2013-08-11 Paolo Carlini <paolo.carlini@oracle.com>
[official-gcc.git] / gcc / ChangeLog
blob728ea482cc85ebf8f927a68b4f77ba2a705370ad
1 2013-08-10  Jan Hubicka  <jh@suse.cz>
3         Workaround binutils PR14342
4         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
5         (gimple_init_edge_profiler): Likewise.
6         (gimple_gen_ic_func_profiler): Likewise.
8 2013-08-09  Jan Hubicka  <jh@suse.cz>
10         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
12 2013-08-09  Xinliang David Li  <davidxl@google.com>
14         * config/i386/stringop.def: New file.
15         * config/i386/stringop.opt: New file.
16         * config/i386/i386-opts.h: Include stringopt.def.
17         * config/i386/i386.opt: Include stringopt.opt.
18         * config/i386/i386.c (ix86_option_override_internal):
19         Override default size based stringop inline strategies
20         with options.
21         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
22         New function.
24 2013-08-09  Jan Hubicka  <jh@suse.cz>
26          * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
28 2013-08-09  Jan Hubicka  <jh@suse.cz>
30         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
31         CGRAPH_FREQ_MAX.
32         (dump_cgraph_node): Dump profile-id.
33         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
34         and common_target_probability.
35         * lto-cgraph.c (lto_output_edge): Stream common targets.
36         (lto_output_node): Stream profile ids.
37         (input_node): Stream profile ids.
38         (input_edge): Stream common targets.
39         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
40         * ipa.c: Include value-prof.h
41         (ipa_profile_generate_summary): Turn indirect call statement histograms
42         into common targets.
43         (ipa_profile): Turn common targets into speculative edges.
45 2013-08-09  Jan Hubicka  <jh@suse.cz>
47         * cgraph.h (cgraph_node): Add profile_id.
48         * value-prof.c (cgraph_node_map): Turn into pointer_map.
49         (init_node_map): Rewrite to handle hashes increas of incremental
50         IDs.
51         (del_node_map): Update.
52         (find_func_by_funcdef_no): Replace by ...
53         (find_func_by_profile_id): ... this one.
54         (gimple_ic_transform): Do not remove useful histograms when
55         speculation is not done; dump info when indirect call removal
56         can happen at LTO.
57         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
58         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
59         (__gcov_indirect_call_profiler_v2): .. this one.
60         * profile.h (init_node_map): Update.
61         * coverage.c (coverage_compute_profile_id): New function.
62         * coverage.h (coverage_compute_profile_id): Declare.
63         * tree-profile.c (init_ic_make_global_vars): Make
64         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
65         (gimple_init_edge_profiler): Update prototype of
66         __gcov_indirect_call_profiler.
67         (gimple_gen_ic_func_profiler): Simplify.
68         (tree_profiling): Use init_node_map
70 2013-08-09  Jan Hubicka  <jh@suse.cz>
72         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only non-speculative
73         refs.
74         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
75         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
76         (cgraph_set_call_stmt): Likewise.
77         (cgraph_create_edge_1): Fix release checking compilatoin;
78         clear lto_stmt_uid.
79         (cgraph_free_edge): Free indirect info.
80         (cgraph_turn_edge_to_speculative): New function.
81         (cgraph_speculative_call_info): New function.
82         (cgraph_make_edge_direct): Return direct edge; handle speculation.
83         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative
84         edges.
85         (dump_cgraph_node): Dump speculation.
86         (verify_edge_count_and_frequency): Accept speculative edges.
87         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
88         (verify_cgraph_node): Handle speculation.
89         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
90         (cgraph_set_call_stmt): Update prototype.
91         (cgraph_make_edge_direct): Update prototype.
92         (cgraph_speculative_call_info): Declare.
93         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
94         to change; update call of ipa_find_references.
95         * ipa-ref.c (ipa_record_reference): Fix return value; clear
96         lto_stmt_uid and speculative flags.
97         (ipa_dump_references): Dump speculation.
98         (ipa_clone_references): Clone speculative flag.
99         (ipa_clone_referring): Likewise.
100         (ipa_clone_ref): New function.
101         (ipa_find_reference): Look into lto_stmt_uids
102         (ipa_clear_stmts_in_references): Do not clear speculative calls.
103         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
104         (ipa_find_reference): Update declaration.
105         (ipa_clone_ref): Declare.
106         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
107         stream speculative flag.
108         (lto_output_ref): Stream statements uids and speculation.
109         (input_ref): Likewise.
110         (input_edge): Stream speuclation.
111         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
112         (cgraph_set_call_stmt_including_clones): Handle speculation.
113         * ipa-inline.c (heap_edge_removal_hook): New function.
114         (inline_small_functions): Register it.
115         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
116         also initialize refs.
117         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
118         edge to change.
119         (try_make_edge_direct_simple_call): Likewise.
120         (try_make_edge_direct_simple_call): Likewise.
121         (update_indirect_edges_after_inlining): Likewise.
122         (remove_described_reference): Look proper lto_stmt_uid.
123         (propagate_controlled_uses): Likewise.
124         (propagate_controlled_uses): Liekwise.
125         * tree-inline.c (copy_bb): Copy speculative edges.
126         (redirect_all_calls): New function.
127         (copy_cfg_body): Do redirection after loop info
128         is updated.
129         (delete_unreachable_blocks_update_callgraph): Updadte
130         speculation.
132 2013-08-09  Jan Hubicka  <jh@suse.cz>
134         * lto-streamer-out.c (output_function): Renumber PHIs.
135         * lto-streamer-in.c (input_function): Likewise.
137 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
139         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
140         (get_lane_unsigned): Likewise.
141         (dup_lane_scalar): Likewise.
142         (get_lane): enable for VALL.
143         * config/aarch64/aarch64-simd.md
144         (aarch64_dup_lane_scalar<mode>): Remove.
145         (aarch64_get_lane_signed<mode>): Likewise.
146         (aarch64_get_lane_unsigned<mode>): Likewise.
147         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
148         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
149         (aarch64_get_lane<mode>): Enable for all vector modes.
150         (aarch64_get_lanedi): Remove misleading constraints.
151         * config/aarch64/arm_neon.h
152         (__aarch64_vget_lane_any): Define.
153         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
154         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
155         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
156         * config/aarch64/iterators.md (VDQQH): New.
157         (VDQQHS): Likewise.
158         (vwcore): Likewise.
160 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
162         * configure.ac: Add GAS check for LEON instructions on SPARC.
163         * configure: Regenerate.
164         * config.in: Likewise.
165         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
166         sparc*-*-* block.
167         * config/sparc/sparc.opt (LEON, LEON3): New masks.
168         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
169         for LEON or LEON3.
170         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
171         (AS_LEON_FLAG): New macro.
172         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
173         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
174         Deal with LEON and LEON3 for the memory model.
175         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
176         (atomic_compare_and_swap<mode>_1): Likewise.
177         (*atomic_compare_and_swap<mode>_1): Likewise.
179 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
181         * config/arm/neon.md (vcond): Fix floating-point vector
182         comparisons against 0.
184 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
186         * lra-constraints.c (emit_spill_move): Remove assert.
187         (process_alt_operands): Add more debugging
188         output.  Increase reject for spilling into memory.  Decrease
189         reject for reloading scratch.
190         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
192 2013-08-08  Steve Ellcey  <sellcey@mips.com>
194         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
195         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
196         micromips incompatible.  Add nan2008.
197         (MULTILIB_DIRNAMES): Add nan2008.
198         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
199         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
200         and micromips incompatible.  Add nan2008.
201         (MULTILIB_DIRNAMES): Add nan2008.
202         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
204 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
206         PR rtl-optimization/58079
207         * combine.c (combine_simplify_rtx): Avoid using SUBST if
208         simplify_comparison has widened a comparison with an integer.
210 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
212         * config/arm/neon.md (movmisalign<mode>): Disable when we
213         don't allow unaligned accesses.
214         (*movmisalign<mode>_neon_store): Likewise.
215         (*movmisalign<mode>_neon_load): Likewise.
216         (*movmisalign<mode>_neon_store): Likewise.
217         (*movmisalign<mode>_neon_load): Likewise.
219 2013-08-08  Jan Hubicka  <jh@suse.cz>
221         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
222         (make_pass_rebuild_cgraph_edges): Also clear references.
223         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
224         * ipa-inline-transform.c (inline_transform): Remove all references
225         after inlining.
226         * cgraphunit.c (expand_function): Remove all references after expansion.
227         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
228         (ipa_find_reference): Rewrite to iterator.
229         (remove_stmt_references): Likewise.
230         (ipa_clear_stmts_in_references): New function.
231         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
232         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
233         clear references.
234         * ipa-split.c (split_function): Remove references in split function.
236 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
238         PR target/57431
239         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
240         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
242 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
244         PR target/56979
245         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
246         suggested mode for the assignment isn't compatible with the
247         registers required.
249 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
251         PR target/58065
252         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
254 2013-08-07  Xinliang David Li  <davidxl@google.com>
256         * config/i386/i386.opt: New option -mtune-ctrl=.
257         * config/i386/x86-tune.def: New file.
258         * config/i386/i386.h: include x86-tune.def.
259         * config/i386/i386.c (ix86_option_override_internal):
260         Parsing -mtune-ctrl= option and set tune features.
262 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
264         PR other/12081
265         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
266         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
267         to gen_2arg_fn_t.
269 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
271         * rtl.h (update_alignments): Declare.
272         * final.c (grow_label_align): New function extracted from...
273         (shorten_branches): ...here.  Call it.
274         (update_alignments): New function.
275         * reorg.c (sibling_labels): New variable.
276         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
277         the new label along with it onto the sibling_labels vector.
278         (fill_simple_delay_slots): Adjust call to get_label_before.
279         (fill_slots_from_thread): Likewise.
280         (relax_delay_slots): Likewise.
281         (make_return_insns): Likewise.
282         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
284 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
286         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
287         document its semantics.
288         (diagnostic_report_diagnostic): Adjust accordingly.
290 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
292         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
293         (sparc_option_override): ...and port to new C++ pass API.
294         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
296 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
298         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
300 2013-08-06  Caroline Tice  <cmtice@google.com>
302         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
303         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
304         * tree-pass.h: Add pass_vtable_verify.
305         * varasm.c (assemble_variable): Add code to properly set the comdat
306         section and name for the .vtable_map_vars section.
307         (assemble_vtyv_preinit_initializer): New function.
308         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
309         LINK_ONCE flag.
310         * output.h: Add function decl for assemble_vtv_preinit_initializer.
311         * vtable-verify.c: New file.
312         * vtable-verify.h: New file.
313         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
314         initialiation levels.
315         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
316         * passes.def: Insert pass_vtable_verify.
317         * aclocal.m4: Reorder includes.
318         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
319         -fvtv-counts options.
320         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
321         as appropriate, if -fvtable-verify=... is used.
322         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
323         -fvtable-verify=... is used.
324         * Makefile.in (OBJS):  Add vtable-verify.o to list.
325         (vtable-verify.o): Add new build rule.
326         (GTFILES): Add vtable-verify.c to list.
327         * common.opt (fvtable-verify=): New flag.
328         (vtv_priority): Values for fvtable-verify= flag.
329         (fvtv-counts): New flag.
330         (fvtv-debug): New flag.
331         * tree.h (save_vtable_map_decl): New extern function decl.
333 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
335         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
336         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
337         (pass_data_rl78_devirt): ...new pass_data instance and...
338         (make_pass_rl78_devirt): ...new function.
339         (rl78_asm_file_start): Port pass registration to new C++ API.
341 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
343         * coretypes.h (rtl_opt_pass): Add.
344         (gcc::context): Add.
345         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
346         (epiphany_init): Port to new C++ pass API.
347         (epiphany_optimize_mode_switching): Likewise.
348         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
349         (pass_manager::get_pass_mode_switching): New.
350         (pass_manager::get_pass_peephole2): New.
351         * mode-switching.c (pass_mode_switching): Add clone method.
352         * recog.c (pass_peephole2): Add clone method.
353         (pass_split_all_insns): Add clone method.
355 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
357         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
358         (mips_option_override): ...here, porting to new C++ API for passes.
360 2013-08-06  Jan Hubicka  <jh@suse.cz>
362         * cgraph.c (cgraph_get_body): New function based on lto.c
363         implementation.
364         * cgraph.h (cgraph_get_body): Declare.
365         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
366         LTO paths.
367         * cgraphunit.c (expand_function): Get body prior expanding.
368         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
369         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
370         really need.
371         * passes.c (do_per_function_toporder): Get body.
372         * tree-inline.c (expand_call_inline): Get body prior inlining it.
373         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
375 2013-08-06  Martin Jambor  <mjambor@suse.cz>
377         PR fortran/57987
378         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
379         is not re-finalized.  Rename second parameter to no_collect.
381 2013-08-06  Martin Jambor  <mjambor@suse.cz>
383         PR middle-end/58041
384         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
385         MEM_REF has proper alignment information.
387 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
389         PR other/12081
390         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
391         class insn_gen_fn.
392         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
393         rtx (*) (rtx, ...) with insn_gen_fn.
394         * genoutput.c (output_insn_data): Cast gen_? function pointers to
395         insn_gen_fn::stored_funcptr.  Add initializer braces.
397 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
399         Rewrite how instances of passes are cloned to remove assumptions
400         about their sizes (thus allowing pass subclasses to have
401         additional data fields, albeit non-GC-managed ones at this point).
403         * passes.c (make_pass_instance): Now that passes have clone
404         methods, rewrite this function to eliminate XNEW and memcpy
405         calls that used hardcoded sizes.  Since this function no longer
406         creates pass instances, rename it to...
407         (add_pass_instance): ...this.  Document the old way that passes were
408         numbered and flagged, and rework this function to continue using it.
409         (next_pass_1): Add an initial_pass argument for use by
410         add_pass_instance.
411         (position_pass): When adding multiple instances of a pass, use
412         the pass's clone method, rather than relying on the XNEW/memcpy
413         within the former make_pass_instance (now add_pass_instance).
414         (pass_manager::pass_manager): When invoking next_pass_1, also supply
415         the initial instance of the current pass within the pass manager.
417 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
419         This is the automated part of the conversion of passes from C
420         structs to C++ classes.
422         Patch autogenerated by refactor_passes.py from
423         https://github.com/davidmalcolm/gcc-refactoring-scripts
424         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
426         * asan.c (pass_asan): Convert from a global struct to a subclass of
427         gimple_opt_pass along with...
428         (pass_data_asan): ...new pass_data instance and...
429         (make_pass_asan): ...new function.
430         (pass_asan_O0): Convert from a global struct to a subclass of
431         gimple_opt_pass along with...
432         (pass_data_asan_O0): ...new pass_data instance and...
433         (make_pass_asan_O0): ...new function.
434         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
435         subclass of rtl_opt_pass along with...
436         (pass_data_inc_dec): ...new pass_data instance and...
437         (make_pass_inc_dec): ...new function.
438         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
439         a subclass of rtl_opt_pass along with...
440         (pass_data_reorder_blocks): ...new pass_data instance and...
441         (make_pass_reorder_blocks): ...new function.
442         (pass_duplicate_computed_gotos): Convert from a global struct to a
443         subclass of rtl_opt_pass along with...
444         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
445         (make_pass_duplicate_computed_gotos): ...new function.
446         (pass_partition_blocks): Convert from a global struct to a subclass of
447         rtl_opt_pass along with...
448         (pass_data_partition_blocks): ...new pass_data instance and...
449         (make_pass_partition_blocks): ...new function.
450         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
451         struct to a subclass of rtl_opt_pass along with...
452         (pass_data_branch_target_load_optimize1): ...new pass_data instance
453         and...
454         (make_pass_branch_target_load_optimize1): ...new function.
455         (pass_branch_target_load_optimize2): Convert from a global struct to a
456         subclass of rtl_opt_pass along with...
457         (pass_data_branch_target_load_optimize2): ...new pass_data instance
458         and...
459         (make_pass_branch_target_load_optimize2): ...new function.
460         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
461         of rtl_opt_pass along with...
462         (pass_data_jump): ...new pass_data instance and...
463         (make_pass_jump): ...new function.
464         (pass_jump2): Convert from a global struct to a subclass of
465         rtl_opt_pass along with...
466         (pass_data_jump2): ...new pass_data instance and...
467         (make_pass_jump2): ...new function.
468         * cfgexpand.c (pass_expand): Convert from a global struct to a
469         subclass of rtl_opt_pass along with...
470         (pass_data_expand): ...new pass_data instance and...
471         (make_pass_expand): ...new function.
472         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
473         of rtl_opt_pass along with...
474         (pass_data_free_cfg): ...new pass_data instance and...
475         (make_pass_free_cfg): ...new function.
476         (pass_into_cfg_layout_mode): Convert from a global struct to a
477         subclass of rtl_opt_pass along with...
478         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
479         (make_pass_into_cfg_layout_mode): ...new function.
480         (pass_outof_cfg_layout_mode): Convert from a global struct to a
481         subclass of rtl_opt_pass along with...
482         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
483         (make_pass_outof_cfg_layout_mode): ...new function.
484         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
485         struct to a subclass of gimple_opt_pass along with...
486         (pass_data_build_cgraph_edges): ...new pass_data instance and...
487         (make_pass_build_cgraph_edges): ...new function.
488         (pass_rebuild_cgraph_edges): Convert from a global struct to a
489         subclass of gimple_opt_pass along with...
490         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
491         (make_pass_rebuild_cgraph_edges): ...new function.
492         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
493         subclass of gimple_opt_pass along with...
494         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
495         and...
496         (make_pass_remove_cgraph_callee_edges): ...new function.
497         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
498         struct to a subclass of rtl_opt_pass along with...
499         (pass_data_stack_adjustments): ...new pass_data instance and...
500         (make_pass_stack_adjustments): ...new function.
501         * combine.c (pass_combine): Convert from a global struct to a subclass
502         of rtl_opt_pass along with...
503         (pass_data_combine): ...new pass_data instance and...
504         (make_pass_combine): ...new function.
505         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
506         global struct to a subclass of rtl_opt_pass along with...
507         (pass_data_compare_elim_after_reload): ...new pass_data instance
508         and...
509         (make_pass_compare_elim_after_reload): ...new function.
510         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
511         of rtl_opt_pass along with...
512         (pass_data_rtl_cprop): ...new pass_data instance and...
513         (make_pass_rtl_cprop): ...new function.
514         * cse.c (pass_cse): Convert from a global struct to a subclass of
515         rtl_opt_pass along with...
516         (pass_data_cse): ...new pass_data instance and...
517         (make_pass_cse): ...new function.
518         (pass_cse2): Convert from a global struct to a subclass of
519         rtl_opt_pass along with...
520         (pass_data_cse2): ...new pass_data instance and...
521         (make_pass_cse2): ...new function.
522         (pass_cse_after_global_opts): Convert from a global struct to a
523         subclass of rtl_opt_pass along with...
524         (pass_data_cse_after_global_opts): ...new pass_data instance and...
525         (make_pass_cse_after_global_opts): ...new function.
526         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
527         of rtl_opt_pass along with...
528         (pass_data_ud_rtl_dce): ...new pass_data instance and...
529         (make_pass_ud_rtl_dce): ...new function.
530         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
531         rtl_opt_pass along with...
532         (pass_data_fast_rtl_dce): ...new pass_data instance and...
533         (make_pass_fast_rtl_dce): ...new function.
534         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
535         a subclass of rtl_opt_pass along with...
536         (pass_data_df_initialize_opt): ...new pass_data instance and...
537         (make_pass_df_initialize_opt): ...new function.
538         (pass_df_initialize_no_opt): Convert from a global struct to a
539         subclass of rtl_opt_pass along with...
540         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
541         (make_pass_df_initialize_no_opt): ...new function.
542         (pass_df_finish): Convert from a global struct to a subclass of
543         rtl_opt_pass along with...
544         (pass_data_df_finish): ...new pass_data instance and...
545         (make_pass_df_finish): ...new function.
546         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
547         rtl_opt_pass along with...
548         (pass_data_rtl_dse1): ...new pass_data instance and...
549         (make_pass_rtl_dse1): ...new function.
550         (pass_rtl_dse2): Convert from a global struct to a subclass of
551         rtl_opt_pass along with...
552         (pass_data_rtl_dse2): ...new pass_data instance and...
553         (make_pass_rtl_dse2): ...new function.
554         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
555         subclass of rtl_opt_pass along with...
556         (pass_data_dwarf2_frame): ...new pass_data instance and...
557         (make_pass_dwarf2_frame): ...new function.
558         * except.c (pass_set_nothrow_function_flags): Convert from a global
559         struct to a subclass of rtl_opt_pass along with...
560         (pass_data_set_nothrow_function_flags): ...new pass_data instance
561         and...
562         (make_pass_set_nothrow_function_flags): ...new function.
563         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
564         subclass of rtl_opt_pass along with...
565         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
566         and...
567         (make_pass_convert_to_eh_region_ranges): ...new function.
568         * final.c (pass_compute_alignments): Convert from a global struct to a
569         subclass of rtl_opt_pass along with...
570         (pass_data_compute_alignments): ...new pass_data instance and...
571         (make_pass_compute_alignments): ...new function.
572         (pass_final): Convert from a global struct to a subclass of
573         rtl_opt_pass along with...
574         (pass_data_final): ...new pass_data instance and...
575         (make_pass_final): ...new function.
576         (pass_shorten_branches): Convert from a global struct to a subclass of
577         rtl_opt_pass along with...
578         (pass_data_shorten_branches): ...new pass_data instance and...
579         (make_pass_shorten_branches): ...new function.
580         (pass_clean_state): Convert from a global struct to a subclass of
581         rtl_opt_pass along with...
582         (pass_data_clean_state): ...new pass_data instance and...
583         (make_pass_clean_state): ...new function.
584         * function.c (pass_instantiate_virtual_regs): Convert from a global
585         struct to a subclass of rtl_opt_pass along with...
586         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
587         (make_pass_instantiate_virtual_regs): ...new function.
588         (pass_leaf_regs): Convert from a global struct to a subclass of
589         rtl_opt_pass along with...
590         (pass_data_leaf_regs): ...new pass_data instance and...
591         (make_pass_leaf_regs): ...new function.
592         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
593         subclass of rtl_opt_pass along with...
594         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
595         and...
596         (make_pass_thread_prologue_and_epilogue): ...new function.
597         (pass_match_asm_constraints): Convert from a global struct to a
598         subclass of rtl_opt_pass along with...
599         (pass_data_match_asm_constraints): ...new pass_data instance and...
600         (make_pass_match_asm_constraints): ...new function.
601         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
602         subclass of rtl_opt_pass along with...
603         (pass_data_rtl_fwprop): ...new pass_data instance and...
604         (make_pass_rtl_fwprop): ...new function.
605         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
606         rtl_opt_pass along with...
607         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
608         (make_pass_rtl_fwprop_addr): ...new function.
609         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
610         rtl_opt_pass along with...
611         (pass_data_rtl_pre): ...new pass_data instance and...
612         (make_pass_rtl_pre): ...new function.
613         (pass_rtl_hoist): Convert from a global struct to a subclass of
614         rtl_opt_pass along with...
615         (pass_data_rtl_hoist): ...new pass_data instance and...
616         (make_pass_rtl_hoist): ...new function.
617         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
618         subclass of gimple_opt_pass along with...
619         (pass_data_lower_cf): ...new pass_data instance and...
620         (make_pass_lower_cf): ...new function.
621         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
622         from a global struct to a subclass of gimple_opt_pass along with...
623         (pass_data_strength_reduction): ...new pass_data instance and...
624         (make_pass_strength_reduction): ...new function.
625         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
626         of rtl_opt_pass along with...
627         (pass_data_rtl_ifcvt): ...new pass_data instance and...
628         (make_pass_rtl_ifcvt): ...new function.
629         (pass_if_after_combine): Convert from a global struct to a subclass of
630         rtl_opt_pass along with...
631         (pass_data_if_after_combine): ...new pass_data instance and...
632         (make_pass_if_after_combine): ...new function.
633         (pass_if_after_reload): Convert from a global struct to a subclass of
634         rtl_opt_pass along with...
635         (pass_data_if_after_reload): ...new pass_data instance and...
636         (make_pass_if_after_reload): ...new function.
637         * init-regs.c (pass_initialize_regs): Convert from a global struct to
638         a subclass of rtl_opt_pass along with...
639         (pass_data_initialize_regs): ...new pass_data instance and...
640         (make_pass_initialize_regs): ...new function.
641         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
642         of ipa_opt_pass_d along with...
643         (pass_data_ipa_cp): ...new pass_data instance and...
644         (make_pass_ipa_cp): ...new function.
645         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
646         global struct to a subclass of gimple_opt_pass along with...
647         (pass_data_inline_parameters): ...new pass_data instance and...
648         (make_pass_inline_parameters): ...new function.
649         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
650         subclass of gimple_opt_pass along with...
651         (pass_data_early_inline): ...new pass_data instance and...
652         (make_pass_early_inline): ...new function.
653         (pass_ipa_inline): Convert from a global struct to a subclass of
654         ipa_opt_pass_d along with...
655         (pass_data_ipa_inline): ...new pass_data instance and...
656         (make_pass_ipa_inline): ...new function.
657         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
658         struct to a subclass of gimple_opt_pass along with...
659         (pass_data_local_pure_const): ...new pass_data instance and...
660         (make_pass_local_pure_const): ...new function.
661         (pass_ipa_pure_const): Convert from a global struct to a subclass of
662         ipa_opt_pass_d along with...
663         (pass_data_ipa_pure_const): ...new pass_data instance and...
664         (make_pass_ipa_pure_const): ...new function.
665         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
666         to a subclass of ipa_opt_pass_d along with...
667         (pass_data_ipa_reference): ...new pass_data instance and...
668         (make_pass_ipa_reference): ...new function.
669         * ipa-split.c (pass_split_functions): Convert from a global struct to
670         a subclass of gimple_opt_pass along with...
671         (pass_data_split_functions): ...new pass_data instance and...
672         (make_pass_split_functions): ...new function.
673         (pass_feedback_split_functions): Convert from a global struct to a
674         subclass of gimple_opt_pass along with...
675         (pass_data_feedback_split_functions): ...new pass_data instance and...
676         (make_pass_feedback_split_functions): ...new function.
677         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
678         global struct to a subclass of simple_ipa_opt_pass along with...
679         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
680         instance and...
681         (make_pass_ipa_function_and_variable_visibility): ...new function.
682         (pass_ipa_free_inline_summary): Convert from a global struct to a
683         subclass of simple_ipa_opt_pass along with...
684         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
685         (make_pass_ipa_free_inline_summary): ...new function.
686         (pass_ipa_whole_program_visibility): Convert from a global struct to a
687         subclass of ipa_opt_pass_d along with...
688         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
689         and...
690         (make_pass_ipa_whole_program_visibility): ...new function.
691         (pass_ipa_profile): Convert from a global struct to a subclass of
692         ipa_opt_pass_d along with...
693         (pass_data_ipa_profile): ...new pass_data instance and...
694         (make_pass_ipa_profile): ...new function.
695         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
696         ipa_opt_pass_d along with...
697         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
698         (make_pass_ipa_cdtor_merge): ...new function.
699         * ira.c (pass_ira): Convert from a global struct to a subclass of
700         rtl_opt_pass along with...
701         (pass_data_ira): ...new pass_data instance and...
702         (make_pass_ira): ...new function.
703         (pass_reload): Convert from a global struct to a subclass of
704         rtl_opt_pass along with...
705         (pass_data_reload): ...new pass_data instance and...
706         (make_pass_reload): ...new function.
707         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
708         subclass of rtl_opt_pass along with...
709         (pass_data_cleanup_barriers): ...new pass_data instance and...
710         (make_pass_cleanup_barriers): ...new function.
711         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
712         of rtl_opt_pass along with...
713         (pass_data_loop2): ...new pass_data instance and...
714         (make_pass_loop2): ...new function.
715         (pass_rtl_loop_init): Convert from a global struct to a subclass of
716         rtl_opt_pass along with...
717         (pass_data_rtl_loop_init): ...new pass_data instance and...
718         (make_pass_rtl_loop_init): ...new function.
719         (pass_rtl_loop_done): Convert from a global struct to a subclass of
720         rtl_opt_pass along with...
721         (pass_data_rtl_loop_done): ...new pass_data instance and...
722         (make_pass_rtl_loop_done): ...new function.
723         (pass_rtl_move_loop_invariants): Convert from a global struct to a
724         subclass of rtl_opt_pass along with...
725         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
726         (make_pass_rtl_move_loop_invariants): ...new function.
727         (pass_rtl_unswitch): Convert from a global struct to a subclass of
728         rtl_opt_pass along with...
729         (pass_data_rtl_unswitch): ...new pass_data instance and...
730         (make_pass_rtl_unswitch): ...new function.
731         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
732         subclass of rtl_opt_pass along with...
733         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
734         and...
735         (make_pass_rtl_unroll_and_peel_loops): ...new function.
736         (pass_rtl_doloop): Convert from a global struct to a subclass of
737         rtl_opt_pass along with...
738         (pass_data_rtl_doloop): ...new pass_data instance and...
739         (make_pass_rtl_doloop): ...new function.
740         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
741         a subclass of rtl_opt_pass along with...
742         (pass_data_lower_subreg): ...new pass_data instance and...
743         (make_pass_lower_subreg): ...new function.
744         (pass_lower_subreg2): Convert from a global struct to a subclass of
745         rtl_opt_pass along with...
746         (pass_data_lower_subreg2): ...new pass_data instance and...
747         (make_pass_lower_subreg2): ...new function.
748         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
749         struct to a subclass of ipa_opt_pass_d along with...
750         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
751         (make_pass_ipa_lto_gimple_out): ...new function.
752         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
753         of ipa_opt_pass_d along with...
754         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
755         (make_pass_ipa_lto_finish_out): ...new function.
756         * mode-switching.c (pass_mode_switching): Convert from a global struct
757         to a subclass of rtl_opt_pass along with...
758         (pass_data_mode_switching): ...new pass_data instance and...
759         (make_pass_mode_switching): ...new function.
760         * modulo-sched.c (pass_sms): Convert from a global struct to a
761         subclass of rtl_opt_pass along with...
762         (pass_data_sms): ...new pass_data instance and...
763         (make_pass_sms): ...new function.
764         * omp-low.c (pass_expand_omp): Convert from a global struct to a
765         subclass of gimple_opt_pass along with...
766         (pass_data_expand_omp): ...new pass_data instance and...
767         (make_pass_expand_omp): ...new function.
768         (pass_lower_omp): Convert from a global struct to a subclass of
769         gimple_opt_pass along with...
770         (pass_data_lower_omp): ...new pass_data instance and...
771         (make_pass_lower_omp): ...new function.
772         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
773         of gimple_opt_pass along with...
774         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
775         (make_pass_diagnose_omp_blocks): ...new function.
776         * passes.c (pass_early_local_passes): Convert from a global struct to
777         a subclass of simple_ipa_opt_pass along with...
778         (pass_data_early_local_passes): ...new pass_data instance and...
779         (make_pass_early_local_passes): ...new function.
780         (pass_all_early_optimizations): Convert from a global struct to a
781         subclass of gimple_opt_pass along with...
782         (pass_data_all_early_optimizations): ...new pass_data instance and...
783         (make_pass_all_early_optimizations): ...new function.
784         (pass_all_optimizations): Convert from a global struct to a subclass
785         of gimple_opt_pass along with...
786         (pass_data_all_optimizations): ...new pass_data instance and...
787         (make_pass_all_optimizations): ...new function.
788         (pass_all_optimizations_g): Convert from a global struct to a subclass
789         of gimple_opt_pass along with...
790         (pass_data_all_optimizations_g): ...new pass_data instance and...
791         (make_pass_all_optimizations_g): ...new function.
792         (pass_rest_of_compilation): Convert from a global struct to a subclass
793         of rtl_opt_pass along with...
794         (pass_data_rest_of_compilation): ...new pass_data instance and...
795         (make_pass_rest_of_compilation): ...new function.
796         (pass_postreload): Convert from a global struct to a subclass of
797         rtl_opt_pass along with...
798         (pass_data_postreload): ...new pass_data instance and...
799         (make_pass_postreload): ...new function.
800         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
801         subclass of rtl_opt_pass along with...
802         (pass_data_gcse2): ...new pass_data instance and...
803         (make_pass_gcse2): ...new function.
804         * postreload.c (pass_postreload_cse): Convert from a global struct to
805         a subclass of rtl_opt_pass along with...
806         (pass_data_postreload_cse): ...new pass_data instance and...
807         (make_pass_postreload_cse): ...new function.
808         * predict.c (pass_profile): Convert from a global struct to a subclass
809         of gimple_opt_pass along with...
810         (pass_data_profile): ...new pass_data instance and...
811         (make_pass_profile): ...new function.
812         (pass_strip_predict_hints): Convert from a global struct to a subclass
813         of gimple_opt_pass along with...
814         (pass_data_strip_predict_hints): ...new pass_data instance and...
815         (make_pass_strip_predict_hints): ...new function.
816         * recog.c (pass_peephole2): Convert from a global struct to a subclass
817         of rtl_opt_pass along with...
818         (pass_data_peephole2): ...new pass_data instance and...
819         (make_pass_peephole2): ...new function.
820         (pass_split_all_insns): Convert from a global struct to a subclass of
821         rtl_opt_pass along with...
822         (pass_data_split_all_insns): ...new pass_data instance and...
823         (make_pass_split_all_insns): ...new function.
824         (pass_split_after_reload): Convert from a global struct to a subclass
825         of rtl_opt_pass along with...
826         (pass_data_split_after_reload): ...new pass_data instance and...
827         (make_pass_split_after_reload): ...new function.
828         (pass_split_before_regstack): Convert from a global struct to a
829         subclass of rtl_opt_pass along with...
830         (pass_data_split_before_regstack): ...new pass_data instance and...
831         (make_pass_split_before_regstack): ...new function.
832         (pass_split_before_sched2): Convert from a global struct to a subclass
833         of rtl_opt_pass along with...
834         (pass_data_split_before_sched2): ...new pass_data instance and...
835         (make_pass_split_before_sched2): ...new function.
836         (pass_split_for_shorten_branches): Convert from a global struct to a
837         subclass of rtl_opt_pass along with...
838         (pass_data_split_for_shorten_branches): ...new pass_data instance
839         and...
840         (make_pass_split_for_shorten_branches): ...new function.
841         * ree.c (pass_ree): Convert from a global struct to a subclass of
842         rtl_opt_pass along with...
843         (pass_data_ree): ...new pass_data instance and...
844         (make_pass_ree): ...new function.
845         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
846         subclass of rtl_opt_pass along with...
847         (pass_data_stack_regs): ...new pass_data instance and...
848         (make_pass_stack_regs): ...new function.
849         (pass_stack_regs_run): Convert from a global struct to a subclass of
850         rtl_opt_pass along with...
851         (pass_data_stack_regs_run): ...new pass_data instance and...
852         (make_pass_stack_regs_run): ...new function.
853         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
854         subclass of rtl_opt_pass along with...
855         (pass_data_cprop_hardreg): ...new pass_data instance and...
856         (make_pass_cprop_hardreg): ...new function.
857         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
858         subclass of rtl_opt_pass along with...
859         (pass_data_reginfo_init): ...new pass_data instance and...
860         (make_pass_reginfo_init): ...new function.
861         * regmove.c (pass_regmove): Convert from a global struct to a subclass
862         of rtl_opt_pass along with...
863         (pass_data_regmove): ...new pass_data instance and...
864         (make_pass_regmove): ...new function.
865         * regrename.c (pass_regrename): Convert from a global struct to a
866         subclass of rtl_opt_pass along with...
867         (pass_data_regrename): ...new pass_data instance and...
868         (make_pass_regrename): ...new function.
869         * reorg.c (pass_delay_slots): Convert from a global struct to a
870         subclass of rtl_opt_pass along with...
871         (pass_data_delay_slots): ...new pass_data instance and...
872         (make_pass_delay_slots): ...new function.
873         (pass_machine_reorg): Convert from a global struct to a subclass of
874         rtl_opt_pass along with...
875         (pass_data_machine_reorg): ...new pass_data instance and...
876         (make_pass_machine_reorg): ...new function.
877         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
878         of rtl_opt_pass along with...
879         (pass_data_sched): ...new pass_data instance and...
880         (make_pass_sched): ...new function.
881         (pass_sched2): Convert from a global struct to a subclass of
882         rtl_opt_pass along with...
883         (pass_data_sched2): ...new pass_data instance and...
884         (make_pass_sched2): ...new function.
885         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
886         to a subclass of rtl_opt_pass along with...
887         (pass_data_stack_ptr_mod): ...new pass_data instance and...
888         (make_pass_stack_ptr_mod): ...new function.
889         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
890         to a subclass of rtl_opt_pass along with...
891         (pass_data_rtl_store_motion): ...new pass_data instance and...
892         (make_pass_rtl_store_motion): ...new function.
893         * tracer.c (pass_tracer): Convert from a global struct to a subclass
894         of gimple_opt_pass along with...
895         (pass_data_tracer): ...new pass_data instance and...
896         (make_pass_tracer): ...new function.
897         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
898         to a subclass of gimple_opt_pass along with...
899         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
900         (make_pass_diagnose_tm_blocks): ...new function.
901         (pass_lower_tm): Convert from a global struct to a subclass of
902         gimple_opt_pass along with...
903         (pass_data_lower_tm): ...new pass_data instance and...
904         (make_pass_lower_tm): ...new function.
905         (pass_tm_init): Convert from a global struct to a subclass of
906         gimple_opt_pass along with...
907         (pass_data_tm_init): ...new pass_data instance and...
908         (make_pass_tm_init): ...new function.
909         (pass_tm_mark): Convert from a global struct to a subclass of
910         gimple_opt_pass along with...
911         (pass_data_tm_mark): ...new pass_data instance and...
912         (make_pass_tm_mark): ...new function.
913         (pass_tm_edges): Convert from a global struct to a subclass of
914         gimple_opt_pass along with...
915         (pass_data_tm_edges): ...new pass_data instance and...
916         (make_pass_tm_edges): ...new function.
917         (pass_tm_memopt): Convert from a global struct to a subclass of
918         gimple_opt_pass along with...
919         (pass_data_tm_memopt): ...new pass_data instance and...
920         (make_pass_tm_memopt): ...new function.
921         (pass_ipa_tm): Convert from a global struct to a subclass of
922         simple_ipa_opt_pass along with...
923         (pass_data_ipa_tm): ...new pass_data instance and...
924         (make_pass_ipa_tm): ...new function.
925         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
926         subclass of gimple_opt_pass along with...
927         (pass_data_call_cdce): ...new pass_data instance and...
928         (make_pass_call_cdce): ...new function.
929         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
930         subclass of gimple_opt_pass along with...
931         (pass_data_build_cfg): ...new pass_data instance and...
932         (make_pass_build_cfg): ...new function.
933         (pass_split_crit_edges): Convert from a global struct to a subclass of
934         gimple_opt_pass along with...
935         (pass_data_split_crit_edges): ...new pass_data instance and...
936         (make_pass_split_crit_edges): ...new function.
937         (pass_warn_function_return): Convert from a global struct to a
938         subclass of gimple_opt_pass along with...
939         (pass_data_warn_function_return): ...new pass_data instance and...
940         (make_pass_warn_function_return): ...new function.
941         (pass_warn_function_noreturn): Convert from a global struct to a
942         subclass of gimple_opt_pass along with...
943         (pass_data_warn_function_noreturn): ...new pass_data instance and...
944         (make_pass_warn_function_noreturn): ...new function.
945         (pass_warn_unused_result): Convert from a global struct to a subclass
946         of gimple_opt_pass along with...
947         (pass_data_warn_unused_result): ...new pass_data instance and...
948         (make_pass_warn_unused_result): ...new function.
949         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
950         a subclass of gimple_opt_pass along with...
951         (pass_data_merge_phi): ...new pass_data instance and...
952         (make_pass_merge_phi): ...new function.
953         * tree-complex.c (pass_lower_complex): Convert from a global struct to
954         a subclass of gimple_opt_pass along with...
955         (pass_data_lower_complex): ...new pass_data instance and...
956         (make_pass_lower_complex): ...new function.
957         (pass_lower_complex_O0): Convert from a global struct to a subclass of
958         gimple_opt_pass along with...
959         (pass_data_lower_complex_O0): ...new pass_data instance and...
960         (make_pass_lower_complex_O0): ...new function.
961         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
962         subclass of gimple_opt_pass along with...
963         (pass_data_lower_eh): ...new pass_data instance and...
964         (make_pass_lower_eh): ...new function.
965         (pass_refactor_eh): Convert from a global struct to a subclass of
966         gimple_opt_pass along with...
967         (pass_data_refactor_eh): ...new pass_data instance and...
968         (make_pass_refactor_eh): ...new function.
969         (pass_lower_resx): Convert from a global struct to a subclass of
970         gimple_opt_pass along with...
971         (pass_data_lower_resx): ...new pass_data instance and...
972         (make_pass_lower_resx): ...new function.
973         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
974         of gimple_opt_pass along with...
975         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
976         (make_pass_lower_eh_dispatch): ...new function.
977         (pass_cleanup_eh): Convert from a global struct to a subclass of
978         gimple_opt_pass along with...
979         (pass_data_cleanup_eh): ...new pass_data instance and...
980         (make_pass_cleanup_eh): ...new function.
981         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
982         to a subclass of simple_ipa_opt_pass along with...
983         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
984         (make_pass_ipa_lower_emutls): ...new function.
985         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
986         a subclass of gimple_opt_pass along with...
987         (pass_data_if_conversion): ...new pass_data instance and...
988         (make_pass_if_conversion): ...new function.
989         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
990         subclass of gimple_opt_pass along with...
991         (pass_data_build_ssa): ...new pass_data instance and...
992         (make_pass_build_ssa): ...new function.
993         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
994         global struct to a subclass of gimple_opt_pass along with...
995         (pass_data_loop_distribution): ...new pass_data instance and...
996         (make_pass_loop_distribution): ...new function.
997         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
998         subclass of gimple_opt_pass along with...
999         (pass_data_mudflap_1): ...new pass_data instance and...
1000         (make_pass_mudflap_1): ...new function.
1001         (pass_mudflap_2): Convert from a global struct to a subclass of
1002         gimple_opt_pass along with...
1003         (pass_data_mudflap_2): ...new pass_data instance and...
1004         (make_pass_mudflap_2): ...new function.
1005         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1006         subclass of gimple_opt_pass along with...
1007         (pass_data_mudflap_1): ...new pass_data instance and...
1008         (make_pass_mudflap_1): ...new function.
1009         (pass_mudflap_2): Convert from a global struct to a subclass of
1010         gimple_opt_pass along with...
1011         (pass_data_mudflap_2): ...new pass_data instance and...
1012         (make_pass_mudflap_2): ...new function.
1013         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1014         gimple_opt_pass along with...
1015         (pass_data_nrv): ...new pass_data instance and...
1016         (make_pass_nrv): ...new function.
1017         (pass_return_slot): Convert from a global struct to a subclass of
1018         gimple_opt_pass along with...
1019         (pass_data_return_slot): ...new pass_data instance and...
1020         (make_pass_return_slot): ...new function.
1021         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1022         to a subclass of gimple_opt_pass along with...
1023         (pass_data_object_sizes): ...new pass_data instance and...
1024         (make_pass_object_sizes): ...new function.
1025         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1026         global struct to a subclass of gimple_opt_pass along with...
1027         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1028         and...
1029         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1030         (pass_fixup_cfg): Convert from a global struct to a subclass of
1031         gimple_opt_pass along with...
1032         (pass_data_fixup_cfg): ...new pass_data instance and...
1033         (make_pass_fixup_cfg): ...new function.
1034         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1035         (make_pass_mudflap_1): ...new function.
1036         (pass_mudflap_2): Replace declaration with that of...
1037         (make_pass_mudflap_2): ...new function.
1038         (pass_asan): Replace declaration with that of...
1039         (make_pass_asan): ...new function.
1040         (pass_asan_O0): Replace declaration with that of...
1041         (make_pass_asan_O0): ...new function.
1042         (pass_tsan): Replace declaration with that of...
1043         (make_pass_tsan): ...new function.
1044         (pass_tsan_O0): Replace declaration with that of...
1045         (make_pass_tsan_O0): ...new function.
1046         (pass_lower_cf): Replace declaration with that of...
1047         (make_pass_lower_cf): ...new function.
1048         (pass_refactor_eh): Replace declaration with that of...
1049         (make_pass_refactor_eh): ...new function.
1050         (pass_lower_eh): Replace declaration with that of...
1051         (make_pass_lower_eh): ...new function.
1052         (pass_lower_eh_dispatch): Replace declaration with that of...
1053         (make_pass_lower_eh_dispatch): ...new function.
1054         (pass_lower_resx): Replace declaration with that of...
1055         (make_pass_lower_resx): ...new function.
1056         (pass_build_cfg): Replace declaration with that of...
1057         (make_pass_build_cfg): ...new function.
1058         (pass_early_tree_profile): Replace declaration with that of...
1059         (make_pass_early_tree_profile): ...new function.
1060         (pass_cleanup_eh): Replace declaration with that of...
1061         (make_pass_cleanup_eh): ...new function.
1062         (pass_sra): Replace declaration with that of...
1063         (make_pass_sra): ...new function.
1064         (pass_sra_early): Replace declaration with that of...
1065         (make_pass_sra_early): ...new function.
1066         (pass_early_ipa_sra): Replace declaration with that of...
1067         (make_pass_early_ipa_sra): ...new function.
1068         (pass_tail_recursion): Replace declaration with that of...
1069         (make_pass_tail_recursion): ...new function.
1070         (pass_tail_calls): Replace declaration with that of...
1071         (make_pass_tail_calls): ...new function.
1072         (pass_tree_loop): Replace declaration with that of...
1073         (make_pass_tree_loop): ...new function.
1074         (pass_tree_loop_init): Replace declaration with that of...
1075         (make_pass_tree_loop_init): ...new function.
1076         (pass_lim): Replace declaration with that of...
1077         (make_pass_lim): ...new function.
1078         (pass_tree_unswitch): Replace declaration with that of...
1079         (make_pass_tree_unswitch): ...new function.
1080         (pass_predcom): Replace declaration with that of...
1081         (make_pass_predcom): ...new function.
1082         (pass_iv_canon): Replace declaration with that of...
1083         (make_pass_iv_canon): ...new function.
1084         (pass_scev_cprop): Replace declaration with that of...
1085         (make_pass_scev_cprop): ...new function.
1086         (pass_empty_loop): Replace declaration with that of...
1087         (make_pass_empty_loop): ...new function.
1088         (pass_record_bounds): Replace declaration with that of...
1089         (make_pass_record_bounds): ...new function.
1090         (pass_graphite): Replace declaration with that of...
1091         (make_pass_graphite): ...new function.
1092         (pass_graphite_transforms): Replace declaration with that of...
1093         (make_pass_graphite_transforms): ...new function.
1094         (pass_if_conversion): Replace declaration with that of...
1095         (make_pass_if_conversion): ...new function.
1096         (pass_loop_distribution): Replace declaration with that of...
1097         (make_pass_loop_distribution): ...new function.
1098         (pass_vectorize): Replace declaration with that of...
1099         (make_pass_vectorize): ...new function.
1100         (pass_slp_vectorize): Replace declaration with that of...
1101         (make_pass_slp_vectorize): ...new function.
1102         (pass_complete_unroll): Replace declaration with that of...
1103         (make_pass_complete_unroll): ...new function.
1104         (pass_complete_unrolli): Replace declaration with that of...
