type)
7620         (write_state_struct_union_type): ditto, renaming to...
7621         (state_writer::write_state_struct_union_type)
7622         (write_state_struct_type): ditto, renaming to...
7623         (state_writer::write_state_struct_type)
7624         (write_state_user_struct_type): ditto, renaming to...
7625         (state_writer::write_state_user_struct_type)
7626         (write_state_lang_struct_type): ditto, renaming to...
7627         (state_writer::write_state_lang_struct_type)
7628         (write_state_param_struct_type): ditto, renaming to...
7629         (state_writer::write_state_param_struct_type)
7630         (write_state_pointer_type): ditto, renaming to...
7631         (state_writer::write_state_pointer_type)
7632         (write_state_array_type): ditto, renaming to...
7633         (state_writer::write_state_array_type)
7634         (write_state_gc_used): ditto, renaming to...
7635         (state_writer::write_state_gc_used)
7636         (write_state_common_type_content): ditto, renaming to...
7637         (state_writer::write_state_common_type_content)
7638         (write_state_type): ditto, renaming to...
7639         (state_writer::write_state_type)
7640         (write_state_pair_list): ditto, renaming to...
7641         (state_writer::write_state_pair_list)
7642         (write_state_pair): ditto, renaming to...
7643         (state_writer::write_state_pair)
7644         (write_state_typedefs): ditto, renaming to...
7645         (state_writer::write_state_typedefs)
7646         (write_state_structures): ditto, renaming to...
7647         (state_writer::write_state_structures)
7648         (write_state_param_structs): ditto, renaming to...
7649         (state_writer::write_state_param_structs)
7650         (write_state_variables): ditto, renaming to...
7651         (state_writer::write_state_variables)
7652         (write_state_srcdir): ditto, renaming to...
7653         (state_writer::write_state_srcdir)
7654         (write_state_files_list): ditto, renaming to...
7655         (state_writer::write_state_files_list)
7656         (write_state_languages): ditto, renaming to...
7657         (state_writer::write_state_languages)
7658         (write_state): create a state_writer instance and use it when
7659         writing out the state file
7661 2013-05-17  Mike Stump  <mikestump@comcast.net>
7663         PR rtl-optimization/57304
7664         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7665         accessing DF_REF_REAL_LOC.
7667 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7669         PR rtl-optimization/57281
7670         PR rtl-optimization/57300
7671         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7672         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7673         what the other splitter did if the registers are dead.
7675 2013-05-17  Richard Biener  <rguenther@suse.de>
7677         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7678         MEM_REF offsets.
7680 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7682         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7683         linking.
7685 2013-05-17  Marek Polacek  <polacek@redhat.com>
7687         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7688         length when doing non-zero store of storing '\0' to '\0'.
7690 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7692         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7693         vect_external_def oprnd1 with loop_vinfo, try to emit
7694         optional cast, negation and and stmts on the loop preheader
7695         edge instead of into the pattern def seq.
7697         PR tree-optimization/57051
7698         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7699         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7701 2013-05-16  Nick Clifton  <nickc@redhat.com>
7703         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7704         (rl78_is_naked_func): New function.
7705         (rl78_expand_prologue): Skip prologue generation for naked functions.
7706         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
7707         * doc/extend.texi (naked): Add RL78 to the list of processors
7708         that supports this attribute.
7710 2013-05-16  Jeff Law  <law@redhat.com>
7712         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7714 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7716         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7717         cache parameters using detect_caches_amd also for CYRIX,
7718         NSC and TM2 signatures.
7720 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7721             Dzianis Kahanovich  <mahatma@eu.by>
7723         PR target/45359
7724         PR target/46396
7725         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7726         VIA/Centaur processors and determine their cache parameters
7727         using detect_caches_amd.
7729 2013-05-16  Teresa Johnson  <tejohnson@google.com>
7731         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7732         (rtl_verify_edges): New function.
7733         (rtl_verify_bb_insns): Ditto.
7734         (rtl_verify_bb_pointers): Ditto.
7735         (rtl_verify_bb_insn_chain): Ditto.
7736         (rtl_verify_fallthru): Ditto.
7737         (rtl_verify_bb_layout): Ditto.
7738         (rtl_verify_flow_info_1): Outline checks into new functions.
7739         (rtl_verify_flow_info): Ditto.
7741 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
7743         * cfghooks.c (copy_bbs): Add update_dominance argument.
7744         * cfghooks.h (copy_bbs): Update prototype.
7745         * tree-cfg.c (gimple_duplicate_sese_region):
7746         Add update_dominance argument.
7747         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7748         * tree-ssa-loop-ch.c (copy_loop_headers): Update
7749         gimple_duplicate_sese_region call.
7750         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7751         Update copy_bbs call.
7752         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7753         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7755 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7757         * tree-vectorizer.h (NUM_PATTERNS): Increment.
7758         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7759         vect_recog_rotate_pattern.
7760         (vect_recog_rotate_pattern): New function.
7762 2013-05-16  Jason Merrill  <jason@redhat.com>
7764         * Makefile.in (LLINKER): New variable.
7765         (mostlyclean): Remove link mutex.
7766         * configure.ac: Handle --enable-link-mutex.
7767         * lock-and-run.sh: New script.
7769 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7771         PR target/19599
7772         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7773         for NULL decl.
7775 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7777         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7779 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
7781         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7782         * config/arm/arm.c (next_consecutive_mem): New function.
7783         (gen_movmem_ldrd_strd): Likewise.
7784         * config/arm/arm.md (movmemqi): Update condition and code.
7785         (unaligned_loaddi, unaligned_storedi): New patterns.
7787 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7789         * config.gcc: Obsolete *-*-solaris2.9*.
7790         * doc/install.texi (Specific, *-*-solaris2*): Document it.
7792 2013-05-16  Richard Biener  <rguenther@suse.de>
7794         * passes.c (init_optimization_passes): Move pass_parallelize_loops
7795         earlier, after GRAPHITE transforms and IV canonicalization.
7797 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7799         * omp-low.c (extract_omp_for_data): For collapsed loops,
7800         if at least one of the loops is known at compile time to
7801         iterate zero times, set count to 0.
7802         (expand_omp_regimplify_p): New function.
7803         (expand_omp_for_generic): For collapsed loops, if at least
7804         one of the loops isn't known to iterate at least once,
7805         add runtime check with setting count to 0.
7806         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7807         For unsigned types if it isn't known at compile time that
7808         the loop will iterate at least once, add runtime check to bypass
7809         the whole loop if initial condition isn't true.
7811 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
7813         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7815 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
7817         PR middle-end/57286
7818         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7819         transformations to avoid an infinite loop.
7821 2013-05-16  Marek Polacek  <polacek@redhat.com>
7823         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7825 2013-05-15  Leif Ekblad  <leif@rdos.net>
7827         * config/i386/i386.c (ix86_decompose_address): Use
7828         DEFAULT_TLS_SEG_REG to access TLS segment register.
7829         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7830         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7831         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7833 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
7835         PR target/57260
7836         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7837         sibling calls to functions that would normally be lazily bound,
7838         unless $gp is call-clobbered.
7840 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
7842         * config/i386/i386.c (ix86_option_override_internal): Update
7843         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
7844         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7845         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
7846         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7847         of TARGET_3DNOW.
7848         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7850 2013-05-15  Andreas Schwab  <schwab@suse.de>
7852         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7853         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
7854         third operand.
7856 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7858         * loop-unroll.c (report_unroll_peel): Check decision before
7859         emitting unroll/peel message.
7861 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7863         * function.h (has_bb_partition): New rtl_data flag.
7864         (bb_reorder_complete): Ditto.
7865         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7866         instead of flag_reorder_blocks_and_partition.
7867         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7868         with some enhancements.
7869         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7870         * bb-reorder.c (connect_traces): Check for has_bb_partition
7871         instead of flag_reorder_blocks_and_partition.
7872         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7873         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7874         verify_hot_cold_block_grouping.
7875         (partition_hot_cold_basic_blocks): Set has_bb_partition.
7877 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7879         PR target/19599
7880         * config/arm/predicates.md (call_insn_operand): New predicate.
7881         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
7882         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7883         if insn is not a tail call.
7884         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
7885         registers.
7886         * config/arm/arm.h (enum reg_class): New caller save register class.
7887         (REG_CLASS_NAMES): Likewise.
7888         (REG_CLASS_CONTENTS): Likewise.
7889         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
7890         without decls.
7892 2013-05-15  Richard Biener  <rguenther@suse.de>
7894         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
7895         of MSG_OPTIMIZED_LOCATIONS.
7896         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
7897         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
7898         message.
7899         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
7900         of MSG_OPTIMIZED_LOCATIONS.
7901         (execute_vect_slp): Likewise.
7902         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
7903         (vect_create_cond_for_alias_checks): Likewise.
7904         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
7905         (vect_recog_widen_mult_pattern): Likewise.
7906         (vect_recog_widen_sum_pattern): Likewise.
7907         (vect_recog_over_widening_pattern): Likewise.
7908         (vect_recog_widen_shift_pattern): Likewise.
7909         (vect_recog_vector_vector_shift_pattern): Likewise.
7910         (vect_recog_divmod_pattern): Likewise.
7911         (vect_recog_mixed_size_cond_pattern): Likewise.
7912         (vect_recog_bool_pattern): Likewise.
7913         (vect_pattern_recog_1): Likewise.
7915 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7917         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
7918         non-functions to builtin_unreachable.
7919         * ipa-inline-transform.c (inline_call): Do not assert estimates were
7920         correct when new direct edges were discovered.
7922 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7924         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
7925         header, print symbol order instead of node uid, print more information
7926         about indirect edge targets.
7927         (ipa_make_edge_direct_to_target): Print symbol order instead of node
7928         uids.
7929         (ipa_make_edge_direct_to_target): Likewise.
7930         (remove_described_reference): Likewise.
7931         (propagate_controlled_uses): Likewise.
7932         (ipa_print_node_params): Also print symbol order.
7933         (ipcp_transform_function): Print symbol order instead of node uids.
7934         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
7935         (cgraph_get_create_real_symbol_node): Likewise.
7936         * ipa-cp.c (print_lattice): Likewise.
7937         (print_all_lattices): Likewise.
7938         (determine_versionability): Likewise.
7939         (initialize_node_lattices): Likewise.
7940         (estimate_local_effects): Likewise.
7941         (update_profiling_info): Likewise.
7942         (create_specialized_node): Likewise.
7943         (perhaps_add_new_callers): Likewise.
7944         (decide_about_value): Likewise.
7945         (decide_whether_version_node): Likewise.
7946         (identify_dead_nodes): Likewise.
7947         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
7948         (dump_inline_summary): Likewise.
7949         (estimate_node_size_and_time): Likewise.
7950         (inline_analyze_function): Likewise.
7951         * ipa-inline.c (report_inline_failed_reason): Likewise.
7952         (want_early_inline_function_p): Likewise.
7953         (edge_badness): Likewise.
7954         (update_edge_key): Likewise.
7955         (inline_small_functions): Likewise.  Add dumping of order to two other
7956         dumps.
7957         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
7958         instead of node uids.
7959         (propagate_pure_const): Likewise.
7960         (propagate_pure_const): Likewise.
7961         * ipa-utils.c (dump_cgraph_node_set): Likewise.
7962         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
7963         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
7964         of node uids.
7965         * tree-pretty-print.c (dump_function_header): Likewise.
7966         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
7967         Print symbol order instead of node uids.
7969 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7971         * config/s390/s390.c (s390_register_move_cost): Don't impose the
7972         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
7974 2013-05-15  Richard Biener  <rguenther@suse.de>
7976         PR tree-optimization/57275
7977         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
7978         return value for fail to do runtime alias checks for gather loads.
7980 2013-05-15  Jan Hubicka  <jh@suse.cz>
7982         PR lto/57038
7983         PR lto/47375
7984         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
7985         weakrefs are not external.
7986         (lto_symtab_merge_decls): Fix thinko when dealing with
7987         non-lto_symtab decls.
7988         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
7989         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
7990         * varpool.c (dump_varpool_node): Dump more flags.
7992 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7994         * config/i386/i386.c (processor_alias_table): Add instruction
7995         FSGSBASE for AMD bdver3 architecture.
7997 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
7999         * tree.c (warn_deprecated_use): Print file:line using locus color.
8000         * diagnostic.c (diagnostic_report_current_module): Print file:line
8001         and file:line:column using locus color.
8003 2013-05-14  Mike Stump  <mikestump@comcast.net>
8005         * gdbinit.in: Add __null.
8007 2013-05-14  Mike Stump  <mikestump@comcast.net>
8009         * recog.h: Rename struct recog_data to Recog_data.
8010         * recog.c: Likewise.
8011         * reload.c (can_reload_into): Likewise.
8012         * config/picochip/picochip.c: Likewise.
8014 2013-05-14  Mike Stump  <mikestump@comcast.net>
8016         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8018 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8020         * resource.h (struct resources): Remove unch_memory member.
8021         (CLEAR_RESOURCE): Don't clear unch_memory.
8022         * resource.c (mark_referenced_resources): Don't set it.
8023         (mark_set_resources): Likewise.
8024         (mark_target_live_regs): Don't clear it.
8025         (init_resource_info): Likewise.
8026         * reorg.c (resource_conflicts_p): Don't compare it.
8027         (redundant_insn): Don't set it.
8029         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8030         Remove prototypes.
8031         * emit-rtl.c (next_label): Remove unused function.
8032         (skip_consecutive_labels, link_cc0_insns): Move to ...
8033         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8034         only place where these functions are used, and make them static.
8036 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8038         * fold-const.c (fold_negate_expr): Handle vectors.
8039         (fold_truth_not_expr): Make it static.
8040         (fold_invert_truthvalue): New static function.
8041         (invert_truthvalue_loc): Handle vectors. Do not call
8042         fold_truth_not_expr directly.
8043         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8044         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8045         (fold_comparison): Handle vector constants.
8046         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8047         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8048         * tree.h (fold_truth_not_expr): Remove declaration.
8050 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8052         * config/aarch64/aarch64-simd.md
8053         (aarch64_vcond_internal<mode>): Rename to...
8054         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8055         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8056         float modes. Clarify all iterator modes.
8057         (vcond<mode><mode>): Use new name for vcond expanders.
8058         (vcond<v_cmp_result><mode>): Likewise.
8059         (vcondu<mode><mode>: Likewise.
8060         * config/aarch64/iterators.md (VDQF_COND): New.
8062 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8064         PR bootstrap/57266
8065         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8066         variable for the shift amount. Check that we shift by non-negative
8067         amounts.
8069 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8071         PR target/42017
8072         * config/arm/arm.h (EPILOGUE_USES): Only return true
8073         for LR_REGNUM after epilogue_completed.
8075 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8077         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8078         is error_mark_node.
8080 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8082         PR target/57261
8083         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8084         and Solaris 11+/x86 with gld.
8085         * configure: Regenerate.
8087 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8089         * expmed.c (expand_shift_1): Canonicalize rotates by
8090         constant bitsize / 2 to bitsize - 1.
8091         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8092         case ROTATERT>: Likewise.
8094         Revert:
8095         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8097         * config/i386/i386.md (rotateinv): New code attr.
8098         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8099         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8100         roll $31, %eax, etc.
8102 2013-05-14  Richard Biener  <rguenther@suse.de>
8104         PR middle-end/57235
8105         * tree-eh.c (sink_clobbers): Give up for successors with
8106         multiple predecessors and no virtual uses.
8108 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8110         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8111         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8113 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8115         PR middle-end/57251
8116         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8117         the case when both op0 and op1 have VOIDmode.
8119 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8121         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8122         in multiply-accumulate mode.
8124 2013-05-13  Guozhi Wei  <carrot@google.com>
8126         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8128 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8130         PR target/56975
8131         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8132         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8133         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8134         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8135         (ix86_expand_prologue): Likewise.
8136         (ix86_expand_split_stack_prologue): Likewise.
8137         (legitimate_pic_address_disp_p): Likewise.
8138         (legitimize_pic_address): Likewise.
8139         (legitimize_tls_address): Likewise.
8140         (legitimize_pe_coff_symbol): Likewise.
8141         (output_pic_addr_const): Likewise.
8142         (construct_plt_address): Likewise.
8143         (ix86_expand_call): Likewise.
8144         (x86_output_mi_thunk): Likewise.
8145         (x86_function_profiler): Likewise.
8147 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8149         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8150         similar switch cases.
8151         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8152         (aarch64_simd_mov_to_<mode>low): Delete.
8153         (aarch64_simd_mov_to_<mode>high): Delete.
8154         (move_lo_quad_<mode>): Add w<-r alternative.
8155         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8156         (aarch64_simd_mov_from_*): Update type attribute.
8157         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8158         statement.
8160 2013-05-13  Jan Hubicka  <jh@suse.cz>
8162         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8163         * config/i386/i386.c (ix86_compute_frame_layout,
8164         ix86_expand_epilogue, emit_i387_cw_initialization,
8165         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8166         ix86_local_alignment): Fix use of size/speed predicates.
8168 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8170         PR tree-optimization/45216
8171         PR tree-optimization/57157
8172         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8173         the (-Y) & (B - 1) variant if OP is |.
8174         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8175         return shifted.  Use (-op1) & (prec - 1) as other_amount
8176         instead of prec - op1.
8178 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8180         PR middle-end/42371
8181         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8182         (ipa_constant_data): New type.
8183         (ipa_jump_func): Use ipa_constant_data to hold information about
8184         constant jump functions.
8185         (ipa_get_jf_constant): Adjust to jump function type changes.
8186         (ipa_get_jf_constant_rdesc): New function.
8187         (ipa_param_descriptor): New field controlled_uses.
8188         (ipa_get_controlled_uses): New function.
8189         (ipa_set_controlled_uses): Likewise.
8190         * ipa-ref.h (ipa_find_reference): Declare.
8191         * ipa-prop.c (ipa_cst_ref_desc): New type.
8192         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8193         changes.
8194         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8195         New parameter cs.  Adjust all callers.
8196         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8197         (remove_described_reference): New function.
8198         (jfunc_rdesc_usable): Likewise.
8199         (try_make_edge_direct_simple_call): Decrement controlled use count,
8200         attempt to remove reference if it hits zero.
8201         (combine_controlled_uses_counters): New function.
8202         (propagate_controlled_uses): Likewise.
8203         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8204         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8205         (ipa_print_node_params): Print described use counter.
8206         (ipa_write_jump_function): Adjust to jump function type changes.
8207         (ipa_read_jump_function): New parameter CS, pass it to
8208         ipa_set_jf_constant.  Adjust caller.
8209         (ipa_write_node_info): Stream controlled use count
8210         (ipa_read_node_info): Likewise.
8211         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8212         asserting.
8213         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8214         count.  Remove cloning-added reference if it reaches zero.
8215         * ipa-ref.c (ipa_find_reference): New function.
8217 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8219         * config/i386/i386.c (processor_target_table): Modified default
8220         alignment values for AMD BD and BT architectures.
8222 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8224         * tree-vect-generic.c (uniform_vector_p): Move ...
8225         * tree.c (uniform_vector_p): ... here.
8226         * tree.h (uniform_vector_p): Declare it.
8227         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8228         into a scalar.
8230 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8232         PR tree-optimization/57230
8233         * tree-ssa-strlen.c (handle_char_store): Record length for
8234         array store from STRING_CST.
8236         PR tree-optimization/57230
8237         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8238         check.
8240 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8242         * config/epiphany/epiphany.c (epiphany_init): Check size of
8243         NUM_MODES_FOR_MODE_SWITCHING.
8244         (epiphany_expand_prologue):
8245         Remove CONFIG_REGNUM initial value handling code.
8246         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8247         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8248         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8249         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8250         Don't return 1 for FP_MODE_NONE.
8251         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8252         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8253         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8254         * config/epiphany/epiphany.md (save_config): New pattern.
8256 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8258         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8260 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8262         * config/i386/i386.md (memory): Handle sseishft1.
8263         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8264         (*vec_extractv2di_1): Ditto.
8266 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8268         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8269         saved registers.
8271 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8273         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8274         Add mthumb/march=armv7-a multilib.
8275         Add mthumb/march=armv7-r multilib.
8276         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8278 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8280         * config/v850/t-rtems: Add more multilibs.
8282 2013-05-10  Richard Biener  <rguenther@suse.de>
8284         PR tree-optimization/57214
8285         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8286         not propagate from SSA names that occur in abnormal PHI nodes.
8288 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8290         * stor-layout.c (element_precision): New function.
8291         * machmode.h (element_precision): Declare it.
8292         * tree.c (build_minus_one_cst): New function.
8293         (element_precision): Likewise.
8294         * tree.h (build_minus_one_cst): Declare new function.
8295         (element_precision): Likewise.
8296         * fold-const.c (operand_equal_p): Use element_precision.
8297         (fold_binary_loc): Handle vector types.
8298         * convert.c (convert_to_integer): Use element_precision.
8299         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8300         separately.
8302 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8304         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8305         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8306         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8307         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8308         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8309         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8310         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8311         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8312         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8313         (Uuw8): New constraints.
8314         (Usb4): Move into alphabetical order.
8315         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8316         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8317         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8318         previously unnamed patterns.
8319         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8320         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8321         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8322         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8323         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8324         of specific lengths.
8326 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8328         * config/i386/i386.md (rotateinv): New code attr.
8329         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8330         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8331         roll $31, %eax, etc.
8333         PR tree-optimization/45216
8334         PR tree-optimization/57157
8335         * tree-ssa-forwprop.c (simplify_rotate): New function.
8336         (ssa_forward_propagate_and_combine): Call it.
8338 2013-05-10  Richard Biener  <rguenther@suse.de>
8340         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8341         disable peeling when we version for aliasing.
8342         (vector_alignment_reachable_p): Honor explicit user alignment.
8343         (vect_supportable_dr_alignment): Likewise.
8344         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8345         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8346         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8347         then peeling to arrange for the cost-model check to come first.
8349 2013-05-10  Alan Modra  <amodra@gmail.com>
8351         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8352         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8353         * configure: Regenerate.
8355 2013-05-10  Alan Modra  <amodra@gmail.com>
8357         PR target/55033
8358         * varasm.c (default_elf_select_section): Move !DECL_P check..
8359         (get_named_section): ..to here before calling get_section_name.
8360         Adjust assertion.
8361         (default_section_type_flags): Add DECL_P check.
8362         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8363         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8365 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8367         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8368         When using gen_stack_adjust_str with a register offset, add a
8369         REG_FRAME_RELATED_EXPR note.
8371 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8373         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8374         (*vec_extractv4si_zext_mem): Ditto.
8375         (*vec_extractv2di): Add 0->x and x->x alternatives.
8376         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8377         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8379 2013-05-09  Jason Merrill  <jason@redhat.com>
8381         N3639 C++1y VLA support
8382         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8383         DECL_VALUE_EXPR.
8385         * tree.c (build_constructor_va): New.
8386         * tree.h: Declare it.
8388 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8390         PR lto/57084
8391         * gimple-fold.c (canonicalize_constructor_val): Call
8392         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8394 2013-05-09  Jan Hubicka  <jh@suse.cz>
8395             Richard Biener  <rguenther@suse.de>
8397         PR lto/54095
8398         * symtab.c (symtab_make_decl_local): Do not add private names.
8400 2013-05-09  Jan Hubicka  <jh@suse.cz>
8402         PR lto/54095
8403         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8404         (unlink_from_assembler_name_hash): Likewise.
8405         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8406         symtab_unregister_node, symtab_initialize_asm_name_hash,
8407         change_decl_assembler_name): Update.
8409 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8411         * config/aarch64/aarch64.md: New movtf split.
8412         (*movtf_aarch64): Update.
8413         (aarch64_movdi_tilow): Handle TF modes and rename to
8414         aarch64_movdi_<mode>low.
8415         (aarch64_movdi_tihigh): Handle TF modes and rename to
8416         aarch64_movdi_<mode>high
8417         (aarch64_movtihigh_di): Handle TF modes and rename to
8418         aarch64_mov<mode>high_di
8419         (aarch64_movtilow_di): Handle TF modes and rename to
8420         aarch64_mov<mode>low_di
8421         (aarch64_movtilow_tilow): Remove spurious whitespace.
8422         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8423         splits.
8424         (aarch64_print_operand): Update.
8426 2013-05-09  Alan Modra  <amodra@gmail.com>
8428         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8429         powerpc64le.
8430         * configure: Regenerate.
8432 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8434         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8435         splitter preparation statements.
8436         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8437         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8438         adjust_address_nv.
8440 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8442         * gimple-ssa-strength-reduction.c (count_candidates): Change
8443         return value to int.
8444         (analyze_candidates_and_replace): Change type of length to int.
8446 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8448         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8449         (*vec_extract<mode>): Use VI12_128 mode iterator.
8450         (*vec_extract<mode>_mem): Ditto.
8451         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8452         attribute.
8454 2013-05-08  Diego Novillo  <dnovillo@google.com>
8456         PR bootstrap/54659
8458         Revert:
8459         2012-08-17  Diego Novillo  <dnovillo@google.com>
8461         PR bootstrap/54281
8462         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8463         * config.in: Regenerate.
8464         * configure: Regenerate.
8465         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8467 2013-05-08  Jan Hubicka  <jh@suse.cz>
8469         PR lto/54095
8470         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8471         * cgraph.h (symtab_node_base): Add unique_name.
8472         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8473         input_overwrite_node, input_varpool_node): Stream unique_name.
8474         * cgraphclones.c (cgraph_create_virtual_clone,
8475         cgraph_function_versioning): Set unique_name.
8476         * ipa.c (function_and_variable_visibility): Set unique_name.
8478 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8480         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8481         (alloc_cand_and_find_basis): Restrict conditional candidate
8482         processing to CAND_MULTs.
8484 2013-05-08  Jan Hubicka  <jh@suse.cz>
8486         PR lto/54095
8487         lto-symtab.c (lto_symtab_symbol_p): New function.
8488         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8489         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8490         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8491         Skip static symbols.
8493 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8495         PR tree-optimization/57200
8496         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8497         Only call inform if the preceding warning_at returns true.
8499 2013-05-07  Han Shen  <shenhan@google.com>
8501         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8502         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8503         * common.opt (fstack-protector-strong): New option.
8504         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8505         * doc/invoke.texi (Optimization Options): Document
8506         "-fstack-protector-strong".
8507         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8509 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8511         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8513 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8515         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8516         it is not lhs of the insn.
8517         (setup_reg_equiv): Remove insn having equiv memory which it is not
8518         lhs of the insn.
8519         * lra-constraints.c (process_address): Try to improve generation
8520         code for address base + disp.
8521         (lra_constraints): Make correct the code for checking insn setting
8522         up backward equivalence.  Remove insn only if it is in the init
8523         insn list.
8524         * lra-eliminations.c (update_reg_eliminate): Change return value.
8525         (lra_eliminate): Use the result.
8527 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8529         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8530         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8531         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8532         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8533         register target operands.
8534         (*vec_extractv8hi_sse2): New pattern.
8535         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8536         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8537         (*vec_extract<mode>_mem): New insn and split pattern.
8539 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8541         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8542         (TARGET_ASAN_SHADOW_OFFSET): Define.
8543         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8544         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8546 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8548         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8549         (incr_vec_index): Return -1 if increment not found.
8550         (create_add_on_incoming_edge): Assert if increment not found.
8551         (record_increment): Limit number of increments recorded.
8552         (all_phi_incrs_profitable): Return false if an increment not found.
8553         (replace_profitable_candidates): Don't process increments that were
8554         not recorded.
8555         (analyze_candidates_and_replace): Limit size of incr_vec.
8557 2013-05-07  Richard Biener  <rguenther@suse.de>
8559         * calls.c (special_function_p): setjmp-like functions are leaf.
8560         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8561         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8563 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
8565         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8566         (aarch64_simd_mov<mode>): New expander.
8567         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8568         (aarch64_simd_mov_to_<mode>high): Likewise.
8569         (aarch64_simd_mov_from_<mode>low): Likewise.
8570         (aarch64_simd_mov_from_<mode>high): Likewise.
8571         (aarch64_dup_lane<mode>): Update.
8572         (aarch64_dup_lanedi): New instruction pattern.
8573         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8574         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8576 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8578         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8579         (replace_mult_candidate): Remove unnecessary argument; remove
8580         unnecessary parameter from call to introduce_cast_before_cand.
8581         (replace_unconditional_candidate): Remove unnecessary parameter
8582         from call to replace_mult_candidate.
8583         (replace_conditional_candidate): Likewise.
8584         (insert_initializers): Use make_temp_ssa_name.
8585         (introduce_cast_before_cand): Remove unnecessary argument; use
8586         make_temp_ssa_name.
8587         (replace_one_candidate): Remove unnecessary argument; remove
8588         unnecessary parameter from calls to introduce_cast_before_cand.
8589         (replace_profitable_candidates): Remove unnecessary parameters
8590         from calls to replace_one_candidate.
8592 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8594         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8595         phi def as possibly hiding a basis for a CAND_ADD whose operands
8596         have been commuted in the analysis.
8597         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8599 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
8601         * config/aarch64/aarch64.md
8602         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8603         shift value between 0-4.
8605 2013-05-07  Richard Biener  <rguenther@suse.de>
8607         * double-int.h (rshift): New overload.
8608         * double-int.c (rshift): New function.
8609         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8610         (create_reference_ops_from_ref): Remove.
8611         (vn_reference_insert): Use shared ops for constructing the
8612         reference and copy it.
8614 2013-05-07  Richard Biener  <rguenther@suse.de>
8616         PR middle-end/57190
8617         * tree-eh.c (sink_clobbers): Properly propagate
8618         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8620 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
8622         PR tree-optimization/57149
8623         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8624         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8625         collect_phi_def_edges, execute_late_warn_uninitialized): Use
8626         uninit_undefined_value_p instead of ssa_undefined_value_p.
8628         PR debug/57184
8629         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8630         for modifier == EXPAND_INITIALIZER.
8632 2013-05-07  Anton Blanchard  <anton@samba.org>
8634         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8635         for powerpc64 little endian.
8636         * configure: Regenerate.
8638 2013-05-06  Graham Stott  <grahams@btinternet.com>
8640         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8641         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8642         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8643         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8645 2013-05-06  Graham Stott  <grahams@btinternet.com>
8647         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8648         codes which allow non-lvalues.
8650 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
8652         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8653         components are all 1s.
8654         (integer_minus_onep): New function.
8655         * tree.h (integer_minus_onep): Declare it.
8656         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8657         integer_minus_onep instead of integer_all_onesp.
8659 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8661         PR target/52933
8662         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8663         variations of these patterns.
8665 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8667         * config/i386/i386.md (isa): Add x64_sse4 member.
8668         (enabled): Handle x64_sse4.
8669         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8670         instruction for 64bit SSE4_1 targets.  Update insn attributes.
8671         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8672         instruction for SSE4_1 targets.  Update insn attributes.
8673         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8674         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8675         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8676         const_1 selector.
8677         (*vec_extractv4si): Rename from *sse4_1_pextrd.
8678         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8679         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8681 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8683         PR target/57108
8684         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8686 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
8688         * final.c (do_assembler_dialects): Don't handle curly braces and
8689         vertical bar escaped by % as dialect delimiters.
8690         (output_asm_insn): Print curly braces and vertical bar if escaped
8691         by % and ASSEMBLER_DIALECT defined.
8692         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8693         * doc/tm.texi: Regenerated.
8695 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8697         * config/mips/mips.c: Include tree-pass.h.
8698         (mips_reorg): Split in pre- and post-dbr_schedule parts.
8699         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8700         (pass_mips_machine_reorg2): New machine specific pass.
8701         (insert_pass_mips_machine_reorg2): New pass plugin definition.
8702         (mips_option_override): Register the new pass.
8703         * rtl.h (cleanup_barriers): Remove prototype.
8704         (dbr_schedule): Likewise.
8705         * jump.c (cleanup_barriers): Make static.
8706         * reorg.c (dbr_schedule): Likewise.
8708 2013-05-06  Richard Biener  <rguenther@suse.de>
8710         PR tree-optimization/57185
8711         * tree-parloops.c (add_field_for_reduction): Handle anonymous
8712         SSA names properly.
8714 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8716         PR target/57106
8717         * config/i386/i386.c (add_parameter_dependencies): Add dependence
8718         between "first_arg" and "insn", not "last" and "insn".
8720 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8722         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8723         (find_candidates_in_block): Re-enable slsr_process_phi.
8724         (create_phi_basis): Fix double counting of candidate adjustment.
8726 2013-05-06  Richard Biener  <rguenther@suse.de>
8728         PR middle-end/57147
8729         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8730         the edge is also fallthru, preserve it and just clear the
8731         abnormal flag.
8732         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8733         also complex, preserve that and just clear the fallthru flag.
8734         * tree-inline.c (update_ssa_across_abnormal_edges): Also
8735         update virtual operands.
8737 2013-05-06  Alan Modra  <amodra@gmail.com>
8739         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8740         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8741         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8742         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8743         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8744         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8746 2013-05-06  Alan Modra  <amodra@gmail.com>
8748         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8749         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
8750         (DEFAULT_ASM_ENDIAN): Define.
8751         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8752         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8753         Update -K PIC clause from sysv4.h.
8754         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8755         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8757 2013-05-06  Alan Modra  <amodra@gmail.com>
8759         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8760         twice for little-endian.
8761         (ashrdi3_no_power, ashrdi3): Support little-endian.
8763 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8765         PR target/55303
8766         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8767         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8768         related expanders.
8769         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8770         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8771         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8772         New predicates.
8774 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
8775             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
8777         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8778         * config/pa/pa.opt: Make mbig-switch a no-op.
8779         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8780         (CASE_VECTOR_MODE): Always return SImode.
8781         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8782         for the !TARGET_BIG_SWITCH case.
8783         * config/pa/pa-linux.h: Likewise.
8784         * config/pa/pa-openbsd.h: Likewise.
8785         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8786         * config/pa/pa.md (short_jump): Remove define_insn.
8787         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8788         (casesi0): Remove define_insn.
8789         (type): Remove btable_branch.
8790         (pa_combine_type): Likewise.
8791         (in_nullified_branch_delay): Likewise.
8792         (in_call_delay): Likewise.
8793         (define_delay): Likewise.
8794         (define_insn_reservation "Z3"): Likewise.
8795         (define_insn_reservation "Z4"): Likewise.
8796         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8797         (pa_adjust_insn_length): Remove adjustment for btable branches.
8798         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8799         and mno-big-switch
8801 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
8803         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8804         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8805         Add m->r,x alternatives.
8806         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8807         splitters using SWI48x mode iterator.
8808         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
8809         TARGET_64BIT.  Add m->x alternative.
8810         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8811         Add o->x alternative.  Enable for TARGET_SSE.
8812         (sse_storeq): Remove expander.
8813         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8814         with memory input operand.
8815         (*vec_extractv2di_1 splitter): New.
8816         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8817         * config/i386/i386.md (ssevecmodelower): New mode attribute.
8819 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8821         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
8822         (INT_LOWPART): Delete.
8823         (extract_MB): Adjust.
8824         (extract_ME): Adjust.
8825         (print_operand): Adjust.
8827 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8829         * config/rs6000/predicates.md (reg_or_add_cint_operand,
8830         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8831         (reg_or_logical_cint_operand, easy_fp_constant,
8832         logical_const_operand): Delete "CONST_DOUBLE" case.
8833         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8834         "HOST_BITS_PER_WIDE_INT == 64" test.
8835         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
8836         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8837         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8838         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8839         test.
8840         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8841         CONST_DOUBLE DImode/VOIDmode case.
8842         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8843         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8844         CONST_DOUBLE VOIDmode case.
8845         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8846         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8847         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8848         Delete CONST_DOUBLE case.
8849         (splitters for mov FMOVE64 const_double): Delete
8850         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8851         "HOST_BITS_PER_WIDE_INT >= 64" test.
8852         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8853         case.
8854         (mov DI const_double): Delete.
8856 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
8858         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8859         on op shows all bits zero in mode of a lowpart subreg, return zero.
8861 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
8863         PR target/57150
8864         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8865         to save TFmode registers and DImode to save TImode registers for
8866         caller save operations.
8867         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8868         mark being partially clobbered since they only use the first
8869         double word.
8871         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8872         and TDmode only use the upper 64-bits of each VSX register.
8874 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8876         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8877         (find_candidates_in_block): Disable slsr_process_phi.
8879 2013-05-03  Guozhi Wei  <carrot@google.com>
8881         * coverage.c (coverage_obj_init): Move the construction of gcov
8882         constructor to ...
8883         (build_init_ctor): ... here.
8885 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8887         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
8888         (slsr_cand_d): Redefine def_phi.
8889         (stride_status, phi_adjust_status, count_phis_status): New enums.
8890         (find_phi_def): New.
8891         (find_basis_for_base_expr): New.
8892         (find_basis_for_candidate): Handle hidden bases.
8893         (alloc_cand_and_find_basis): Handle phi candidates.
8894         (slsr_process_phi): New.
8895         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
8896         (create_mul_imm_cand): Likewise.
8897         (create_add_ssa_cand): Exclude phi base candidates.
8898         (create_add_imm_cand): Likewise.
8899         (slsr_process_cast): Likewise.
8900         (slsr_process_copy): Likewise.
8901         (find_candidates_in_block): Handle phi candidates.
8902         (dump_candidate): Likewise.
8903         (unconditional_cands): Delete.
8904         (unconditional_cands_with_known_stride_p): Delete.
8905         (phi_dependent_cand_p): New.
8906         (cand_increment): Handle phi-dependent candidates.
8907         (replace_dependent): Delete.
8908         (replace_mult_candidate): New.
8909         (replace_unconditional_candidate): New.
8910         (incr_vec_index): Move to avoid forward reference.
8911         (create_add_on_incoming_edge): New.
8912         (create_phi_basis): New.
8913         (replace_dependents): Delete.
8914         (replace_conditional_candidate): New.
8915         (phi_add_costs): New.
8916         (replace_uncond_cands_and_profitable_phis): New.
8917         (record_increment): Handle phi adjustments.
8918         (record_phi_increments): New.
8919         (record_increments): Handle phi adjustments.
8920         (phi_incr_cost): New.
8921         (lowest_cost_path): Handle phis.
8922         (total_savings): Likewise.
8923         (analyze_increments): Likewise.
8924         (ncd_with_phi): New.
8925         (ncd_of_cand_and_phis): New.
8926         (nearest_common_dominator_for_cands): Handle phi increments.
8927         (all_phi_incrs_profitable): New.
8928         (replace_profitable_candidates): Handle phi-dependent candidates.
8929         (analyze_candidates_and_replace): Likewise.
8931 2013-05-03  Teresa Johnson  <tejohnson@google.com>
8933         PR bootstrap/57154
8934         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
8935         do not exceed REG_BR_PROB_BASE.