1105         (make_pass_complete_unrolli): ...new function.
1106         (pass_parallelize_loops): Replace declaration with that of...
1107         (make_pass_parallelize_loops): ...new function.
1108         (pass_loop_prefetch): Replace declaration with that of...
1109         (make_pass_loop_prefetch): ...new function.
1110         (pass_iv_optimize): Replace declaration with that of...
1111         (make_pass_iv_optimize): ...new function.
1112         (pass_tree_loop_done): Replace declaration with that of...
1113         (make_pass_tree_loop_done): ...new function.
1114         (pass_ch): Replace declaration with that of...
1115         (make_pass_ch): ...new function.
1116         (pass_ccp): Replace declaration with that of...
1117         (make_pass_ccp): ...new function.
1118         (pass_phi_only_cprop): Replace declaration with that of...
1119         (make_pass_phi_only_cprop): ...new function.
1120         (pass_build_ssa): Replace declaration with that of...
1121         (make_pass_build_ssa): ...new function.
1122         (pass_build_alias): Replace declaration with that of...
1123         (make_pass_build_alias): ...new function.
1124         (pass_build_ealias): Replace declaration with that of...
1125         (make_pass_build_ealias): ...new function.
1126         (pass_dominator): Replace declaration with that of...
1127         (make_pass_dominator): ...new function.
1128         (pass_dce): Replace declaration with that of...
1129         (make_pass_dce): ...new function.
1130         (pass_dce_loop): Replace declaration with that of...
1131         (make_pass_dce_loop): ...new function.
1132         (pass_cd_dce): Replace declaration with that of...
1133         (make_pass_cd_dce): ...new function.
1134         (pass_call_cdce): Replace declaration with that of...
1135         (make_pass_call_cdce): ...new function.
1136         (pass_merge_phi): Replace declaration with that of...
1137         (make_pass_merge_phi): ...new function.
1138         (pass_split_crit_edges): Replace declaration with that of...
1139         (make_pass_split_crit_edges): ...new function.
1140         (pass_pre): Replace declaration with that of...
1141         (make_pass_pre): ...new function.
1142         (pass_profile): Replace declaration with that of...
1143         (make_pass_profile): ...new function.
1144         (pass_strip_predict_hints): Replace declaration with that of...
1145         (make_pass_strip_predict_hints): ...new function.
1146         (pass_lower_complex_O0): Replace declaration with that of...
1147         (make_pass_lower_complex_O0): ...new function.
1148         (pass_lower_complex): Replace declaration with that of...
1149         (make_pass_lower_complex): ...new function.
1150         (pass_lower_vector): Replace declaration with that of...
1151         (make_pass_lower_vector): ...new function.
1152         (pass_lower_vector_ssa): Replace declaration with that of...
1153         (make_pass_lower_vector_ssa): ...new function.
1154         (pass_lower_omp): Replace declaration with that of...
1155         (make_pass_lower_omp): ...new function.
1156         (pass_diagnose_omp_blocks): Replace declaration with that of...
1157         (make_pass_diagnose_omp_blocks): ...new function.
1158         (pass_expand_omp): Replace declaration with that of...
1159         (make_pass_expand_omp): ...new function.
1160         (pass_expand_omp_ssa): Replace declaration with that of...
1161         (make_pass_expand_omp_ssa): ...new function.
1162         (pass_object_sizes): Replace declaration with that of...
1163         (make_pass_object_sizes): ...new function.
1164         (pass_strlen): Replace declaration with that of...
1165         (make_pass_strlen): ...new function.
1166         (pass_fold_builtins): Replace declaration with that of...
1167         (make_pass_fold_builtins): ...new function.
1168         (pass_stdarg): Replace declaration with that of...
1169         (make_pass_stdarg): ...new function.
1170         (pass_early_warn_uninitialized): Replace declaration with that of...
1171         (make_pass_early_warn_uninitialized): ...new function.
1172         (pass_late_warn_uninitialized): Replace declaration with that of...
1173         (make_pass_late_warn_uninitialized): ...new function.
1174         (pass_cse_reciprocals): Replace declaration with that of...
1175         (make_pass_cse_reciprocals): ...new function.
1176         (pass_cse_sincos): Replace declaration with that of...
1177         (make_pass_cse_sincos): ...new function.
1178         (pass_optimize_bswap): Replace declaration with that of...
1179         (make_pass_optimize_bswap): ...new function.
1180         (pass_optimize_widening_mul): Replace declaration with that of...
1181         (make_pass_optimize_widening_mul): ...new function.
1182         (pass_warn_function_return): Replace declaration with that of...
1183         (make_pass_warn_function_return): ...new function.
1184         (pass_warn_function_noreturn): Replace declaration with that of...
1185         (make_pass_warn_function_noreturn): ...new function.
1186         (pass_cselim): Replace declaration with that of...
1187         (make_pass_cselim): ...new function.
1188         (pass_phiopt): Replace declaration with that of...
1189         (make_pass_phiopt): ...new function.
1190         (pass_forwprop): Replace declaration with that of...
1191         (make_pass_forwprop): ...new function.
1192         (pass_phiprop): Replace declaration with that of...
1193         (make_pass_phiprop): ...new function.
1194         (pass_tree_ifcombine): Replace declaration with that of...
1195         (make_pass_tree_ifcombine): ...new function.
1196         (pass_dse): Replace declaration with that of...
1197         (make_pass_dse): ...new function.
1198         (pass_nrv): Replace declaration with that of...
1199         (make_pass_nrv): ...new function.
1200         (pass_rename_ssa_copies): Replace declaration with that of...
1201         (make_pass_rename_ssa_copies): ...new function.
1202         (pass_sink_code): Replace declaration with that of...
1203         (make_pass_sink_code): ...new function.
1204         (pass_fre): Replace declaration with that of...
1205         (make_pass_fre): ...new function.
1206         (pass_check_data_deps): Replace declaration with that of...
1207         (make_pass_check_data_deps): ...new function.
1208         (pass_copy_prop): Replace declaration with that of...
1209         (make_pass_copy_prop): ...new function.
1210         (pass_vrp): Replace declaration with that of...
1211         (make_pass_vrp): ...new function.
1212         (pass_uncprop): Replace declaration with that of...
1213         (make_pass_uncprop): ...new function.
1214         (pass_return_slot): Replace declaration with that of...
1215         (make_pass_return_slot): ...new function.
1216         (pass_reassoc): Replace declaration with that of...
1217         (make_pass_reassoc): ...new function.
1218         (pass_rebuild_cgraph_edges): Replace declaration with that of...
1219         (make_pass_rebuild_cgraph_edges): ...new function.
1220         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1221         (make_pass_remove_cgraph_callee_edges): ...new function.
1222         (pass_build_cgraph_edges): Replace declaration with that of...
1223         (make_pass_build_cgraph_edges): ...new function.
1224         (pass_local_pure_const): Replace declaration with that of...
1225         (make_pass_local_pure_const): ...new function.
1226         (pass_tracer): Replace declaration with that of...
1227         (make_pass_tracer): ...new function.
1228         (pass_warn_unused_result): Replace declaration with that of...
1229         (make_pass_warn_unused_result): ...new function.
1230         (pass_diagnose_tm_blocks): Replace declaration with that of...
1231         (make_pass_diagnose_tm_blocks): ...new function.
1232         (pass_lower_tm): Replace declaration with that of...
1233         (make_pass_lower_tm): ...new function.
1234         (pass_tm_init): Replace declaration with that of...
1235         (make_pass_tm_init): ...new function.
1236         (pass_tm_mark): Replace declaration with that of...
1237         (make_pass_tm_mark): ...new function.
1238         (pass_tm_memopt): Replace declaration with that of...
1239         (make_pass_tm_memopt): ...new function.
1240         (pass_tm_edges): Replace declaration with that of...
1241         (make_pass_tm_edges): ...new function.
1242         (pass_split_functions): Replace declaration with that of...
1243         (make_pass_split_functions): ...new function.
1244         (pass_feedback_split_functions): Replace declaration with that of...
1245         (make_pass_feedback_split_functions): ...new function.
1246         (pass_strength_reduction): Replace declaration with that of...
1247         (make_pass_strength_reduction): ...new function.
1248         (pass_ipa_lower_emutls): Replace declaration with that of...
1249         (make_pass_ipa_lower_emutls): ...new function.
1250         (pass_ipa_function_and_variable_visibility): Replace declaration with
1251         that of...
1252         (make_pass_ipa_function_and_variable_visibility): ...new function.
1253         (pass_ipa_tree_profile): Replace declaration with that of...
1254         (make_pass_ipa_tree_profile): ...new function.
1255         (pass_early_local_passes): Replace declaration with that of...
1256         (make_pass_early_local_passes): ...new function.
1257         (pass_ipa_whole_program_visibility): Replace declaration with that
1258         of...
1259         (make_pass_ipa_whole_program_visibility): ...new function.
1260         (pass_ipa_lto_gimple_out): Replace declaration with that of...
1261         (make_pass_ipa_lto_gimple_out): ...new function.
1262         (pass_ipa_increase_alignment): Replace declaration with that of...
1263         (make_pass_ipa_increase_alignment): ...new function.
1264         (pass_ipa_inline): Replace declaration with that of...
1265         (make_pass_ipa_inline): ...new function.
1266         (pass_ipa_free_lang_data): Replace declaration with that of...
1267         (make_pass_ipa_free_lang_data): ...new function.
1268         (pass_ipa_free_inline_summary): Replace declaration with that of...
1269         (make_pass_ipa_free_inline_summary): ...new function.
1270         (pass_ipa_cp): Replace declaration with that of...
1271         (make_pass_ipa_cp): ...new function.
1272         (pass_ipa_reference): Replace declaration with that of...
1273         (make_pass_ipa_reference): ...new function.
1274         (pass_ipa_pure_const): Replace declaration with that of...
1275         (make_pass_ipa_pure_const): ...new function.
1276         (pass_ipa_pta): Replace declaration with that of...
1277         (make_pass_ipa_pta): ...new function.
1278         (pass_ipa_lto_finish_out): Replace declaration with that of...
1279         (make_pass_ipa_lto_finish_out): ...new function.
1280         (pass_ipa_tm): Replace declaration with that of...
1281         (make_pass_ipa_tm): ...new function.
1282         (pass_ipa_profile): Replace declaration with that of...
1283         (make_pass_ipa_profile): ...new function.
1284         (pass_ipa_cdtor_merge): Replace declaration with that of...
1285         (make_pass_ipa_cdtor_merge): ...new function.
1286         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1287         of...
1288         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1289         (pass_init_datastructures): Replace declaration with that of...
1290         (make_pass_init_datastructures): ...new function.
1291         (pass_fixup_cfg): Replace declaration with that of...
1292         (make_pass_fixup_cfg): ...new function.
1293         (pass_expand): Replace declaration with that of...
1294         (make_pass_expand): ...new function.
1295         (pass_instantiate_virtual_regs): Replace declaration with that of...
1296         (make_pass_instantiate_virtual_regs): ...new function.
1297         (pass_rtl_fwprop): Replace declaration with that of...
1298         (make_pass_rtl_fwprop): ...new function.
1299         (pass_rtl_fwprop_addr): Replace declaration with that of...
1300         (make_pass_rtl_fwprop_addr): ...new function.
1301         (pass_jump): Replace declaration with that of...
1302         (make_pass_jump): ...new function.
1303         (pass_jump2): Replace declaration with that of...
1304         (make_pass_jump2): ...new function.
1305         (pass_lower_subreg): Replace declaration with that of...
1306         (make_pass_lower_subreg): ...new function.
1307         (pass_cse): Replace declaration with that of...
1308         (make_pass_cse): ...new function.
1309         (pass_fast_rtl_dce): Replace declaration with that of...
1310         (make_pass_fast_rtl_dce): ...new function.
1311         (pass_ud_rtl_dce): Replace declaration with that of...
1312         (make_pass_ud_rtl_dce): ...new function.
1313         (pass_rtl_dce): Replace declaration with that of...
1314         (make_pass_rtl_dce): ...new function.
1315         (pass_rtl_dse1): Replace declaration with that of...
1316         (make_pass_rtl_dse1): ...new function.
1317         (pass_rtl_dse2): Replace declaration with that of...
1318         (make_pass_rtl_dse2): ...new function.
1319         (pass_rtl_dse3): Replace declaration with that of...
1320         (make_pass_rtl_dse3): ...new function.
1321         (pass_rtl_cprop): Replace declaration with that of...
1322         (make_pass_rtl_cprop): ...new function.
1323         (pass_rtl_pre): Replace declaration with that of...
1324         (make_pass_rtl_pre): ...new function.
1325         (pass_rtl_hoist): Replace declaration with that of...
1326         (make_pass_rtl_hoist): ...new function.
1327         (pass_rtl_store_motion): Replace declaration with that of...
1328         (make_pass_rtl_store_motion): ...new function.
1329         (pass_cse_after_global_opts): Replace declaration with that of...
1330         (make_pass_cse_after_global_opts): ...new function.
1331         (pass_rtl_ifcvt): Replace declaration with that of...
1332         (make_pass_rtl_ifcvt): ...new function.
1333         (pass_into_cfg_layout_mode): Replace declaration with that of...
1334         (make_pass_into_cfg_layout_mode): ...new function.
1335         (pass_outof_cfg_layout_mode): Replace declaration with that of...
1336         (make_pass_outof_cfg_layout_mode): ...new function.
1337         (pass_loop2): Replace declaration with that of...
1338         (make_pass_loop2): ...new function.
1339         (pass_rtl_loop_init): Replace declaration with that of...
1340         (make_pass_rtl_loop_init): ...new function.
1341         (pass_rtl_move_loop_invariants): Replace declaration with that of...
1342         (make_pass_rtl_move_loop_invariants): ...new function.
1343         (pass_rtl_unswitch): Replace declaration with that of...
1344         (make_pass_rtl_unswitch): ...new function.
1345         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1346         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1347         (pass_rtl_doloop): Replace declaration with that of...
1348         (make_pass_rtl_doloop): ...new function.
1349         (pass_rtl_loop_done): Replace declaration with that of...
1350         (make_pass_rtl_loop_done): ...new function.
1351         (pass_web): Replace declaration with that of...
1352         (make_pass_web): ...new function.
1353         (pass_cse2): Replace declaration with that of...
1354         (make_pass_cse2): ...new function.
1355         (pass_df_initialize_opt): Replace declaration with that of...
1356         (make_pass_df_initialize_opt): ...new function.
1357         (pass_df_initialize_no_opt): Replace declaration with that of...
1358         (make_pass_df_initialize_no_opt): ...new function.
1359         (pass_reginfo_init): Replace declaration with that of...
1360         (make_pass_reginfo_init): ...new function.
1361         (pass_inc_dec): Replace declaration with that of...
1362         (make_pass_inc_dec): ...new function.
1363         (pass_stack_ptr_mod): Replace declaration with that of...
1364         (make_pass_stack_ptr_mod): ...new function.
1365         (pass_initialize_regs): Replace declaration with that of...
1366         (make_pass_initialize_regs): ...new function.
1367         (pass_combine): Replace declaration with that of...
1368         (make_pass_combine): ...new function.
1369         (pass_if_after_combine): Replace declaration with that of...
1370         (make_pass_if_after_combine): ...new function.
1371         (pass_ree): Replace declaration with that of...
1372         (make_pass_ree): ...new function.
1373         (pass_partition_blocks): Replace declaration with that of...
1374         (make_pass_partition_blocks): ...new function.
1375         (pass_match_asm_constraints): Replace declaration with that of...
1376         (make_pass_match_asm_constraints): ...new function.
1377         (pass_regmove): Replace declaration with that of...
1378         (make_pass_regmove): ...new function.
1379         (pass_split_all_insns): Replace declaration with that of...
1380         (make_pass_split_all_insns): ...new function.
1381         (pass_fast_rtl_byte_dce): Replace declaration with that of...
1382         (make_pass_fast_rtl_byte_dce): ...new function.
1383         (pass_lower_subreg2): Replace declaration with that of...
1384         (make_pass_lower_subreg2): ...new function.
1385         (pass_mode_switching): Replace declaration with that of...
1386         (make_pass_mode_switching): ...new function.
1387         (pass_sms): Replace declaration with that of...
1388         (make_pass_sms): ...new function.
1389         (pass_sched): Replace declaration with that of...
1390         (make_pass_sched): ...new function.
1391         (pass_ira): Replace declaration with that of...
1392         (make_pass_ira): ...new function.
1393         (pass_reload): Replace declaration with that of...
1394         (make_pass_reload): ...new function.
1395         (pass_clean_state): Replace declaration with that of...
1396         (make_pass_clean_state): ...new function.
1397         (pass_branch_prob): Replace declaration with that of...
1398         (make_pass_branch_prob): ...new function.
1399         (pass_value_profile_transformations): Replace declaration with that
1400         of...
1401         (make_pass_value_profile_transformations): ...new function.
1402         (pass_postreload_cse): Replace declaration with that of...
1403         (make_pass_postreload_cse): ...new function.
1404         (pass_gcse2): Replace declaration with that of...
1405         (make_pass_gcse2): ...new function.
1406         (pass_split_after_reload): Replace declaration with that of...
1407         (make_pass_split_after_reload): ...new function.
1408         (pass_branch_target_load_optimize1): Replace declaration with that
1409         of...
1410         (make_pass_branch_target_load_optimize1): ...new function.
1411         (pass_thread_prologue_and_epilogue): Replace declaration with that
1412         of...
1413         (make_pass_thread_prologue_and_epilogue): ...new function.
1414         (pass_stack_adjustments): Replace declaration with that of...
1415         (make_pass_stack_adjustments): ...new function.
1416         (pass_peephole2): Replace declaration with that of...
1417         (make_pass_peephole2): ...new function.
1418         (pass_if_after_reload): Replace declaration with that of...
1419         (make_pass_if_after_reload): ...new function.
1420         (pass_regrename): Replace declaration with that of...
1421         (make_pass_regrename): ...new function.
1422         (pass_cprop_hardreg): Replace declaration with that of...
1423         (make_pass_cprop_hardreg): ...new function.
1424         (pass_reorder_blocks): Replace declaration with that of...
1425         (make_pass_reorder_blocks): ...new function.
1426         (pass_branch_target_load_optimize2): Replace declaration with that
1427         of...
1428         (make_pass_branch_target_load_optimize2): ...new function.
1429         (pass_leaf_regs): Replace declaration with that of...
1430         (make_pass_leaf_regs): ...new function.
1431         (pass_split_before_sched2): Replace declaration with that of...
1432         (make_pass_split_before_sched2): ...new function.
1433         (pass_compare_elim_after_reload): Replace declaration with that of...
1434         (make_pass_compare_elim_after_reload): ...new function.
1435         (pass_sched2): Replace declaration with that of...
1436         (make_pass_sched2): ...new function.
1437         (pass_stack_regs): Replace declaration with that of...
1438         (make_pass_stack_regs): ...new function.
1439         (pass_stack_regs_run): Replace declaration with that of...
1440         (make_pass_stack_regs_run): ...new function.
1441         (pass_df_finish): Replace declaration with that of...
1442         (make_pass_df_finish): ...new function.
1443         (pass_compute_alignments): Replace declaration with that of...
1444         (make_pass_compute_alignments): ...new function.
1445         (pass_duplicate_computed_gotos): Replace declaration with that of...
1446         (make_pass_duplicate_computed_gotos): ...new function.
1447         (pass_variable_tracking): Replace declaration with that of...
1448         (make_pass_variable_tracking): ...new function.
1449         (pass_free_cfg): Replace declaration with that of...
1450         (make_pass_free_cfg): ...new function.
1451         (pass_machine_reorg): Replace declaration with that of...
1452         (make_pass_machine_reorg): ...new function.
1453         (pass_cleanup_barriers): Replace declaration with that of...
1454         (make_pass_cleanup_barriers): ...new function.
1455         (pass_delay_slots): Replace declaration with that of...
1456         (make_pass_delay_slots): ...new function.
1457         (pass_split_for_shorten_branches): Replace declaration with that of...
1458         (make_pass_split_for_shorten_branches): ...new function.
1459         (pass_split_before_regstack): Replace declaration with that of...
1460         (make_pass_split_before_regstack): ...new function.
1461         (pass_convert_to_eh_region_ranges): Replace declaration with that
1462         of...
1463         (make_pass_convert_to_eh_region_ranges): ...new function.
1464         (pass_shorten_branches): Replace declaration with that of...
1465         (make_pass_shorten_branches): ...new function.
1466         (pass_set_nothrow_function_flags): Replace declaration with that of...
1467         (make_pass_set_nothrow_function_flags): ...new function.
1468         (pass_dwarf2_frame): Replace declaration with that of...
1469         (make_pass_dwarf2_frame): ...new function.
1470         (pass_final): Replace declaration with that of...
1471         (make_pass_final): ...new function.
1472         (pass_rtl_seqabstr): Replace declaration with that of...
1473         (make_pass_rtl_seqabstr): ...new function.
1474         (pass_release_ssa_names): Replace declaration with that of...
1475         (make_pass_release_ssa_names): ...new function.
1476         (pass_early_inline): Replace declaration with that of...
1477         (make_pass_early_inline): ...new function.
1478         (pass_inline_parameters): Replace declaration with that of...
1479         (make_pass_inline_parameters): ...new function.
1480         (pass_update_address_taken): Replace declaration with that of...
1481         (make_pass_update_address_taken): ...new function.
1482         (pass_convert_switch): Replace declaration with that of...
1483         (make_pass_convert_switch): ...new function.
1484         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1485         to a subclass of simple_ipa_opt_pass along with...
1486         (pass_data_ipa_tree_profile): ...new pass_data instance and...
1487         (make_pass_ipa_tree_profile): ...new function.
1488         * tree-sra.c (pass_sra_early): Convert from a global struct to a
1489         subclass of gimple_opt_pass along with...
1490         (pass_data_sra_early): ...new pass_data instance and...
1491         (make_pass_sra_early): ...new function.
1492         (pass_sra): Convert from a global struct to a subclass of
1493         gimple_opt_pass along with...
1494         (pass_data_sra): ...new pass_data instance and...
1495         (make_pass_sra): ...new function.
1496         (pass_early_ipa_sra): Convert from a global struct to a subclass of
1497         gimple_opt_pass along with...
1498         (pass_data_early_ipa_sra): ...new pass_data instance and...
1499         (make_pass_early_ipa_sra): ...new function.
1500         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1501         subclass of gimple_opt_pass along with...
1502         (pass_data_ccp): ...new pass_data instance and...
1503         (make_pass_ccp): ...new function.
1504         (pass_fold_builtins): Convert from a global struct to a subclass of
1505         gimple_opt_pass along with...
1506         (pass_data_fold_builtins): ...new pass_data instance and...
1507         (make_pass_fold_builtins): ...new function.
1508         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1509         subclass of gimple_opt_pass along with...
1510         (pass_data_copy_prop): ...new pass_data instance and...
1511         (make_pass_copy_prop): ...new function.
1512         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1513         global struct to a subclass of gimple_opt_pass along with...
1514         (pass_data_rename_ssa_copies): ...new pass_data instance and...
1515         (make_pass_rename_ssa_copies): ...new function.
1516         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1517         subclass of gimple_opt_pass along with...
1518         (pass_data_dce): ...new pass_data instance and...
1519         (make_pass_dce): ...new function.
1520         (pass_dce_loop): Convert from a global struct to a subclass of
1521         gimple_opt_pass along with...
1522         (pass_data_dce_loop): ...new pass_data instance and...
1523         (make_pass_dce_loop): ...new function.
1524         (pass_cd_dce): Convert from a global struct to a subclass of
1525         gimple_opt_pass along with...
1526         (pass_data_cd_dce): ...new pass_data instance and...
1527         (make_pass_cd_dce): ...new function.
1528         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1529         subclass of gimple_opt_pass along with...
1530         (pass_data_dominator): ...new pass_data instance and...
1531         (make_pass_dominator): ...new function.
1532         (pass_phi_only_cprop): Convert from a global struct to a subclass of
1533         gimple_opt_pass along with...
1534         (pass_data_phi_only_cprop): ...new pass_data instance and...
1535         (make_pass_phi_only_cprop): ...new function.
1536         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1537         subclass of gimple_opt_pass along with...
1538         (pass_data_dse): ...new pass_data instance and...
1539         (make_pass_dse): ...new function.
1540         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1541         a subclass of gimple_opt_pass along with...
1542         (pass_data_forwprop): ...new pass_data instance and...
1543         (make_pass_forwprop): ...new function.
1544         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1545         struct to a subclass of gimple_opt_pass along with...
1546         (pass_data_tree_ifcombine): ...new pass_data instance and...
1547         (make_pass_tree_ifcombine): ...new function.
1548         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1549         subclass of gimple_opt_pass along with...
1550         (pass_data_ch): ...new pass_data instance and...
1551         (make_pass_ch): ...new function.
1552         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1553         subclass of gimple_opt_pass along with...
1554         (pass_data_tree_loop): ...new pass_data instance and...
1555         (make_pass_tree_loop): ...new function.
1556         (pass_tree_loop_init): Convert from a global struct to a subclass of
1557         gimple_opt_pass along with...
1558         (pass_data_tree_loop_init): ...new pass_data instance and...
1559         (make_pass_tree_loop_init): ...new function.
1560         (pass_lim): Convert from a global struct to a subclass of
1561         gimple_opt_pass along with...
1562         (pass_data_lim): ...new pass_data instance and...
1563         (make_pass_lim): ...new function.
1564         (pass_tree_unswitch): Convert from a global struct to a subclass of
1565         gimple_opt_pass along with...
1566         (pass_data_tree_unswitch): ...new pass_data instance and...
1567         (make_pass_tree_unswitch): ...new function.
1568         (pass_predcom): Convert from a global struct to a subclass of
1569         gimple_opt_pass along with...
1570         (pass_data_predcom): ...new pass_data instance and...
1571         (make_pass_predcom): ...new function.
1572         (pass_vectorize): Convert from a global struct to a subclass of
1573         gimple_opt_pass along with...
1574         (pass_data_vectorize): ...new pass_data instance and...
1575         (make_pass_vectorize): ...new function.
1576         (pass_graphite): Convert from a global struct to a subclass of
1577         gimple_opt_pass along with...
1578         (pass_data_graphite): ...new pass_data instance and...
1579         (make_pass_graphite): ...new function.
1580         (pass_graphite_transforms): Convert from a global struct to a subclass
1581         of gimple_opt_pass along with...
1582         (pass_data_graphite_transforms): ...new pass_data instance and...
1583         (make_pass_graphite_transforms): ...new function.
1584         (pass_check_data_deps): Convert from a global struct to a subclass of
1585         gimple_opt_pass along with...
1586         (pass_data_check_data_deps): ...new pass_data instance and...
1587         (make_pass_check_data_deps): ...new function.
1588         (pass_iv_canon): Convert from a global struct to a subclass of
1589         gimple_opt_pass along with...
1590         (pass_data_iv_canon): ...new pass_data instance and...
1591         (make_pass_iv_canon): ...new function.
1592         (pass_scev_cprop): Convert from a global struct to a subclass of
1593         gimple_opt_pass along with...
1594         (pass_data_scev_cprop): ...new pass_data instance and...
1595         (make_pass_scev_cprop): ...new function.
1596         (pass_record_bounds): Convert from a global struct to a subclass of
1597         gimple_opt_pass along with...
1598         (pass_data_record_bounds): ...new pass_data instance and...
1599         (make_pass_record_bounds): ...new function.
1600         (pass_complete_unroll): Convert from a global struct to a subclass of
1601         gimple_opt_pass along with...
1602         (pass_data_complete_unroll): ...new pass_data instance and...
1603         (make_pass_complete_unroll): ...new function.
1604         (pass_complete_unrolli): Convert from a global struct to a subclass of
1605         gimple_opt_pass along with...
1606         (pass_data_complete_unrolli): ...new pass_data instance and...
1607         (make_pass_complete_unrolli): ...new function.
1608         (pass_parallelize_loops): Convert from a global struct to a subclass
1609         of gimple_opt_pass along with...
1610         (pass_data_parallelize_loops): ...new pass_data instance and...
1611         (make_pass_parallelize_loops): ...new function.
1612         (pass_loop_prefetch): Convert from a global struct to a subclass of
1613         gimple_opt_pass along with...
1614         (pass_data_loop_prefetch): ...new pass_data instance and...
1615         (make_pass_loop_prefetch): ...new function.
1616         (pass_iv_optimize): Convert from a global struct to a subclass of
1617         gimple_opt_pass along with...
1618         (pass_data_iv_optimize): ...new pass_data instance and...
1619         (make_pass_iv_optimize): ...new function.
1620         (pass_tree_loop_done): Convert from a global struct to a subclass of
1621         gimple_opt_pass along with...
1622         (pass_data_tree_loop_done): ...new pass_data instance and...
1623         (make_pass_tree_loop_done): ...new function.
1624         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1625         struct to a subclass of gimple_opt_pass along with...
1626         (pass_data_cse_reciprocals): ...new pass_data instance and...
1627         (make_pass_cse_reciprocals): ...new function.
1628         (pass_cse_sincos): Convert from a global struct to a subclass of
1629         gimple_opt_pass along with...
1630         (pass_data_cse_sincos): ...new pass_data instance and...
1631         (make_pass_cse_sincos): ...new function.
1632         (pass_optimize_bswap): Convert from a global struct to a subclass of
1633         gimple_opt_pass along with...
1634         (pass_data_optimize_bswap): ...new pass_data instance and...
1635         (make_pass_optimize_bswap): ...new function.
1636         (pass_optimize_widening_mul): Convert from a global struct to a
1637         subclass of gimple_opt_pass along with...
1638         (pass_data_optimize_widening_mul): ...new pass_data instance and...
1639         (make_pass_optimize_widening_mul): ...new function.
1640         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1641         subclass of gimple_opt_pass along with...
1642         (pass_data_phiopt): ...new pass_data instance and...
1643         (make_pass_phiopt): ...new function.
1644         (pass_cselim): Convert from a global struct to a subclass of
1645         gimple_opt_pass along with...
1646         (pass_data_cselim): ...new pass_data instance and...
1647         (make_pass_cselim): ...new function.
1648         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1649         subclass of gimple_opt_pass along with...
1650         (pass_data_phiprop): ...new pass_data instance and...
1651         (make_pass_phiprop): ...new function.
1652         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1653         subclass of gimple_opt_pass along with...
1654         (pass_data_pre): ...new pass_data instance and...
1655         (make_pass_pre): ...new function.
1656         (pass_fre): Convert from a global struct to a subclass of
1657         gimple_opt_pass along with...
1658         (pass_data_fre): ...new pass_data instance and...
1659         (make_pass_fre): ...new function.
1660         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1661         subclass of gimple_opt_pass along with...
1662         (pass_data_reassoc): ...new pass_data instance and...
1663         (make_pass_reassoc): ...new function.
1664         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1665         subclass of gimple_opt_pass along with...
1666         (pass_data_sink_code): ...new pass_data instance and...
1667         (make_pass_sink_code): ...new function.
1668         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1669         subclass of gimple_opt_pass along with...
1670         (pass_data_strlen): ...new pass_data instance and...
1671         (make_pass_strlen): ...new function.
1672         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1673         struct to a subclass of gimple_opt_pass along with...
1674         (pass_data_build_alias): ...new pass_data instance and...
1675         (make_pass_build_alias): ...new function.
1676         (pass_build_ealias): Convert from a global struct to a subclass of
1677         gimple_opt_pass along with...
1678         (pass_data_build_ealias): ...new pass_data instance and...
1679         (make_pass_build_ealias): ...new function.
1680         (pass_ipa_pta): Convert from a global struct to a subclass of
1681         simple_ipa_opt_pass along with...
1682         (pass_data_ipa_pta): ...new pass_data instance and...
1683         (make_pass_ipa_pta): ...new function.
1684         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1685         subclass of gimple_opt_pass along with...
1686         (pass_data_uncprop): ...new pass_data instance and...
1687         (make_pass_uncprop): ...new function.
1688         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1689         global struct to a subclass of gimple_opt_pass along with...
1690         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1691         (make_pass_late_warn_uninitialized): ...new function.
1692         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1693         to a subclass of gimple_opt_pass along with...
1694         (pass_data_init_datastructures): ...new pass_data instance and...
1695         (make_pass_init_datastructures): ...new function.
1696         (pass_early_warn_uninitialized): Convert from a global struct to a
1697         subclass of gimple_opt_pass along with...
1698         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1699         (make_pass_early_warn_uninitialized): ...new function.
1700         (pass_update_address_taken): Convert from a global struct to a
1701         subclass of gimple_opt_pass along with...
1702         (pass_data_update_address_taken): ...new pass_data instance and...
1703         (make_pass_update_address_taken): ...new function.
1704         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1705         struct to a subclass of gimple_opt_pass along with...
1706         (pass_data_release_ssa_names): ...new pass_data instance and...
1707         (make_pass_release_ssa_names): ...new function.
1708         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1709         subclass of gimple_opt_pass along with...
1710         (pass_data_stdarg): ...new pass_data instance and...
1711         (make_pass_stdarg): ...new function.
1712         * tree-switch-conversion.c (pass_convert_switch): Convert from a
1713         global struct to a subclass of gimple_opt_pass along with...
1714         (pass_data_convert_switch): ...new pass_data instance and...
1715         (make_pass_convert_switch): ...new function.
1716         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1717         to a subclass of gimple_opt_pass along with...
1718         (pass_data_tail_recursion): ...new pass_data instance and...
1719         (make_pass_tail_recursion): ...new function.
1720         (pass_tail_calls): Convert from a global struct to a subclass of
1721         gimple_opt_pass along with...
1722         (pass_data_tail_calls): ...new pass_data instance and...
1723         (make_pass_tail_calls): ...new function.
1724         * tree-vect-generic.c (pass_lower_vector): Convert from a global
1725         struct to a subclass of gimple_opt_pass along with...
1726         (pass_data_lower_vector): ...new pass_data instance and...
1727         (make_pass_lower_vector): ...new function.
1728         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1729         gimple_opt_pass along with...
1730         (pass_data_lower_vector_ssa): ...new pass_data instance and...
1731         (make_pass_lower_vector_ssa): ...new function.
1732         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1733         to a subclass of gimple_opt_pass along with...
1734         (pass_data_slp_vectorize): ...new pass_data instance and...
1735         (make_pass_slp_vectorize): ...new function.
1736         (pass_ipa_increase_alignment): Convert from a global struct to a
1737         subclass of simple_ipa_opt_pass along with...
1738         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1739         (make_pass_ipa_increase_alignment): ...new function.
1740         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1741         gimple_opt_pass along with...
1742         (pass_data_vrp): ...new pass_data instance and...
1743         (make_pass_vrp): ...new function.
1744         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1745         subclass of simple_ipa_opt_pass along with...
1746         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1747         (make_pass_ipa_free_lang_data): ...new function.
1748         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1749         gimple_opt_pass along with...
1750         (pass_data_tsan): ...new pass_data instance and...
1751         (make_pass_tsan): ...new function.
1752         (pass_tsan_O0): Convert from a global struct to a subclass of
1753         gimple_opt_pass along with...
1754         (pass_data_tsan_O0): ...new pass_data instance and...
1755         (make_pass_tsan_O0): ...new function.
1756         * var-tracking.c (pass_variable_tracking): Convert from a global
1757         struct to a subclass of rtl_opt_pass along with...
1758         (pass_data_variable_tracking): ...new pass_data instance and...
1759         (make_pass_variable_tracking): ...new function.
1760         * web.c (pass_web): Convert from a global struct to a subclass of
1761         rtl_opt_pass along with...
1762         (pass_data_web): ...new pass_data instance and...
1763         (make_pass_web): ...new function.
1764         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1765         declaration with that of...
1766         (make_pass_mode_switch_use): ...new function.
1767         (pass_resolve_sw_modes): Replace declaration with that of...
1768         (make_pass_resolve_sw_modes): ...new function.
1769         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1770         from a global struct to a subclass of rtl_opt_pass along with...
1771         (pass_data_mode_switch_use): ...new pass_data instance and...
1772         (make_pass_mode_switch_use): ...new function.
1773         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1774         from a global struct to a subclass of rtl_opt_pass along with...
1775         (pass_data_resolve_sw_modes): ...new pass_data instance and...
1776         (make_pass_resolve_sw_modes): ...new function.
1777         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1778         struct to a subclass of rtl_opt_pass along with...
1779         (pass_data_insert_vzeroupper): ...new pass_data instance and...
1780         (make_pass_insert_vzeroupper): ...new function.
1781         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1782         global struct to a subclass of rtl_opt_pass along with...
1783         (pass_data_work_around_errata): ...new pass_data instance and...
1784         (make_pass_work_around_errata): ...new function.
1785         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1786         struct to a subclass of rtl_opt_pass along with...
1787         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1788         (make_pass_mips_machine_reorg2): ...new function.
1790 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1792         * passes.c (pass_manager::operator new): New.
1794 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1796         Handwritten part of conversion of passes to C++ classes.
1798         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1799         (toplev.o): Add dep on PASS_MANAGER_H.
1800         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1801         of early local pases to reflect this moving from a global to a
1802         member of gcc::pass_manager.
1803         (cgraph_add_new_function): Likewise.
1804         * lto-cgraph.c (lto_output_node): Update for conversion of
1805         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1806         * passes.c (opt_pass::clone): New.
1807         (opt_pass::gate): New.
1808         (opt_pass::execute): New.
1809         (opt_pass::opt_pass): New.
1810         (pass_manager::execute_early_local_passes): New.
1811         (pass_manager::execute_pass_mode_switching): new.
1812         (finish_optimization_passes): Convert to...
1813         (pass_manager::finish_optimization_passes): ...this.
1814         (finish_optimization_passes): Update for conversion of passes to
1815         C++ classes.
1816         (register_dump_files_1): Use has_gate since we cannot portably
1817         check a vtable entry against NULL.
1818         (dump_one_pass): Likewise.
1819         (ipa_write_summaries_2): Likewise.
1820         (ipa_write_optimization_summaries_1): Likewise.
1821         (ipa_read_summaries_1): Likewise.
1822         (ipa_read_optimization_summaries_1): Likewise.
1823         (execute_ipa_stmt_fixups): Likewise.
1824         (pass_manager::pass_manager): Rewrite pass-creation, invoking
1825         pass-creation functions rather than wiring up globals, and
1826         storing the results in fields of pass_manager generated using
1827         pass-instances.def.
1828         (pass_manager::dump_profile_report): Update for conversion of
1829         passes to C++ classes.
1830         (pass_manager::execute_ipa_summary_passes): Likewise.
1831         (execute_one_ipa_transform_pass): Likewise.
1832         (execute_one_pass): Use has_gate and has_execute since we cannot
1833         portably check a vtable entry against NULL.
1834         * pass_manager.h (pass_manager::finish_optimization_passes): New.
1835         (pass_manager): Use pass-instances.def to add fields for the
1836         various pass instances.
1837         * toplev.c (finalize): Update for move of
1838         finish_optimization_passes to a method of gcc::pass_manager.
1839         * toplev.h (finish_optimization_passes): Move to method of class
1840         pass_manager.
1841         * tree-pass.h (struct pass_data): New.
1842         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
1843         (opt_pass::gate): Convert to virtual function.
1844         (opt_pass::~opt_pass): New.
1845         (opt_pass::clone): New.
1846         (opt_pass::execute): Convert to virtual function.
1847         (opt_pass::opt_pass): New.
1848         (opt_pass::ctxt_): new.
1849         (gimple_opt_pass): Convert to subclass of opt_pass.
1850         (gimple_opt_pass::gimple_opt_pass): New.
1851         (rtl_opt_pass): Convert to subclass of opt_pass.
1852         (rtl_opt_pass::rtl_opt_pass): New.
1853         (ipa_opt_pass_d): Convert to subclass of opt_pass.
1854         (ipa_opt_pass_d::ipa_opt_pass_d): New.
1855         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
1856         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
1857         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
1858         invocation of pass_mode_switching to reflect this moving from a
1859         global to a member of gcc::pass_manager.
1860         (ix86_option_override): Rework how pass_insert_vzeroupper is
1861         added to the pass_manager to reflect autogenerated changes.
1862         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
1864 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
1866         PR rtl-optimization/57708
1867         * recog.c (peep2_find_free_register): Validate all regs in a
1868         multi-reg mode.
1870 2013-08-05  Jan Hubicka  <jh@suse.cz>
1872         PR lto/57602
1873         * cgraph.c (verify_cgraph_node): Accept local flags from other
1874         partitions.
1875         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
1876         (function_and_variable_visibility): Likewise.
1877         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
1879 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1881         * graph.c (init_graph_slim_pretty_print): Remove.
1882         (print_graph_cfg): Do not call it.  Use local pretty printer.
1883         (start_graph_dump): Likewise.
1885 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1887         * gimple-pretty-print.c (buffer): Remove.
1888         (initialized): Likewise.
1889         (maybe_init_pretty_print): Likewise.
1890         (print_gimple_stmt): Do not call it.  Use non-static local
1891         pretty_printer variable.
1892         (print_gimple_expr): Likewise.
1893         (print_gimple_seq): Likewise.
1894         (gimple_dump_bb): Likewise.
1896 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1898         * asan.c (asan_pp): Remove.
1899         (asan_pp_initialized): Likewise.
1900         (asan_pp_initialize): Likewise.
1901         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
1902         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
1903         (asan_add_global): Likewise.
1905 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1907         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
1908         * diagnostic.h (diagnostic_flush_buffer): Adjust.
1909         * pretty-print.c (pp_formatted_text_data): Likewise.
1910         (pp_indent): Rename from pp_base_indent.
1911         (pp_format): Rename from pp_base_format.
1912         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
1913         (pp_format_verbatim): Rename from pp_base_format_verbatim.
1914         (pp_flush): Rename from pp_base_flush.
1915         (pp_set_line_maximum_length): Rename from
1916         pp_base_set_line_maximum_length.
1917         (pp_clear_output_area): Rename from pp_base_clear_output_area.
1918         (pp_set_prefix): Rename from pp_base_set_prefix.
1919         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
1920         (pp_emit_prefix): Rename from pp_base_emit_prefix.
1921         (pp_append_text): Rename from pp_base_append_text.
1922         (pp_formatted_text): Rename from pp_base_formatted_text.
1923         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
1924         (pp_remaining_character_count_for_line): Rename from
1925         pp_base_remaining_character_count_for_line.
1926         (pp_newline): Rename from pp_base_newline.
1927         (pp_character): Rename from pp_base_character.
1928         (pp_string): Rename from pp_base_string.
1929         (pp_maybe_space): Rename from pp_base_maybe_space.
1930         * asan.c (asan_pp_string): Adjust.
1931         (asan_emit_stack_protection): Likewise.
1932         (asan_add_global): Likewise.
1933         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
1934         * tree-mudflap.c (mf_varname_tree): Likewise.
1935         * tree-pretty-print.c (pp_tree_identifier): Rename from
1936         pp_base_tree_identifier.
1937         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
1938         Declare as function.
1940 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1942         * pretty-print.h (pp_bar_bar): New.
1943         (pp_ampersand_ampersand): Likewise.
1944         (pp_less_equal): Likewise.
1945         (pp_greater_equal): Likewise.
1946         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
1947         printer functions instead of pp_string or operators and punctuators.
1948         (dump_gimple_call): Likewise.
1949         (dump_gimple_omp_for): Likewise.
1950         (dump_gimple_transaction): Likewise.
1951         (dump_gimple_phi): Likewise.
1952         (pp_gimple_stmt_1): Likewise.
1953         * sched-vis.c (print_insn): Likewise.
1954         * tree-mudflap.c (mf_varname_tree): Likewise.
1955         * tree-pretty-print.c (dump_block_node): Likewise.
1956         (dump_generic_node): Likewise.
1958 2013-08-02  Jan Hubicka  <jh@suse.cz>
1960         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
1961         boundaries.
1962         * lto-streamer-out.c (tree_is_indexable): Results decls and
1963         parm decls are not indexable.
1964         (DFS_write_tree_body): Do not follow args and results.
1965         (hash_tree): Likewise.
1966         (output_functions): Rearrange so struct function is needed
1967         only when real body is output; be able to also ouptut abstract
1968         functions; output DECL_ARGUMENTS and DECL_RESULT.
1969         (lto_output): When not in WPA, ale store abstract functions.
1970         (write_symbol): Do not care about RESULT_DECL.
1971         (output_symbol_p): Handle correctly sbtract decls.
1972         * lto-streamer-in.c (input_function): Rearrange so struct
1973         function can be NULL at entry; allow streaming of
1974         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
1975         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
1976         sanity check during LTO.
1977         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
1978         RESULT_DECl and DECL_ARGUMENTS.
1979         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
1980         Likewise.
1982 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1984         * pretty-print.h (pp_underscore): New.
1985         (pp_comma): Tidy.
1986         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
1987         printer functions instead of pp_character.
1988         (dump_binary_rhs): Likewise.
1989         (dump_ternary_rhs): Likewise.
1990         (dump_gimple_call_args): Likewise.
1991         (pp_points_to_solution): Likewise.
1992         (dump_gimple_call): Likewise.
1993         (dump_gimple_switch): Likewise.
1994         (dump_gimple_cond): Likewise.
1995         (dump_gimple_bind): Likewise.
1996         (dump_gimple_try): Likewise.
1997         (dump_gimple_omp_for): Likewise.
1998         (dump_gimple_omp_continue): Likewise.
1999         (dump_gimple_omp_single): Likewise.
2000         (dump_gimple_omp_sections): Likewise.
2001         (dump_gimple_omp_block): Likewise.
2002         (dump_gimple_omp_critical): Likewise.
2003         (dump_gimple_transaction): Likewise.
2004         (dump_gimple_asm): Likewise.
2005         (dump_gimple_phi): Likewise.
2006         (dump_gimple_omp_parallel): Likewise.
2007         (dump_gimple_omp_task): Likewise.
2008         (dump_gimple_omp_atomic_load): Likewise.
2009         (dump_gimple_omp_atomic_store): Likewise.
2010         (dump_gimple_mem_ops): Likewise.
2011         (pp_gimple_stmt_1): Likewise.
2012         (pp_cfg_jump): Likewise.
2013         (dump_implicit_edges): Likewise.
2014         (gimple_dump_bb_for_graph): Likewise.
2015         * graph.c (draw_cfg_node): Likewise.
2016         * langhooks.c (lhd_print_error_function): Likewise.
2017         * sched-vis.c (print_exp): Likewise.
2018         (print_value): Likewise.
2019         (print_pattern): Likewise.
2020         (print_insn): Likewise.
2021         (rtl_dump_bb_for_graph): Likewise.
2022         * tree-pretty-print.c (dump_function_declaration): Likewise.
2023         (dump_array_domain): Likewise.
2024         (dump_omp_clause): Likewise.
2025         (dump_location): Likewise.
2026         (dump_generic_node): Likewise.
2027         (print_struct_decl): Likewise.
2028         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2030 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2032         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2033         candidate table when replacing a candidate statement.
2034         (replace_rhs_if_not_dup): Likewise.
2035         (replace_one_candidate): Likewise.
2037 2013-08-02  Jan Hubicka  <jh@suse.cz>
2038             Martin Liska  <marxin.liska@gmail.com>
2040         * cgraphunit.c (add_new_function): Fix logic when adding from
2041         late IPA pass.
2042         (assemble_thunk): Rename to ...
2043         (expand_thunk); .. this one; export; get it working with
2044         general functions; make produced gimple valid.
2045         * cgraph.h (expand_thunk): Declare.
2047 2013-08-02  Jan Hubicka  <jh@suse.cz>
2049         * ipa-cp.c (gather_context_independent_values): Use
2050         ipa_get_param_move_cost.
2051         (get_replacement_map): Remove PARAM; move parameter folding
2052         into tree-inline.c
2053         (create_specialized_node): Update.
2054         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2055         assert that we have gimple body; update move_cost.
2056         (count_formal_params): Assert that we have gimple body.
2057         (ipa_dump_param): New function.
2058         (ipa_alloc_node_params): Break out from ...
2059         (ipa_initialize_node_params): ... here.
2060         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2061         (ipa_write_node_info): Stream move costs.
2062         (ipa_read_node_info): Read move costs.
2063         (ipa_update_after_lto_read): Do not recompute node params.
2064         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2065         (ipa_get_param): Check we are not in WPA.
2066         (ipa_get_param_move_cost): New.
2067         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2068         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2069         parm numbers to be present.
2071 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2073         PR rtl-optimization/58048
2074         * lra-constraints.c (process_alt_operands): Don't check asm
2075         operand on register.
2077 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2079         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2080         the implied StoreLoad barrier for atomic operations if before.
2082 2013-08-02  Jan Hubicka  <jh@suse.cz>
2083             Martin Liska  <marxin.liska@gmail.com>
2085         * cgraph.c (cgraph_function_body_availability): Do not check
2086         cgraph flags.
2087         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2088         symtab_node_availability): Declare.
2089         * ipa.c (can_replace_by_local_alias): New.
2090         (function_and_variable_visibility): Use it.
2091         * symtab.c (symtab_for_node_and_aliases,
2092         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2094 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2096         PR rtl-optimization/57963
2097         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2098         (lra_constraints): Use them.
2100 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2102         * config/arm/types.md (define_attr "type"): Add "load_acq"
2103         and "store_rel".
2104         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2105         changes.
2106         (cortex_a53_store1): Likewise.
2108 2013-08-01  Jan Hubicka  <jh@suse.cz>
2110         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2111         partitions are not needed.
2113 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
2115         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2116         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2117         MAYBE_NON_Q_CLASS_P where appropriate.
2119 2013-08-01  Jan Hubicka  <jh@suse.cz>
2121         * cgraph.h (release_function_body): Declare.
2122         * tree.c (free_lang_data_in_decl): Free, parameters and return values
2123         of unused delcarations.
2125 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2127         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2128         RTL form when subtracting a constant.
2130 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2132         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2133         Generate canonical plus rtx with negated immediate instead of minus
2134         where appropriate.
2135         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2137 2013-08-01  Jan Hubicka  <jh@suse.cz>
2139         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2140         (cgraph_release_function_body): Likewise.
2141         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2142         * cgraph.h (cgrpah_node): Rename abstract_and_needed
2143         to used_as_abstract_origin.
2144         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2145         symbols used as abstract origins.
2146         * cgraphunit.c (analyze_functions): Update.
2147         * ipa.c (symtab_remove_unreachable_nodes): Recompute
2148         used_as_abstract_origin.
2149         * tree-inline.c (tree_function_versioning): Update
2150         used_as_abstract_origin; be ready for DECL_RESULT and
2151         DECL_ARGUMENTS to be NULL.
2153         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2154         for abstract functions.
2155         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2156         real symbols.
2158 2013-08-01  Jan Hubicka  <jh@suse.cz>
2160         * profile.c (compute_value_histograms): Fix thinko.
2162 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2164         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
2165         aarch-common-protos.h to extra_headers.
2166         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2167         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2168         * config/aarch64/t-aarch64 (aarch-common.o): Define.
2170 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
2172         * config/aarch64/aarch64.md (define_attr "type"): Delete.
2173         Include "../arm/types.md".  Define "type" attribute for all patterns.
2174         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2175         attribute changes.
2177 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
2179         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2180         to support power8 load fusion.
2181         (fusion_gpr_mem_load): Likewise.
2183         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2185         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2186         declarations for power8 load fusion.
2187         (emit_fusion_gpr_load): Likewise.
2189         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2190         tuning for power8, turn on fusion mode by default.  Turn on sign
2191         extending fusion mode if normal fusion mode is on, and we are at
2192         -O2 or -O3.
2193         (fusion_gpr_load_p): New function, return true if we can fuse an
2194         addis instruction with a dependent load to a GPR.
2195         (emit_fusion_gpr_load): Emit the instructions for power8 load
2196         fusion to GPRs.
2198         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2199         (VSX load fusion peepholes): New peepholes to fuse together an
2200         addi instruction with a VSX load instruction.
2202         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2203         peepholes to fuse an addis instruction with a load to a GPR base
2204         register.  If we are supporting sign extending fusions, convert
2205         sign extending loads to zero extending loads and add an explicit
2206         sign extension.
2208 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2210         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
2211         aarch-common-protos.h to extra_headers.
2212         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2213         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2214         (arm_early_store_addr_dep): Likewise.
2215         (arm_no_early_alu_shift_dep): Likewise.
2216         (arm_no_early_alu_shift_value_dep): Likewise.
2217         (arm_no_early_mul_dep): Likewise.
2218         (arm_no_early_store_addr_dep): Likewise.
2219         (arm_mac_accumulator_is_mul_result): Likewise.
2220         (arm_mac_accumulator_is_result): Likewise.
2221         * config/arm/aarch-common.c: ... here.  New file.
2222         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2223         here to ...
2224         (arm_early_store_addr_dep): Likewise.
2225         (arm_no_early_alu_shift_dep): Likewise.
2226         (arm_no_early_alu_shift_value_dep): Likewise.
2227         (arm_no_early_mul_dep): Likewise.
2228         (arm_no_early_store_addr_dep): Likewise.
2229         (arm_mac_accumulator_is_mul_result): Likewise.
2230         (arm_mac_accumulator_is_result): Likewise.
2231         * config/arm/aarch-common-protos.h: ... here.  New file.
2232         * config/arm/t-arm (aarch-common.o): Define.
2234 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
2236         * config/arm/arm.md: Include new file "types.md".
2237         (define_attr "type"): Move from here to ...
2238         (define_attr "mul32"): Likewise.
2239         (define_attr "mul64"): Likewise.
2240         * config/arm/types.md: ... here.  New file.
2242 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2244         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2245         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2247 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2249         * gen-pass-instances.awk: Fix offset of substr().
2251 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
2253         * Makefile.in (pass-instances.def): New.
2254         (passes.o): Replace dependency on passes.def with one on
2255         pass-instances.def
2257         * gen-pass-instances.awk: New.
2259         * passes.c (pass_manager::pass_manager): Use pass-instances.def
2260         rather than passes.def, updating local definition of NEXT_PASS
2261         macro to add an extra NUM parameter (currently unused).
2263 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
2265         * Makefile.in (PASS_MANAGER_H): New.
2266         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2267         (passes.o): Likewise.
2268         (statistics.o): Likewise.
2269         (cgraphunit.o): Likewise.
2270         (context.o): Depend on PASS_MANAGER_H.
2272         * pass_manager.h: New.
2274         * cgraphunit.c (cgraph_add_new_function): Update for moves
2275         of globals to fields of pass_manager.
2276         (analyze_function): Likewise.
2277         (expand_function): Likewise.
2278         (ipa_passes): Likewise.
2279         (compile): Likewise.
2281         * context.c (context::context): New.
2282         * context.h  (context::context): New.
2283         (context::get_passes): New.
2284         (context::passes_): New.
2286         * lto-cgraph.c (input_node): Update for moves of globals to
2287         fields of pass_manager.
2289         * passes.c (all_passes): Remove, in favor of a field of the
2290         same name within the new class pass_manager.
2291         (all_small_ipa_passes): Likewise.
2292         (all_lowering_passes): Likewise.
2293         (all_regular_ipa_passes): Likewise.
2294         (all_late_ipa_passes): Likewise.
2295         (all_lto_gen_passes): Likewise.
2296         (passes_by_id): Likewise.
2297         (passes_by_id_size): Likewise.
2298         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2299         the new class pass_manager.
2300         (set_pass_for_id): Convert to...
2301         (pass_manager::set_pass_for_id): ...method.
2302         (get_pass_for_id): Convert to...
2303         (pass_manager::get_pass_for_id): ...method.
2304         (register_one_dump_file): Move body of implementation into...
2305         (pass_manager::register_one_dump_file): ...here.
2306         (register_dump_files_1): Convert to...
2307         (pass_manager::register_dump_files_1): ...method.
2308         (register_dump_files): Convert to...
2309         (pass_manager::register_dump_files): ...method.
2310         (create_pass_tab): Update for moves of globals to fields of
2311         pass_manager.
2312         (dump_passes): Move body of implementation into...
2313         (pass_manager::dump_passes): ...here.
2314         (register_pass): Move body of implementation into...
2315         (pass_manager::register_pass): ...here.
2316         (init_optimization_passes): Convert into...
2317         (pass_manager::pass_manager): ...constructor for new
2318         pass_manager class, and initialize the pass_lists array.
2319         (check_profile_consistency): Update for moves of globals to
2320         fields of pass_manager.
2321         (dump_profile_report): Move body of implementation into...
2322         (pass_manager::dump_profile_report): ...here.
2323         (ipa_write_summaries_1): Update for moves of pass lists from
2324         being globals to fields of pass_manager.
2325         (ipa_write_optimization_summaries): Likewise.
2326         (ipa_read_summaries):  Likewise.
2327         (ipa_read_optimization_summaries): Likewise.
2328         (execute_all_ipa_stmt_fixups): Likewise.
2330         * statistics.c (statistics_fini): Update for moves of globals to
2331         fields of pass_manager.
2333         * toplev.c (general_init): Replace call to
2334         init_optimization_passes with construction of the pass_manager
2335         instance.
2337         * tree-pass.h (all_passes): Remove, in favor of a field of the
2338         same name within the new class pass_manager.
2339         (all_small_ipa_passes): Likewise.
2340         (all_lowering_passes): Likewise.
2341         (all_regular_ipa_passes): Likewise.
2342         (all_lto_gen_passes): Likewise.
2343         (all_late_ipa_passes): Likewise.
2344         (passes_by_id): Likewise.
2345         (passes_by_id_size): Likewise.
2346         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2347         the new class pass_manager.
2348         (get_pass_for_id): Remove.
2350 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2352         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2353         configs.
2355 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
2357         * arm.md (mulhi3): New expand pattern.
2359 2013-07-30  Jan Hubicka  <jh@suse.cz>
2360             Martin Liska  <marxin.liska@gmail.com>
2362         * profile.c (compute_value_histograms): Do not ICE when
2363         there is mismatch only on some counters.
2365 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2367         PR rtl-optimization/57637
2368         * function.c (move_insn_for_shrink_wrap): Also check the
2369         GEN set of the LIVE problem for the liveness analysis
2370         if it exists, otherwise give up.
2372 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2374         PR tree-optimization/57993
2375         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2376         replaced statement in the candidate table.
2377         (phi_add_costs): Return infinite cost when the hidden basis does
2378         not dominate all phis on which the candidate is dependent.
2379         (replace_one_candidate): Record replaced statement in the
2380         candidate table.
2382 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
2384         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2385         (ashlv2si3): New expander.
2386         (*ashlv2si3_i): New define_insn_and_split.
2387         * predicates.md (float_operation): Allow patterns with three
2388         basic sub-patterns.
2390         PR rtl-optimization/58021
2391         * mode-switching.c (create_pre_exit): Always split off preceding
2392         insns if we are not at the basic block head.
2394 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
2396         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2397         (UCLIBC_DYNAMIC_LINKER): New macro.
2398         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2399         `-mnan=2008'.
2400         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2401         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
2402         `-mnan=2008'.
2403         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2404         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2405         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2406         for SF and DF modes.  Use ieee_quad_format for TF mode.
2407         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2408         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2409         (mips_option_override): Handle `-mnan=legacy'.
2410         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2411         `-mabs=2008' and `-mnan=2008'.
2412         (OPTION_DEFAULT_SPECS): Add "nan" default.
2413         (ASM_SPEC): Handle `-mnan='.
2414         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2415         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2416         comment accordingly.
2417         (neg<mode>2): Likewise.
2418         * config/mips/mips.opt (mabs, mnan): New options.
2419         * doc/install.texi (Configuration): Document `--with-nan=' option.
2420         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2421         `-mnan=' options.
2422         (MIPS Options): Document them.
2423         * config.gcc <mips*-*-*>: Handle `--with-nan='.
2424         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2425         * configure: Regenerate.
2426         * config.in: Regenerate.
2428 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2430         * config/i386/i386.md (float post-reload splitters): Do not check
2431         for subregs of SSE registers.
2433 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
2434             H.J. Lu  <hongjiu.lu@intel.com>
2436         PR target/57954
2437         PR target/57988
2438         * config/i386/i386.md (post-reload splitter
2439         to avoid partial SSE reg dependency stalls): New pattern.
2441 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2443         * config/s390/s390.md ("movcc"): Swap load and store instructions.
2445 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
2447         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2448         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2450 2013-07-26  Cary Coutant  <ccoutant@google.com>
2452         * dwarf2out.c (die_checksum_ordered): Don't include template
2453         instantiations in signature.
2454         (is_template_parameter): New function.
2455         (is_template_instantiation): New function.
2456         (generate_skeleton_bottom_up): Don't include template instantiations
2457         in type unit DIE.
2458         (generate_skeleton): Likewise.
2459         (break_out_comdat_types): Move recursive call to break out nested
2460         types earlier.
2461         (prune_unused_types_mark_generic_parms_dies): Call
2462         is_template_parameter.
2464 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
2466         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2467         uses vector registers.
2468         * config/aarch64/iterators.md: Add attributes rtn and vas.
2470 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2471             Richard Earnshaw  <richard.earnshaw@arm.com>
2473         * combine.c (simplify_comparison): Re-canonicalize operands
2474         where appropriate.
2475         * config/arm/arm.md (movcond_addsi): New splitter.
2477 2013-07-25  Sterling Augustine  <saugustine@google.com>
2479         * dwarf2out.c (size_of_pubnames): Move code to...
2480         (include_pubname_in_output): ...here.  New.
2481         (want_pubnames): Rearrange.
2482         (output_pubnames): Call include_pubname_in_output.  Move assertion.
2484 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2486         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2488 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
2490         PR target/38836
2491         * doc/extend.texi: Remove obsolete builtins. Fix
2492         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2494 2013-07-25  Jan Hubicka  <jh@suse.cz>
2496         * cgraph.c (release_function_body): Break out from ...
2497         (cgraph_release_function_body): ... this one; also release DECL_RESULT
2498         and DECL_ARGUMENTS.
2499         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2500         old_tree in the map.
2501         (create_specialized_node): Update.
2502         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2503         into index.
2504         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2505         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2506         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2507         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2508         DECL_RESULT.
2510 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2512         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2513         addsi3_carryin_alt2_<optab>): Correct output template.
2515 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2517         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2518         Adjust for arm_restrict_it.
2519         Remove trailing whitespace.
2521 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
2523         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2524         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2526         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2528 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
2530         PR rtl-optimization/57960
2531         * lra-constraints.c (process_alt_operands): Use the right mode
2532         when checking strict_low.
2534 2013-07-25  Jan Hubicka  <jh@suse.cz>
2536         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2537         * cgraph.c (cgraph_remove_node): Do not release function body
2538         when in cgraph streaming.
2539         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2540         in other partitions are not considered reachable; fix handling of
2541         clones.
2543 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2545         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2547 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2549         PR target/19599
2550         PR target/57731
2551         PR target/57837
2552         * config/arm/arm.md ("*sibcall_insn): Replace use of
2553         Ss with US. Adjust output for v5 and v4t.
2554         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2556         * config/arm/constraints.md ("Ss"): Rename to US.
2558 2013-07-25  Terry Guo  <terry.guo@arm.com>
2560         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2561         shift_add/shift_sub0/shift_sub1 RTXs.
2563 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
2564             Anton Blanchard <anton@au1.ibm.com>
2566         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2567         (altivec_vpks<VI_char>ss): Likewise.
2568         (altivec_vpks<VI_char>us): Likewise.
2569         (altivec_vpku<VI_char>us): Likewise.
2570         (altivec_vpku<VI_char>um): Likewise.
2572 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
2574         Introduce context class.
2576         * Makefile.in (CONTEXT_H): New.
2577         (OBJS): Add context.o.
2578         (toplev.o): Add CONTEXT_H to dependencies.
2579         (context.o): New.
2581         * toplev.c (general_init): Create the singleton gcc::context instance.
2583         * context.c: New.
2585         * context.h: New.
2587 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2589         PR rtl-optimization/57968
2590         * mode-switching.c (create_pre_exit): Allow instructions that
2591         don't set a return register to need a non-exit mode.
2593 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2594             Anton Blanchard <anton@au1.ibm.com>
2596         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2597         operands to vperm for little endian.
2598         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2599         of lvsl to create the control mask for a vperm for little endian.
2601 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2602             Anton Blanchard <anton@au1.ibm.com>
2604         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2605         two operands for little-endian.
2607 2013-07-23  Steve Ellcey  <sellcey@mips.com>
2609         * config/mips/mips.c (mips_case_values_threshold): New.
2610         (TARGET_CASE_VALUES_THRESHOLD): Define.
2612 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2613             Anton Blanchard <anton@au1.ibm.com>
2615         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2616         selection of field for vector splat in little endian mode.
2618 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
2620         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2621         expanders to rs6000.md.
2622         (ior<mode>3): Likewise.
2623         (and<mode>3): Likewise.
2624         (one_cmpl<mode>2): Likewise.
2625         (nor<mode>3): Likewise.
2626         (andc<mode>3): Likewise.
2627         (eqv<mode>3): Likewise.
2628         (nand<mode>3): Likewise.
2629         (orc<mode>3): Likewise.
2631         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2632         declaration.
2634         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2635         to split multi-word logical operations.
2636         (rs6000_split_logical_di): Likewise.
2637         (rs6000_split_logical): Likewise.
2639         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2640         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2641         and allow TImode operations in 32-bit.
2642         (vsx_and<mode>3_64bit): Likewise.
2643         (vsx_ior<mode>3_32bit): Likewise.
2644         (vsx_ior<mode>3_64bit): Likewise.
2645         (vsx_xor<mode>3_32bit): Likewise.
2646         (vsx_xor<mode>3_64bit): Likewise.
2647         (vsx_one_cmpl<mode>2_32bit): Likewise.
2648         (vsx_one_cmpl<mode>2_64bit): Likewise.
2649         (vsx_nor<mode>3_32bit): Likewise.
2650         (vsx_nor<mode>3_64bit): Likewise.
2651         (vsx_andc<mode>3_32bit): Likewise.
2652         (vsx_andc<mode>3_64bit): Likewise.
2653         (vsx_eqv<mode>3_32bit): Likewise.
2654         (vsx_eqv<mode>3_64bit): Likewise.
2655         (vsx_nand<mode>3_32bit): Likewise.
2656         (vsx_nand<mode>3_64bit): Likewise.
2657         (vsx_orc<mode>3_32bit): Likewise.
2658         (vsx_orc<mode>3_64bit): Likewise.
2660         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2661         logical types in GPRs.
2663         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2664         logical insns to rs6000.md, and allow TImode operations in
2665         32-bit.
2666         (altivec_ior<mode>3): Likewise.
2667         (altivec_xor<mode>3): Likewise.
2668         (altivec_one_cmpl<mode>2): Likewise.
2669         (altivec_nor<mode>3): Likewise.
2670         (altivec_andc<mode>3): Likewise.
2672         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2673         attributes for moving the 128-bit logical operations into
2674         rs6000.md.
2675         (BOOL_REGS_OUTPUT): Likewise.
2676         (BOOL_REGS_OP1): Likewise.
2677         (BOOL_REGS_OP2): Likewise.
2678         (BOOL_REGS_UNARY): Likewise.
2679         (BOOL_REGS_AND_CR0): Likewise.
2680         (one_cmpl<mode>2): Add support for DI logical operations on
2681         32-bit, splitting the operations to 32-bit.
2682         (anddi3): Likewise.
2683         (iordi3): Likewise.
2684         (xordi3): Likewise.
2685         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2686         changes to combine the 32/64-bit code, allow logical operations on
2687         TI mode in 32-bit, and to use similar match_operator patterns like
2688         scalar mode uses.  Combine the Altivec and VSX code for logical
2689         operations, and move it here.
2690         (ior<mode>3, 128-bit types): Likewise.
2691         (xor<mode>3, 128-bit types): Likewise.
2692         (one_cmpl<mode>3, 128-bit types): Likewise.
2693         (nor<mode>3, 128-bit types): Likewise.
2694         (andc<mode>3, 128-bit types): Likewise.
2695         (eqv<mode>3, 128-bit types): Likewise.
2696         (nand<mode>3, 128-bit types): Likewise.
2697         (orc<mode>3, 128-bit types): Likewise.
2698         (and<mode>3_internal): Likewise.
2699         (bool<mode>3_internal): Likewise.
2700         (boolc<mode>3_internal1): Likewise.
2701         (boolc<mode>3_internal2): Likewise.
2702         (boolcc<mode>3_internal1): Likewise.
2703         (boolcc<mode>3_internal2): Likewise.
2704         (eqv<mode>3_internal1): Likewise.
2705         (eqv<mode>3_internal2): Likewise.
2706         (one_cmpl1<mode>3_internal): Likewise.
2708 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2710         * config/microblaze/microblaze.c (microblaze_expand_prologue):
2711         Rename flag_stack_usage to flag_stack_usage_info.
2713 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
2715         * config/microblaze/sync.md: New file.
2716         * config/microblaze/microblaze.md: Include sync.md
2717         * config/microblaze/microblaze.c: Add print_operand 'y'.
2718         * config/microblaze/constraints.md: Add memory_contraint
2719         'Q' which is a single register.
2721 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
2723         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2725 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2727         * reload.c (find_reloads): Exit loop once we find this operand
2728         cannot be reloaded somehow for this alternative.
2730         * reload.c (find_reloads): Exit loop once we find a hard register.
2732         * rtlanal.c (computed_jump_p): Exit loop once we find label
2733         reference is used.
2735         * i386.c (ix86_pad_returns): Exit loop after setting replace.
2737         * cfgloopmanip.c (remove_path): Exit loop after setting
2738         irred_invalidated.
2740         * gensupport.c (subst_dup): Avoid loop if code is not
2741         MATCH_DUP nor MATCH_OP_DUP.
2743 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
2745         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2747 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2749         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2750         true for SP_REGNUM if mode == ptr_mode.
2751         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2752         with value R0_REGNUM + 31.
2754 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2756         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2757         pad pointer-typed argument downward.
2759 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2761         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2762         and __ILP32__ when the ILP32 model is in use.
2764 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2766         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2767         (aarch64_load_symref_appropriately): In the case of
2768         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2769         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2770         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2771         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2772         if 'mode' doesn't equal to 'ptr_mode'.
2773         (aarch64_output_mi_thunk): Add an assertion on the alignment of
2774         'vcall_offset'; change to call aarch64_emit_move differently depending
2775         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2776         to calculate the upper bound of 'vcall_offset'.
2777         (aarch64_cannot_force_const_mem): Change to also return true if
2778         mode != ptr_mode.
2779         (aarch64_legitimize_reload_address): In the case of large
2780         displacements, add new local variable 'xmode' and an assertion
2781         based on it; change to use 'xmode' to generate the new rtx and
2782         reload.
2783         (aarch64_asm_trampoline_template): Change to generate the template
2784         differently depending on TARGET_ILP32 or not; change to use
2785         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2786         (aarch64_trampoline_size): Removed.
2787         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2788         and replace immediate literals with it.  Change to use 'ptr_mode'
2789         instead of 'DImode' and call convert_memory_address if the mode
2790         of 'fnaddr' doesn't equal to 'ptr_mode'.
2791         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2792         to output symbol.
2793         (aarch64_elf_asm_destructor): Likewise.
2794         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2795         on TARGET_ILP32 instead of aarch64_trampoline_size.
2796         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2797         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2798         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2799         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
2800         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2801         (storewb_pair<GPI:mode>_<P:mode>): ... this.
2802         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2803         depending on the value of 'mode'.
2804         (add_losym_<mode>): New.
2805         (ldr_got_small_<mode>): New, based on ldr_got_small.
2806         (ldr_got_small): Remove.
2807         (ldr_got_small_sidi): New.
2808         * config/aarch64/iterators.md (P): New.
2809         (PTR): Change to 'ptr_mode' in the condition.
2811 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
2813         * config.gcc (aarch64*-*-*): Support --with-abi.
2814         (aarch64*-*-elf): Support --with-multilib-list.
2815         (aarch64*-*-linux*): Likewise.
2816         (supported_defaults): Add abi to aarch64*-*-*.
2817         * configure.ac: Mention AArch64 for --with-multilib-list.
2818         * configure: Re-generated.
2819         * config/aarch64/biarchilp32.h: New file.
2820         * config/aarch64/biarchlp64.h: New file.
2821         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2822         (ABI_SPEC): Ditto.
2823         (MULTILIB_DEFAULTS): Ditto.
2824         (DRIVER_SELF_SPECS): Ditto.
2825         (ASM_SPEC): Update to also substitute -mabi.
2826         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2827         file whose name depends on -mabi= and -mbig-endian.
2828         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2829         TARGET_ILP32.
2830         (POINTER_SIZE): New define.
2831         (POINTERS_EXTEND_UNSIGNED): Ditto.
2832         (enum aarch64_abi_type): New enumeration tag.
2833         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2834         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2835         (TARGET_ILP32): New define.
2836         * config/aarch64/aarch64.opt (mabi): New.
2837         (aarch64_abi): New.
2838         (ilp32, lp64): New values for -mabi.
2839         * config/aarch64/t-aarch64 (comma): New define.
2840         (MULTILIB_OPTIONS): Ditto.
2841         (MULTILIB_DIRNAMES): Ditto.
2842         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2843         * doc/invoke.texi: Document -mabi for AArch64.
2845 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
2847         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
2849 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2850             Anton Blanchard <anton@au1.ibm.com>
2852         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
2853         endianness when selecting field to splat.
2855 2013-07-22  Eric Christopher  <echristo@gmail.com>
2857         * dwarf2out.c (die_odr_checksum): New function to use
2858         CHECKSUM_ macros and ULEB128 for DIE tag.
2859         (generate_type_signature): Use.
2861 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
2863         * config.gcc (sparc*-*-*): Accept leon3 processor.
2864         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
2865         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
2866         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
2867         * config/sparc/sparc.opt (enum processor_type): Add leon3.
2868         (mfix-ut699): Adjust comment.
2869         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
2870         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
2871         (CPP_CPU_SPEC): Likewise.
2872         (ASM_CPU_SPEC): Likewise.
2873         * config/sparc/sparc.c (leon3_cost): New constant.
2874         (sparc_option_override): Add leon3 support.
2875         (mem_ref): New function.
2876         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
2877         (sparc_do_work_around_errata): Look into the instruction in the delay
2878         slot and adjust accordingly.  Add fix for the data cache nullify issues
2879         of the UT699.  Change insertion position for the NOP.
2880         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
2881         (leon3_load): New reservation.
2882         (leon_store): Bump latency to 2.
2883         (grfpu): New automaton.
2884         (grfpu_alu): New unit.
2885         (grfpu_ds): Likewise.
2886         (leon_fp_alu): Adjust.
2887         (leon_fp_mult): Delete.
2888         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
2889         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
2890         * config/sparc/sparc.md (cpu): Add leon3.
2891         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
2892         (swapsi): Likewise.
2893         (atomic_test_and_set): Likewise.
2894         (ldstub): Likewise.
2896 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
2898         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
2899         default for R5900 targets.
2900         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
2901         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
2902         * config/mips/mips.c (mips_option_override): Report an error for
2903         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
2904         for -march=r5900 -mhard-float.
2906 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
2908         * df-problems.c (can_move_insns_across): Exit loop once we
2909         find a non-fixed, non-global register.
2911         * ipa-pure-const.c (propagate_nothrow): Exit loop after
2912         setting can_throw.
2914         * omega.c (omega_eliminate_red): Break after setting red_found.
2915         (omega_problem_has_red_equations): Similarly after setting found.
2916         (omega_query_variable): Similarly after setting coupled.
2918 2013-07-22  Marek Polacek  <polacek@redhat.com>
2920         * gimplify.c: Don't include gimple.h twice.
2922 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2924         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
2925         instead of TARGET_THUMB1.
2926         (Pz): New constraint.
2927         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
2928         encodings.
2929         (compare_negsi_si): Likewise.
2930         (compare_addsi2_op0): Likewise.
2931         (compare_addsi2_op1): Likewise.
2932         (addsi3_carryin_<optab>): Likewise.
2933         (addsi3_carryin_alt2_<optab>): Likewise.
2934         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
2935         for arm_restrict_it.
2936         (subsi3_carryin): Likewise.
2937         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
2938         (minmax_arithsi): Disable for arm_restrict_it.
2939         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
2940         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
2941         (satsi_<SAT:code>_shift): Likewise.
2942         (arm_shiftsi3): Add alternative for 16-bit encoding.
2943         (arm32_movhf): Disable for arm_restrict_it.
2944         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
2945         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
2947 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
2949         * config/arm/arm.md (attribute "insn"): Delete.
2950         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
2951         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
2952         (not_shiftsi): Update for attribute change.
2953         (not_shiftsi_compare0): Likewise.
2954         (not_shiftsi_compare0_scratch): Likewise.
2955         (arm_one_cmplsi2): Likewise.
2956         (thumb1_one_cmplsi2): Likewise.
2957         (notsi_compare0): Likewise.
2958         (notsi_compare0_scratch): Likewise.
2959         (thumb1_movdi_insn): Likewise.
2960         (arm_movsi_insn): Likewise.
2961         (movhi_insn_arch4): Likewise.
2962         (movhi_bytes): Likewise.
2963         (arm_movqi_insn): Likewise.
2964         (thumb1_movqi_insn): Likewise.
2965         (arm32_movhf): Likewise.
2966         (thumb1_movhf): Likewise.
2967         (arm_movsf_soft_insn): Likewise.
2968         (thumb1_movsf_insn): Likewise.
2969         (thumb_movdf_insn): Likewise.
2970         (movsicc_insn): Likewise.
2971         (movsfcc_soft_insn): Likewise.
2972         (and_scc): Likewise.
2973         (cond_move): Likewise.
2974         (if_move_not): Likewise.
2975         (if_not_move): Likewise.
2976         (if_shift_move): Likewise.
2977         (if_move_shift): Likewise.
2978         (if_shift_shift): Likewise.
2979         (if_not_arith): Likewise.
2980         (if_arith_not): Likewise.
2981         (cond_move_not): Likewise.
2982         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
2983         (neon_mov<mode>): Likewise.
2984         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
2985         (thumb2_movsi_vfp): Likewise.
2986         (movsf_vfp): Likewise.
2987         (thumb2_movsf_vfp): Likewise.
2988         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
2989         change.
2990         (cortexa7_older_only): Likewise.
2991         (cortexa7_younger): Likewise.
2992         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
2993         (1020alu_shift_op): Likewise.
2994         (1020alu_shift_reg_op): Likewise.
2995         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
2996         (alu_shift_op): Likewise.
2997         (alu_shift_reg_op): Likewise.
2998         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
2999         (11_alu_shift_op): Likewise.
3000         (11_alu_shift_reg_op): Likewise.
3001         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3002         (9_alu_shift_reg_op): Likewise.
3003         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3004         change.
3005         (cortex_a15_alu_shift): Likewise.
3006         (cortex_a15_alu_shift_reg): Likewise.
3007         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3008         (cortex_a5_alu_shift): Likewise.
3009         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3010         change.
3011         (cortex_a53_alu_shift): Likewise.
3012         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3013         change.
3014         (cortex_a7_alu_reg): Likewise.
3015         (cortex_a7_alu_shift): Likewise.
3016         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3017         (cortex_a8_alu_shift): Likewise.
3018         (cortex_a8_alu_shift_reg): Likewise.
3019         (cortex_a8_mov): Likewise.
3020         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3021         (cortex_a9_dp_shift): Likewise.
3022         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3023         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3024         (cortex_r4_mov): Likewise.
3025         (cortex_r4_alu_shift): Likewise.
3026         (cortex_r4_alu_shift_reg): Likewise.
3027         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3028         (526_alu_shift_op): Likewise.
3029         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3030         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3031         (626te_alu_shift_op): Likewise.
3032         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3033         (726te_alu_op): Likewise.
3034         (726te_alu_shift_op): Likewise.
3035         (726te_alu_shift_reg_op): Likewise.
3036         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3037         (mp626_alu_shift_op): Likewise.
3038         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3039         (pj4_alu_e1_conds): Likewise.
3040         (pj4_alu): Likewise.
3041         (pj4_alu_conds): Likewise.
3042         (pj4_shift): Likewise.
3043         (pj4_shift_conds): Likewise.
3044         (pj4_alu_shift): Likewise.
3045         (pj4_alu_shift_conds): Likewise.
3047 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3049         * config/arm/predicates.md (shiftable_operator_strict_it):
3050         New predicate.
3051         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3052         Disable cond_exec version for arm_restrict_it.
3053         (thumb2_smaxsi3): Convert to generate cond_exec.
3054         (thumb2_sminsi3): Likewise.
3055         (thumb32_umaxsi3): Likewise.
3056         (thumb2_uminsi3): Likewise.
3057         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3058         (thumb2_neg_abssi2): Likewise.
3059         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3060         (thumb2_movsicc_insn): Adjust alternatives.
3061         (thumb2_mov_negscc): Disable for arm_restrict_it.
3062         (thumb2_mov_negscc_strict_it): New pattern.
3063         (thumb2_mov_notscc_strict_it): New pattern.
3064         (thumb2_mov_notscc): Disable for arm_restrict_it.
3065         (thumb2_ior_scc): Likewise.
3066         (thumb2_ior_scc_strict_it): New pattern.
3067         (thumb2_cond_move): Adjust for arm_restrict_it.
3068         (thumb2_cond_arith): Disable for arm_restrict_it.
3069         (thumb2_cond_arith_strict_it): New pattern.
3070         (thumb2_cond_sub): Adjust for arm_restrict_it.
3071         (thumb2_movcond): Likewise.
3072         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3073         (thumb2_zero_extendhisi2_v6): Likewise.
3074         (thumb2_zero_extendqisi2_v6): Likewise.
3075         (orsi_notsi_si): Likewise.
3076         (orsi_not_shiftsi_si): Likewise.
3078 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3080         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3081         instruction sequence is 1 byte shorter.
3083 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3085         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3086         it is not needed after split.
3088 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3090         PR target/51784
3091         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3092         second label for nonlocal goto receivers. Don't output pic base labels
3093         unless we're producing PIC; mark that action unreachable().
3094         (ix86_save_reg): If the function contains a nonlocal label, save the
3095         PIC base reg.
3096         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3097         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3098         (update_pic_label_number_if_needed): New.
3099         (machopic_output_function_base_name): Adjust for nonlocal receiver
3100         case.
3101         (machopic_should_output_picbase_label): New.
3102         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3103         (nonlocal_goto_receiver): New insn and split.
3105 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3107         * config/aarch64/aarch64-builtins.c
3108         (aarch64_fold_builtin): Fold abs in all modes.
3109         * config/aarch64/aarch64-simd-builtins.def
3110         (abs): Enable for all modes.
3111         * config/aarch64/arm_neon.h
3112         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3113         (vabs_f64): Add missing intrinsic.
3115 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
3117         * config/aarch64/arm_neon.h (vabs_s64): New function
3119 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
3121         PR target/57516
3122         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3123         * config/avr/avr.md (adjust_len): Add `round'.
3124         * config/avr/avr-protos.h (avr_out_round): New prototype.
3125         (avr_out_plus): Add `out_label' argument.
3126         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3127         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3128         Handle the case where `insn' is just a pattern.
3129         (avr_out_bitop): Handle the case where `insn' is just a pattern.
3130         (avr_out_round): New function.
3131         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3133 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
3135         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3136         Add check for flag_stack_usage to handle -fstack-usage support
3138 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
3140         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3141         interaction for new Power8 flags and VSX.
3143 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
3145         PR middle-end/57698
3146         * tree-inline.c (expand_call_inline): Emit errors during
3147         early_inlining only if optimization is not turned on.
3149 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3151         * passes.def: New.
3153         * passes.c (init_optimization_passes): Move the construction of
3154         the pass hierarchy into a new passes.def file.
3156         * Makefile.in (passes.o): Add dependency on passes.def.
3158 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3160         * passes.c (init_optimization_passes): Introduce macros for
3161         constructing the tree of passes (INSERT_PASSES_AFTER,
3162         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3163         TERMINATE_PASS_LIST).
3165 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
3166             Wei Mi  <wmi@google.com>
3168         PR rtl-optimization/57878
3169         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3170         top.
3171         (reload_pseudo_compare_func): Check nregs first for reload
3172         pseudos.
3174 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3176         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3178 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
3180         * read-rtl.c (validate_const_int): Once an invalid character is
3181         seen, quit the loop.
3183         * gengtype.c (write_roots): Similarly once we find the "deletable"
3184         or "if_marked" option.
3186 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3188         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3189         "xtab" and "sat".  Move value "clz" from here to ...
3190         (attriubte "type"): ... here.
3191         (satsi_<SAT:code>): Delete "insn" attribute.
3192         (satsi_<SAT:code>_shift): Likewise.
3193         (arm_zero_extendqisi2addsi): Likewise.
3194         (arm_extendqisi2addsi): Likewise.
3195         (clzsi2): Update for attribute changes.
3196         (rbitsi2): Likewise.
3197         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3198         attribute.
3199         (arm_usatsihi): Likewise.
3200         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3202 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
3204         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3205         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
3206         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
3207         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
3208         in alphabetical order.
3209         (attribute "core_cycles"): Update for attribute changes.
3210         (arm_addsi3): Likewise.
3211         (addsi3_compare0): Likewise.
3212         (addsi3_compare0_scratch): Likewise.
3213         (addsi3_compare_op1): Likewise.
3214         (addsi3_compare_op2): Likewise.
3215         (compare_addsi2_op0): Likewise.
3216         (compare_addsi2_op1): Likewise.
3217         (addsi3_carryin_shift_<optab>): Likewise.
3218         (subsi3_carryin_shift): Likewise.
3219         (rsbsi3_carryin_shift): Likewise.
3220         (arm_subsi3_insn): Likewise.
3221         (subsi3_compare0): Likewise.
3222         (subsi3_compare): Likewise.
3223         (arm_andsi3_insn): Likewise.
3224         (thumb1_andsi3_insn): Likewise.
3225         (andsi3_compare0): Likewise.
3226         (andsi3_compare0_scratch): Likewise.
3227         (zeroextractsi_compare0_scratch
3228         (andsi_not_shiftsi_si): Likewise.
3229         (iorsi3_insn): Likewise.
3230         (iorsi3_compare0): Likewise.
3231         (iorsi3_compare0_scratch): Likewise.
3232         (arm_xorsi3): Likewise.
3233         (thumb1_xorsi3_insn): Likewise.
3234         (xorsi3_compare0): Likewise.
3235         (xorsi3_compare0_scratch): Likewise.
3236         (satsi_<SAT:code>_shift): Likewise.
3237         (rrx): Likewise.
3238         (arm_shiftsi3): Likewise.
3239         (shiftsi3_compare0): Likewise.
3240         (not_shiftsi): Likewise.
3241         (not_shiftsi_compare0): Likewise.
3242         (not_shiftsi_compare0_scratch): Likewise.
3243         (arm_one_cmplsi2): Likewise.
3244         (thumb_one_complsi2): Likewise.
3245         (notsi_compare0): Likewise.
3246         (notsi_compare0_scratch): Likewise.
3247         (thumb1_zero_extendhisi2): Likewise.
3248         (arm_zero_extendhisi2): Likewise.
3249         (arm_zero_extendhisi2_v6): Likewise.
3250         (arm_zero_extendhisi2addsi): Likewise.
3251         (thumb1_zero_extendqisi2): Likewise.
3252         (thumb1_zero_extendqisi2_v6): Likewise.
3253         (arm_zero_extendqisi2): Likewise.
3254         (arm_zero_extendqisi2_v6): Likewise.
3255         (arm_zero_extendqisi2addsi): Likewise.
3256         (thumb1_extendhisi2): Likewise.
3257         (arm_extendhisi2): Likewise.
3258         (arm_extendhisi2_v6): Likewise.
3259         (arm_extendqisi): Likewise.
3260         (arm_extendqisi_v6): Likewise.
3261         (arm_extendqisi2addsi): Likewise.
3262         (thumb1_extendqisi2): Likewise.
3263         (thumb1_movdi_insn): Likewise.
3264         (arm_movsi_insn): Likewise.
3265         (movsi_compare0): Likewise.
3266         (movhi_insn_arch4): Likewise.
3267         (movhi_bytes): Likewise.
3268         (arm_movqi_insn): Likewise.
3269         (thumb1_movqi_insn): Likewise.
3270         (arm32_movhf): Likewise.
3271         (thumb1_movhf): Likewise.
3272         (arm_movsf_soft_insn): Likewise.
3273         (thumb1_movsf_insn): Likewise.
3274         (movdf_soft_insn): Likewise.
3275         (thumb_movdf_insn): Likewise.
3276         (arm_cmpsi_insn): Likewise.
3277         (cmpsi_shiftsi): Likewise.
3278         (cmpsi_shiftsi_swp): Likewise.
3279         (arm_cmpsi_negshiftsi_si): Likewise.
3280         (movsicc_insn): Likewise.
3281         (movsfcc_soft_insn): Likewise.
3282         (arith_shiftsi): Likewise.
3283         (arith_shiftsi_compare0
3284         (arith_shiftsi_compare0_scratch
3285         (sub_shiftsi): Likewise.
3286         (sub_shiftsi_compare0
3287         (sub_shiftsi_compare0_scratch
3288         (and_scc): Likewise.
3289         (cond_move): Likewise.
3290         (if_plus_move): Likewise.
3291         (if_move_plus): Likewise.
3292         (if_move_not): Likewise.
3293         (if_not_move): Likewise.
3294         (if_shift_move): Likewise.
3295         (if_move_shift): Likewise.
3296         (if_shift_shift): Likewise.
3297         (if_not_arith): Likewise.
3298         (if_arith_not): Likewise.
3299         (cond_move_not): Likewise.
3300         (thumb1_ashlsi3): Set type attribute.
3301         (thumb1_ashrsi3): Likewise.
3302         (thumb1_lshrsi3): Likewise.
3303         (thumb1_rotrsi3): Likewise.
3304         (shiftsi3_compare0_scratch): Likewise.
3305         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3306         (neon_mov<mode>): Likewise.
3307         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3308         attribute changes.
3309         (thumb2_movsi_insn): Likewise.
3310         (thumb2_cmpsi_neg_shiftsi): Likewise.
3311         (thumb2_extendqisi_v6): Likewise.
3312         (thumb2_zero_extendhisi2_v6): Likewise.
3313         (thumb2_zero_extendqisi2_v6): Likewise.
3314         (thumb2_shiftsi3_short): Likewise.
3315         (thumb2_addsi3_compare0_scratch): Likewise.
3316         (orsi_not_shiftsi_si): Likewise.
3317         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3318         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3319         changes.
3320         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3321         (1020alu_shift_op): Likewise.
3322         (1020alu_shift_reg_op): Likewise.
3323         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3324         (alu_shift_op): Likewise.
3325         (alu_shift_reg_op): Likewise.
3326         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3327         (11_alu_shift_op): Likewise.
3328         (11_alu_shift_reg_op): Likewise.
3329         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3330         (9_alu_shift_reg_op): Likewise.
3331         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3332         changes.
3333         (cortex_a15_alu_shift): Likewise.
3334         (cortex_a15_alu_shift_reg): Likewise.
3335         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3336         changes.
3337         (cortex_a5_alu_shift): Likewise.
3338         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3339         changes.
3340         (cortex_a53_alu_shift): Likewise.
3341         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3342         changes.
3343         (cortex_a7_alu_reg): Likewise.
3344         (cortex_a7_alu_shift): Likewise.
3345         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3346         changes.
3347         (cortex_a8_alu_shift): Likewise.
3348         (cortex_a8_alu_shift_reg): Likewise.
3349         (cortex_a8_mov): Likewise.
3350         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3351         (cortex_a9_dp_shift): Likewise.
3352         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3353         changes.
3354         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3355         changes.
3356         (cortex_r4_mov): Likewise.
3357         (cortex_r4_alu_shift): Likewise.
3358         (cortex_r4_alu_shift_reg): Likewise.
3359         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3360         (526_alu_shift_op): Likewise.
3361         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3362         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3363         (626te_alu_shift_op): Likewise.
3364         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3365         (726te_alu_op): Likewise.
3366         (726te_alu_shift_op): Likewise.
3367         (726te_alu_shift_reg_op): Likewise.
3368         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3369         (mp626_alu_shift_op): Likewise.
3370         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3371         (pj4_alu_e1_conds): Likewise.
3372         (pj4_alu): Likewise.
3373         (pj4_alu_conds): Likewise.
3374         (pj4_shift): Likewise.
3375         (pj4_shift_conds): Likewise.
3376         (pj4_alu_shift): Likewise.
3377         (pj4_alu_shift_conds): Likewise.
3378         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3379         changes.
3380         (cortexa7_older_only): Likewise.
3381         (cortexa7_younger): Likewise.
3383 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
3385         * ipa-pure-const.c (generate_summary): Rename to...
3386         (pure_const_generate_summary): ... this.
3388 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
3390         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3392 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
3394         PR target/57909
3395         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3396         usage in HI mode.
3398 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3400         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3401         enabled without -march=zEC12.
3402         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3403         flags to be set.
3405 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3407         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3408         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3409         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3410         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3411         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3412         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3413         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3414         ISA_HAS_FP4.
3415         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3416         and ISA_HAS_NMADD3_NMSUB3.
3417         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3418         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3419         (nmsub4<mode>, nmsub3<mode>): Likewise.
3420         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3422 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3424         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3425         TARGET_MIPS5400 checking.
3427 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
3428             Peter Bergner  <bergner@vnet.ibm.com>
3430         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3431         registers in the comment.
3432         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3433         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3434         rather than FIRST_PSEUDO_REGISTERS.
3436 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
3438         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3439         enable extra ISA flags with TARGET_HTM.
3441 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
3443         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3444         Fix comment typos.
3446 2013-07-15  Cong Hou  <congh@google.com>
3448         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3449         in compare function for sorting.
3451 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
3453         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3454         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3455         * config/rs6000/rs6000.opt: Add -mhtm option.
3456         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3457         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3458         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3459         __HTM__ if the HTM instructions are available.
3460         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3461         htm_spr_reg_operand): New define_predicates.
3462         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3463         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3464         Include htm.md.
3465         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3466         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3467         HTM builtin functions.
3468         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3469         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3470         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3471         (rs6000_builtin_mask_calculate): Likewise.
3472         (rs6000_option_override_internal): Likewise.
3473         (bdesc_htm): Add new HTM builtin support.
3474         (htm_spr_num): New function.
3475         (htm_spr_regno): Likewise.
3476         (rs6000_htm_spr_icode): Likewise.
3477         (htm_expand_builtin): Likewise.
3478         (htm_init_builtins): Likewise.
3479         (rs6000_expand_builtin): Add support for HTM builtin functions.
3480         (rs6000_init_builtins): Likewise.
3481         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3482         option.
3483         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3484         (TARGET_HTM, MASK_HTM): Define macros.
3485         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3486         (FIXED_REGISTERS): Likewise.
3487         (CALL_USED_REGISTERS): Likewise.