8937 2013-05-03  Jeff Law  <law@redhat.com>
8939         PR tree-optimization/57144
8940         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
8941         operand of the condition will bit into the new type when eliminating
8942         a cast feeding a condition.
8944 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
8946         PR rtl-optimization/57130
8947         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
8948         of COMPARE as in_code to the recursive call if needed.
8950 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
8952         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
8953         (enabled): Handle new members.
8954         * config/i386/sse.md (*vec_concatv2si): Merge from
8955         *vec_concatv2si_sse2 and vec_concatv2si_sse.
8956         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
8958 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
8960         PR tree-optimization/57027
8961         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
8962         for fnms opportunity, check we got the prerequisite kind
8963         of tree / gimple before using accessor functions.
8965 2013-05-03  Richard Biener  <rguenther@suse.de>
8967         * double-int.h (lshift): New overload without precision
8968         and arith argument.
8969         (operator *=, operator +=, operator -=): Move ...
8970         * double-int.c (operator *=, operator +=, operator -=): ... here
8971         and implement more efficiently.
8972         (mul_double_with_sign): Remove.
8973         (lshift_double): Adjust to take unsinged shift argument, push
8974         dispatching code to callers.
8975         (mul_double_wide_with_sign): Add early out for callers that
8976         are not interested in high parts or overflow.
8977         (lshift): New function.
8978         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
8979         dispatch code here.
8980         (lrotate, rrotate): Use logical shifts.
8981         * expr.c (get_inner_reference): Use lshift.
8982         * fixed-value.c (do_fixed_divide): Likewise.
8983         * tree-dfa.c (get_ref_base_and_extent): Likewise.
8984         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
8985         (indirect_refs_may_alias_p): Likewise.
8986         (stmt_kills_ref_p_1): Likewise.
8988 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8990         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
8992 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8994         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
8995         scalar form of FABD instruction.
8997 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
8999         * lra-constraints.c (process_alt_operands): Add checking alt
9000         number to choose the best alternative.
9002 2013-05-02  Richard Biener  <rguenther@suse.de>
9004         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9005         bitmap and its handling.
9006         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9008 2013-05-02  Richard Biener  <rguenther@suse.de>
9010         PR middle-end/57140
9011         * tree-inline.c (copy_loops): Properly handle removed loops.
9012         (copy_cfg_body): Mark destination loops for fixup if source
9013         loops needed fixup.
9015 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9017         PR target/56732
9018         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9019         generating simple_return for naked functions.
9021 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9023         PR middle-end/56988
9024         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9025         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9026         flags match.
9027         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9028         ipa_agg_replacement_value structures.
9029         (known_aggs_to_agg_replacement_list): Likewise.
9030         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9031         (read_agg_replacement_chain): Likewise.
9032         (ipcp_transform_function): Also check that by_ref flags match.
9034 2013-05-02  Richard Biener  <rguenther@suse.de>
9036         * graphds.h (struct graph): Add obstack member.
9037         * graphds.c (new_graph): Initialize obstack and allocate
9038         vertices from it.
9039         (add_edge): Allocate edge from the obstack.
9040         (free_graph): Free the obstack instead of all edges and vertices.
9042 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9044         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9045         divides.
9046         * cfg.c (update_bb_profile_for_threading): Ditto.
9047         * tree-inline.c (copy_bb): Ditto.
9048         (copy_edges_for_bb): Ditto.
9049         (initialize_cfun): Ditto.
9050         (copy_cfg_body): Ditto.
9051         (expand_call_inline): Ditto.
9052         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9053         (estimate_node_size_and_time): Ditto.
9054         (inline_merge_summary): Ditto.
9055         * cgraphclones.c (cgraph_clone_edge): Ditto.
9056         (cgraph_clone_node): Ditto.
9057         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9058         (compute_trg_info): Ditto.
9060 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9062         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9063         S reg when fp attribute set.
9064         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9066 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9068         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9069         New pattern.
9070         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9071         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9072         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9074 2013-05-02  Richard Biener  <rguenther@suse.de>
9076         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9077         (struct instantiate_cache_entry): New type.
9078         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9079         (struct instantiate_cache_type): New type.
9080         (set_instantiated_value, get_instantiated_value): Remove.
9081         (get_instantiated_value_entry): New function.
9082         (instantiate_scev_name): Use the new cache and adjust.
9083         (instantiate_scev_poly): Adjust.
9084         (instantiate_scev_binary): Likewise.
9085         (instantiate_array_ref): Likewise.
9086         (instantiate_scev_convert): Likewise.
9087         (instantiate_scev_not): Likewise.
9088         (instantiate_scev_3): Likewise.
9089         (instantiate_scev_2): Likewise.
9090         (instantiate_scev_r): Likewise.
9091         (instantiate_scev): Likewise.
9092         (resolve_mixers): Likewise.
9094 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9096         PR target/57091
9097         * lra-constraints.c (best_small_class_operands_num): Remove.
9098         (process_alt_operands): Remove small_class_operands_num.  Take
9099         small classes operands into losers and only if the operand is not
9100         matched.  Modify debugging output.
9101         (curr_insn_transform): Remove best_small_class_operands_num.
9102         Print insn name.
9104 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9106         * config/aarch64/aarch64-builtins.c
9107         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9108         * config/aarch64/aarch64-simd-builtins.def
9109         (reduc_splus_): Add new modes.
9110         (reduc_uplus_): New.
9111         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9112         (reduc_uplus_v4sf): Likewise.
9113         (reduc_splus_v4sf): Likewise.
9114         (aarch64_addv<mode>): Likewise.
9115         (reduc_uplus_<mode>): Likewise.
9116         (reduc_splus_<mode>): Likewise.
9117         (aarch64_addvv2di): Likewise.
9118         (reduc_uplus_v2di): Likewise.
9119         (reduc_splus_v2di): Likewise.
9120         (aarch64_addvv2si): Likewise.
9121         (reduc_uplus_v2si): Likewise.
9122         (reduc_splus_v2si): Likewise.
9123         (reduc_<sur>plus_<mode>): New.
9124         (reduc_<sur>plus_v2di): Likewise.
9125         (reduc_<sur>plus_v2si): Likewise.
9126         (reduc_<sur>plus_v4sf): Likewise.
9127         (aarch64_addpv4sf): Likewise.
9128         * config/aarch64/arm_neon.h
9129         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9130         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9131         add UNSPEC_SADDV, UNSPEC_UADDV.
9132         (SUADDV): New.
9133         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9135 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9137         * config/aarch64/arm_neon.h
9138         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9140 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9142         * config/aarch64/aarch64-builtins
9143         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9145 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9147         * config/aarch64/aarch64-simd-builtins.def
9148         (reduc_smax_): New.
9149         (reduc_smin_): Likewise.
9150         (reduc_umax_): Likewise.
9151         (reduc_umin_): Likewise.
9152         (reduc_smax_nan_): Likewise.
9153         (reduc_smin_nan_): Likewise.
9154         (fmax): Remove.
9155         (fmin): Likewise.
9156         (smax): Update for V2SF, V4SF and V2DF modes.
9157         (smin): Likewise.
9158         (smax_nan): New.
9159         (smin_nan): Likewise.
9160         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9161         (<su><maxmin><mode>3): ...This, refactor.
9162         (s<maxmin><mode>3): New.
9163         (<maxmin_uns><mode>3): Likewise.
9164         (reduc_<maxmin_uns>_<mode>): Refactor.
9165         (reduc_<maxmin_uns>_v4sf): Likewise.
9166         (reduc_<maxmin_uns>_v2si): Likewise.
9167         (aarch64_<fmaxmin><mode>: Remove.
9168         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9169         new builtin names.
9170         (vmin<q>_f<32,64>): Likewise.
9171         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9172         (FMAXMIN): New.
9173         (su): Add mappings for smax, smin, umax, umin.
9174         (maxmin): New.
9175         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9176         (FMAXMIN): Rename as...
9177         (FMAXMIN_UNS): ...This.
9178         (maxminv): Remove.
9179         (fmaxminv): Likewise.
9180         (fmaxmin): Likewise.
9181         (maxmin_uns): New.
9182         (maxmin_uns_op): Likewise.
9184 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9186         * config/aarch64/arm_neon.h
9187         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9188         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9189         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9191 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9193         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9194         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9196 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9198         * config/aarch64/aarch64-simd.md
9199         (vcond<mode>_internal): Handle special cases for constant masks.
9200         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9201         (vcondu<mode><mode>): Likewise.
9202         (vcond<v_cmp_result><mode>): New.
9204 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9206         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9207         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9208         * config/aarch64/aarch64-simd-builtins.def
9209         (cmeq): Update to BUILTIN_VALLDI.
9210         (cmgt): Likewise.
9211         (cmge): Likewise.
9212         (cmle): Likewise.
9213         (cmlt): Likewise.
9214         * config/aarch64/arm_neon.h
9215         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9216         to builtins or C as appropriate.
9218 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9220         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9221         (cmgeu): ...This.
9222         (cmhi): Rename to...
9223         (cmgtu): ...This.
9224         * config/aarch64/aarch64-simd.md
9225         (simd_mode): Add SF.
9226         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9227         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9228         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9229         (cstore<mode>_neg): ...This.
9230         * config/aarch64/iterators.md
9231         (VALLF): new.
9232         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9233         (COMPARISONS): New.
9234         (UCOMPARISONS): Likewise.
9235         (optab): Add missing comparisons.
9236         (n_optab): New.
9237         (cmp_1): Likewise.
9238         (cmp_2): Likewise.
9239         (CMP): Likewise.
9240         (cmp): Remove.
9241         (VCMP_S): Likewise.
9242         (VCMP_U): Likewise.
9243         (V_cmp_result): Add DF, SF modes.
9244         (v_cmp_result): Likewise.
9245         (v): Likewise.
9246         (vmtype): Likewise.
9247         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9249 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9251         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9252         define_insn to define_insn_and_split.
9253         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9254         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9255         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9256         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9257         (thumb2_negscc): Likewise.
9259 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9261         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9263 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9265         * config/arm/thumb2.md: Remove trailing whitespaces.
9267 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9269         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9270         Use gen_int_mode rather than GEN_INT.
9272 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9274         * value-prof.c (stream_in_histogram_value): Remove the strayed
9275         debug_gimple_stmt.
9277 2013-04-30  Richard Biener  <rguenther@suse.de>
9279         PR middle-end/57122
9280         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9282 2013-04-30  Richard Biener  <rguenther@suse.de>
9284         PR middle-end/57107
9285         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9287 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9289         PR rtl-optimization/56957
9290         PR rtl-optimization/57105
9291         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9292         variable.  Use just INSN_UID for determining whether an insn
9293         should be only disconnected from the insn stream.
9294         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9296 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9298         PR tree-optimization/57104
9299         * tsan.c (instrument_expr): Don't instrument accesses to
9300         DECL_HARD_REGISTER VAR_DECLs.
9302 2013-04-30  Richard Biener  <rguenther@suse.de>
9304         * function.h (loops_for_fn): New inline function.
9305         (set_loops_for_fn): Likewise.
9306         * cfgloop.h (place_new_loop): Add struct function parameter.
9307         (get_loop): Likewise.
9308         (get_loops): Likewise.
9309         (number_of_loops): Likewise.
9310         (fel_next): Adjust.
9311         (fel_init): Likewise.
9312         * cfg.c (get_loop_copy): Adjust.
9313         * cfgloop.c (flow_loops_dump): Likewise.
9314         (record_loop_exits): Likewise.
9315         (verify_loop_structure): Likewise.
9316         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9317         (estimate_reg_pressure_cost): Likewise.
9318         (mark_loop_exit_edges): Likewise.
9319         * cfgloopmanip.c (place_new_loop): Likewise.
9320         (add_loop): Likewise.
9321         (duplicate_loop): Likewise.
9322         * graph.c (draw_cfg_nodes): Likewise.
9323         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9324         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9325         (extract_affine_chrec): Likewise.
9326         (build_scop_iteration_domain): Likewise.
9327         * graphite.c (graphite_initialize): Likewise.
9328         * ira-build.c (create_loop_tree_nodes): Likewise.
9329         (more_one_region_p): Likewise.
9330         (rebuild_regno_allocno_maps): Likewise.
9331         (mark_loops_for_removal): Likewise.
9332         (mark_all_loops_for_removal): Likewise.
9333         (remove_unnecessary_regions): Likewise.
9334         (ira_build): Likewise.
9335         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9336         * loop-init.c (fix_loop_structure): Likewise.
9337         (gate_rtl_move_loop_invariants): Likewise.
9338         (gate_rtl_unswitch): Likewise.
9339         (gate_rtl_unroll_and_peel_loops): Likewise.
9340         (rtl_doloop): Likewise.
9341         * lto-streamer-in.c (input_cfg): Likewise.
9342         * lto-streamer-out.c (output_cfg): Likewise.
9343         * modulo-sched.c (sms_schedule): Likewise.
9344         * predict.c (tree_estimate_probability): Likewise.
9345         (tree_estimate_probability_driver): Likewise.
9346         (estimate_loops): Likewise.
9347         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9348         (move_sese_region_to_fn): Likewise.
9349         (debug_loop_num): Likewise.
9350         * tree-chrec.c (chrec_evaluate): Likewise.
9351         (hide_evolution_in_other_loops_than_loop): Likewise.
9352         (chrec_component_in_loop_num): Likewise.
9353         (reset_evolution_in_loop): Likewise.
9354         (evolution_function_is_invariant_rec_p): Likewise.
9355         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9356         * tree-inline.c (copy_loops): Likewise.
9357         (copy_cfg_body): Likewise.
9358         (tree_function_versioning): Likewise.
9359         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9360         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9361         Likewise.
9362         (add_to_evolution_1): Likewise.
9363         (scev_const_prop): Likewise.
9364         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9365         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9366         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9367         (tree_ssa_lim_initialize): Likewise.
9368         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9369         (verify_loop_closed_ssa): Likewise.
9370         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9371         (tree_ssa_loop_im): Likewise.
9372         (tree_ssa_loop_unswitch): Likewise.
9373         (tree_vectorize): Likewise.
9374         (check_data_deps): Likewise.
9375         (tree_ssa_loop_ivcanon): Likewise.
9376         (tree_ssa_loop_bounds): Likewise.
9377         (tree_complete_unroll): Likewise.
9378         (tree_complete_unroll_inner): Likewise.
9379         (tree_parallelize_loops): Likewise.
9380         (tree_ssa_loop_prefetch): Likewise.
9381         (tree_ssa_loop_ivopts): Likewise.
9382         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9383         * tree-vectorizer.c (vectorize_loops): Likewise.
9385 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9387         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9388         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9389         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9390         with EABI_LINK_SPEC.
9392 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9394         PR target/44578
9395         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9396         alternative.
9398 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9400         PR target/57097
9401         * lra-constraints.c (process_alt_operands): Discourage a bit more
9402         using memory for pseudos.  Print cost dump for alternatives.
9403         Modify cost values for conflicts with early clobbers.
9404         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9406 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9408         PR target/57098
9409         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9411 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9413         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9414         from/to S register.
9415         (movdi_aarch64): Support LDR/STR from/to D register.
9417 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9419         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9420         or higher by default.
9422 2013-04-29  Richard Biener  <rguenther@suse.de>
9424         PR middle-end/57075
9425         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9426         even if not adding abnormal edges for calls that can make
9427         abnormal gotos.
9429 2013-04-29  Richard Biener  <rguenther@suse.de>
9431         PR middle-end/57103
9432         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9433         TREE_BLOCK.
9434         (move_stmt_r): Remove redundant checking.
9436 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9438         PR bootstrap/57077
9439         * basic-block.h (apply_scale): New function.
9440         (apply_probability): Use apply_scale.
9441         * gimple-streamer-in.c (input_bb): Ditto.
9442         * lto-streamer-in.c (input_cfg): Ditto.
9443         * lto-cgraph.c (merge_profile_summaries): Ditto.
9444         * tree-optimize.c (execute_fixup_cfg): Ditto.
9445         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9446         (copy_edges_for_bb): Ditto.
9447         (copy_cfg_body): Ditto.
9449 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9451         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9452         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9453         (tail_merge_optimize): Handle current_loops == NULL.
9455 2013-04-26  Jeff Law  <law@redhat.com>
9457         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9458         (simplify_cond_using_ranges): Generalize code to simplify
9459         COND_EXPRs where one argument is a constant and the other
9460         is an SSA_NAME created by an integral type conversion.
9462 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9464         * config/arm/arm.md (store_minmaxsi): Use only when
9465         optimize_insn_for_size_p.
9467 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9469         PR target/57108
9470         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9472 2013-04-29  Richard Biener  <rguenther@suse.de>
9474         PR middle-end/57089
9475         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9476         loop tree make sure to schedule a fixup for the child as well.
9477         (expand_omp_for_generic): Properly add loops.
9478         (expand_omp_for_static_nochunk): Likewise.
9479         (expand_omp_for_static_chunk): Likewise.
9480         (expand_omp_for): For the degenerate case fixup loops.
9481         (expand_omp_sections): Fix default bb placement in loops.
9482         (expand_omp_atomic_pipeline): Properly add loops.
9484 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9486         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9488 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9490         * tree-ssa-tail-merge.c: Update header comment.
9492 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9494         * config/aarch64/arm_neon.h
9495         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9496         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9497         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9498         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9499         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9500         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9501         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9502         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9504 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9506         * config/aarch64/aarch64-simd.md
9507         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9508         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9509         fix_trunc, fixuns_trunc.
9510         (ftrunc<VDQF:mode>2): New.
9511         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9512         (fix_trunc_optab): New.
9514 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9516         * config/aarch64/aarch64-builtins.c
9517         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9518         iceilf, lround, iroundf.
9520 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9522         PR target/54349
9523         * config/i386/i386.h (enum ix86_tune_indices)
9524         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9525         New, split from X86_TUNE_INTER_UNIT_MOVES.
9526         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9527         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9528         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9529         (TARGET_INTER_UNIT_MOVES): Remove.
9530         * config/i386/i386.c (initial_ix86_tune_features): Update.
9531         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9532         (ix86_expand_convert_uns_didf_sse): Use
9533         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9534         (ix86_expand_vector_init_one_nonzero): Ditto.
9535         (ix86_expand_vector_init_interleave): Ditto.
9536         (inline_secondary_memory_needed): Return true for moves from SSE class
9537         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9538         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9539         * config/i386/constraints.md (Yi, Ym): Depend on
9540         TARGET_INTER_UNIT_MOVES_TO_VEC.
9541         (Yj, Yn): New constraints.
9542         * config/i386/i386.md (*movdi_internal): Change constraints of
9543         operand 1 from Yi to Yj and from Ym to Yn.
9544         (*movsi_internal): Ditto.
9545         (*movdf_internal): Ditto.
9546         (*movsf_internal): Ditto.
9547         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9548         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9549         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9550         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9551         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9552         * config/i386/sse.md (movdi_to_sse): Ditto.
9553         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9554         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9555         TARGET_INTER_UNIT_MOVES.
9556         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9557         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9558         instead of TARGET_INTER_UNIT_MOVES.
9559         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9560         operand 1 from Yi to Yj and from Ym to Yn.
9562 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9564         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9565         (float_truncate_hi_): Likewise.
9566         (float_extend_lo_): Likewise.
9567         (float_truncate_lo_): Likewise.
9568         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9569         (aarch64_float_extend_lo_v2df): Likewise.
9570         (vec_unpacks_hi_v4sf): Likewise.
9571         (aarch64_float_truncate_lo_v2sf): Likewise.
9572         (aarch64_float_truncate_hi_v4sf): Likewise.
9573         (vec_pack_trunc_v2df): Likewise.
9574         (vec_pack_trunc_df): Likewise.
9576 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9578         * config/aarch64/aarch64-builtins.c
9579         (aarch64_fold_builtin): Fold float conversions.
9580         * config/aarch64/aarch64-simd-builtins.def
9581         (floatv2si, floatv4si, floatv2di): New.
9582         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9583         * config/aarch64/aarch64-simd.md
9584         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9585         * config/aarch64/iterators.md (FLOATUORS): New.
9586         (optab): Add float, floatuns.
9587         (su_optab): Likewise.
9589 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9591         * config/aarch64/aarch64-builtins.c
9592         (aarch64_builtin_vectorized_function): Use new names for
9593         fcvt builtins.
9594         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9595         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9596         (fcvtzu): Split as...
9597         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9598         (fcvtas): Split as...
9599         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9600         (fcvtau): Split as...
9601         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9602         (fcvtps): Split as...
9603         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9604         (fcvtpu): Split as...
9605         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9606         (fcvtms): Split as...
9607         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9608         (fcvtmu): Split as...
9609         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9610         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9611         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9612         (lfrintnusf, lfrintnudf): Likewise.
9613         * config/aarch64/aarch64-simd.md
9614         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9615         define_insn.
9616         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9617         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9618         (fcvt_pattern): Likewise.
9620 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9622         * config/aarch64/aarch64-simd.md
9623         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9624         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9626 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9628         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9629         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9630         (vrnd<a,m,n,p>_f32): Implement using builtins.
9631         (vrnd<i,x><q>_f<32, 64>): New.
9633 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9635         * config/aarch64/aarch64-builtins.c
9636         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9637         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9638         (frintz): Rename to...
9639         (btrunc): ...this.
9640         (frintp): Rename to...
9641         (ceil): ...this.
9642         (frintm): Rename to...
9643         (floor): ...this.
9644         (frinti): Rename to...
9645         (nearbyint): ...this.
9646         (frintx): Rename to...
9647         (rint): ...this.
9648         (frinta): Rename to...
9649         (round): ...this.
9650         * config/aarch64/aarch64-simd.md
9651         (aarch64_frint<frint_suffix><mode>): Delete.
9652         (<frint_pattern><mode>2): Convert to insn.
9653         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9654         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9655         (frint_pattern): Likewise.
9656         (frint_suffix): Likewise.
9658 2013-04-29  Richard Biener  <rguenther@suse.de>
9660         PR tree-optimization/57081
9661         * loop-init.c: Include tree-flow.h.
9662         (loop_optimizer_finalize): Free number of iteration estimates.
9663         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9665 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
9667         PR tree-optimization/57083
9668         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9669         non-singleton shift count range, zero extend low_bound for uns case.
9671         * config/i386/predicates.md (general_vector_operand): New predicate.
9672         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9673         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9674         if they aren't nonimmediate operands.  If their original values
9675         satisfy const_vector_equal_evenodd_p, don't shift them.
9676         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9677         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
9678         if not nonimmediate_operand.
9679         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9680         instead of register_operand.
9681         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9683 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
9685         * stor-layout.c (finalize_size_functions): Allocate a structure and
9686         reset cfun before dumping the functions.
9688 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
9690         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9692         PR target/56866
9693         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9694         use xop_pmacsdqh if uns_p.
9695         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9696         the immediate rotate count.
9698 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
9700         * rtl.h (struct rtx_def): Add comment for field jump.
9701         (LRA_SUBREG_P): New macro.
9702         * recog.c (register_operand): Check LRA_SUBREG_P.
9703         * lra.c (lra): Add note at the end of RTL code. Align non-empty
9704         stack frame.
9705         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9706         (lra_final_code_change): Skip subreg change for operators.
9707         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9708         if there are no operand changes.
9709         * lra-constraints.c (curr_insn_set): New.
9710         (match_reload): Set LRA_SUBREG_P.
9711         (emit_spill_move): Ditto.
9712         (check_and_process_move): Use curr_insn_set. Process only single
9713         set insns.  Don't initialize sec_mem_p and change_p.
9714         (simplify_operand_subreg): Use LRA_SUBREG_P.
9715         (reg_in_class_p): New function.
9716         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
9717         of #ifdef.  Add code to remove cycling.
9718         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9719         non-null disp.  Reload inner instead of disp when base and index
9720         are null.  Try to put lo_sum into register.
9721         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9722         (check_and_process_move): Move code for move cost check to
9723         simple_move_p.  Remove equiv_substitution.
9724         (simple_move_p): New function.
9725         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
9726         curr_insn_set.  Call check_and_process_move only for single set
9727         insns.  Use the new function.  Move call of check_and_process_move
9728         after operand equiv substitution and address process.
9730 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9732         PR go/57045
9733         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9734         with nonlocal goto receivers or returns twice calls, ignore
9735         unininitialized values from abnormal edges to nl goto receiver
9736         or returns twice call.
9738 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9740         PR tree-optimization/57051
9741         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9742         and VEC_RSHIFT_EXPR if shift count is a multiple of element
9743         bitsize.
9745 2013-04-26  Richard Biener  <rguenther@suse.de>
9747         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9748         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
9749         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9750         (fixup_loop_arrays_after_move): New function.
9751         (move_sese_region_to_fn): Properly outline the loop tree parts
9752         of the SESE region.
9754 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
9756         * config/i386/i386.md (type, unit): Fix long lines.
9758 2013-04-26  Richard Biener  <rguenther@suse.de>
9760         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9761         (lto-streamer-out.o): Likewise.
9762         * cfgloop.c (init_loops_structure): Export, add struct function
9763         argument and adjust.
9764         (flow_loops_find): Adjust.
9765         * cfgloop.h (enum loop_estimation): Add EST_LAST.
9766         (init_loops_structure): Declare.
9767         * lto-streamer-in.c: Include cfgloop.h.
9768         (input_cfg): Input the loop tree.
9769         * lto-streamer-out.c: Include cfgloop.h.
9770         (output_cfg): Output the loop tree.
9771         (output_struct_function_base): Do not drop PROP_loops.
9773 2013-03-26  Richard Biener  <rguenther@suse.de>
9775         * tree-cfg.c (execute_build_cfg): Build the loop tree.
9776         (pass_build_cfg): Provide PROP_loops.
9777         (move_sese_region_to_fn): Remove loops that are outlined into fn
9778         for now.
9779         * tree-inline.c: Include cfgloop.h.
9780         (initialize_cfun): Do not drop PROP_loops.
9781         (copy_loops): New function.
9782         (copy_cfg_body): Copy loop structure.
9783         (tree_function_versioning): Initialize destination loop tree.
9784         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9785         (pass_parallelize_loops): Do IL verification.
9786         * loop-init.c (loop_optimizer_init): Fixup loops if required.
9787         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9788         the CFG make sure we fixup loops as well.
9789         * tree-ssa-tail-merge.c: Include cfgloop.h.
9790         (replace_block_by): When merging loop latches mark loops for fixup.
9791         * lto-streamer-out.c (output_struct_function_base): Drop
9792         PROP_loops for now.
9793         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9794         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9795         * ipa-split.c: Include cfgloop.h.
9796         (split_function): Add the new return block to the loop tree root.
9797         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9798         whether we have removed the forwarder block.
9799         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9800         * cfgloop.h (place_new_loop): Declare.
9801         * cfgloopmanip.c (place_new_loop): Export.
9802         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9803         (tree-switch-conversion.o): Likewise.
9804         (tree-complex.o): Likewise.
9805         (tree-inline.o): Likewise.
9806         (tree-ssa-tailmerge.o): Likewise.
9807         (ipa-split.o): Likewise.
9808         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9809         (tree-ssa-copy.o): Likewise.
9810         * tree-switch-conversion.c: Include cfgloop.h
9811         (process_switch): If we emit a bit-test cascade, schedule loops
9812         for fixup.
9813         * tree-complex.c: Include cfgloop.h.
9814         (expand_complex_div_wide): Properly add new basic-blocks to loops.
9815         * asan.c: Include cfgloop.h.
9816         (create_cond_insert_point): Properly add new basic-blocks to
9817         loops, schedule loop fixup.
9818         * cfgloop.c (verify_loop_structure): Check that looks are not
9819         marked for fixup.
9820         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9821         to loops.
9822         (expand_omp_for_generic): Likewise.
9823         (expand_omp_sections): Likewise.
9824         (expand_omp_atomic_pipeline): Schedule loops for fixup.
9825         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9826         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9827         is initialized, not when loops are present.
9828         * tree-parloops.c (parallelize_loops): Remove checking here.
9829         * passes.c (init_optimization_passes): Schedule a copy-propagation
9830         pass before complete unrolling of inner loops.
9832 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9834         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9835         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9836         (colorize_init): Add argument to _WIN32 version.
9837         * toplev.c: Include diagnostic-color.h.
9838         (process_options): Default to -fdiagnostics-color=auto if
9839         GCC_COLORS env var is in the environment.
9840         * common.opt (fdiagnostics-color=): Add Var and Init.
9841         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9842         env var is in the environment, the default is auto rather than never.
9844         * diagnostic.h (file_name_as_prefix): Add context argument.
9845         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
9846         the string as locus.
9847         * langhooks.c (lhd_print_error_function): Adjust caller.
9849 2013-04-25  Lawrence Crowl  <crowl@google.com>
9851         * var-tracking.c (shared_hash_def::htab):
9852         Change type to hash_table.  Update dependent calls and types.
9854 2013-04-25  Lawrence Crowl  <crowl@google.com>
9856         * Makefile.in: Update as needed below.
9858         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9859         Move declaration to after the type's method definitons.
9861         * attribs.c (htab_t scoped_attributes::attribute_hash):
9862         Change type to hash_table.  Update dependent calls and types.
9864         * bitmap.c (htab_t bitmap_desc_hash):
9865         Change type to hash_table.  Update dependent calls and types.
9867         * cselib.c (htab_t cselib_hash_table):
9868         Change type to hash_table.  Update dependent calls and types.
9870         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9871         (hash_string_slot_node): Move implementation into lto-streamer.h
9872         struct string_slot_hasher.
9873         (eq_string_slot_node): Likewise.
9875         * data-streamer-out.c: Update output_block::string_hash_table
9876         dependent calls and types.
9878         * dwarf2cfi.c (htab_t trace_index):
9879         Change type to hash_table.  Update dependent calls and types.
9881         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9882         Change type to hash_table.  Update dependent calls and types.
9883         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
9884         (htab_t optimize_external_refs::map): Likewise.
9885         (htab_t output_comp_unit::extern_map): Likewise.
9886         (htab_t output_comdat_type_unit::extern_map): Likewise.
9887         (htab_t output_macinfo::macinfo_htab): Likewise.
9888         (htab_t optimize_location_lists::htab): Likewise.
9889         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
9891         * except.c (htab_t ehspec_hash_type):
9892         Change type to hash_table.  Update dependent calls and types.
9893         (assign_filter_values::ttypes): Likewise.
9894         (assign_filter_values::ehspec): Likewise.
9895         (sjlj_assign_call_site_values::ar_hash): Likewise.
9896         (convert_to_eh_region_ranges::ar_hash): Likewise.
9898         * gcse.c (htab_t pre_ldst_table):
9899         Change type to hash_table.  Update dependent calls and types.
9901         * ggc-common.c (htab_t saving_htab):
9902         Change type to hash_table.  Update dependent calls and types.
9903         (htab_t loc_hash): Likewise.
9904         (htab_t ptr_hash): Likewise.
9905         (call_count): Rename ggc_call_count.
9906         (call_alloc): Rename ggc_call_alloc.
9907         (loc_descriptor): Rename make_loc_descriptor.
9908         (add_statistics): Rename ggc_add_statistics.
9910         * ggc-common.c (saving_htab):
9911         Change type to hash_table.  Update dependent calls and types.
9913         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
9914         (push_gimplify_context): Likewise.
9915         (pop_gimplify_context): Likewise.
9916         (struct gimple_temp_hash_elt): Added.
9917         (struct gimplify_hasher): Likewise.
9918         (struct gimplify_ctx.temp_htab):
9919         Change type to hash_table.  Update dependent calls and types.
9921         * gimple-fold.c: Include gimplify-ctx.h.
9923         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
9924         Change type to hash_table.  Update dependent calls and types.
9925         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
9926         avoid potential global name collision.
9928         * gimplify.c: Include gimplify-ctx.h.
9929         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
9930         (htab_t gimplify_ctx::temp_htab):
9931         Update dependent calls and types for new type hash_table.
9932         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
9933         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
9935         * gimplify-ctx.h: New.
9936         (struct gimple_temp_hash_elt): Move from gimplify.c.
9937         (class gimplify_hasher): New.
9938         (struct gimplify_ctx): Move from gimple.h.
9939         (htab_t gimplify_ctx::temp_htab):
9940         Change type to hash_table.  Update dependent calls and types.
9942         * graphite-clast-to-gimple.c: Include graphite-htab.h.
9943         (htab_t ivs_params::newivs_index):
9944         Change type to hash_table.  Update dependent calls and types.
9945         (htab_t ivs_params::params_index): Likewise.
9946         (htab_t print_generated_program::params_index): Likewise.
9947         (htab_t gloog::newivs_index): Likewise.
9948         (htab_t gloog::params_index): Likewise.
9950         * graphite.c: Include graphite-htab.h.
9951         4htab_t graphite_transform_loops::bb_pbb_mapping):
9952         Change type to hash_table.  Update dependent calls and types.
9954         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
9955         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
9956         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
9958         * graphite-dependences.c: Include graphite-htab.h.
9959         (loop_is_parallel_p): Change hash table type of parameter.
9961         * graphite-htab.h: New.
9962         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
9963         (extern find_pbb_via_hash): Move from graphite-poly.h.
9964         (extern loop_is_parallel_p): Move from graphite-poly.h.
9965         (extern get_loop_body_pbbs): Move from graphite-poly.h.
9967         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
9968         (extern loop_is_parallel_p): Move to graphite-htab.h.
9969         (extern get_loop_body_pbbs): Move to graphite-htab.h.
9971         * haifa-sched.c (htab_t delay_htab):
9972         Change type to hash_table.  Update dependent calls and types.
9973         (htab_t delay_htab_i2): Likewise.
9975         * ira-color.c (htab_t allocno_hard_regs_htab):
9976         Change type to hash_table.  Update dependent calls and types.
9978         * ira-costs.c (htab_t cost_classes_htab):
9979         Change type to hash_table.  Update dependent calls and types.
9981         * loop-invariant.c (htab_t merge_identical_invariants::eq):
9982         Change type to hash_table.  Update dependent calls and types.
9984         * loop-iv.c (htab_t bivs):
9985         Change type to hash_table.  Update dependent calls and types.
9987         * loop-unroll.c (htab_t opt_info::insns_to_split):
9988         Change type to hash_table.  Update dependent calls and types.
9989         (htab_t opt_info::insns_with_var_to_expand): Likewise.
9991         * lto-streamer.h (struct string_slot): Move from data-streamer.h
9992         (struct string_slot_hasher): New.
9993         (htab_t output_block::string_hash_table):
9994         Change type to hash_table.  Update dependent calls and types.
9996         * lto-streamer-in.c (freeing_string_slot_hasher): New.
9997         (htab_t file_name_hash_table):
9998         Change type to hash_table.  Update dependent calls and types.
10000         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10001         calls and types.
10003         * lto-streamer.c (htab_t tree_htab):
10004         Change type to hash_table.  Update dependent calls and types.
10006         * omp-low.c: Include gimplify-ctx.h.
10008         * passes.c (htab_t name_to_pass_map):
10009         Change type to hash_table.  Update dependent calls and types.
10010         (pass_traverse): Rename to passes_pass_traverse.
10012         * plugin.c (htab_t event_tab):
10013         Change type to hash_table.  Update dependent calls and types.
10015         * postreload-gcse.c (htab_t expr_table):
10016         Change type to hash_table.  Update dependent calls and types.
10017         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10019         * sese.c (debug_rename_map_1): Make extern.
10020         (htab_t copy_bb_and_scalar_dependences::rename_map):
10021         Change type to hash_table.  Update dependent calls and types.
10023         * sese.h (extern debug_rename_map): Move to .c file.
10025         * store-motion.c (htab_t store_motion_mems_table):
10026         Change type to hash_table.  Update dependent calls and types.
10028         * trans-mem.c (htab_t tm_new_mem_hash):
10029         Change type to hash_table.  Update dependent calls and types.
10031         * tree-browser.c (htab_t TB_up_ht):
10032         Change type to hash_table.  Update dependent calls and types.
10034         * tree-cfg.c (htab_t discriminator_per_locus):
10035         Change type to hash_table.  Update dependent calls and types.
10037         * tree-complex.c: Include tree-hasher.h
10038         (htab_t complex_variable_components):
10039         Change type to hash_table.  Update dependent calls and types.
10041         * tree-eh.c (htab_t finally_tree):
10042         Change type to hash_table.  Update dependent calls and types.
10044         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10045         struct int_tree_hasher.
10046         (extern int_tree_map_eq): Likewise.
10047         (uid_decl_map_hash): Removed.
10048         (extern decl_tree_map_eq): Likewise.
10050         * tree-hasher.h: New.
10051         (struct int_tree_hasher): New.
10052         (typedef int_tree_htab_type): New.
10054         * tree-inline.c: Include gimplify-ctx.h.
10056         * tree-mudflap.c: Include gimplify-ctx.h.
10058         * tree-parloops.c: Include tree-hasher.h.
10059         (htab_t eliminate_local_variables_stmt::decl_address):
10060         Change type to hash_table.  Update dependent calls and types.
10061         (htab_t separate_decls_in_region::decl_copies): Likewise.
10063         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10064         Change type to hash_table.  Update dependent calls and types.
10066         * tree-sra.c (candidates):
10067         Change type to hash_table.  Update dependent calls and types.
10069         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10070         in tree-flow.h.
10071         (int_tree_map_hash): Likewise.
10073         * tree-ssa-dom.c (htab_t avail_exprs):
10074         Change type to hash_table.  Update dependent calls and types.
10076         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10077         Change type to hash_table.  Update dependent calls and types.
10079         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10080         Change type to hash_table.  Update dependent calls and types.
10082         * tree-ssa-phiopt.c (seen_ssa_names):
10083         Change type to hash_table.  Update dependent calls and types.
10085         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10086         Change type to hash_table.  Update dependent calls and types.
10088         * tree-ssa-uncprop.c (equiv):
10089         Change type to hash_table.  Update dependent calls and types.
10091 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10093         PR rtl-optimization/57003
10094         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10095         call note_stores with kill_clobbered_value callback again after
10096         killing regs_invalidated_by_call.
10098 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10100         * config/aarch64/aarch64-simd.md
10101         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10102         (aarch64_simd_bsl<mode>): Likewise.
10103         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10105 2013-04-25  Marek Polacek  <polacek@redhat.com>
10107         PR tree-optimization/57066
10108         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10110 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10112         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10114 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10116         * config/aarch64/aarch64-builtins.c
10117         (aarch64_fold_builtin): New.
10118         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10119         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10120         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10121         * config/aarch64/arm_neon.h
10122         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10124 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10125             Tejas Belagod  <tejas.belagod@arm.com>
10127         * config/aarch64/aarch64-builtins.c
10128         (aarch64_gimple_fold_builtin): New.
10129         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10130         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10131         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10132         (addvv4sf): Update.
10133         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10135 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10137         * config/aarch64/aarch64.md
10138         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10140 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10142         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10143         (*ngcsi_uxtw): New pattern.
10145 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10146             Julian Brown  <julian@codesourcery.com>
10148         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10149         (TB_DREG): Add T_V4HF.