3488         (CALL_REALLY_USED_REGISTERS): Likewise.
3489         (REG_ALLOC_ORDER): Likewise.
3490         (enum reg_class): Likewise.
3491         (REG_CLASS_NAMES): Likewise.
3492         (REG_CLASS_CONTENTS): Likewise.
3493         (REGISTER_NAMES): Likewise.
3494         (ADDITIONAL_REGISTER_NAMES): Likewise.
3495         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3496         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3497         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3498         * config/rs6000/htm.md: New file.
3499         * config/rs6000/htmintrin.h: New file.
3500         * config/rs6000/htmxlintrin.h: New file.
3502 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3504         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3505         Define SYMBOL_TINY_GOT, update comment.
3506         * config/aarch64/aarch64.c
3507         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3508         (aarch64_expand_mov_immediate): Likewise.
3509         (aarch64_print_operand): Likewise.
3510         (aarch64_classify_symbol): Likewise.
3511         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3512         (ldr_got_tiny): Define.
3514 2013-07-13  Tobias Grosser  <tobias@grosser.es>
3516         PR tree-optimization/54094
3517         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3518           scheduling dimension for the parallelism check from the polyhedral
3519           information in the AST.
3520         * graphite-dependences.c (carries_deps): Do not assume the schedule is
3521           in 2D + 1 form.
3523 2013-07-13  Jason Merrill  <jason@redhat.com>
3525         * print-tree.c (debug_vec_tree): Use debug_raw.
3526         (debug_raw (vec<tree, va_gc> &)): New.
3527         (debug_raw (vec<tree, va_gc> *)): New.
3528         * tree.h: Declare them.
3530 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
3532         * ifcvt.c (ifcvt_after_combine): New static variable.
3533         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3534         for size.
3535         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
3536         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3537         rest_of_handle_if_after_reload): Pass new argument for if_convert.
3539 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
3541         * config/mips/mips.c (mips_expand_call): Remove empty statement.
3543 2013-07-12  Michael Matz  <matz@suse.de>
3545         PR middle-end/55771
3546         * convert.c (convert_to_real): Reject non-float inner types.
3548 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
3550         * config/aarch64/aarch64-protos.h
3551         (aarch64_simd_immediate_valid_for_move): Remove.
3552         * config/aarch64/aarch64.c (simd_immediate_info): New member.
3553         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3554         cases.
3555         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3557 2013-07-11  Steve Ellcey  <sellcey@mips.com>
3559         * config/mips/mips.c (mips_conditional_register_usage): Do not
3560         use t[0-7] registers in MIPS16 mode when optimizing for size.
3562 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
3564         * config/i386/i386.c (dispatch_function_versions): Fix array
3565         indexing of function_version_info to match actual_versions.
3567 2013-07-11  Teresa Johnson  <tejohnson@google.com>
3569         * vec.h (struct va_gc): Move release out-of-line.
3570         (va_gc::release): Call ggc_free on released vec.
3572 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3574         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3575         Require GOT register as additional operand in UNSPEC.
3576         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3577         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3578         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3579         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3580         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3581         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3582         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3584 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
3586         PR target/57631
3587         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3588         name seen by assembler/linker rather if available.
3590 2013-07-11  Andreas Schwab  <schwab@suse.de>
3592         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3594 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
3596         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
3598 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
3600         * doc/tm.texi.in: Move hook documentation to ....
3601         * target.def: ... here.
3603         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3604         text on @hook line.
3605         * doc/tm.texi: Regenerate.
3607 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
3609         PR c++/57869
3610         * doc/invoke.texi: Document -Wconditionally-supported.
3612 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3614         PR target/57844
3615         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3616         of my_fp.
3618 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3620         PR target/57506
3621         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3622         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3623         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3624         Remove duplicate devices.
3625         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3626         * config/avr/t-multilib: Regenerate.
3627         * config/avr/avr-tables.opt: Regenerate.
3628         * doc/avr-mmcu.texi: Regenerate.
3630 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
3632         PR target/56987
3633         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3635 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
3637         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3638         the cost of MULT when optimizing for size.
3640 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3642         * config/cr16/cr16-protos.h: Don't include target.h.
3644 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
3646         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3647         adjust register size for TDmode and TFmode for VSX registers.
3649 2013-07-08  Kai Tietz  <ktietz@redhat.com>
3651         PR target/56892
3652         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3653         hook_bool_const_tree_true.
3655 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3657         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3658         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3659         * config/s390/s390.md: Define FPR*_REGNUM constants.
3660         Fix FPR2_REGNUM constant (18 -> 17).
3661         ("*trunc<BFP:mode><DFP_ALL:mode>2")
3662         ("*trunc<DFP_ALL:mode><BFP:mode>2")
3663         ("trunc<BFP:mode><DFP_ALL:mode>2")
3664         ("trunc<DFP_ALL:mode><BFP:mode>2")
3665         ("*extend<BFP:mode><DFP_ALL:mode>2")
3666         ("*extend<DFP_ALL:mode><BFP:mode>2")
3667         ("extend<BFP:mode><DFP_ALL:mode>2")
3668         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3669         FPR4_REGNUM.
3671 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
3673         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
3675 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3677         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3678         and cfun_fpr_bit_p to cfun_fpr_save_p.
3679         (s390_frame_area, s390_register_info, s390_frame_info)
3680         (s390_emit_prologue, s390_emit_epilogue)
3681         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3682         register numbers.
3683         * config/s390/s390.h: Define *_REGNUM macros for floating point
3684         register numbers.
3686 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
3688         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3690 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
3692         PR rtl-optimization/57786
3693         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3694         and break out of the loop when it is set to false.
3696 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
3698         PR target/57819
3699         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3700         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3701         (const_int 63)) 0)).
3702         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3703         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3704         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3706         PR rtl-optimization/57829
3707         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3708         mask bits outside of mode are just sign-extension from mode to HWI.
3710 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
3712         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3713         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3714         adjust_address instead of change_address to keep info about alignment.
3715         (emit_strmov): Remove.
3716         (emit_memmov): New function.
3717         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3718         (expand_movmem_epilogue): Likewise and return updated rtx for
3719         destination.
3720         (expand_constant_movmem_prologue): Likewise and return updated rtx for
3721         destination and source.
3722         (decide_alignment): Refactor, handle vector_loop.
3723         (ix86_expand_movmem): Likewise.
3724         (ix86_expand_setmem): Likewise.
3725         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3727 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
3729         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3730         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3732 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3734         * config/i386/sse.md (sse_movlhps): Change alternative 3
3735         of operand 2 to "m".
3737 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
3739         PR target/57807
3740         * config/i386/sse.md (iptr): New mode attribute.
3741         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3742         (<sse>_vm<plusminus_insn><mode>3): Ditto.
3743         (<sse>_vmmul<mode>3): Ditto.
3744         (<sse>_vmdiv<mode>3): Ditto.
3745         (sse_vmrcpv4sf2): Ditto.
3746         (<sse>_vmsqrt<mode>2): Ditto.
3747         (sse_vmrsqrtv4sf2): Ditto.
3748         (<sse>_vm<code><mode>3): Ditto.
3749         (avx_vmcmp<mode>3): Ditto.
3750         (<sse>_vmmaskcmp<mode>3): Ditto.
3751         (<sse>_comi): Ditto.
3752         (<sse>_ucomi): Ditto.
3753         (*xop_vmfrcz_<mode>): Ditto.
3754         (*fmai_fmadd_<mode>): Ditto.
3755         (*fmai_fmsub_<mode>): Ditto.
3756         (*fmai_fnmadd_<mode>): Ditto.
3757         (*fmai_fnmsub_<mode>): Ditto.
3758         (*fma4i_vmfmadd_<mode>): Ditto.
3759         (*fma4i_vmfmsub_<mode>): Ditto.
3760         (*fma4i_vmfnmadd_<mode>): Ditto.
3761         (*fma4i_vmfnmsub_<mode>): Ditto.
3762         (*xop_vmfrcz_<mode>): Ditto.
3763         (sse_cvtps2pi): Ditto.
3764         (sse_cvttps2pi): Ditto.
3765         (sse_cvtss2si): Ditto.
3766         (sse_cvtss2si_2): Ditto.
3767         (sse_cvtss2siq_2): Ditto.
3768         (sse_cvttss2si): Ditto.
3769         (sse_cvttss2siq): Ditto.
3770         (sse_cvtsd2si): Ditto.
3771         (sse_cvtsd2si_2): Ditto.
3772         (sse_cvtsd2siq_2): Ditto.
3773         (sse_cvttsd2si): Ditto.
3774         (sse_cvttsd2siq): Ditto.
3775         (sse_cvtsd2ss): Ditto.
3776         (sse_cvtss2sd): Ditto.
3777         (avx2_pbroadcast<mode>): Ditto.
3778         (avx2_pbroadcast<mode>_1): Ditto.
3779         (*avx_vperm_broadcast_v4sf): Ditto.
3781         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3782         (sse_movlhps): Ditto.
3783         (sse_storehps): Ditto.
3784         (sse_loadhps): Ditto.
3785         (sse_storelps): Ditto.
3786         (sse_loadlps): Ditto.
3787         (*vec_concatv4sf): Ditto.
3788         (*vec_interleave_highv2df): Ditto.
3789         (*vec_interleave_lowv2df): Ditto.
3790         (*vec_extractv2df_1_sse): Ditto.
3791         (*vec_extractv2df_0_sse): Ditto.
3792         (sse2_storelpd): Ditto.
3793         (sse2_loadlpd): Ditto.
3794         (sse2_movsd): Ditto.
3795         (*vec_concatv4si): Ditto.
3796         (vec_concatv2di): Ditto.
3798         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3799         for Intel asm dialect.
3800         (mmx_punpcklwd): Ditto.
3801         (mmx_punpckldq): Ditto.
3803         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3804         for intel assembler dialect.
3806 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
3808         PR target/29776
3809         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3810         for BUILT_IN_C{LZ,LRSB}*.
3811         * tree.h (CASE_INT_FN): Add FN##IMAX case.
3812         * tree-vrp.c (extract_range_basic): Handle
3813         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
3814         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3815         fall thru to code calling set_value*.
3816         * builtins.c (expand_builtin): Remove *IMAX cases.
3817         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3818         if width is bigger than 2*HWI.
3820 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
3822         PR rtl-optimization/55342
3823         * lra-int.h (lra_subreg_reload_pseudos): New.
3824         * lra.c: Add undoing optional reloads to the block diagram.
3825         (lra_subreg_reload_pseudos): New.
3826         (lra_optional_reload_pseudos): Change comments.
3827         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
3828         lra_optional_reload_pseudos after undo transformations.
3829         * lra-assigns.c (pseudo_prefix_title): New.
3830         (lra_setup_reg_renumber): Use it.
3831         (spill_for): Ditto.  Check subreg reload pseudos too.
3832         (assign_by_spills): Consider subreg reload pseudos too.
3833         * lra-constraints.c (simplify_operand_subreg): Use
3834         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3835         (curr_insn_transform): Recognize and do optional reloads.
3836         (undo_optional_reloads): New.
3837         (lra_undo_inheritance): Call undo_optional_reloads.
3839 2013-07-05  Thomas Quinot  <quinot@adacore.com>
3841         * tree-complex.c (expand_complex_operations_1): Fix typo.
3843 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
3845         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3846         (tune_params): New member 'const vec_costs'.
3847         * config/aarch64/aarch64.c (generic_vector_cost): New.
3848         (generic_tunings): New member 'generic_vector_cost'.
3849         (aarch64_builtin_vectorization_cost): New.
3850         (aarch64_add_stmt_cost): New.
3851         (TARGET_VECTORIZE_ADD_STMT_COST): New.
3852         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
3854 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
3856         PR target/57777
3857         * config/i386/predicates.md (vsib_address_operand): Disallow
3858         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
3860 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
3862         PR middle-end/55030
3863         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
3864         expand_builtin_setjmp_receiver.
3865         (expand_label): Adjust, call expand_builtin_setjmp_receiver
3866         with NULL for the label parameter.
3867         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
3868         the frame-pointer.  Adjust comments.
3869         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
3870         only if LABEL is non-NULL.
3872 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
3874         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
3875         (ARM_ABI_AAPCS64): Ditto.
3876         (arm_abi): Ditto.
3877         (ARM_DEFAULT_ABI): Ditto.
3879 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
3881         * config/aarch64/aarch64-builtins.c
3882         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
3883         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
3884         (st1): Likewise.
3885         * config/aarch64/aarch64-simd.md
3886         (aarch64_ld1<VALL:mode>): New.
3887         (aarch64_st1<VALL:mode>): Likewise.
3888         * config/aarch64/arm_neon.h
3889         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
3891 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
3893         * config/i386/i386.c (gate_insert_vzeroupper): Check if
3894         target ISA is AVX.
3895         (ix86_option_override_internal):Turn on all -mavx target flags by
3896         default as they are dependent on AVX anyway.
3898 2013-07-02  Cary Coutant  <ccoutant@google.com>
3900         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
3901         deterministic hash.
3902         (loc_checksum_ordered): Likewise.
3903         (hash_loc_operands): Remove inline keyword.
3905 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
3907         PR tree-optimization/57741
3908         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
3909         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
3910         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
3911         Allow REAL_CST step_exprs if flag_associative_math.
3912         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
3914 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3916         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
3918 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
3920         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
3922 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3924         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
3925         encoding.
3926         (iorsi3_insn): Likewise.
3927         (arm_xorsi3): Likewise.
3929 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
3931         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
3932         to ...
3933         (attribute "type"): ... here, and prefix with "wmmx_".
3934         (attribute "core_cycles"): Update for attribute changes.
3935         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
3936         (tbcstv4hi): Likewise.
3937         (tbcstv2si): Likewise.
3938         (iwmmxt_iordi3): Likewise.
3939         (iwmmxt_xordi3): Likewise.
3940         (iwmmxt_anddi3): Likewise.
3941         (iwmmxt_nanddi3): Likewise.
3942         (iwmmxt_arm_movdi): Likewise.
3943         (iwmmxt_movsi_insn): Likewise.
3944         (mov<mode>_internal): Likewise.
3945         (and<mode>3_iwmmxt): Likewise.
3946         (ior<mode>3_iwmmxt): Likewise.
3947         (xor<mode>3_iwmmxt): Likewise.
3948         (add<mode>3_iwmmxt): Likewise.
3949         (ssaddv8qi3): Likewise.
3950         (ssaddv4hi3): Likewise.
3951         (ssaddv2si3): Likewise.
3952         (usaddv8qi3): Likewise.
3953         (usaddv4hi3): Likewise.
3954         (usaddv2si3): Likewise.
3955         (sub<mode>3_iwmmxt): Likewise.
3956         (sssubv8qi3): Likewise.
3957         (sssubv4hi3): Likewise.
3958         (sssubv2si3): Likewise.
3959         (ussubv8qi3): Likewise.
3960         (ussubv4hi3): Likewise.
3961         (ussubv2si3): Likewise.
3962         (mulv4hi3_iwmmxt): Likewise.
3963         (smulv4hi3_highpart): Likewise.
3964         (umulv4hi3_highpart): Likewise.
3965         (iwmmxt_wmacs): Likewise.
3966         (iwmmxt_wmacsz): Likewise.
3967         (iwmmxt_wmacu): Likewise.
3968         (iwmmxt_wmacuz): Likewise.
3969         (iwmmxt_clrdi): Likewise.
3970         (iwmmxt_clrv8qi): Likewise.
3971         (iwmmxt_clr4hi): Likewise.
3972         (iwmmxt_clr2si): Likewise.
3973         (iwmmxt_uavgrndv8qi3): Likewise.
3974         (iwmmxt_uavgrndv4hi3): Likewise.
3975         (iwmmxt_uavgv8qi3): Likewise.
3976         (iwmmxt_uavgv4hi3): Likewise.
3977         (iwmmxt_tinsrb): Likewise.
3978         (iwmmxt_tinsrh): Likewise.
3979         (iwmmxt_tinsrw): Likewise.
3980         (iwmmxt_textrmub): Likewise.
3981         (iwmmxt_textrmsb): Likewise.
3982         (iwmmxt_textrmuh): Likewise.
3983         (iwmmxt_textrmsh): Likewise.
3984         (iwmmxt_textrmw): Likewise.
3985         (iwmxxt_wshufh): Likewise.
3986         (eqv8qi3): Likewise.
3987         (eqv4hi3): Likewise.
3988         (eqv2si3): Likewise.
3989         (gtuv8qi3): Likewise.
3990         (gtuv4hi3): Likewise.
3991         (gtuv2si3): Likewise.
3992         (gtv8qi3): Likewise.
3993         (gtv4hi3): Likewise.
3994         (gtv2si3): Likewise.
3995         (smax<mode>3_iwmmxt): Likewise.
3996         (umax<mode>3_iwmmxt): Likewise.
3997         (smin<mode>3_iwmmxt): Likewise.
3998         (umin<mode>3_iwmmxt): Likewise.
3999         (iwmmxt_wpackhss): Likewise.
4000         (iwmmxt_wpackwss): Likewise.
4001         (iwmmxt_wpackdss): Likewise.
4002         (iwmmxt_wpackhus): Likewise.
4003         (iwmmxt_wpackwus): Likewise.
4004         (iwmmxt_wpackdus): Likewise.
4005         (iwmmxt_wunpckihb): Likewise.
4006         (iwmmxt_wunpckihh): Likewise.
4007         (iwmmxt_wunpckihw): Likewise.
4008         (iwmmxt_wunpckilb): Likewise.
4009         (iwmmxt_wunpckilh): Likewise.
4010         (iwmmxt_wunpckilw): Likewise.
4011         (iwmmxt_wunpckehub): Likewise.
4012         (iwmmxt_wunpckehuh): Likewise.
4013         (iwmmxt_wunpckehuw): Likewise.
4014         (iwmmxt_wunpckehsb): Likewise.
4015         (iwmmxt_wunpckehsh): Likewise.
4016         (iwmmxt_wunpckehsw): Likewise.
4017         (iwmmxt_wunpckelub): Likewise.
4018         (iwmmxt_wunpckeluh): Likewise.
4019         (iwmmxt_wunpckeluw): Likewise.
4020         (iwmmxt_wunpckelsb): Likewise.
4021         (iwmmxt_wunpckelsh): Likewise.
4022         (iwmmxt_wunpckelsw): Likewise.
4023         (ror<mode>3): Likewise.
4024         (ashr<mode>3_iwmmxt): Likewise.
4025         (lshr<mode>3_iwmmxt): Likewise.
4026         (ashl<mode>3_iwmmxt): Likewise.
4027         (ror<mode>3_di): Likewise.
4028         (ashr<mode>3_di): Likewise.
4029         (lshr<mode>3_di): Likewise.
4030         (ashl<mode>3_di): Likewise.
4031         (iwmmxt_wmadds): Likewise.
4032         (iwmmxt_wmaddu): Likewise.
4033         (iwmmxt_tmia): Likewise.
4034         (iwmmxt_tmiaph): Likewise.
4035         (iwmmxt_tmiabb): Likewise.
4036         (iwmmxt_tmiatb): Likewise.
4037         (iwmmxt_tmiabt): Likewise.
4038         (iwmmxt_tmiatt): Likewise.
4039         (iwmmxt_tmovmskb): Likewise.
4040         (iwmmxt_tmovmskh): Likewise.
4041         (iwmmxt_tmovmskw): Likewise.
4042         (iwmmxt_waccb): Likewise.
4043         (iwmmxt_wacch): Likewise.
4044         (iwmmxt_waccw): Likewise.
4045         (iwmmxt_waligni): Likewise.
4046         (iwmmxt_walignr): Likewise.
4047         (iwmmxt_walignr0): Likewise.
4048         (iwmmxt_walignr1): Likewise.
4049         (iwmmxt_walignr2): Likewise.
4050         (iwmmxt_walignr3): Likewise.
4051         (iwmmxt_wsadb): Likewise.
4052         (iwmmxt_wsadh): Likewise.
4053         (iwmmxt_wsadbz): Likewise.
4054         (iwmmxt_wsadhz): Likewise.
4055         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4056         (iwmmxt_wabsdiffb): Likewise.
4057         (iwmmxt_wabsdiffh): Likewise.
4058         (iwmmxt_wabsdiffw): Likewise.
4059         (iwmmxt_waddsubhx): Likewise
4060         (iwmmxt_wsubaddhx): Likewise.
4061         (addc<mode>3): Likewise.
4062         (iwmmxt_avg4): Likewise.
4063         (iwmmxt_avg4r): Likewise.
4064         (iwmmxt_wmaddsx): Likewise.
4065         (iwmmxt_wmaddux): Likewise.
4066         (iwmmxt_wmaddsn): Likewise.
4067         (iwmmxt_wmaddun): Likewise.
4068         (iwmmxt_wmulwsm): Likewise.
4069         (iwmmxt_wmulwum): Likewise.
4070         (iwmmxt_wmulsmr): Likewise.
4071         (iwmmxt_wmulumr): Likewise.
4072         (iwmmxt_wmulwsmr): Likewise.
4073         (iwmmxt_wmulwumr): Likewise.
4074         (iwmmxt_wmulwl): Likewise.
4075         (iwmmxt_wqmulm): Likewise.
4076         (iwmmxt_wqmulwm): Likewise.
4077         (iwmmxt_wqmulmr): Likewise.
4078         (iwmmxt_wqmulwmr): Likewise.
4079         (iwmmxt_waddbhusm): Likewise.
4080         (iwmmxt_waddbhusl): Likewise.
4081         (iwmmxt_wqmiabb): Likewise.
4082         (iwmmxt_wqmiabt): Likewise.
4083         (iwmmxt_wqmiatb): Likewise.
4084         (iwmmxt_wqmiatt): Likewise.
4085         (iwmmxt_wqmiabbn): Likewise.
4086         (iwmmxt_wqmiabtn): Likewise.
4087         (iwmmxt_wqmiatbn): Likewise.
4088         (iwmmxt_wqmiattn): Likewise.
4089         (iwmmxt_wmiabb): Likewise.
4090         (iwmmxt_wmiabt): Likewise.
4091         (iwmmxt_wmiatb): Likewise.
4092         (iwmmxt_wmiatt): Likewise.
4093         (iwmmxt_wmiabbn): Likewise.
4094         (iwmmxt_wmiabtn): Likewise.
4095         (iwmmxt_wmiatbn): Likewise.
4096         (iwmmxt_wmiattn): Likewise.
4097         (iwmmxt_wmiawbb): Likewise.
4098         (iwmmxt_wmiawbt): Likewise.
4099         (iwmmxt_wmiawtb): Likewise.
4100         (iwmmxt_wmiawtt): Likewise.
4101         (iwmmxt_wmiawbbn): Likewise.
4102         (iwmmxt_wmiawbtn): Likewise.
4103         (iwmmxt_wmiawtbn): Likewise.
4104         (iwmmxt_wmiawttn): Likewise.
4105         (iwmmxt_wmerge): Likewise.
4106         (iwmmxt_tandc<mode>3): Likewise.
4107         (iwmmxt_torc<mode>3): Likewise.
4108         (iwmmxt_torvsc<mode>3): Likewise.
4109         (iwmmxt_textrc<mode>3): Likewise.
4110         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4111         (wmmxt_pack): Likewise.
4112         (wmmxt_mult_c1): Likewise.
4113         (wmmxt_mult_c2): Likewise.
4114         (wmmxt_alu_c1): Likewise.
4115         (wmmxt_alu_c2): Likewise.
4116         (wmmxt_alu_c3): Likewise.
4117         (wmmxt_transfer_c1): Likewise.
4118         (wmmxt_transfer_c2): Likewise.
4119         (wmmxt_transfer_c3): Likewise.
4120         (marvell_f_iwmmxt_wstr): Likewise.
4121         (marvell_f_iwmmxt_wldr): Likewise.
4123 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
4125         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4127 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4129         Revert:
4130         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4131         * lra-constraints.c (need_for_split_p): Check call used hard regs
4132         living through calls.
4134         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4135         call used regs for call insn.
4137 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
4139         PR target/57736
4140         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4141         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4143 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
4145         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4147 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
4149         * lra-constraints.c (need_for_split_p): Check call used hard regs
4150         living through calls.
4152 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
4154         PR target/57744
4155         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4156         to tie with any other modes.  Eliminate Altivec vector mode tests,
4157         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
4158         code, to return 0 if testing MODE2 for a condition, if we've
4159         already tested MODE1 for the same condition.
4161 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4163         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4164         layout.
4166 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4168         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4169         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4171 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4173         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4174         Define.
4175         (aarch64_symbolic_constant_p): Remove.
4176         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4177         static.  Fix line length and white space.
4178         (aarch64_symbolic_constant_p): Remove.
4179         * config/aarch64/predicates.md (aarch64_valid_symref):
4180         Use aarch64_classify_symbol_expression.
4182 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4184         * config/arm/constraints.md (Ts): New constraint.
4185         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4186         16-bit encodings.
4187         (compare_scc): Use "Ts" constraint for operand 0.
4188         (ior_scc_scc): Likewise.
4189         (and_scc_scc): Likewise.
4190         (and_scc_scc_nodom): Likewise.
4191         (ior_scc_scc_cmp): Likewise for operand 7.
4192         (and_scc_scc_cmp): Likewise.
4193         * config/arm/thumb2.md (thumb2_movsi_insn):
4194         Add alternatives for 16-bit encodings.
4195         (thumb2_movhi_insn): Likewise.
4196         (thumb2_movsicc_insn): Likewise.
4197         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
4198         (thumb2_negscc): Use "Ts" constraint.
4199         Move mvn instruction outside cond_exec block.
4200         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4201         for 16-bit encodings.
4203 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4205         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4206         encoding.
4207         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4208         (mulsi3subsi): Likewise.
4209         (mulsidi3adddi): Likewise.
4210         (mulsidi3_v6): Likewise.
4211         (umulsidi3_v6): Likewise.
4212         (umulsidi3adddi_v6): Likewise.
4213         (smulsi3_highpart_v6): Likewise.
4214         (umulsi3_highpart_v6): Likewise.
4215         (mulhisi3tb): Likewise.
4216         (mulhisi3bt): Likewise.
4217         (mulhisi3tt): Likewise.
4218         (maddhisi4): Likewise.
4219         (maddhisi4tb): Likewise.
4220         (maddhisi4tt): Likewise.
4221         (maddhidi4): Likewise.
4222         (maddhidi4tb): Likewise.
4223         (maddhidi4tt): Likewise.
4224         (zeroextractsi_compare0_scratch): Likewise.
4225         (insv_zero): Likewise.
4226         (insv_t2): Likewise.
4227         (anddi_notzesidi_di): Likewise.
4228         (anddi_notsesidi_di): Likewise.
4229         (andsi_notsi_si): Likewise.
4230         (iordi_zesidi_di): Likewise.
4231         (xordi_zesidi_di): Likewise.
4232         (andsi_iorsi3_notsi): Likewise.
4233         (smax_0): Likewise.
4234         (smax_m1): Likewise.
4235         (smin_0): Likewise.
4236         (not_shiftsi): Likewise.
4237         (unaligned_loadsi): Likewise.
4238         (unaligned_loadhis): Likewise.
4239         (unaligned_loadhiu): Likewise.
4240         (unaligned_storesi): Likewise.
4241         (unaligned_storehi): Likewise.
4242         (extv_reg): Likewise.
4243         (extzv_t2): Likewise.
4244         (divsi3): Likewise.
4245         (udivsi3): Likewise.
4246         (arm_zero_extendhisi2addsi): Likewise.
4247         (arm_zero_extendqisi2addsi): Likewise.
4248         (compareqi_eq0): Likewise.
4249         (arm_extendhisi2_v6): Likewise.
4250         (arm_extendqisi2addsi): Likewise.
4251         (arm_movt): Likewise.
4252         (thumb2_ldrd): Likewise.
4253         (thumb2_ldrd_base): Likewise.
4254         (thumb2_ldrd_base_neg): Likewise.
4255         (thumb2_strd): Likewise.
4256         (thumb2_strd_base): Likewise.
4257         (thumb2_strd_base_neg): Likewise.
4258         (arm_negsi2): Add alternative for 16-bit encoding.
4259         (arm_one_cmplsi2): Likewise.
4261 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4263         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4264         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4265         (movdfcc): Likewise.
4266         * config/arm/vfp.md (*thumb2_movsf_vfp):
4267         Disable predication for arm_restrict_it.
4268         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4269         (*thumb2_movdfcc_vfp): Likewise.
4270         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4271         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4272         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4273         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4274         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4275         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4276         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4277         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4278         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4279         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4280         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4281         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4282         Disable predication for arm_restrict_it.
4284 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
4286         * config/i386/bmiintrin.h (_bextr_u32): New.
4287         (_bextr_u64): Ditto.
4289 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
4291         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4292         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4293         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4294         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4295         * config/mips/n32-elf.h: ...this new file.
4297 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
4299         PR target/57224
4300         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4301         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4303 2013-06-27  Catherine Moore <clm@codesourcery.com>
4305         * config/mips/mips-tables.opt: Regenerate.
4306         * config/mips/mips-cpus.def: Add m14ke and m14kec.
4307         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4308         * doc/invoke.texi: Add -m14kc.
4310 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4312         PR target/57623
4313         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4314         constraints of operand 1 and 2.
4316         PR target/57623
4317         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4318         to match RTL canonicalization.  Swap predicates and
4319         constraints of operand 1 and 2.
4321 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
4323         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4324         Process OP_INOUT regs for splitting too.
4326 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
4328         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4329         decl before the loop, initialize to NULL.
4330         (vectorizable_load): Initialize ptr_incr to NULL.
4332 2013-06-27  Martin Jambor  <mjambor@suse.cz>
4334         PR lto/57208
4335         * ipa-ref.h (ipa_maybe_record_reference): Declare.
4336         * ipa-ref.c (ipa_maybe_record_reference): New function.
4337         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4338         * ipa-cp.c (create_specialized_node): Record potential references from
4339         aggvals.
4340         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4342 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4344         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4345         parameter 'mode' of type 'enum machine_mode mode'; change to pass
4346         'mode' to force_reg.
4347         (aarch64_add_offset): Update calls to aarch64_force_temporary.
4348         (aarch64_expand_mov_immediate): Likewise.
4350 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
4352         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4353         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4355 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4357         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4358         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4359         (struct machine_function): Add tbegin_p.
4360         (s390_canonicalize_comparison): Fold CC mode compares to
4361         conditional jump if possible.
4362         (s390_emit_jump): Return the emitted jump.
4363         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4364         Handle CCRAWmode compares.
4365         (s390_option_override): Default to -mhtm if available.
4366         (s390_reg_clobbered_rtx): Handle floating point regs as well.
4367         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4368         FPRs instead of df_regs_ever_live_p.
4369         (s390_optimize_nonescaping_tx): New function.
4370         (s390_init_frame_layout): Extend clobbered_regs array to cover
4371         FPRs as well.
4372         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4373         (s390_expand_tbegin): New function.
4374         (enum s390_builtin): New enum definition.
4375         (code_for_builtin): New array definition.
4376         (s390_init_builtins): New function.
4377         (s390_expand_builtin): New function.
4378         (TARGET_INIT_BUILTINS): Define.
4379         (TARGET_EXPAND_BUILTIN): Define.
4380         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4381         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4382         (s390_alc_comparison): Likewise.
4383         * config/s390/s390-modes.def: Add CCRAWmode.
4384         * config/s390/s390.h (processor_flags): Add PF_TX.
4385         (TARGET_CPU_HTM): Define macro.
4386         (TARGET_HTM): Define macro.
4387         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4388         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4389         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4390         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4391         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4392         (TBEGIN_MASK, TBEGINC_MASK): New constants.
4393         ("*cc_to_int"): Move up.
4394         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4395         constants other than 0.
4396         ("*ccraw_to_int"): New insn and splitter definition.
4397         ("tbegin", "tbegin_nofloat", "tbegin_retry")
4398         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4399         ("tx_assist"): New expander.
4400         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4401         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4402         * config/s390/s390.opt: Add -mhtm option.
4403         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4404         * config/s390/htmxlintrin.h: New file.
4405         * config/s390/htmintrin.h: New file.
4406         * config/s390/s390intrin.h: New file.
4407         * doc/extend.texi: Document htm builtins.
4408         * config.gcc: Add the new header files to extra_headers.
4410 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
4412         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4413         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4415 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
4416             Pat Haugen <pthaugen@us.ibm.com>
4417             Peter Bergner <bergner@vnet.ibm.com>
4419         * config/rs6000/power8.md: New.
4420         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4421         setting for power8 entry.
4422         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4423         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4424         test for Power4/Power5 only.
4425         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4426         support.
4427         (force_new_group): Adjust comment.
4428         * config/rs6000/rs6000.md: Include power8.md.
4430 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
4432         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4433         * config/arm/arm-protos.h (arm_max_conditional_execute): New
4434         declaration.
4435         (tune_params): Update comment.
4436         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4437         (arm_max_conditional_execute): New function.
4438         (thumb2_final_prescan_insn): Use max_insn_skipped and
4439         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4441 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
4443         PR tree-optimization/57705
4444         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4445         SSA_NAME step, provided that it is not defined inside the loop.
4446         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4447         (get_initial_def_for_induction): Handle SSA_NAME IV step.
4449 2013-06-25  Martin Jambor  <mjambor@suse.cz>
4451         PR middle-end/57670
4452         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4453         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4454         calls in the dump.
4455         (ipa_note_param_call): Initialize member_ptr flag.
4456         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4457         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4458         (ipa_write_indirect_edge_info): Stream member_ptr flag.
4459         (ipa_read_indirect_edge_info): Likewise.
4461 2013-06-25  Richard Biener  <rguenther@suse.de>
4463         PR middle-end/56977
4464         * passes.c (init_optimization_passes): Move pass_fold_builtins
4465         and pass_dce earlier with -Og.
4467 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
4469         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4470         <BIT_FIELD_REF>: Remove trailing TAB.
4471         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4472         remove blank line.
4474 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4476         PR tree-optimization/57358
4477         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4478         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4479         (ipa_analyze_params_uses): Generate pessimistic info when true.
4481 2013-06-24  Martin Jambor  <mjambor@suse.cz>
4483         PR tree-optimization/57539
4484         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4485         global.inlined_to of the new node to it.  All callers changed.
4486         * ipa-inline-transform.c (clone_inlined_nodes): New variable
4487         inlining_into, pass it to cgraph_clone_node.
4488         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4489         ipa_free_edge_args_substructures.
4490         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4491         rdesc linked list.  Do not assert rdesc edges have inlined caller.
4492         Assert we have found an rdesc in the rdesc list.
4494 2013-06-24  Richard Biener  <rguenther@suse.de>
4496         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4497         (pointer_set_lookup): Declare.
4498         (class pointer_map): New template class implementing a
4499         generic pointer to T map.
4500         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4501         pointer_map<T>::contains, pointer_map<T>::insert,
4502         pointer_map<T>::traverse): New functions.
4503         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4504         (pointer_set_lookup): New function.
4505         (pointer_set_contains): Use pointer_set_lookup.
4506         (pointer_set_insert): Likewise.
4507         (insert_aux): Remove.
4508         (struct pointer_map_t): Embed a pointer_set_t.
4509         (pointer_map_create): Adjust.
4510         (pointer_map_destroy): Likewise.
4511         (pointer_map_contains): Likewise.
4512         (pointer_map_insert): Likewise.
4513         (pointer_map_traverse): Likewise.
4514         * tree-streamer.h (struct streamer_tree_cache_d): Use a
4515         pointer_map<unsigned> instead of a pointer_map_t.
4516         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4517         (streamer_tree_cache_lookup): Likewise.
4518         (streamer_tree_cache_create): Likewise.
4519         (streamer_tree_cache_delete): Likewise.
4520         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4521         pointer_map<unsigned> instead of a pointer_map_t.
4522         (lto_init_tree_ref_encoder): Adjust.
4523         (lto_destroy_tree_ref_encoder): Likewise.
4524         * lto-section-out.c (lto_output_decl_index): Likewise.
4525         (lto_record_function_out_decl_state): Likewise.
4526         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4528 2013-06-24  Richard Biener  <rguenther@suse.de>
4530         PR tree-optimization/57488
4531         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4533 2013-06-24  Alan Modra  <amodra@gmail.com>
4535         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4536         (gen_easy_altivec_constant): Likewise.
4537         * config/rs6000/predicates.md (easy_vector_constant_add_self,
4538         easy_vector_constant_msb): Likewise.
4540 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
4542         PR target/57688
4543         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4544         add missing return true.
4546 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
4548         PR target/52483
4549         * config/sh/predicates.md (general_extend_operand): Invoke
4550         general_movsrc_operand for memory operands.
4551         (general_movsrc_operand): Allow reg+reg addressing, do not use
4552         general_operand for memory operands.
4554 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
4556         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
4557         when current target options does not apply.
4558         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4559         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4560         * config/i386/bmiintrin.h: Pass appropriate target
4561         attributes to header.
4562         * config/i386/mmintrin.h: Ditto.
4563         * config/i386/nmmintrin.h: Ditto.
4564         * config/i386/avx2intrin.h: Ditto.
4565         * config/i386/fxsrintrin.h: Ditto.
4566         * config/i386/tbmintrin.h: Ditto.
4567         * config/i386/xsaveintrin.h: Ditto.
4568         * config/i386/f16cintrin.h: Ditto.
4569         * config/i386/xtestintrin.h: Ditto.
4570         * config/i386/xsaveoptintrin.h: Ditto.
4571         * config/i386/bmi2intrin.h: Ditto.
4572         * config/i386/lzcntintrin.h: Ditto.
4573         * config/i386/smmintrin.h: Ditto.
4574         * config/i386/wmmintrin.h: Ditto.
4575         * config/i386/x86intrin.h: Remove all header include guards.
4576         * config/i386/prfchwintrin.h: Ditto.
4577         * config/i386/pmmintrin.h: Ditto.
4578         * config/i386/tmmintrin.h: Ditto.
4579         * config/i386/xmmintrin.h: Ditto.
4580         * config/i386/popcntintrin.h: Ditto.
4581         * config/i386/rdseedintrin.h: Ditto.
4582         * config/i386/ammintrin.h: Ditto.
4583         * config/i386/emmintrin.h: Ditto.
4584         * config/i386/immintrin.h: Remove all header include guards.
4585         * config/i386/fma4intrin.h: Ditto.
4586         * config/i386/lwpintrin.h: Ditto.
4587         * config/i386/xopintrin.h: Ditto.
4588         * config/i386/ia32intrin.h: Ditto.
4589         * config/i386/avxintrin.h: Ditto.
4590         * config/i386/rtmintrin.h: Ditto.
4591         * config/i386/fmaintrin.h: Ditto.
4592         * config/i386/mm3dnow.h: Ditto.
4594 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
4596         * common/config/i386/i386-common.c: Handle LZCNT.
4598 2013-06-22  Andi Kleen  <ak@linux.intel.com>
4600         * doc/extend.texi: Use __atomic_store_n instead of
4601         __atomic_store in HLE example.
4603 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
4605         * config/sh/sh.c: Remove <cstdlib> workaround.
4607 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4609         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
4611 2013-06-21  Andi Kleen  <ak@linux.intel.com>
4613         * doc/extend.texi: Document that __atomic_clear and
4614           __atomic_test_and_set should only be used with bool.
4616 2013-06-20  Jan Hubicka  <jh@suse.cz>
4618         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4619         types_same_for_odr.
4620         * tree.c (decls_same_for_odr): New function.
4621         (same_for_edr): New function.
4622         (types_same_for_odr): New function.
4623         (get_binfo_at_offset): Use it.
4624         * tree.h (types_same_for_odr): Declare.
4626 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
4627             Jason Merrill  <jason@redhat.com>
4629         * system.h: Include <cstdlib> as well as <stdlib.h>.
4631 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
4633         PR target/57655
4634         * config/i386/i386.c (construct_container): Report error if
4635         long double is used with disabled x87 float returns.
4637 2013-06-20  Jan Hubicka  <jh@suse.cz>
4639         * lto-cgraph.c (input_symtab): Do not set cgraph state.
4641 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
4643         PR rtl-optimization/57425
4644         PR rtl-optimization/57569
4645         * alias.c (write_dependence_p): Remove parameters mem_mode and
4646         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
4647         Changed all callers.
4648         (canon_anti_dependence): Get comments and semantics in sync.
4649         Add parameter mem_canonicalized.  Changed all callers.
4650         * rtl.h (canon_anti_dependence): Update prototype.
4652 2013-06-20  Richard Biener  <rguenther@suse.de>
4654         * data-streamer-in.c (streamer_read_uhwi): Optimize single
4655         byte case, inline streamer_read_uchar and defer section
4656         overrun check.
4658 2013-06-20  Richard Biener  <rguenther@suse.de>
4660         PR tree-optimization/57584
4661         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4662         SSA names into the expanded expression that take part in
4663         abnormal coalescing.
4665 2013-06-19  Sharad Singhai  <singhai@google.com>
4667         * gcov.c (print_usage): Handle new option.
4668         (process_args): Ditto.
4669         (get_gcov_intermediate_filename): New function.
4670         (output_intermediate_file): New function.
4671         (output_gcov_file): New function
4672         (generate_results): Handle new option.
4673         (release_function): Relase demangled name.
4674         (read_graph_file): Handle demangled name.
4675         (output_lines): Ditto.
4676         * doc/gcov.texi: Document gcov intermediate format.
4678 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
4680         PR bootstrap/57604
4681         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4682         (lra_emit_add): Use the functions.  Add comment about Y as an
4683         address segment.
4685 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
4687         PR driver/57652
4688         * collect2.c (collect_atexit): New.
4689         (collect_exit): Delete.
4690         (main): Register collect_atexit with atexit.
4691         (collect_wait): Change collect_exit to exit.
4692         (do_wait): Same.
4693         * collect2.h (collect_exit): Delete.
4694         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4696 2013-06-19  Wei Mi  <wmi@google.com>
4698         PR rtl-optimization/57518
4699         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4700         if regno is used in paradoxical subreg.
4701         (update_equiv_regs): Check pdx_subregs[regno] before
4702         set a reg to be equivalent with a mem.
4704 2013-06-19  Matthias Klose  <doko@ubuntu.com>
4706         PR driver/57651
4707         * file-find.h (find_a_file): Add a mode parameter.
4708         * file-find.c (find_a_file): Likewise.
4709         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4710         with X_OK for the executables.
4711         * collect2.c (main): Call find_a_file with X_OK.
4713 2013-06-19  Steve Ellcey  <sellcey@mips.com>
4715         PR target/56942
4716         * config/mips/mips.md (casesi_internal_mips16_<mode>):
4717         Use NEXT_INSN instead of next_real_insn.
4719 2013-06-19  Jan Hubicka  <jh@suse.cz>
4721         * cgraph.h (const_value_known_p): Replace by ...
4722         (ctor_for_folding): .. this one.
4723         * cgraphunit.c (process_function_and_variable_attributes): Use it.
4724         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4725         * expr.c (expand_expr_real_1): Likewise.
4726         (string_constant): Likewise.
4727         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4728         * ipa.c (process_references): Likewise.
4729         (symtab_remove_unreachable_nodes): Likewise.
4730         * ipa-inline-analysis.c (param_change_prob): Likewise.
4731         * gimple-fold.c (canonicalize_constructor_val): Likewise.
4732         (get_base_constructor): Likwise.
4733         * varpool.c (varpool_remove_node): Likewise.
4734         (varpool_remove_initializer): LIkewise.
4735         (dump_varpool_node): LIkwise.
4736         (const_value_known_p): Rewrite to ...
4737         (ctor_for_folding): ... this one.
4739 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
4741         PR driver/57651
4742         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4743         PERSONALITY in $PATH derived prefixes.
4745 2013-06-19  Jeff Law  <law@redhat.com>
4747         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4748         in comment.
4750         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4751         (simplify_bitwise_binary): Use it to simpify certain binary ops on
4752         booleans.
4754 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
4756         * config/arm/vfp.md: Move VFP instruction classification documentation
4757         to ...
4758         * config/arm/arm.md: ... here.  Update instruction classification
4759         documentation.
4761 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
4763         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4764         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4765         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
4767 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
4769         PR target/57609
4770         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4771         with NEXT_INSN.  Use tablejump_p to check for jump table data
4772         insns.
4774 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
4776         PR c++/56544
4777         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4778         that now in C++ the value is correct per the C++ standards.
4780 2013-06-19  Richard Biener  <rguenther@suse.de>
4782         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4783         for global context.
4785 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4787         Revert:
4788         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4790         PR target/57609
4791         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4792         with next_active_insn.
4794 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
4796         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4797         functions are inlined during failures to flag an error.
4798         * tree-inline.c (expand_call_inline): Allow the error to be flagged
4799         in early inline pass.
4801 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
4803         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4804         in comments.
4806 2013-06-18  Julian Brown  <julian@codesourcery.com>
4808         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4809         Permit virtual register pre-reload if !strict.
4810         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4811         change.
4812         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4813         prototype.
4814         * config/arm/neon.md (movmisalign<mode>): Use
4815         neon_perm_struct_or_reg_operand instead of
4816         neon_struct_or_register_operand.
4817         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4818         neon_permissive_struct_operand instead of neon_struct_operand.
4819         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4820         neon_vector_mem_operand.
4821         * config/arm/predicates.md (neon_struct_operand): Adjust call to
4822         neon_vector_mem_operand.
4823         (neon_permissive_struct_operand): New.
4824         (neon_struct_or_register_operand): Rename to...
4825         (neon_perm_struct_or_reg_operand): This. Adjust call to
4826         neon_vector_mem_operand.
4828 2013-06-18  Richard Biener  <rguenther@suse.de>
4830         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4831         * lto-streamer.h: Include pointer-set.h.
4832         (struct lto_decl_slot): Remove.
4833         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4834         Remove next_index entry.
4835         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4836         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4837         (lto_init_tree_ref_encoder): Adjust.
4838         (lto_destroy_tree_ref_encoder): Likewise.
4839         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4840         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4841         (lto_output_decl_index): Adjust.
4842         (lto_new_out_decl_state): Likewise.
4843         (lto_record_function_out_decl_state): Likewise.
4844         * lto-streamer-out.c (copy_function): Likewise.
4846 2013-06-18  Richard Biener  <rguenther@suse.de>
4848         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
4849         * cgraphunit.c: Include cfgloop.h.
4850         (init_lowered_empty_function): Initialize the loop tree.
4851         (assemble_thunk): Insert new BBs into loops.
4853 2013-06-18  Richard Biener  <rguenther@suse.de>
4855         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
4856         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
4857         the map from cache entry to cache index optional.
4858         (streamer_tree_cache_replace_tree): Adjust accordingly.
4859         (streamer_tree_cache_append): Likewise.
4860         (streamer_tree_cache_delete): Likewise.
4861         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
4862         streamer cache map from cache entry to cache index.
4863         * lto-streamer-out.c (create_output_block): Adjust.
4865 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
4867         * config/arm/arm.md (attribute "insn"): Move multiplication and
4868         division attributes to...
4869         (attribute "type"): ... here.  Remove mult.
4870         (attribute "mul32"): New attribute.
4871         (attribute "mul64"): Add umaal.
4872         (*arm_mulsi3): Update attributes.
4873         (*arm_mulsi3_v6): Likewise.
4874         (*thumb_mulsi3): Likewise.
4875         (*thumb_mulsi3_v6): Likewise.
4876         (*mulsi3_compare0): Likewise.
4877         (*mulsi3_compare0_v6): Likewise.
4878         (*mulsi_compare0_scratch): Likewise.
4879         (*mulsi_compare0_scratch_v6): Likewise.
4880         (*mulsi3addsi): Likewise.
4881         (*mulsi3addsi_v6): Likewise.
4882         (*mulsi3addsi_compare0): Likewise.
4883         (*mulsi3addsi_compare0_v6): Likewise.
4884         (*mulsi3addsi_compare0_scratch): Likewise.
4885         (*mulsi3addsi_compare0_scratch_v6): Likewise.
4886         (*mulsi3subsi): Likewise.
4887         (*mulsidi3adddi): Likewise.
4888         (*mulsi3addsi_v6): Likewise.
4889         (*mulsidi3adddi_v6): Likewise.
4890         (*mulsidi3_nov6): Likewise.
4891         (*mulsidi3_v6): Likewise.
4892         (*umulsidi3_nov6): Likewise.
4893         (*umulsidi3_v6): Likewise.
4894         (*umulsidi3adddi): Likewise.
4895         (*umulsidi3adddi_v6): Likewise.
4896         (*smulsi3_highpart_nov6): Likewise.
4897         (*smulsi3_highpart_v6): Likewise.
4898         (*umulsi3_highpart_nov6): Likewise.
4899         (*umulsi3_highpart_v6): Likewise.
4900         (mulhisi3): Likewise.
4901         (*mulhisi3tb): Likewise.
4902         (*mulhisi3bt): Likewise.
4903         (*mulhisi3tt): Likewise.
4904         (maddhisi4): Likewise.
4905         (*maddhisi4tb): Likewise.
4906         (*maddhisi4tt): Likewise.
4907         (maddhidi4): Likewise.
4908         (*maddhidi4tb): Likewise.
4909         (*maddhidi4tt): Likewise.
4910         (divsi3): Likewise.
4911         (udivsi3): Likewise.
4912         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
4913         (thumb2_mulsi_short_compare0): Likewise.
4914         (thumb2_mulsi_short_compare0_scratch): Likewise.
4915         * config/arm/arm1020e.md (1020mult1): Update attribute change.
4916         (1020mult2): Likewise.
4917         (1020mult3): Likewise.
4918         (1020mult4): Likewise.
4919         (1020mult5): Likewise.
4920         (1020mult6): Likewise.
4921         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
4922         change.
4923         (cortex_a15_mult64): Likewise.
4924         (cortex_a15_sdiv): Likewise.
4925         (cortex_a15_udiv): Likewise.
4926         * config/arm/arm1026ejs.md (mult1): Update attribute change.
4927         (mult2): Likewise.
4928         (mult3): Likewise.
4929         (mult4): Likewise.
4930         (mult5): Likewise.
4931         (mult6): Likewise.
4932         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
4933         (pj4_ir_div): Likewise.
4934         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
4935         (11_mult2): Likewise.
4936         (11_mult3): Likewise.
4937         (11_mult4): Likewise.
4938         (11_mult5): Likewise.
4939         (11_mult6): Likewise.
4940         (11_mult7): Likewise.
4941         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
4942         (cortex_a8_mla): Likewise.
4943         (cortex_a8_mull): Likewise.
4944         (cortex_a8_smulwy): Likewise.
4945         (cortex_a8_smlald): Likewise.
4946         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
4947         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
4948         (cortex_r4_mul_3): Likewise.
4949         (cortex_r4_mla_4): Likewise.
4950         (cortex_r4_mla_3): Likewise.
4951         (cortex_r4_smlald): Likewise.
4952         (cortex_r4_mull): Likewise.
4953         (cortex_r4_sdiv): Likewise.
4954         (cortex_r4_udiv): Likewise.
4955         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
4956         (cortex_a7_idiv): Likewise.
4957         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
4958         (9_mult2): Likewise.
4959         (9_mult3): Likewise.
4960         (9_mult4): Likewise.
4961         (9_mult5): Likewise.
4962         (9_mult6): Likewise.
4963         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
4964         (cortex_a53_sdiv): Likewise.
4965         (cortex_a53_udiv): Likewise.
4966         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
4967         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
4968         (mp626_mult2): Likewise.
4969         (mp626_mult3): Likewise.
4970         (mp626_mult4): Likewise.
4971         * config/arm/fa526.md (526_mult1): Update attribute change.
4972         (526_mult2): Likewise.
4973         * config/arm/arm-generic.md (mult): Update attribute change.
4974         (mult_ldsched_strongarm): Likewise.
4975         (mult_ldsched): Likewise.
4976         (multi_cycle): Likewise.
4977         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
4978         * config/arm/fa606te.md (606te_mult1): Update attribute change.
4979         (606te_mult2): Likewise.
4980         (606te_mult3): Likewise.
4981         (606te_mult4): Likewise.
4982         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
4983         (cortex_a9_mac16): Likewise.
4984         (cortex_a9_multiply): Likewise.
4985         (cortex_a9_mac): Likewise.
4986         (cortex_a9_multiply_long): Likewise.
4987         * config/arm/fa626te.md (626te_mult1): Update attribute change.
4988         (626te_mult2): Likewise.
4989         (626te_mult3): Likewise.
4990         (626te_mult4): Likewise.
4992 2013-06-18  Richard Biener  <rguenther@suse.de>
4994         PR lto/57334
4995         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
4997 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4999         PR target/57609
5000         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5001         with next_active_insn.
5003 2013-06-18  Alan Modra  <amodra@gmail.com>
5005         * config/rs6000/rs6000.h (enum data_align): New.
5006         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5007         (DATA_ABI_ALIGNMENT): Define.
5008         (CONSTANT_ALIGNMENT): Correct comment.
5009         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5010         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5012 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5014         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5015         ATTRIBUTE_UNUSED marking.
5017 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5019         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5020         alternative and update.
5021         (aarch64_dup_lanedi): Delete.
5022         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5023         * config/aarch64/aarch64-simd-builtins.def: Update.
5025 2013-06-17  Richard Biener  <rguenther@suse.de>
5027         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5028         (lto_input_scc): Declare.
5029         (lto_input_tree_1): Likewise.
5030         (struct lto_stats_d): Add num_tree_bodies_output and
5031         num_pickle_refs_output.
5032         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5033         (lto_read_tree_1): Split out from ...
5034         (lto_read_tree): ... this.
5035         (lto_input_scc): New function.
5036         (lto_input_tree_1): Split out from ...
5037         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5038         (lto_data_in_create): Create the streamer cache without hashes.
5039         * lto-streamer-out.c (create_output_block): Create the streamer
5040         cache with hashes when not doing WPA.
5041         (lto_write_tree_1): Split out from ...
5042         (lto_write_tree): ... this.
5043         (get_symbol_initial_value): New function.
5044         (lto_output_tree_1): Split out from ...
5045         (lto_output_tree): ... this.  Write trees as series of SCCs
5046         using a DFS walk via DFS_write_tree.
5047         (struct sccs, struct scc_entry): New types.
5048         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5049         (DFS_write_tree_body): New function.
5050         (DFS_write_tree): Likewise.
5051         (hash_tree): Likewise.
5052         (scc_entry_compare): Likewise.
5053         (hash_scc): Likewise.
5054         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5055         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5056         TREE_CHAIN as regular reference.
5057         (streamer_read_integer_cst): Remove.
5058         (streamer_get_pickled_tree): Adjust.
5059         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5060         of DECL_EXTERNALs in BLOCK_VARS for now.
5061         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5062         reference.
5063         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5064         Add hash value argument and record that if hashes are recorded
5065         in the cache.
5066         (streamer_tree_cache_insert_1): Adjust.
5067         (streamer_tree_cache_insert): Likewise.
5068         (streamer_tree_cache_insert_at): Rename to ...
5069         (streamer_tree_cache_replace_tree): ... this and adjust.
5070         (streamer_tree_cache_append): Adjust.
5071         (record_common_node): Likewise.
5072         (streamer_tree_cache_create): Add argument whether to
5073         record hash values together with trees.
5074         (streamer_tree_cache_delete): Adjust.
5075         * tree-streamer.h (struct streamer_tree_cache_d): Add
5076         vector of hashes.
5077         (streamer_read_integer_cst): Remove.
5078         (streamer_tree_cache_insert): Adjust.
5079         (streamer_tree_cache_append): Likewise.
5080         (streamer_tree_cache_insert_at): Rename to ...
5081         (streamer_tree_cache_replace_tree): ... this and adjust.
5082         (streamer_tree_cache_create): Add argument whether to record hashes.
5083         (streamer_tree_cache_get): Rename to ...
5084         (streamer_tree_cache_get_tree): ... this.
5085         (streamer_tree_cache_get_hash): New function.
5086         * tree.c (cache_integer_cst): New function.
5087         * tree.h (cache_integer_cst): Declare.
5088         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5089         * lto-symtab.c (lto_varpool_replace_node): Only release
5090         DECL_INITIAL of non-prevailing decls.
5091         * varpool.c (varpool_remove_initializer): Do not release
5092         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5094 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5096         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5097         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5098         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5099         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5100         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5101         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5102         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5103         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5104         instead of TARGET_64BIT.
5105         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5106         Require ISA_HAS_<D>DIV.
5108 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
5110         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5111         (mips*-*-linux*): Move default with_llsc setting to where other
5112         defaults are set.
5113         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5114         with_arch block.
5115         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5116         Likewise.  Remove default with_tune setting.  Move default float
5117         setting to its own block.  Handle with_llsc in the same block as above.
5119 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
5121         PR rtl-optimization/57425
5122         PR rtl-optimization/57569
5123         * alias.c (write_dependence_p): Add new parameters mem_mode,
5124         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
5125         Changed all callers.
5126         (canon_anti_dependence): New function.
5127         * cse.c (check_dependence): Use canon_anti_dependence.
5128         * cselib.c (cselib_invalidate_mem): Likewise.
5129         * rtl.h (canon_anti_dependence): Declare.
5131 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5133         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5134         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5135         ".set mips3" for 64-bit targets.
5137 2013-06-15  Dehao Chen  <dehao@google.com>
5139         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5140         * gimple-low.c (gimple_check_call_matching_types): Likewise.
5141         (gimple_check_call_args): Likewise.
5142         * value-prof.c (check_ic_target): Likewise.
5143         * ipa-inline.c (early_inliner): Likewise.
5144         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5145         * cgraph.c (cgraph_create_edge_1): Likewise.
5146         (cgraph_make_edge_direct): Likewise.
5148 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
5150         PR target/57615
5151         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5152         rs6000_output_move_128bit to handle emitting quad memory
5153         operations.  Set attribute length to 8 bytes.
5155 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
5157         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5158         New pattern.
5159         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5160         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5161         (aarch64_<su>mlsl<mode>): Likewise.
5163 2013-06-14  Mike Stump  <mikestump@comcast.net>
5165         * Makefile.in (TARGET_H): Add insn-codes.h.
5167 2013-06-14  Alan Modra  <amodra@gmail.com>
5169         PR middle-end/57134
5170         PR middle-end/57586
5171         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5172         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
5173         bitfield expansion when EXPAND_MEMORY.
5174         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5176 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
5178         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5179         test for clearing quad memory on 32-bit later.
5181 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5183         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5184         (fold_negate_expr): Likewise.
5185         (fold_real_zero_addition_p): Handle vectors.
5186         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5188 2013-06-14  Alan Modra  <amodra@gmail.com>
5190         * varasm.c (force_const_mem): Revert 2013-06-07 change.
5192 2013-06-13  Jan Hubicka  <jh@suse.cz>
5194         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5195         Local comdats are not externally visible.
5196         * symtab.c (dump_symtab_base): Dump externally visible.
5197         (verify_symtab_base): Verify back links in the symtab hash.
5199 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5201         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5202         CONVERT_EXPR as equal nodes.
5204 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
5206         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5208 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
5210         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5211         Generalize to complex and vector.
5212         * tree.c (build_all_ones_cst): New function.
5213         * tree.h (build_all_ones_cst): Declare it.
5215 2013-06-13  Alan Modra  <amodra@gmail.com>
5217         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5218         * config/rs6000/rs6000.md (signbittf2): New insn.
5219         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5220         (abstf2_internal, cmptf_internal2): Likewise.
5221         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5223 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5224             Pat Haugen <pthaugen@us.ibm.com>
5225             Peter Bergner <bergner@vnet.ibm.com>
5227         * config/rs6000/rs6000.c (emit_load_locked): Add support for
5228         power8 byte, half-word, and quad-word atomic instructions.
5229         (emit_store_conditional): Likewise.
5230         (rs6000_expand_atomic_compare_and_swap): Likewise.
5231         (rs6000_expand_atomic_op): Likewise.
5233         * config/rs6000/sync.md (larx): Add new modes for power8.
5234         (stcx): Likewise.
5235         (AINT): New mode iterator to include TImode as well as normal
5236         integer modes on power8.
5237         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5238         that VSX registers are not considered.  Use AINT mode iterator
5239         instead of INT1 to allow inclusion of quad word atomic operations
5240         on power8.
5241         (load_locked<mode>): Likewise.
5242         (store_conditional<mode>): Likewise.
5243         (atomic_compare_and_swap<mode>): Likewise.
5244         (atomic_exchange<mode>): Likewise.
5245         (atomic_nand<mode>): Likewise.
5246         (atomic_fetch_<fetchop_name><mode>): Likewise.
5247         (atomic_nand_fetch<mode>): Likewise.
5248         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5249         each type.
5250         (ATOMIC): On power8, add QImode, HImode modes.
5251         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5252         modes that promote to SImode.
5253         (load_lockedti): Convert TImode arguments to PTImode, so that we
5254         get a guaranteed even/odd register pair.
5255         (load_lockedpti): Likewise.
5256         (store_conditionalti): Likewise.
5257         (store_conditionalpti): Likewise.
5259         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5260         atomic load/store instructions.
5261         (HSI): Likewise.
5263 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
5265         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5266         loads.
5267         (insn_count): New attribute, with most cases extracted from...
5268         (length): ...here.  Redefine most cases in terms of insn_count.
5269         (single_insn): Delete.
5270         (can_delay): Use insn_count to check for single instructions.
5271         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5272         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5273         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5274         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5275         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5276         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5277         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5278         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5279         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5280         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5281         rather than "length".
5282         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
5283         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5284         Use "insn_count" rather than "length".
5285         * config/mips/mips-dsp.md
5286         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5287         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5288         length attributes.
5290 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
5292         PR tree-optimization/57361
5293         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5295 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
5297         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5298         to split.
5299         (aarch64_simd_combine<mode>): New instruction expansion.
5300         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5301         function prototype.
5302         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5303         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5305 2013-06-12  Jan Hubicka  <jh@suse.cz>
5307         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5308         decl has when in streaming stage.
5309         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5310         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5312 2013-06-12  Roland Stigge <stigge@antcom.de>
5314         PR target/57578
5315         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5317 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5319         PR tree-optimization/57537
5320         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5321         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5323 2013-06-12  Richard Biener  <rguenther@suse.de>
5325         * data-streamer.h (streamer_write_char_stream): CSE
5326         obs->current_pointer.
5327         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5328         streamer_write_char_stream manually and optimize the resulting loop.
5329         (streamer_write_hwi_stream): Likewise.
5331 2013-06-12  Jan Hubicka  <jh@suse.cz>
5333         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5334         * cgraph.h (varpool_create_empty_node): Declare.
5335         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5336         duplicated nodes.
5337         * symtab.c (symtab_unregister_node): Be lax about missin entries
5338         in node hash.
5339         (symtab_get_node): Update comment.
5340         * varpool.c (varpool_create_empty_node): Break out from ...
5341         (varpool_node_for_decl): ... here.
5342         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5344 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
5346         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5347         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5348         part.  Use straight-line flow at the end.
5349         <COMPONENT_REF>: Remove superfluous else.
5350         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5352 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
5354         PR target/56564
5355         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5356         target hook even for !TREE_PUBLIC decls.  If no resolution info
5357         is available, return false for common and external decls.
5359 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
5361         * config/rl78/constraints.md (U): New constraint.
5362         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5363         valloc attribute.
5365 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5367         PR target/57589
5368         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5369         to allow returning address to AT_PLATFORM name.
5371 2013-06-11  Jan Hubicka  <jh@suse.cz>
5373         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5374         * cgraph.h (symtab_node_base): Add weakref flag.
5375         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5376         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5377         (output_weakrefs): Use weakref flag.
5378         * fold-const.c (simple_operand_p): Handle WEAK.
5379         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5380         * ipa.c (varpool_externally_visible_p): Drop weakref.
5381         (function_and_variable_visibility): Update comment; fix weakref
5382         sanity checks; do not clear DECL_WEAK on them.
5383         * lto-cgraph.c (lto_output_node): update.
5384         (lto_output_varpool_node): Update.
5385         (input_overwrite_node): Update.
5386         (input_node): Update.
5387         (input_varpool_node): Update.
5388         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5389         (lto_symtab_merge_symbols): Add sanity check.
5390         (lto_symtab_prevailing_decl): Do not special case weakrefs.
5391         * passes.c (rest_of_decl_compilation): Set static flag, too.
5392         * symtab.c (dump_symtab_base): Dump weakref.
5393         (verify_symtab_base): Sanity check weakrefs.
5394         (symtab_make_decl_local): Remove duplicated code.
5395         (symtab_alias_ultimate_target): Simplify.
5396         * varpool.c (varpool_create_variable_alias): Set weakref flag.
5398 2013-06-11  DJ Delorie  <dj@redhat.com>
5400         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5401         (rl78_unwind_word_mode): New.
5403 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5405         * final.c (debug_prefix_maps): Make static.
5407 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
5409         * function.c (initial_trampoline): Remove stray copy.
5411 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
5413         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5415 2013-06-11  Martin Jambor  <mjambor@suse.cz>
5417         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5418         within bounds at the beginning of the function.
5420 2013-06-11  Alan Modra  <amodra@gmail.com>
5422         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
5423         reporting.
5424         (get_named_section): Don't NULL !DECL_P decl.
5426 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
5428         * doc/invoke.texi (core-avx2): Document.
5429         (slm): Likewise.
5430         (atom): Updated with MOVBE.
5432 2013-06-11  Richard Biener  <rguenther@suse.de>
5434         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5436 2013-06-11  Anton Blanchard  <anton@samba.org>
5438         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5439         correct shift value in little-endian mode.
5441 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
5443         PR target/56564
5444         * varasm.c (get_variable_align): Move #endif to the right place.
5446 2013-06-10  Cary Coutant  <ccoutant@google.com>
5448         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5449         for hash so that hash table traversal order is deterministic.
5451 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
5452             Pat Haugen <pthaugen@us.ibm.com>
5453             Peter Bergner <bergner@vnet.ibm.com>
5455         * config/rs6000/vector.md (GPR move splitter): Do not split moves
5456         of vectors in GPRS if they are direct moves or quad word load or
5457         store moves.
5459         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5460         declaration.
5461         (direct_move_p): Likewise.
5462         (quad_load_store_p): Likewise.
5464         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5465         classes into bins based on the physical register type.
5466         (reg_class_to_reg_type): Likewise.
5467         (IS_STD_REG_TYPE): Likewise.
5468         (IS_FP_VECT_REG_TYPE): Likewise.
5469         (reload_fpr_gpr): Arrays to determine what insn to use if we can
5470         use direct move instructions.
5471         (reload_gpr_vsx): Likewise.
5472         (reload_vsx_gpr): Likewise.
5473         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5474         information that is a simplification of register classes.  Also
5475         precalculate direct move reload helpers.
5476         (direct_move_p): New function to return true if the operation can
5477         be done as a direct move instruciton.
5478         (quad_load_store_p): New function to return true if the operation
5479         is a quad memory operation.
5480         (rs6000_legitimize_address): If quad memory, only allow register
5481         indirect for TImode addresses.
5482         (rs6000_legitimate_address_p): Likewise.
5483         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5484         (rs6000_reload_register_type): Likewise.
5485         (register_to_reg_type): Return register type.
5486         (rs6000_secondary_reload_simple_move): New helper function for
5487         secondary reload and secondary memory needed to identify anything
5488         that is a simple move, and does not need reloading.
5489         (rs6000_secondary_reload_direct_move): New helper function for
5490         secondary reload to identify cases that can be done with several
5491         instructions via the direct move instructions.
5492         (rs6000_secondary_reload_move): New helper function for secondary
5493         reload to identify moves between register types that can be done.
5494         (rs6000_secondary_reload): Add support for quad memory operations
5495         and for direct move.
5496         (rs6000_secondary_memory_needed): Likewise.
5497         (rs6000_debug_secondary_memory_needed): Change argument names.
5498         (rs6000_output_move_128bit): New function to return the move to
5499         use for 128-bit moves, including knowing about the various
5500         limitations of quad memory operations.
5502         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5503         memory operations.  call rs6000_output_move_128bit for the actual
5504         instruciton(s) to generate.
5505         (vsx_movti_64bit): Likewise.
5507         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5508         (UNSPEC_P8V_MTVSRWZ): Likewise.
5509         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5510         (UNSPEC_P8V_MTVSRD): Likewise.
5511         (UNSPEC_P8V_XXPERMDI): Likewise.
5512         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5513         (UNSPEC_FUSION_GPR): Likewise.
5514         (FMOVE128_GPR): New iterator for direct move.
5515         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5516         (f32_sv): Likewise.
5517         (f32_dm): Likewise.
5518         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5519         loads and direct move instructions.
5520         (zero_extendsidi2_lfiwzx): Likewise.
5521         (extendsidi2_lfiwax): Likewise.
5522         (extendsidi2_nocell): Likewise.
5523         (floatsi<mode>2_lfiwax): Likewise.
5524         (lfiwax): Likewise.
5525         (floatunssi<mode>2_lfiwzx): Likewise.
5526         (lfiwzx): Likewise.
5527         (fix_trunc<mode>_stfiwx): Likewise.
5528         (fixuns_trunc<mode>_stfiwx): Likewise.
5529         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5530         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5531         (parity<mode>2_cmpb): Set length/type attr.
5532         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5533         for 'mr.' to fast_compare.
5534         (bpermd_<mode>): Change type attr to popcnt.
5535         (p8_fmrgow_<mode>): New insns for power8 direct move support.
5536         (p8_mtvsrwz_1): Likewise.
5537         (p8_mtvsrwz_2): Likewise.
5538         (reload_fpr_from_gpr<mode>): Likewise.
5539         (p8_mtvsrd_1): Likewise.
5540         (p8_mtvsrd_2): Likewise.
5541         (p8_xxpermdi_<mode>): Likewise.
5542         (reload_vsx_from_gpr<mode>): Likewise.
5543         (reload_vsx_from_gprsf): Likewise.
5544         (p8_mfvsrd_3_<mode>): LIkewise.
5545         (reload_gpr_from_vsx<mode>): Likewise.
5546         (reload_gpr_from_vsxsf): Likewise.
5547         (p8_mfvsrd_4_disf): Likewise.
5548         (multi-word GPR splits): Do not split direct moves or quad memory
5549         operations.
5551 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
5553         * tree-into-ssa.c (interesting_blocks): Make static.
5555 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
5557         PR target/56564
5558         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5559         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5560         Use DATA_ABI_ALIGNMENT for that case instead if defined.
5561         (get_variable_align): New function.
5562         (get_variable_section, emit_bss, emit_common,
5563         assemble_variable_contents, place_block_symbol): Use
5564         get_variable_align instead of DECL_ALIGN.
5565         (assemble_noswitch_variable): Add align argument, use it
5566         instead of DECL_ALIGN.
5567         (assemble_variable): Adjust caller.  Use get_variable_align
5568         instead of DECL_ALIGN.
5569         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5570         caller.
5571         (DATA_ABI_ALIGNMENT): Define.
5572         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5573         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
5574         opt is false, only return the psABI mandated alignment increase.
5575         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5576         (DATA_ABI_ALIGNMENT): ... this.
5577         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5578         (DATA_ABI_ALIGNMENT): ... this.
5579         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5580         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5581         (DATA_ABI_ALIGNMENT): ... this.
5582         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5583         * doc/tm.texi: Regenerated.
5585 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
5587         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5588         cmp_code to construct REG_EQUAL note.
5590 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
5592         PR target/57568
5593         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5594         that operands[2] doesn't overlap with operands[0].
5596 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
5597             Jan Hubicka  <jh@suse.cz>
5599         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5600         hack to mark symbols as used.
5602 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
5604         PR rtl-optimization/57559
5605         * lra-constraints.c (process_alt_operands): Don't discourage
5606         memory with known offset for offsetable memory constraint.
5607         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5609 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
5611         * varasm.c (struct oc_local_state): Reorder fields.
5612         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5613         and adjust accordingly.
5614         (output_constructor): Reorder initialization code and adjust call to
5615         output_constructor_bitfield.
5617 2013-06-07  Jan Hubicka  <jh@suse.cz>
5619         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5621 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
5623         * tree-object-size.c (unknown): Make const.
5625 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5627         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5628         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5629         for last alternative in the cpu_facility attribute.
5631 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5633         PR target/56315
5634         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5635         (xordi3): Change operand 2 constraint to arm_xordi_operand.
5636         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5637         * config/arm/constraints.md (Dg): New constraint.
5638         * config/arm/neon.md (xordi3_neon): Remove.
5639         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5640         * config/arm/predicates.md (arm_xordi_operand): New predicate.
5642 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5644         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5645         Clean up alternatives.
5647 2013-06-07  Alan Modra  <amodra@gmail.com>
5649         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5650         va_list_gpr_size.
5652 2013-06-07  Alan Modra  <amodra@gmail.com>
5654         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5656 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5658         * config/arm/constraints.md (Df): New constraint.
5659         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5660         Correct length attribute for last two alternatives.
5662 2013-06-07  Alan Modra  <amodra@gmail.com>
5664         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5665         override user -mfp-in-toc.
5666         (offsettable_ok_by_alignment): Consider just the current access
5667         rather than the whole object, unless BLKmode.  Handle
5668         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5669         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5670         for -mcmodel=medium.
5671         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5672         override user -mfp-in-toc or -msum-in-toc.  Default to
5673         -mno-fp-in-toc for -mcmodel=medium.
5675 2013-06-06  DJ Delorie  <dj@redhat.com>
5677         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5678         TARGET_VALID_POINTER_MODE.
5680 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
5681             Pat Haugen <pthaugen@us.ibm.com>
5682             Peter Bergner <bergner@vnet.ibm.com>
5684         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5685         Document new power8 builtins.
5687         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5688         condition code register, to allow 128-bit logical operations to be
5689         done in the VSX or GPR registers.
5690         (nor<mode>3): Use the canonical form for nor.
5691         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5692         vclz*, and vpopcnt* vector instructions.
5693         (nand<mode>3): Likewise.
5694         (orc<mode>3): Likewise.
5695         (clz<mode>2): LIkewise.
5696         (popcount<mode>2): Likewise.
5698         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5699         that only the GPRs are recognized.
5701         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5702         support for new power8 builtins.
5704         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5705         builtin functions.
5706         (xscvdpspn): Likewise.
5707         (vclz): Likewise.
5708         (vclzb): Likewise.
5709         (vclzh): Likewise.
5710         (vclzw): Likewise.
5711         (vclzd): Likewise.
5712         (vpopcnt): Likewise.
5713         (vpopcntb): Likewise.
5714         (vpopcnth): Likewise.
5715         (vpopcntw): Likewise.
5716         (vpopcntd): Likewise.
5717         (vgbbd): Likewise.
5718         (vmrgew): Likewise.
5719         (vmrgow): Likewise.
5720         (eqv): Likewise.
5721         (eqv_v16qi3): Likewise.
5722         (eqv_v8hi3): Likewise.
5723         (eqv_v4si3): Likewise.
5724         (eqv_v2di3): Likewise.
5725         (eqv_v4sf3): Likewise.
5726         (eqv_v2df3): Likewise.
5727         (nand): Likewise.
5728         (nand_v16qi3): Likewise.
5729         (nand_v8hi3): Likewise.
5730         (nand_v4si3): Likewise.
5731         (nand_v2di3): Likewise.
5732         (nand_v4sf3): Likewise.
5733         (nand_v2df3): Likewise.
5734         (orc): Likewise.
5735         (orc_v16qi3): Likewise.
5736         (orc_v8hi3): Likewise.
5737         (orc_v4si3): Likewise.
5738         (orc_v2di3): Likewise.
5739         (orc_v4sf3): Likewise.
5740         (orc_v2df3): Likewise.
5742         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5743         allow power8 quad mode in 64-bit.
5744         (rs6000_builtin_vectorized_function): Add support to vectorize
5745         ISA 2.07 count leading zeros, population count builtins.
5746         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5747         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5748         (builtin_function_type): Add vgbbd builtin function which takes an
5749         unsigned argument.
5750         (altivec_expand_vec_perm_const): Add support for new power8 merge
5751         instructions.
5753         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5754         that does not include TImdoe for use with 32-bit.
5755         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5756         instructions.
5757         (UNSPEC_VSX_CVDPSPN): Likewise.
5758         (vsx_xscvdpspn): Likewise.
5759         (vsx_xscvspdpn): Likewise.
5760         (vsx_xscvdpspn_scalar): Likewise.
5761         (vsx_xscvspdpn_directmove): Likewise.
5762         (vsx_and<mode>3): Split logical operations into 32-bit and
5763         64-bit. Add support to do logical operations on TImode as well as
5764         VSX vector types.  Allow logical operations to be done in either
5765         VSX registers or in general purpose registers in 64-bit mode.  Add
5766         splitters if GPRs were used. For AND, add clobber of CCmode to
5767         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
5768         encoding.
5769         (vsx_and<mode>3_32bit): Likewise.
5770         (vsx_and<mode>3_64bit): Likewise.
5771         (vsx_ior<mode>3): Likewise.
5772         (vsx_ior<mode>3_32bit): Likewise.
5773         (vsx_ior<mode>3_64bit): Likewise.
5774         (vsx_xor<mode>3): Likewise.
5775         (vsx_xor<mode>3_32bit): Likewise.
5776         (vsx_xor<mode>3_64bit): Likewise.
5777         (vsx_one_cmpl<mode>2): Likewise.
5778         (vsx_one_cmpl<mode>2_32bit): Likewise.
5779         (vsx_one_cmpl<mode>2_64bit): Likewise.
5780         (vsx_nor<mode>3): Likewise.
5781         (vsx_nor<mode>3_32bit): Likewise.
5782         (vsx_nor<mode>3_64bit): Likewise.
5783         (vsx_andc<mode>3): Likewise.
5784         (vsx_andc<mode>3_32bit): Likewise.
5785         (vsx_andc<mode>3_64bit): Likewise.
5786         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5787         and xxlorc instructions.
5788         (vsx_eqv<mode>3_64bit): Likewise.
5789         (vsx_nand<mode>3_32bit): Likewise.
5790         (vsx_nand<mode>3_64bit): Likewise.
5791         (vsx_orc<mode>3_32bit): Likewise.
5792         (vsx_orc<mode>3_64bit): Likewise.
5794         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5796         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5797         instruction.
5798         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5799         (p8_vmrgow): Likewise.
5800         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5801         GPRs to be split under VSX.
5802         (p8v_clz<mode>2): Add power8 count leading zero support.
5803         (p8v_popcount<mode>2): Add power8 population count support.
5804         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5805         support.
5807         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5808         instruction.
5810         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5811         builtin functions.
5812         (vec_nand): Likewise.
5813         (vec_vclz): Likewise.
5814         (vec_vclzb): Likewise.
5815         (vec_vclzd): Likewise.
5816         (vec_vclzh): Likewise.
5817         (vec_vclzw): Likewise.
5818         (vec_vgbbd): Likewise.
5819         (vec_vmrgew): Likewise.
5820         (vec_vmrgow): Likewise.
5821         (vec_vpopcnt): Likewise.
5822         (vec_vpopcntb): Likewise.
5823         (vec_vpopcntd): Likewise.
5824         (vec_vpopcnth): Likewise.
5825         (vec_vpopcntw): Likewise.
5827 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5829         PR rtl-optimization/57468
5830         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5831         spilled pseudos.
5833 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5835         PR rtl-optimization/57459
5836         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5837         type when setting live regs.
5839 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
5841         * config/s390/s390.opt (mlra): New option.
5842         * config/s390/s390.c (s390_decompose_address): Check displacement
5843         for all registers for LRA.
5844         (s390_secondary_reload): Don't used secondary reloads for LRA.
5845         (s390_lra_p): New function.
5846         (TARGET_LRA_P): Define.
5847         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5848         of attribute cpu_facility to zarch for the last alternative.
5849         (*cmpmem_short): Ditto.
5851 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
5853         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
5854         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
5855         (arm_expand_prologue): Likewise.
5857 2013-06-06  Teresa Johnson  <tejohnson@google.com>
5859         PR c++/53743
5860         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
5861         as this is now done by redirect_edge_and_branch_force.
5862         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
5863         barriers, and fix interaction with splitting.
5864         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
5865         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
5866         reflect changes made in the routine.
5867         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
5868         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
5869         since this is called in cfglayout mode, and replace partition fixup
5870         with assert as that is now done by force_nonfallthru_and_redirect.
5871         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
5872         already be marked with region crossing note.
5873         (insert_section_boundary_note): Make non-static, gate on flag
5874         has_bb_partition, rewrite to also check for multiple partitions.
5875         (rest_of_handle_reorder_blocks): Remove call to
5876         insert_section_boundary_note, now done later during free_cfg.
5877         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
5878         * bb-reorder.h (insert_section_boundary_note): Declare.
5879         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
5880         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
5881         invoke insert_section_boundary_note.
5882         (try_redirect_by_replacing_jump): Remove unnecessary
5883         check for region crossing note.
5884         (fixup_partition_crossing): New function.
5885         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
5886         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
5887         in non-cfglayout mode.
5888         (force_nonfallthru_and_redirect): Fixup partition boundaries,
5889         remove old code that tried to do this. Emit barrier correctly
5890         when we are in cfglayout mode.
5891         (last_bb_in_partition): New function.
5892         (rtl_split_edge): Correctly fixup partition boundaries.
5893         (commit_one_edge_insertion): Remove old code that tried to
5894         fixup region crossing edge since this is now handled in
5895         split_block, and set up insertion point correctly since
5896         block may now end in a jump.
5897         (verify_hot_cold_block_grouping): Guard against checking when not in
5898         linearized RTL mode.
5899         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
5900         notes.
5901         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
5902         rtl_verify_flow_info, so not called in cfglayout mode.
5903         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
5904         (fixup_reorder_chain): Remove old code that attempted to fixup region
5905         crossing note as this is now handled in force_nonfallthru_and_redirect.
5906         (duplicate_insn_chain): Don't duplicate switch section notes.
5907         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
5908         note.
5909         * basic-block.h (emit_barrier_after_bb): Declare.
5911 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5913         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
5914         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
5915         arm_usatsihi): Adjust alternatives for arm_restrict_it.
5917 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5919         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
5920         where appropriate.
5921         * config/arm/ldmstm.md: Regenerate.
5923 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5925         * config/arm/sync.md (atomic_loaddi_1):
5926         Disable predication for arm_restrict_it.
5927         (arm_load_exclusive<mode>): Likewise.
5928         (arm_load_exclusivesi): Likewise.
5929         (arm_load_exclusivedi): Likewise.
5930         (arm_load_acquire_exclusive<mode>): Likewise.
5931         (arm_load_acquire_exclusivesi): Likewise.
5932         (arm_load_acquire_exclusivedi): Likewise.
5933         (arm_store_exclusive<mode>): Likewise.
5934         (arm_store_exclusive<mode>): Likewise.
5935         (arm_store_release_exclusivedi): Likewise.
5936         (arm_store_release_exclusive<mode>): Likewise.
5938 2013-06-06  Richard Biener  <rguenther@suse.de>
5940         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
5941         after LTO_null.
5942         (lto_tag_is_tree_code_p): Adjust.
5943         (lto_tag_is_gimple_code_p): Likewise.
5944         (lto_gimple_code_to_tag): Likewise.
5945         (lto_tag_to_gimple_code): Likewise.
5946         (lto_tree_code_to_tag): Likewise.
5947         (lto_tag_to_tree_code): Likewise.
5948         * data-streamer.h (streamer_write_hwi_in_range): Use
5949         uhwi streaming to stream the normalized range.
5950         (streamer_read_hwi_in_range): Likewise.
5952 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5954         * config/arm/arm.md (enabled_for_depr_it): New attribute.
5955         (predicable_short_it): Likewise.
5956         (predicated): Likewise.
5957         (enabled): Handle above.
5958         (define_cond_exec): Set predicated attribute to yes.
5960 2013-06-05  Mike Stump  <mikestump@comcast.net>
5962         * gdbinit.in (__FUNCTION__): Add.
5964 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
5966         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
5967         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
5969 2013-06-05  Jan Hubicka  <jh@suse.cz>
5971         * varasm.c (mark_decl_referenced): Revert the removal until targets
5972         are fixed.
5974 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
5976         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
5977         instead of mark_decl_referenced.
5979 2013-06-05  Jan Hubicka  <jh@suse.cz>
5981         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
5982         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
5983         and symtab_used_from_object_file_p.
5984         (cgraph_make_node_local_1): Clear forced_by_abi.
5985         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
5986         * cgraph.h (symtab_node_base): Add forced_by_abi.
5987         (decide_is_variable_needed): Remove.
5988         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
5989         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
5990         (decide_is_symbol_needed): ... this one; handle symbols in general;
5991         always analyze virtuals; honnor forced_by_abi.
5992         (cgraph_finalize_function): Update.
5993         (varpool_finalize_decl): Update.
5994         (symbol_defined_and_needed): Remove.
5995         (analyze_functions): Update.
5996         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
5997         output_refs, input_overwrite_node): Handle forced_by_abi.
5998         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
5999         (address_taken_from_non_vtable_p): ... this one.
6000         (comdat_can_be_unshared_p_1): New function.
6001         (cgraph_comdat_can_be_unshared_p): Rename to ...
6002         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6003         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6004         (function_and_variable_visibility): Clear forced_by_abi as needed.
6005         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6006         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6007         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6008         * varpool.c (decide_is_variable_needed): Remove.
6010 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6012         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6013         (arm_option_override): Override arm_restrict_it where appropriate.
6014         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6015         * config/arm/arm.opt (mrestrict-it): New command-line option.
6016         * doc/invoke.texi: Document -mrestrict-it.
6018 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6020         * tsan.c (tsan_atomic_table): Make const.
6022 2013-06-05  Richard Biener  <rguenther@suse.de>
6024         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6025         index associated with the tree we are supposed to replace.
6026         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6027         TREE_ASM_WRITTEN as zero for everything but SSA names.
6029 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6031         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6033 2013-06-04  Jan Hubicka  <jh@suse.cz>
6035         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6036         (input_node, input_varpool_node): Handle correctly external same
6037         body aliases.
6038         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6039         nodes at ltrans stage.
6041 2013-06-04  Jan Hubicka  <jh@suse.cz>
6043         * ipa-inline.c (update_caller_keys): Fix availability test.
6044         (update_callee_keys): Likewise.
6045         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6046         to follow ELF standard.
6048 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6050         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6051         (mips64r5900el-*-elf*): New configurations.
6052         * config/mips/mips-cpus.def (r5900): New processor.
6053         * config/mips/mips-tables.opt: Regenerate.
6054         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6055         (mips_issue_rate): Handle PROCESSOR_R5900.
6056         (mips_reorg_process_insns): Force reorder mode for the R5900.
6057         * config/mips/mips.h (TARGET_MIPS5900): Define.
6058         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6059         TARGET_MIPS5900.
6060         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6061         TARGET_MIPS5900.
6062         * config/mips/mips.md (processor): Add r5900.
6063         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6065 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6067         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6068         into function to generate MOVI instruction.
6069         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6070         (aarch64_preferred_simd_mode): Turn into wrapper.
6071         (aarch64_output_scalar_simd_mov_immediate): New function.
6072         * config/aarch64/aarch64-protos.h: Add prototype for above.
6074 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6076         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6077         element_char member.
6078         (sizetochar): Return signed char.
6079         (aarch64_simd_valid_immediate): Remove elchar and other
6080         unnecessary variables.
6081         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6082         Calculate element_char as required.
6083         * config/aarch64/aarch64-protos.h: Update and move prototype
6084         for aarch64_output_simd_mov_immediate.
6085         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6086         Update arguments.
6088 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6090         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6091         information completed by aarch64_simd_valid_immediate.
6092         (aarch64_legitimate_constant_p): Update arguments.
6093         (aarch64_simd_valid_immediate): Work with struct rather than many
6094         pointers.
6095         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6096         (aarch64_simd_make_constant): Update arguments.
6097         (aarch64_output_simd_mov_immediate): Work with struct rather than
6098         many pointers.  Output immediate directly rather than as operand.
6099         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6100         Update prototype.
6101         * config/aarch64/constraints.md (Dn): Update arguments.
6103 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6105         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6106         longer static.
6107         (aarch64_simd_immediate_valid_for_move): Remove.
6108         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6109         (aarch64_simd_make_constant): Update call.
6110         (aarch64_output_simd_mov_immediate): Update call.
6111         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6112         Add prototype.
6113         * config/aarch64/constraints.md (Dn): Update call.
6115 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6117         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6118         return type to bool for prototype.
6119         (aarch64_legitimate_constant_p): Check for true instead of not -1.
6120         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6121         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6123 2013-06-04  Catherine Moore  <clm@codesourcery.com>
6125         * config/mips/mips.opt (meva): New.
6126         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6127         (ASM_SPEC): Handle -meva.
6128         * doc/invoke.texi (meva):  Document.
6130 2013-06-04  Alan Modra  <amodra@gmail.com>
6132         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6133         constant output.
6135 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6137         * rtl.def: Add extra fourth optional field to define_cond_exec.
6138         * gensupport.c (process_one_cond_exec): Process attributes from
6139         define_cond_exec.
6140         * doc/md.texi: Document fourth field in define_cond_exec.
6142 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
6144         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6145         out the processing order as in store_bit_field_1.
6147 2013-06-04  Jan Hubicka  <jh@suse.cz>
6149         PR middle-end/57500
6150         * cgraphunit.c (cgraph_process_same_body_aliases): Create
6151         non-VAR_DECL node if it does not exist yet.
6153 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
6155         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6156         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6157         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6158         target_cpu_default setting.
6160 2013-06-03  Teresa Johnson  <tejohnson@google.com>
6162         * dumpfile.c (opt_info_switch_p): Change -fopt-info
6163         default to -fopt-info=optimized instead of all.
6164         * doc/invoke.texi: Ditto.
6165         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6166         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6167         (execute_vect_slp): Emit BB vectorization success under
6168         MSG_OPTIMIZED_LOCATIONS.
6169         * tree-vect-slp.c (vect_slp_transform_bb): Change
6170         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6171         * tree-vect-loop.c (vect_transform_loop): Ditto.
6173 2013-06-03  Jason Merrill  <jason@redhat.com>
6175         PR c++/57415
6176         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6177         Use TARGET_EXPR for C++.
6179 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
6181         PR rtl-optimization/57268
6182         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6183         if DEBUG_INSN_P (insn).
6185         Reapply
6186         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6188         PR rtl-optimization/57268
6189         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6190         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6192 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6194         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6195         (ix86_avoid_lea_for_addr): Likewise.
6196         (exact_dependency_1): Likewise.
6197         (ix86_adjust_cost): Likewise.
6198         (swap_top_of_ready_list): Fix formatting and !reload_completed check
6199         removed.
6200         (do_reorder_for_imul): Fix typo, formatting and
6201         !reload_completed check removed.
6202         (ix86_sched_reorder): Fix typo and formatting.
6203         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6204         list.
6206 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
6208         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6210 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
6212         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6213         <STRING_CST>: Likewise.
6214         <VECTOR_CST>: Likewise.
6216 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
6217             Mikael Morin  <mikael@gcc.gnu.org>
6219         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6220         * config.in: Regenerated.
6221         * configure: Regenerated.
6223 2013-06-01  Jan Hubicka  <jh@suse.cz>
6225         PR middle-end/57366
6226         * cgraphunit.c (compile): When weakref is not supported,
6227         set up transparent aliases before final output pass.
6228         * varasm.c (assemble_alias): Do not try to do it here.
6230 2013-06-01  Jan Hubicka  <jh@suse.cz>
6232         PR middle-end/57467
6233         * passes.c (for_per_function): Skip unanalyzed functions.
6235 2013-06-01  Jan Hubicka  <jh@suse.cz>
6237         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6238         (lto_symtab_merge_symbols_1): ... this one.
6239         (lto_symtab_merge_cgraph_nodes): Rename to ...
6240         (lto_symtab_merge_symbols): ... this one; simplify.
6241         * cgraph.c (same_body_aliases_done): Rename to ...
6242         (cpp_implicit_aliases_done): ... this one.
6243         (cgraph_create_function_alias): Update.
6244         (cgraph_same_body_alias): Update.
6245         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6246         (verify_edge_corresponds_to_fndecl): Simplify.
6247         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6248         (cgraph_node): Remove same_body_alias.
6249         (varpool_node): Remove alias_of and extra_name_alias.
6250         (same_body_aliases_done): Rename to ..
6251         (cpp_implicit_aliases_done): ... this one.
6252         (symtab_alias_ultimate_target): Add default parameter.
6253         (symtab_resolve_alias): New function.
6254         (fixup_same_cpp_alias_visibility): Declare.
6255         (cgraph_function_node): Add default parameter.
6256         (cgraph_node_asm_name): Likewise.
6257         (cgraph_function_or_thunk_node): Add default parameter; do
6258         not ICE when it is NULL.
6259         (varpool_variable_node): Likewise.
6260         * tree-emutls.c (create_emultls_var): Update.
6261         (ipa_lower_emutls): Update.
6262         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6263         (cgraph_reset_node): Reset alias info.
6264         (cgraph_finalize_function): Update.
6265         (fixup_same_cpp_alias_visibility): Move to symtab.c.
6266         (analyze_function): Simplify.
6267         (cgraph_process_same_body_aliases): Simplify.
6268         (analyze_functions): Fixup same body aliases.
6269         (handle_alias_pairs): Simplify.
6270         (assemble_thunk): Update.
6271         (assemble_thunks_and_aliases): Update.
6272         (output_weakrefs): Rewrite.
6273         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6274         (lto_output_varpool_node): Likewise.
6275         (compute_ltrans_boundary): Remve assert.
6276         (get_alias_symbol): New functoin.
6277         (input_node): Rewrite alias handling.
6278         (input_varpool_node): Likewise.
6279         * ipa-pure-const.c (propagate_pure_const): Fix formating.
6280         * ipa.c (process_references): Handle weakrefs correctly.
6281         (symtab_remove_unreachable_nodes): Likewise.
6282         * trans-mem.c (get_cg_data): Update.
6283         (ipa_tm_create_version_alias): Update.
6284         (ipa_tm_execute): Update.
6285         * symtab.c (dump_symtab_base): Dump aliases.
6286         (verify_symtab_base): Verify aliases.
6287         (symtab_node_availability): New function.
6288         (symtab_alias_ultimate_target): Simplify.
6289         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6290         handle all the fixup cases.
6291         (symtab_resolve_alias): New function.
6292         * passes.c (ipa_write_summaries): Handle weakrefs.
6293         * varpool.c (varpool_analyze_node): Simplify.
6294         (assemble_aliases): Update.
6295         (varpool_create_variable_alias): Simplify.
6296         (varpool_extra_name_alias): Simplify.
6297         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6298         (lto_symtab_merge_symbols): ... this one.
6300 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
6302         Revert
6303         PR rtl-optimization/57268
6304         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6305         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6307 2013-06-01  Tobias Burnus  <burnus@net-b.de>
6309         Partially reverted:
6310         2013-05-31  Tobias Burnus  <burnus@net-b.de>
6312         PR middle-end/57073
6313         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6314         further up.