10150         (v4hf_UP): New macro.
10151         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10152         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10153         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10154         built-ins.
10155         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10156         (arm_vector_mode_supported_p): Handle V4HF.
10157         (arm_mangle_map): Handle V4HFmode.
10158         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10159         * config/arm/arm_neon_builtins.def: Add entries for
10160         vcvtv4hfv4sf, vcvtv4sfv4hf.
10161         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10162         (neon_vcvtv4hfv4sf): Likewise.
10163         * config/arm/neon-gen.ml: Handle half-precision floating point
10164         features.
10165         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10166         * config/arm/arm_neon.h: Regenerate.
10167         * config/arm/neon.ml (type elts): Add F16.
10168         (type vectype): Add T_float16x4, T_floatHF.
10169         (type vecmode): Add V4HF.
10170         (type features): Add Requires_FP_bit feature.
10171         (elt_width): Handle F16.
10172         (elt_class): Likewise.
10173         (elt_of_class_width): Likewise.
10174         (mode_of_elt): Refactor.
10175         (type_for_elt): Handle F16, fix error messages.
10176         (vectype_size): Handle T_float16x4.
10177         (vcvt_sh): New function.
10178         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10179         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10180         (string_of_mode): Handle V4HF.
10181         * doc/arm-neon-intrinsics.texi: Regenerate.
10183 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10185         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10186         format specifier in 'X' case.
10188 2013-04-25  Alan Modra  <amodra@gmail.com>
10190         PR target/57052
10191         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10192         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10193         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10194         Repeat for many other rotate/shift and mask patterns using subregs.
10195         Name lshiftrt insns.
10196         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10197         on WORDS_BIG_ENDIAN.
10199 2013-04-25  Alan Modra  <amodra@gmail.com>
10201         * config.gcc: Support little-endian powerpc-linux targets.
10202         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10203         (LINK_OS_LINUX_SPEC): Define.
10204         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10205         Preserve MASK_LITTLE_ENDIAN.
10206         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10207         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10208         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10209         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10210         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10211         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10212         Correct fp word order for little-endian.  Don't shift toc entries
10213         smaller than a word for little-endian.
10214         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10215         (bswapdi2 splits): Correct low-part subreg for little-endian.
10216         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10217         low/high where such is correct only for be.
10218         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10219         little-endian for -mcall-aixdesc.
10221 2013-04-25  Alan Modra  <amodra@gmail.com>
10223         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10224         replace_equiv_address_nv.
10226 2013-04-25  Alan Modra  <amodra@gmail.com>
10228         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10230 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10232         Revert:
10233         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10234         * rtl.h (struct rtx_def): ...
10236 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10238         PR rtl-optimizations/57046
10239         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10240         for multi-reg splits.
10242 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10244         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10246 2013-04-24  Sterling Augustine  <saugustine@google.com>
10248         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10249         (comp_dir_string, debug_str_dwo_section): New.
10250         (DEBUG_STR_DWO_SECTION): Rename to ...
10251         (DEBUG_DWO_STR_SECTION): ... this.
10252         (DEBUG_NORM_STR_SECTION): Delete.
10253         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10254         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10255         (find_AT_string): Move most logic to ...
10256         (find_AT_string_in_table): ... here.  New.
10257         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10258         add_skeleton_AT_string.  Delete logic.
10259         (output_skeleton_debug_sections): Remove call to
10260         add_top_level_skeleton_die_attrs.
10261         (add_comp_dir_attribute): Move logic to comp_dir_string.
10262         (dwarf2out_init): Initialize debug_str_dwo_section.
10263         (output_indirect_string): Call find_string_form.
10264         (output_indirect_strings): Rewrite.
10265         (prune_unused_types): Empty skeleton_debug_str_hash.
10266         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10267         (dwarf2out_finish):  Call output_indirect_strings.
10269 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10271         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10273 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10275         * rtl.h (struct rtx_def): Add comment for field jump.
10276         (LRA_SUBREG_P): New macro.
10277         * recog.c (register_operand): Check LRA_SUBREG_P.
10278         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10279         stack frame.
10280         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10281         (lra_final_code_change): Skip subreg change for operators.
10282         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10283         if there are no operand changes.
10284         * lra-constraints.c (curr_insn_set): New.
10285         (match_reload): Set LRA_SUBREG_P.
10286         (emit_spill_move): Ditto.
10287         (check_and_process_move): Use curr_insn_set. Process only single
10288         set insns.  Don't initialize sec_mem_p and change_p.
10289         (simplify_operand_subreg): Use LRA_SUBREG_P.
10290         (reg_in_class_p): New function.
10291         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10292         of #ifdef.  Add code to remove cycling.
10293         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10294         non-null disp.  Reload inner instead of disp when base and index
10295         are null.  Try to put lo_sum into register.
10296         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10297         (check_and_process_move): Move code for move cost check to
10298         simple_move_p.  Remove equiv_substitution.
10299         (simple_move_p): New function.
10300         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10301         curr_insn_set.  Call check_and_process_move only for single set
10302         insns.  Use the new function.  Move call of check_and_process_move
10303         after operand equiv substitution and address process.
10305 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10307         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10308         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10309         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10311 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10313         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10315 2013-04-24  Marek Polacek  <polacek@redhat.com>
10317         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10318         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10319         (select_loops_exit_conditions): Likewise.
10320         (number_of_iterations_for_all_loops): Likewise.
10321         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10322         (scev_analysis): Likewise.
10324 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10325             Chao-ying Fu <fu@mips.com>
10327         * config/mips/micromips.md (jraddiusp): New pattern.
10328         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10329         instruction if possible.
10331 2013-04-24  Alan Modra  <amodra@gmail.com>
10333         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10335 2013-04-24  Julian Brown  <julian@codesourcery.com>
10336             Chung-Lin Tang  <cltang@codesourcery.com>
10338         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10339         dependency behavior in enumeration type DIE generation. Add TODO note
10340         to comments about future DW_FORM_sdata/udata re-work of related code.
10342 2013-04-23  Lawrence Crowl  <crowl@google.com>
10344         * Makefile.in: Update as needed below.
10346         * hash-table.h (class hash_table):
10347         Correct many methods with parameter types compare_type to the correct
10348         value_type.  (Correct code was unlikely to notice the change.)
10349         (hash_table::elements_with_deleted) New.
10350         (class hashtable::iterator): New.
10351         (hashtable::begin()): New.
10352         (hashtable::end()): New.
10353         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10355         * statistics.c (statistics_hashes):
10356         Change type to hash_table.  Update dependent calls and types.
10358         * tree-into-ssa.c (var_infos):
10359         Change type to hash_table.  Update dependent calls and types.
10361         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10362         Change type to hash_table.  Update dependent calls and types.
10364         * tree-ssa-loop-im.c (struct mem_ref.refs):
10365         Change type to hash_table.  Update dependent calls and types.
10367         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10368         Change type to hash_table.  Update dependent calls and types.
10370         * tree-ssa-sccvn.c (vn_tables_s::nary):
10371         Change type to hash_table.  Update dependent calls and types.
10372         (vn_tables_s::phis): Likewise.
10373         (vn_tables_s::references): Likewise.
10375         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10376         (vn_reference_eq): Update parameter and return types.
10378         * tree-ssa-structalias.c (pointer_equiv_class_table):
10379         Change type to hash_table.  Update dependent calls and types.
10380         (location_equiv_class_table): Likewise.
10382         * tree-vect-data-refs.c: Consequential changes for making
10383         peeling a hash_table.
10385         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10386         (destroy_loop_vec_info): Dependent hash_table update.
10388         * tree-vectorizer.h (peeling_htab):
10389         Change type to hash_table.  Update dependent calls and types.
10391 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10393         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10394         to check the register content is equal or not.
10395         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10396         to assign register content record.
10397         * lra-eliminations.c (update_reg_eliminate): Use
10398         lra_update_reg_val_offset to update register content offset.
10399         * lra-int.h (struct lra_reg): Add offset member.
10400         (lra_reg_val_equal_p): New static inline function.
10401         (lra_update_reg_val_offset): New static inline function.
10402         (lra_assign_reg_val): New static inline function.
10403         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10404         to assign register content record.
10405         (initialize_lra_reg_info_element): Initial offset to zero.
10407 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10409         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10410         operands.  Record compression.
10412 2013-04-23  Xinliang David Li  <davidxl@google.com>
10414         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10416 2013-04-23  Richard Biener  <rguenther@suse.de>
10418         PR middle-end/57036
10419         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10420         parameter, only add abnormal goto edges from the copied body
10421         if the call could perform abnormal gotos.
10422         (copy_cfg_body): Adjust.
10424 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10426         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10428 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10430         * coretypes.h (gimple_stmt_iterator): Add struct to make
10431         compatible with C.
10433 2013-04-23  Richard Biener  <rguenther@suse.de>
10435         PR tree-optimization/57026
10436         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10437         from SSA names occuring in abnormal PHI nodes.
10439 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10441         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10443 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10445         * lto/lto.c (print_lto_report_1): Declare early.
10446         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10448 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10450         * common.opt (-flto-report-wpa): Add.
10451         * doc/invoke.texi (-flto-report-wpa): Add.
10452         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10453         (lto_main): dito.
10455 2013-04-22  Xinliang David Li  <davidxl@google.com>
10457         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10458         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10459         * Makefile.in: New dependency
10461             David Daney <ddaney.cavm@gmail.com>
10463         * configure.ac (gcc_cv_as_micromips_support):  Use the
10464         --fatal-warnings option.
10465         * configure: Regenerate.
10467 2013-04-22  Marek Polacek  <polacek@redhat.com>
10469         PR sanitizer/56990
10470         * tsan.c (instrument_expr): Don't instrument expression
10471         in case its size is zero.
10473 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10475         PR target/57032
10476         Revert:
10477         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10479         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10481 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10483         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10484         (gimple_stmt_iterator): New typedef.
10485         * gimple.h (gimple_stmt_iterator): Rename to...
10486         (gimple_stmt_iterator_d): ... This.
10487         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10488         trees be valid for GIMPLE and GENERIC.
10489         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10490         * gimple-fold.c (gimple_fold_call): Call target hook
10491         gimple_fold_builtin.
10492         * hooks.c (hook_bool_gsiptr_false): New.
10493         * hooks.h (hook_bool_gsiptr_false): New.
10494         * target.def (fold_stmt): New.
10495         * doc/tm.texi: Regenerate.
10497 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10499         PR target/57018
10500         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10501         a set sp if no stack realignment.
10503 2013-04-22  Nick Clifton  <nickc@redhat.com>
10505         * config.gcc (tilegx-linux): Extend extra_objs rather than
10506         overwriting it.
10507         (tilepro-linux): Likewise.
10509 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10511         * config/aarch64/aarch64-builtins.c
10512         (CF): Remove.
10513         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10514         (VAR<1-12>): Add MAP parameter.
10515         (BUILTIN_*): Likewise.
10516         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10517         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10518         (aarch64_ushl_n<mode>): Likewise.
10519         (aarch64_sshr_n<mode>): Likewise.
10520         (aarch64_ushr_n<mode>): Likewise.
10521         (aarch64_<maxmin><mode>): Likewise.
10522         (aarch64_sqrt<mode>): Likewise.
10523         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10524         (vshr<q>_n_*): Likewise.
10526 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10528         * config/aarch64/aarch64-builtins.c
10529         (aarch64_simd_builtin_type_mode): Handle SF types.
10530         (sf_UP): Define.
10531         (BUILTIN_GPF): Define.
10532         (aarch64_init_simd_builtins): Handle SF types.
10533         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10534         (frecps): Likewise.
10535         (frecpx): Likewise.
10536         * config/aarch64/aarch64-simd.md
10537         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10538         (aarch64_frecpe<mode>): New.
10539         (aarch64_frecps<mode>): Likewise.
10540         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10541         (v8type): Add frecp<esx>.
10542         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10543         (aarch64_frecps<mode>): Likewise.
10544         * config/aarch64/iterators.md (FRECP): New.
10545         (frecp_suffix): Likewise.
10546         * config/aarch64/arm_neon.h
10547         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10549 2013-04-22  Christian Bruel  <christian.bruel@st.com>
10551         PR target/56995
10552         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10553         (REG_CLASS_NAMES): Idem.
10554         (REG_CLASS_CONTENTS): Idem.
10555         (REGCLASS_HAS_FP_REG): Idem.
10556         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10557         (sh_conditional_register_usage): Idem.
10559 2013-04-21  Jeff Law  <law@redhat.com>
10561         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10562         (ssa_forward_propagate_and_combine): Use it.
10564 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10566         * lra.c: Update the flow chart diagram.
10568 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10570         PR rtl-optimization/56847
10571         * lra-constraints.c (process_alt_operands): Discourage alternative
10572         with non-matche doffsettable memory constraint fro memory with
10573         known offset.
10575 2013-04-19  Richard Biener  <rguenther@suse.de>
10577         PR tree-optimization/56982
10578         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10579         function.
10580         * gimplify.c (gimplify_call_expr): Notice special calls.
10581         (gimplify_modify_expr): Likewise.
10582         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10583         abnormal control flow receivers.
10584         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10585         in the same way as cfun->has_nonlocal_labels.
10586         (gimple_purge_dead_abnormal_call_edges): Likewise.
10587         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10588         receivers start a basic-block.
10590 2013-04-19  Richard Biener  <rguenther@suse.de>
10592         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10593         member ...
10594         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
10595         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10596         (SLP_TREE_LOAD_PERMUTATION): Add.
10597         (vect_transform_slp_perm_load): Adjust prototype.
10598         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10599         (vect_free_slp_instance): Likewise.
10600         (vect_create_new_slp_node): Likewise.
10601         (vect_supported_slp_permutation_p): Remove.
10602         (vect_slp_rearrange_stmts): Adjust.
10603         (vect_supported_load_permutation_p): Likewise.  Inline
10604         vect_supported_slp_permutation_p here.
10605         (vect_analyze_slp_instance): Compute load permutations per
10606         slp node instead of per instance.
10607         (vect_get_slp_defs): Adjust.
10608         (vect_transform_slp_perm_load): Likewise.
10609         (vect_schedule_slp_instance): Remove redundant code.
10610         (vect_schedule_slp): Remove hack for PR56270, add it ...
10611         * tree-vect-stmts.c (vectorizable_load): ... here, do not
10612         CSE loads for SLP.  Adjust.
10614 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10616         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10617         spelling in two comments.
10619 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10621         PR target/56797
10622         * config/arm/arm.c (load_multiple_sequence): Require SP
10623         as base register for loads if SP is in the register list.
10625 2013-04-19  Martin Jambor  <mjambor@suse.cz>
10627         PR tree-optimization/56718
10628         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10629         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10630         and made public.  Adjusted all callers.
10631         (ipa_intraprocedural_devirtualization): New function.
10632         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10633         (ipa_intraprocedural_devirtualization): Likewise.
10634         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10636 2013-04-19  Richard Biener  <rguenther@suse.de>
10638         PR tree-optimization/57000
10639         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10641 2013-04-19  Terry Guo  <terry.guo@arm.com>
10643         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10644         Replace with ...
10645         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
10646         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10647         (cortex_m4_fmacs): Use new reservations.
10648         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10650 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
10652         PR rtl-optimization/56999
10653         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10654         related code.
10655         (lra_coalesce): Remove split_origin_bitmap and related code.
10656         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
10657         ranges if necessary.
10659 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
10661         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10662         New array.
10663         (ix86_expand_call): Remove clobbered_registers array and use
10664         x86_64_ms_sysv_extra_clobbered_registers instead.
10665         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10666         Declare here.
10667         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10668         predicate.
10669         * config/i386/i386.md (*call_rex64_ms_sysv): Use
10670         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
10671         (*call_value_rex64_ms_sysv): Ditto.
10673 2013-04-18  Cary Coutant  <ccoutant@google.com>
10675         * dwarf2out.c (output_pubnames): Check die_perennial_p of
10676         parent instead of die_mark.
10678 2013-04-18  Diego Novillo  <dnovillo@google.com>
10680         * gimple.c (create_gimple_tmp): New.
10681         (get_expr_type): New.
10682         (build_assign): New.
10683         (build_type_cast): New.
10684         * gimple.h (enum ssa_mode): Define.
10685         (gimple_seq_set_location): New.
10686         * asan.c (build_check_stmt): Change some gimple_build_* calls
10687         to use build_assign and build_type_cast.
10689 2013-04-18  Richard Biener  <rguenther@suse.de>
10691         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10692         handle negative step.  Remove redundant checks.
10693         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10694         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10695         for negative step and grouped loads fail to vectorize.
10697 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10699         * emit-rtl.c (reset_insn_used_flags): New function.
10700         (reset_all_used_flags): Use it.
10701         (verify_insn_sharing): New function.
10702         (verify_rtl_sharing): Fix verification for SEQUENCEs.
10704 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
10706         PR tree-optimization/56984
10707         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10708         and (x >> M) >= N don't register any assertion if N << M is the
10709         minimum value.
10711 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10713         * lower-subreg.c (resolve_simple_move): If called self-recursive,
10714         do not delete_insn insns that have not yet been emitted, only
10715         unlink them with remove_insn.
10716         * df-scan.c (df_insn_delete): Revert r197492.
10718 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
10720         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10721         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10723 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10725         * config/arm/arm.md (movsicc_insn): Convert define_insn into
10726         define_insn_and_split.
10727         (and_scc,ior_scc,negscc): Likewise.
10728         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10730 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10732         * config/arm/arm.c (use_return_insn): Return 0 for targets that
10733         can benefit from using a sequence of LDRD instructions in epilogue
10734         instead of a single LDM instruction.
10736 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
10738         PR 45688
10739         * doc/extend.texi: Fix typo.
10741 2013-04-17  Richard Biener  <rguenther@suse.de>
10743         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10744         (vect_build_slp_tree): ... here.
10745         (vect_build_slp_tree_1): Compute which stmts of the SLP group
10746         match.  Remove special-casing of mismatched complex loads.
10747         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10748         re-try the match with swapped commutative operands.
10749         (vect_supported_load_permutation_p): Remove special-casing of
10750         mismatched complex loads.
10751         (vect_analyze_slp_instance): Adjust.
10753 2013-04-17  Richard Biener  <rguenther@suse.de>
10755         PR rtl-optimization/56921
10756         * cfgloop.h (struct loop): Add simple_loop_desc member.
10757         (struct niter_desc): Mark with GTY(()).
10758         (simple_loop_desc): Do not use aux field but simple_loop_desc.
10759         * loop-iv.c (get_simple_loop_desc): Likewise.
10760         (free_simple_loop_desc): Likewise.
10762         Revert
10763         2013-04-16  Richard Biener  <rguenther@suse.de>
10765         PR rtl-optimization/56921
10766         * loop-init.c (pass_rtl_move_loop_invariants): Add
10767         TODO_do_not_ggc_collect to todo_flags_finish.
10768         (pass_rtl_unswitch): Same.
10769         (pass_rtl_unroll_and_peel_loops): Same.
10770         (pass_rtl_doloop): Same.
10772 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
10774         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10775         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10776         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10777         references.
10778         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10779         * tree-streamer.c (record_common_node): Adjust reference in comment.
10781 2013-04-17  Terry Guo  <terry.guo@arm.com>
10783         * config/arm/cortex-m4.md: Add a new bypass.
10785 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10787         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10788         New pattern.
10789         (*subs_<optab><mode>_multp2): New pattern.
10790         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10791         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10793 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10795         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10796         (*subs_mul_imm_<mode>): New pattern.
10798 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
10800         PR target/56948
10801         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10802         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
10803         (vsx_movti_32bit): Change j->wa to O->wa.
10805 2013-04-16  Richard Biener  <rguenther@suse.de>
10807         PR rtl-optimization/56921
10808         * loop-init.c (pass_rtl_move_loop_invariants): Add
10809         TODO_do_not_ggc_collect to todo_flags_finish.
10810         (pass_rtl_unswitch): Same.
10811         (pass_rtl_unroll_and_peel_loops): Same.
10812         (pass_rtl_doloop): Same.
10814 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
10816         * config/arm/arm.c (emit_multi_reg_push): New declaration
10817         for an existing function.
10818         (arm_emit_strd_push): New function.
10819         (arm_expand_prologue): Used here.
10820         (arm_emit_ldrd_pop): New function.
10821         (arm_expand_epilogue): Used here.
10822         (arm_get_frame_offsets): Update condition.
10823         (arm_emit_multi_reg_pop): Add a special case for load of a single
10824         register with writeback.
10826 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
10828         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10829         description.
10831 2013-04-16  Richard Biener  <rguenther@suse.de>
10833         PR tree-optimization/56756
10834         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10835         (first_mem_ref_loc): New.
10836         (execute_sm): Place the load temporarily before a previous
10837         access instead of in the latch edge to ensure its SSA dependencies
10838         are defined at points dominating the load.
10840 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
10842         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10843         correct fix by moving header and footer insn to the footer of
10844         the merged basic block.  Clear BB_END of the merged-away block.
10846         PR middle-end/43631
10847         * emit-rtl.c (make_note_raw): New function.
10848         (link_insn_into_chain): New static inline function.
10849         (add_insn): Use it.
10850         (add_insn_before, add_insn_after): Factor insn chain linking code...
10851         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10852         using link_insn_into_chain.
10853         (note_outside_basic_block_p): New helper function for emit_note_after
10854         and emit_note_before.
10855         (emit_note_after): Use nobb variant of add_insn_after if the note
10856         should not be contained in a basic block.
10857         (emit_note_before): Use nobb variant of add_insn_before if the note
10858         should not be contained in a basic block.
10859         (emit_note_copy): Use make_note_raw.
10860         (emit_note): Likewise.
10861         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10862         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10863         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10864         the moved barrier the tail of the basic block it follows.
10865         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10867 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
10869         PR tree-optimization/56962
10870         * gimple-ssa-strength-reduction.c (record_increment): Only set
10871         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10872         either rhs1 or rhs2 is equal to c->base_expr.
10874 2013-04-15  Richard Biener  <rguenther@suse.de>
10876         PR tree-optimization/56933
10877         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10878         member.
10879         (GROUP_READ_WRITE_DEPENDENCE): Remove.
10880         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10881         * tree-vect-data-refs.c (vect_analyze_group_access): Move
10882         dependence check ...
10883         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
10884         ... here.
10885         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
10886         GROUP_READ_WRITE_DEPENDENCE.
10888 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10890         * emit-rtl.c (reset_all_used_flags): New function.
10891         (verify_rtl_sharing): Call reset_all_used_flags before and after
10892         performing the checks.
10894 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10896         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
10897         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
10898         * config/arm/constraints.md (De): New constraint.
10899         * config/arm/neon.md (anddi3_neon): Delete.
10900         (neon_vand<mode>): Expand to standard anddi3 pattern.
10901         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
10902         Move earlier in the file.
10903         (neon_inv_logic_op2): Likewise.
10904         (arm_anddi_operand_neon): New predicate.
10906 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10908         * configure.ac (gcc_cv_ld_as_needed): Set
10909         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
10910         Use -z ignore, -z record on *-*-solaris2*.
10911         (HAVE_LD_AS_NEEDED): Update comment.
10912         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
10913         * configure: Regenerate.
10914         * config.in: Regenerate.
10915         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
10916         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
10917         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
10918         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
10919         equivalents.  Fix markup.
10920         * doc/tm.texi: Regenerate.
10922 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
10924         * config/i386/i386.opt: New option mstack-protector-guard=.
10925         * config/i386/i386-opts.h: Add enum stack_protector_guard.
10926         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
10927         TARGET_SSP_TLS_GUARD.
10928         * config/i386/i386.c (ix86_option_override_internal): Set
10929         ix86_stack_protector_guard.
10930         * config/i386/i386.md (stack_protect_set): Enable for
10931         TARGET_SSP_TLS_GUARD only.
10932         (stack_protect_set_<mode>): Ditto.
10933         (stack_protect_test): Ditto.
10934         (stack_protect_test_<mode>): Ditto.
10935         * doc/invoke.texi (i386 Option): Document.
10937 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
10939         PR target/56890
10940         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
10941         (S_MODES): Set H_MODE bit.
10942         (SF_MODES): Set only S_MODE and SF_MODE bits.
10943         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
10944         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
10945         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
10946         <MODE_FLOAT>: Likewise.
10948 2013-04-15  Joey Ye  <joey.ye@arm.com>
10950         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
10952 2013-04-15  Joey Ye  <joey.ye@arm.com>
10954         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
10955         for real far jump.
10956         (thumb_far_jump_used_p): Count instruction size and set
10957         far_jump_used.
10959 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
10961         * reorg.c (fill_simple_delay_slots): Reindent block of code.
10962         * resource.c (mark_target_live_regs): Reformat conditional block.
10964 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10966         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
10967         notes, they are emitted only just before final.
10968         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
10970 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10972         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
10973         * cfgrtl.c (delete_insn): Call it here instead.
10974         * lra-spills.c (lra_final_code_change): Use delete_insn.
10975         * haifa-sched.c (sched_remove_insn): Likewise.
10976         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
10977         returning to the nop pool.
10978         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
10979         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
10981 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10983         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
10984         * doc/tm.texi: Regenerated.
10986 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
10988         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
10989         QImode checks.
10991 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10993         * df-core.c (df_find_def): Compare register numbers.
10994         (df_find_use): Likewise.
10996 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
10998         PR target/56903
10999         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11000         lra_in_progress for return.
11002 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11004         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11005         define_insn into define_insn_and_split and emit movsicc patterns.
11007 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11009         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11011 2013-04-12  Richard Biener  <rguenther@suse.de>
11013         * tree-pass.h (TODO_do_not_ggc_collect): New.
11014         * passes.c (execute_one_ipa_transform_pass): Honor
11015         TODO_do_not_ggc_collect.
11016         (execute_one_pass): Likewise.
11018         Revert
11019         2013-04-10  Richard Biener  <rguenther@suse.de>
11021         * passes.c (init_optimization_passes): Remove reload pass.
11022         * ira.c (do_reload): Merge into ...
11023         (ira): ... this.
11024         (rest_of_handle_reload): Remove.
11025         (pass_reload): Likewise.
11026         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11027         of reload for vzeroupper pass placement.
11029 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11031         PR tree-optimization/56918
11032         PR tree-optimization/56920
11033         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11034         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11035         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11036         use wide_mul_with_sign method.
11038 2013-04-12  Richard Biener  <rguenther@suse.de>
11040         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11041         not be considered a gimple constant.
11043 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11045         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11046         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11047         operations.
11049 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11050             Jakub Jelinek  <jakub@redhat.com>
11052         * opts.c: Include diagnostic-color.h.
11053         (common_handle_option): Handle OPT_fdiagnostics_color_.
11054         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11055         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11056         (diagnostic-color.o): New.
11057         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11058         (diagnostic_color_rule): New enum.
11059         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11060         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11061         the location string.
11062         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11063         either NULL, or color kind.
11064         * diagnostic-color.c: New file.
11065         * diagnostic-color.h: New file.
11066         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11067         arguments.
11068         * doc/invoke.texi (-fdiagnostics-color): Document.
11069         * pretty-print.h (pp_show_color): Define.
11070         (struct pretty_print_info): Add show_color field.
11071         * diagnostic.c: Include diagnostic-color.h.
11072         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11073         macros.  Colorize error:, warning: etc. strings and also the location
11074         string.
11075         (diagnostic_show_locus): Colorize the caret line.
11076         * pretty-print.c: Include diagnostic-color.h.
11077         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11078         inside of %< %> quotes or quoted through q format modifier.
11080 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11082         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11084 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11086         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11087         code in CC_NZ mode.
11088         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11089         pattern.
11091 2013-04-11  Marek Polacek  <polacek@redhat.com>
11093         PR tree-optimization/48184
11094         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11096 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11098         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11099         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11100         (skip_simple_arithmetic): Tidy up.
11101         * tree.h (skip_simple_constant_arithmetic): Declare.
11103 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11105         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11107 2013-04-11  Richard Biener  <rguenther@suse.de>
11109         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11110         generate vector constants.
11112 2013-04-11  Richard Biener  <rguenther@suse.de>
11114         PR tree-optimization/56878
11115         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11116         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11117         New function.
11118         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11119         Prefer to align the DR with the most invariant base address.
11121 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11123         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11124         comment.
11126 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11128         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11129         floating-point vector comparisons against 0.
11131 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11133         PR tree-optimization/56899
11134         * fold-const.c (extract_muldiv_1): Apply distributive law
11135         only if TYPE_OVERFLOW_WRAPS (ctype).
11137 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11139         PR target/56124
11140         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11141         loading has side effect.
11143 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11145         * config/sparc/sparc.c: Include tree-pass.h.
11146         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11147         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11148         head of file.  Change return type.  Split off gate function.
11149         (sparc_gate_work_around_errata): New function.
11150         (pass_work_around_errata): New pass definition.
11151         (insert_pass_work_around_errata) New pass insert definition to
11152         insert pass_work_around_errata just after delayed-branch scheduling.
11153         (sparc_option_override): Insert the pass.
11154         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11156 2013-04-10  David S. Miller  <davem@davemloft.net>
11158         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11159         or -mcpu=hypersparc.
11161         * target.def (cstore_mode): New hook.
11162         * target.h: Include insn-codes.h
11163         * targhooks.c: Likewise.
11164         (default_cstore_mode): New function.
11165         * targhooks.h: Declare it.
11166         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11167         * doc/tm.texi: Rebuild.
11168         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11169         target hook, rather than inspecting the insn_data.
11170         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11171         (TARGET_CSTORE_MODE): Redefine.
11172         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11173         result patterns.
11174         * config/sparc/predicates.md (cstore_result_operand): New special
11175         predicate.
11176         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11177         Use it for operand 0.
11178         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11179         (*snesi_special): Likewise.
11180         (*snesi_zero): Likewise.
11181         (*seqsi_zero): Likewise.
11182         (*sltu_insn): Likewise.
11183         (*sgeu_insn): Likewise.
11184         (*seqdi_special): Make operand 0 and comparison operation be of
11185         DImode.
11186         (*snedi_special): Likewise.
11187         (*snedi_special_vis3): Likewise.
11188         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11189         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11190         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11191         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11192         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11193         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11194         (*sltu_extend_sp64): Likewise.
11195         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11196         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11197         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11198         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11199         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11201 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11203         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11204         (aarch64_start_file): Use the new function.
11206 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11207             Jason Merrill  <jason@redhat.com>
11209         * common.opt: Add -gdwarf.
11210         * opts.c (common_handle_option): Handle it.
11211         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11213 2013-04-10  Richard Biener  <rguenther@suse.de>
11215         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11216         (execute_one_ipa_transform_pass): But unconditionally here.
11217         (execute_one_pass): And here.
11218         (init_optimization_passes): Remove reload pass.
11219         * tree-pass.h (TODO_ggc_collect): Remove.
11220         (pass_reload): Likewise.
11221         * ira.c (do_reload): Merge into ...
11222         (ira): ... this.
11223         (rest_of_handle_reload): Remove.
11224         (pass_reload): Likewise.
11225         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11226         of reload for vzeroupper pass placement.
11227         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11228         and todo_flags_finish of all passes.
11230 2013-04-10  Richard Biener  <rguenther@suse.de>
11232         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11233         first_const_oprnd field, rename first_def_type to first_op_type.
11234         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11235         (vect_get_and_check_slp_defs): Always use the type of the
11236         operand.  Allow mixed vect_external_def, vect_constant_def types.
11237         (vect_get_constant_vectors): Handle mixed vect_external_def,
11238         vect_constant_def types.
11240 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11242         PR tree-optimization/55524
11243         * tree-ssa-math-opts.c
11244         (convert_mult_to_fma): Don't use an fms construct
11245         when we don't have an fms operation, but fnma, and it looks
11246         likely that we'll be able to use the latter.
11248 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11250         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11251         function.
11252         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11253         inline fail caused by overwritable functions.
11255 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11257         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11258         unnecessary bits in the constant power of two case.
11260 2013-04-10  Richard Biener  <rguenther@suse.de>
11262         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11263         broken code swapping operands.
11264         (vect_build_slp_tree): Do not compute load permutations here.
11265         (vect_analyze_slp_instance): Compute load permutations here,
11266         after building the SLP tree.
11268 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11270         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11271         of next/prev_real_insn.
11273 2013-04-09  Jan Hubicka  <jh@suse.cz>
11275         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11276         Drop aliased parameter.
11277         (function_and_variable_visibility): Do not handle alias pairs.
11278         * cgraph.c (varpool_externally_visible_p): Update prototype.
11279         * varpool.c (varpool_add_new_variable): Update.
11281 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11283         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11285 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11287         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11289         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11291 2013-04-09  Marek Polacek  <polacek@redhat.com>
11293         PR tree-optimization/48762
11294         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11296 2013-04-09  Richard Biener  <rguenther@suse.de>
11298         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11299         dealing with cost.
11300         (vect_build_slp_tree): Likewise.
11301         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11302         calculating the cost of a SLP instance.
11303         (vect_analyze_slp_instance): Use it from here, after building
11304         the SLP tree.
11306 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11308         PR middle-end/56883
11309         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11310         expand_omp_for_static_chunk): Use simple_p = true in
11311         force_gimple_operand_gsi calls when assigning to addressable decls.
11313 2013-04-09  Jeff Law  <law@redhat.com>
11315         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11316         when the boolean was created by converting a wider object which
11317         had a boolean range.
11319 2013-04-09  Richard Biener  <rguenther@suse.de>
11321         * tree-vectorizer.h (slp_void_p): Remove.
11322         (slp_tree): Typedef before _slp_tree declaration.
11323         (struct _slp_tree): Use a vector of slp_tree as children.
11324         (vect_get_place_in_interleaving_chain): Remove.
11325         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11326         Move ...
11327         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11328         and make static.
11329         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11330         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11331         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11332         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11333         Use slp_node instead of slp_void_p and adjust.
11335 2013-04-09  Richard Biener  <rguenther@suse.de>
11337         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11338         work that is not necessary.
11340 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11342         PR tree-optimization/56854
11343         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11344         forward into clobber stmts if it would change MEM_REF lhs into
11345         non-MEM_REF.
11347 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11349         * tree.c (type_hash_lookup, type_hash_add): Make static.
11350         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11352 2013-04-09  Richard Biener  <rguenther@suse.de>
11354         * tree.h (unsave_expr_now): Remove.
11355         * tree-inline.c (mark_local_for_remap_r): Remove.
11356         (unsave_expr_1): Likewise.
11357         (unsave_r): Likewise.
11358         (unsave_expr_now): Likewise.
11359         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11360         (propagate_tree_value): Likewise.
11362 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11364         * doc/rtl.texi (sequence): Rewrite documentation to match the
11365         current use of SEQUENCE rtl objects.
11366         * rtl.def (SEQUENCE): Likewise.
11368         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11369         Update documentation.
11370         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11371         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11373         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11375 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11377         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11378         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11379         methods.
11380         (estimate_edge_size_and_time): Add comment to suggest using rounding
11381         methods.
11382         (estimate_node_size_and_time): Ditto.
11383         (remap_edge_change_prob): Use helper rounding divide methods.
11384         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11385         (gimple_mod_pow2_value_transform): Ditto.
11386         (gimple_mod_subtract_transform): Ditto.
11387         (gimple_ic_transform): Ditto.
11388         (gimple_stringops_transform): Ditto.
11389         * stmt.c (conditional_probability): Ditto.
11390         (emit_case_dispatch_table): Ditto.
11391         * lto-cgraph.c (merge_profile_summaries): Ditto.
11392         * tree-optimize.c (execute_fixup_cfg): Ditto.
11393         * cfgcleanup.c (try_forward_edges): Ditto.
11394         * cfgloopmanip.c (scale_loop_profile): Ditto.
11395         (loopify): Ditto.
11396         (duplicate_loop_to_header_edge): Ditto.
11397         (lv_adjust_loop_entry_edge): Ditto.
11398         * tree-vect-loop.c (vect_transform_loop): Ditto.
11399         * profile.c (compute_branch_probabilities): Ditto.
11400         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11401         * lto-streamer-in.c (input_cfg): Ditto.
11402         * gimple-streamer-in.c (input_bb): Ditto.
11403         * ipa-cp.c (update_profiling_info): Ditto.
11404         (update_specialized_profile): Ditto.
11405         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11406         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11407         rounding methods.
11408         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11409         (compute_trg_info): Ditto.
11410         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11411         (purge_dead_edges): Ditto.
11412         * loop-unswitch.c (unswitch_loop): Ditto.
11413         * cgraphclones.c (cgraph_clone_edge): Ditto.
11414         (cgraph_clone_node): Ditto.
11415         * tree-inline.c (copy_bb): Ditto.
11416         (copy_edges_for_bb): Ditto.
11417         (initialize_cfun): Ditto.
11418         (copy_cfg_body): Ditto.
11419         (expand_call_inline): Ditto.
11421 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11423         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11424         TARGET_CYGWIN64 by TARGET_64BIT.
11426 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11428         * config/epiphany/epiphany.md (GPR_1): New constant.
11429         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11430         * config/epiphany/epiphany.c (gen_compare_reg):
11431         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11432         is already in place.
11433         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11434         Don't require being called during rtl expansion; If y operlaps r0,
11435         return 0.
11436         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11437         (epiphany_expand_epilogue): Likewise.
11439         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11440         Don't use CC_FPmode for ORDERED / UNORDERED.
11441         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11443         * config/epiphany/constraints.md (CnL): New constraint.
11444         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11445         * config/epiphany/predicates.md (add_operand): Allow 1024.
11447         * config/epiphany/epiphany.md (logical_op): New code iterator.
11448         (op_mnc): New code attribute.
11449         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11450         (mov_f+1, mov_f+2): New peephole2 patterns.
11452         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11453         (cstoresi4): Also allow re-use of zero result when doing a NE
11454         comparison to a non-zero operand.
11455         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11457         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11458         Use gen_addsi3_i / gen_subsi3_i.
11460 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11462         PR c++/34949
11463         PR c++/50243
11464         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11465         contain anything but clobbers, at most one __builtin_stack_restore,
11466         optionally debug stmts and final resx, and if it has at least one
11467         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11468         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11469         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11470         which isn't defaut definition, remove them.
11471         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11472         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11473         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11474         with MEM_REF LHS with SSA_NAME address.
11476 2013-04-08  Jeff Law  <law@redhat.com>
11478         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11480 2013-04-08  Richard Biener  <rguenther@suse.de>
11482         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11483         extra newline.
11484         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11485         determined vector type.
11486         (vect_analyze_data_refs): Likewise.
11487         (vect_get_new_vect_var): Adjust.
11488         (vect_create_destination_var): Preserve SSA name versions.
11489         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11490         not dump anything here.
11492 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11494         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11495         Add member lr_slot_known.
11496         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11497         if necessary.
11498         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11499         Remove code that sets lr_slot_offset according to what a previous
11500         version of epiphany_emit_save_restore used to do.