6316 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
6318         PR rtl-optimization/57268
6319         * sched-deps.c (sched_analyze_2): Flush dependence lists if
6320         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6322 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6324         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6325         unordered comparison operators when -fno-trapping-math is in effect
6326         on the e500.
6327         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6328         and implement unordered comparison operators properly on the e500.
6330 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
6332         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6333         for constant scalar integers.
6334         (simplify_relational_operation_1): Likewise.
6336 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
6338         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6339         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
6340         Fix comment.
6342 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6343             Igor Zamyatin  <igor.zamyatin@intel.com>
6345         Silvermont (SLM) architecture performance tuning.
6346         * config/i386/i386.h (enum ix86_tune_indices): Add
6347         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6348         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6350         * config/i386/i386.c (initial_ix86_tune_features)
6351         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6352         (ix86_lea_outperforms): Handle Silvermont tuning.
6353         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6354         call.
6355         (ix86_use_lea_for_mov): Likewise.
6356         (ix86_avoid_lea_for_addr): Likewise.
6357         (ix86_lea_for_add_ok): Likewise.
6358         (exact_dependency_1): New function.
6359         (exact_store_load_dependency): Likewise.
6360         (ix86_adjust_cost): Handle Silvermont tuning.
6361         (do_reoder_for_imul): Likewise.
6362         (swap_top_of_ready_list): New function.
6363         (ix86_sched_reorder): Changed to handle Silvermont tuning.
6365         * config/i386/i386.md (peepholes that split memory operand in fp
6366         converts): New.
6368 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6370         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6371         Remove un-necessary braces.
6373 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6375         * config/aarch64/aarch64.c (aarch64_classify_symbol):
6376         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6378 2013-05-31  Tobias Burnus  <burnus@net-b.de>
6380         PR middle-end/57073
6381         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6383 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6385         PR target/56315
6386         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6387         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6388         * config/arm/neon.md (iordi3_neon): Remove.
6389         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6390         * config/arm/predicates.md (imm_for_neon_logic_operand):
6391         Move to earlier in the file.
6392         (neon_logic_op2): Likewise.
6393         (arm_iordi_operand_neon): New predicate.
6395 2013-05-31  Richard Biener  <rguenther@suse.de>
6397         PR tree-optimization/57478
6398         PR tree-optimization/57453
6399         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6400         are life as well.
6402 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6404         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6405         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6407 2013-05-30  Tobias Burnus  <burnus@net-b.de>
6408             Thomas Koenig  <tkoenig@gcc.gnu.org>
6410         PR middle-end/57073
6411         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6412         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6414 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
6416         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6418 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
6420         * target.def (register_usage_leveling_p): New hook.
6421         * targhooks.c (default_register_usage_leveling_p): New.
6422         * targhooks.h (default_register_usage_leveling_p): New prototype.
6423         * lra-assigns.c (register_usage_leveling_p): Use the hook.
6424         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6425         * doc/tm.texi: Update.
6426         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6428 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
6430         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6431         (*insv_reg<mode>): New define_insn.
6433 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
6435         PR rtl-optimization/57439
6436         * postreload.c (move2add_valid_value_p): Check that we have
6437         a zero subreg_regno_offset when accessing the register in
6438         the requested mode.
6440 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
6441             Igor Zamyatin  <igor.zamyatin@intel.com>
6443         Silvermont (SLM) architecture pipeline model, tuning and
6444         insn selection.
6445         * config.gcc: Add slm config options and target.
6447         * config/i386/slm.md: New.
6449         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6451         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
6452         PROCESSOR_SLM.
6453         (ix86_target_macros_internal): Likewise.
6455         * gcc/config/i386/i386.c (slm_cost): New cost.
6456         (m_SLM): New macro flag.
6457         (initial_ix86_tune_features): Set m_SLM.
6458         (x86_accumulate_outgoing_args): Likewise.
6459         (x86_arch_always_fancy_math_387): Likewise.
6460         (processor_target_table): Add slm cost.
6461         (cpu_names): Add slm cpu name.
6462         (x86_option_override_internal): Set SLM ISA.
6463         (ix86_issue_rate): New case PROCESSOR_SLM.
6464         (ia32_multipass_dfa_lookahead): Likewise.
6465         (fold_builtin_cpu): Add slm.
6467         * config/i386/i386.h (TARGET_SLM): New target macro.
6468         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6469         (processor_type): Add PROCESSOR_SLM.
6471         * config/i386/i386.md (cpu): Add new value "slm".
6472         (slm.md): Include slm.md.
6474 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
6475             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6477         * config/arm/arm-protos.h: Add and update function protos.
6478         * config/arm/arm.c (use_simple_return_p): New added.
6479         (thumb2_expand_return): Check simple_return flag.
6480         * config/arm/arm.md: Add simple_return and conditional simple_return.
6481         * config/arm/iterators.md: Add iterator for return and simple_return.
6483 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6485         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6486         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6487         (arm_emit_vfp_multi_reg_pop): Likewise.
6488         (thumb2_emit_ldrd_pop): Likewise.
6489         (arm_expand_epilogue): Add misc REG_CFA notes.
6490         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6492 2013-05-29  Lawrence Crowl  <crowl@google.com>
6494         * config/arm/t-arm: Update for below.
6496         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6497         Change type to hash_table.  Update dependent calls and types.
6499         * config/i386/t-cygming: Update for below.
6501         * config/i386/t-interix: Update for below.
6503         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6504         Change type to hash_table.  Update dependent calls and types.
6505         (i386_find_on_wrapper_list::wrappers): Likewise.
6507         * config/ia64/t-ia64: Update for below.
6509         * config/ia64/ia64.c (bundle_state_table):
6510         Change type to hash_table.  Update dependent calls and types.
6512         * config/mips/mips.c (mips_reorg_process_insns::htab):
6513         Change type to hash_table.  Update dependent calls and types.
6515         * config/sol2.c (solaris_comdat_htab):
6516         Change type to hash_table.  Update dependent calls and types.
6518         * config/t-sol2: Update for above.
6520 2013-05-29  Teresa Johnson  <tejohnson@google.com>
6522         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6523         functions are not yet marked as defined.
6525 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
6526             Pat Haugen <pthaugen@us.ibm.com>
6527             Peter Bergner <bergner@vnet.ibm.com>
6529         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6530         instructions.
6531         (VEC_A): Likewise.
6532         (VEC_C): Likewise.
6533         (vrotl<mode>3): Likewise.
6534         (vashl<mode>3): Likewise.
6535         (vlshr<mode>3): Likewise.
6536         (vashr<mode>3): Likewise.
6538         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6539         support for power8 V2DI builtins.
6541         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6542         power8 V2DI builtins.
6543         (vupkhsw): Likewise.
6544         (vupklsw): Likewise.
6545         (vaddudm): Likewise.
6546         (vminsd): Likewise.
6547         (vmaxsd): Likewise.
6548         (vminud): Likewise.
6549         (vmaxud): Likewise.
6550         (vpkudum): Likewise.
6551         (vpksdss): Likewise.
6552         (vpkudus): Likewise.
6553         (vpksdus): Likewise.
6554         (vrld): Likewise.
6555         (vsld): Likewise.
6556         (vsrd): Likewise.
6557         (vsrad): Likewise.
6558         (vsubudm): Likewise.
6559         (vcmpequd): Likewise.
6560         (vcmpgtsd): Likewise.
6561         (vcmpgtud): Likewise.
6562         (vcmpequd_p): Likewise.
6563         (vcmpgtsd_p): Likewise.
6564         (vcmpgtud_p): Likewise.
6565         (vupkhsw): Likewise.
6566         (vupklsw): Likewise.
6567         (vaddudm): Likewise.
6568         (vmaxsd): Likewise.
6569         (vmaxud): Likewise.
6570         (vminsd): Likewise.
6571         (vminud): Likewise.
6572         (vpksdss): Likewise.
6573         (vpksdus): Likewise.
6574         (vpkudum): Likewise.
6575         (vpkudus): Likewise.
6576         (vrld): Likewise.
6577         (vsld): Likewise.
6578         (vsrad): Likewise.
6579         (vsrd): Likewise.
6580         (vsubudm): Likewise.
6582         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6583         support for power8 V2DI instructions.
6585         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6586         power8 V2DI instructions.  Combine pack and unpack insns to use an
6587         iterator for each mode.  Check whether a particular mode supports
6588         Altivec instructions instead of just checking TARGET_ALTIVEC.
6589         (UNSPEC_VPKUWUM): Likewise.
6590         (UNSPEC_VPKSHSS): Likewise.
6591         (UNSPEC_VPKSWSS): Likewise.
6592         (UNSPEC_VPKUHUS): Likewise.
6593         (UNSPEC_VPKSHUS): Likewise.
6594         (UNSPEC_VPKUWUS): Likewise.
6595         (UNSPEC_VPKSWUS): Likewise.
6596         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6597         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6598         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6599         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6600         (UNSPEC_VUPKHSB): Likewise.
6601         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6602         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6603         (UNSPEC_VUPKHSH): Likewise.
6604         (UNSPEC_VUPKLSB): Likewise.
6605         (UNSPEC_VUPKLSH): Likewise.
6606         (VI2): Likewise.
6607         (VI_char): Likewise.
6608         (VI_scalar): Likewise.
6609         (VI_unit): Likewise.
6610         (VP): Likewise.
6611         (VP_small): Likewise.
6612         (VP_small_lc): Likewise.
6613         (VU_char): Likewise.
6614         (add<mode>3): Likewise.
6615         (altivec_vaddcuw): Likewise.
6616         (altivec_vaddu<VI_char>s): Likewise.
6617         (altivec_vadds<VI_char>s): Likewise.
6618         (sub<mode>3): Likewise.
6619         (altivec_vsubcuw): Likewise.
6620         (altivec_vsubu<VI_char>s): Likewise.
6621         (altivec_vsubs<VI_char>s): Likewise.
6622         (altivec_vavgs<VI_char>): Likewise.
6623         (altivec_vcmpbfp): Likewise.
6624         (altivec_eq<mode>): Likewise.
6625         (altivec_gt<mode>): Likewise.
6626         (altivec_gtu<mode>): Likewise.
6627         (umax<mode>3): Likewise.
6628         (smax<mode>3): Likewise.
6629         (umin<mode>3): Likewise.
6630         (smin<mode>3): Likewise.
6631         (altivec_vpkuhum): Likewise.
6632         (altivec_vpkuwum): Likewise.
6633         (altivec_vpkshss): Likewise.
6634         (altivec_vpkswss): Likewise.
6635         (altivec_vpkuhus): Likewise.
6636         (altivec_vpkshus): Likewise.
6637         (altivec_vpkuwus): Likewise.
6638         (altivec_vpkswus): Likewise.
6639         (altivec_vpks<VI_char>ss): Likewise.
6640         (altivec_vpks<VI_char>us): Likewise.
6641         (altivec_vpku<VI_char>us): Likewise.
6642         (altivec_vpku<VI_char>um): Likewise.
6643         (altivec_vrl<VI_char>): Likewise.
6644         (altivec_vsl<VI_char>): Likewise.
6645         (altivec_vsr<VI_char>): Likewise.
6646         (altivec_vsra<VI_char>): Likewise.
6647         (altivec_vsldoi_<mode>): Likewise.
6648         (altivec_vupkhsb): Likewise.
6649         (altivec_vupkhs<VU_char>): Likewise.
6650         (altivec_vupkls<VU_char>): Likewise.
6651         (altivec_vupkhsh): Likewise.
6652         (altivec_vupklsb): Likewise.
6653         (altivec_vupklsh): Likewise.
6654         (altivec_vcmpequ<VI_char>_p): Likewise.
6655         (altivec_vcmpgts<VI_char>_p): Likewise.
6656         (altivec_vcmpgtu<VI_char>_p): Likewise.
6657         (abs<mode>2): Likewise.
6658         (vec_unpacks_hi_v16qi): Likewise.
6659         (vec_unpacks_hi_v8hi): Likewise.
6660         (vec_unpacks_lo_v16qi): Likewise.
6661         (vec_unpacks_hi_<VP_small_lc>): Likewise.
6662         (vec_unpacks_lo_v8hi): Likewise.
6663         (vec_unpacks_lo_<VP_small_lc>): Likewise.
6664         (vec_pack_trunc_v8h): Likewise.
6665         (vec_pack_trunc_v4si): Likewise.
6666         (vec_pack_trunc_<mode>): Likewise.
6668         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6669         V2DI builtins.
6670         (vec_vmaxsd): Likewise.
6671         (vec_vmaxud): Likewise.
6672         (vec_vminsd): Likewise.
6673         (vec_vminud): Likewise.
6674         (vec_vpksdss): Likewise.
6675         (vec_vpksdus): Likewise.
6676         (vec_vpkudum): Likewise.
6677         (vec_vpkudus): Likewise.
6678         (vec_vrld): Likewise.
6679         (vec_vsld): Likewise.
6680         (vec_vsrad): Likewise.
6681         (vec_vsrd): Likewise.
6682         (vec_vsubudm): Likewise.
6683         (vec_vupkhsw): Likewise.
6684         (vec_vupklsw): Likewise.
6686 2013-05-29  Jan Hubicka  <jh@suse.cz>
6688         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6689         flags; reorder rest of fields in more consistent way.
6690         (varpool_node): Remove analyzed, finalized and alias.
6691         (cgraph_ndoe): Likewise.
6692         (symtab_alias_ultimate_target): New function.
6693         (cgraph_function_node): Move offline.
6694         (cgraph_reset_node): Declare.
6695         (cgraph_comdat_can_be_unshared_p): Remove.
6696         (varpool_remove_initializer): Declare.
6697         (varpool_first_defined_variable, varpool_next_defined_variable
6698         cgraph_first_defined_function, cgraph_next_defined_function): Update.
6699         (cgraph_function_with_gimple_body_p): Update.
6700         (varpool_all_refs_explicit_p): Update.
6701         (symtab_alias_target): New function.
6702         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6703         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
6704         (cgraph_function_or_thunk_node): Simplify using
6705         symtab_alias_ultimate_target.
6706         (varpool_variable_node): Likewise.
6707         * cgraph.c (cgraph_create_function_alias): Update.
6708         (cgraph_add_thunk): Update.
6709         (cgraph_remove_node): Update.
6710         (dump_cgraph_node): Do not dump removed flags.
6711         (cgraph_function_body_availability): Update.
6712         (cgraph_propagate_frequency): Update.
6713         (verify_cgraph_node): Check sanity of local flag.
6714         (cgraph_function_node): Move here from cgraph.h; revamp for
6715         cgraph_function_or_thunk_node.
6716         * lto-symtab.c (lto_varpool_replace_node): Update.
6717         (lto_symtab_resolve_can_prevail_p): Update.
6718         (lto_symtab_merge_cgraph_nodes): Update.
6719         * ipa-cp.c (determine_versionability, initialize_node_lattices,
6720         propagate_constants_accross_call, devirtualization_time_bonus,
6721         ipcp_propagate_stage): Update.
6722         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
6723         * ipa-inline-transform.c (clone_inlined_nodes,
6724         preserve_function_body_p): Update.
6725         * ipa-reference.c (propagate): Update.
6726         (write_node_summary_p): Update.
6727         * toplev.c (wrapup_global_declaration_2): Update.
6728         * cgraphunit.c (cgraph_analyze_function): Rename to ...
6729         (analyze_function) ... this one.
6730         (cgraph_process_new_functions): Update.
6731         (cgraph_reset_node): Export.
6732         (cgraph_finalize_function): Update.
6733         (cgraph_add_new_function): Update.
6734         (process_function_and_variable_attributes): Update.
6735         (varpool_finalize_decl): Update.
6736         (symbol_finalized): Remove.
6737         (symbol_finalized_and_needed): Rename to ...
6738         (symbol_defined_and_needed): ... update.
6739         (cgraph_analyze_functions): Update.
6740         (handle_alias_pairs): Update.
6741         (mark_functions_to_output): Update.
6742         (assemble_thunk): Update.
6743         (output_in_order): Update.
6744         (output_weakrefs): Update.
6745         (finalize_compilation_unit): Update.
6746         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6747         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6748         input_node, input_varpool_node): Update.
6749         * dbxout.c (dbxout_expand_expr): Update.
6750         * cgraphclones.c (cgraph_clone_node): Update.
6751         (cgraph_copy_node_for_versioning): Update.
6752         (cgraph_materialize_clone): Update.
6753         (cgraph_materialize_all_clones): Update.
6754         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6755         propagate_pure_const, propagate_nothrow): Update.
6756         * lto-streamer-out.c (lto_output, write_symbol): Update.
6757         * ipa-utils.c (ipa_reverse_postorder): Update.
6758         * ipa-inline.c (can_inline_edge_p): Update.
6759         (update_caller_keys, ipa_inline): Update.
6760         * dwarf2out.c (reference_to_unused,
6761         premark_types_used_by_global_vars_helper): Update.
6762         * tree-eh.c (tree_could_trap_p): Update.
6763         * ipa-split.c (consider_split, execute_split_functions): Update.
6764         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
6765         has_addr_references_p): Update; move ahead in file for better
6766         readability.
6767         (process_references): Simplify.
6768         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6769         bodies are removed.
6770         (cgraph_comdat_can_be_unshared_p): Make static.
6771         (cgraph_externally_visible_p): Update.
6772         (varpool_externally_visible_p): Update.
6773         (function_and_variable_visibility): Update.
6774         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6775         ipa_tm_mark_force_output_node): Update.
6776         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6777         estimate_edge_devirt_benefit, inline_generate_summary,
6778         inline_write_summary): Update.
6779         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6780         * ipa-prop.c (ipa_compute_jump_functions): Update.
6781         (ipa_print_node_params, ipa_prop_read_section,
6782         ipa_update_after_lto_read, read_replacements_section): Update.
6783         * varasm.c (mark_decl_referenced): Update.
6784         (assemble_alias, dump_tm_clone_pairs): Update.
6785         * tree-inline.c (copy_bb): Update.
6786         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6787         Update.
6788         * symtab.c (dump_symtab_base): Print new flags.
6789         (verify_symtab_base): Verify new flags.
6790         (symtab_alias_ultimate_target): New function.
6791         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6792         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6793         Update.
6794         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6795         Update.
6796         * i386.c (ix86_get_function_versions_dispatcher,
6797         ix86_generate_version_dispatcher_body): Update.
6798         (fold_builtin_cpu): Use varpool_add_new_variable.
6799         * varpool.c (varpool_remove_initializer): Break out from ...
6800         (varpool_remove_node): ... this one.
6801         (dump_varpool_node, varpool_node_for_asm,
6802         cgraph_variable_initializer_availability, varpool_analyze_node,
6803         varpool_assemble_decl, varpool_remove_unreferenced_decls,
6804         varpool_finalize_named_section_flags, varpool_create_variable_alias):
6805         Update.
6807 2013-05-29  Jan Hubicka  <jh@suse.cz>
6809         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6811 2013-05-29  Easwaran Raman  <eraman@google.com>
6813         PR tree-optimization/57442
6814         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6815         when control exits the main loop.
6817 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
6819         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6820         and RX600.
6821         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
6822         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6823         * rx/t-rx: Add rx100 under multi library matches option for nofpu
6824         option.
6826 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6828         PR tree-optimization/57441
6829         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6830         Don't limit size of incr_vec to number of candidates.
6832 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
6834         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6835         and mips16 directories.
6836         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
6837         (MULTILIB_DIRNAMES): Ditto.
6838         (MULTILIB_EXCEPTIONS): Add new exceptions.
6839         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6840         (MULTILIB_DIRNAMES): Ditto.
6841         (MULTILIB_EXCEPTIONS): Add new exceptions.
6843 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6844             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6846         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6847         SYMBOL_TINY_ABSOLUTE.
6848         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
6849         SYMBOL_TINY_ABSOLUTE.
6850         (aarch64_expand_mov_immediate): Likewise.
6851         (aarch64_classify_symbol): Likewise.
6852         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
6853         Permit SYMBOL_TINY_ABSOLUTE.
6854         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
6856 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6857             Marcus Shawcroft  <marcus.shawcroft@arm.com>
6859         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
6860         Refactor if/switch.  Replace gcc_assert with if.
6862 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
6864         * config/i386/i386.c (initial_ix86_tune_features): Enable
6865         FP Reassociation for AMD bdver1 and bdver2.
6867 2013-05-29  Martin Jambor  <mjambor@suse.cz>
6869         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
6870         and IMAGPART_EXPR do not occur within other handled_components.
6872 2013-05-29  Richard Biener  <rguenther@suse.de>
6874         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
6875         access on whether the use is in the BB we currently try to
6876         vectorize.
6877         (vect_bb_vectorization_profitable_p): Pass the BB we currently
6878         vectorize to vect_bb_slp_scalar_cost.
6880 2013-05-29  Richard Biener  <rguenther@suse.de>
6882         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
6883         computing scalar cost offsetted by stmts that are kept live
6884         by scalar uses.
6885         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
6886         for computation of scalar cost.
6888 2013-05-28  Steve Ellcey  <sellcey@mips.com>
6890         * config/mips/mips-cpus.def (mips32r2): Change processor type.
6892 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6894         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
6895         array notation built-in reduction functions.
6896         * doc/passes.texi (Passes): Added documentation about changes done
6897         for Cilk Plus.
6898         * doc/invoke.texi (C Dialect Options): Added documentation about
6899         the -fcilkplus flag.
6900         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
6901         (BUILTINS_DEF): Depend on cilkplus.def.
6902         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
6903         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
6904         * cilkplus.def: New file.
6906 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
6908         PR rtl-optimization/57439
6909         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
6911 2013-05-28  Easwaran Raman  <eraman@google.com>
6913         PR tree-optimization/57337
6914         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
6915         (find_insert_point): Correctly identify the insertion point
6916         when two statements with the same UID is compared.
6918 2013-05-28  Richard Biener  <rguenther@suse.de>
6920         PR tree-optimization/56787
6921         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
6922         from the list of data references.
6923         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
6924         clobbers.
6925         (vect_analyze_loop_operations): Likewise.
6926         (vect_transform_loop): Remove clobbers.
6928 2013-05-28  Martin Jambor  <mjambor@suse.cz>
6930         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
6931         and REALPART_EXPRs have scalar type.
6933 2013-05-28  Richard Biener  <rguenther@suse.de>
6935         PR tree-optimization/57411
6936         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
6937         virtual operands.
6938         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
6939         virtual operand propagation.
6941 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6943         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
6944         destination register for bmasksi_vis.
6945         (vector_init_bshuffle): Likewise.
6946         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
6948 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
6950         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
6951         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
6952         mode if the instruction isn't available in the original mode.
6953         * config/sparc/sparc.opt (mfix-ut699): New option.
6954         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
6955         (divdf3): Turn into expander.
6956         (divdf3_nofix): New insn.
6957         (divdf3_fix): Likewise.
6958         (divsf3): Disable if -mfix-ut699.
6959         (sqrtdf2): Turn into expander.
6960         (sqrtdf2_nofix): New insn.
6961         (sqrtdf2_fix): Likewise.
6962         (sqrtsf2): Disable if -mfix-ut699.
6964 2013-05-27  Richard Biener  <rguenther@suse.de>
6966         PR middle-end/57412
6967         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
6968         block for the new loop.
6970 2013-05-27  Richard Biener  <rguenther@suse.de>
6972         PR tree-optimization/57343
6973         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
6974         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
6975         (number_of_iterations_cond): Do not build the folded tree.
6977 2013-05-27  Richard Biener  <rguenther@suse.de>
6979         Revert
6980         PR middle-end/57381
6981         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
6982         OEP_CONSTANT_ADDRESS_OF retained.
6984         PR tree-optimization/57417
6985         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
6986         for unchanged base.
6987         (set_ssa_val_to): Compare addresses using
6988         get_addr_base_and_unit_offset.
6990 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
6992         PR rtl-optimization/56833
6993         * postreload.c (move2add_record_mode): New function.
6994         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
6995         (move2add_use_add2_insn): Use move2add_record_sym_value.
6996         (move2add_use_add3_insn): Likewise.
6997         (reload_cse_move2add): Use move2add_valid_value_p and
6998         move2add_record_mode.  Invalidate call-clobbered and REG_INC
6999         affected regs by setting reg_mode to VOIDmode.
7000         (move2add_note_store): Don't pretend the inside of a SUBREG is
7001         the actual destination.  Invalidate single/leading registers by
7002         setting reg_mode to VOIDmode.
7003         Use move2add_record_sym_value, move2add_valid_value_p and
7004         move2add_record_mode.
7006 2013-05-27  Richard Biener  <rguenther@suse.de>
7008         PR tree-optimization/57396
7009         * tree-affine.c (double_int_constant_multiple_p): Properly
7010         return false for val == 0 and div != 0.
7012 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7014         * config/mips/mips.h: Use #elif in preprocessor conditions.
7016 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7018         PR target/53916
7019         * config/mips/constraints.md (kl): New constraint.
7020         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7021         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7022         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7023         from LO for MIPS16.
7024         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7026 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7028         PR target/55777
7029         * config/mips/mips.c (mips_can_inline_p): New function.
7030         (TARGET_CAN_INLINE_P): Define.
7032 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7034         * sched-int.h (ds_t, dw_t): Make unsigned int.
7035         Fix documentation that describes how all the ds_t bits are used.
7036         Reserve the last bit for delayed-branch scheduling.
7037         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7038         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7039         (gen_dep_weak_1): Remove prototype.
7040         * sched-deps.c (get_dep_weak_1): Make static.
7041         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7042         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7043         * doc/tm.texi: Regenerate.
7044         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7046 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7048         PR debug/56950
7049         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7051 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7052             Sandra Loosemore  <sandra@codesourcery.com>
7054         * config.gcc (powerpc-*): Allow native for with-cpu.
7056 2013-05-24  Jeff Law  <law@redhat.com>
7058         PR tree-optimization/57124
7059         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7060         conversion feeding a condition if the range has an overflow
7061         if -fstrict-overflow.  Add warnings for when we do make the
7062         transformation.
7064 2013-05-24  Dehao Chen  <dehao@google.com>
7066         * tree-cfg.c (locus_discrim_map): Fix the typo.
7067         (locus_discrim_hasher): Likewise.
7068         (locus_discrim_hasher::hash): Likewise.
7069         (locus_discrim_hasher::equal): Likewise.
7071 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7073         PR tree-optimization/57294
7074         * cgraph.h (ipa_record_stmt_references): Declare.
7075         * cgraphbuild.c (ipa_record_stmt_references): New function.
7076         (build_cgraph_edges): Use ipa_record_stmt_references.
7077         (rebuild_cgraph_edges): Likewise.
7078         (cgraph_rebuild_references): Likewise.
7079         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7080         associated with the old statement and build references from the
7081         newly built statements.
7082         * ipa-ref.c (ipa_remove_stmt_references): New function.
7083         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7085 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7087         * lra-constraints.c (emit_spill_move): Use smaller mode for
7088         mem-mem moves.
7089         (check_and_process_move): Consider mem-reg moves for secondary
7090         too.
7091         (curr_insn_transform): Don't lose insns emitted before for
7092         secondary memory moves.
7093         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7094         reg set up in the current insn.
7096 2013-05-24  Dehao Chen  <dehao@google.com>
7098         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7099         hash function.
7100         (locus_descrim_hasher::equal): Likewise.
7101         (build_gimple_cfg): New discriminator assignment algorithm.
7102         (make_edges): Likewise.
7103         (next_discriminator_for_locus): Likewise.
7104         (same_line_p): Likewise.
7105         (assign_discriminators): Likewise.
7106         (make_cond_expr_edges): Likewise.
7107         (make_gimple_switch_edges): Likewise.
7108         (make_goto_expr_edges): Likewise.
7109         (make_gimple_asm_edges): Likewise.
7111 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
7113         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7114         X format specifier to only display bottom 16 bits.
7115         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7116         immediate to match for operand 2, since it will be masked.
7118 2013-05-24  Richard Biener  <rguenther@suse.de>
7120         PR tree-optimization/57287
7121         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7122         all SSA names that occur in abnormal PHIs.
7124 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7126         PR tree-ssa/57385
7127         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7128         that index is not negative.
7130 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
7132         PR rtl-optimization/55177
7133         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7134         (simplify_byte_swapping_operation): New.
7135         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7136         (simplify_relational_operation_1): Deal with BSWAP.
7138 2013-05-23  Richard Henderson  <rth@redhat.com>
7140         PR target/56742
7141         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7142         (ix86_reorg): Call it.
7144 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
7146         PR target/57379
7147         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7148         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7149         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7151 2013-05-23  Christian Bruel  <christian.bruel@st.com>
7153         PR debug/57351
7154         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7156 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7157             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7159         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7160         * config/aarch64/constraints.md (Usa): Remove.
7161         * doc/md.texi (AArch64 Usa): Remove.
7163 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7164             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7166         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7167         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7168         * config/aarch64/predicates.md (aarch64_const_address): Remove.
7169         (aarch64_mov_operand): Use aarch64_mov_operand_p.
7171 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
7173         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7174         instruction (AdvSIMD).
7175         * config/aarch64/aarch64-builtins.c
7176         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7177         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7179 2013-05-23  Martin Jambor  <mjambor@suse.cz>
7181         PR middle-end/57347
7182         * tree.h (contains_bitfld_component_ref_p): Declare.
7183         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7184         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
7185         caller.
7186         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7187         not access a bit-field.  Assert all final offsets are byte-aligned.
7189 2013-05-23  Richard Biener  <rguenther@suse.de>
7191         PR tree-optimization/57380
7192         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7193         least one invariant or re-used load.
7194         * passes.c (init_optimization_passes): Move pass_phiprop before
7195         pass_forwprop.
7197 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
7199         * config/aarch64/aarch64-simd.md
7200         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7202 2013-05-23  Richard Biener  <rguenther@suse.de>
7204         PR middle-end/57381
7205         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7206         OEP_CONSTANT_ADDRESS_OF retained.
7208 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
7210         PR middle-end/57344
7211         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7212         don't lower unit.  Handle unit not being always BITS_PER_WORD.
7214 2013-05-23  Richard Biener  <rguenther@suse.de>
7216         PR rtl-optimization/57341
7217         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7218         instead of true_dependence.
7220 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
7222         * bb-reorder.c (branch_threshold): Make const.
7223         (exec_threshold): Ditto.
7225 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7226             Pat Haugen <pthaugen@us.ibm.com>
7227             Peter Bergner <bergner@vnet.ibm.com>
7229         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7230         documentation for the power8 crypto builtins.
7232         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7234         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7235         macros for defining power8 builtin functions.
7236         (BU_P8V_AV_2): Likewise.
7237         (BU_P8V_AV_P): Likewise.
7238         (BU_P8V_VSX_1): Likewise.
7239         (BU_P8V_OVERLOAD_1): Likewise.
7240         (BU_P8V_OVERLOAD_2): Likewise.
7241         (BU_CRYPTO_1): Likewise.
7242         (BU_CRYPTO_2): Likewise.
7243         (BU_CRYPTO_3): Likewise.
7244         (BU_CRYPTO_OVERLOAD_1): Likewise.
7245         (BU_CRYPTO_OVERLOAD_2): Likewise.
7246         (XSCVSPDP): Fix typo, point to the correct instruction.
7247         (VCIPHER): Add power8 crypto builtins.
7248         (VCIPHERLAST): Likewise.
7249         (VNCIPHER): Likewise.
7250         (VNCIPHERLAST): Likewise.
7251         (VPMSUMB): Likewise.
7252         (VPMSUMH): Likewise.
7253         (VPMSUMW): Likewise.
7254         (VPERMXOR_V2DI): Likewise.
7255         (VPERMXOR_V4SI: Likewise.
7256         (VPERMXOR_V8HI: Likewise.
7257         (VPERMXOR_V16QI: Likewise.
7258         (VSHASIGMAW): Likewise.
7259         (VSHASIGMAD): Likewise.
7260         (VPMSUM): Likewise.
7261         (VPERMXOR): Likewise.
7262         (VSHASIGMA): Likewise.
7264         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7265         __CRYPTO__ if the crypto instructions are available.
7266         (altivec_overloaded_builtins): Add support for overloaded power8
7267         builtins.
7269         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7270         support for power8 crypto builtins.
7271         (builtin_function_type): Likewise.
7272         (altivec_init_builtins): Add support for builtins that take vector
7273         long long (V2DI) arguments.
7275         * config/rs6000/crypto.md: New file, define power8 crypto
7276         instructions.
7278 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
7279             Pat Haugen <pthaugen@us.ibm.com>
7280             Peter Bergner <bergner@vnet.ibm.com>
7282         * doc/invoke.texi (Option Summary): Add power8 options.
7283         (RS/6000 and PowerPC Options): Likewise.
7285         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7286         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
7287         wm, wn, wr documentation.
7289         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7290         registers if direct move instructions are enabled.
7291         (wn): New constraint for no registers.
7292         (wq): New constraint for quad word even GPR registers.
7293         (wr): New constraint if 64-bit instructions are enabled.
7294         (wv): New constraint if power8 vector instructions are enabled.
7295         (wQ): New constraint for quad word memory locations.
7297         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7298         constraint for 0..15 for crypto instructions.
7299         (gpc_reg_operand): If VSX allow registers in VSX registers as well
7300         as GPR and floating point registers.
7301         (int_reg_operand): New predicate to match only GPR registers.
7302         (base_reg_operand): New predicate to match base registers.
7303         (quad_int_reg_operand): New predicate to match even GPR registers
7304         for quad memory operations.
7305         (vsx_reg_or_cint_operand): New predicate to allow vector logical
7306         operations in both GPR and VSX registers.
7307         (quad_memory_operand): New predicate for quad memory operations.
7308         (reg_or_indexed_operand): New predicate for direct move support.
7310         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7311         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7312         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7313         (POWERPC_MASKS): Add power8 options.
7314         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7315         various options.
7317         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7318         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7320         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7321         (-mpower8-fusion): New power8 options.
7322         (-mpower8-fusion-sign): Likewise.
7323         (-mpower8-vector): Likewise.
7324         (-mcrypto): Likewise.
7325         (-mdirect-move): Likewise.
7326         (-mquad-memory): Likewise.
7328         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7329         power8.
7330         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7331         registers.
7332         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7333         (rs6000_debug_vector_unit): Add p8_vector.
7334         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7335         definitions.  Also print fusion state.
7336         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7337         (rs6000_builtin_mask_calculate): Add power8 builtin support.
7338         (rs6000_option_override_internal): Add support for power8.
7339         (rs6000_common_init_builtins): Add debugging for skipped builtins
7340         if -mdebug=builtin.
7341         (rs6000_adjust_cost): Add power8 support.
7342         (rs6000_issue_rate): Likewise.
7343         (insn_must_be_first_in_group): Likewise.
7344         (insn_must_be_last_in_group): Likewise.
7345         (force_new_group): Likewise.
7346         (rs6000_register_move_cost): Likewise.
7347         (rs6000_opt_masks): Likewise.
7349         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7350         power8 capable assembler, default to power7 options.
7351         (TARGET_DIRECT_MOVE): Likewise.
7352         (TARGET_CRYPTO): Likewise.
7353         (TARGET_P8_VECTOR): Likewise.
7354         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7355         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7356         (VECTOR_MEM_P8_VECTOR_P): Likewise.
7357         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7358         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7359         (TARGET_XSCVDPSPN): Likewise.
7360         (TARGET_XSCVSPDPN): Likewsie.
7361         (TARGET_SYNC_HI_QI): Likewise.
7362         (TARGET_SYNC_TI): Likewise.
7363         (MASK_CRYPTO): Likewise.
7364         (MASK_DIRECT_MOVE): Likewise.
7365         (MASK_P8_FUSION): Likewise.
7366         (MASK_P8_VECTOR): Likewise.
7367         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7368         temporary used by some of the direct move instructions to get two FP
7369         temporary registers does not force creation of a stack frame.
7370         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7371         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7372         that any VSX registers are tieable, even if they are also an
7373         Altivec vector mode.
7374         (r6000_reg_class_enum): Add wm, wr, wv constraints.
7375         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7376         (RS6000_BTM_CRYPTO): Likewise.
7377         (RS6000_BTM_COMMON): Likewise.
7379         * config/rs6000/rs6000.md (cpu attribute): Add power8.
7380         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7381         (enum rs6000_vector): Add power8 vector support.
7383 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7385         PR target/19599
7386         PR target/57340
7387         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7388         (any_sibcall_could_use_r3): this and handle indirect calls.
7389         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7391 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7393         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7395 2013-05-22  Richard Biener  <rguenther@suse.de>
7397         PR middle-end/57349
7398         * profile.c (branch_prob): Do not split blocks that are
7399         abnormally receiving from ECF_RETURNS_TWICE functions.
7401 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7403         * recog.c (offsettable_address_addr_space_p): Fix calculation of
7404         address mode.  Move pointer mode initialization to the same place.
7406 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7408         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7409         while it has any effect.
7411 2013-05-21  Easwaran Raman  <eraman@google.com>
7413         PR tree-optimization/57322
7414         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7415         UID of the statement added to the BB to be 1.
7417 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
7419         PR tree-optimization/57331
7420         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7421         of conversion from pointer type to integral type with integer.
7423 2013-05-21  Martin Jambor  <mjambor@suse.cz>
7425         PR lto/57289
7426         * ipa-prop.c (ipa_read_node_info): Process param_used and
7427         controlled_uses in the same order as when writing.
7429 2013-05-21  Magnus Granberg  <baldrick@free.fr>
7431         PR plugins/56754
7432         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7434 2013-05-21  Richard Biener  <rguenther@suse.de>
7436         PR tree-optimization/57318
7437         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7438         estimate stmts with side-effects as likely eliminated.
7440 2013-05-21  Richard Biener  <rguenther@suse.de>
7442         PR tree-optimization/57330
7443         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7444         preserve the call stmts fntype.
7446 2013-05-21  Richard Biener  <rguenther@suse.de>
7448         PR tree-optimization/57303
7449         * tree-ssa-sink.c (statement_sink_location): Improve killing
7450         stmt detection and properly handle self-assignments.
7452 2013-05-21  Christian Bruel  <christian.bruel@st.com>
7454         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7455         spanning registers. LEAF_REG_REMAP is supported only for contiguous
7456         registers. Set register size out of the PARALLEL loop.
7458 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
7460         PR target/56547
7461         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7462         (*fmasf4, *fmasf4_media): New insns.
7464 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7466         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7467         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7468         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7469         (mips_idiv_insns): Update the comments to say that the returned
7470         instruction counts are in units of BASE_INSN_LENGTH.
7471         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7472         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
7473         using 2 rather than 4 as the length of indirect MIPS16 and
7474         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
7475         length of a NOP.  Don't divide MIPS16 lengths by 2.
7476         (mips16_split_long_branches): Assume a branch is long if the
7477         length is greater than 4 rather than 8.
7478         * config/mips/mips.md (length): Give MIPS16 lengths directly,
7479         rather than multiplying them by 2.  Multiply instruction counts
7480         by BASE_INSN_LENGTH rather than 4.
7481         (*jump_mips16, tls_get_tp_mips16_<mode>)
7482         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7484 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7486         * config/mips/mips.md (extended_mips16): Remove branch case.
7487         (length): Remove duplicated extended_mips16 test.
7489 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
7491         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7493 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
7495         * recog.h (Recog_data): Rename to...
7496         (recog_data_d): ...this.
7497         (recog_data): Update accordingly.
7498         * recog.c (recog_data): Likewise.
7499         * reload.c (save_recog_data): Likewise.
7500         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7501         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7503 2013-05-17  Julian Brown  <julian@codesourcery.com>
7505         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7506         found in a REG_EQUAL note, invalidate it.
7508 2013-05-17  Easwaran Raman  <eraman@google.com>
7510         * tree-ssa-reassoc.c (find_insert_point): New function.
7511         (insert_stmt_after): Likewise.
7512         (get_def_stmt): Likewise.
7513         (ensure_ops_are_available): Likewise.
7514         (not_dominated_by): Likewise.
7515         (rewrite_expr_tree): Do not move statements beyond what is
7516         necessary. Remove call to swap_ops_for_binary_stmt...
7517         (reassociate_bb): ... and move it here.
7518         (build_and_add_sum): Assign UIDs for new statements.
7519         (linearize_expr): Likewise.
7520         (do_reassoc): Renumber gimple statement UIDs.
7522 2013-05-17  Jan Hubicka  <jh@suse.cz>
7524         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7525         weakrefs.
7526         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7527         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7528         weakrefs.
7529         (output_weakrefs): Update.
7531 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
7532             Martin Jambor  <mjambor@suse.cz>
7534         PR middle-end/57276
7535         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7536         value that corresponds to the given aggval is found in values vector.
7538 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
7540         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7541         sse, sse2, sse3, ssse3 and sse4a flags to options.
7543 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
7545         * gengtype-state.c: (s_expr_writer): New class, to handle
7546         prettifying of output layout of s-expressions.
7547         (state_writer): New class, to write out gtype.state.
7548         (state_written_type_count): Move this variable into member data of
7549         state_writer.
7550         (s_expr_writer::s_expr_writer): New code: constructor for new class
7551         (state_writer::state_writer(): ditto
7552         (s_expr_writer::write_new_line): New function
7553         (s_expr_writer::write_any_indent): ditto
7554         (s_expr_writer::begin_s_expr): ditto
7555         (s_expr_writer::end_s_expr): ditto
7556         (write_state_fileloc): convert to method of state_writer...
7557         (state_writer:: write_state_fileloc): ...and use methods of
7558         s_expr_writer to write indentation into the gtype.state output file
7559         to visually represent the hierarchical structure of the list
7560         structures
7561         (write_state_fields): ditto, renaming to...
7562         (state_writer::write_state_fields)
7563         (write_state_a_string): ditto, renaming to...
7564         (state_writer::write_state_a_string)
7565         (write_state_string_option): ditto, renaming to...
7566         (state_writer::write_state_string_option)
7567         (write_state_type_option): ditto, renaming to...
7568         (state_writer::write_state_type_option)
7569         (write_state_nested_option): ditto, renaming to...
7570         (state_writer::write_state_nested_option)
7571         (write_state_option): ditto, renaming to...
7572         (state_writer::write_state_option)
7573         (write_state_options): ditto, renaming to...
7574         (state_writer::write_state_options)
7575         (write_state_lang_bitmap): ditto, renaming to...
7576         (state_writer::write_state_lang_bitmap)
7577         (write_state_version): ditto, renaming to...
7578         (state_writer::write_state_version)
7579         (write_state_scalar_type): ditto, renaming to...
7580         (state_writer::write_state_scalar_type)
7581         (write_state_string_type): ditto, renaming to...
7582         (state_writer::write_state_string_type)
7583         (write_state_undefined_type): ditto, renaming to...
7584         (state_writer::write_state_undefined_type)
7585         (write_state_struct_union_type): ditto, renaming to...
7586         (state_writer::write_state_struct_union_type)
7587         (write_state_struct_type): ditto, renaming to...
7588         (state_writer::write_state_struct_type)
7589         (write_state_user_struct_type): ditto, renaming to...
7590         (state_writer::write_state_user_struct_type)
7591         (write_state_lang_struct_type): ditto, renaming to...
7592         (state_writer::write_state_lang_struct_type)
7593         (write_state_param_struct_type): ditto, renaming to...
7594         (state_writer::write_state_param_struct_type)
7595         (write_state_pointer_type): ditto, renaming to...
7596         (state_writer::write_state_pointer_type)
7597         (write_state_array_type): ditto, renaming to...
7598         (state_writer::write_state_array_type)
7599         (write_state_gc_used): ditto, renaming to...
7600         (state_writer::write_state_gc_used)
7601         (write_state_common_type_content): ditto, renaming to...
7602         (state_writer::write_state_common_type_content)
7603         (write_state_type): ditto, renaming to...
7604         (state_writer::write_state_type)
7605         (write_state_pair_list): ditto, renaming to...
7606         (state_writer::write_state_pair_list)
7607         (write_state_pair): ditto, renaming to...
7608         (state_writer::write_state_pair)
7609         (write_state_typedefs): ditto, renaming to...
7610         (state_writer::write_state_typedefs)
7611         (write_state_structures): ditto, renaming to...
7612         (state_writer::write_state_structures)
7613         (write_state_param_structs): ditto, renaming to...
7614         (state_writer::write_state_param_structs)
7615         (write_state_variables): ditto, renaming to...
7616         (state_writer::write_state_variables)
7617         (write_state_srcdir): ditto, renaming to...
7618         (state_writer::write_state_srcdir)
7619         (write_state_files_list): ditto, renaming to...
7620         (state_writer::write_state_files_list)
7621         (write_state_languages): ditto, renaming to...
7622         (state_writer::write_state_languages)
7623         (write_state): create a state_writer instance and use it when
7624         writing out the state file
7626 2013-05-17  Mike Stump  <mikestump@comcast.net>
7628         PR rtl-optimization/57304
7629         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7630         accessing DF_REF_REAL_LOC.
7632 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7634         PR rtl-optimization/57281
7635         PR rtl-optimization/57300
7636         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7637         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7638         what the other splitter did if the registers are dead.
7640 2013-05-17  Richard Biener  <rguenther@suse.de>
7642         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7643         MEM_REF offsets.
7645 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7647         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7648         linking.
7650 2013-05-17  Marek Polacek  <polacek@redhat.com>
7652         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7653         length when doing non-zero store of storing '\0' to '\0'.
7655 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
7657         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7658         vect_external_def oprnd1 with loop_vinfo, try to emit
7659         optional cast, negation and and stmts on the loop preheader
7660         edge instead of into the pattern def seq.
7662         PR tree-optimization/57051
7663         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7664         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7666 2013-05-16  Nick Clifton  <nickc@redhat.com>
7668         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7669         (rl78_is_naked_func): New function.
7670         (rl78_expand_prologue): Skip prologue generation for naked functions.
7671         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
7672         * doc/extend.texi (naked): Add RL78 to the list of processors
7673         that supports this attribute.
7675 2013-05-16  Jeff Law  <law@redhat.com>
7677         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7679 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7681         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7682         cache parameters using detect_caches_amd also for CYRIX,
7683         NSC and TM2 signatures.
7685 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
7686             Dzianis Kahanovich  <mahatma@eu.by>
7688         PR target/45359
7689         PR target/46396
7690         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7691         VIA/Centaur processors and determine their cache parameters
7692         using detect_caches_amd.
7694 2013-05-16  Teresa Johnson  <tejohnson@google.com>
7696         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7697         (rtl_verify_edges): New function.
7698         (rtl_verify_bb_insns): Ditto.
7699         (rtl_verify_bb_pointers): Ditto.
7700         (rtl_verify_bb_insn_chain): Ditto.
7701         (rtl_verify_fallthru): Ditto.
7702         (rtl_verify_bb_layout): Ditto.
7703         (rtl_verify_flow_info_1): Outline checks into new functions.
7704         (rtl_verify_flow_info): Ditto.
7706 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
7708         * cfghooks.c (copy_bbs): Add update_dominance argument.
7709         * cfghooks.h (copy_bbs): Update prototype.
7710         * tree-cfg.c (gimple_duplicate_sese_region):
7711         Add update_dominance argument.
7712         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7713         * tree-ssa-loop-ch.c (copy_loop_headers): Update
7714         gimple_duplicate_sese_region call.
7715         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7716         Update copy_bbs call.
7717         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7718         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7720 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7722         * tree-vectorizer.h (NUM_PATTERNS): Increment.
7723         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7724         vect_recog_rotate_pattern.
7725         (vect_recog_rotate_pattern): New function.
7727 2013-05-16  Jason Merrill  <jason@redhat.com>
7729         * Makefile.in (LLINKER): New variable.
7730         (mostlyclean): Remove link mutex.
7731         * configure.ac: Handle --enable-link-mutex.
7732         * lock-and-run.sh: New script.
7734 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7736         PR target/19599
7737         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7738         for NULL decl.
7740 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7742         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7744 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
7746         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7747         * config/arm/arm.c (next_consecutive_mem): New function.
7748         (gen_movmem_ldrd_strd): Likewise.
7749         * config/arm/arm.md (movmemqi): Update condition and code.
7750         (unaligned_loaddi, unaligned_storedi): New patterns.
7752 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7754         * config.gcc: Obsolete *-*-solaris2.9*.
7755         * doc/install.texi (Specific, *-*-solaris2*): Document it.
7757 2013-05-16  Richard Biener  <rguenther@suse.de>
7759         * passes.c (init_optimization_passes): Move pass_parallelize_loops
7760         earlier, after GRAPHITE transforms and IV canonicalization.
7762 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
7764         * omp-low.c (extract_omp_for_data): For collapsed loops,
7765         if at least one of the loops is known at compile time to
7766         iterate zero times, set count to 0.
7767         (expand_omp_regimplify_p): New function.
7768         (expand_omp_for_generic): For collapsed loops, if at least
7769         one of the loops isn't known to iterate at least once,
7770         add runtime check with setting count to 0.
7771         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7772         For unsigned types if it isn't known at compile time that
7773         the loop will iterate at least once, add runtime check to bypass
7774         the whole loop if initial condition isn't true.
7776 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
7778         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7780 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
7782         PR middle-end/57286
7783         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7784         transformations to avoid an infinite loop.
7786 2013-05-16  Marek Polacek  <polacek@redhat.com>
7788         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7790 2013-05-15  Leif Ekblad  <leif@rdos.net>
7792         * config/i386/i386.c (ix86_decompose_address): Use
7793         DEFAULT_TLS_SEG_REG to access TLS segment register.
7794         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7795         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7796         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7798 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
7800         PR target/57260
7801         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7802         sibling calls to functions that would normally be lazily bound,
7803         unless $gp is call-clobbered.
7805 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
7807         * config/i386/i386.c (ix86_option_override_internal): Update
7808         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
7809         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7810         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
7811         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7812         of TARGET_3DNOW.
7813         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7815 2013-05-15  Andreas Schwab  <schwab@suse.de>
7817         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7818         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
7819         third operand.
7821 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7823         * loop-unroll.c (report_unroll_peel): Check decision before
7824         emitting unroll/peel message.
7826 2013-05-15  Teresa Johnson  <tejohnson@google.com>
7828         * function.h (has_bb_partition): New rtl_data flag.
7829         (bb_reorder_complete): Ditto.
7830         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7831         instead of flag_reorder_blocks_and_partition.
7832         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7833         with some enhancements.
7834         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7835         * bb-reorder.c (connect_traces): Check for has_bb_partition
7836         instead of flag_reorder_blocks_and_partition.
7837         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7838         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7839         verify_hot_cold_block_grouping.
7840         (partition_hot_cold_basic_blocks): Set has_bb_partition.
7842 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7844         PR target/19599
7845         * config/arm/predicates.md (call_insn_operand): New predicate.
7846         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
7847         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7848         if insn is not a tail call.
7849         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
7850         registers.
7851         * config/arm/arm.h (enum reg_class): New caller save register class.
7852         (REG_CLASS_NAMES): Likewise.
7853         (REG_CLASS_CONTENTS): Likewise.
7854         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
7855         without decls.
7857 2013-05-15  Richard Biener  <rguenther@suse.de>
7859         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
7860         of MSG_OPTIMIZED_LOCATIONS.
7861         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
7862         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
7863         message.
7864         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
7865         of MSG_OPTIMIZED_LOCATIONS.
7866         (execute_vect_slp): Likewise.
7867         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
7868         (vect_create_cond_for_alias_checks): Likewise.
7869         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
7870         (vect_recog_widen_mult_pattern): Likewise.
7871         (vect_recog_widen_sum_pattern): Likewise.
7872         (vect_recog_over_widening_pattern): Likewise.
7873         (vect_recog_widen_shift_pattern): Likewise.
7874         (vect_recog_vector_vector_shift_pattern): Likewise.
7875         (vect_recog_divmod_pattern): Likewise.
7876         (vect_recog_mixed_size_cond_pattern): Likewise.
7877         (vect_recog_bool_pattern): Likewise.
7878         (vect_pattern_recog_1): Likewise.
7880 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7882         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
7883         non-functions to builtin_unreachable.
7884         * ipa-inline-transform.c (inline_call): Do not assert estimates were
7885         correct when new direct edges were discovered.
7887 2013-05-15  Martin Jambor  <mjambor@suse.cz>
7889         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
7890         header, print symbol order instead of node uid, print more information
7891         about indirect edge targets.
7892         (ipa_make_edge_direct_to_target): Print symbol order instead of node
7893         uids.
7894         (ipa_make_edge_direct_to_target): Likewise.
7895         (remove_described_reference): Likewise.
7896         (propagate_controlled_uses): Likewise.
7897         (ipa_print_node_params): Also print symbol order.
7898         (ipcp_transform_function): Print symbol order instead of node uids.
7899         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
7900         (cgraph_get_create_real_symbol_node): Likewise.
7901         * ipa-cp.c (print_lattice): Likewise.
7902         (print_all_lattices): Likewise.
7903         (determine_versionability): Likewise.
7904         (initialize_node_lattices): Likewise.
7905         (estimate_local_effects): Likewise.
7906         (update_profiling_info): Likewise.
7907         (create_specialized_node): Likewise.
7908         (perhaps_add_new_callers): Likewise.
7909         (decide_about_value): Likewise.
7910         (decide_whether_version_node): Likewise.
7911         (identify_dead_nodes): Likewise.
7912         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
7913         (dump_inline_summary): Likewise.
7914         (estimate_node_size_and_time): Likewise.
7915         (inline_analyze_function): Likewise.
7916         * ipa-inline.c (report_inline_failed_reason): Likewise.
7917         (want_early_inline_function_p): Likewise.
7918         (edge_badness): Likewise.
7919         (update_edge_key): Likewise.
7920         (inline_small_functions): Likewise.  Add dumping of order to two other
7921         dumps.
7922         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
7923         instead of node uids.
7924         (propagate_pure_const): Likewise.
7925         (propagate_pure_const): Likewise.
7926         * ipa-utils.c (dump_cgraph_node_set): Likewise.
7927         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
7928         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
7929         of node uids.
7930         * tree-pretty-print.c (dump_function_header): Likewise.
7931         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
7932         Print symbol order instead of node uids.
7934 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7936         * config/s390/s390.c (s390_register_move_cost): Don't impose the
7937         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
7939 2013-05-15  Richard Biener  <rguenther@suse.de>
7941         PR tree-optimization/57275
7942         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
7943         return value for fail to do runtime alias checks for gather loads.
7945 2013-05-15  Jan Hubicka  <jh@suse.cz>
7947         PR lto/57038
7948         PR lto/47375
7949         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
7950         weakrefs are not external.
7951         (lto_symtab_merge_decls): Fix thinko when dealing with
7952         non-lto_symtab decls.
7953         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
7954         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
7955         * varpool.c (dump_varpool_node): Dump more flags.
7957 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7959         * config/i386/i386.c (processor_alias_table): Add instruction
7960         FSGSBASE for AMD bdver3 architecture.
7962 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
7964         * tree.c (warn_deprecated_use): Print file:line using locus color.
7965         * diagnostic.c (diagnostic_report_current_module): Print file:line
7966         and file:line:column using locus color.
7968 2013-05-14  Mike Stump  <mikestump@comcast.net>
7970         * gdbinit.in: Add __null.
7972 2013-05-14  Mike Stump  <mikestump@comcast.net>
7974         * recog.h: Rename struct recog_data to Recog_data.
7975         * recog.c: Likewise.
7976         * reload.c (can_reload_into): Likewise.
7977         * config/picochip/picochip.c: Likewise.
7979 2013-05-14  Mike Stump  <mikestump@comcast.net>
7981         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
7983 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
7985         * resource.h (struct resources): Remove unch_memory member.
7986         (CLEAR_RESOURCE): Don't clear unch_memory.
7987         * resource.c (mark_referenced_resources): Don't set it.
7988         (mark_set_resources): Likewise.
7989         (mark_target_live_regs): Don't clear it.
7990         (init_resource_info): Likewise.
7991         * reorg.c (resource_conflicts_p): Don't compare it.
7992         (redundant_insn): Don't set it.
7994         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
7995         Remove prototypes.
7996         * emit-rtl.c (next_label): Remove unused function.
7997         (skip_consecutive_labels, link_cc0_insns): Move to ...
7998         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
7999         only place where these functions are used, and make them static.
8001 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8003         * fold-const.c (fold_negate_expr): Handle vectors.
8004         (fold_truth_not_expr): Make it static.
8005         (fold_invert_truthvalue): New static function.
8006         (invert_truthvalue_loc): Handle vectors. Do not call
8007         fold_truth_not_expr directly.
8008         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8009         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8010         (fold_comparison): Handle vector constants.
8011         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8012         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8013         * tree.h (fold_truth_not_expr): Remove declaration.
8015 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8017         * config/aarch64/aarch64-simd.md
8018         (aarch64_vcond_internal<mode>): Rename to...
8019         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8020         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8021         float modes. Clarify all iterator modes.
8022         (vcond<mode><mode>): Use new name for vcond expanders.
8023         (vcond<v_cmp_result><mode>): Likewise.
8024         (vcondu<mode><mode>: Likewise.
8025         * config/aarch64/iterators.md (VDQF_COND): New.
8027 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8029         PR bootstrap/57266
8030         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8031         variable for the shift amount. Check that we shift by non-negative
8032         amounts.
8034 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8036         PR target/42017
8037         * config/arm/arm.h (EPILOGUE_USES): Only return true
8038         for LR_REGNUM after epilogue_completed.
8040 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8042         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8043         is error_mark_node.
8045 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8047         PR target/57261
8048         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8049         and Solaris 11+/x86 with gld.
8050         * configure: Regenerate.
8052 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8054         * expmed.c (expand_shift_1): Canonicalize rotates by
8055         constant bitsize / 2 to bitsize - 1.
8056         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8057         case ROTATERT>: Likewise.
8059         Revert:
8060         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8062         * config/i386/i386.md (rotateinv): New code attr.
8063         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8064         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8065         roll $31, %eax, etc.
8067 2013-05-14  Richard Biener  <rguenther@suse.de>
8069         PR middle-end/57235
8070         * tree-eh.c (sink_clobbers): Give up for successors with
8071         multiple predecessors and no virtual uses.
8073 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8075         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8076         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8078 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8080         PR middle-end/57251
8081         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8082         the case when both op0 and op1 have VOIDmode.
8084 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8086         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8087         in multiply-accumulate mode.
8089 2013-05-13  Guozhi Wei  <carrot@google.com>
8091         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8093 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8095         PR target/56975
8096         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8097         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8098         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8099         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8100         (ix86_expand_prologue): Likewise.
8101         (ix86_expand_split_stack_prologue): Likewise.
8102         (legitimate_pic_address_disp_p): Likewise.
8103         (legitimize_pic_address): Likewise.
8104         (legitimize_tls_address): Likewise.
8105         (legitimize_pe_coff_symbol): Likewise.
8106         (output_pic_addr_const): Likewise.
8107         (construct_plt_address): Likewise.
8108         (ix86_expand_call): Likewise.
8109         (x86_output_mi_thunk): Likewise.
8110         (x86_function_profiler): Likewise.
8112 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
8114         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8115         similar switch cases.
8116         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8117         (aarch64_simd_mov_to_<mode>low): Delete.
8118         (aarch64_simd_mov_to_<mode>high): Delete.
8119         (move_lo_quad_<mode>): Add w<-r alternative.
8120         (aarch64_simd_move_hi_quad_<mode>): Likewise.
8121         (aarch64_simd_mov_from_*): Update type attribute.
8122         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8123         statement.
8125 2013-05-13  Jan Hubicka  <jh@suse.cz>
8127         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8128         * config/i386/i386.c (ix86_compute_frame_layout,
8129         ix86_expand_epilogue, emit_i387_cw_initialization,
8130         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8131         ix86_local_alignment): Fix use of size/speed predicates.
8133 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8135         PR tree-optimization/45216
8136         PR tree-optimization/57157
8137         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8138         the (-Y) & (B - 1) variant if OP is |.
8139         * expmed.c (expand_shift_1): For rotations by const0_rtx just
8140         return shifted.  Use (-op1) & (prec - 1) as other_amount
8141         instead of prec - op1.
8143 2013-05-13  Martin Jambor  <mjambor@suse.cz>
8145         PR middle-end/42371
8146         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8147         (ipa_constant_data): New type.
8148         (ipa_jump_func): Use ipa_constant_data to hold information about
8149         constant jump functions.
8150         (ipa_get_jf_constant): Adjust to jump function type changes.
8151         (ipa_get_jf_constant_rdesc): New function.
8152         (ipa_param_descriptor): New field controlled_uses.
8153         (ipa_get_controlled_uses): New function.
8154         (ipa_set_controlled_uses): Likewise.
8155         * ipa-ref.h (ipa_find_reference): Declare.
8156         * ipa-prop.c (ipa_cst_ref_desc): New type.
8157         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8158         changes.
8159         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
8160         New parameter cs.  Adjust all callers.
8161         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8162         (remove_described_reference): New function.
8163         (jfunc_rdesc_usable): Likewise.
8164         (try_make_edge_direct_simple_call): Decrement controlled use count,
8165         attempt to remove reference if it hits zero.
8166         (combine_controlled_uses_counters): New function.
8167         (propagate_controlled_uses): Likewise.
8168         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8169         (ipa_edge_duplication_hook): Duplicate reference descriptions.
8170         (ipa_print_node_params): Print described use counter.
8171         (ipa_write_jump_function): Adjust to jump function type changes.
8172         (ipa_read_jump_function): New parameter CS, pass it to
8173         ipa_set_jf_constant.  Adjust caller.
8174         (ipa_write_node_info): Stream controlled use count
8175         (ipa_read_node_info): Likewise.
8176         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8177         asserting.
8178         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8179         count.  Remove cloning-added reference if it reaches zero.
8180         * ipa-ref.c (ipa_find_reference): New function.
8182 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8184         * config/i386/i386.c (processor_target_table): Modified default
8185         alignment values for AMD BD and BT architectures.
8187 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
8189         * tree-vect-generic.c (uniform_vector_p): Move ...
8190         * tree.c (uniform_vector_p): ... here.
8191         * tree.h (uniform_vector_p): Declare it.
8192         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8193         into a scalar.
8195 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
8197         PR tree-optimization/57230
8198         * tree-ssa-strlen.c (handle_char_store): Record length for
8199         array store from STRING_CST.
8201         PR tree-optimization/57230
8202         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8203         check.
8205 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
8207         * config/epiphany/epiphany.c (epiphany_init): Check size of
8208         NUM_MODES_FOR_MODE_SWITCHING.
8209         (epiphany_expand_prologue):
8210         Remove CONFIG_REGNUM initial value handling code.
8211         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8212         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8213         (emit_set_fp_mode, epiphany_mode_after): Likewise.
8214         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8215         Don't return 1 for FP_MODE_NONE.
8216         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8217         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8218         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8219         * config/epiphany/epiphany.md (save_config): New pattern.
8221 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
8223         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8225 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
8227         * config/i386/i386.md (memory): Handle sseishft1.
8228         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8229         (*vec_extractv2di_1): Ditto.
8231 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
8233         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8234         saved registers.
8236 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
8238         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8239         Add mthumb/march=armv7-a multilib.
8240         Add mthumb/march=armv7-r multilib.
8241         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8243 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
8245         * config/v850/t-rtems: Add more multilibs.
8247 2013-05-10  Richard Biener  <rguenther@suse.de>
8249         PR tree-optimization/57214
8250         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8251         not propagate from SSA names that occur in abnormal PHI nodes.
8253 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
8255         * stor-layout.c (element_precision): New function.
8256         * machmode.h (element_precision): Declare it.
8257         * tree.c (build_minus_one_cst): New function.
8258         (element_precision): Likewise.
8259         * tree.h (build_minus_one_cst): Declare new function.
8260         (element_precision): Likewise.
8261         * fold-const.c (operand_equal_p): Use element_precision.
8262         (fold_binary_loc): Handle vector types.
8263         * convert.c (convert_to_integer): Use element_precision.
8264         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8265         separately.
8267 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
8269         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8270         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8271         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8272         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8273         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8274         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8275         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8276         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8277         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8278         (Uuw8): New constraints.
8279         (Usb4): Move into alphabetical order.
8280         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8281         (sd8_operand, ub8_operand, uw8_operand): New predicates.
8282         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8283         previously unnamed patterns.
8284         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8285         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8286         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8287         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8288         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
8289         of specific lengths.
8291 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8293         * config/i386/i386.md (rotateinv): New code attr.
8294         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8295         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8296         roll $31, %eax, etc.
8298         PR tree-optimization/45216
8299         PR tree-optimization/57157
8300         * tree-ssa-forwprop.c (simplify_rotate): New function.
8301         (ssa_forward_propagate_and_combine): Call it.
8303 2013-05-10  Richard Biener  <rguenther@suse.de>
8305         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8306         disable peeling when we version for aliasing.
8307         (vector_alignment_reachable_p): Honor explicit user alignment.
8308         (vect_supportable_dr_alignment): Likewise.
8309         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8310         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8311         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8312         then peeling to arrange for the cost-model check to come first.
8314 2013-05-10  Alan Modra  <amodra@gmail.com>
8316         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8317         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8318         * configure: Regenerate.
8320 2013-05-10  Alan Modra  <amodra@gmail.com>
8322         PR target/55033
8323         * varasm.c (default_elf_select_section): Move !DECL_P check..
8324         (get_named_section): ..to here before calling get_section_name.
8325         Adjust assertion.
8326         (default_section_type_flags): Add DECL_P check.
8327         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8328         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8330 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
8332         * config/epiphany/epiphany.c (epiphany_expand_prologue):
8333         When using gen_stack_adjust_str with a register offset, add a
8334         REG_FRAME_RELATED_EXPR note.
8336 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
8338         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8339         (*vec_extractv4si_zext_mem): Ditto.
8340         (*vec_extractv2di): Add 0->x and x->x alternatives.
8341         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8342         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8344 2013-05-09  Jason Merrill  <jason@redhat.com>
8346         N3639 C++1y VLA support
8347         * gimplify.c (gimplify_vla_decl): Don't touch an existing
8348         DECL_VALUE_EXPR.
8350         * tree.c (build_constructor_va): New.
8351         * tree.h: Declare it.
8353 2013-05-09  Martin Jambor  <mjambor@suse.cz>
8355         PR lto/57084
8356         * gimple-fold.c (canonicalize_constructor_val): Call
8357         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8359 2013-05-09  Jan Hubicka  <jh@suse.cz>
8360             Richard Biener  <rguenther@suse.de>
8362         PR lto/54095
8363         * symtab.c (symtab_make_decl_local): Do not add private names.
8365 2013-05-09  Jan Hubicka  <jh@suse.cz>
8367         PR lto/54095
8368         * symtab.c (insert_to_assembler_name_hash): Handle clones.
8369         (unlink_from_assembler_name_hash): Likewise.
8370         (symtab_prevail_in_asm_name_hash, symtab_register_node,
8371         symtab_unregister_node, symtab_initialize_asm_name_hash,
8372         change_decl_assembler_name): Update.
8374 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
8376         * config/aarch64/aarch64.md: New movtf split.
8377         (*movtf_aarch64): Update.
8378         (aarch64_movdi_tilow): Handle TF modes and rename to
8379         aarch64_movdi_<mode>low.
8380         (aarch64_movdi_tihigh): Handle TF modes and rename to
8381         aarch64_movdi_<mode>high
8382         (aarch64_movtihigh_di): Handle TF modes and rename to
8383         aarch64_mov<mode>high_di
8384         (aarch64_movtilow_di): Handle TF modes and rename to
8385         aarch64_mov<mode>low_di
8386         (aarch64_movtilow_tilow): Remove spurious whitespace.
8387         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8388         splits.
8389         (aarch64_print_operand): Update.
8391 2013-05-09  Alan Modra  <amodra@gmail.com>
8393         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8394         powerpc64le.
8395         * configure: Regenerate.
8397 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8399         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8400         splitter preparation statements.
8401         * config/i386/sse.md (*vec_extract* splitters): Ditto.
8402         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8403         adjust_address_nv.
8405 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8407         * gimple-ssa-strength-reduction.c (count_candidates): Change
8408         return value to int.
8409         (analyze_candidates_and_replace): Change type of length to int.
8411 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
8413         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8414         (*vec_extract<mode>): Use VI12_128 mode iterator.
8415         (*vec_extract<mode>_mem): Ditto.
8416         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8417         attribute.
8419 2013-05-08  Diego Novillo  <dnovillo@google.com>
8421         PR bootstrap/54659
8423         Revert:
8424         2012-08-17  Diego Novillo  <dnovillo@google.com>
8426         PR bootstrap/54281
8427         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8428         * config.in: Regenerate.
8429         * configure: Regenerate.
8430         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8432 2013-05-08  Jan Hubicka  <jh@suse.cz>
8434         PR lto/54095
8435         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8436         * cgraph.h (symtab_node_base): Add unique_name.
8437         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8438         input_overwrite_node, input_varpool_node): Stream unique_name.
8439         * cgraphclones.c (cgraph_create_virtual_clone,
8440         cgraph_function_versioning): Set unique_name.
8441         * ipa.c (function_and_variable_visibility): Set unique_name.
8443 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8445         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8446         (alloc_cand_and_find_basis): Restrict conditional candidate
8447         processing to CAND_MULTs.
8449 2013-05-08  Jan Hubicka  <jh@suse.cz>
8451         PR lto/54095
8452         lto-symtab.c (lto_symtab_symbol_p): New function.
8453         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8454         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8455         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8456         Skip static symbols.
8458 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
8460         PR tree-optimization/57200
8461         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8462         Only call inform if the preceding warning_at returns true.
8464 2013-05-07  Han Shen  <shenhan@google.com>
8466         * cfgexpand.c (record_or_union_type_has_array_p): New function.
8467         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8468         * common.opt (fstack-protector-strong): New option.
8469         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8470         * doc/invoke.texi (Optimization Options): Document
8471         "-fstack-protector-strong".
8472         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8474 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8476         * config/mips/mips.c (mips_machine_reorg2): Return 0.
8478 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
8480         * ira.c (update_equiv_regs): Add insn having equiv memory even if
8481         it is not lhs of the insn.
8482         (setup_reg_equiv): Remove insn having equiv memory which it is not
8483         lhs of the insn.
8484         * lra-constraints.c (process_address): Try to improve generation
8485         code for address base + disp.
8486         (lra_constraints): Make correct the code for checking insn setting
8487         up backward equivalence.  Remove insn only if it is in the init
8488         insn list.
8489         * lra-eliminations.c (update_reg_eliminate): Change return value.
8490         (lra_eliminate): Use the result.
8492 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
8494         * config/i386/sse.md (ssescalarnummask): New mode attribute.
8495         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8496         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8497         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8498         register target operands.
8499         (*vec_extractv8hi_sse2): New pattern.
8500         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8501         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8502         (*vec_extract<mode>_mem): New insn and split pattern.
8504 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
8506         * config/arm/arm.c (arm_asan_shadow_offset): New function.
8507         (TARGET_ASAN_SHADOW_OFFSET): Define.
8508         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8509         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8511 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8513         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8514         (incr_vec_index): Return -1 if increment not found.
8515         (create_add_on_incoming_edge): Assert if increment not found.
8516         (record_increment): Limit number of increments recorded.
8517         (all_phi_incrs_profitable): Return false if an increment not found.
8518         (replace_profitable_candidates): Don't process increments that were
8519         not recorded.
8520         (analyze_candidates_and_replace): Limit size of incr_vec.
8522 2013-05-07  Richard Biener  <rguenther@suse.de>
8524         * calls.c (special_function_p): setjmp-like functions are leaf.
8525         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8526         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8528 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
8530         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8531         (aarch64_simd_mov<mode>): New expander.
8532         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8533         (aarch64_simd_mov_to_<mode>high): Likewise.
8534         (aarch64_simd_mov_from_<mode>low): Likewise.
8535         (aarch64_simd_mov_from_<mode>high): Likewise.
8536         (aarch64_dup_lane<mode>): Update.
8537         (aarch64_dup_lanedi): New instruction pattern.
8538         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8539         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8541 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8543         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8544         (replace_mult_candidate): Remove unnecessary argument; remove
8545         unnecessary parameter from call to introduce_cast_before_cand.
8546         (replace_unconditional_candidate): Remove unnecessary parameter
8547         from call to replace_mult_candidate.
8548         (replace_conditional_candidate): Likewise.
8549         (insert_initializers): Use make_temp_ssa_name.
8550         (introduce_cast_before_cand): Remove unnecessary argument; use
8551         make_temp_ssa_name.
8552         (replace_one_candidate): Remove unnecessary argument; remove
8553         unnecessary parameter from calls to introduce_cast_before_cand.
8554         (replace_profitable_candidates): Remove unnecessary parameters
8555         from calls to replace_one_candidate.
8557 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8559         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8560         phi def as possibly hiding a basis for a CAND_ADD whose operands
8561         have been commuted in the analysis.
8562         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8564 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
8566         * config/aarch64/aarch64.md
8567         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8568         shift value between 0-4.
8570 2013-05-07  Richard Biener  <rguenther@suse.de>
8572         * double-int.h (rshift): New overload.
8573         * double-int.c (rshift): New function.
8574         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8575         (create_reference_ops_from_ref): Remove.
8576         (vn_reference_insert): Use shared ops for constructing the
8577         reference and copy it.
8579 2013-05-07  Richard Biener  <rguenther@suse.de>
8581         PR middle-end/57190
8582         * tree-eh.c (sink_clobbers): Properly propagate
8583         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8585 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
8587         PR tree-optimization/57149
8588         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8589         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8590         collect_phi_def_edges, execute_late_warn_uninitialized): Use
8591         uninit_undefined_value_p instead of ssa_undefined_value_p.
8593         PR debug/57184
8594         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8595         for modifier == EXPAND_INITIALIZER.
8597 2013-05-07  Anton Blanchard  <anton@samba.org>
8599         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8600         for powerpc64 little endian.
8601         * configure: Regenerate.
8603 2013-05-06  Graham Stott  <grahams@btinternet.com>
8605         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8606         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8607         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8608         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8610 2013-05-06  Graham Stott  <grahams@btinternet.com>
8612         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8613         codes which allow non-lvalues.
8615 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
8617         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8618         components are all 1s.
8619         (integer_minus_onep): New function.
8620         * tree.h (integer_minus_onep): Declare it.
8621         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8622         integer_minus_onep instead of integer_all_onesp.
8624 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8626         PR target/52933
8627         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8628         variations of these patterns.
8630 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8632         * config/i386/i386.md (isa): Add x64_sse4 member.
8633         (enabled): Handle x64_sse4.
8634         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8635         instruction for 64bit SSE4_1 targets.  Update insn attributes.
8636         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8637         instruction for SSE4_1 targets.  Update insn attributes.
8638         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8639         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8640         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8641         const_1 selector.
8642         (*vec_extractv4si): Rename from *sse4_1_pextrd.
8643         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8644         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8646 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8648         PR target/57108
8649         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8651 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
8653         * final.c (do_assembler_dialects): Don't handle curly braces and
8654         vertical bar escaped by % as dialect delimiters.
8655         (output_asm_insn): Print curly braces and vertical bar if escaped
8656         by % and ASSEMBLER_DIALECT defined.
8657         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8658         * doc/tm.texi: Regenerated.
8660 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
8662         * config/mips/mips.c: Include tree-pass.h.
8663         (mips_reorg): Split in pre- and post-dbr_schedule parts.
8664         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8665         (pass_mips_machine_reorg2): New machine specific pass.
8666         (insert_pass_mips_machine_reorg2): New pass plugin definition.
8667         (mips_option_override): Register the new pass.
8668         * rtl.h (cleanup_barriers): Remove prototype.
8669         (dbr_schedule): Likewise.
8670         * jump.c (cleanup_barriers): Make static.
8671         * reorg.c (dbr_schedule): Likewise.
8673 2013-05-06  Richard Biener  <rguenther@suse.de>
8675         PR tree-optimization/57185
8676         * tree-parloops.c (add_field_for_reduction): Handle anonymous
8677         SSA names properly.
8679 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
8681         PR target/57106
8682         * config/i386/i386.c (add_parameter_dependencies): Add dependence
8683         between "first_arg" and "insn", not "last" and "insn".
8685 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8687         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8688         (find_candidates_in_block): Re-enable slsr_process_phi.
8689         (create_phi_basis): Fix double counting of candidate adjustment.
8691 2013-05-06  Richard Biener  <rguenther@suse.de>
8693         PR middle-end/57147
8694         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8695         the edge is also fallthru, preserve it and just clear the
8696         abnormal flag.
8697         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8698         also complex, preserve that and just clear the fallthru flag.
8699         * tree-inline.c (update_ssa_across_abnormal_edges): Also
8700         update virtual operands.
8702 2013-05-06  Alan Modra  <amodra@gmail.com>
8704         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8705         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8706         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8707         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8708         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8709         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8711 2013-05-06  Alan Modra  <amodra@gmail.com>
8713         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8714         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
8715         (DEFAULT_ASM_ENDIAN): Define.
8716         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8717         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8718         Update -K PIC clause from sysv4.h.
8719         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8720         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8722 2013-05-06  Alan Modra  <amodra@gmail.com>
8724         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8725         twice for little-endian.
8726         (ashrdi3_no_power, ashrdi3): Support little-endian.
8728 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
8730         PR target/55303
8731         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8732         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8733         related expanders.
8734         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8735         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8736         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8737         New predicates.
8739 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
8740             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
8742         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8743         * config/pa/pa.opt: Make mbig-switch a no-op.
8744         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8745         (CASE_VECTOR_MODE): Always return SImode.
8746         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8747         for the !TARGET_BIG_SWITCH case.
8748         * config/pa/pa-linux.h: Likewise.
8749         * config/pa/pa-openbsd.h: Likewise.
8750         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8751         * config/pa/pa.md (short_jump): Remove define_insn.
8752         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8753         (casesi0): Remove define_insn.
8754         (type): Remove btable_branch.
8755         (pa_combine_type): Likewise.
8756         (in_nullified_branch_delay): Likewise.
8757         (in_call_delay): Likewise.
8758         (define_delay): Likewise.
8759         (define_insn_reservation "Z3"): Likewise.
8760         (define_insn_reservation "Z4"): Likewise.
8761         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8762         (pa_adjust_insn_length): Remove adjustment for btable branches.
8763         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8764         and mno-big-switch
8766 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
8768         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8769         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8770         Add m->r,x alternatives.
8771         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8772         splitters using SWI48x mode iterator.
8773         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
8774         TARGET_64BIT.  Add m->x alternative.
8775         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8776         Add o->x alternative.  Enable for TARGET_SSE.
8777         (sse_storeq): Remove expander.
8778         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8779         with memory input operand.
8780         (*vec_extractv2di_1 splitter): New.
8781         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8782         * config/i386/i386.md (ssevecmodelower): New mode attribute.
8784 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8786         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
8787         (INT_LOWPART): Delete.
8788         (extract_MB): Adjust.
8789         (extract_ME): Adjust.
8790         (print_operand): Adjust.
8792 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
8794         * config/rs6000/predicates.md (reg_or_add_cint_operand,
8795         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8796         (reg_or_logical_cint_operand, easy_fp_constant,
8797         logical_const_operand): Delete "CONST_DOUBLE" case.
8798         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8799         "HOST_BITS_PER_WIDE_INT == 64" test.
8800         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
8801         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8802         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8803         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8804         test.
8805         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8806         CONST_DOUBLE DImode/VOIDmode case.
8807         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8808         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8809         CONST_DOUBLE VOIDmode case.
8810         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8811         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8812         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8813         Delete CONST_DOUBLE case.
8814         (splitters for mov FMOVE64 const_double): Delete
8815         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
8816         "HOST_BITS_PER_WIDE_INT >= 64" test.
8817         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8818         case.
8819         (mov DI const_double): Delete.
8821 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
8823         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8824         on op shows all bits zero in mode of a lowpart subreg, return zero.
8826 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
8828         PR target/57150
8829         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8830         to save TFmode registers and DImode to save TImode registers for
8831         caller save operations.
8832         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8833         mark being partially clobbered since they only use the first
8834         double word.
8836         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8837         and TDmode only use the upper 64-bits of each VSX register.
8839 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8841         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8842         (find_candidates_in_block): Disable slsr_process_phi.
8844 2013-05-03  Guozhi Wei  <carrot@google.com>
8846         * coverage.c (coverage_obj_init): Move the construction of gcov
8847         constructor to ...
8848         (build_init_ctor): ... here.
8850 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8852         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
8853         (slsr_cand_d): Redefine def_phi.
8854         (stride_status, phi_adjust_status, count_phis_status): New enums.
8855         (find_phi_def): New.
8856         (find_basis_for_base_expr): New.
8857         (find_basis_for_candidate): Handle hidden bases.
8858         (alloc_cand_and_find_basis): Handle phi candidates.
8859         (slsr_process_phi): New.
8860         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
8861         (create_mul_imm_cand): Likewise.
8862         (create_add_ssa_cand): Exclude phi base candidates.
8863         (create_add_imm_cand): Likewise.
8864         (slsr_process_cast): Likewise.
8865         (slsr_process_copy): Likewise.
8866         (find_candidates_in_block): Handle phi candidates.
8867         (dump_candidate): Likewise.
8868         (unconditional_cands): Delete.
8869         (unconditional_cands_with_known_stride_p): Delete.
8870         (phi_dependent_cand_p): New.
8871         (cand_increment): Handle phi-dependent candidates.
8872         (replace_dependent): Delete.
8873         (replace_mult_candidate): New.
8874         (replace_unconditional_candidate): New.
8875         (incr_vec_index): Move to avoid forward reference.
8876         (create_add_on_incoming_edge): New.
8877         (create_phi_basis): New.
8878         (replace_dependents): Delete.
8879         (replace_conditional_candidate): New.
8880         (phi_add_costs): New.
8881         (replace_uncond_cands_and_profitable_phis): New.
8882         (record_increment): Handle phi adjustments.
8883         (record_phi_increments): New.
8884         (record_increments): Handle phi adjustments.
8885         (phi_incr_cost): New.
8886         (lowest_cost_path): Handle phis.
8887         (total_savings): Likewise.
8888         (analyze_increments): Likewise.
8889         (ncd_with_phi): New.
8890         (ncd_of_cand_and_phis): New.
8891         (nearest_common_dominator_for_cands): Handle phi increments.
8892         (all_phi_incrs_profitable): New.
8893         (replace_profitable_candidates): Handle phi-dependent candidates.
8894         (analyze_candidates_and_replace): Likewise.
8896 2013-05-03  Teresa Johnson  <tejohnson@google.com>
8898         PR bootstrap/57154
8899         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
8900         do not exceed REG_BR_PROB_BASE.
8902 2013-05-03  Jeff Law  <law@redhat.com>
8904         PR tree-optimization/57144
8905         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
8906         operand of the condition will bit into the new type when eliminating
8907         a cast feeding a condition.
8909 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
8911         PR rtl-optimization/57130
8912         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
8913         of COMPARE as in_code to the recursive call if needed.
8915 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
8917         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
8918         (enabled): Handle new members.
8919         * config/i386/sse.md (*vec_concatv2si): Merge from
8920         *vec_concatv2si_sse2 and vec_concatv2si_sse.
8921         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
8923 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
8925         PR tree-optimization/57027
8926         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
8927         for fnms opportunity, check we got the prerequisite kind
8928         of tree / gimple before using accessor functions.
8930 2013-05-03  Richard Biener  <rguenther@suse.de>
8932         * double-int.h (lshift): New overload without precision
8933         and arith argument.
8934         (operator *=, operator +=, operator -=): Move ...
8935         * double-int.c (operator *=, operator +=, operator -=): ... here
8936         and implement more efficiently.
8937         (mul_double_with_sign): Remove.
8938         (lshift_double): Adjust to take unsinged shift argument, push
8939         dispatching code to callers.
8940         (mul_double_wide_with_sign): Add early out for callers that
8941         are not interested in high parts or overflow.
8942         (lshift): New function.
8943         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
8944         dispatch code here.
8945         (lrotate, rrotate): Use logical shifts.
8946         * expr.c (get_inner_reference): Use lshift.
8947         * fixed-value.c (do_fixed_divide): Likewise.
8948         * tree-dfa.c (get_ref_base_and_extent): Likewise.
8949         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
8950         (indirect_refs_may_alias_p): Likewise.
8951         (stmt_kills_ref_p_1): Likewise.
8953 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8955         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
8957 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
8959         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
8960         scalar form of FABD instruction.
8962 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
8964         * lra-constraints.c (process_alt_operands): Add checking alt
8965         number to choose the best alternative.
8967 2013-05-02  Richard Biener  <rguenther@suse.de>
8969         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
8970         bitmap and its handling.
8971         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
8973 2013-05-02  Richard Biener  <rguenther@suse.de>
8975         PR middle-end/57140
8976         * tree-inline.c (copy_loops): Properly handle removed loops.
8977         (copy_cfg_body): Mark destination loops for fixup if source
8978         loops needed fixup.
8980 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
8982         PR target/56732
8983         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
8984         generating simple_return for naked functions.
8986 2013-05-02  Martin Jambor  <mjambor@suse.cz>
8988         PR middle-end/56988
8989         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
8990         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
8991         flags match.
8992         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
8993         ipa_agg_replacement_value structures.
8994         (known_aggs_to_agg_replacement_list): Likewise.
8995         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
8996         (read_agg_replacement_chain): Likewise.
8997         (ipcp_transform_function): Also check that by_ref flags match.
8999 2013-05-02  Richard Biener  <rguenther@suse.de>
9001         * graphds.h (struct graph): Add obstack member.
9002         * graphds.c (new_graph): Initialize obstack and allocate
9003         vertices from it.
9004         (add_edge): Allocate edge from the obstack.
9005         (free_graph): Free the obstack instead of all edges and vertices.
9007 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9009         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9010         divides.
9011         * cfg.c (update_bb_profile_for_threading): Ditto.
9012         * tree-inline.c (copy_bb): Ditto.
9013         (copy_edges_for_bb): Ditto.
9014         (initialize_cfun): Ditto.
9015         (copy_cfg_body): Ditto.
9016         (expand_call_inline): Ditto.
9017         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9018         (estimate_node_size_and_time): Ditto.
9019         (inline_merge_summary): Ditto.
9020         * cgraphclones.c (cgraph_clone_edge): Ditto.
9021         (cgraph_clone_node): Ditto.
9022         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9023         (compute_trg_info): Ditto.
9025 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9027         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9028         S reg when fp attribute set.
9029         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9031 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9033         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9034         New pattern.
9035         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9036         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9037         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9039 2013-05-02  Richard Biener  <rguenther@suse.de>
9041         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9042         (struct instantiate_cache_entry): New type.
9043         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9044         (struct instantiate_cache_type): New type.
9045         (set_instantiated_value, get_instantiated_value): Remove.
9046         (get_instantiated_value_entry): New function.
9047         (instantiate_scev_name): Use the new cache and adjust.
9048         (instantiate_scev_poly): Adjust.
9049         (instantiate_scev_binary): Likewise.
9050         (instantiate_array_ref): Likewise.
9051         (instantiate_scev_convert): Likewise.
9052         (instantiate_scev_not): Likewise.
9053         (instantiate_scev_3): Likewise.
9054         (instantiate_scev_2): Likewise.
9055         (instantiate_scev_r): Likewise.
9056         (instantiate_scev): Likewise.
9057         (resolve_mixers): Likewise.
9059 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9061         PR target/57091
9062         * lra-constraints.c (best_small_class_operands_num): Remove.
9063         (process_alt_operands): Remove small_class_operands_num.  Take
9064         small classes operands into losers and only if the operand is not
9065         matched.  Modify debugging output.
9066         (curr_insn_transform): Remove best_small_class_operands_num.
9067         Print insn name.
9069 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9071         * config/aarch64/aarch64-builtins.c
9072         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9073         * config/aarch64/aarch64-simd-builtins.def
9074         (reduc_splus_): Add new modes.
9075         (reduc_uplus_): New.
9076         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9077         (reduc_uplus_v4sf): Likewise.
9078         (reduc_splus_v4sf): Likewise.
9079         (aarch64_addv<mode>): Likewise.
9080         (reduc_uplus_<mode>): Likewise.
9081         (reduc_splus_<mode>): Likewise.
9082         (aarch64_addvv2di): Likewise.
9083         (reduc_uplus_v2di): Likewise.
9084         (reduc_splus_v2di): Likewise.
9085         (aarch64_addvv2si): Likewise.
9086         (reduc_uplus_v2si): Likewise.
9087         (reduc_splus_v2si): Likewise.
9088         (reduc_<sur>plus_<mode>): New.
9089         (reduc_<sur>plus_v2di): Likewise.
9090         (reduc_<sur>plus_v2si): Likewise.
9091         (reduc_<sur>plus_v4sf): Likewise.
9092         (aarch64_addpv4sf): Likewise.
9093         * config/aarch64/arm_neon.h
9094         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9095         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9096         add UNSPEC_SADDV, UNSPEC_UADDV.
9097         (SUADDV): New.
9098         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9100 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9102         * config/aarch64/arm_neon.h
9103         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9105 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9107         * config/aarch64/aarch64-builtins
9108         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9110 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9112         * config/aarch64/aarch64-simd-builtins.def
9113         (reduc_smax_): New.
9114         (reduc_smin_): Likewise.
9115         (reduc_umax_): Likewise.
9116         (reduc_umin_): Likewise.
9117         (reduc_smax_nan_): Likewise.
9118         (reduc_smin_nan_): Likewise.
9119         (fmax): Remove.
9120         (fmin): Likewise.
9121         (smax): Update for V2SF, V4SF and V2DF modes.
9122         (smin): Likewise.
9123         (smax_nan): New.
9124         (smin_nan): Likewise.
9125         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9126         (<su><maxmin><mode>3): ...This, refactor.
9127         (s<maxmin><mode>3): New.
9128         (<maxmin_uns><mode>3): Likewise.
9129         (reduc_<maxmin_uns>_<mode>): Refactor.
9130         (reduc_<maxmin_uns>_v4sf): Likewise.
9131         (reduc_<maxmin_uns>_v2si): Likewise.
9132         (aarch64_<fmaxmin><mode>: Remove.
9133         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9134         new builtin names.
9135         (vmin<q>_f<32,64>): Likewise.
9136         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9137         (FMAXMIN): New.
9138         (su): Add mappings for smax, smin, umax, umin.
9139         (maxmin): New.
9140         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9141         (FMAXMIN): Rename as...
9142         (FMAXMIN_UNS): ...This.
9143         (maxminv): Remove.
9144         (fmaxminv): Likewise.
9145         (fmaxmin): Likewise.
9146         (maxmin_uns): New.
9147         (maxmin_uns_op): Likewise.
9149 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9151         * config/aarch64/arm_neon.h
9152         (vac<ge, gt><sd>_f<32, 64>): Rename to...
9153         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9154         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9156 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9158         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9159         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9161 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9163         * config/aarch64/aarch64-simd.md
9164         (vcond<mode>_internal): Handle special cases for constant masks.
9165         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9166         (vcondu<mode><mode>): Likewise.
9167         (vcond<v_cmp_result><mode>): New.
9169 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9171         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9172         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9173         * config/aarch64/aarch64-simd-builtins.def
9174         (cmeq): Update to BUILTIN_VALLDI.
9175         (cmgt): Likewise.
9176         (cmge): Likewise.
9177         (cmle): Likewise.
9178         (cmlt): Likewise.
9179         * config/aarch64/arm_neon.h
9180         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9181         to builtins or C as appropriate.
9183 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9185         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9186         (cmgeu): ...This.
9187         (cmhi): Rename to...
9188         (cmgtu): ...This.
9189         * config/aarch64/aarch64-simd.md
9190         (simd_mode): Add SF.
9191         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9192         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9193         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9194         (cstore<mode>_neg): ...This.
9195         * config/aarch64/iterators.md
9196         (VALLF): new.
9197         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9198         (COMPARISONS): New.
9199         (UCOMPARISONS): Likewise.
9200         (optab): Add missing comparisons.
9201         (n_optab): New.
9202         (cmp_1): Likewise.
9203         (cmp_2): Likewise.
9204         (CMP): Likewise.
9205         (cmp): Remove.
9206         (VCMP_S): Likewise.
9207         (VCMP_U): Likewise.
9208         (V_cmp_result): Add DF, SF modes.
9209         (v_cmp_result): Likewise.
9210         (v): Likewise.
9211         (vmtype): Likewise.
9212         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9214 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
9216         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9217         define_insn to define_insn_and_split.
9218         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9219         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9220         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9221         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9222         (thumb2_negscc): Likewise.
9224 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9226         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9228 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
9230         * config/arm/thumb2.md: Remove trailing whitespaces.
9232 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9234         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9235         Use gen_int_mode rather than GEN_INT.
9237 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
9239         * value-prof.c (stream_in_histogram_value): Remove the strayed
9240         debug_gimple_stmt.
9242 2013-04-30  Richard Biener  <rguenther@suse.de>
9244         PR middle-end/57122
9245         * cfghooks.c (split_edge): Properly check for the loop latch edge.
9247 2013-04-30  Richard Biener  <rguenther@suse.de>
9249         PR middle-end/57107
9250         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9252 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
9254         PR rtl-optimization/56957
9255         PR rtl-optimization/57105
9256         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9257         variable.  Use just INSN_UID for determining whether an insn
9258         should be only disconnected from the insn stream.
9259         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9261 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
9263         PR tree-optimization/57104
9264         * tsan.c (instrument_expr): Don't instrument accesses to
9265         DECL_HARD_REGISTER VAR_DECLs.
9267 2013-04-30  Richard Biener  <rguenther@suse.de>
9269         * function.h (loops_for_fn): New inline function.
9270         (set_loops_for_fn): Likewise.
9271         * cfgloop.h (place_new_loop): Add struct function parameter.
9272         (get_loop): Likewise.
9273         (get_loops): Likewise.
9274         (number_of_loops): Likewise.
9275         (fel_next): Adjust.
9276         (fel_init): Likewise.
9277         * cfg.c (get_loop_copy): Adjust.
9278         * cfgloop.c (flow_loops_dump): Likewise.
9279         (record_loop_exits): Likewise.
9280         (verify_loop_structure): Likewise.
9281         * cfgloopanal.c (mark_irreducible_loops): Likewise.
9282         (estimate_reg_pressure_cost): Likewise.
9283         (mark_loop_exit_edges): Likewise.
9284         * cfgloopmanip.c (place_new_loop): Likewise.
9285         (add_loop): Likewise.
9286         (duplicate_loop): Likewise.
9287         * graph.c (draw_cfg_nodes): Likewise.
9288         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9289         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9290         (extract_affine_chrec): Likewise.
9291         (build_scop_iteration_domain): Likewise.
9292         * graphite.c (graphite_initialize): Likewise.
9293         * ira-build.c (create_loop_tree_nodes): Likewise.
9294         (more_one_region_p): Likewise.