11501         (epiphany_emit_save_restore): When doing an lr save or restore,
11502         set/verify lr_slot_known and lr_slot_offset.
11504 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11506         PR target/54338
11507         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11508         in ALL_REGS.
11510 2013-04-08  Richard Biener  <rguenther@suse.de>
11512         * alias.c (find_base_term): Fix thinko in previous change.
11514 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11516         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11517         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11518         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11519         if possible to compute val.
11520         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11521         For QImode integers don't require anything about precision.  Use
11522         const_with_all_bytes_same to find out if the constant doesn't have
11523         repeated bytes in it.
11525 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11527         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11528         within mode size.
11530 2013-04-08  Marek Polacek  <polacek@redhat.com>
11532         PR rtl-optimization/48182
11533         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11534         value to 1.
11536 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11538         PR target/55487
11539         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11540         nuses, make sure we have a label.
11542 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11544         PR target/56843
11545         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11546         (rs6000_emit_swdiv_low_precision): Remove.
11547         (rs6000_emit_swdiv): Rewrite to handle between one and four
11548         iterations of Newton-Raphson generally; modify required number of
11549         iterations for some cases.
11550         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11552 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
11554         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11555         set-but-unused variable.
11557         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11558         basic blocks of released function bodies garbage-collectable.
11560         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11561         (struct rtl_opt_pass): Add TODO_df_finish.
11563         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11565 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11567         * config/arm/constraints.md (q): New constraint.
11568         * config/arm/ldrdstrd.md: New file.
11569         * config/arm/arm.md (ldrdstrd.md) New include.
11570         (arm_movdi): Use "q" instead of "r" constraint
11571         for double-word memory access.
11572         (movdf_soft_insn): Likewise.
11573         * config/arm/vfp.md (movdi_vfp): Likewise.
11574         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11575         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11576         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11577         (mem_ok_for_ldrd_strd): Likewise.
11578         (output_move_double): Update assertion.
11580 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11582         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11584 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11586         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11587         define_insn_and_split.
11588         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11590 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11592         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11593         define_insn_and_split.
11594         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11595         (shiftsi3_compare): New pattern.
11596         (rrx): New pattern.
11597         * config/arm/unspecs.md (UNSPEC_RRX): New.
11599 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11601         * config/arm/arm.md (negdi_extendsidi): New pattern.
11602         (negdi_zero_extendsidi): Likewise.
11604 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11606         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11607         define_insn_and_split.
11608         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11609         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11611 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11613         * config/arm/arm.md (arm_subdi3): Convert define_insn into
11614         define_insn_and_split.
11615         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11616         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11618 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11620         * config/arm/arm.md (subsi3_carryin): New pattern.
11621         (subsi3_carryin_const): Likewise.
11622         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11623         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11625 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11627         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11629 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11631         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
11632         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
11634 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11636         * config/arm/arm.c (arm_expand_builtin): Change fcode
11637         type to unsigned int.
11639 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11641         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11643 2013-04-04  Ian Lance Taylor  <iant@google.com>
11645         * doc/standards.texi (Standards): The Go frontend supports the Go 1
11646         language standard.
11648 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
11650         PR middle-end/56729
11651         * df-scan.c (df_insn_delete): Disable failing assert.
11653 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11655         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11656         New function prototype.
11657         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11658         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11659         (arm_builtin_vectorized_function): New function.
11661 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11663         * config/arm/arm_neon_builtins.def: New file.
11664         * config/arm/arm.c (neon_builtin_data): Move contents to
11665         arm_neon_builtins.def.
11666         (enum arm_builtins): Include neon builtin definitions.
11667         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
11668         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
11670 2013-04-04  Marek Polacek  <polacek@redhat.com>
11672         PR tree-optimization/48186
11673         * predict.c (maybe_hot_frequency_p): Return false if
11674         HOT_BB_FREQUENCY_FRACTION is 0.
11675         (cgraph_maybe_hot_edge_p): Likewise.
11677 2013-04-04  Richard Biener  <rguenther@suse.de>
11679         PR tree-optimization/56826
11680         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11681         more accurately.
11683 2013-04-04  Richard Biener  <rguenther@suse.de>
11685         PR tree-optimization/56213
11686         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
11687         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
11689 2013-04-04  Richard Biener  <rguenther@suse.de>
11691         PR tree-optimization/56837
11692         * tree-loop-distribution.c (classify_partition): For non-zero
11693         values require that the value has the same precision as its
11694         mode to be useful as memset value.
11696 2013-04-03  Nick Clifton  <nickc@redhat.com>
11698         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
11699         (fmssf4): Use fmsf.s on E3V5 architectures.
11700         (fnmasf4): Use fnmaf.s on E3V5 architectures.
11701         (fnmssf4): Use fnmsf.s on E3V5 architectures.
11703 2013-04-03  Jeff Law  <law@redhat.com>
11705         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11706         (lra-eliminations.o): Likewise.
11708 2013-04-03  Teresa Johnson  <tejohnson@google.com>
11710         * gcov-io.c (compute_working_sets): Moved most of body of old
11711         compute_working_sets here from profile.c.
11712         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
11713         (gcov_working_set_t): Moved typedef here from basic-block.h
11714         (compute_working_set): Declare.
11715         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11716         (get_working_sets): Renamed from compute_working_set,
11717         replace most of body with call to new compute_working_sets.
11718         (get_exec_counts): Replace call to compute_working_sets
11719         to get_working_sets.
11720         * profile.h (get_working_sets): Renamed from compute_working_set.
11721         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
11722         to get_working_sets.
11723         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11724         * gcov-dump.c (dump_working_sets): New function.
11726 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
11728         * hwint.c (sext_hwi, zext_hwi): New functions.
11729         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11730         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11731         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11732         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11733         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11734         (sext_hwi, zext_hwi): New functions.
11736 2013-04-03  Jeff Law  <law@redhat.com>
11738         PR tree-optimization/56799
11739         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11740         back test for widening conversion erroneously dropped in prior change.
11742 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11744         PR target/56809
11745         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11746         instead of next_real_insn.
11748 2013-04-03  Marek Polacek  <polacek@redhat.com>
11750         PR sanitizer/55702
11751         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
11753 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11755         PR target/56809
11756         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11757         next_real_insn.
11758         (thumb1_output_casesi): Likewise.
11759         (thumb2_output_casesi): Likewise.
11761 2013-04-03  Richard Biener  <rguenther@suse.de>
11763         PR tree-optimization/56817
11764         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11765         Split out ...
11766         (tree_unroll_loops_completely_1): ... new function to manually
11767         walk the loop tree, properly defering outer loops of unrolled
11768         loops to later iterations.
11770 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11772         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11773         (vectorizable_load): Likewise.
11774         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11775         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11777 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11779         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11780         BIT_FIELD_REF.
11782 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
11784         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11786 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
11788         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11790 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11792         PR tree-optimization/56790
11793         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11794         folding.
11796 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11798         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11799         Handle VEC_MERGE.
11800         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11801         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11802         equal arguments.
11804 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
11806         PR c/19449
11807         * tree.h (force_folding_builtin_constant_p): New decl.
11808         * builtins.c (force_folding_builtin_constant_p): New variable.
11809         (fold_builtin_constant_p): Fold immediately also if
11810         force_folding_builtin_constant_p.
11812 2013-04-03  Richard Biener  <rguenther@suse.de>
11814         PR tree-optimization/56812
11815         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11816         DRs of the same interleaving chain are independent.
11818 2013-04-02  Jason Merrill  <jason@redhat.com>
11820         * gdbinit.in (pbb): Use debug fn.
11822 2013-04-02  Lawrence Crowl  <crowl@google.com>
11824         * sese.h (struct ivtype_map_elt_s): Remove unused.
11825         (extern debug_ivtype_map): Remove unused.
11826         (extern eq_ivtype_map_elts): Remove unused.
11827         * sese.c (debug_ivtype_map): Removed unused.
11828         (debug_ivtype_map_1): Removed unused.
11829         (debug_ivtype_elt): Remove unused.
11830         (eq_ivtype_map_elts): Remove unused.
11833 2013-04-02  Kai Tietz  <ktietz@redhat.com>
11835         PR target/52790
11836         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11837         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
11838         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11839         function.
11840         (legitimize_pe_coff_symbol): Likewise.
11841         (is_imported_p): New helper-function.
11842         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11843         for Windows x64 targets.
11844         (ix86_expand_prologue): Optimize for pe-coff targets.
11845         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11846         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11847         medium/large code-model.
11848         (legitimize_pic_address): Likewise.
11849         (legitimize_tls_address): Likewise.
11850         (ix86_expand_call): Likewise.
11851         (x86_output_mi_thunk): Likewise.
11852         (get_dllimport_decl): Add new beimport argument.
11853         (construct_plt_address): Don't assert for x64 pe-coff targets.
11854         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11855         targets.
11856         (SYMBOL_FLAG_STUBVAR): New macro.
11857         (SYMBOL_REF_STUBVAR_P): Likewise.
11858         * config/i386/winnt.c (stub_list): New structure.
11859         (stub_head): New local variable.
11860         (i386_pe_record_stub): New function.
11861         (i386_pe_file_end): Emit refptr-stubs.
11863 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
11865         PR rtl-optimization/56745
11866         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11867         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11869         PR c++/34949
11870         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11871         and both of them are MEM_REFs, just compare first argument for
11872         equality and attempt to deal even with differing offsets.
11874         PR c++/34949
11875         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11876         of gimple_clobber_p to be MEM_REF.
11877         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11878         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
11879         after gimplification.
11880         * asan.c (get_mem_ref_of_assignment): Don't instrument
11881         gimple_clobber_p stmts.
11882         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11883         gimple_clobber_p stmt if they have MEM_REF lhs and
11884         are dead because of another gimple_clobber_p stmt.
11885         * tree-ssa-live.c (clear_unused_block_pointer): Treat
11886         gimple_clobber_p stmts like debug stmts.
11887         (remove_unused_locals): Remove clobbers with MEM_REF lhs
11888         that refer to unused VAR_DECLs or uninitialized values.
11889         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
11890         gimple_clobber_p stmts if they refer to removed parameters.
11891         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
11892         formatting.
11894 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
11896         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
11897         using SWI48 mode attribute.
11899 2013-04-02  Wei Mi  <wmi@google.com>
11901         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
11902         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
11903         *<rotate_insn><mode>3_mask in i386.md.
11905 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11907         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
11909 2013-04-02  Richard Biener  <rguenther@suse.de>
11911         PR tree-optimization/56778
11912         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11913         Runtime alias tests are not supported for gather loads.
11914         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
11915         stmts referenced from SSA operands before updating SSA form.
11917 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
11918             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11920         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
11921         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
11922         * config/arm/cortex-a53.md: New file.
11923         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
11924         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
11925         * config/arm/arm.c (arm_issue_rate): Likewise.
11926         * config/arm/arm-tune.md: Regenerate
11927         * config/arm/arm-tables.opt: Regenerate.
11928         * config/arm/arm-cores.def: Add cortex-a53.
11930 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
11932         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
11933         non-static link.
11935 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11937         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
11938         scalar load/store operations using B/H registers.
11939         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
11941 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11943         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
11944         scalar move.
11945         * config/aarch64/aarch64.c
11946         (aarch64_simd_scalar_immediate_valid_for_move): New.
11947         * config/aarch64/aarch64-protos.h
11948         (aarch64_simd_scalar_immediate_valid_for_move): New.
11949         * config/aarch64/constraints.md (Dh, Dq): New.
11950         * config/aarch64/iterators.md (hq): New.
11952 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
11954         * reorg.c (get_branch_condition): Deal with conditional returns.
11955         (fill_simple_delay_slots): Remove dead code dealing with jumps.
11957 2013-04-01  Wei Mi  <wmi@google.com>
11959         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
11960         Truncate operand 2 using %b asm operand modifier.
11961         (*<shift_insn><mode>3_mask): Ditto.
11962         (*<rotate_insn><mode>3_mask): Ditto.
11964 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
11966         PR middle-end/56798
11967         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
11969 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
11971         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
11972         of next_real_insn.
11973         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
11975 2013-03-30  Lawrence Crowl  <crowl@google.com>
11977         * dse.c (clear_alias_sets): Remove never set.
11978         (disqualified_clear_alias_sets): Remove never set.
11979         (clear_alias_mode_pool): Remove never set.
11980         (dse_step0): Remove condition that is never true.
11981         (canon_address): Remove condition that is never true.
11982         (dse_step7): Remove condition that is never true.
11983         (rest_of_handle_dse): Remove condition that is never true.
11984         (rest_of_handle_dse::did_global): Remove never read from above.
11985         (dse_step2_spill): Remove never called from above.
11986         (dse_step5_spill): Remove never called from above.
11988 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
11990         * doc/md.texi (Standard Names) <casesi>: Update documentation for
11991         JUMP_TABLE_DATA changes.
11992         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
11993         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
11994         (Insns) <jump_table_data>: New entry.
11995         * doc/tm.texi: Regenerate.
11997         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
11999         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12000         for table jump at the end of a basic block using tablejump_p.
12001         * targhooks.c (default_invalid_within_doloop): Likewise.
12002         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12003         target hook implementation that is identical to the default hook.
12004         (rs6000_invalid_within_doloop): Remove.
12006         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12007         unused variable from tablejump_p call.
12009         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12010         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12011         (INSN_DELETED_P): Likewise.
12012         (emit_jump_table_data): New prototype.
12013         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12014         after 4th as unused.
12015         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12016         * sched-vis.c (print_insn): Likewise.
12017         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12018         insn for compatibility with back ends that use next_active_insn to
12019         identify jump table data.
12020         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12021         (remove_insn): Likewise.
12022         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12023         to be emitted.
12024         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12025         (emit_jump_table_data): New function.
12027         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12028         basic block, a JUMP_TABLE_DATA never is.
12029         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12030         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12031         off from code handling real insns.
12032         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12033         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12034         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12035         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12036         is not a NONDEBUG_INSN_P.
12037         * ira-costs.c (scan_one_insn): Likewise.
12038         * jump.c (mark_all_labels): Likewise.
12039         (mark_jump_label_1): Likewise.
12040         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12041         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12042         a basic block.
12043         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12044         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12045         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12046         JUMP_TABLE_DATA_P insns.
12047         (calculate_elim_costs_all_insns): Likewise.
12048         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12049         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12050         (delete_output_reload): Code style fixups.
12051         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12052         insn flags on this non-insn.
12053         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12054         as scheduling barriers, for pre-change compatibility.
12055         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12056         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12058         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12059         redundant JUMP_TABLE_DATA_P test.
12060         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12061         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12062         (frv_for_each_packet): Likewise.
12063         * config/i386/i386.c (min_insn_size): Likewise.
12064         (ix86_avoid_jump_mispredicts): Likewise.
12065         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12066         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12067         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12068         (mips16_insn_length): Robustify.
12069         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12070         (mips16_split_long_branches): Likewise.
12071         * config/pa/pa.c (pa_combine_instructions): Likewise.
12072         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12073         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12074         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12075         as contributing to pool range lengths.
12076         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12077         Remove redundant JUMP_TABLE_DATA_P test.
12078         (sh_loop_align): Likewise.
12079         (split_branches): Likewise.
12080         (sh_insn_length_adjustment): Likewise.
12081         * config/spu/spu.c (get_branch_target): Likewise.
12083 2013-03-29  Jan Hubicka  <jh@suse.cz>
12085         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12086         gcov streaming; stream hot bb threshold to ltrans.
12087         * predict.c (get_hot_bb_threshold): Break out from ....
12088         (maybe_hot_count_p): ... here.
12089         (set_hot_bb_threshold): New function.
12090         * lto-section-in.c (lto_section_name): Add profile.
12091         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12092         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12093         and data-streamer.h
12094         (histogram_entry): New structure.
12095         (histogram, histogram_pool): New global vars.
12096         (histogram_hash): New structure.
12097         (histogram_hash::hash): New method.
12098         (histogram_hash::equal): Likewise.
12099         (account_time_size): New function.
12100         (cmp_counts): New function.
12101         (dump_histogram): New function.
12102         (ipa_profile_generate_summary): New function.
12103         (ipa_profile_write_summary): New function.
12104         (ipa_profile_read_summary): New function.
12105         (ipa_profile): Decide on threshold.
12106         (pass_ipa_profile): Add ipa_profile_write_summary and
12107         ipa_profile_read_summary.
12108         * Makefile.in (ipa.o): Update dependencies.
12109         * lto-streamer.h (LTO_section_ipa_profile): New section.
12111 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12113         * tree.h (VAR_P): New.
12115 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12117         PR lto/56777
12118         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12120 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12122         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12123         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12124         (control_flow_insn_p): Likewise.
12125         * cfgrtl.c (duplicate_insn_chain): Likewise.
12126         * final.c (get_attr_length_1): Likewise.
12127         (shorten_branches): Likewise.
12128         (final_scan_insn): Likewise.
12129         * function.c (instantiate_virtual_regs): Likewise.
12130         * gcse.c (insert_insn_end_basic_block): Likewise.
12131         * ira-costs.c (scan_one_insn): Likewise.
12132         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12133         * lra.c (check_rtl): Likewise.
12134         * reload1.c (elimination_costs_in_insn): Likewise.
12135         * reorg.c (follow_jumps): Likewise.
12137         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12138         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12139         (thumb_far_jump_used_p): Likewise.
12140         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12141         (workaround_speculation): Likewise.
12142         (add_sched_insns_for_speculation): Likewise.
12143         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12144         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12145         (frv_for_each_packet): Likewise.
12146         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12147         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12148         (final_emit_insn_group_barriers): Likewise.
12149         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12150         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12151         (mips16_insn_length): Likewise.
12152         * config/pa/pa.c (pa_reorg): Likewise.
12153         (pa_combine_instructions): Likewise.
12154         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12155         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12156         (sh_reorg): Likewise.
12157         (split_branches): Likewise.
12158         * config/spu/spu.c (get_branch_target): Likewise.
12160         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12161         JUMP_TABLE_DATA_P.
12163 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12165         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12166         Fix declaration name.
12168 2013-03-28  Lawrence Crowl  <crowl@google.com>
12170         * graphds.h (struct graph.indicies): Remove unused.
12171         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12172         (SCOP_ORIGINAL_PDDRS): Remove unused.
12173         * sese.h (extern insert_loop_close_phis): Removed unused.
12174         (extern insert_guard_phis): Removed unused.
12175         (extern ivtype_map_elt_info): Removed unused.
12176         (new_ivtype_map_elt): Removed unused.
12177         * sese.c (ivtype_map_elt_info): Removed unused.
12179 2013-03-28  Lawrence Crowl  <crowl@google.com>
12181         * Makefile.in: Add several missing include dependences.
12182         (DUMPFILE_H): New.
12183         (test-dump.o): New.  This object is not added to any executable,
12184         but is present for ad-hoc testing.
12185         * bitmap.c
12186         (debug (const bitmap_head_def &)): New.
12187         (debug (const bitmap_head_def *)): New.
12188         * bitmap.h
12189         (extern debug (const bitmap_head_def &)): New.
12190         (extern debug (const bitmap_head_def *)): New.
12191         * cfg.c
12192         (debug (edge_def &)): New.
12193         (debug (edge_def *)): New.
12194         * cfghooks.c
12195         (debug (basic_block_def &)): New.
12196         (debug (basic_block_def *)): New.
12197         * dumpfile.h
12198         (dump_node (const_tree, int, FILE *)): Correct source file.
12199         * dwarf2out.c
12200         (debug (die_struct &)): New.
12201         (debug (die_struct *)): New.
12202         * dwarf2out.h
12203         (extern debug (die_struct &)): New.
12204         (extern debug (die_struct *)): New.
12205         * gimple-pretty-print.c
12206         (debug (gimple_statement_d &)): New.
12207         (debug (gimple_statement_d *)): New.
12208         * gimple-pretty-print.h
12209         (extern debug (gimple_statement_d &)): New.
12210         (extern debug (gimple_statement_d *)): New.
12211         * ira-build.c
12212         (debug (ira_allocno_copy &)): New.
12213         (debug (ira_allocno_copy *)): New.
12214         (debug (ira_allocno &)): New.
12215         (debug (ira_allocno *)): New.
12216         * ira-int.h
12217         (extern debug (ira_allocno_copy &)): New.
12218         (extern debug (ira_allocno_copy *)): New.
12219         (extern debug (ira_allocno &)): New.
12220         (extern debug (ira_allocno *)): New.
12221         * ira-lives.c
12222         (debug (live_range &)): New.
12223         (debug (live_range *)): New.
12224         * lra-int.h
12225         (debug (lra_live_range &)): New.
12226         (debug (lra_live_range *)): New.
12227         * lra-lives.c
12228         (debug (lra_live_range &)): New.
12229         (debug (lra_live_range *)): New.
12230         * omega.c
12231         (debug (omega_pb_d &)): New.
12232         (debug (omega_pb_d *)): New.
12233         * omega.h
12234         (extern debug (omega_pb_d &)): New.
12235         (extern debug (omega_pb_d *)): New.
12236         * print-rtl.c
12237         (debug (const rtx_def &)): New.
12238         (debug (const rtx_def *)): New.
12239         * print-tree.c
12240         (debug_tree (tree): Move within file.
12241         (debug_raw (const tree_node &)): New.
12242         (debug_raw (const tree_node *)): New.
12243         (dump_tree_via_hooks (const tree_node *, int)): New.
12244         (debug (const tree_node &)): New.
12245         (debug (const tree_node *)): New.
12246         (debug_verbose (const tree_node &)): New.
12247         (debug_verbose (const tree_node *)): New.
12248         (debug_head (const tree_node &)): New.
12249         (debug_head (const tree_node *)): New.
12250         (debug_body (const tree_node &)): New.
12251         (debug_body (const tree_node *)): New.
12252         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12253         (debug (vec<tree, va_gc> &)): New.
12254         (debug (vec<tree, va_gc> *)): New.
12255         * rtl.h
12256         (extern debug (const rtx_def &)): New.
12257         (extern debug (const rtx_def *)): New.
12258         * sbitmap.c
12259         (debug_raw (simple_bitmap_def &)): New.
12260         (debug_raw (simple_bitmap_def *)): New.
12261         (debug (simple_bitmap_def &)): New.
12262         (debug (simple_bitmap_def *)): New.
12263         * sbitmap.h
12264         (extern debug (simple_bitmap_def &)): New.
12265         (extern debug (simple_bitmap_def *)): New.
12266         (extern debug_raw (simple_bitmap_def &)): New.
12267         (extern debug_raw (simple_bitmap_def *)): New.
12268         * sel-sched-dump.c
12269         (debug (vinsn_def &)): New.
12270         (debug (vinsn_def *)): New.
12271         (debug_verbose (vinsn_def &)): New.
12272         (debug_verbose (vinsn_def *)): New.
12273         (debug (expr_def &)): New.
12274         (debug (expr_def *)): New.
12275         (debug_verbose (expr_def &)): New.
12276         (debug_verbose (expr_def *)): New.
12277         (debug (vec<rtx> &)): New.
12278         (debug (vec<rtx> *)): New.
12279         * sel-sched-dump.h
12280         (extern debug (vinsn_def &)): New.
12281         (extern debug (vinsn_def *)): New.
12282         (extern debug_verbose (vinsn_def &)): New.
12283         (extern debug_verbose (vinsn_def *)): New.
12284         (extern debug (expr_def &)): New.
12285         (extern debug (expr_def *)): New.
12286         (extern debug_verbose (expr_def &)): New.
12287         (extern debug_verbose (expr_def *)): New.
12288         (extern debug (vec<rtx> &)): New.
12289         (extern debug (vec<rtx> *)): New.
12290         * sel-sched-ir.h
12291         (_list_iter_cond_expr): Make inline instead of static.
12292         * sreal.c
12293         (debug (sreal &)): New.
12294         (debug (sreal *)): New.
12295         * sreal.h
12296         (extern debug (sreal &)): New.
12297         (extern debug (sreal *)): New.
12298         * tree.h
12299         (extern debug_raw (const tree_node &)): New.
12300         (extern debug_raw (const tree_node *)): New.
12301         (extern debug (const tree_node &)): New.
12302         (extern debug (const tree_node *)): New.
12303         (extern debug_verbose (const tree_node &)): New.
12304         (extern debug_verbose (const tree_node *)): New.
12305         (extern debug_head (const tree_node &)): New.
12306         (extern debug_head (const tree_node *)): New.
12307         (extern debug_body (const tree_node &)): New.
12308         (extern debug_body (const tree_node *)): New.
12309         (extern debug (vec<tree, va_gc> &)): New.
12310         (extern debug (vec<tree, va_gc> *)): New.
12311         * tree-cfg.c
12312         (debug (struct loop &)): New.
12313         (debug (struct loop *)): New.
12314         (debug_verbose (struct loop &)): New.
12315         (debug_verbose (struct loop *)): New.
12316         * tree-dump.c: Add header dependence.
12317         * tree-flow.h
12318         (extern debug (struct loop &)): New.
12319         (extern debug (struct loop *)): New.
12320         (extern debug_verbose (struct loop &)): New.
12321         (extern debug_verbose (struct loop *)): New.
12322         * tree-data-ref.c
12323         (debug (data_reference &)): New.
12324         (debug (data_reference *)): New.
12325         (debug (vec<data_reference_p> &)): New.
12326         (debug (vec<data_reference_p> *)): New.
12327         (debug (vec<ddr_p> &)): New.
12328         (debug (vec<ddr_p> *)): New.
12329         * tree-data-ref.h
12330         (extern debug (data_reference &)): New.
12331         (extern debug (data_reference *)): New.
12332         (extern debug (vec<data_reference_p> &)): New.
12333         (extern debug (vec<data_reference_p> *)): New.
12334         (extern debug (vec<ddr_p> &)): New.
12335         (extern debug (vec<ddr_p> *)): New.
12336         * tree-ssa-alias.c
12337         (debug (pt_solution &)): New.
12338         (debug (pt_solution *)): New.
12339         * tree-ssa-alias.h
12340         (extern debug (pt_solution &)): New.
12341         (extern debug (pt_solution *)): New.
12342         * tree-ssa-alias.c
12343         (debug (_var_map &)): New.
12344         (debug (_var_map *)): New.
12345         (debug (tree_live_info_d &)): New.
12346         (debug (tree_live_info_d *)): New.
12347         * tree-ssa-alias.h
12348         (extern debug (_var_map &)): New.
12349         (extern debug (_var_map *)): New.
12350         (extern debug (tree_live_info_d &)): New.
12351         (extern debug (tree_live_info_d *)): New.
12353 2013-03-28  Jan Hubicka  <jh@suse.cz>
12355         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12357 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12359         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12360         record only when desired or required.
12362 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12364         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12365         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12367 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12369         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12370         (*andsi3_compare0_uxtw): New pattern.
12371         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12372         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12374 2013-03-28  Jan Hubicka  <jh@suse.cz>
12376         * data-streamer-in.c (streamer_read_gcov_count): New function.
12377         * gimple-streamer-out.c: Include value-prof.h.
12378         (output_gimple_stmt): Output histogram.
12379         (output_bb): Use streamer_write_gcov_count.
12380         * value-prof.c: Include data-streamer.h
12381         (dump_histogram_value): Add HIST_TYPE_MAX.
12382         (stream_out_histogram_value): New function.
12383         (stream_in_histogram_value): New function.
12384         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12385         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12386         * data-streamer-out.c (streamer_write_gcov_count): New function.
12387         (streamer_write_gcov_count_stream): New function.
12388         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12389         (lto_output_node): Likewise.
12390         (input_node, input_edge): Likewise.
12391         * lto-streamer-out.c (output_cfg): Update streaming.
12392         * lto-streamer-in.c (input_cfg): Likewise.
12393         * data-streamer.h (streamer_write_gcov_count,
12394         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12395         * gimple-streamer-in.c: Include value-prof.h
12396         (input_gimple_stmt): Input histograms.
12397         (input_bb): Update profile streaming.
12399 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12401         * genmodes.c (emit_max_int): New function.
12402         (emit_insn_modes_h): Added call to emit_max_function.
12403         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12404         Added doc.
12405         * machmode.def: Fixed comment.
12407 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12409         * combine.c (try_combine): Removed useless assert.
12410         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12412 2013-03-28  Marek Polacek  <polacek@redhat.com>
12413             Richard Biener  <rguenther@suse.de>
12415         PR tree-optimization/56695
12416         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12417         build signed result of a vector comparison.
12418         * tree-cfg.c (verify_gimple_comparison): Check that a result
12419         of a vector comparison has signed type.
12421 2013-03-28  Richard Biener  <rguenther@suse.de>
12423         PR tree-optimization/37021
12424         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12425         do not restrict gaps between groups.
12426         * tree-vect-stmts.c (vectorizable_load): Properly account for
12427         a gap between groups.
12429 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12431         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12432         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12433         is not enabled.
12435 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12437         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12438         * doc/extend.texi (Named Address Spaces): Ditto.
12439         (Variable Attributes): Ditto.
12441 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12443         * config.build: Add support for cygwin x64 target.
12444         * config.gcc: Likewise.
12445         * config.host: Likewise.
12446         * configure.ac: Likewise
12447         * configure: Regenerated.
12449 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12451         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12452         * config/i386/t-cygwin-w64: New file.
12453         * config/i386/cygwin-w64.h: New file.
12454         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12455         and add support for x64-cygwin target.
12456         (CPP_SPEC): Likewise.
12457         (CXX_WRAP_SPEC_LIST): Undefine before define.
12458         (LIBGCJ_SONAME): Use 15 as version.
12460 2013-03-27  Richard Biener  <rguenther@suse.de>
12462         PR tree-optimization/56716
12463         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12464         dumping for ref nodes.
12466 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12468         PR tree-optimization/55334
12469         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12470         restricted pointers to arrays.
12472 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12474         * Makefile.in (.SUFFIXES): Add .cc.
12475         (.c.o): Apply same recipe for implicit rule .cc.o.
12477 2013-03-27  Richard Biener  <rguenther@suse.de>
12479         PR tree-optimization/37021
12480         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12481         REALPART/IMAGPART_EXPRs around the supported refs.
12482         * tree-ssa-structalias.c (find_func_aliases): Assume that
12483         floating-point values are not used to transfer pointers.
12485 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12487         * target.def (TARGET_HAS_IFUNC_P): New target hook.
12488         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12489         * doc/tm.texi: Regenerate.
12490         * targhooks.h (default_has_ifunc_p): New.
12491         * targhooks.c (default_has_ifunc_p): Ditto.
12492         * config/linux-protos.h: New file.
12493         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12494         this hook for linux which disables support of indirect functions in
12495         android.
12496         * config/linux-android.c: New file.
12497         * config/t-linux-android.c: Ditto.
12498         * config.gcc: Added new object file linux-android.o.
12499         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
12500         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12501         * varasm.c (do_assemble_alias): Likewise.
12502         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12503         doesn't support indirect functions.
12504         * configure: Regenerate.
12506 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12508         PR target/56102
12509         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12510         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12511         mult-word mode.
12513 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12515         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12517 2013-03-27  Terry Guo  <terry.guo@arm.com>
12519         * config/arm/arm-cores.def: Added core cortex-r7.
12520         * config/arm/arm-tune.md: Regenerated.
12521         * config/arm/arm-tables.opt: Regenerated.
12522         * doc/invoke.texi: Added entry for core cortex-r7.
12524 2013-03-27  Walter Lee  <walt@tilera.com>
12526         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12527         double-decrement of next_scratch_regno.
12529 2013-03-27  Walter Lee  <walt@tilera.com>
12531         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12532         input operands.
12533         (insn_v1mulus): Ditto.
12534         (insn_v2muls): Ditto.
12536 2013-03-27  Walter Lee  <walt@tilera.com>
12538         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12539         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12541 2013-03-27  Walter Lee  <walt@tilera.com>
12543         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12544         (*sibcall_value): Ditto.
12546 2013-03-27  Walter Lee  <walt@tilera.com>
12548         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12549         (insn_mnz_v8qi): ... this ...
12550         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
12551         vector equivalent.
12552         (insn_v<n>mnz): Replaced by ...
12553         (insn_v1mnz): ... this ...
12554         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
12555         equivalent.
12556         (insn_mz_<mode>): Replaced by ...
12557         (insn_mz_v8qi): ... this ...
12558         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
12559         vector equivalent.
12560         (insn_v<n>mz): Replaced by ...
12561         (insn_v1mz): ... this ...
12562         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
12563         equivalent.
12565 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
12567         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12569 2013-03-26  Roland McGrath  <mcgrathr@google.com>
12571         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12572         than fprintf with a non-constant, non-format string.
12574 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12576         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12577         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
12578         operand 0 predicate.
12579         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12580         attribute.  Use general_x64nomem_operand as operand 1 predicate.
12581         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12582         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12583         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12584         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12585         (mov<mode>_insv_1): Remove expander.  Merge insn with
12586         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12587         Use general_x64nomem_operand as operand 1 predicate.
12588         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12589         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12590         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12591         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12592         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
12593         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12594         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12595         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12596         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
12597         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12598         (general_x64nomem_operand): Ditto.
12600 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12602         * config/rtems.opt: Add -pthread option.
12604 2013-03-26  Richard Biener  <rguenther@suse.de>
12606         * alias.c (find_base_term): Avoid redundant and not used recursion.
12607         (base_alias_check): Get the initial base term from the caller.
12608         (true_dependence_1): Compute and pass base terms to base_alias_check.
12609         (write_dependence_p): Likewise.
12610         (may_alias_p): Likewise.
12612 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
12614         * config/aarch64/aarch64.c (aarch64_classify_address): Support
12615         PC-relative load in SI modes and above only.
12617 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
12619         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12620         * config/arm/iwmmxt.md (WCGR0): Update.
12621         (WCGR1, WCGR2, WCGR3): Likewise.
12623 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12625         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12626         Use x64 and nox64 isa attributes.
12628 2013-03-26  Richard Biener  <rguenther@suse.de>
12630         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12631         alignment computations and rely on get_object_alignment_1
12632         for the !TYPE_P case.
12633         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12635 2013-03-26  Walter Lee  <walt@tilera.com>
12637         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12638         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12640 2013-03-25  Jeff Law  <law@redhat.com>
12642         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12643         check for INTEGRAL_TYPE_P that was missing due to checking in
12644         wrong version of prior patch.
12646 2013-03-25  Walter Lee  <walt@tilera.com>
12648         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12649         TILEGX_INSN_SHUFFLEBYTES1.
12650         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12651         shufflebytes1.
12652         (tilegx_builtins): Ditto.
12653         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12655 2013-03-25  Walter Lee  <walt@tilera.com>
12657         * config/tilegx/tilegx.md (floatsisf2): New pattern.
12658         (floatunssisf2): New pattern.
12659         (floatsidf2): New pattern.
12660         (floatunssidf2): New pattern.
12662 2013-03-25  Walter Lee  <walt@tilera.com>
12664         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12665         tests for constraint J, K, N, P.
12667 2013-03-25  Walter Lee  <walt@tilera.com>
12669         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12670         Use indirect/pcrel encoding.
12671         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12672         Ditto.
12674 2013-03-25  Steve Ellcey  <sellcey@mips.com>
12676         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12677         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12678         * config/mips/mips.c (mips_option_override): Set IMADD default.
12679         * config/mips/mips.h (PTF_AVOID_IMADD): New.
12680         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12681         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12682         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12683         * doc/invoke.texi (-mimadd/-mno-imadd): New.
12685 2013-03-25  Jeff Law  <law@redhat.com>
12687         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12688         slightly to avoid creating and folding useless trees.  Simplify
12689         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12691 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
12693         * config/i386/i386.md (*zero_extendsidi2): Merge with
12694         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
12695         * config/i386/predicates.md (x86_64_zext_operand): Rename from
12696         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
12697         targets.  Clarify comment.
12699 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12701         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12702         pass-through jump functions differently.
12703         (ipa_read_jump_function): Likewise.  Also use setter functions to set
12704         up jump functions.
12706 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12708         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12709         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12710         process it.
12711         (ipa_get_indirect_edge_target): New function.
12712         (devirtualization_time_bonus): New parameter known_aggs, pass it to
12713         ipa_get_indirect_edge_target.  Update all callers.
12714         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
12715         ipa_get_indirect_edge_target_1 instead of calling
12716         ipa_get_indirect_edge_target.
12717         (create_specialized_node): Pass aggvlas to
12718         ipcp_discover_new_direct_edges.
12720 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12722         * config/arm/arm.md (f_sels, f_seld): New types.
12723         (*cmov<mode>): New pattern.
12724         * config/arm/predicates.md (arm_vsel_comparison_operator): New
12725         predicate.
12727 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12729         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12730         POSIX-printf for mingw-hosted  builds.
12732 2013-03-25  Richard Biener  <rguenther@suse.de>
12734         PR middle-end/56694
12735         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12736         must-not-throw stmt location.
12738 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12740         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12741         Emit load-acquire versions when acq is true.
12742         (arm_emit_store_exclusive): Add rel parameter.
12743         Emit store-release versions when rel is true.
12744         (arm_split_compare_and_swap): Use acquire-release instructions
12745         instead.
12746         of barriers when appropriate.
12747         (arm_split_atomic_op): Likewise.
12748         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12749         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12750         (VUNSPEC_SLX): Likewise.
12751         (VUNSPEC_LDA): Likewise.
12752         (VUNSPEC_STL): Likewise.
12753         * config/arm/sync.md (atomic_load<mode>): New pattern.
12754         (atomic_store<mode>): Likewise.
12755         (arm_load_acquire_exclusive<mode>): Likewise.
12756         (arm_load_acquire_exclusivesi): Likewise.
12757         (arm_load_acquire_exclusivedi): Likewise.
12758         (arm_store_release_exclusive<mode>): Likewise.
12760 2013-03-25  Catherine Moore  <clm@codesourcery.com>
12762         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12763         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12764         * config/mip/predicates.md (lwsp_swsp_operand,
12765         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12766         sb16_operand, db4_operand, db7_operand, ib3_operand,
12767         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12768         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12769         andi16_operand): New predicates.
12770         * config/mips/mips.md (compression): New attribute.
12771         (enabled): New attribute.
12772         (length): Consider compression in computing length.
12773         (shift_compression): New code attribute.
12774         (*add<mode>3): New operands. Record compression.
12775         (sub<mode>3): Likewise.
12776         (one_cmpl<mode>2): Likewise.
12777         (*and<mode>3): Likewise.
12778         (*ior<mode>3): Likewise.
12779         (unnamed pattern for xor): Likewise.
12780         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12781         (*<optab><mode>3): Likewise.
12782         (*mov<mode>_internal: Likewise.
12783         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12784         (mips_unsigned_immediate_p): New.
12785         (umips_lwsp_swsp_address_p): New.
12786         (m16_based_address_p): New.
12787         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12788         (mips_unsigned_immediate_p): New prototype.
12789         (lwsp_swsp_address_p): New prototype.
12790         (m16_based_address_p): New prototype.
12791         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12792         (mips_signed_immediate_p): New function.
12793         (m16_based_address_p): New function.
12794         (lwsp_swsp_address_p): New function.
12795         (mips_print_operand_punctuation): Recognize short delay slot insns
12796         for microMIPS.add<mode>3"
12798 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12800         PR target/56720
12801         * config/arm/iterators.md (v_cmp_result): New mode attribute.
12802         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12804 2013-03-25  Richard Biener  <rguenther@suse.de>
12806         PR tree-optimization/56689
12807         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12808         any edge.
12810 2013-03-25  Richard Biener  <rguenther@suse.de>
12812         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12813         of bitmap.
12814         (memory_references): Likewise.
12815         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12816         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12817         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12818         (gather_mem_refs_in_loops): Fold into ...
12819         (analyze_memory_references): ... this.  Move initialization
12820         to tree_ssa_lim_initialize.
12821         (fill_always_executed_in): Rename to ...
12822         (fill_always_executed_in_1): ... this.
12823         (fill_always_executed_in): Move contains_call computation to
12824         this new function from ...
12825         (tree_ssa_lim_initialize): ... here.
12826         (tree_ssa_lim): Call fill_always_executed_in.
12828 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12830         * postreload.c (reload_combine): Fix code detecting returns.
12832 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12834         * function.c (emit_use_return_register_into_block): On cc0 targets,
12835         do not emit the sequence between cc0 setter and user.
12837 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12839         * config/i386/predicates.md (local_symbolic_operand): Interpret
12840         dll-imported symbols as none-local.
12842 2013-03-25  Richard Biener  <rguenther@suse.de>
12844         * tree-ssa-loop-im.c (struct depend): Remove.
12845         (struct lim_aux_data): Make depends a vec of gimples.
12846         (free_lim_aux_data): Adjust.
12847         (add_dependency): Likewise.
12848         (set_level): Likewise.
12850 2013-03-25  Richard Biener  <rguenther@suse.de>
12852         PR middle-end/56434
12853         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12854         the pointer returned by calls with ECF_MALLOC set.
12856 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12858         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12860 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12862         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
12863         using MMXMODE mode iterator.
12864         (*move<mode>_internal): Merge with *movv2sf_internal and
12865         *movv2sf_internal_rex64 using MMXMODE mode iterator.
12867 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12869         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12870         (record_last_mem_set_info): Likewise.
12872         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12873         of XNEWVEC followed by memset.
12874         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12876 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12878         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12879         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12880         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12881         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12882         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12883         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
12884         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
12885         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
12886         BARRIER_P instead of GET_CODE.
12888 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12890         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
12891         inaccuracy in the probing code.
12893         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
12894         (ctrapdi4): Likewise.
12896 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12898         * calls.c (expand_call): Add missing guard to code handling return
12899         of non-BLKmode structures in MSB.
12900         * function.c (expand_function_end): Likewise.
12902 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12904         * combine.c (try_combine): Adjust comment.  Do not add the set of
12905         insn #0 if the destination indirectly is set or dies in insn #2.
12906         Tidy up code to distribute a new note.
12908 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12910         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
12911         also for alternatives 16 and 17.
12913 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12915         * config/i386/sse.md (*mov<mode>_internal): Merge with
12916         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
12917         Emit insn template depending on type attribute.  Use
12918         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
12919         movd instead of movq mnemonic for interunit moves.  Rewrite mode
12920         attribute calculation.  Remove unit attribute calculation.
12921         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
12922         Set prefix_data16 attribute for DImode ssemov types.
12923         Use Ym instead of y for SSE-MMX conversion alternatives.
12924         Reorder operand constraints.
12926 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
12928         * df.h (df_insn_delete): Adjust prototype.
12929         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
12930         and let it decide whether mark the basic block dirty.
12931         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
12932         * df-scan.c (df_insn_info_delete): New helper function, split
12933         off from df_insn_delete.
12934         (df_scan_free_bb_info): Use it.
12935         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
12936         Likewise.
12937         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
12938         that the insn is actually an insn and it has a non-NULL basic block.
12939         Do not mark basic block dirty if only deleting a DEBUG_INSN.
12941 2013-03-22  Richard Biener  <rguenther@suse.de>
12943         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
12944         dep_ref members.
12945         (mem_ref_alloc): Do not allocate them.
12946         (refs_independent_p): Do not query or maintain a cache.
12948 2013-03-22  Richard Biener  <rguenther@suse.de>
12950         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
12951         (gather_mem_refs_in_loops): Do not compute it.
12952         (analyze_memory_references): Do not allocate it.
12953         (tree_ssa_lim_finalize): Do not free it.
12954         (for_all_locs_in_loop): Do not query all_refs_in_loop.
12956 2013-03-22  Richard Biener  <rguenther@suse.de>
12958         * is-a.h (as_a): Use gcc_checking_assert.
12960 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
12962         * config/aarch64/aarch64.c (aarch64_print_operand): New
12963         format specifier for printing a constant in hex.
12964         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
12965         format specifier for printing second operand.
12967 2013-03-22  Richard Biener  <rguenther@suse.de>
12969         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
12970         bitmaps.
12971         (gather_mem_refs_in_loops): Perform store accumulation here.
12972         (create_vop_ref_mapping_loop): Remove.
12973         (create_vop_ref_mapping): Likewise.
12974         (analyze_memory_references): Initialize refs_stored_in_loop.
12975         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
12976         (record_indep_loop): Remove.
12977         (record_dep_loop): New function.
12978         (ref_indep_loop_p_1): Adjust to only walk over references
12979         in the loop, not its subloops.
12980         (ref_indep_loop_p): Rename to ...
12981         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
12982         maintaining a more fine-grained cache.
12983         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
12984         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
12986 2013-03-22  Richard Biener  <rguenther@suse.de>
12988         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
12989         (struct mem_ref): Make accesses_in_loop a vec of a vec of
12990         aggregate mem_ref_loc.
12991         (free_mem_ref_locs): Inline into ...
12992         (memref_free): ... this and adjust.
12993         (mem_ref_alloc): Adjust.
12994         (mem_ref_locs_alloc): Remove.
12995         (record_mem_ref_loc): Adjust.
12996         (get_all_locs_in_loop): Rewrite into ...
12997         (for_all_locs_in_loop): ... this iterator.
12998         (rewrite_mem_ref_loc): New functor.
12999         (rewrite_mem_refs): Use for_all_locs_in_loop.
13000         (sm_set_flag_if_changed): New functor.
13001         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13002         (ref_always_accessed): New functor.
13003         (ref_always_accessed_p): Use for_all_locs_in_loop.
13005 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13007         * tree-pass.h (PROP_gimple_lvec): New.
13008         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13009         (init_optimization_passes): Move pass_lower_vector.
13010         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13011         PROP_gimple_lvec.
13012         (pass_lower_vector): Provide PROP_gimple_lvec.
13013         (pass_lower_vector_ssa): Likewise.
13014         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13016 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13018         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13020 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13022         * config/i386/i386.md (*movdi_internal): Disparage slightly
13023         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13024         conversion alternatives.
13026 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13028         PR middle-end/48087
13029         * diagnostic.def (DK_WERROR): New kind.
13030         * diagnostic.h (werrorcount): Define.
13031         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13032         promoted to DK_ERROR, increment DK_WERROR counter instead of
13033         DK_ERROR counter.
13034         * toplev.c (toplev_main): Call print_ignored_options even if
13035         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13036         even if just werrorcount is non-zero.
13038         PR debug/55608
13039         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13040         on failure.
13041         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13042         (string_cst_pool_decl): New function.
13043         (optimize_one_addr_into_implicit_ptr): New function.
13044         (resolve_addr_in_expr): Optimize DWARF location expression
13045         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13046         which doesn't live in memory, but has DW_AT_location or
13047         DW_AT_const_value, or refers to a string literal, into
13048         DW_OP_GNU_implicit_pointer.
13049         (optimize_location_into_implicit_ptr): New function.
13050         (resolve_addr): If removing DW_AT_location of a variable because
13051         it was DW_OP_addr of address of the variable, but the variable doesn't
13052         live in memory, try to emit const value attribute for the initializer.
13054 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13056         * tree.h (VECTOR_TYPE_P): New macro.
13057         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13058         TYPE_MODE): Use it.
13059         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13060         VEC_COND_EXPR cannot be lvalues.
13061         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13063 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13065         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13066         Restrict the transformation to equal modes.
13068 2013-03-21  Richard Biener  <rguenther@suse.de>
13070         PR tree-optimization/39326
13071         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13072         (MEM_ANALYZABLE): Adjust.
13073         (record_mem_ref_loc): Move bitmap ops ...
13074         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13075         unanalyzable refs, do not record locations for it.
13076         (analyze_memory_references): Allocate ref zero as shared
13077         unanalyzable ref.
13078         (refs_independent_p): Do not test for unanalyzed mems here.
13079         (ref_indep_loop_p_1): Special-case disambiguation against
13080         the unanalyzed ref.
13081         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13083 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13085         * config/arm/arm-protos.h (tune_params): Add
13086         prefer_neon_for_64bits field.
13087         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13088         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13089         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13090         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13091         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13092         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13093         (arm_option_override): Handle -mneon-for-64bits new option.
13094         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13095         (prefer_neon_for_64bits): Declare new variable.
13096         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13097         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13098         (arch_enabled): Handle new arch types. Remove support for onlya8
13099         and nota8.
13100         (one_cmpldi2): Use new arch names.
13101         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13102         * config/arm/arm.opt (mneon-for-64bits): Add option.
13103         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13104         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13105         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13106         of onlya8.
13107         * doc/invoke.texi (-mneon-for-64bits): Document.
13109 2013-03-21  Richard Biener  <rguenther@suse.de>
13111         PR tree-optimization/39326
13112         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13113         (sort_bbs_in_loop_postorder_cmp): New function.
13114         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13116 2013-03-21  Richard Biener  <rguenther@suse.de>
13118         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13119         (vect_insert_into_interleaving_chain): Likewise.
13120         (vect_drs_dependent_in_basic_block): Inline ...
13121         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13122         split out from ...
13123         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13124         (vect_check_interleaving): Simplify.
13125         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13126         (vect_slp_analyze_data_ref_dependences): ... this new function.
13127         (dr_group_sort_cmp): New function.
13128         (vect_analyze_data_ref_accesses): Compute data-reference groups
13129         here instead of in vect_analyze_data_ref_dependence.  Use
13130         a more efficient algorithm.
13131         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13132         vect_slp_analyze_data_ref_dependences.  Call
13133         vect_analyze_data_ref_accesses earlier.
13134         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13135         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13136         (vect_slp_analyze_data_ref_dependences): New prototype.
13138 2013-03-21  Richard Biener  <rguenther@suse.de>
13140         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13141         ref is stored in the loop.
13142         (find_refs_for_sm): Walk only over all stores.
13143         (store_motion_loop): Allocate from lim_bitmap_obstack.
13144         (store_motion): Likewise.
13146 2013-03-21  Richard Biener  <rguenther@suse.de>
13148         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13149         Update virtual SSA form.
13151 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13153         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13154         * configure: Regenerate.
13155         * config.in: Regenerate.
13156         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13157         if !HAVE_LD_EH_FRAME_CIEV3.
13159 2013-03-21  Richard Biener  <rguenther@suse.de>
13161         * tree-cfg.c (verify_expr_no_block): New function.
13162         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13163         nor DECL_VALUE_EXPR have locations with associated blocks.
13164         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13165         (clear_unused_block_pointer): Remove code dealing with
13166         blocks in DECL_DEBUG_EXPR locations.
13168 2013-03-21  Richard Biener  <rguenther@suse.de>
13170         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13171         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13172         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13173         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13174         instead of DECL_DEBUG_EXPR_IS_FROM.
13175         * gimplify.c (gimplify_modify_expr): Likewise.
13176         * tree-cfg.c (verify_expr_location_1): Likewise.
13177         * tree-complex.c (create_one_component_var): Likewise.
13178         * tree-sra.c (create_access_replacement): Likewise.
13179         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13180         (clear_unused_block_pointer): Likewise.
13181         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13182         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13183         * var-tracking.c (var_debug_decl): Likewise.
13184         (track_expr_p): Likewise.
13185         * tree-inline.c (add_local_variables): Likewise.  Set
13186         DECL_HAS_DEBUG_EXPR_P after copying it.
13187         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13188         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13190 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13192         PR bootstrap/56656
13193         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13194         * configure: Regenerate.
13195         * config.in: Regenerate.
13196         * config/i386/i386.md (*movdf_internal): Use
13197         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13198         movd instead of movq mnemonic for interunit moves.
13199         (*movdi_internal): Ditto.
13201 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13203         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13204         (abd<mode>_3): New pattern.
13205         (aba<mode>_3): New pattern.
13206         (fabd<mode>_3): New pattern.
13208 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13210         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13211         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13212         occurrence of REGISTER_PREFIX as its empty string.
13214 2013-03-20  Jeff Law  <law@redhat.com>
13216         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13217         addititional equivalences for equality comparisons between an SSA_NAME
13218         and a constant where the SSA_NAME was set from a widening conversion.
13220 2013-03-20  Walter Lee  <walt@tilera.com>
13222         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13224 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13226         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13227         depending on type attribute.
13228         (*movti_internal): Ditto.
13229         (*movtf_internal): Ditto.
13230         (*movxf_internal): Ditto.
13231         (*movdf_internal): Ditto.
13232         (*movsf_internal): Ditto.
13234 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13236         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13237         maybe_vex for sselog1 and ssemov types.
13238         (*movdi_internal): Reorder operand constraints.
13239         (*movsi_internal): Ditto.  Set prefix attribute to
13240         maybe_vex for sselog1 and ssemov types.
13241         (*movtf_internal): Set prefix attribute to maybe_vex
13242         for sselog1 and ssemov types.
13243         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13244         DImode ssemov types.  Reorder operand constraints.
13245         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13246         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13247         attribute for SImode ssemov types.  Reorder operand constraints.
13249 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13251         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13252         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13254 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13256         * config/rs6000/predicates.md (indexed_address, update_address_mem
13257         update_indexed_address_mem): New predicates.
13258         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13259         attribute for load/store instructions.
13260         * config/rs6000/dfp.md (movsd_store): Likewise.
13261         (movsd_load): Likewise.
13262         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13263         (unnamed HI->DI extend define_insn): Likewise.
13264         (unnamed SI->DI extend define_insn): Likewise.
13265         (unnamed QI->SI extend define_insn): Likewise.
13266         (unnamed QI->HI extend define_insn): Likewise.
13267         (unnamed HI->SI extend define_insn): Likewise.
13268         (unnamed HI->SI extend define_insn): Likewise.
13269         (extendsfdf2_fpr): Likewise.
13270         (movsi_internal1): Likewise.
13271         (movsi_internal1_single): Likewise.
13272         (movhi_internal): Likewise.
13273         (movqi_internal): Likewise.
13274         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13275         attribute for load/store instructions.
13276         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13277         instructions.
13278         (mov<mode>_softfloat): Likewise.
13279         (mov<mode>_hardfloat32): Likewise.
13280         (mov<mode>_hardfloat64): Likewise.
13281         (mov<mode>_softfloat64): Likewise.
13282         (movdi_internal32): Likewise.
13283         (movdi_internal64): Likewise.
13284         (probe_stack_<mode>): Likewise.
13286 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13288         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13289         floating point, and decimal floating point to reload iterator.
13291         * config/rs6000/constraints.md (wl constraint): New constraints to
13292         return FLOAT_REGS if certain options are used to reduce the number
13293         of separate patterns that exist in the file.
13294         (wx constraint): Likewise.
13295         (wz constraint): Likewise.
13297         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13298         -mdebug=reg, print wg, wl, wx, and wz constraints.
13299         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13300         Initialize the reload functions for 64-bit binary/decimal floating
13301         point types.
13302         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13303         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13304         create the buffer on the stack to overcome not having a 32-bit
13305         load and store.
13306         (rs6000_emit_move): Likewise.
13307         (rs6000_secondary_memory_needed_rtx): Likewise.
13308         (rs6000_alloc_sdmode_stack_slot): Likewise.
13309         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13310         via xxlxor, just like DFmode 0.0.
13312         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13313         define as 1 if we are running on a power7 or newer.
13314         (enum r6000_reg_class_enum): Add new constraints.
13316         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13317         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13318         with other moves by using conditional constraits (wg).  Use LFIWZX
13319         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13320         (movsd splitter): Likewise.
13321         (movsd_hardfloat): Likewise.
13322         (movsd_softfloat): Likewise.
13324         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13325         binary and decimal floating point moves.
13326         (fmove_ok): New attributes to combine binary and decimal floating
13327         point moves, and to combine power6x (mfpgpr) moves along normal
13328         floating moves.
13329         (real_value_to_target): Likewise.
13330         (f32_lr): Likewise.
13331         (f32_lm): Likewise.
13332         (f32_li): Likewise.
13333         (f32_sr): Likewise.
13334         (f32_sm): Likewise.
13335         (f32_si): Likewise.
13336         (movsf): Combine binary and decimal floating point moves.  Combine
13337         power6x (mfpgpr) moves with other moves by using conditional
13338         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13339         (mov<mode> for SFmode/SDmode); Likewise.
13340         (SFmode/SDmode splitters): Likewise.
13341         (movsf_hardfloat): Likewise.
13342         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13343         (movsf_softfloat): Likewise.
13344         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13346         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13347         wx and wz constraints.
13349         * config/rs6000/constraints.md (wg constraint): New constraint to
13350         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13352         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13353         constraint.
13355         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13356         -mdebug=reg, print wg, wl, wx, and wz constraints.
13357         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13358         Initialize the reload functions for 64-bit binary/decimal floating
13359         point types.
13360         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13361         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13362         create the buffer on the stack to overcome not having a 32-bit
13363         load and store.
13364         (rs6000_emit_move): Likewise.
13365         (rs6000_secondary_memory_needed_rtx): Likewise.
13366         (rs6000_alloc_sdmode_stack_slot): Likewise.
13367         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13368         via xxlxor, just like DFmode 0.0.
13370         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13371         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13372         with other moves by using conditional constraits (wg).  Use LFIWZX
13373         and STFIWX for loading SDmode on power7.
13374         (movdd splitters): Likewise.
13375         (movdd_hardfloat32): Likewise.
13376         (movdd_softfloat32): Likewise.
13377         (movdd_hardfloat64_mfpgpr): Likewise.
13378         (movdd_hardfloat64): Likewise.
13379         (movdd_softfloat64): Likewise.
13381         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13382         64-bit binary and decimal floating point moves.
13383         (FMOVE64X): Likewise.
13384         (movdf): Combine 64-bit binary and decimal floating point moves.
13385         Combine power6x (mfpgpr) moves with other moves by using
13386         conditional constraits (wg).
13387         (mov<mode> for DFmode/DDmode): Likewise.
13388         (DFmode/DDmode splitters): Likewise.
13389         (movdf_hardfloat32): Likewise.
13390         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13391         (movdf_softfloat32): Likewise.
13392         (movdf_hardfloat64_mfpgpr): Likewise.
13393         (movdf_hardfloat64): Likewise.
13394         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13395         (movdf_softfloat64): Likewise.
13396         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13397         (reload_<mode>_load): Move to later in the file so they aren't in
13398         the middle of the floating point move insns.
13399         (reload_<mode>_store): Likewise.
13401         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13402         constraint.
13404         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13405         constraint if -mdebug=reg.
13406         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13407         Enable using dd reload support if needed.
13409         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13410         binary and decimal floating point moves in rs6000.md.
13411         (movtd_internal): Likewise.
13413         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13414         decimal floating point moves.
13415         (movtf): Likewise.
13416         (movtf_internal): Likewise.
13417         (mov<mode>_internal, TDmode/TFmode): Likewise.
13418         (movtf_softfloat): Likewise.
13419         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13421         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13422         movdi_internal64, using wg constraint for move direct operations.
13423         (movdi_internal64): Likewise.
13425         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13426         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13427         the various virtual registers. Use GPR/FPR first/last values,
13428         instead of hard coding the register numbers.  Print which modes
13429         have reload functions registered.
13430         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13431         settings before/after setting cpu, target and subtarget settings.
13432         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13433         and for secondary reload failures in rs6000_secondary_reload_inner.
13434         (rs6000_secondary_reload_fail): Likewise.
13435         (rs6000_secondary_reload_inner): Likewise.
13437         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13438         macros for first/last GPR and FPR registers.
13439         (LAST_GPR_REGNO): Likewise.
13440         (FIRST_FPR_REGNO): Likewise.
13441         (LAST_FPR_REGNO): Likewise.
13443         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13444         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13445         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13446         (vcond<mode><mode>): Likewise.
13447         (vcondu<mode><mode>): Likewise.
13448         (vector_gtu<mode>): Likewise.
13449         (vector_gte<mode>): Likewise.
13450         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13451         to prevent the compiler from converting DImode operations to TImode.
13452         (ior<mode>3): Likewise.
13453         (and<mode>3): Likewise.
13454         (one_cmpl<mode>2): Likewise.
13455         (nor<mode>3): Likewise.
13456         (andc<mode>3): Likewise.
13458         * config/rs6000/constraints.md (wt constraint): New constraint
13459         that returns VSX_REGS if TImode is allowed in VSX registers.
13461         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13462         constant under VSX.
13464         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13465         similar to TImode, but it is restricted to being in the GPRs.
13467         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13468         TImode to occupy a single VSX register.
13470         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13471         -mvsx-timode for power7/power8.
13472         (power7 cpu): Likewise.
13473         (power8 cpu): Likewise.
13475         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13476         sure that TFmode/TDmode take up two registers if they are ever
13477         allowed in the upper VSX registers.
13478         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13479         registers.
13480         (rs6000_init_hard_regno_mode_ok): Likewise.
13481         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13482         constraint.  Print if LRA is turned on.
13483         (rs6000_option_override_internal): Give an error if -mvsx-timode
13484         and VSX is not enabled.
13485         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13486         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13487         to reg+offset addressing.  Use PTImode when checking offset
13488         addresses for validity.
13489         (reg_offset_addressing_ok_p): Likewise.
13490         (rs6000_legitimate_offset_address_p): Likewise.
13491         (rs6000_legitimize_address): Likewise.
13492         (rs6000_legitimize_reload_address): Likewise.
13493         (rs6000_legitimate_address_p): Likewise.
13494         (rs6000_eliminate_indexed_memrefs): Likewise.
13495         (rs6000_emit_move): Likewise.
13496         (rs6000_secondary_reload): Likewise.
13497         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13498         reloads to fpr registers to continue to use reg+offset addressing,
13499         but 64-bit reloads to altivec registers need reg+reg addressing.
13500         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13501         it.  Treat LO_SUM like a PLUS operation.
13502         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13503         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13504         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13505         registers to share a register with a smaller sized type, since VSX
13506         puts scalars in the upper 64-bits.
13507         (print_operand): Add support for PTImode.
13508         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13509         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13510         registers, but don't have arithmetic support.
13511         (rs6000_memory_move_cost): Add test for VSX.
13512         (rs6000_opt_masks): Add -mvsx-timode.
13514         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13515         for TImode.
13516         (VSs): Likewise.
13517         (VSr): Use wt constraint for TImode.
13518         (VSv): Drop TImode support.
13519         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13520         (vsx_movti_64bit): Likewise.
13521         (vsx_movti_32bit): Likewise.
13522         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13523         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13524         one '?' on the appropriate output constraint.  Do not allow TImode
13525         logical operations on 32-bit systems.
13526         (vsx_ior<mode>3): Likewise.
13527         (vsx_xor<mode>3): Likewise.
13528         (vsx_one_cmpl<mode>2): Likewise.
13529         (vsx_nor<mode>3): Likewise.
13530         (vsx_andc<mode>3): Likewise.
13531         (vsx_concat_<mode>): Likewise.
13532         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13534         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13535         OPTION_MASK_VSX_TIMODE.
13536         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13537         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13539         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13540         (TI2 iterator): New iterator for TImode, PTImode.
13541         (wd mode attribute): Add values for vector types.
13542         (movti_string): Replace TI move operations with operations for TImode
13543         and PTImode.  Add support for TImode being allowed in VSX registers.
13544         (mov<mode>_string, TImode/PTImode): Likewise.
13545         (movti_ppc64): Likewise.
13546         (mov<mode>_ppc64, TImode/PTImode): Likewise.
13547         (TI mode splitters): Likewise.
13549         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13550         constraint.
13552 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
13554         PR tree-optimization/56355
13555         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13556         Also handle integers with undefined overflow.
13558 2013-03-20  Catherine Moore  <clm@codesourcery.com>
13559             Maciej W. Rozycki  <macro@codesourcery.com>
13560             Tom de Vries  <tom@codesourcery.com>
13561             Nathan Sidwell <nathan@codesourcery.com>
13562             Iain Sandoe  <iain@codesourcery.com>
13563             Nathan Froyd  <froydnj@codesourcery.com>
13564             Chao-ying Fu <fu@mips.com>
13566         * doc/extend.texi: (micromips, nomicromips, nocompression):
13567         Document new function attributes.
13568         * doc/invoke.texi (minterlink-compressed, mmicromips,
13569         m14k, m14ke, m14kec): Document new options.
13570         (minterlink-mips16): Update documentation.
13571         * doc/md.texi (ZC, ZD): Document new constraints.
13572         * configure.ac (gcc_cv_as_micromips): Check if linker
13573         supports the .set micromips directive.
13574         * configure: Regenerate.
13575         * config.in: Regenerate.
13576         * config/mips/mips-tables.opt: Regenerate.
13577         * config/mips/micromips.md: New file.
13578         * constraints.md (ZC, ZD): New constraints.
13579         * config/mips/predicates.md (movep_src_register): New predicate.
13580         (movep_src_operand): New predicate.
13581         (non_volatile_mem_operand): New predicate.
13582         * config/mips/mips.md (multimem): New type.
13583         (length): Differentiate between 17-bit and 18-bit branch offsets.
13584         (MOVEP1, MOVEP2): New mode iterator.
13585         (mov_<load>l): Use ZC constraint.
13586         (mov_<load>r): Likewise.
13587         (mov_<store>l): Likewise.
13588         (mov_<store>r): Likewise.
13589         (*branch_equality<mode>_inverted): Add microMIPS support.
13590         (*branch_equality<mode>): Likewise.
13591         (*jump_absolute): Likewise.
13592         (indirect_jump_<mode>): Likewise.
13593         (tablejump_<mode>): Likewise.
13594         (<optab>_internal): Likewise.
13595         (sibcall_internal): Likewise.
13596         (sibcall_value_internal): Likewise.
13597         (prefetch): Use constraint ZD.
13598         * config/mips/mips.opt (minterlink-compressed): New option.
13599         (minterlink-mips16): Now an alias for minterlink-compressed.
13600         (mmicromips): New option.
13601         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13602         (compare_and_swap_12): Likewise.
13603         (sync_add<mode>): Likewise.
13604         (sync_<optab>_12): Likewise.
13605         (sync_old_<optab>_12): Likewise.
13606         (sync_new_<optab>_12): Likewise.
13607         (sync_nand_12): Likewise.
13608         (sync_old_nand_12): Likewise.
13609         (sync_new_nand_12): Likewise.
13610         (sync_sub<mode>): Likewise.
13611         (sync_old_add<mode>): Likewise.
13612         (sync_old_sub<mode>): Likewise.
13613         (sync_new_add<mode>): Likewise.
13614         (sync_new_sub<mode>): Likewise.
13615         (sync_<optab><mode>): Likewise.
13616         (sync_old_<optab><mode>): Likewise.
13617         (sync_new_<optab><mode>): Likewise.
13618         (sync_nand<mode>): Likewise.
13619         (sync_old_nand<mode>): Likewise.
13620         (sync_new_nand<mode>): Likewise.
13621         (sync_lock_test_and_set<mode>): Likewise.
13622         (test_and_set_12): Likewise.
13623         (atomic_compare_and_swap<mode>): Likewise.
13624         (atomic_exchange<mode>_llsc): Likewise.
13625         (atomic_fetch_add<mode>_llsc): Likewise.
13626         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13627         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13628         (umips_save_restore_pattern_p): Likewise.
13629         (umips_load_store_pair_p): Likewise.
13630         (umips_output_load_store_pair): Likewise.
13631         (umips_movep_target_p): Likewise.
13632         (umips_12bit_offset_address_p): Likewise.
13633         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13634         (mips_base_mips16): Rename this...
13635         (mips_base_compression_flags): ...to this. Update all uses.
13636         (mips_attribute_table): Add micromips, nomicromips and nocompression.
13637         (mips_mips16_decl_p): Delete.
13638         (mips_nomips16_decl_p): Delete.
13639         (mips_get_compress_on_flags): New function.
13640         (mips_get_compress_off_flags): New function.
13641         (mips_get_compress_mode): New function.
13642         (mips_get_compress_on_name): New function.
13643         (mips_get_compress_off_name): New function.
13644         (mips_insert_attributes): Support multiple compression types.
13645         (mips_merge_decl_attributes): Likewise.
13646         (umips_12bit_offset_address_p): New function.
13647         (mips_start_function_definition): Emit .set micromips directive.
13648         (mips_call_may_need_jalx_p): New function.
13649         (mips_function_ok_for_sibcall): Add microMIPS support.
13650         (mips_print_operand_punctuation): Support short delay slots and
13651         compact jumps.
13652         (umips_swm_mask, umips_swm_encoding): New.
13653         (umips_build_save_restore): New function.
13654         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13655         (was_mips16_p): Remove.
13656         (old_compression_mode): New.
13657         (mips_set_compression_mode): New function.
13658         (mips_set_current_function): Add microMIPS support.
13659         (mips_option_override): Likewise.
13660         (umips_save_restore_pattern_p): New function.
13661         (umips_output_save_restore): New function.
13662         (umips_load_store_pair_p_1): New function.
13663         (umips_load_store_pair_p): New function.
13664         (umips_output_load_store_pair_1): New function.
13665         (umips_output_load_store_pair): New function.
13666         (umips_movep_target_p) New function.
13667         (mips_prepare_pch_save): Add microMIPS support.
13668         * config/mips/mips.h (TARGET_COMPRESSION): New.
13669         (TARGET_CPU_CPP_BUILTINS): Update macro
13670         to use new compression flags and to support microMIPS.
13671         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13672         (MIPS_ARCH_FLOAT_SPEC): Likewise.
13673         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13674         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13675         (ASM_SPEC): Support mmicromips and mno-micromips.
13676         (M16STORE_REG_P): New macro.
13677         (MIPS_CALL): Support TARGET_MICROMIPS.
13678         (MICROMIPS_J): New macro.
13679         (mips_base_mips16): Rename this...
13680         (mips_base_compression_flags): ...to this.
13681         (UMIPS_12BIT_OFFSET_P): New macro.
13682         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13683         (MULTILIB_DIRNAMES): Likewise.
13684 2013-03-20  Richard Biener  <rguenther@suse.de>
13686         PR tree-optimization/56661
13687         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13688         the result does not have to be distinct.
13690 2013-03-20  Richard Biener  <rguenther@suse.de>
13692         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13693         remap_gimple_op_r.
13695 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13696             Steven Bosscher <steven@gcc.gnu.org>
13698         PR rtl-optimization/56605
13699         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13701 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13703         PR bootstrap/56656
13704         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13705         that require movd instead of movq.
13707 2013-03-20  Richard Biener  <rguenther@suse.de>
13709         * tree-ssa-structalias.c (struct variable_info): Add pointer
13710         to the first field of an aggregate with sub-vars.  Make
13711         this and the pointer to the next subfield its ID.
13712         (vi_next): New function.
13713         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13714         storedanything_id, integer_id): Increment by one.
13715         (new_var_info, get_call_vi, lookup_call_clobber_vi,
13716         get_call_clobber_vi): Adjust.
13717         (solution_set_expand): Simplify and speedup.
13718         (solution_set_add): Inline into ...
13719         (set_union_with_increment): ... this.  Adjust accordingly.
13720         (do_sd_constraint): Likewise.
13721         (do_ds_constraint): Likewise.
13722         (do_complex_constraint): Simplify.
13723         (build_pred_graph): Adjust.
13724         (solve_graph): Likewise.  Simplify and speedup.
13725         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13726         get_constraint_for_component_ref, get_constraint_for_1,
13727         first_vi_for_offset, first_or_preceding_vi_for_offset,
13728         create_function_info_for, create_variable_info_for_1,
13729         create_variable_info_for, intra_create_variable_infos): Adjust.
13730         (init_base_vars): Push NULL for ID zero.
13731         (compute_points_to_sets): Adjust.
13733 2013-03-20  Richard Biener  <rguenther@suse.de>
13735         * cfgloop.c (verify_loop_structure): Streamline and avoid
13736         ICEing on corrupt loop tree.
13737         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13738         loop tree.
13740 2013-03-20  Richard Biener  <rguenther@suse.de>
13742         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13743         check whether an SSA update is needed.
13745 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
13747         * config/mips/constraints.md (T): Rename to...
13748         (Yf): ...this.
13749         (U): Rename to...
13750         (Yd): ...this.
13751         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13752         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13754 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13756         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13757         (*subsi3_carryin_uxtw): Likewise.
13759 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13761         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13762         (*rorsi3_insn_uxtw): Likewise.
13764 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13766         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13767         (*extrsi5_insn_uxtw): Likewise.
13769 2013-03-19  Richard Biener  <rguenther@suse.de>
13771         PR tree-optimization/56273
13772         * passes.c (init_optimization_passes): Move second VRP after DOM.
13774 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
13776         * config/i386/i386.md (*movti_internal): Merge from
13777         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
13778         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
13779         nox64 isa attributes.
13781 2013-03-18  Richard Biener  <rguenther@suse.de>
13783         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13784         (unite): Likewise.
13785         (merge_node_constraints): Likewise.
13786         (build_succ_graph): Likewise.
13787         (valid_graph_edge): Inline into single caller.
13788         (unify_nodes): Likewise.  Use bitmap_set_bit return value
13789         and cache varinfo.
13790         (scc_visit): Fix formatting and variable use.
13791         (do_sd_constraint): Use gcc_checking_assert.
13792         (do_ds_constraint): Likewise.
13793         (do_complex_constraint): Likewise.
13794         (condense_visit): Likewise.  Cleanup.
13795         (dump_pred_graph): New function.
13796         (perform_var_substitution): Dump the pred-graph before
13797         variable substitution.
13798         (find_equivalent_node): Use gcc_checking_assert.
13799         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13801 2013-03-18  Richard Biener  <rguenther@suse.de>
13803         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13804         Remove cond_expr_stmt_list argument and do not gimplify the
13805         built expression.
13806         (vect_loop_versioning): Adjust.
13807         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13808         Cleanup to use less temporaries.
13809         (vect_create_data_ref_ptr): Cleanup.
13811 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13813         PR tree-optimization/56635
13814         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13815         require types_compatible_p types.
13817 2013-03-18  Nick Clifton  <nickc@redhat.com>
13819         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13820         spurious backslash.
13822         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13823         Add missing line to comment describing function.
13825 2013-03-18  Richard Biener  <rguenther@suse.de>
13827         PR tree-optimization/56210
13828         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13829         Handle string / character search functions.
13830         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13832 2013-03-18  Richard Biener  <rguenther@suse.de>
13834         PR middle-end/56483
13835         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13836         and implement properly.
13837         * gimple.h (gimple_cond_single_var_p): Remove.
13839 2013-03-18  Richard Biener  <rguenther@suse.de>
13841         * tree-data-ref.h (find_data_references_in_loop): Declare.
13842         * tree-data-ref.c (get_references_in_stmt): Use a stack
13843         vector pre-allocated in the callers.
13844         (find_data_references_in_stmt): Adjust.
13845         (graphite_find_data_references_in_stmt): Likewise.
13846         (create_rdg_vertices): Likewise.
13847         (find_data_references_in_loop): Export.
13848         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13849         Compute dependences here...
13850         (vect_analyze_data_refs): ...not here.  When we encounter
13851         a non-vectorizable data reference in basic-block vectorization
13852         truncate the data reference vector.  Do not bother to
13853         fixup data-dependence information for gather loads.
13854         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13855         of data references, as reported.
13857 2013-03-18  Richard Biener  <rguenther@suse.de>
13859         PR tree-optimization/3713
13860         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
13861         has_constants and expr.
13862         (stmt_has_constants): Properly valueize SSA names when deciding
13863         whether the stmt has constants.
13865 2013-03-18  Richard Biener  <rguenther@suse.de>
13867         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13868         whole function when there is nothing to do.
13869         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13870         * tree-vectorizer.c (vectorize_loops): Update virtual and
13871         loop-closed SSA once.
13872         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13874 2013-03-18  Richard Biener  <rguenther@suse.de>
13876         PR middle-end/56113
13877         * domwalk.c (bb_postorder): New global static.
13878         (cmp_bb_postorder): New function.
13879         (walk_dominator_tree): Replace scheme imposing an order for
13880         visiting dominator sons by one sorting them at the time they
13881         are pushed on the stack.
13883 2013-03-18  Richard Biener  <rguenther@suse.de>
13885         PR tree-optimization/39326
13886         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
13887         (struct mem_ref): Replace mem member with ao_ref typed member.
13888         (MEM_ANALYZABLE): Adjust.
13889         (memref_eq): Likewise.
13890         (mem_ref_alloc): Likewise.
13891         (gather_mem_refs_stmt): Likewise.
13892         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
13893         (execute_sm_if_changed_flag_set): Adjust.
13894         (execute_sm): Likewise.
13895         (ref_always_accessed_p): Likewise.
13896         (refs_independent_p): Likewise.
13897         (can_sm_ref_p): Likewise.
13899 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13901         PR c/56566
13902         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
13903         return 1 even for !unsignedp.
13905 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13907         * config/i386/i386.md (isa): Add x64 and nox64.
13908         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
13909         (*pushtf): Enable *roF alternative for x64 isa only.
13910         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
13911         mode attribute of integer alternatives to DImode for TARGET_64BIT.
13912         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
13913         (*movtf_internal): Merge from *movtf_internal_rex64 and
13914         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
13915         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
13916         nox64 isa attributes.
13917         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
13918         nox64 isa attributes.
13919         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
13921 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13923         * config/alpha/alpha.c (TARGET_LRA_P): New define.
13925 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
13927         PR target/56640
13928         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
13929         class names.  Remove trailing comma after "ALL_REGS".
13931 2013-03-16  Jan Hubicka  <jh@suse.cz>
13933         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
13934         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
13935         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
13936         of cgraph_get_create_node.
13937         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13939 2013-03-16  Jason Merrill  <jason@redhat.com>
13941         PR debug/49090
13942         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
13943         with DW_AT_default_value.
13945 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
13947         * BASE-VER: Set to 4.9.0.
13949 2013-03-14  Andi Kleen  <ak@linux.intel.com>
13951         PR target/56619
13952         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
13953         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
13954         Document _x* TSX intrinsics.
13956 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
13957             David Holsgrove  <david.holsgrove@xilinx.com>
13959         * configure.ac: Add MicroBlaze TLS support detection.
13960         * configure: Regenerate.
13961         * config/microblaze/microblaze-protos.h
13962         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
13963         symbol_mentioned_p, label_mentioned_p): Add prototypes.
13964         * config/microblaze/microblaze.c (microblaze_address_type): Add
13965         ADDRESS_TLS and tls_reloc address types.
13966         (microblaze_address_info): Add tls_reloc.
13967         (TARGET_HAVE_TLS): Define.
13968         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
13969          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
13970          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
13971         load_tls_operand,  microblaze_call_tls_get_addr,
13972         microblaze_legitimize_tls_address): New functions.
13973         (microblaze_classify_unspec): Handle UNSPEC_TLS.
13974         (get_base_reg): Use microblaze_tls_symbol_p.
13975         (microblaze_classify_address): Handle TLS.
13976         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
13977         label_mentioned_p and microblaze_tls_referenced_p.
13978         (microblaze_legitimize_address): Handle TLS.
13979         (microblaze_address_insns): Handle ADDRESS_TLS.
13980         (pic_address_needs_scratch): Handle TLS.
13981         (print_operand_address): Handle TLS.
13982         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
13983         (microblaze_expand_move): Handle TLS.
13984         (microblaze_legitimate_constant_p): Check
13985         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
13986         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13987         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
13988         (PIC_OFFSET_TABLE_REGNUM): Set.
13989         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
13990         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
13991         (addsi3, movsi_internal2, movdf_internal): Update constraints
13992         * config/microblaze/predicates.md (arith_plus_operand): Define
13993         (move_operand): Redefine as move_src_operand,
13994         check microblaze_tls_referenced_p.
13996 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
13998         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
13999         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14001 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14003         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14004         CC mode for AND.
14006 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14008         PR tree-optimization/53265
14009         * common.opt (Waggressive-loop-optimizations): New option.
14010         * tree-ssa-loop-niter.c: Include tree-pass.h.
14011         (do_warn_aggressive_loop_optimizations): New function.
14012         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14013         if number_of_latch_executions returned constant.
14014         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14015         early.  If number_of_latch_executions returned constant, set
14016         nb_iterations_upper_bound back to it.
14017         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14018         field.
14019         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14020         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14022         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14023         (MULTILIB_OSDIRNAMES): Set.
14024         * genmultilib: If defaultosdirname doesn't start with :: , set
14025         defaultosdirname2 instead, clear it and emit two . multilib_raw
14026         entries instead of just one.
14028 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14030         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14031         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14032         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14033         (SUBTARGET_OVERRIDE_OPTIONS): New.
14035 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14037         PR target/49880
14038         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14039         (musermode): Convert to Var(TARGET_USERMODE).
14040         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14041         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14042         * config/sh/sh.c (sh_option_override): Use
14043         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14044         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14045         condition.
14046         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14047         TARGET_SH4.
14048         (udivsi3_i4_single, divsi3_i4_single): Use
14049         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14051 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14053         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14054         default setting.
14056 2013-03-13  Richard Biener  <rguenther@suse.de>
14058         PR tree-optimization/56608
14059         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14060         calls when vectorizing basic-blocks.
14062 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14064         PR plugins/45078
14065         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14066         tm_file.
14068 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14070         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14072 2013-03-11  Jan Hubicka  <jh@suse.cz>
14074         PR lto/56557
14075         * lto-streamer-out.c (output_symbol_p): Skip references from
14076         constructors of external variables.
14078 2013-03-11  Jan Hubicka  <jh@suse.cz>
14080         PR middle-end/56571
14081         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14082         from pseudos.
14083         * emit-rtl.c (verify_rtx_sharing): Likewise.
14084         (copy_insn_1): Likewise.
14085         * rtl.c (copy_rtx): Likewise.
14087 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14089         PR target/56591
14090         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14091         output_operand_lossage message.
14093 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14095         PR target/56470
14096         * arm.c (shift_op): Validate RTL pattern on the fly.
14097         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14098         the RTL.
14100 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14102         PR target/56347
14103         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14104         new call patterns that clobber %fr12.
14105         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14106         split and postreload patterns.
14107         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14108         registers %fr12 and %fr12R as call used.
14110 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14112         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14113         (canon_address, record_store, replace_read, check_mem_read_rtx,
14114         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14115         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14116         rest_of_handle_dse): Likewise.
14118 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14120         PR middle-end/56524
14121         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14122         Add base_optabs.
14123         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14124         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14125         (save_optabs_if_changed): Replace with...
14126         (init_tree_optimization_optabs): ...this.
14127         * optabs.c (save_optabs_if_changed): Rename to...
14128         (init_tree_optimization_optabs): ...this.  Take the optimization node
14129         as argument.  Do nothing if the base optabs are already correct.
14130         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14131         to recompute optabs.
14132         * function.h (function): Remove optabs field.
14133         * function.c (invoke_set_current_function_hook): Call
14134         init_tree_optimization_optabs.  Use the result to initialize
14135         this_fn_optabs.
14137 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14139         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14140         if GTMA_HAS_NO_INSTRUMENTATION.
14141         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14142         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14143         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14144         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14145         GTMA_HAS_NO_INSTRUMENTATION.
14147 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14149         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14150         libasan_preinit.o.
14152 2013-03-08  Marek Polacek  <polacek@redhat.com>
14153             Jakub Jelinek  <jakub@redhat.com>
14155         PR tree-optimization/56478
14156         * predict.c (is_comparison_with_loop_invariant_p): Change the
14157         type of loop_step to tree.
14158         (predict_loops): Adjust.
14159         (predict_iv_comparison): Perform the computations on double_ints.
14161 2013-03-08  Richard Biener  <rguenther@suse.de>
14163         PR tree-optimization/56570
14164         * tree-cfg.c (verify_expr_location_1): Verify locations for
14165         DECL_DEBUG_EXPR.
14166         * tree-sra.c (create_access_replacement): Strip locations
14167         from DECL_DEBUG_EXPRs.
14169 2013-03-08  Richard Biener  <rguenther@suse.de>
14171         * tree-inline.c (expand_call_inline): Do not associate
14172         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14173         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14175 2013-03-08  Richard Biener  <rguenther@suse.de>
14177         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14178         or block changes with -Og.  Fix for location / block encoding
14179         changes and PHI arguments with locations.
14181 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14183         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14184         for all counters.
14185         (struct output_info): Likewise.
14186         (register_overhead): Remove bad gcc_assert.
14187         (bitmap_find_bit): If there is only a single bitmap element, do not
14188         count a miss as a search.
14189         (print_statistics): Update for counter type changes.
14190         (dump_bitmap_statistics): Likewise.  Print headers such that they
14191         are properly lined up with the printed counters.
14193 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14195         PR tree-optimization/56559
14196         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14197         check that it has only a single use.
14199 2013-03-07  Richard Biener  <rguenther@suse.de>
14201         * doc/invoke.texi (fwhole-program): Discourage use in combination
14202         with -flto.
14204 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14206         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14208         PR tree-optimization/56539
14209         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14210         instead of GSI_CONTINUE_LINKING as last argument to
14211         force_gimple_operand_gsi.  Adjust function comment.
14213         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14214         aarch64-cores.def.
14216         PR middle-end/56548
14217         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14218         promoted mode, convert the result back to the original mode.
14220 2013-03-06  Richard Biener  <rguenther@suse.de>
14222         PR middle-end/56294
14223         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14224         (insert_updated_phi_nodes_compare_uids): New function.
14225         (update_ssa): Sort symbols_to_rename after UID before
14226         traversing it to insert PHI nodes.
14228 2013-03-06  Richard Biener  <rguenther@suse.de>
14230         PR middle-end/50494
14231         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14232         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14234         Revert
14235         2013-02-13  Richard Biener  <rguenther@suse.de>
14237         PR lto/50494
14238         * varasm.c (output_constant_def_1): Get the decl representing
14239         the constant as argument.
14240         (output_constant_def): Wrap output_constant_def_1.
14241         (make_decl_rtl): Use output_constant_def_1 with the decl
14242         representing the constant.
14243         (build_constant_desc): Optionally re-use a decl already
14244         representing the constant.
14245         (tree_output_constant_def): Adjust.
14247 2013-03-06  Joey Ye  <joey.ye@arm.com>
14249         PR lto/50293
14250         * gcc.c (convert_white_space): New function.
14251         (main): Handles white space in function name.
14253 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14255         PR target/56529
14256         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14257         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14258         to SH_DIV_CALL_TABLE for TARGET_SH2.
14259         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14260         list.
14261         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14262         call-table options.
14264 2013-03-05  Sterling Augustine  <saugustine@google.com>
14265             Cary Coutant  <ccoutant@google.com>
14267         PR debug/55364
14268         * dwarf2out.c (resolve_addr): Don't call
14269         remove_loc_list_addr_table_entries a second time for the same
14270         expression.
14272 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14274         PR debug/56510
14275         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14276         (avoid_complex_debug_insns): New function.
14277         (expand_debug_locations): Call it.
14279         PR rtl-optimization/56484
14280         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14281         lifetimes of hard registers on small register class machines.
14283 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14285         * config/microblaze/microblaze-protos.h: Rename
14286         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14287         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14288         fast_interrupt.
14289         (microblaze_fast_interrupt_function_p): New function.
14290         (microblaze_is_interrupt_handler): Rename to
14291         microblaze_is_interrupt_variant and add fast_interrupt check.
14292         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14293         (save_restore_insns): Likewise.
14294         (compute_frame_size): Likewise.
14295         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14296         (microblaze_globalize_label): Likewise.
14297         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14298         * config/microblaze/microblaze.md: Use wrapper
14299         microblaze_is_interrupt_variant.
14301 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14303         * sdbout.c (sdbout_one_type): Switch to current function's section
14304         supporting cold/hot.
14306 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14308         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14309         -mxl-reorder.
14311 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14313         PR middle-end/56461
14314         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14315         if VALGRIND_GET_VBITS is defined, temporarily make object
14316         memory all defined, and restore previous valgrind addressability
14317         and definability afterwards.  Free this_object at the end.
14319         PR middle-end/56461
14320         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14321         right before calling lra_create_live_ranges, also call it
14322         when clearing live_p.  Only call lra_clear_live_ranges
14323         at the end if live_p.
14325         PR middle-end/56461
14326         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14328 2013-03-05  Richard Biener  <rguenther@suse.de>
14330         PR tree-optimization/56521
14331         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14332         value-id.
14334 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14336         PR c++/55135
14337         * except.h (remove_unreachable_eh_regions): New prototype.
14338         * except.c (remove_eh_handler_splicer): New function, split out
14339         of remove_eh_handler.
14340         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14341         warning about running it on many EH regions one at a time.
14342         (remove_unreachable_eh_regions_worker): New function, walk the
14343         EH tree in depth-first order and remove non-marked regions.
14344         (remove_unreachable_eh_regions): New function.
14345         * tree-eh.c (mark_reachable_handlers): New function, split out
14346         from remove_unreachable_handlers.
14347         (remove_unreachable_handlers): Use mark_reachable_handlers and
14348         remove_unreachable_eh_regions.
14349         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14350         and remove_unreachable_eh_regions.
14352 2013-03-05  Richard Biener  <rguenther@suse.de>
14354         PR middle-end/56525
14355         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14356         not freeing them until the end.
14358 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14360         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14362 2013-03-05  Richard Biener  <rguenther@suse.de>
14364         PR tree-optimization/56270
14365         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14366         of loads after scheduling an SLP instance.
14368 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14370         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14371         tic6x.exp.
14372         (check_gcc_parallelize): Run guality.exp as a separate job from
14373         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14374         struct-layout-1.exp with stackalign.exp.
14376         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14378         PR middle-end/56461
14379         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14380         load_index sbitmap even if some bit in it isn't set.
14382         PR middle-end/56461
14383         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14384         (discover_iteration_bound_by_body_walk): Change queues to
14385         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14386         spelling in comment.  Call safe_push on queues[bound_index] directly.
14387         Release queues[queue_index] in every iteration unconditionally.
14388         Release bounds vector.
14390         PR middle-end/56461
14391         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14392         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14393         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14394         inner_phis vector.
14396 2013-03-05  Richard Biener  <rguenther@suse.de>
14398         PR lto/56515
14399         * tree-inline.c (remap_blocks_to_null): New function.
14400         (expand_call_inline): When expanding a call stmt without
14401         an associated block inline remap all callee blocks to NULL.
14403 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14405         PR rtl-optimization/56494
14406         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14407         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14408         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14410         PR middle-end/56461
14411         * sel-sched-ir.c (free_sched_pools): Release
14412         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14413         if succs_info_pool.max_top isn't -1.
14415         PR bootstrap/56509
14416         * opts.c (opts_obstack, opts_concat): Moved to...
14417         * opts-common.c (opts_obstack, opts_concat): ... here.
14419 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14421         PR middle-end/56461
14422         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14424 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14426         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14427         all appropriate places.
14429 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14431         PR tree-optimization/56424
14432         * ipa-split.c (split_function): Do not set the RSO flag if result is
14433         not by reference and its type is a register type.
14435 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14437         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14438         (microblaze_legitimate_pic_operand): Likewise
14439         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14440         new function microblaze_legitimate_pic_operand
14441         * config/microblaze/microblaze-protos.h
14442         (microblaze_legitimate_pic_operand): Declare.
14444 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14446         * config/microblaze/predicates.md (call_insn_simple_operand):
14447         New predicate for supported rtx code types.
14448         * config/microblaze/microblaze.md (call_internal1): Use
14449         call_insn_simple_operand predicate.
14451 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14453         PR middle-end/56461
14454         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14455         partitions.ordered_remove.
14457         PR middle-end/56461
14458         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14459         vec_oprnds0.create (1) for modifier == NONE.
14461         PR middle-end/56461
14462         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14463         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14464         vec_oprnds1 right before pushing anything to it for
14465         scalar_shift_arg.
14467         PR middle-end/56461
14468         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14469         set nbbs to 0 instead of having separate code path.
14470         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14471         instead of false as last argument if returning NULL.
14473 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14475         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14476         the attribute is now called "target" instead of "option".
14477         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14478         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14479         attribute/pragma name for TARGET_OPTION_VALID_P and
14480         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14481         * doc/tm.texi: Regenerated.
14483 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14485         * config/microblaze/microblaze.c:
14486         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14487         * config/microblaze/microblaze.h: Add -mxl-reorder to
14488         DRIVER_SELF_SPECS.
14489         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14490         instructions emitted if TARGET_REORDER.
14491         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14492         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14493         separately.
14495 2013-03-01  Xinliang David Li  <davidxl@google.com>
14497         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14498         walk length.
14500 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14502         PR middle-end/56461
14503         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14504         vector even when returning true.  Fix up function comment formatting.
14506         PR middle-end/56461
14507         * ira-build.c (ira_loop_nodes_count): New variable.
14508         (create_loop_tree_nodes): Initialize it.
14509         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14511         PR middle-end/56461
14512         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14513         method on dr_chain and result_chain.
14514         * tree-vect-stmts.c (vectorizable_store): Only call
14515         result_chain.create if j == 0.
14517         PR middle-end/56461
14518         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14519         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14520         before overwriting it.
14522 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14524         * doc/extended.texi (C Extensions): Change order in @menu
14525         to match @node.
14526         (Other MIPS Built-in Functions): Move last MIPS entry before
14527         "picoChip Built-in Functions".
14528         (SH Built-in Functions): Move after RX Built-in Functions.
14529         * doc/gcc.texi (Introduction): Change order in @menu
14530         to match @node.
14531         * doc/md.texi (Constraints): Ditto.
14532         * gty.texi (Type Information): Ditto.
14533         (User-provided marking routines for template types): Make
14534         subsection.
14535         * doc/invoke.texi (AArch64 Options): Move before
14536         "Adapteva Epiphany Options".
14538 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14539             Jakub Jelinek  <jakub@redhat.com>
14541         PR sanitizer/56454
14542         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14543         no_address_safety_analysis attribute.
14544         * doc/extend.texi (no_address_safety_attribute): Rename to
14545         no_sanitize_address attribute, mention no_address_safety_analysis
14546         attribute as deprecated alias.
14548 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
14550         PR middle-end/56461
14551         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14552         type to vec<vec<tree> > *.
14553         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
14554         to be vec<tree> instead of vec<tree> *, set vec_defs
14555         to vNULL and call vec_defs.create (number_of_vects), adjust other
14556         uses of vec_defs.
14557         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14558         vectorizable_condition): Adjust vect_get_slp_defs callers.
14560 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14562         * config/aarch64/aarch64.c
14563         (aarch64_float_const_representable): Remove unused variable.
14565 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14567         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14569 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14571         * config/aarch64/aarch64-builtins.c
14572         (aarch64_init_simd_builtins): Make static.
14574 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14576         * config/aarch64/aarch64.c
14577         (aarch64_simd_make_constant): Make static.
14579 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14581         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14582         with no initialization to the RHS of debug statements.
14584 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14586         PR tree-optimization/56294
14587         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14588         Adjust dumping.
14589         (get_access_replacement): Do not call create_access_replacement.
14590         Assert a replacement exists.
14591         (get_repl_default_def_ssa_name): Create the replacement declaration
14592         itself.
14594 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14596         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14597         final_end_function.
14599 2013-02-28  Marek Polacek  <polacek@redhat.com>
14601         PR rtl-optimization/56466
14602         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14603         if we're changing a loop.
14604         (peel_loops_completely): Likewise.
14606 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
14608         PR c++/55813
14609         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14611 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14613         PR target/56445
14614         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14615         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14616         INTX_FTYPE_FX, FX_FTYPE_INTX.
14617         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14619 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14621         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14622         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14623         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14624         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14625         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14626         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14627         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14628         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14629         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14630         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14631         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14632         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14633         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14634         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14635         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14636         (avrxmega6): Increase max flash segments from 5 to 6.
14637         * config/avr/t-multilib: Regenerate.
14638         * config/avr/avr-tables.opt: Regenerate.
14639         * doc/avr-mmcu.texi: Regenerate.
14641 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14643         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14644         (avr_device_to_arch): Rename to avr_device_to_ld.
14645         (avr_device_to_as): New prototype.
14646         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14647         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14648         * config/avr/driver-avr.c (avr_device_to_as): New.
14649         (avr_device_to_arch): Rename to avr_device_to_ld.
14651 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14653         PR middle-end/56461
14654         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14655         method on dr_chain and result_chain.
14657         PR middle-end/56461
14658         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14659         pointer_set_destroy on not_executed_last_iteration.
14661         PR middle-end/56461
14662         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
14664         PR middle-end/56461
14665         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14666         FOR_EACH_DEFINED_FUNCTION when freeing state.
14668         PR middle-end/56461
14669         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14670         pool_free.
14671         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14672         overwriting it.
14674         PR middle-end/56461
14675         * ipa-cp.c (decide_whether_version_node): Call vec_free on
14676         known_aggs[i].items and release known_aggs vector.
14678         PR middle-end/56461
14679         * ipa-reference.c (propagate): Free node_info even for alias nodes.
14681 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14683         * config/microblaze/microblaze.c (microblaze_emit_compare):
14684         Use xor for EQ/NE comparisions.
14685         * config/microblaze/microblaze.md (cstoresf4): Add constraints
14686         (cbranchsf4): Adjust operator to comparison_operator.
14688 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14690         PR middle-end/56461
14691         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14692         vector.
14693         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14694         vec_safe_push, always update *slot.
14695         (redirect_edge_var_map_clear): Use vec_free.
14696         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14697         (free_var_map_entry): Use vec_free.
14698         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14699         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14701 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
14703         PR middle-end/45472
14704         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14705         when the may_trap_p bit of the exprs being merged differs.
14706         Reorder tests for speculativeness in the logical and operator.
14708 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14710         * incpath.c (add_standard_paths): Use reconcat instead of concat
14711         where appropriate and avoid leaking memory.
14713         * opts.h: Include obstack.h.
14714         (opts_concat): New prototype.
14715         (opts_obstack): New declaration.
14716         * opts.c (opts_concat): New function.
14717         (opts_obstack): New variable.
14718         (init_options_struct): Call gcc_init_obstack on opts_obstack.
14719         (finish_options): Use opts_concat instead of concat
14720         and XOBNEWVEC instead of XNEWVEC.
14721         * opts-common.c (generate_canonical_option, decode_cmdline_option,
14722         generate_option): Likewise.
14723         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14724         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14726         PR target/56455
14727         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14728         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14730 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14732         PR middle-end/56461
14733         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14735 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14737         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14738         (arm_block_move_unaligned_straight): Likewise.
14739         (arm_adjust_block_mem): Likewise.
14741 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14743         PR target/48901
14744         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14745         temp, cond and label.
14746         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14748         PR target/52500
14749         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14750         * config/c6x/c6x.h (dbx_register_map): Update declaration.
14752         PR target/52501
14753         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14754         of prologue/epilogue functions.
14756         PR target/52550
14757         * config/tilegx/tilegx.c (tilegx_expand_prologue):
14758         Remove unused variable cfa_offset.
14759         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14761         PR target/54639
14762         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14763         type promotion to unsigned.
14765         PR target/54640
14766         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14767         for HOST_WIDE_INT of 32 bit / same size as int.
14768         (arm_block_move_unaligned_straight): Likewise.
14769         (arm_adjust_block_mem): Likewise.
14771         PR target/54662
14772         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14773         ALL_CFLAGS.
14775 2013-02-26  Marek Polacek  <polacek@redhat.com>
14777         PR tree-optimization/56426
14778         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
14780 2013-02-26  Richard Biener  <rguenther@suse.de>
14782         PR target/56444
14783         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14784         unused variable loops.
14786 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14788         PR tree-optimization/56448
14789         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14790         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14791         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14792         later operands of the references, or even first operand for
14793         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14795         PR tree-optimization/56443
14796         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14797         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14798         to type_for_mode langhook.
14800 2013-02-25  Matt Turner  <mattst88@gmail.com>
14802         * doc/invoke.texi: Document r4700.
14804 2013-02-25  Richard Biener  <rguenther@suse.de>
14806         PR tree-optimization/56175
14807         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14808         split out from ...
14809         (simplify_bitwise_binary): ... here.  Also guard the conversion
14810         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14812 2013-02-25  Catherine Moore  <clm@codesourcery.com>
14814         Revert:
14815         2013-02-24  Catherine Moore  <clm@codesourcery.com>
14816             Maciej W. Rozycki  <macro@codesourcery.com>
14817             Tom de Vries  <tom@codesourcery.com>
14818             Nathan Sidwell  <nathan@codesourcery.com>
14819             Iain Sandoe  <iain@codesourcery.com>
14820             Nathan Froyd  <froydnj@codesourcery.com>
14821             Chao-ying Fu  <fu@mips.com>
14823         * doc/extend.texi: (micromips, nomicromips, nocompression):
14824         Document new function attributes.
14825         * doc/invoke.texi (minterlink-compressed, mmicromips,
14826         m14k, m14ke, m14kec): Document new options.
14827         (minterlink-mips16): Update documentation.
14828         * doc/md.texi (ZC, ZD): Document new constraints.
14829         * configure.ac (gcc_cv_as_micromips): Check if linker
14830         supports the .set micromips directive.
14831         * configure: Regenerate.
14832         * config.in: Regenerate.
14833         * config/mips/mips-tables.opt: Regenerate.
14834         * config/mips/micromips.md: New file.
14835         * constraints.md (ZC, AD): New constraints.
14836         * config/mips/predicates.md (movep_src_register): New predicate.
14837         (movep_src_operand): New predicate.
14838         (non_volatile_mem_operand): New predicate.
14839         * config/mips/mips.md (multimem): New type.
14840         (length): Differentiate between 17-bit and 18-bit branch offsets.
14841         (MOVEP1, MOVEP2): New mode iterator.
14842         (mov_<load>l): Use ZC constraint.
14843         (mov_<load>r): Likewise.
14844         (mov_<store>l): Likewise.
14845         (mov_<store>r): Likewise.
14846         (*branch_equality<mode>_inverted): Add microMIPS support.
14847         (*branch_equality<mode>): Likewise.
14848         (*jump_absolute): Likewise.
14849         (indirect_jump_<mode>): Likewise.
14850         (tablejump_<mode>): Likewise.
14851         (<optab>_internal): Likewise.
14852         (sibcall_internal): Likewise.
14853         (sibcall_value_internal): Likewise.
14854         (prefetch): Use constraint ZD.
14855         * config/mips/mips.opt (minterlink-compressed): New option.
14856         (minterlink-mips16): Now an alias for minterlink-compressed.
14857         (mmicromips): New option.
14858         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14859         (compare_and_swap_12): Likewise.
14860         (sync_add<mode>): Likewise.
14861         (sync_<optab>_12): Likewise.
14862         (sync_old_<optab>_12): Likewise.
14863         (sync_new_<optab>_12): Likewise.
14864         (sync_nand_12): Likewise.
14865         (sync_old_nand_12): Likewise.
14866         (sync_new_nand_12): Likewise.
14867         (sync_sub<mode>): Likewise.
14868         (sync_old_add<mode>): Likewise.
14869         (sync_old_sub<mode>): Likewise.
14870         (sync_new_add<mode>): Likewise.
14871         (sync_new_sub<mode>): Likewise.
14872         (sync_<optab><mode>): Likewise.
14873         (sync_old_<optab><mode>): Likewise.
14874         (sync_new_<optab><mode>): Likewise.
14875         (sync_nand<mode>): Likewise.
14876         (sync_old_nand<mode>): Likewise.
14877         (sync_new_nand<mode>): Likewise.
14878         (sync_lock_test_and_set<mode>): Likewise.
14879         (test_and_set_12): Likewise.
14880         (atomic_compare_and_swap<mode>): Likewise.
14881         (atomic_exchange<mode>_llsc): Likewise.
14882         (atomic_fetch_add<mode>_llsc): Likewise.
14883         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14884         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14885         (umips_save_restore_pattern_p): Likewise.
14886         (umips_load_store_pair_p): Likewise.
14887         (umips_output_load_store_pair): Likewise.
14888         (umips_movep_target_p): Likewise.
14889         (umips_12bit_offset_address_p): Likewise.
14890         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14891         (mips_base_mips16): Rename this...
14892         (mips_base_compression_flags): ...to this. Update all uses.
14893         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14894         (mips_mips16_decl_p): Delete.
14895         (mips_nomips16_decl_p): Delete.
14896         (mips_get_compress_on_flags): New function.
14897         (mips_get_compress_off_flags): New function.
14898         (mips_get_compress_mode): New function.
14899         (mips_get_compress_on_name): New function.
14900         (mips_get_compress_off_name): New function.
14901         (mips_insert_attributes): Support multiple compression types.
14902         (mips_merge_decl_attributes): Likewise.
14903         (umips_12bit_offset_address_p): New function.
14904         (mips_start_function_definition): Emit .set micromips directive.
14905         (mips_call_may_need_jalx_p): New function.
14906         (mips_function_ok_for_sibcall): Add microMIPS support.
14907         (mips_print_operand_punctuation): Support short delay slots and
14908         compact jumps.
14909         (umips_swm_mask, umips_swm_encoding): New.
14910         (umips_build_save_restore): New function.
14911         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14912         (was_mips16_p): Remove.
14913         (old_compression_mode): New.
14914         (mips_set_compression_mode): New function.
14915         (mips_set_current_function): Add microMIPS support.
14916         (mips_option_override): Likewise.
14917         (umips_save_restore_pattern_p): New function.
14918         (umips_output_save_restore): New function.
14919         (umips_load_store_pair_p_1): New function.
14920         (umips_load_store_pair_p): New function.
14921         (umips_output_load_store_pair_1): New function.
14922         (umips_output_load_store_pair): New function.
14923         (umips_movep_target_p) New function.
14924         (mips_prepare_pch_save): Add microMIPS support.
14925         * config/mips/mips.h (TARGET_COMPRESSION): New.
14926         (TARGET_CPU_CPP_BUILTINS): Update macro
14927         to use new compression flags and to support microMIPS.
14928         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14929         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14930         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14931         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14932         (ASM_SPEC): Support mmicromips and mno-micromips.
14933         (M16STORE_REG_P): New macro.
14934         (MIPS_CALL): Support TARGET_MICROMIPS.
14935         (MICROMIPS_J): New macro.
14936         (mips_base_mips16): Rename this...
14937         (mips_base_compression_flags): ...to this.
14938         (UMIPS_12BIT_OFFSET_P): New macro.
14939         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14940         (MULTILIB_DIRNAMES): Likewise.
14942 2013-02-25  Tom de Vries  <tom@codesourcery.com>
14944         PR rtl-optimization/56131
14945         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
14946         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
14947         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
14949 2013-02-25  Tobias Burnus  <burnus@net-b.de>
14951         * doc/invoke.texi (-fsanitize=): Move from optimization
14952         to debugging options.
14954 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14956         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
14958 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14959             Alexander Monakov  <amonakov@ispras.ru>
14961         PR middle-end/56077
14962         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
14963         flush pending lists also on non-jumps.  Adjust comment.
14965 2013-02-24  Catherine Moore  <clm@codesourcery.com>
14966             Maciej W. Rozycki  <macro@codesourcery.com>
14967             Tom de Vries  <tom@codesourcery.com>
14968             Nathan Sidwell  <nathan@codesourcery.com>
14969             Iain Sandoe  <iain@codesourcery.com>
14970             Nathan Froyd  <froydnj@codesourcery.com>
14971             Chao-ying Fu  <fu@mips.com>
14973         * doc/extend.texi: (micromips, nomicromips, nocompression):
14974         Document new function attributes.
14975         * doc/invoke.texi (minterlink-compressed, mmicromips,
14976         m14k, m14ke, m14kec): Document new options.
14977         (minterlink-mips16): Update documentation.
14978         * doc/md.texi (ZC, ZD): Document new constraints.
14979         * configure.ac (gcc_cv_as_micromips): Check if linker
14980         supports the .set micromips directive.
14981         * configure: Regenerate.
14982         * config.in: Regenerate.
14983         * config/mips/mips-tables.opt: Regenerate.
14984         * config/mips/micromips.md: New file.
14985         * constraints.md (ZC, AD): New constraints.
14986         * config/mips/predicates.md (movep_src_register): New predicate.
14987         (movep_src_operand): New predicate.
14988         (non_volatile_mem_operand): New predicate.
14989         * config/mips/mips.md (multimem): New type.
14990         (length): Differentiate between 17-bit and 18-bit branch offsets.
14991         (MOVEP1, MOVEP2): New mode iterator.
14992         (mov_<load>l): Use ZC constraint.
14993         (mov_<load>r): Likewise.
14994         (mov_<store>l): Likewise.
14995         (mov_<store>r): Likewise.
14996         (*branch_equality<mode>_inverted): Add microMIPS support.
14997         (*branch_equality<mode>): Likewise.
14998         (*jump_absolute): Likewise.
14999         (indirect_jump_<mode>): Likewise.
15000         (tablejump_<mode>): Likewise.
15001         (<optab>_internal): Likewise.
15002         (sibcall_internal): Likewise.
15003         (sibcall_value_internal): Likewise.
15004         (prefetch): Use constraint ZD.
15005         * config/mips/mips.opt (minterlink-compressed): New option.
15006         (minterlink-mips16): Now an alias for minterlink-compressed.
15007         (mmicromips): New option.
15008         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15009         (compare_and_swap_12): Likewise.
15010         (sync_add<mode>): Likewise.
15011         (sync_<optab>_12): Likewise.
15012         (sync_old_<optab>_12): Likewise.
15013         (sync_new_<optab>_12): Likewise.
15014         (sync_nand_12): Likewise.
15015         (sync_old_nand_12): Likewise.
15016         (sync_new_nand_12): Likewise.
15017         (sync_sub<mode>): Likewise.
15018         (sync_old_add<mode>): Likewise.
15019         (sync_old_sub<mode>): Likewise.
15020         (sync_new_add<mode>): Likewise.
15021         (sync_new_sub<mode>): Likewise.
15022         (sync_<optab><mode>): Likewise.
15023         (sync_old_<optab><mode>): Likewise.
15024         (sync_new_<optab><mode>): Likewise.
15025         (sync_nand<mode>): Likewise.
15026         (sync_old_nand<mode>): Likewise.
15027         (sync_new_nand<mode>): Likewise.
15028         (sync_lock_test_and_set<mode>): Likewise.
15029         (test_and_set_12): Likewise.
15030         (atomic_compare_and_swap<mode>): Likewise.
15031         (atomic_exchange<mode>_llsc): Likewise.
15032         (atomic_fetch_add<mode>_llsc): Likewise.
15033         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15034         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15035         (umips_save_restore_pattern_p): Likewise.
15036         (umips_load_store_pair_p): Likewise.
15037         (umips_output_load_store_pair): Likewise.
15038         (umips_movep_target_p): Likewise.
15039         (umips_12bit_offset_address_p): Likewise.
15040         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15041         (mips_base_mips16): Rename this...
15042         (mips_base_compression_flags): ...to this. Update all uses.
15043         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15044         (mips_mips16_decl_p): Delete.
15045         (mips_nomips16_decl_p): Delete.
15046         (mips_get_compress_on_flags): New function.
15047         (mips_get_compress_off_flags): New function.
15048         (mips_get_compress_mode): New function.
15049         (mips_get_compress_on_name): New function.
15050         (mips_get_compress_off_name): New function.
15051         (mips_insert_attributes): Support multiple compression types.
15052         (mips_merge_decl_attributes): Likewise.
15053         (umips_12bit_offset_address_p): New function.
15054         (mips_start_function_definition): Emit .set micromips directive.
15055         (mips_call_may_need_jalx_p): New function.
15056         (mips_function_ok_for_sibcall): Add microMIPS support.
15057         (mips_print_operand_punctuation): Support short delay slots and
15058         compact jumps.
15059         (umips_swm_mask, umips_swm_encoding): New.
15060         (umips_build_save_restore): New function.
15061         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15062         (was_mips16_p): Remove.
15063         (old_compression_mode): New.
15064         (mips_set_compression_mode): New function.
15065         (mips_set_current_function): Add microMIPS support.
15066         (mips_option_override): Likewise.
15067         (umips_save_restore_pattern_p): New function.
15068         (umips_output_save_restore): New function.
15069         (umips_load_store_pair_p_1): New function.
15070         (umips_load_store_pair_p): New function.
15071         (umips_output_load_store_pair_1): New function.
15072         (umips_output_load_store_pair): New function.
15073         (umips_movep_target_p) New function.
15074         (mips_prepare_pch_save): Add microMIPS support.
15075         * config/mips/mips.h (TARGET_COMPRESSION): New.
15076         (TARGET_CPU_CPP_BUILTINS): Update macro
15077         to use new compression flags and to support microMIPS.
15078         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15079         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15080         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15081         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15082         (ASM_SPEC): Support mmicromips and mno-micromips.
15083         (M16STORE_REG_P): New macro.
15084         (MIPS_CALL): Support TARGET_MICROMIPS.
15085         (MICROMIPS_J): New macro.
15086         (mips_base_mips16): Rename this...
15087         (mips_base_compression_flags): ...to this.
15088         (UMIPS_12BIT_OFFSET_P): New macro.
15089         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15090         (MULTILIB_DIRNAMES): Likewise.
15092 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15094         PR target/52555
15095         * target-globals.c (save_target_globals): For init_reg_sets and
15096         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15098 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15100         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15101         * config/aarch64/t-aarch64
15102         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15104 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15106         PR inline-asm/56148
15107         * lra-constraints.c (process_alt_operands): Reload operand
15108         conflicting with earlier clobber only if no more other conflicting
15109         operands.
15111 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15113         PR sanitizer/56393
15114         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15115         if not linking a shared library.
15117 2013-02-22  Seth LaForge  <sethml@google.com>
15119         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15121 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15123         * config/arm/arm.md (split for extendsidi): Update condition.
15124         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15125         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15126         (qhs_zextenddi_cstr): Likewise.
15128 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15130         PR middle-end/56420
15131         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15132         avoid signed wrapping.
15133         (expand_mult): Handle properly multiplication by
15134         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15135         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15136         in the compiler if coeff is HOST_WIDE_INT_MIN.
15137         (expand_divmod): Don't make ext_op1 static, change it's type to
15138         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15140         PR rtl-optimization/50339
15141         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15142         field.
15143         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15144         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15145         into splitting_ashiftrt field.
15146         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15147         ASHIFTRT.
15148         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15149         choices.
15151 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15153         PR middle-end/56108
15154         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15155         are sure to go irrevocable.
15157 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15159         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15160         scalars are valid operands.
15162 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15164         PR tree-optimization/56310
15165         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15166         only matching indices and non-negative final offsets.
15167         (intersect_aggregates_with_edge): Pass src_idx to
15168         agg_replacements_to_vector.  Pass src_idx insstead of index to
15169         intersect_with_agg_replacements.
15171 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15173         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15174         instead of hard-wired defaults.
15176 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15178         * doc/invoke.texi (MIPS Options): Update documentation of the
15179         floating-point multiply-accumulate instruction restrictions.
15181 2013-02-21  Kostya Serebryany  <kcc@google.com>
15183         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15184         asan_shadow_offset on x86_64 linux.
15186 2013-02-21  Richard Biener  <rguenther@suse.de>
15188         PR tree-optimization/56415
15189         Revert
15190         2013-02-11  Richard Biener  <rguenther@suse.de>
15192         PR tree-optimization/56273
15193         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15194         first VRP run.
15196 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15198         PR bootstrap/56258
15199         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15200         instead of @itemx.
15202         PR inline-asm/56405
15203         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15204         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15206 2013-02-20  Jan Hubicka  <jh@suse.cz>
15208         PR tree-optimization/56265
15209         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15210         when target is referenced for first time.
15212 2013-02-20  Richard Biener  <rguenther@suse.de>
15214         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15215         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15216         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15217         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15218         not return anything.
15219         (rename_ssa_copies): Do not remove unused locals.
15220         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15221         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15222         * passes.c (execute_function_todo): Do not schedule unused locals
15223         removal if cleanup_tree_cfg did something.
15224         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15225         about the number of removed locals.
15227 2013-02-20  Richard Biener  <rguenther@suse.de>
15229         PR tree-optimization/56398
15230         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15232 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15234         PR tree-optimization/55334
15235         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15236         restricted pointers to arrays.
15238 2013-02-20  Richard Biener  <rguenther@suse.de>
15239             Jakub Jelinek  <jakub@redhat.com>
15241         PR tree-optimization/56396
15242         * tree-ssa-ccp.c (n_const_val): New static variable.
15243         (get_value): Return NULL for SSA names we don't have a lattice
15244         entry for.
15245         (ccp_initialize): Initialize n_const_val.
15246         * tree-ssa-copy.c (n_copy_of): New static variable.
15247         (init_copy_prop): Initialize n_copy_of.
15248         (get_value): Return NULL_TREE for SSA names we don't have a
15249         lattice entry for.
15251 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15253         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15255 2013-02-20  Richard Biener  <rguenther@suse.de>
15257         * genpreds.c (write_lookup_constraint): Do not compare first
15258         letter of the constraint again.
15260 2013-02-20  Richard Biener  <rguenther@suse.de>
15262         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15263         and ceil_log2.
15264         (get_use_iv_cost): Terminate hashtable walk when coming across
15265         an empty entry.
15267 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15269         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15270         reassociation for avx2 targets.
15272 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15274         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15275         Add version check for v8.10.a to enable microblaze_has_clz
15276         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15277         version and TARGET_PATTERN_COMPARE check
15278         * config/microblaze/microblaze.md: New clzsi2 instruction
15280 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15282         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15283         function before branching.
15285 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15287         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15288         DUMP_INSN_RTX_UID.
15289         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15291 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15293         PR middle-end/55889
15294         * sel-sched.c: Include ira.h.
15295         (implicit_clobber_conflict_p): New function.
15296         (moveup_expr): Use it.
15297         * Makefile.in (sel-sched.o): Depend on ira.h.
15299 2013-02-19  Richard Biener  <rguenther@suse.de>
15301         PR tree-optimization/56384
15302         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15303         (vn_hash_type): Split out from ...
15304         (vn_hash_constant_with_type): ... here.
15305         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15306         (vn_phi_eq): Compare types from vn_phi_s structure.
15307         (vn_phi_lookup): Populate vn_phi_s type.
15308         (vn_phi_insert): Likewise.
15310 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15312         PR tree-optimization/56350
15313         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15314         if haven't found reduction or nested cycle operand, rather than
15315         asserting we must find it.
15317         PR tree-optimization/56381
15318         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15319         to fold_build3.
15321 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15322             Jakub Jelinek  <jakub@redhat.com>
15324         PR target/52555
15325         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15326         (swap_optab_enable): Same.
15327         (init_all_optabs): Use argument instead of global.
15328         * tree.h (struct tree_optimization_option): New field target_optabs.
15329         * expr.h (init_all_optabs): Add argument to prototype.
15330         (TREE_OPTIMIZATION_OPTABS): New.
15331         (save_optabs_if_changed): Protoize.
15332         * optabs.h: Declare this_fn_optabs.
15333         * optabs.c (save_optabs_if_changed): New.
15334         Declare this_fn_optabs.
15335         (init_optabs): Add argument to init_all_optabs() call.
15336         * function.c (invoke_set_current_function_hook): Handle per
15337         function optabs.
15338         * function.h (struct function): New field optabs.
15339         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15340         optimization_current_node has changed.
15341         * target-globals.h (save_target_globals_default_opts): Protoize.
15342         * target-globals.c (save_target_globals_default_opts): New.
15344 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15346         PR target/56347
15347         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15348         registers %fr12 and %fr12R as call used.
15350         PR target/56214
15351         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15352         and HImode, require all displacements to be an integer multiple of
15353         their mode size.
15354         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15355         only allow QImode and HImode when reload is in progress and strict is
15356         true.  Likewise for symbolic addresses.  Use base14_operand to check
15357         displacements in REG+BASE addresses.
15359 2013-02-18  Richard Biener  <rguenther@suse.de>
15361         PR tree-optimization/56366
15362         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15363         handle sign-conversion of outer-loop initial induction value.
15365 2013-02-18  Richard Biener  <rguenther@suse.de>
15367         PR middle-end/56349
15368         * cfghooks.c (merge_blocks): If we merge a latch into another
15369         block adjust references to it.
15370         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15371         (verify_loop_structure): Verify that a recorded latch is in fact
15372         a latch.
15374 2013-02-18  Richard Biener  <rguenther@suse.de>
15376         PR tree-optimization/56321
15377         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15378         order SSA name release and virtual operand unlinking.
15380 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15382         * config/microblaze/microblaze.md (save_stack_block): Define.
15383         (restore_stack_block): Likewise.
15385 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15387         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15388         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15389         * config/microblaze/microblaze.c (microblaze_option_override):
15390         Bail out early for PIC modes when target does not support PIC.
15392 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15394         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15395         Replace with a microblaze version.
15396         (microblaze_trampoline_init): Adapt for microblaze.
15397         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15398         microblaze.
15400 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15401             Dodji Seketeli  <dodji@redhat.com>
15403         PR asan/56330
15404         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15405         (instrument_mem_region_access): Do not forget to always put
15406         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15407         0) statement, even for cases where either 'base' or 'base + len'
15408         are not instrumented -- because they have been previously
15409         instrumented.  Simplify the logic by putting all the statements
15410         instrument 'base + len' inside a sequence, and then insert that
15411         sequence right before the current insertion point.  Then, to
15412         instrument 'base + len', just get an iterator on that statement.
15413         And do not forget to update the pointer to iterator the function
15414         received as argument.
15416 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15418         PR rtl-optimization/56348
15419         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15421 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15423         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15424         (clean_graph_dump_file): Pass base to start_graph_dump.
15426 2013-02-14  Richard Henderson  <rth@redhat.com>
15428         PR target/55941
15429         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15431 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15433         * collect2-aix.h: Define F_LOADONLY.
15435 2013-02-14  Richard Biener  <rguenther@suse.de>
15437         PR lto/50494
15438         * varasm.c (output_constant_def_1): Get the decl representing
15439         the constant as argument.
15440         (output_constant_def): Wrap output_constant_def_1.
15441         (make_decl_rtl): Use output_constant_def_1 with the decl
15442         representing the constant.
15443         (build_constant_desc): Optionally re-use a decl already
15444         representing the constant.
15445         (tree_output_constant_def): Adjust.
15447 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15449         Fix an asan crash
15450         * asan.c (instrument_builtin_call):  Really put the length of the
15451         second source argument into src1_len.
15453 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15455         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15456         argument.  If it is false, don't create edge from then_bb to
15457         fallthru_bb.
15458         (insert_if_then_before_iter): Pass true to it.
15459         (build_check_stmt): Pass false to it.
15460         (transform_statements): Flush hash table only on extended basic
15461         block boundaries, rather than at the beginning of every bb.
15462         Don't flush hash table on nonfreeing_call_p calls.
15463         * tree-flow.h (nonfreeing_call_p): New prototype.
15464         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15466 2013-02-13  David S. Miller  <davem@davemloft.net>
15468         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15470 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15472         PR target/56184
15473         * ira.c (max_regno_before_ira): Move from ...
15474         (ira): ... here.
15475         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15476         vec_safe_length.
15478 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15480         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15482 2013-02-13  Richard Biener  <rguenther@suse.de>
15484         PR lto/56295
15485         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15486         globals in MEM_REFs.
15488 2013-02-13  Richard Biener  <rguenther@suse.de>
15490         * loop-init.c (loop_optimizer_init): Clear loop state when
15491         re-initializing preserved loops.
15492         * loop-unswitch.c (unswitch_single_loop): Return whether
15493         we unswitched the loop.  Do not verify loop state here.
15494         (unswitch_loops): When we unswitched a loop discover new loops.
15496 2013-02-13  Kostya Serebryany  <kcc@google.com>
15498         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15499         on x86_64 linux.
15500         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15502 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15504         Avoid instrumenting duplicated memory access in the same basic block
15505         * Makefile.in (asan.o): Add new dependency on hash-table.h
15506         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15507         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15508         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15509         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15510         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15511         (get_mem_ref_of_assignment): New functions.
15512         (get_mem_refs_of_builtin_call): Extract from
15513         instrument_builtin_call and tweak a little bit to make it fit with
15514         the new signature.
15515         (instrument_builtin_call): Use the new
15516         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15517         of is_gimple_builtin_call.
15518         (instrument_derefs, instrument_mem_region_access): Insert the
15519         instrumented memory reference into the hash table.
15520         (maybe_instrument_assignment): Renamed instrument_assignment into
15521         this, and change it to advance the iterator when instrumentation
15522         actually happened and return true in that case.  This makes it
15523         homogeneous with maybe_instrument_assignment, and thus give a
15524         chance to callers to be more 'regular'.
15525         (transform_statements): Clear the memory reference hash table
15526         whenever we enter a new BB, when we cross a function call, or when
15527         we are done transforming statements.  Use
15528         maybe_instrument_assignment instead of instrumentation.  No more
15529         need to special case maybe_instrument_assignment and advance the
15530         iterator after calling it; it's now handled just like
15531         maybe_instrument_call.  Update comment.
15533 2013-02-13  Richard Biener  <rguenther@suse.de>
15535         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15536         Fix loop discovery code.
15538 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15540         PR inline-asm/56148
15541         * lra-constraints.c (process_alt_operands): Match early clobber
15542         operand with itself.  Check conflicts with earlyclobber only if
15543         the operand is not reloaded.  Prefer to reload conflicting operand
15544         if earlyclobber and matching operands are the same.
15546 2013-02-12  Richard Biener  <rguenther@suse.de>
15548         PR lto/56297
15549         * lto-streamer-out.c (write_symbol): Do not output symbols
15550         for hard register variables.
15552 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
15554         PR target/54222
15555         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15556         (umulsidi3_insn, mulsidi3_insn): New insns.
15558 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
15560         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15561         (struct tune_params): Add vec_costs field.
15562         * config/arm/arm.c (arm_builtin_vectorization_cost)
15563         (arm_add_stmt_cost): New functions.
15564         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15565         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15566         (arm_default_vec_cost): New struct of type cpu_vec_costs.
15567         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15568         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15569         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15570         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15572 2013-02-12  Richard Biener  <rguenther@suse.de>
15574         PR lto/56295
15575         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15576         decls again if possible.
15578 2013-02-12  Richard Biener  <rguenther@suse.de>
15580         PR middle-end/56288
15581         * tree-ssa.c (verify_ssa_name): Fix check, move
15582         SSA_NAME_IN_FREE_LIST check up.
15584 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
15585             Steven Bosscher   <steven@gcc.gnu.org>
15587         PR rtl-optimization/56151
15588         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15589         equal to op0 or op1, and last_insn pattern is CODE operation
15590         with MEM dest and one of the operands matches that MEM.
15592 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
15594         * doc/extend.texi: Document Function Multiversioning and "default"
15595         parameter string to target attribute.
15596         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15597         target attribute parameter is "default".
15598         (ix86_compare_version_priority): Remove checks for target attribute.
15599         (ix86_mangle_function_version_assembler_name): Change error to sorry.
15600         Remove check for target attribute equal to NULL. Add assert.
15601         (ix86_generate_version_dispatcher_body): Change error to sorry.
15603 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
15604             Jack Howarth  <howarth@bromo.med.uc.edu>
15605             Patrick Marlier  <patrick.marlier@gmail.com>
15607         PR libitm/55693
15608         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15609         define ENDFILE_SPEC as TM_DESTRUCTOR.
15610         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15612 2013-02-11  Alexander Potapenko  <glider@google.com>
15613             Jack Howarth  <howarth@bromo.med.uc.edu>
15614             Jakub Jelinek  <jakub@redhat.com>
15616         PR sanitizer/55617
15617         * config/darwin.c (cdtor_record): Rename ctor_record.
15618         (sort_cdtor_records): Rename sort_ctor_records.
15619         (finalize_dtors): New routine to sort destructors by
15620         priority before use in assemble_integer.
15621         (machopic_asm_out_destructor): Use finalize_dtors if needed.
15623 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
15625         PR rtl-optimization/56275
15626         * simplify-rtx.c (avoid_constant_pool_reference): Check that
15627         offset is non-negative and less than cmode size before
15628         calling simplify_subreg.
15630 2013-02-11  Richard Biener  <rguenther@suse.de>
15632         PR tree-optimization/56264
15633         * cfgloop.h (fix_loop_structure): Adjust prototype.
15634         * loop-init.c (fix_loop_structure): Return the number of
15635         newly discovered loops.
15636         * tree-cfgcleanup.c (repair_loop_structures): When new loops
15637         are discovered, do a full loop-closed SSA rewrite.
15639 2013-02-11  Richard Biener  <rguenther@suse.de>
15641         PR tree-optimization/56273
15642         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15643         first VRP run.
15644         (check_array_ref): Fix missing newline in dumps.
15645         (search_for_addr_array): Likewise.
15647 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
15649         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15651 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
15653         PR target/56256
15654         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15656 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
15658         PR rtl-optimization/56246
15659         * lra-constraints.c (simplify_operand_subreg): Try to reuse
15660         reload pseudo.
15661         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15662         constraints are satisfied.
15664 2013-02-08  Jeff Law  <law@redhat.com>
15666         PR debug/53948
15667         * emit-rtl.c (reg_is_parm_p): New function.
15668         * regs.h (reg_is_parm_p): New prototype.
15669         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15670         callee-clobbered registers.
15672 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
15674         PR target/56043
15675         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15676         If there is no implicit builtin declaration, just return NULL.
15678 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15680         * config/i386/sse.md (FMAMODEM): New mode iterator.
15681         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15682         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15684 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15686         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15687         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15688         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15690 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15692         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15693         (microblaze*-*-elf): Likewise.
15694         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15695         LINK_SPEC.
15696         * config/microblaze/microblaze-c.c: Add builtin defines for
15697         _LITTLE_ENDIAN and _BIG_ENDIAN.
15698         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15699         add to TARGET_DEFAULT flags.
15700         Expand ASM_SPEC and LINK_SPEC.
15701         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15702         * config/microblaze/microblaze.md: Update extendsidi2 and
15703         movdi_internal instructions to use low-order / high-order reg
15704         print_operands.
15705         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15706         options and inversemask / mask of LITTLE_ENDIAN.
15707         * config/microblaze/t-microblaze: Expand multilib options to
15708         include mlittle-endian (le) and update exceptions patterns.
15710 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
15712         PR rtl-optimization/56195
15713         * lra-constraints.c (get_reload_reg): Don't reuse regs
15714         if they have smaller mode than requested, if they have
15715         wider mode than requested, try to return a SUBREG.
15717         PR tree-optimization/56250
15718         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15719         if type is unsigned and code isn't MULT_EXPR.
15721 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15723         PR tree-optimization/56064
15724         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15725         bits according to mode.
15726         * fixed-value.h (fixed_from_double_int)
15727         (const_fixed_from_double_int): Adjust comments.
15729 2013-02-08  Richard Biener  <rguenther@suse.de>
15731         PR lto/56231
15732         * lto-streamer.h (struct data_in): Remove current_file, current_line
15733         and current_col members.
15734         * lto-streamer-out.c (lto_output_location): Stream changed bits
15735         en-block for efficiency.
15736         * lto-streamer-in.c (clear_line_info): Remove.
15737         (lto_input_location): Cache current file, line and column
15738         globally via local statics.  Read changed bits en-block.
15739         (input_function): Do not call clear_line_info.
15740         (lto_read_body): Likewise.
15741         (lto_input_toplevel_asms): Likewise.
15743 2013-02-08  Michael Matz  <matz@suse.de>
15745         PR tree-optimization/52448
15746         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15747         (nt_call_phase): New static.
15748         (add_or_mark_expr): Only mark accesses with newer phase than any
15749         call seen.
15750         (nonfreeing_call_p): New.
15751         (nt_init_block): Update nt_call_phase, mark blocks as visited.
15752         (nt_fini_block): Keep blocks marked as visited.
15753         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15755 2013-02-08  Richard Biener  <rguenther@suse.de>
15757         * ira.c (ira): Free broken dominator information.
15759 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15761         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15763 2013-02-08  Marek Polacek  <polacek@redhat.com>
15765         * cfgloop.c (verify_loop_structure): Add more checking of headers.
15767 2013-02-08  Richard Biener  <rguenther@suse.de>
15769         PR middle-end/56181
15770         * cfgloop.h (flow_loops_find): Adjust.
15771         (bb_loop_header_p): Declare.
15772         * cfgloop.c (bb_loop_header_p): New function split out from ...
15773         (flow_loops_find): ... here.  Adjust function signature,
15774         support incremental loop structure update.
15775         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
15776         * cfgloopmanip.c (fix_loop_structure): Move ...
15777         * loop-init.c (fix_loop_structure): ... here.
15778         (apply_loop_flags): Split out from ...
15779         (loop_optimizer_init): ... here.
15780         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
15781         in incremental mode, only remove dead loops here.
15783 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15785         PR target/54222
15786         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15787         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15788         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15789         (*round<mode>3.libgcc): New insns for fixed-modes.
15790         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15791         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15792         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15793         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15794         implementations.  Define to __builtin_avr_absFX,
15795         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15796         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15797         __builtin_avr_countlsFX, respectively.
15798         * config/avr/avr-c.c (target.h): Include it.
15799         (enum avr_builtin_id): New enum.
15800         (avr_resolve_overloaded_builtin): New static function.
15801         (avr_register_target_pragmas): Use it to set
15802         targetm.resolve_overloaded_builtin.
15803         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15804         tree nodes used by DEF_BUILTIN.
15805         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15806         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15807         <AVR_BUILTIN_xxBITS>: Same.
15809 2013-02-08  Richard Biener  <rguenther@suse.de>
15811         * cfgloop.c (verify_loop_structure): Properly handle
15812         a loop exiting to another loop header.
15813         * ira-int.h (ira_loops): Remove.
15814         * ira.c (ira_loops): Remove.
15815         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15816         (do_reload): Use loop_optimizer_finalize.
15817         * ira-build.c (create_loop_tree_nodes): Use get_loops and
15818         number_of_loops to access the loop tree.
15819         (more_one_region_p): Likewise.
15820         (finish_loop_tree_nodes): Likewise.
15821         (rebuild_regno_allocno_maps): Likewise.
15822         (mark_loops_for_removal): Likewise.
15823         (mark_all_loops_for_removal): Likewise.
15824         (remove_unnecessary_regions): Likewise.
15825         (ira_build): Likewise.
15826         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15828 2013-02-08  Richard Biener  <rguenther@suse.de>
15830         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15831         * ipa-pure-const.c (analyze_function): Avoid calling
15832         mark_irreducible_loops twice.
15833         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
15835 2013-02-07  David S. Miller  <davem@davemloft.net>
15837         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15838         on 'reg'.
15839         * var-tracking.c (vt_add_function_parameter): Test the presence of
15840         HAVE_window_save properly and do not remap argument registers when
15841         we have a leaf function.
15843 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
15845         PR bootstrap/56227
15846         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15847         instead of "ll".
15848         * config/i386/i386.c (ix86_print_operand): Ditto.
15850 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15852         * lra-constraints.c (process_alt_operands): Fix recently added comment.
15854 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15856         PR rtl-optimization/56225
15857         * lra-constraints.c (process_alt_operands): Check that reload hard
15858         reg can hold value for strict_low_part.
15860 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15862         PR debug/56154
15863         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15864         dwarf2out_end_function.
15865         (in_first_function_p, maybe_at_text_label_p,
15866         first_loclabel_num_not_at_text_label): New variables.
15867         (dwarf2out_var_location): In the first function find out
15868         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
15869         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15870         functions.
15872 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
15874         PR rtl-optimization/56178
15875         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15876         SUBREG of a register.  Tidy up related block of code.
15877         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15878         note if the source is a register or a SUBREG of a register.
15880 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15882         PR target/56228
15883         * config/rs6000/rs6000.md (ptrm): New mode attr.
15884         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
15885         call_value_indirect_aix<pttrsize>,
15886         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
15887         m in constraints.
15889 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
15891         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
15892         if -bnortl. Convert to strcmp and strncmp.
15894 2013-02-07  Alan Modra  <amodra@gmail.com>
15896         PR target/54009
15897         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
15898         addresses won't wrap when offsetting.
15899         (rs6000_secondary_reload): Provide secondary reloads needed for
15900         wrapping LO_SUM addresses.
15902 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
15904         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
15905         MACH, just __MACH__.
15907 2013-02-06  Richard Biener  <rguenther@suse.de>
15909         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
15910         instead of calling fix_loop_structure.
15912 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
15914         PR middle-end/56217
15915         * omp-low.c (use_pointer_for_field): Return false if
15916         lower_send_shared_vars doesn't generate any copy-out code.
15918 2013-02-06  Tom de Vries  <tom@codesourcery.com>
15920         PR rtl-optimization/56131
15921         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
15922         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
15923         of the label is NULL.  Add comment.
15925 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
15927         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
15929         PR sanitizer/55374
15930         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
15931         (STATIC_LIBTSAN_LIBS): Likewise.
15932         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
15933         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
15934         is defined, don't add anything else beyond that.
15935         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
15936         (LINK_COMMAND_SPEC): Use them.
15938         PR tree-optimization/56205
15939         * tree-stdarg.c (check_all_va_list_escapes): Return true if
15940         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
15941         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
15943 2013-02-05  Richard Biener  <rguenther@suse.de>
15945         PR tree-optimization/53342
15946         PR tree-optimization/53185
15947         * tree-vectorizer.h (vect_check_strided_load): Remove.
15948         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
15949         not disallow peeling for vectorized strided loads.
15950         (vect_check_strided_load): Make static and simplify.
15951         (vect_analyze_data_refs): Adjust.
15952         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
15953         correctly when vectorizing strided loads.
15955 2013-02-05  Richard Biener  <rguenther@suse.de>
15957         * doc/install.texi: Refer to ISL, not PPL.
15959 2013-02-05  Jan Hubicka  <jh@suse.cz>
15961         PR tree-optimization/55789
15962         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
15964 2013-02-05  Jan Hubicka  <jh@suse.cz>
15966         PR tree-optimization/55789
15967         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
15968         the dead call anyway.
15970 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
15972         PR sanitizer/55374
15973         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
15975 2013-02-04  Alexander Potapenko  <glider@google.com>
15976             Jack Howarth  <howarth@bromo.med.uc.edu>
15977             Jakub Jelinek  <jakub@redhat.com>
15979         PR sanitizer/55617
15980         * config/darwin.c (sort_ctor_records): Stabilized qsort
15981         on constructor priority by using original position.
15982         (finalize_ctors): New routine to sort constructors by
15983         priority before use in assemble_integer.
15984         (machopic_asm_out_constructor): Use finalize_ctors if needed.
15986 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
15988         PR libstdc++/54314
15989         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
15990         about visibility on artificial decls.
15991         * config/sol2.c (solaris_assemble_visibility): Likewise.
15993 2013-02-04  Kai Tietz  <ktietz@redhat.com>
15995         PR target/56186
15996         * config/i386/i386.c (function_value_ms_64): Add additional valtype
15997         argument and improve checking of return-argument types for 16-byte
15998         modes.
15999         (ix86_function_value_1): Add additional valtype argument on call
16000         of function_value_64.
16001         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16002         handling infunction_value_64 function.
16004 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16006         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16008 2013-02-04  Richard Biener  <rguenther@suse.de>
16010         PR tree-optimization/56188
16011         * tree-ssa-structalias.c (label_visit): Consider case with
16012         initially non-empty points-to set.
16013         (perform_var_substitution): Dump node mapping and clean up.
16015 2013-02-04  Richard Guenther  <rguenther@suse.de>
16017         PR lto/56168
16018         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16019         node prevail as last resort.
16020         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16021         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16023 2013-02-04  Richard Biener  <rguenther@suse.de>
16025         PR tree-optimization/56113
16026         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16027         Merge into ...
16028         (equiv_class_lookup_or_add): ... this.
16029         (label_visit): Adjust and fix error in previous patch.
16030         (perform_var_substitution): Adjust.
16032 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16034         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16035         * config/sh/sh4-300.md: Likewise.
16036         * config/sh/sh4a.md: Likewise.
16037         * config/sh/constraints.md: Likewise.
16038         * config/sh/sh.md: Likewise.
16039         * config/sh/netbsd-elf.h: Likewise.
16040         * config/sh/predicates.md: Likewise.
16041         * config/sh/sh-protos.h: Likewise.
16042         * config/sh/ushmedia.h: Likewise.
16043         * config/sh/linux.h: Likewise.
16044         * config/sh/sh.c: Likewise.
16045         * config/sh/superh.h: Likewise.
16046         * config/sh/elf.h: Likewise.
16047         * config/sh/sh4.md: Likewise.
16048         * config/sh/sh.h: Likewise.
16050 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16052         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16053         constraint to match_test floating_point_store_memory_operand().
16054         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16055         (base14_operand): New.
16056         (floating_point_store_memory_operand): New.
16057         (integer_store_memory_operand): Revise to use base14_operand and
16058         reg_plus_base_memory_operand.
16059         (move_dest_operand): Allow symbolic_memory_operands.
16060         (symbolic_memory_operand): Check for LO_SOM.
16061         (symbolic_operand): Change default case to break.
16062         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16063         CONST_DOUBLE values to be reloaded by putting them into memory when
16064         the destination is a floating point register.
16065         (movdf): Remove code to handle CONST_DOUBLE.
16066         (movsf): Likewise.
16067         (reload_indf_r1): New.
16068         (reload_insf_r1): New.
16069         Consistently use "Q" and "T" constraints with integer and floating
16070         point move instructions, respectively.
16071         (movdi): Remove FAIL.
16072         Change predicate for source operand unamed DImode move from
16073         general_operand to move_src_operand.
16074         (umulsidi3): Change predicate for destination operand to
16075         register_operand.
16076         Likewise for similar unamed patterns.
16077         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16078         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16079         (hppa_legitimize_address): Simplify mask calculation.
16080         (pa_emit_move_sequence): Revised handling of secondary reloads from
16081         REG+D addresses for floating point loads and stores.  Directly handle
16082         loading CONST0_RTX (mode) to a floating point register.
16083         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16084         to floating point registers.  Don't restrict secondary reloads to
16085         floating point registers to integer modes.  Revise some comments and
16086         cleanup some code.
16087         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16088         (pa_legitimate_address_p): New.
16089         (pa_legitimize_reload_address): New.
16090         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16091         (STRICT_REG_OK_FOR_BASE_P): New.
16092         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16093         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16095 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16096             Andrew Dixie  <andrewd@gentrack.com>
16098         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16099         flag set.
16101 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16103         * expmed.c (extract_bit_field_1): Pass the full width of the
16104         structure to get_best_reg_extraction_insn.
16106 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16108         PR target/54601
16109         * configure.ac (use_cxa_atexit): Add AIX.
16110         * configure: Regenerate.
16112         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16114 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16116         PR debug/54793
16117         * final.c (need_profile_function): New variable.
16118         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16119         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16120         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16121         notes, targetm.asm_out.function_prologue doesn't emit anything,
16122         HAVE_prologue and profiler should be emitted before prologue,
16123         set need_profile_function instead of emitting it.
16124         (final_scan_insn): If need_profile_function, emit
16125         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16126         NOTE_INSN_FUNCTION_BEG note.
16128 2013-02-01  Richard Henderson  <rth@redhat.com>
16130         * config/rs6000/rs6000.md (smulditi3): New.
16131         (umulditi3): New.
16133         * config/alpha/alpha.md (umulditi3): New.
16135 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16137         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16138         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16140 2013-02-01  Richard Biener  <rguenther@suse.de>
16142         PR tree-optimization/56113
16143         * tree-ssa-structalias.c (label_visit): Reduce work for
16144         single-predecessor nodes.
16146 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16148         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16149         range isn't testing for zero.
16151 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16153         PR middle-end/56113
16154         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16156 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16157             Nick Clifton  <nickc@redhat.com>
16159         * config/v850/constraints.md (Q): Define as a memory constraint.
16160         * config/v850/predicates.md (label_ref_operand): New predicate.
16161         (e3v5_shift_operand): New predicate.
16162         (ior_operator): New predicate.
16163         * config/v850/t-v850: Add e3v5 multilib.
16164         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16165         (v850_gen_movdi): Prototype.
16166         * config/v850/v850.c: Add support for e3v5 architecture.
16167         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16168         TARGET_V850E_UP.
16169         (construct_save_jarl): Add e3v5 long JARL support.
16170         (v850_adjust_insn_length): New function.  Adjust length of call
16171         insns when using e3v5 instructions.
16172         (v850_gen_movdi): New function: Generate instructions to move a
16173         DImode value.
16174         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16175         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16176         (TARGET_USE_FPU): Enable for e3v5.
16177         (CONST_OK_FOR_W): New macro.
16178         (ADJUST_INSN_LENGTH): Define.
16179         * config/v850/v850.md (UNSPEC_LOOP): Define.
16180         (attr cpu): Add v850e3v5.
16181         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16182         (movdi): New pattern.
16183         (movdi_internal): New pattern.
16184         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16185         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16186         (cstoresf4): Likewise.
16187         (cstoredf4): Likewise.
16188         (insv): New pattern.
16189         (rotlso3_a): New pattern.
16190         (rotlsi3_b): New pattern
16191         (rotlsi3_v850e3v5): New pattern.
16192         (doloop_begin): New pattern.
16193         (fix_loop_counter): New pattern.
16194         (doloop_end): New pattern.
16195         (branch_normal): Add e3v5 long branch support.
16196         (branch_invert): Likewise.
16197         (branch_z_normal): Likewise.
16198         (branch_z_invert): Likewise.
16199         (branch_nz_normal): Likewise.
16200         (branch_nz_invert): Likewise.
16201         (call_internal_short): Add e3v5 register-indirect JARL support.
16202         (call_internal_long): Likewise.
16203         (call_value_internal_short): Likewise.
16204         (call_value_internal_long): Likewise.
16205         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16206         (mloop): New option.
16207         * config.gcc: Add support for configuring v840e3v5 target.
16208         * doc/invoke.texi: Document new v850 specific command line options.
16210 2013-01-31  Paul Koning  <ni1d@arrl.net>
16212         PR debug/55059
16213         PR debug/54508
16214         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16215         children if parent is a class.
16216         (prune_unused_types_prune): Don't add DW_AT_declaration.
16218 2013-01-31  Richard Biener  <rguenther@suse.de>
16220         PR tree-optimization/56157
16221         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16222         match up operand with SLP child.
16224 2013-01-31  Jason Merrill  <jason@redhat.com>
16226         PR debug/54410
16227         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16228         parameters the first time.
16229         (gen_scheduled_generic_parms_dies): Check completeness here.
16231 2013-01-31  Richard Biener  <rguenther@suse.de>
16233         PR middle-end/53073
16234         * common.opt (faggressive-loop-optimizations): New flag,
16235         enabled by default.
16236         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16237         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16238         infer_loop_bounds_from_undefined by it.
16240 2013-01-31  Richard Biener  <rguenther@suse.de>
16242         PR tree-optimization/56150
16243         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16244         visit virtual operands.
16245         (find_uses_to_rename_bb): Likewise.
16247 2013-01-31  Richard Biener  <rguenther@suse.de>
16249         PR tree-optimization/56150
16250         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16251         mixed store non-store stmts.
16253 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16255         PR sanitizer/55374
16256         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16257         LIBASAN_EARLY_SPEC is defined.
16258         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16259         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16260         before %o.
16261         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16263         PR c++/55742
16264         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16265         invalid args instead of ICEing on it.
16266         (ix86_valid_target_attribute_tree): Return error_mark_node if
16267         ix86_valid_target_attribute_inner_p failed.
16268         (ix86_valid_target_attribute_p): Return false only if
16269         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16270         target("default") attribute.
16271         (sorted_attr_string): Change argument from const char * to tree,
16272         merge in all target attribute arguments rather than just one.
16273         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16274         instead of free.  Avoid using strcat.
16275         (ix86_mangle_function_version_assembler_name): Mangle
16276         target("default") as if no target attribute is present.  Adjust
16277         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16278         instead of xmalloc and XDELETEVEC instead of free.
16279         (ix86_function_versions): Don't return true if one of the decls
16280         doesn't have target attribute.  If they don't and one of the decls
16281         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16282         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16283         (ix86_supports_function_versions): Remove.
16284         (make_name): Fix up formatting.
16285         (make_dispatcher_decl): Remove resolver_name and its initialization.
16286         Avoid leaking memory.
16287         (is_function_default_version): Return true if there is
16288         target("default") attribute rather than no target attribute at all.
16289         (make_resolver_func): Avoid leaking memory.
16290         (ix86_generate_version_dispatcher_body): Likewise.
16291         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16292         * target.def (supports_function_versions): Remove.
16293         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16294         * doc/tm.texi: Regenerated.
16296 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16298         PR rtl-optimization/56144
16299         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16300         for values with side effects.
16302 2013-01-30  Richard Biener  <rguenther@suse.de>
16304         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16305         (sparseset_pop): Likewise.
16306         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16307         to be able to use quick_push in the worker loop.
16309 2013-01-30  Marek Polacek  <polacek@redhat.com>
16311         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16313 2013-01-30  Richard Biener  <rguenther@suse.de>
16315         PR lto/56147
16316         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16318 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16320         PR tree-optimization/56064
16321         * fixed-value.c (fixed_from_double_int): New function.
16322         * fixed-value.h (fixed_from_double_int): New prototype.
16323         (const_fixed_from_double_int): New static inline function.
16324         * fold-const.c (native_interpret_fixed): New static function.
16325         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16326         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16327         (native_encode_fixed): New static function.
16328         (native_encode_expr) <FIXED_CST>: Use it.
16329         (native_interpret_int): Move double_int worker code to...
16330         * double-int.c (double_int::from_buffer): ...this new static method.
16331         * double-int.h (double_int::from_buffer): Prototype it.
16333 2013-01-30  Richard Biener  <rguenther@suse.de>
16335         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16336         New pointer-map and obstack.
16337         (init_alias_vars): Allocate pointer-map and obstack.
16338         (delete_points_to_sets): Free them.
16339         (find_what_var_points_to): Cache result.
16340         (find_what_p_points_to): Adjust for changed interface of
16341         find_what_var_points_to.
16342         (compute_points_to_sets): Likewise.
16343         (ipa_pta_execute): Likewise.
16345 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16347         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16348         * configure: Regenerate.
16349         * config.in: Regenerate.
16350         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16351         #nobits/#progbits if supported.
16353 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16355         PR target/56121
16356         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16357         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16358         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16360 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16362         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16363         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16365 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16367         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16368         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16370 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16372         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16373         declaration.
16374         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16375         * config/arm/cortex-a7.md: New bypasses using
16376         arm_mac_accumulator_is_result.
16378 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16380         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16381         (cortex_a7_neon_mla): Likewise.
16382         (cortex_a7_fpfmad): New reservation.
16383         (cortex_a7_fpmacs): Use ffmas and update required units.
16384         (cortex_a7_fpmuld): Update required units and latency.
16385         (cortex_a7_fpmacd): Likewise.
16386         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16387         (cortex_a7_neon). Likewise.
16388         (bypass) Update participating units.
16390 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16392         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16393         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16394         from fmac to ffma.
16395         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16396         (vfp_fmul): Use ffmad.
16397         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16398         (cortex_r4_fmacd): Use ffmad.
16399         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16400         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16401         (cortex_a9_fmacd): Use ffmad.
16402         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16403         (cortex_a8_vfp_macd): Use ffmad.
16404         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16405         (cortex_a5_fpmacd): Use ffmad.
16406         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16407         (cortex_a15_vfp_macd): Use ffmad.
16408         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16410 2013-01-29  Jason Merrill  <jason@redhat.com>
16412         PR libstdc++/54314
16413         * varasm.c (default_assemble_visibility): Don't warn about
16414         visibility on artificial decls.
16416 2013-01-29  Richard Biener  <rguenther@suse.de>
16418         PR tree-optimization/56113
16419         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16420         the bitmap leader.
16421         (label_visit): Free duplicate bitmaps and record the leader instead.
16422         (perform_var_substitution): Adjust.
16424 2013-01-29  Richard Biener  <rguenther@suse.de>
16426         PR tree-optimization/55270
16427         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16428         the CFG, schedule loops for fixup.
16430 2013-01-29  Nick Clifton  <nickc@redhat.com>
16432         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16433         SP_REG.
16435 2013-01-28  Leif Ekblad  <leif@rdos.net>
16437         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16438         * config/i386/i386.h (TARGET_RDOS): New macro.
16439         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16440         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16441         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16442         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16443         DEFAULT_LARGE_SECTION_THRESHOLD.
16444         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16445         * config/i386/rdos.h: New file.
16446         * config/i386/rdos64.h: New file.
16448 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16450         PR other/54814
16451         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16452         TEST_HARD_REG_BIT.
16454 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16456         PR rtl-optimization/56117
16457         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16458         call cselib_lookup_from_insn on the MEM before calling
16459         add_insn_mem_dependence.
16461 2013-01-28  Richard Biener  <rguenther@suse.de>
16463         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16464         to a stmt that didn't have one.
16465         (copy_phis_for_bb): Likewise for PHI arguments.
16466         (copy_debug_stmt): Likewise for debug stmts.
16468 2013-01-28  Richard Biener  <rguenther@suse.de>
16470         PR tree-optimization/56034
16471         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16472         (partition_builtin_p): Adjust.
16473         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16474         it is the last partition.
16475         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16476         up the vertex for the definition.
16477         (classify_partition): Classify whether a partition is a
16478         PKIND_REDUCTION, thus has uses outside of the loop.
16479         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16480         Merge all PKIND_REDUCTION partitions into the last partition.
16481         (tree_loop_distribution): Seed partitions from reductions as well.
16483 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16485         PR tree-optimization/56125
16486         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16487         pow(x,c) into sqrt(x) * powi(x, n/2) or
16488         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16489         optimizing for size.
16490         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16491         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16492         integer.
16494         PR tree-optimization/56094
16495         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16496         to UNKNOWN_LOCATION while gimplifying expr.
16498 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16500         PR target/56114
16501         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16502         operand 0 in movabs insn template for -masm=intel asm alternative.
16503         (*movabs<mode>_2): Ditto for operand 1.
16505 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16507         PR target/54663
16508         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16509         of microblaze-c.o
16511 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16513         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16514         tm_file.
16516 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16518         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16519         Undef to avoid warning.
16521 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16523         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16524         * configure: Regenerate.
16526 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16528         PR tree-optimization/56098
16529         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16530         for stmts with volatile ops.
16531         (cond_store_replacement): Don't optimize if assign has volatile ops.
16532         (cond_if_else_store_replacement_1): Don't optimize if either
16533         then_assign or else_assign have volatile ops.
16534         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16535         volatile ops.
16537 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16539         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16541 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16543         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16544         missing ':' in asm example.
16546 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
16548         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16549         entries into lane and laneq entries.
16550         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16551         Remove AdvSIMD scalar modes.
16552         (aarch64_sq<r>dmulh_laneq<mode>): New.
16553         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16554         modes.
16555         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16556         builtin implementations to relfect changes in RTL in aarch64-simd.md.
16557         * config/aarch64/iterators.md (VCOND): New.
16558         (VCONQ): New.
16560 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16562         PR target/54222
16563         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16564         Add NULL LIBNAME argument to existing definitions.
16565         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16566         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16567         * config/avr/avr.c (DEF_BUILTIN): Same.
16568         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16569         (avr_expand_builtin): Expand to a vanilla call if a libgcc
16570         implementation is available (DECL_ASSEMBLER_NAME is set).
16571         (avr_fold_absfx): New static function.
16572         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16573         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16574         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16575         AVR_BUILTIN_ABSLLK.
16576         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16577         (abshk, absk, abslk, absllk): Provide as static inline functions.
16579 2013-01-25  Marek Polacek  <polacek@redhat.com>
16581         PR tree-optimization/56035
16582         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16584 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
16586         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16587         (*movtf_internal_rex64): Add (!o,C) alternative
16588         (*movxf_internal_rex64): Ditto.
16589         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16591 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
16593         * doc/invoke.texi: fix typo.
16594         * doc/objc.texi: fix typo.
16596 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
16598         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16599         for the first two alternatives.
16601 2013-01-24  Diego Novillo  <dnovillo@google.com>
16603         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
16604         (ggc-zone.o): Remove.
16605         * configure.ac: Remove option --with-gc.
16606         * configure: Re-generate.
16607         * doc/install.texi: Remove documentation for --with-gc.
16608         * gengtype.c (write_enum_defn): Remove.  Update all users.
16609         (write_Types_process_field): Remove generation of gt_e_* argument.
16610         (output_type_enum): Remove.  Update all users.
16611         (write_enum_defn): Remove.  Update all users.
16612         (enum alloc_zone): Remove.  Update all users.
16613         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16614         * ggc-common.c (ggc_splay_alloc): Remove first argument.
16615         Update all callers.
16616         (struct ptr_data): Remove field TYPE.  Update all users.
16617         (gt_pch_note_object): Remove argument TYPE.  Update all users.
16618         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16619         Update all users.
16620         * ggc-none.c (ggc_alloc_typed_stat): Remove.
16621         (struct alloc_zone): Remove.
16622         (ggc_internal_alloc_zone_stat): Remove.
16623         (ggc_internal_cleared_alloc_zone_stat): Remove.
16624         * ggc-page.c (ggc_alloc_typed_stat): Remove.
16625         (ggc_pch_count_object): Remove last argument.  Update all users.
16626         (ggc_pch_alloc_object): Remove last argument.  Update all users.
16627         (struct alloc_zone): Remove.
16628         * ggc-zone.c: Remove.
16629         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
16630         (struct alloc_zone): Remove.
16631         (ggc_alloc_typed_stat): Remove.
16632         (ggc_alloc_typed): Remove.
16633         (ggc_splay_alloc): Remove first argument.
16634         (rtl_zone): Remove.  Update all users.
16635         (tree_zone): Remove.  Update all users.
16636         (tree_id_zone): Remove.  Update all users.
16637         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
16638         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
16639         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
16640         * tree-ssanames.c: Remove references to zone allocator in comments.
16642 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
16644         * config/avr/avr.c (avr_out_fract): Make register numbers that
16645         might be outside of source operand signed.
16647 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
16649         * config/i386/constraints.md (Yf): New constraint.
16650         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16651         of f constraint to conditionaly disable x87 register preferences.
16652         (*movdf_internal): Ditto.
16653         (*movsf_internal): Ditto.
16655 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
16657         PR inline-asm/55934
16658         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16659         that have operands with impossible constraints.
16660         Add a FIXME for a speed-up opportunity.
16661         * lra-constraints.c (process_alt_operands): Verify that a class
16662         selected from constraints on asms is valid for the operand mode.
16663         (curr_insn_transform): Remove incorrect comment.
16665 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
16667         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16668         TOC operand is a valid symbol ref in the constant pool.
16670 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16672         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
16674 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
16676         PR target/54222
16677         * config/avr/stdfix.h: New file.
16678         * t-avr (stdfix-gcc.h): New rule to build it.
16679         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16681 2013-01-23  Kostya Serebryany  <kcc@google.com>
16683         * config/darwin.h: remove dependency on
16684         CoreFoundation (asan on Mac OS).
16686 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
16688         PR target/49069
16689         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16690         instead of cmpdi_operand for first comparison operand.
16691         Don't assert that comparison operands aren't both constants.
16693 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16695         * doc/install.texi (Downloading the Source): Update references to
16696         downloading separate components.
16698 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16700         * doc/extend.texi (__int128): Improve grammar.
16702 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
16704         PR target/56028
16705         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16706         alternative to (o,r).
16707         (*movdi_internal_rex64): Remove (!o,n) alternative.
16708         (DImode immediate->memory splitter): Remove.
16709         (DImode immediate->memory peephole2): Remove.
16710         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16711         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16712         alternative to (!o,*r).
16713         (*movtf_internal_sse): New pattern.
16714         (*movxf_internal_rex64): New pattern.
16715         (*movxf_internal): Disable for TARGET_64BIT.
16716         (*movdf_internal_rex64): Remove (!o,F) alternative.
16718 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
16720         PR middle-end/56074
16721         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16722         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16723         * tree-vect-loop-manip.c (find_loop_location): Also ignore
16724         stmt locations where LOCATION_LOCUS of the stmt location is
16725         UNKNOWN_LOCATION or BUILTINS_LOCATION.
16727         PR target/55686
16728         * config/i386/i386.md (UNSPEC_STOS): New.
16729         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16730         *strsetqi_1): Add UNSPEC_STOS.
16732 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
16734         PR c++/56067
16735         * doc/invoke.texi: Remove left over -Wsynth example.
16737 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16739         PR tree-optimization/56051
16740         * fold-const.c (fold_binary_loc): Don't fold
16741         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16742         a narrowing conversion, or widening conversion from signed
16743         to unsigned.
16745 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16747         PR rtl-optimization/56023
16748         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16749         dependent on debug instruction.
16751 2013-01-21  Martin Jambor  <mjambor@suse.cz>
16753         PR middle-end/56022
16754         * function.c (allocate_struct_function): Call
16755         invoke_set_current_function_hook earlier.
16757 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16759         * reload1.c (init_reload): Only initialize reload_obstack
16760         during the first call.
16762 2013-01-21  Marek Polacek  <polacek@redhat.com>
16764         * cfgloop.c (verify_loop_structure): Fix up grammar.
16766 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
16768         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16769         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16771 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16773         PR target/56058
16774         * config/arm/marvell-pj4.md: Update copyright year.
16775         Fix up use of alu to alu_reg and simple_alu_imm.
16777 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16779         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16781 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
16783         PR target/55433
16784         * lra-constraints.c (curr_insn_transform): Don't reuse original
16785         insn for secondary memory move when memory mode should be different.
16787 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16789         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16790         atomic_storedi_1): New patterns.
16792 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
16794         btver2 pipeline descriptions.
16795         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16796         descriptions.
16797         * config/i386/i386.md (btver2_decode): New type attributes.
16798         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
16799         type attributes.
16800         * config/i386/btver2.md: New file describing btver2 pipelines.
16802 2013-01-19  Andrew Pinski  <apinski@cavium.com>
16804         PR tree-optimization/52631
16805         * tree-ssa-sccvn (visit_use): Before looking up the original
16806         statement, try looking up the simplified expression.
16808 2013-01-19  Anthony Green  <green@moxielogic.com>
16810         * config/moxie/moxie.c (moxie_expand_prologue): Set
16811         current_function_static_stack_size.
16813 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16815         PR tree-optimization/56029
16816         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16817         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16819 2013-01-18  Sharad Singhai  <singhai@google.com>
16821         PR tree-optimization/55995
16822         * dumpfile.c (dump_loc): Print location only if available.
16823         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16825 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
16827         PR target/55433
16828         * lra-constraints.c (curr_insn_transform): Reuse original insn for
16829         secondary memory move.
16830         (inherit_reload_reg): Use rclass instead of cl for
16831         check_secondary_memory_needed_p.
16833 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16835         PR middle-end/56015
16836         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
16837         the case where writing real complex part of target modifies op1.
16839 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
16841         * config/aarch64/aarch64-simd.md
16842         (aarch64_vcond_internal<mode>): Handle unordered cases.
16843         * config/aarch64/iterators.md (v_cmp_result): New.
16845 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
16846             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16848         * config/arm/marvell-pj4.md: New file.
16849         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16850         * config/arm/arm.md (generic_sched): Add marvell_pj4.
16851         (generic_vfp): Likewise.
16852         * config/arm/arm-cores.def: Add marvell-pj4.
16853         * config/arm/arm-tune.md: Regenerate.
16854         * config/arm/arm-tables.opt: Regenerate.
16855         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16856         * doc/invoke.texi: Document marvell-pj4.
16858 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
16860         * config/aarch64/arm_neon.h: Map scalar types to standard types.
16862 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16864         PR debug/54114
16865         PR debug/54402
16866         PR debug/49888
16867         * var-tracking.c (negative_power_of_two_p): New.
16868         (global_get_addr_cache, local_get_addr_cache): New.
16869         (get_addr_from_global_cache, get_addr_from_local_cache): New.
16870         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
16871         heading comment.
16872         (vt_stack_offset_p): Remove.
16873         (vt_canon_true_dep): Always canonicalize loc's address.
16874         (clobber_overlapping_mems): Make sure we have a MEM.
16875         (local_get_addr_clear_given_value): New.
16876         (val_reset): Clear local cached entries.
16877         (compute_bb_dataflow): Create and release the local cache.
16878         Disable duplicate MEMs clobbering.
16879         (emit_notes_in_bb): Clobber MEMs likewise.
16880         (vt_emit_notes): Create and release the local cache.
16881         (vt_initialize, vt_finalize): Create and release the global
16882         cache, respectively.
16883         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
16885 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16887         PR libmudflap/53359
16888         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
16889         not found in the symtab.
16891 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16893         PR debug/56006
16894         PR rtl-optimization/55547
16895         PR rtl-optimization/53827
16896         PR debug/53671
16897         PR debug/49888
16898         * alias.c (offset_overlap_p): New, factored out of...
16899         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
16900         the conservative special case for symbolic constants.  Don't
16901         adjust zero sizes on alignment.
16903 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
16905         PR rtl-optimization/52573
16906         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
16907         REG_UNUSED for the same register.
16909 2013-01-17  Richard Biener  <rguenther@suse.de>
16910             Marek Polacek  <polacek@redhat.com>
16912         PR rtl-optimization/55833
16913         * loop-unswitch.c (unswitch_loops): Move loop verification...
16914         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
16915         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
16916         Set it to true when we're removing a loop from hierarchy tree in
16917         an irreducible region.
16918         (fix_bb_placements): Adjust caller.
16919         (fix_loop_placements): Likewise.
16921 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
16923         * config/avr/builtins.def (DEF_BUILTIN): Factor out
16924         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
16925         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
16926         Remove ID.  Adjust comments.
16927         * config/avr/avr-c.c (avr_builtin_name): Remove.
16928         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
16929         * config/avr/avr.c (avr_tolower): New static function.
16930         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
16931         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
16932         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
16933         default expansion.
16935 2013-01-17  Jan Hubicka  <jh@suse.cz>
16937         PR tree-optimization/55273
16938         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
16940 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
16942         PR target/55981
16943         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
16944         store through atomic_store<mode>_1.
16945         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
16947 2013-01-17  Martin Jambor  <mjambor@suse.cz>
16949         PR tree-optimizations/55264
16950         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
16951         for virtual methods.
16952         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
16953         virtual methods before inlining is over.
16954         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
16955         virtual functions.
16956         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
16957         non-virtual.
16959 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
16961         PR rtl-optimization/56005
16962         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
16963         pending reads for prefetch.
16965 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
16967         * config/aarch64/aarch64.md
16968         (*cstoresi_neg_uxtw): New pattern.
16969         (*cmovsi_insn_uxtw): New pattern.
16970         (*<optab>si3_uxtw): New pattern.
16971         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
16972         (*<optab>si3_insn_uxtw): New pattern.
16973         (*bswapsi2_uxtw): New pattern.
16975 2013-01-16  Richard Biener  <rguenther@suse.de>
16977         * tree-inline.c (tree_function_versioning): Remove set but
16978         never used variable.
16980 2013-01-16  Richard Biener  <rguenther@suse.de>
16982         PR tree-optimization/55964
16983         * tree-flow.h (rename_variables_in_loop): Remove.
16984         (rename_variables_in_bb): Likewise.
16985         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
16986         (copy_loop_before): Adjust and delete update-ssa status.
16987         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
16988         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
16989         (rename_variables_in_loop): Remove.
16990         (slpeel_update_phis_for_duplicate_loop): Likewise.
16991         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
16992         use available cfg machinery instead of duplicating it.
16993         Update PHI nodes and perform poor-mans SSA update here.
16994         (slpeel_tree_peel_loop_to_edge): Adjust.
16996 2013-01-16  Richard Biener  <rguenther@suse.de>
16998         PR tree-optimization/54767
16999         PR tree-optimization/53465
17000         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17001         (vrp_visit_phi_node): For PHI arguments coming via backedges
17002         drop all symbolical range information.
17003         (execute_vrp): Compute backedges.
17005 2013-01-16  Richard Biener  <rguenther@suse.de>
17007         * doc/install.texi: Update CLooG and ISL requirements to
17008         0.18.0 and 0.11.1.
17010 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17012         PR target/55301
17013         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17014         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17015         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17017 2013-01-16  DJ Delorie  <dj@redhat.com>
17019         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17020         (UNSPECV_SP_SWITCH_E): New.
17021         (sp_switch_1): Change to an unspec.
17022         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17023         replace $r15.
17025 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17027         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17028         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17029         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17030         (expand_mem_thread_fence): Ditto.
17031         (expand_mem_signal_fence): Ditto.
17032         (expand_atomic_load): Ditto.
17033         (expand_atomic_store): Ditto.
17035 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17037         PR rtl-optimization/55547
17038         PR rtl-optimization/53827
17039         PR debug/53671
17040         PR debug/49888
17041         * alias.c (memrefs_conflict_p): Set sizes to negative after
17042         AND adjustments.
17044 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17046         PR target/55940
17047         * function.c (thread_prologue_and_epilogue_insns): Always
17048         add crtl->drap_reg to set_up_by_prologue.set, even if
17049         stack_realign_drap is false.
17051 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17053         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17054         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17055         *call): Fix indention.
17057 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17059         PR target/55876
17060         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17061         Update comment.
17063 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17065         PR rtl-optimization/55153
17066         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17068 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17070         PR tree-optimization/55920
17071         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17072         accesses as grp_to_be_debug_replaced.
17074 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17076         PR tree-optimization/55920
17077         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17078         there is non-useless type conversion needed from debug rhs to lhs,
17079         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17081 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17082             Mikael Pettersson  <mikpe@it.uu.se>
17084         PR target/43961
17085         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17086         Thumb.
17087         (ASM_OUTPUT_CASE_LABEL): Remove.
17088         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17089         * final.c (shorten_branches): Update alignment of labels before
17090         jump tables if CASE_VECTOR_SHORTEN_MODE.
17092 2013-01-15  Richard Biener  <rguenther@suse.de>
17094         PR bootstrap/55961
17095         * system.h: Do not include gmp.h for building host tools.
17097 2013-01-15  Richard Biener  <rguenther@suse.de>
17099         PR middle-end/55882
17100         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17101         account for bitpos when computing alignment.
17103 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17105         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17106         (ix86_target_macros_internal): Likewise.
17108         * config/i386/i386.c (m_CORE2I7): Removed.
17109         (m_CORE_HASWELL): New macro.
17110         (m_CORE_ALL): Likewise.
17111         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17112         (initial_ix86_arch_features): Likewise.
17113         (processor_target_table): Initializations for Core avx2.
17114         (cpu_names): New names "core-avx2".
17115         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17116         PROCESSOR_CORE_HASWELL.
17117         (ix86_issue_rate): New case.
17118         (ia32_multipass_dfa_lookahead): Likewise.
17119         (ix86_sched_init_global): Likewise.
17121         * config/i386/i386.h (TARGET_HASWELL): New macro.
17122         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17123         (processor_type): New PROCESSOR_HASWELL.
17125 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17127         PR tree-optimization/55955
17128         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17129         *SHIFT_EXPR and *ROTATE_EXPR codes.
17131         PR tree-optimization/48766
17132         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17133         -ftrapv disable -fwrapv.
17135 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17137         PR target/55974
17138         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17139         etc. to 1 and not to __flash.
17140         Use LL suffix for __INT24_MAX__ with -mint8.
17141         Use ULL suffix for __UINT24_MAX__ with -mint8.
17143 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17145         * config/avr/avr-arch.h
17146         (struct base_arch_s): Use typedef avr_arch_t instead.
17147         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17148         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17149         * config/avr/avr.c: Same.
17150         * config/avr/avr-devices.c: Same.
17151         * config/avr/driver-avr.c: Same.
17152         * config/avr/gen-avr-mmcu-texi.c: Same.
17153         * config/avr/avr-mcus.def: Adjust comment.
17155 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17157         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17158         * config/aarch64/iterators.md (VALLDI): New.
17160 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17161             Andi Kleen  <ak@linux.intel.com>
17163         PR target/55948
17164         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17165         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17166         memmodel flag.
17168 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17170         * config/avr/avr-stdint.h: Remove trailing blanks.
17171         * config/avr/avr-log.h: Same.
17172         * config/avr/avr-arch.h: Same.
17173         * config/avr/avr-devices.c: Same.
17174         * config/avr/avr-dimode.md: Same.
17175         * config/avr/predicates.md: Same.
17176         * config/avr/avr-c.c: Same.  And fix typo.
17178         * config/avr/avr-protos.h: Same.  And:
17179         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17180         (init_cumulative_args): Rename to avr_init_cumulative_args.
17181         (expand_prologue): Rename to avr_expand_prologue.
17182         (expand_epilogue): Rename to avr_expand_epilogue.
17183         (adjust_insn_length): Rename to avr_adjust_insn_length.
17184         (notice_update_cc): Rename to avr_notice_update_cc.
17185         (final_prescan_insn): Rename to avr_final_prescan_insn.
17186         * config/avr/avr.c: Same.
17187         * config/avr/avr.h: Same.
17188         * config/avr/avr.md: Remove trailing blanks.
17189         (prologue): Use avr_expand_prologue.
17190         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17192 2013-01-14  Richard Biener  <rguenther@suse.de>
17194         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17195         verify_location, collect_subblocks): New functions.
17196         (verify_gimple_in_cfg): Verify that locations only reference
17197         BLOCKs in the functions BLOCK tree.
17199 2013-01-14  Richard Biener  <rguenther@suse.de>
17201         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17202         PHI argument.
17203         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17204         unshare reference.
17205         (insert_out_of_ssa_copy_on_edge): Likewise.
17206         (rewrite_close_phi_out_of_ssa): Likewise.
17207         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17208         debug expressions.
17209         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17210         propagated constants.
17211         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17212         can not be shared.
17214 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17216         * config/avr/avr-modes.def: Add GPL copyright notice.
17218 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17220         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17221         MEMMODEL_MASK to determine memory model.
17222         (atomic_store<mode>): Ditto from operands[2].
17223         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17225 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17227         PR fortran/55935
17228         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17229         (fold_gimple_assign): Don't call unshare_expr here.
17230         (fold_ctor_reference): Call unshare_expr.
17232 2013-01-13  Terry Guo  <terry.guo@arm.com>
17234         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17235         * doc/fragments.texi: Document MULTILIB_REUSE.
17236         * gcc.c (multilib_reuse): New internal spec.
17237         (set_multilib_dir): Also search multilib from multilib_reuse.
17238         * genmultilib (tmpmultilib3): Refactor code.
17239         (tmpmultilib4): Ditto.
17240         (multilib_reuse): New multilib argument.
17242 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17244         * Makefile.in: Update copyright.
17246 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17248         PR middle-end/55890
17249         * calls.c (expand_call): Check if arg_nr is valid.
17251 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17253         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17254         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17255         documentation.  Add missing '__' in front of
17256         __builtin_ia32_packssdw256.
17258 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17260         PR target/55719
17261         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17262         NO_REGS for larl operands.
17263         (s390_reload_larl_operand): Use s390_load_address instead of
17264         emit_move_insn.
17266 2013-01-11  Richard Biener  <rguenther@suse.de>
17268         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17269         (verify_node_sharing): ... here.
17270         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17272 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17274         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17275         Substitute TREECHECKING.
17276         * configure: Regenerate.
17277         * Makefile.in (TREECHECKING): New.
17279 2013-01-11  Richard Guenther  <rguenther@suse.de>
17281         PR tree-optimization/44061
17282         * tree-vrp.c (extract_range_basic): Compute zero as
17283         value-range for __builtin_constant_p of function parameters.
17285 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17287         Update copyright years.
17289 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17291         PR rtl-optimization/55672
17292         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17293         const to be eliminable.
17295 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17297         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17298         * configure: Regenerate.
17300 2013-01-10  Richard Biener  <rguenther@suse.de>
17302         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17304 2013-01-10  Richard Biener  <rguenther@suse.de>
17306         PR bootstrap/55792
17307         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17308         locations for virtual PHI arguments.
17309         (rewrite_update_phi_arguments): Likewise.
17311 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17313         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17314         on to assembler.
17316 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17318         PR tree-optimization/55921
17319         * tree-complex.c (expand_complex_asm): New function.
17320         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17322 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17324         PR target/55718
17325         * config/s390/s390.c (s390_symref_operand_p)
17326         (s390_loadrelative_operand_p): Merge the two functions.
17327         (s390_check_qrst_address, print_operand_address): Add parameters
17328         to s390_loadrelative_operand_p invokation.
17329         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17330         (s390_reload_larl_operand, s390_secondary_reload): Use
17331         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17332         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17334 2013-01-09  Mike Stump  <mikestump@comcast.net>
17336         * dse.c (record_store): Remove unnecessary assert.
17338 2013-01-09  Jan Hubicka  <jh@suse.cz>
17340         PR tree-optimization/55569
17341         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17342         * cfgloop.h (scale_loop_profile): Likewise.
17344 2013-01-09  Jan Hubicka  <jh@suse.cz>
17346         PR lto/45375
17347         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17348         functions.
17349         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17351 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17353         PR middle-end/55114
17354         * expr.h (maybe_emit_group_store): Declare.
17355         * expr.c (maybe_emit_group_store): New function.
17356         * builtins.c (expand_builtin_int_roundingfn): Call it.
17357         (expand_builtin_int_roundingfn_2): Likewise.
17359 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17361         PR rtl-optimization/55829
17362         * lra-constraints.c (match_reload): Add code for absent output.
17363         (curr_insn_transform): Add code for reloads of matched inputs
17364         without output.
17366 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17368         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17369         attribute of movddup insn to DF.
17370         (*vec_interleave_lowv2df): Ditto.
17371         (vec_dupv2df): Ditto.
17373 2013-01-09  Jan Hubicka  <jh@suse.cz>
17375         PR tree-optimiation/55875
17376         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17377         EVERY_ITERATION parameter.
17378         (number_of_iterations_exit): Check if exit is executed every iteration.
17379         (idx_infer_loop_bounds): Similarly here.
17380         (n_of_executions_at_most): Simplify
17381         to only test for cases where statement is dominated by the
17382         particular bound; handle correctly the "postdominance" test.
17383         (scev_probably_wraps_p): Use max loop iterations info
17384         as a global bound first.
17386 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17387             Nick Clifton  <nickc@redhat.com>
17389         * config/v850/v850.md (cbranchsf4): New pattern.
17390         (cstoresf4): New pattern.
17391         (cbranchdf4): New pattern.
17392         (cstoredf4): New pattern.
17393         (movsicc): Disallow floating point comparisons.
17394         (cmpsf_le_insn): Fix order of operators.
17395         (cmpsf_lt_insn): Likewise.
17396         (cmpsf_eq_insn): Likewise.
17397         (cmpdf_le_insn): Likewise.
17398         (cmpdf_lt_insn): Likewise.
17399         (cmpdf_eq_insn): Likewise.
17400         (cmpsf_ge_insn): Use LE comparison.
17401         (cmpdf_ge_insn): Likewise.
17402         (cmpsf_gt_insn): Use LT comparison.
17403         (cmpdf_gt_insn): Likewise.
17404         (cmpsf_ne_insn): Delete pattern.
17405         (cmpdf_ne_insn): Delete pattern.
17406         * config/v850/v850.c (v850_gen_float_compare): Use
17407         gen_cmpdf_eq_insn for NE comparison.
17408         (v850_float_z_comparison_operator)
17409         (v850_float_nz_comparison_operator): Move from here ...
17410         * config/v850/predicates.md: ... to here.  Move GT and GE
17411         comparisons into v850_float_z_comparison_operator.
17412         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17413         Delete prototype.
17414         (v850_float_nz_comparison_operator): Likewise.
17416 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17418         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17419         with calls to gen_insvsi/gen_insvdi.
17421 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17423         * config/i386/i386.c (initial_ix86_tune_features): Set up
17424         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17426 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17427             Jakub Jelinek  <jakub@redhat.com>
17429         PR tree-optimization/48189
17430         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17431         If nitercst is 0, don't predict the exit edge.
17433 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17435         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17436         in asm_fprintf with reg_names.
17437         (aarch64_print_operand_address): Likewise.
17438         (aarch64_return_addr): Likewise.
17439         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17441 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17443         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17444         (INT_U6_BITS): Likewise.
17445         * config/pa/predicates.md (uint6_operand): New predicate.
17446         (shift5_operand, shift6_operand): Likewise.
17447         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17448         arith32_operand.
17449         (lshrdi3): Use shift6_operand.
17450         (shrpsi4, shrpdi4): New insn patterns.
17451         (extzv): Delete expander.
17452         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17453         predicates in unamed zero extract patterns.  Tighten common constraint.
17454         (extv): Delete expander.
17455         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17456         predicates in unamed sign extract patterns.  Tighten common constraint.
17457         (insv): Delete expander.
17458         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17459         predicates in unamed insert patterns.  Tighten common constraint.
17460         Change uint32_operand predicate to uint6_operand predicate in unamed
17461         DImode pattern to insert constant values of type 1...1xxxx.
17463 2013-01-04  Jan Hubicka  <jh@suse.cz>
17465         PR tree-optimization/55823
17466         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17467         issue.
17469 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17470             Uros Bizjak  <ubizjak@gmail.com>
17472         PR rtl-optimization/55845
17473         * df-problems.c (can_move_insns_across): Stop scanning at
17474         volatile_insn_p source instruction or give up if
17475         across_from .. across_to range contains any volatile_insn_p
17476         instructions.
17478 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17480         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17481         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17482         Declare.
17483         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17484         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17486 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17488         PR fortran/55341
17489         * asan.c (asan_clear_shadow): New function.
17490         (asan_emit_stack_protection): Use it.
17492 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17494         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17495         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17496         with tab instead of space.
17498 2013-01-08  Nick Clifton  <nickc@redhat.com>
17500         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17501         register bank 0 at the start of an interrupt handler.
17502         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17503         MDBH registers.
17505 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17507         * config/aarch64/aarch64-simd.md
17508         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17509         (aarch64_simd_bsl): Likewise.
17510         (aarch64_vcond_internal<mode>): Likewise.
17511         (vcond<mode><mode>): Likewise.
17512         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17513         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17515 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17517         * config/aarch64/aarch64-builtins.c
17518         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17520 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17522         PR debug/55579
17523         * tree-sra.c (analyze_access_subtree): Return true also after
17524         potentially creating a debug-only replacement.
17526 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17528         PR middle-end/55890
17529         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17531         PR tree-optimization/54120
17532         * tree-vrp.c (range_fits_type_p): Don't allow
17533         src_precision < precision from signed vr to unsigned_p
17534         if vr->min or vr->max is negative.
17535         (simplify_float_conversion_using_ranges): Test can_float_p
17536         against CODE_FOR_nothing.
17538 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17539             Richard Biener  <rguenther@suse.de>
17541         PR middle-end/55851
17542         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17543         types instead of just INTEGER_TYPE types.
17545 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
17547         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17548         TF_SIZE): Define.
17550 2013-01-07  Steve Ellcey  <sellcey@mips.com>
17552         PR target/42661
17553         * config/mips/mips.opt: Change mad to mmad to match documentation.
17555 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17557         PR target/55897
17558         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17559         .progmemx.data now.
17561 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17563         PR target/55897
17564         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17565         (avr_addrspace_t): Add .section_name field.
17566         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17567         array size.
17568         (avr_addrspace): Same.  Initialize .section_name.  Remove last
17569         NULL entry.  Put __memx into .progmemx.data.
17570         (progmem_section_prefix): Remove.
17571         (avr_asm_init_sections): No need to initialize progmem_section.
17572         (avr_asm_named_section): Use avr_addrspace[].section_name to get
17573         section name prefix.
17574         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
17575         retrieve the progmem section.
17576         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17577         boundary to run over avr_addrspace[].
17578         (avr_register_target_pragmas): Ditto.
17580 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
17582         * varasm.c (output_constant_def_contents): For asan_protect_global
17583         protected strings, adjust DECL_ALIGN if needed, before testing for
17584         anchored symbols.
17585         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17586         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
17587         normal decls.
17588         (output_object_block): For asan protected decls, emit asan padding
17589         after their contents.
17590         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17591         (asan_finish_file): Test it here instead.
17593 2013-01-07  Nick Clifton  <nickc@redhat.com>
17594             Matthias Klose  <doko@debian.org>
17595             Doug Kwan  <dougkwan@google.com>
17596             H.J. Lu  <hongjiu.lu@intel.com>
17598         PR driver/55470
17599         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17601         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17603         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17605         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17607         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17609 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17611         PR target/54461
17612         * doc/install.texi (Cross-Compiler-Specific Options): Document
17613         --with-avrlibc.
17615 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
17617         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17618         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17619         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17620         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17621         vqmovun_high_s64): Fix source operand number and update copyright.
17623 2013-01-07  Richard Biener  <rguenther@suse.de>
17625         PR middle-end/55890
17626         * gimple.h (gimple_call_builtin_p): New overload.
17627         * gimple.c (validate_call): New function.
17628         (gimple_call_builtin_p): Likewise.
17629         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17630         Use gimple_call_builtin_p.
17631         (find_func_clobbers): Likewise.
17632         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17633         (strlen_optimize_stmt): Likewise.
17635 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17637         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17638         (vld1q_dup_*): Likewise.
17639         (vld1_*): Likewise.
17640         (vld1q_*): Likewise.
17641         (vld1_lane_*): Likewise.
17642         (vld1q_lane_*): Likewise.
17644 2013-01-07  Richard Biener  <rguenther@suse.de>
17646         * lto-streamer.h (LTO_minor_version): Bump to 2.
17648 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17650         * config/aarch64/aarch64-protos.h
17651         (aarch64_const_double_zero_rtx_p): Rename to...
17652         (aarch64_float_const_zero_rtx_p): ...this.
17653         (aarch64_float_const_representable_p): New.
17654         (aarch64_output_simd_mov_immediate): Likewise.
17655         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17656         move immediate case.
17657         * config/aarch64/aarch64.c
17658         (aarch64_const_double_zero_rtx_p): Rename to...
17659         (aarch64_float_const_zero_rtx_p): ...this.
17660         (aarch64_print_operand): Allow printing of new constants.
17661         (aarch64_valid_floating_const): New.
17662         (aarch64_legitimate_constant_p): Check for valid floating-point
17663         constants.
17664         (aarch64_simd_valid_immediate): Likewise.
17665         (aarch64_vect_float_const_representable_p): New.
17666         (aarch64_float_const_representable_p): Likewise.
17667         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17668         (aarch64_output_simd_mov_immediate): New.
17669         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17670         (*movdf_aarch64): Likewise.
17671         * config/aarch64/constraints.md (Ufc): New.
17672         (Y): call aarch64_float_const_zero_rtx.
17673         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17675 2013-01-07  Richard Biener  <rguenther@suse.de>
17677         PR tree-optimization/55888
17678         PR tree-optimization/55862
17679         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17680         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17681         not if it is contained therein.
17683 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17685         * config/avr/t-avr: Typo.
17687 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17689         PR55243
17690         * config/avr/t-avr: Don't automatically rebuild
17691         $(srcdir)/config/avr/t-multilib
17692         $(srcdir)/config/avr/avr-tables.opt
17693         $(srcdir)/doc/avr-mmcu.texi
17694         (avr-mcus): New phony target to build them on request.
17695         (s-avr-mlib, s-avr-mmcu-texi): Remove.
17696         * avr/avr-mcus.def: Adjust comments.
17698 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
17700         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17702 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17704         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17706 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17708         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17710 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
17712         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17713         to generate profiling.
17714         * config/rs6000/aix64.h (LIB_SPEC): Same.
17716 2013-01-04  Andrew Pinski  <apinski@cavium.com>
17718         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17719         New function.
17720         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17722 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
17724         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17725         unconditionally.
17726         (ix86_expand_move): Ditto.
17727         (ix86_zero_extend_to_Pmode): Ditto.
17728         (ix86_expand_call): Ditto.
17729         (ix86_expand_special_args_builtin): Ditto.
17730         (ix86_expand_builtin): Ditto.
17732 2013-01-04  Richard Biener  <rguenther@suse.de>
17734         PR tree-optimization/55862
17735         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17736         translating them through PHI nodes.
17738 2013-01-04  Martin Jambor  <mjambor@suse.cz>
17740         PR tree-optimization/55755
17741         * tree-sra.c (sra_modify_assign): Do not check that an access has no
17742         children when trying to avoid producing a VIEW_CONVERT_EXPR.
17744 2013-01-04  Marek Polacek  <polacek@redhat.com>
17746         PR middle-end/55859
17747         * opts.c (default_options_optimization): Clarify error message.
17749 2013-01-04  Richard Biener  <rguenther@suse.de>
17751         PR middle-end/55863
17752         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17753         reassociation.
17755 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17757         PR target/53789
17758         * config/pa/pa.md (movsi): Revert previous change.
17759         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17760         references.
17762 2013-01-03  Richard Henderson  <rth@redhat.com>
17764         * config/i386/i386.c (ix86_expand_move): Always assign to op1
17765         after eliminating TLS symbols.
17767 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
17769         PR bootstrap/50167
17770         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17771         * graphite-poly.c (debug_gmp_value): Likewise.
17773 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
17775         PR target/55712
17776         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17777         selected code model, define __code_mode_small__, __code_model_medium__,
17778         __code_model_large__, __code_model_32__ or __code_model_kernel__.
17779         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17780         xchg temporary register with %k.  Declare temporary register as
17781         early clobbered.
17782         [__x86_64__]: For medium and large code models, preserve %rbx register.
17784 2013-01-03  Richard Biener  <rguenther@suse.de>
17786         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
17787         (dump_subscript): Adjust.
17788         (finalize_ddr_dependent): Do not dump redundant info.
17789         (analyze_siv_subscript): Adjust.
17790         (subscript_dependence_tester): Likewise.
17791         (compute_affine_dependence): Likewise.
17793 2013-01-03  Richard Biener  <rguenther@suse.de>
17795         Revert
17796         2013-01-03  Richard Biener  <rguenther@suse.de>
17798         PR tree-optimization/55857
17799         * tree-vect-stmts.c (vectorizable_load): Do not setup
17800         re-alignment for invariant loads.
17802         2013-01-02  Richard Biener  <rguenther@suse.de>
17804         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17805         invariant load do not generate a vector load from the scalar location.
17807 2013-01-03  Richard Biener  <rguenther@suse.de>
17809         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17810         for not vectorizing.
17811         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17812         not build INDIRECT_REFs, call get_name once only.
17813         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
17814         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17816 2013-01-03  Richard Biener  <rguenther@suse.de>
17818         PR tree-optimization/55857
17819         * tree-vect-stmts.c (vectorizable_load): Do not setup
17820         re-alignment for invariant loads.
17822 2013-01-03  Richard Biener  <rguenther@suse.de>
17824         PR lto/55848
17825         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17826         prefer a built-in decl.
17828 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17830         * gcc.c (process_command): Update copyright notice dates.
17831         * gcov.c (print_version): Likewise.
17832         * gcov-dump.c (print_version): Likewise.
17834         PR rtl-optimization/55838
17835         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17836         iv0.step, iv1.step and step.
17838 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17839             Marc Glisse  <marc.glisse@inria.fr>
17841         PR tree-optimization/55832
17842         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17843         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17844         integer_{one,zero}_node.
17846 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17848         PR debug/54402
17849         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17850         * var-tracking.c (reverse_op): Don't add reverse ops to
17851         VALUEs that have already
17852         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
17854 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
17856         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17858 2013-01-02  Teresa Johnson  <tejohnson@google.com>
17860         * dumpfile.c (dump_loc): Print filename with location.
17861         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17862         new location_t parameter to emit complete unroll message with
17863         new dump framework.
17864         (canonicalize_loop_induction_variables): Compute loops location
17865         and pass to try_unroll_loop_completely.
17866         * loop-unroll.c (report_unroll_peel): New function.
17867         (peel_loops_completely): Use new dump format with location
17868         for main dumpfile message, and invoke report_unroll_peel on success.
17869         (decide_unrolling_and_peeling): Ditto.
17870         (decide_peel_once_rolling): Remove old dumpfile message subsumed
17871         by report_unroll_peel.
17872         (decide_peel_completely): Ditto.
17873         (decide_unroll_constant_iterations): Ditto.
17874         (decide_unroll_runtime_iterations): Ditto.
17875         (decide_peel_simple): Ditto.
17876         (decide_unroll_stupid): Ditto.
17877         * cfgloop.c (get_loop_location): New function.
17878         * cfgloop.h (get_loop_location): Declare.
17880 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17882         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
17883         NULL.
17885 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17887         PR middle-end/55198
17888         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
17889         BLKmode objects when EXPAND_MEMORY is specified.
17891 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17893         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
17894         in loop predicate.
17895         (fold_builtin_cpu): Do not share cpu model decls across statements.
17897 2013-01-02  Jason Merrill  <jason@redhat.com>
17899         PR c++/55804
17900         * tree.c (build_array_type_1): Revert earlier change.
17902 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
17904         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
17905         "cortex-a57".
17906         * config/aarch64/aarch64-tune.md: Re-generate.
17908 2013-01-02  Richard Biener  <rguenther@suse.de>
17910         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17911         invariant load do not generate a vector load from the scalar location.
17913 2013-01-02  Richard Biener  <rguenther@suse.de>
17915         PR bootstrap/55784
17916         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
17917         * configure: Regenerate.
17919 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17921         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
17922         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
17923         (expand_builtin_int_roundingfn_2): Keep the original target around
17924         for the fallback case.
17926 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17928         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
17929         to be clear for sign changes.
17931 2013-01-01  Jan Hubicka  <jh@suse.cz>
17933         * ipa-inline-analysis.c: Fix formatting.
17935 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
17937         PR tree-optimization/55831
17938         * tree-vect-loop.c (get_initial_def_for_induction): Use
17939         gsi_after_labels instead of gsi_start_bb.
17941 Copyright (C) 2013 Free Software Foundation, Inc.
17943 Copying and distribution of this file, with or without modification,
17944 are permitted in any medium without royalty provided the copyright
17945 notice and this notice are preserved.