9295         (rebuild_regno_allocno_maps): Likewise.
9296         (mark_loops_for_removal): Likewise.
9297         (mark_all_loops_for_removal): Likewise.
9298         (remove_unnecessary_regions): Likewise.
9299         (ira_build): Likewise.
9300         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9301         * loop-init.c (fix_loop_structure): Likewise.
9302         (gate_rtl_move_loop_invariants): Likewise.
9303         (gate_rtl_unswitch): Likewise.
9304         (gate_rtl_unroll_and_peel_loops): Likewise.
9305         (rtl_doloop): Likewise.
9306         * lto-streamer-in.c (input_cfg): Likewise.
9307         * lto-streamer-out.c (output_cfg): Likewise.
9308         * modulo-sched.c (sms_schedule): Likewise.
9309         * predict.c (tree_estimate_probability): Likewise.
9310         (tree_estimate_probability_driver): Likewise.
9311         (estimate_loops): Likewise.
9312         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9313         (move_sese_region_to_fn): Likewise.
9314         (debug_loop_num): Likewise.
9315         * tree-chrec.c (chrec_evaluate): Likewise.
9316         (hide_evolution_in_other_loops_than_loop): Likewise.
9317         (chrec_component_in_loop_num): Likewise.
9318         (reset_evolution_in_loop): Likewise.
9319         (evolution_function_is_invariant_rec_p): Likewise.
9320         * tree-if-conv.c (main_tree_if_conversion): Likewise.
9321         * tree-inline.c (copy_loops): Likewise.
9322         (copy_cfg_body): Likewise.
9323         (tree_function_versioning): Likewise.
9324         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9325         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9326         Likewise.
9327         (add_to_evolution_1): Likewise.
9328         (scev_const_prop): Likewise.
9329         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9330         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9331         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9332         (tree_ssa_lim_initialize): Likewise.
9333         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9334         (verify_loop_closed_ssa): Likewise.
9335         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
9336         (tree_ssa_loop_im): Likewise.
9337         (tree_ssa_loop_unswitch): Likewise.
9338         (tree_vectorize): Likewise.
9339         (check_data_deps): Likewise.
9340         (tree_ssa_loop_ivcanon): Likewise.
9341         (tree_ssa_loop_bounds): Likewise.
9342         (tree_complete_unroll): Likewise.
9343         (tree_complete_unroll_inner): Likewise.
9344         (tree_parallelize_loops): Likewise.
9345         (tree_ssa_loop_prefetch): Likewise.
9346         (tree_ssa_loop_ivopts): Likewise.
9347         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9348         * tree-vectorizer.c (vectorize_loops): Likewise.
9350 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
9352         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9353         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9354         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9355         with EABI_LINK_SPEC.
9357 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9359         PR target/44578
9360         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9361         alternative.
9363 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
9365         PR target/57097
9366         * lra-constraints.c (process_alt_operands): Discourage a bit more
9367         using memory for pseudos.  Print cost dump for alternatives.
9368         Modify cost values for conflicts with early clobbers.
9369         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9371 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9373         PR target/57098
9374         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9376 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9378         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9379         from/to S register.
9380         (movdi_aarch64): Support LDR/STR from/to D register.
9382 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
9384         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9385         or higher by default.
9387 2013-04-29  Richard Biener  <rguenther@suse.de>
9389         PR middle-end/57075
9390         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9391         even if not adding abnormal edges for calls that can make
9392         abnormal gotos.
9394 2013-04-29  Richard Biener  <rguenther@suse.de>
9396         PR middle-end/57103
9397         * tree-cfg.c (move_stmt_op): Fix condition under which to update
9398         TREE_BLOCK.
9399         (move_stmt_r): Remove redundant checking.
9401 2013-04-29  Teresa Johnson  <tejohnson@google.com>
9403         PR bootstrap/57077
9404         * basic-block.h (apply_scale): New function.
9405         (apply_probability): Use apply_scale.
9406         * gimple-streamer-in.c (input_bb): Ditto.
9407         * lto-streamer-in.c (input_cfg): Ditto.
9408         * lto-cgraph.c (merge_profile_summaries): Ditto.
9409         * tree-optimize.c (execute_fixup_cfg): Ditto.
9410         * tree-inline.c (copy_bb): Update comment to use apply_scale.
9411         (copy_edges_for_bb): Ditto.
9412         (copy_cfg_body): Ditto.
9414 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9416         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9417         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9418         (tail_merge_optimize): Handle current_loops == NULL.
9420 2013-04-26  Jeff Law  <law@redhat.com>
9422         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9423         (simplify_cond_using_ranges): Generalize code to simplify
9424         COND_EXPRs where one argument is a constant and the other
9425         is an SSA_NAME created by an integral type conversion.
9427 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9429         * config/arm/arm.md (store_minmaxsi): Use only when
9430         optimize_insn_for_size_p.
9432 2013-04-29  Christian Bruel  <christian.bruel@st.com>
9434         PR target/57108
9435         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9437 2013-04-29  Richard Biener  <rguenther@suse.de>
9439         PR middle-end/57089
9440         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9441         loop tree make sure to schedule a fixup for the child as well.
9442         (expand_omp_for_generic): Properly add loops.
9443         (expand_omp_for_static_nochunk): Likewise.
9444         (expand_omp_for_static_chunk): Likewise.
9445         (expand_omp_for): For the degenerate case fixup loops.
9446         (expand_omp_sections): Fix default bb placement in loops.
9447         (expand_omp_atomic_pipeline): Properly add loops.
9449 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9451         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9453 2013-04-29  Tom de Vries  <tom@codesourcery.com>
9455         * tree-ssa-tail-merge.c: Update header comment.
9457 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9459         * config/aarch64/arm_neon.h
9460         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9461         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9462         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9463         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9464         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9465         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9466         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9467         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9469 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9471         * config/aarch64/aarch64-simd.md
9472         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9473         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9474         fix_trunc, fixuns_trunc.
9475         (ftrunc<VDQF:mode>2): New.
9476         * config/aarch64/iterators.md (optab): Add fix, fixuns.
9477         (fix_trunc_optab): New.
9479 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9481         * config/aarch64/aarch64-builtins.c
9482         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9483         iceilf, lround, iroundf.
9485 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
9487         PR target/54349
9488         * config/i386/i386.h (enum ix86_tune_indices)
9489         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9490         New, split from X86_TUNE_INTER_UNIT_MOVES.
9491         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9492         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9493         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9494         (TARGET_INTER_UNIT_MOVES): Remove.
9495         * config/i386/i386.c (initial_ix86_tune_features): Update.
9496         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9497         (ix86_expand_convert_uns_didf_sse): Use
9498         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9499         (ix86_expand_vector_init_one_nonzero): Ditto.
9500         (ix86_expand_vector_init_interleave): Ditto.
9501         (inline_secondary_memory_needed): Return true for moves from SSE class
9502         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9503         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9504         * config/i386/constraints.md (Yi, Ym): Depend on
9505         TARGET_INTER_UNIT_MOVES_TO_VEC.
9506         (Yj, Yn): New constraints.
9507         * config/i386/i386.md (*movdi_internal): Change constraints of
9508         operand 1 from Yi to Yj and from Ym to Yn.
9509         (*movsi_internal): Ditto.
9510         (*movdf_internal): Ditto.
9511         (*movsf_internal): Ditto.
9512         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9513         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9514         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9515         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9516         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9517         * config/i386/sse.md (movdi_to_sse): Ditto.
9518         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9519         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9520         TARGET_INTER_UNIT_MOVES.
9521         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9522         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9523         instead of TARGET_INTER_UNIT_MOVES.
9524         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9525         operand 1 from Yi to Yj and from Ym to Yn.
9527 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9529         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9530         (float_truncate_hi_): Likewise.
9531         (float_extend_lo_): Likewise.
9532         (float_truncate_lo_): Likewise.
9533         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9534         (aarch64_float_extend_lo_v2df): Likewise.
9535         (vec_unpacks_hi_v4sf): Likewise.
9536         (aarch64_float_truncate_lo_v2sf): Likewise.
9537         (aarch64_float_truncate_hi_v4sf): Likewise.
9538         (vec_pack_trunc_v2df): Likewise.
9539         (vec_pack_trunc_df): Likewise.
9541 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9543         * config/aarch64/aarch64-builtins.c
9544         (aarch64_fold_builtin): Fold float conversions.
9545         * config/aarch64/aarch64-simd-builtins.def
9546         (floatv2si, floatv4si, floatv2di): New.
9547         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9548         * config/aarch64/aarch64-simd.md
9549         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9550         * config/aarch64/iterators.md (FLOATUORS): New.
9551         (optab): Add float, floatuns.
9552         (su_optab): Likewise.
9554 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9556         * config/aarch64/aarch64-builtins.c
9557         (aarch64_builtin_vectorized_function): Use new names for
9558         fcvt builtins.
9559         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9560         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9561         (fcvtzu): Split as...
9562         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9563         (fcvtas): Split as...
9564         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9565         (fcvtau): Split as...
9566         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9567         (fcvtps): Split as...
9568         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9569         (fcvtpu): Split as...
9570         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9571         (fcvtms): Split as...
9572         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9573         (fcvtmu): Split as...
9574         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9575         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9576         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9577         (lfrintnusf, lfrintnudf): Likewise.
9578         * config/aarch64/aarch64-simd.md
9579         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9580         define_insn.
9581         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9582         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9583         (fcvt_pattern): Likewise.
9585 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9587         * config/aarch64/aarch64-simd.md
9588         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9589         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9591 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9593         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9594         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9595         (vrnd<a,m,n,p>_f32): Implement using builtins.
9596         (vrnd<i,x><q>_f<32, 64>): New.
9598 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
9600         * config/aarch64/aarch64-builtins.c
9601         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9602         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9603         (frintz): Rename to...
9604         (btrunc): ...this.
9605         (frintp): Rename to...
9606         (ceil): ...this.
9607         (frintm): Rename to...
9608         (floor): ...this.
9609         (frinti): Rename to...
9610         (nearbyint): ...this.
9611         (frintx): Rename to...
9612         (rint): ...this.
9613         (frinta): Rename to...
9614         (round): ...this.
9615         * config/aarch64/aarch64-simd.md
9616         (aarch64_frint<frint_suffix><mode>): Delete.
9617         (<frint_pattern><mode>2): Convert to insn.
9618         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9619         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9620         (frint_pattern): Likewise.
9621         (frint_suffix): Likewise.
9623 2013-04-29  Richard Biener  <rguenther@suse.de>
9625         PR tree-optimization/57081
9626         * loop-init.c: Include tree-flow.h.
9627         (loop_optimizer_finalize): Free number of iteration estimates.
9628         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9630 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
9632         PR tree-optimization/57083
9633         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9634         non-singleton shift count range, zero extend low_bound for uns case.
9636         * config/i386/predicates.md (general_vector_operand): New predicate.
9637         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9638         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9639         if they aren't nonimmediate operands.  If their original values
9640         satisfy const_vector_equal_evenodd_p, don't shift them.
9641         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9642         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
9643         if not nonimmediate_operand.
9644         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9645         instead of register_operand.
9646         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9648 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
9650         * stor-layout.c (finalize_size_functions): Allocate a structure and
9651         reset cfun before dumping the functions.
9653 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
9655         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9657         PR target/56866
9658         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9659         use xop_pmacsdqh if uns_p.
9660         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9661         the immediate rotate count.
9663 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
9665         * rtl.h (struct rtx_def): Add comment for field jump.
9666         (LRA_SUBREG_P): New macro.
9667         * recog.c (register_operand): Check LRA_SUBREG_P.
9668         * lra.c (lra): Add note at the end of RTL code. Align non-empty
9669         stack frame.
9670         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9671         (lra_final_code_change): Skip subreg change for operators.
9672         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9673         if there are no operand changes.
9674         * lra-constraints.c (curr_insn_set): New.
9675         (match_reload): Set LRA_SUBREG_P.
9676         (emit_spill_move): Ditto.
9677         (check_and_process_move): Use curr_insn_set. Process only single
9678         set insns.  Don't initialize sec_mem_p and change_p.
9679         (simplify_operand_subreg): Use LRA_SUBREG_P.
9680         (reg_in_class_p): New function.
9681         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
9682         of #ifdef.  Add code to remove cycling.
9683         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9684         non-null disp.  Reload inner instead of disp when base and index
9685         are null.  Try to put lo_sum into register.
9686         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9687         (check_and_process_move): Move code for move cost check to
9688         simple_move_p.  Remove equiv_substitution.
9689         (simple_move_p): New function.
9690         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
9691         curr_insn_set.  Call check_and_process_move only for single set
9692         insns.  Use the new function.  Move call of check_and_process_move
9693         after operand equiv substitution and address process.
9695 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9697         PR go/57045
9698         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9699         with nonlocal goto receivers or returns twice calls, ignore
9700         unininitialized values from abnormal edges to nl goto receiver
9701         or returns twice call.
9703 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9705         PR tree-optimization/57051
9706         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9707         and VEC_RSHIFT_EXPR if shift count is a multiple of element
9708         bitsize.
9710 2013-04-26  Richard Biener  <rguenther@suse.de>
9712         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9713         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
9714         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9715         (fixup_loop_arrays_after_move): New function.
9716         (move_sese_region_to_fn): Properly outline the loop tree parts
9717         of the SESE region.
9719 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
9721         * config/i386/i386.md (type, unit): Fix long lines.
9723 2013-04-26  Richard Biener  <rguenther@suse.de>
9725         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9726         (lto-streamer-out.o): Likewise.
9727         * cfgloop.c (init_loops_structure): Export, add struct function
9728         argument and adjust.
9729         (flow_loops_find): Adjust.
9730         * cfgloop.h (enum loop_estimation): Add EST_LAST.
9731         (init_loops_structure): Declare.
9732         * lto-streamer-in.c: Include cfgloop.h.
9733         (input_cfg): Input the loop tree.
9734         * lto-streamer-out.c: Include cfgloop.h.
9735         (output_cfg): Output the loop tree.
9736         (output_struct_function_base): Do not drop PROP_loops.
9738 2013-03-26  Richard Biener  <rguenther@suse.de>
9740         * tree-cfg.c (execute_build_cfg): Build the loop tree.
9741         (pass_build_cfg): Provide PROP_loops.
9742         (move_sese_region_to_fn): Remove loops that are outlined into fn
9743         for now.
9744         * tree-inline.c: Include cfgloop.h.
9745         (initialize_cfun): Do not drop PROP_loops.
9746         (copy_loops): New function.
9747         (copy_cfg_body): Copy loop structure.
9748         (tree_function_versioning): Initialize destination loop tree.
9749         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9750         (pass_parallelize_loops): Do IL verification.
9751         * loop-init.c (loop_optimizer_init): Fixup loops if required.
9752         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9753         the CFG make sure we fixup loops as well.
9754         * tree-ssa-tail-merge.c: Include cfgloop.h.
9755         (replace_block_by): When merging loop latches mark loops for fixup.
9756         * lto-streamer-out.c (output_struct_function_base): Drop
9757         PROP_loops for now.
9758         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9759         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9760         * ipa-split.c: Include cfgloop.h.
9761         (split_function): Add the new return block to the loop tree root.
9762         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9763         whether we have removed the forwarder block.
9764         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9765         * cfgloop.h (place_new_loop): Declare.
9766         * cfgloopmanip.c (place_new_loop): Export.
9767         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9768         (tree-switch-conversion.o): Likewise.
9769         (tree-complex.o): Likewise.
9770         (tree-inline.o): Likewise.
9771         (tree-ssa-tailmerge.o): Likewise.
9772         (ipa-split.o): Likewise.
9773         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9774         (tree-ssa-copy.o): Likewise.
9775         * tree-switch-conversion.c: Include cfgloop.h
9776         (process_switch): If we emit a bit-test cascade, schedule loops
9777         for fixup.
9778         * tree-complex.c: Include cfgloop.h.
9779         (expand_complex_div_wide): Properly add new basic-blocks to loops.
9780         * asan.c: Include cfgloop.h.
9781         (create_cond_insert_point): Properly add new basic-blocks to
9782         loops, schedule loop fixup.
9783         * cfgloop.c (verify_loop_structure): Check that looks are not
9784         marked for fixup.
9785         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9786         to loops.
9787         (expand_omp_for_generic): Likewise.
9788         (expand_omp_sections): Likewise.
9789         (expand_omp_atomic_pipeline): Schedule loops for fixup.
9790         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9791         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9792         is initialized, not when loops are present.
9793         * tree-parloops.c (parallelize_loops): Remove checking here.
9794         * passes.c (init_optimization_passes): Schedule a copy-propagation
9795         pass before complete unrolling of inner loops.
9797 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
9799         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9800         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9801         (colorize_init): Add argument to _WIN32 version.
9802         * toplev.c: Include diagnostic-color.h.
9803         (process_options): Default to -fdiagnostics-color=auto if
9804         GCC_COLORS env var is in the environment.
9805         * common.opt (fdiagnostics-color=): Add Var and Init.
9806         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9807         env var is in the environment, the default is auto rather than never.
9809         * diagnostic.h (file_name_as_prefix): Add context argument.
9810         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
9811         the string as locus.
9812         * langhooks.c (lhd_print_error_function): Adjust caller.
9814 2013-04-25  Lawrence Crowl  <crowl@google.com>
9816         * var-tracking.c (shared_hash_def::htab):
9817         Change type to hash_table.  Update dependent calls and types.
9819 2013-04-25  Lawrence Crowl  <crowl@google.com>
9821         * Makefile.in: Update as needed below.
9823         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9824         Move declaration to after the type's method definitons.
9826         * attribs.c (htab_t scoped_attributes::attribute_hash):
9827         Change type to hash_table.  Update dependent calls and types.
9829         * bitmap.c (htab_t bitmap_desc_hash):
9830         Change type to hash_table.  Update dependent calls and types.
9832         * cselib.c (htab_t cselib_hash_table):
9833         Change type to hash_table.  Update dependent calls and types.
9835         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9836         (hash_string_slot_node): Move implementation into lto-streamer.h
9837         struct string_slot_hasher.
9838         (eq_string_slot_node): Likewise.
9840         * data-streamer-out.c: Update output_block::string_hash_table
9841         dependent calls and types.
9843         * dwarf2cfi.c (htab_t trace_index):
9844         Change type to hash_table.  Update dependent calls and types.
9846         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9847         Change type to hash_table.  Update dependent calls and types.
9848         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
9849         (htab_t optimize_external_refs::map): Likewise.
9850         (htab_t output_comp_unit::extern_map): Likewise.
9851         (htab_t output_comdat_type_unit::extern_map): Likewise.
9852         (htab_t output_macinfo::macinfo_htab): Likewise.
9853         (htab_t optimize_location_lists::htab): Likewise.
9854         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
9856         * except.c (htab_t ehspec_hash_type):
9857         Change type to hash_table.  Update dependent calls and types.
9858         (assign_filter_values::ttypes): Likewise.
9859         (assign_filter_values::ehspec): Likewise.
9860         (sjlj_assign_call_site_values::ar_hash): Likewise.
9861         (convert_to_eh_region_ranges::ar_hash): Likewise.
9863         * gcse.c (htab_t pre_ldst_table):
9864         Change type to hash_table.  Update dependent calls and types.
9866         * ggc-common.c (htab_t saving_htab):
9867         Change type to hash_table.  Update dependent calls and types.
9868         (htab_t loc_hash): Likewise.
9869         (htab_t ptr_hash): Likewise.
9870         (call_count): Rename ggc_call_count.
9871         (call_alloc): Rename ggc_call_alloc.
9872         (loc_descriptor): Rename make_loc_descriptor.
9873         (add_statistics): Rename ggc_add_statistics.
9875         * ggc-common.c (saving_htab):
9876         Change type to hash_table.  Update dependent calls and types.
9878         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
9879         (push_gimplify_context): Likewise.
9880         (pop_gimplify_context): Likewise.
9881         (struct gimple_temp_hash_elt): Added.
9882         (struct gimplify_hasher): Likewise.
9883         (struct gimplify_ctx.temp_htab):
9884         Change type to hash_table.  Update dependent calls and types.
9886         * gimple-fold.c: Include gimplify-ctx.h.
9888         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
9889         Change type to hash_table.  Update dependent calls and types.
9890         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
9891         avoid potential global name collision.
9893         * gimplify.c: Include gimplify-ctx.h.
9894         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
9895         (htab_t gimplify_ctx::temp_htab):
9896         Update dependent calls and types for new type hash_table.
9897         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
9898         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
9900         * gimplify-ctx.h: New.
9901         (struct gimple_temp_hash_elt): Move from gimplify.c.
9902         (class gimplify_hasher): New.
9903         (struct gimplify_ctx): Move from gimple.h.
9904         (htab_t gimplify_ctx::temp_htab):
9905         Change type to hash_table.  Update dependent calls and types.
9907         * graphite-clast-to-gimple.c: Include graphite-htab.h.
9908         (htab_t ivs_params::newivs_index):
9909         Change type to hash_table.  Update dependent calls and types.
9910         (htab_t ivs_params::params_index): Likewise.
9911         (htab_t print_generated_program::params_index): Likewise.
9912         (htab_t gloog::newivs_index): Likewise.
9913         (htab_t gloog::params_index): Likewise.
9915         * graphite.c: Include graphite-htab.h.
9916         4htab_t graphite_transform_loops::bb_pbb_mapping):
9917         Change type to hash_table.  Update dependent calls and types.
9919         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
9920         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
9921         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
9923         * graphite-dependences.c: Include graphite-htab.h.
9924         (loop_is_parallel_p): Change hash table type of parameter.
9926         * graphite-htab.h: New.
9927         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
9928         (extern find_pbb_via_hash): Move from graphite-poly.h.
9929         (extern loop_is_parallel_p): Move from graphite-poly.h.
9930         (extern get_loop_body_pbbs): Move from graphite-poly.h.
9932         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
9933         (extern loop_is_parallel_p): Move to graphite-htab.h.
9934         (extern get_loop_body_pbbs): Move to graphite-htab.h.
9936         * haifa-sched.c (htab_t delay_htab):
9937         Change type to hash_table.  Update dependent calls and types.
9938         (htab_t delay_htab_i2): Likewise.
9940         * ira-color.c (htab_t allocno_hard_regs_htab):
9941         Change type to hash_table.  Update dependent calls and types.
9943         * ira-costs.c (htab_t cost_classes_htab):
9944         Change type to hash_table.  Update dependent calls and types.
9946         * loop-invariant.c (htab_t merge_identical_invariants::eq):
9947         Change type to hash_table.  Update dependent calls and types.
9949         * loop-iv.c (htab_t bivs):
9950         Change type to hash_table.  Update dependent calls and types.
9952         * loop-unroll.c (htab_t opt_info::insns_to_split):
9953         Change type to hash_table.  Update dependent calls and types.
9954         (htab_t opt_info::insns_with_var_to_expand): Likewise.
9956         * lto-streamer.h (struct string_slot): Move from data-streamer.h
9957         (struct string_slot_hasher): New.
9958         (htab_t output_block::string_hash_table):
9959         Change type to hash_table.  Update dependent calls and types.
9961         * lto-streamer-in.c (freeing_string_slot_hasher): New.
9962         (htab_t file_name_hash_table):
9963         Change type to hash_table.  Update dependent calls and types.
9965         * lto-streamer-out.c: Update output_block::string_hash_table dependent
9966         calls and types.
9968         * lto-streamer.c (htab_t tree_htab):
9969         Change type to hash_table.  Update dependent calls and types.
9971         * omp-low.c: Include gimplify-ctx.h.
9973         * passes.c (htab_t name_to_pass_map):
9974         Change type to hash_table.  Update dependent calls and types.
9975         (pass_traverse): Rename to passes_pass_traverse.
9977         * plugin.c (htab_t event_tab):
9978         Change type to hash_table.  Update dependent calls and types.
9980         * postreload-gcse.c (htab_t expr_table):
9981         Change type to hash_table.  Update dependent calls and types.
9982         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
9984         * sese.c (debug_rename_map_1): Make extern.
9985         (htab_t copy_bb_and_scalar_dependences::rename_map):
9986         Change type to hash_table.  Update dependent calls and types.
9988         * sese.h (extern debug_rename_map): Move to .c file.
9990         * store-motion.c (htab_t store_motion_mems_table):
9991         Change type to hash_table.  Update dependent calls and types.
9993         * trans-mem.c (htab_t tm_new_mem_hash):
9994         Change type to hash_table.  Update dependent calls and types.
9996         * tree-browser.c (htab_t TB_up_ht):
9997         Change type to hash_table.  Update dependent calls and types.
9999         * tree-cfg.c (htab_t discriminator_per_locus):
10000         Change type to hash_table.  Update dependent calls and types.
10002         * tree-complex.c: Include tree-hasher.h
10003         (htab_t complex_variable_components):
10004         Change type to hash_table.  Update dependent calls and types.
10006         * tree-eh.c (htab_t finally_tree):
10007         Change type to hash_table.  Update dependent calls and types.
10009         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10010         struct int_tree_hasher.
10011         (extern int_tree_map_eq): Likewise.
10012         (uid_decl_map_hash): Removed.
10013         (extern decl_tree_map_eq): Likewise.
10015         * tree-hasher.h: New.
10016         (struct int_tree_hasher): New.
10017         (typedef int_tree_htab_type): New.
10019         * tree-inline.c: Include gimplify-ctx.h.
10021         * tree-mudflap.c: Include gimplify-ctx.h.
10023         * tree-parloops.c: Include tree-hasher.h.
10024         (htab_t eliminate_local_variables_stmt::decl_address):
10025         Change type to hash_table.  Update dependent calls and types.
10026         (htab_t separate_decls_in_region::decl_copies): Likewise.
10028         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10029         Change type to hash_table.  Update dependent calls and types.
10031         * tree-sra.c (candidates):
10032         Change type to hash_table.  Update dependent calls and types.
10034         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10035         in tree-flow.h.
10036         (int_tree_map_hash): Likewise.
10038         * tree-ssa-dom.c (htab_t avail_exprs):
10039         Change type to hash_table.  Update dependent calls and types.
10041         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10042         Change type to hash_table.  Update dependent calls and types.
10044         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10045         Change type to hash_table.  Update dependent calls and types.
10047         * tree-ssa-phiopt.c (seen_ssa_names):
10048         Change type to hash_table.  Update dependent calls and types.
10050         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10051         Change type to hash_table.  Update dependent calls and types.
10053         * tree-ssa-uncprop.c (equiv):
10054         Change type to hash_table.  Update dependent calls and types.
10056 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10058         PR rtl-optimization/57003
10059         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10060         call note_stores with kill_clobbered_value callback again after
10061         killing regs_invalidated_by_call.
10063 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10065         * config/aarch64/aarch64-simd.md
10066         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10067         (aarch64_simd_bsl<mode>): Likewise.
10068         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10070 2013-04-25  Marek Polacek  <polacek@redhat.com>
10072         PR tree-optimization/57066
10073         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10075 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10077         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10079 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10081         * config/aarch64/aarch64-builtins.c
10082         (aarch64_fold_builtin): New.
10083         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10084         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10085         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10086         * config/aarch64/arm_neon.h
10087         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10089 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10090             Tejas Belagod  <tejas.belagod@arm.com>
10092         * config/aarch64/aarch64-builtins.c
10093         (aarch64_gimple_fold_builtin): New.
10094         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10095         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10096         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10097         (addvv4sf): Update.
10098         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10100 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10102         * config/aarch64/aarch64.md
10103         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10105 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10107         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10108         (*ngcsi_uxtw): New pattern.
10110 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10111             Julian Brown  <julian@codesourcery.com>
10113         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10114         (TB_DREG): Add T_V4HF.
10115         (v4hf_UP): New macro.
10116         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10117         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10118         Handle initialisation of V4HF. Adjust initialisation of reinterpret
10119         built-ins.
10120         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10121         (arm_vector_mode_supported_p): Handle V4HF.
10122         (arm_mangle_map): Handle V4HFmode.
10123         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10124         * config/arm/arm_neon_builtins.def: Add entries for
10125         vcvtv4hfv4sf, vcvtv4sfv4hf.
10126         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10127         (neon_vcvtv4hfv4sf): Likewise.
10128         * config/arm/neon-gen.ml: Handle half-precision floating point
10129         features.
10130         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10131         * config/arm/arm_neon.h: Regenerate.
10132         * config/arm/neon.ml (type elts): Add F16.
10133         (type vectype): Add T_float16x4, T_floatHF.
10134         (type vecmode): Add V4HF.
10135         (type features): Add Requires_FP_bit feature.
10136         (elt_width): Handle F16.
10137         (elt_class): Likewise.
10138         (elt_of_class_width): Likewise.
10139         (mode_of_elt): Refactor.
10140         (type_for_elt): Handle F16, fix error messages.
10141         (vectype_size): Handle T_float16x4.
10142         (vcvt_sh): New function.
10143         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10144         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10145         (string_of_mode): Handle V4HF.
10146         * doc/arm-neon-intrinsics.texi: Regenerate.
10148 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10150         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10151         format specifier in 'X' case.
10153 2013-04-25  Alan Modra  <amodra@gmail.com>
10155         PR target/57052
10156         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10157         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10158         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10159         Repeat for many other rotate/shift and mask patterns using subregs.
10160         Name lshiftrt insns.
10161         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10162         on WORDS_BIG_ENDIAN.
10164 2013-04-25  Alan Modra  <amodra@gmail.com>
10166         * config.gcc: Support little-endian powerpc-linux targets.
10167         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10168         (LINK_OS_LINUX_SPEC): Define.
10169         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10170         Preserve MASK_LITTLE_ENDIAN.
10171         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10172         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10173         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10174         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10175         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10176         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10177         Correct fp word order for little-endian.  Don't shift toc entries
10178         smaller than a word for little-endian.
10179         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10180         (bswapdi2 splits): Correct low-part subreg for little-endian.
10181         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10182         low/high where such is correct only for be.
10183         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10184         little-endian for -mcall-aixdesc.
10186 2013-04-25  Alan Modra  <amodra@gmail.com>
10188         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10189         replace_equiv_address_nv.
10191 2013-04-25  Alan Modra  <amodra@gmail.com>
10193         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10195 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10197         Revert:
10198         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10199         * rtl.h (struct rtx_def): ...
10201 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10203         PR rtl-optimizations/57046
10204         * lra-constraints (split_reg): Set up lra_risky_transformations_p
10205         for multi-reg splits.
10207 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
10209         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10211 2013-04-24  Sterling Augustine  <saugustine@google.com>
10213         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10214         (comp_dir_string, debug_str_dwo_section): New.
10215         (DEBUG_STR_DWO_SECTION): Rename to ...
10216         (DEBUG_DWO_STR_SECTION): ... this.
10217         (DEBUG_NORM_STR_SECTION): Delete.
10218         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10219         (DEBUG_STR_DWO_SECTION_FLAGS): New.
10220         (find_AT_string): Move most logic to ...
10221         (find_AT_string_in_table): ... here.  New.
10222         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10223         add_skeleton_AT_string.  Delete logic.
10224         (output_skeleton_debug_sections): Remove call to
10225         add_top_level_skeleton_die_attrs.
10226         (add_comp_dir_attribute): Move logic to comp_dir_string.
10227         (dwarf2out_init): Initialize debug_str_dwo_section.
10228         (output_indirect_string): Call find_string_form.
10229         (output_indirect_strings): Rewrite.
10230         (prune_unused_types): Empty skeleton_debug_str_hash.
10231         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10232         (dwarf2out_finish):  Call output_indirect_strings.
10234 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10236         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10238 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
10240         * rtl.h (struct rtx_def): Add comment for field jump.
10241         (LRA_SUBREG_P): New macro.
10242         * recog.c (register_operand): Check LRA_SUBREG_P.
10243         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10244         stack frame.
10245         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10246         (lra_final_code_change): Skip subreg change for operators.
10247         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10248         if there are no operand changes.
10249         * lra-constraints.c (curr_insn_set): New.
10250         (match_reload): Set LRA_SUBREG_P.
10251         (emit_spill_move): Ditto.
10252         (check_and_process_move): Use curr_insn_set. Process only single
10253         set insns.  Don't initialize sec_mem_p and change_p.
10254         (simplify_operand_subreg): Use LRA_SUBREG_P.
10255         (reg_in_class_p): New function.
10256         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10257         of #ifdef.  Add code to remove cycling.
10258         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10259         non-null disp.  Reload inner instead of disp when base and index
10260         are null.  Try to put lo_sum into register.
10261         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10262         (check_and_process_move): Move code for move cost check to
10263         simple_move_p.  Remove equiv_substitution.
10264         (simple_move_p): New function.
10265         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10266         curr_insn_set.  Call check_and_process_move only for single set
10267         insns.  Use the new function.  Move call of check_and_process_move
10268         after operand equiv substitution and address process.
10270 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
10272         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10273         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10274         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10276 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
10278         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10280 2013-04-24  Marek Polacek  <polacek@redhat.com>
10282         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10283         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10284         (select_loops_exit_conditions): Likewise.
10285         (number_of_iterations_for_all_loops): Likewise.
10286         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10287         (scev_analysis): Likewise.
10289 2013-04-02  Catherine Moore  <clm@codesourcery.com>
10290             Chao-ying Fu <fu@mips.com>
10292         * config/mips/micromips.md (jraddiusp): New pattern.
10293         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10294         instruction if possible.
10296 2013-04-24  Alan Modra  <amodra@gmail.com>
10298         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10300 2013-04-24  Julian Brown  <julian@codesourcery.com>
10301             Chung-Lin Tang  <cltang@codesourcery.com>
10303         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10304         dependency behavior in enumeration type DIE generation. Add TODO note
10305         to comments about future DW_FORM_sdata/udata re-work of related code.
10307 2013-04-23  Lawrence Crowl  <crowl@google.com>
10309         * Makefile.in: Update as needed below.
10311         * hash-table.h (class hash_table):
10312         Correct many methods with parameter types compare_type to the correct
10313         value_type.  (Correct code was unlikely to notice the change.)
10314         (hash_table::elements_with_deleted) New.
10315         (class hashtable::iterator): New.
10316         (hashtable::begin()): New.
10317         (hashtable::end()): New.
10318         (FOR_EACH_HASH_TABLE_ELEMENT): New.
10320         * statistics.c (statistics_hashes):
10321         Change type to hash_table.  Update dependent calls and types.
10323         * tree-into-ssa.c (var_infos):
10324         Change type to hash_table.  Update dependent calls and types.
10326         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10327         Change type to hash_table.  Update dependent calls and types.
10329         * tree-ssa-loop-im.c (struct mem_ref.refs):
10330         Change type to hash_table.  Update dependent calls and types.
10332         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10333         Change type to hash_table.  Update dependent calls and types.
10335         * tree-ssa-sccvn.c (vn_tables_s::nary):
10336         Change type to hash_table.  Update dependent calls and types.
10337         (vn_tables_s::phis): Likewise.
10338         (vn_tables_s::references): Likewise.
10340         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10341         (vn_reference_eq): Update parameter and return types.
10343         * tree-ssa-structalias.c (pointer_equiv_class_table):
10344         Change type to hash_table.  Update dependent calls and types.
10345         (location_equiv_class_table): Likewise.
10347         * tree-vect-data-refs.c: Consequential changes for making
10348         peeling a hash_table.
10350         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10351         (destroy_loop_vec_info): Dependent hash_table update.
10353         * tree-vectorizer.h (peeling_htab):
10354         Change type to hash_table.  Update dependent calls and types.
10356 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
10358         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10359         to check the register content is equal or not.
10360         * lra-constraints.c (match_reload): Use lra_assign_reg_val
10361         to assign register content record.
10362         * lra-eliminations.c (update_reg_eliminate): Use
10363         lra_update_reg_val_offset to update register content offset.
10364         * lra-int.h (struct lra_reg): Add offset member.
10365         (lra_reg_val_equal_p): New static inline function.
10366         (lra_update_reg_val_offset): New static inline function.
10367         (lra_assign_reg_val): New static inline function.
10368         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10369         to assign register content record.
10370         (initialize_lra_reg_info_element): Initial offset to zero.
10372 2013-04-23  Catherine Moore  <clm@codesourcery.com>
10374         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10375         operands.  Record compression.
10377 2013-04-23  Xinliang David Li  <davidxl@google.com>
10379         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10381 2013-04-23  Richard Biener  <rguenther@suse.de>
10383         PR middle-end/57036
10384         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10385         parameter, only add abnormal goto edges from the copied body
10386         if the call could perform abnormal gotos.
10387         (copy_cfg_body): Adjust.
10389 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
10391         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10393 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
10395         * coretypes.h (gimple_stmt_iterator): Add struct to make
10396         compatible with C.
10398 2013-04-23  Richard Biener  <rguenther@suse.de>
10400         PR tree-optimization/57026
10401         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10402         from SSA names occuring in abnormal PHI nodes.
10404 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10406         * lto/lto.c (print_lto_report_1): Fix LTO report names.
10408 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10410         * lto/lto.c (print_lto_report_1): Declare early.
10411         (read_cgraph_and_symbols): Call print_lto_report_1 early.
10413 2013-04-22  Andi Kleen  <ak@linux.intel.com>
10415         * common.opt (-flto-report-wpa): Add.
10416         * doc/invoke.texi (-flto-report-wpa): Add.
10417         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10418         (lto_main): dito.
10420 2013-04-22  Xinliang David Li  <davidxl@google.com>
10422         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10423         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10424         * Makefile.in: New dependency
10426             David Daney <ddaney.cavm@gmail.com>
10428         * configure.ac (gcc_cv_as_micromips_support):  Use the
10429         --fatal-warnings option.
10430         * configure: Regenerate.
10432 2013-04-22  Marek Polacek  <polacek@redhat.com>
10434         PR sanitizer/56990
10435         * tsan.c (instrument_expr): Don't instrument expression
10436         in case its size is zero.
10438 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
10440         PR target/57032
10441         Revert:
10442         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
10444         * config/alpha/alpha.c (TARGET_LRA_P): New define.
10446 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10448         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10449         (gimple_stmt_iterator): New typedef.
10450         * gimple.h (gimple_stmt_iterator): Rename to...
10451         (gimple_stmt_iterator_d): ... This.
10452         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10453         trees be valid for GIMPLE and GENERIC.
10454         (TARGET_GIMPLE_FOLD_BUILTIN): New.
10455         * gimple-fold.c (gimple_fold_call): Call target hook
10456         gimple_fold_builtin.
10457         * hooks.c (hook_bool_gsiptr_false): New.
10458         * hooks.h (hook_bool_gsiptr_false): New.
10459         * target.def (fold_stmt): New.
10460         * doc/tm.texi: Regenerate.
10462 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
10464         PR target/57018
10465         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10466         a set sp if no stack realignment.
10468 2013-04-22  Nick Clifton  <nickc@redhat.com>
10470         * config.gcc (tilegx-linux): Extend extra_objs rather than
10471         overwriting it.
10472         (tilepro-linux): Likewise.
10474 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10476         * config/aarch64/aarch64-builtins.c
10477         (CF): Remove.
10478         (CF0, CF1, CF2, CF3, CF4, CF10): New.
10479         (VAR<1-12>): Add MAP parameter.
10480         (BUILTIN_*): Likewise.
10481         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10482         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10483         (aarch64_ushl_n<mode>): Likewise.
10484         (aarch64_sshr_n<mode>): Likewise.
10485         (aarch64_ushr_n<mode>): Likewise.
10486         (aarch64_<maxmin><mode>): Likewise.
10487         (aarch64_sqrt<mode>): Likewise.
10488         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10489         (vshr<q>_n_*): Likewise.
10491 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
10493         * config/aarch64/aarch64-builtins.c
10494         (aarch64_simd_builtin_type_mode): Handle SF types.
10495         (sf_UP): Define.
10496         (BUILTIN_GPF): Define.
10497         (aarch64_init_simd_builtins): Handle SF types.
10498         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10499         (frecps): Likewise.
10500         (frecpx): Likewise.
10501         * config/aarch64/aarch64-simd.md
10502         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10503         (aarch64_frecpe<mode>): New.
10504         (aarch64_frecps<mode>): Likewise.
10505         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10506         (v8type): Add frecp<esx>.
10507         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10508         (aarch64_frecps<mode>): Likewise.
10509         * config/aarch64/iterators.md (FRECP): New.
10510         (frecp_suffix): Likewise.
10511         * config/aarch64/arm_neon.h
10512         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10514 2013-04-22  Christian Bruel  <christian.bruel@st.com>
10516         PR target/56995
10517         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10518         (REG_CLASS_NAMES): Idem.
10519         (REG_CLASS_CONTENTS): Idem.
10520         (REGCLASS_HAS_FP_REG): Idem.
10521         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10522         (sh_conditional_register_usage): Idem.
10524 2013-04-21  Jeff Law  <law@redhat.com>
10526         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10527         (ssa_forward_propagate_and_combine): Use it.
10529 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10531         * lra.c: Update the flow chart diagram.
10533 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
10535         PR rtl-optimization/56847
10536         * lra-constraints.c (process_alt_operands): Discourage alternative
10537         with non-matche doffsettable memory constraint fro memory with
10538         known offset.
10540 2013-04-19  Richard Biener  <rguenther@suse.de>
10542         PR tree-optimization/56982
10543         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10544         function.
10545         * gimplify.c (gimplify_call_expr): Notice special calls.
10546         (gimplify_modify_expr): Likewise.
10547         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10548         abnormal control flow receivers.
10549         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10550         in the same way as cfun->has_nonlocal_labels.
10551         (gimple_purge_dead_abnormal_call_edges): Likewise.
10552         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10553         receivers start a basic-block.
10555 2013-04-19  Richard Biener  <rguenther@suse.de>
10557         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10558         member ...
10559         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
10560         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10561         (SLP_TREE_LOAD_PERMUTATION): Add.
10562         (vect_transform_slp_perm_load): Adjust prototype.
10563         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10564         (vect_free_slp_instance): Likewise.
10565         (vect_create_new_slp_node): Likewise.
10566         (vect_supported_slp_permutation_p): Remove.
10567         (vect_slp_rearrange_stmts): Adjust.
10568         (vect_supported_load_permutation_p): Likewise.  Inline
10569         vect_supported_slp_permutation_p here.
10570         (vect_analyze_slp_instance): Compute load permutations per
10571         slp node instead of per instance.
10572         (vect_get_slp_defs): Adjust.
10573         (vect_transform_slp_perm_load): Likewise.
10574         (vect_schedule_slp_instance): Remove redundant code.
10575         (vect_schedule_slp): Remove hack for PR56270, add it ...
10576         * tree-vect-stmts.c (vectorizable_load): ... here, do not
10577         CSE loads for SLP.  Adjust.
10579 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10581         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10582         spelling in two comments.
10584 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
10586         PR target/56797
10587         * config/arm/arm.c (load_multiple_sequence): Require SP
10588         as base register for loads if SP is in the register list.
10590 2013-04-19  Martin Jambor  <mjambor@suse.cz>
10592         PR tree-optimization/56718
10593         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10594         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10595         and made public.  Adjusted all callers.
10596         (ipa_intraprocedural_devirtualization): New function.
10597         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10598         (ipa_intraprocedural_devirtualization): Likewise.
10599         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10601 2013-04-19  Richard Biener  <rguenther@suse.de>
10603         PR tree-optimization/57000
10604         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10606 2013-04-19  Terry Guo  <terry.guo@arm.com>
10608         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10609         Replace with ...
10610         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
10611         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10612         (cortex_m4_fmacs): Use new reservations.
10613         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10615 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
10617         PR rtl-optimization/56999
10618         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10619         related code.
10620         (lra_coalesce): Remove split_origin_bitmap and related code.
10621         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
10622         ranges if necessary.
10624 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
10626         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10627         New array.
10628         (ix86_expand_call): Remove clobbered_registers array and use
10629         x86_64_ms_sysv_extra_clobbered_registers instead.
10630         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10631         Declare here.
10632         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10633         predicate.
10634         * config/i386/i386.md (*call_rex64_ms_sysv): Use
10635         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
10636         (*call_value_rex64_ms_sysv): Ditto.
10638 2013-04-18  Cary Coutant  <ccoutant@google.com>
10640         * dwarf2out.c (output_pubnames): Check die_perennial_p of
10641         parent instead of die_mark.
10643 2013-04-18  Diego Novillo  <dnovillo@google.com>
10645         * gimple.c (create_gimple_tmp): New.
10646         (get_expr_type): New.
10647         (build_assign): New.
10648         (build_type_cast): New.
10649         * gimple.h (enum ssa_mode): Define.
10650         (gimple_seq_set_location): New.
10651         * asan.c (build_check_stmt): Change some gimple_build_* calls
10652         to use build_assign and build_type_cast.
10654 2013-04-18  Richard Biener  <rguenther@suse.de>
10656         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10657         handle negative step.  Remove redundant checks.
10658         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10659         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10660         for negative step and grouped loads fail to vectorize.
10662 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10664         * emit-rtl.c (reset_insn_used_flags): New function.
10665         (reset_all_used_flags): Use it.
10666         (verify_insn_sharing): New function.
10667         (verify_rtl_sharing): Fix verification for SEQUENCEs.
10669 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
10671         PR tree-optimization/56984
10672         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10673         and (x >> M) >= N don't register any assertion if N << M is the
10674         minimum value.
10676 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
10678         * lower-subreg.c (resolve_simple_move): If called self-recursive,
10679         do not delete_insn insns that have not yet been emitted, only
10680         unlink them with remove_insn.
10681         * df-scan.c (df_insn_delete): Revert r197492.
10683 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
10685         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10686         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10688 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10690         * config/arm/arm.md (movsicc_insn): Convert define_insn into
10691         define_insn_and_split.
10692         (and_scc,ior_scc,negscc): Likewise.
10693         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10695 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
10697         * config/arm/arm.c (use_return_insn): Return 0 for targets that
10698         can benefit from using a sequence of LDRD instructions in epilogue
10699         instead of a single LDM instruction.
10701 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
10703         PR 45688
10704         * doc/extend.texi: Fix typo.
10706 2013-04-17  Richard Biener  <rguenther@suse.de>
10708         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10709         (vect_build_slp_tree): ... here.
10710         (vect_build_slp_tree_1): Compute which stmts of the SLP group
10711         match.  Remove special-casing of mismatched complex loads.
10712         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10713         re-try the match with swapped commutative operands.
10714         (vect_supported_load_permutation_p): Remove special-casing of
10715         mismatched complex loads.
10716         (vect_analyze_slp_instance): Adjust.
10718 2013-04-17  Richard Biener  <rguenther@suse.de>
10720         PR rtl-optimization/56921
10721         * cfgloop.h (struct loop): Add simple_loop_desc member.
10722         (struct niter_desc): Mark with GTY(()).
10723         (simple_loop_desc): Do not use aux field but simple_loop_desc.
10724         * loop-iv.c (get_simple_loop_desc): Likewise.
10725         (free_simple_loop_desc): Likewise.
10727         Revert
10728         2013-04-16  Richard Biener  <rguenther@suse.de>
10730         PR rtl-optimization/56921
10731         * loop-init.c (pass_rtl_move_loop_invariants): Add
10732         TODO_do_not_ggc_collect to todo_flags_finish.
10733         (pass_rtl_unswitch): Same.
10734         (pass_rtl_unroll_and_peel_loops): Same.
10735         (pass_rtl_doloop): Same.
10737 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
10739         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10740         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10741         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10742         references.
10743         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10744         * tree-streamer.c (record_common_node): Adjust reference in comment.
10746 2013-04-17  Terry Guo  <terry.guo@arm.com>
10748         * config/arm/cortex-m4.md: Add a new bypass.
10750 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10752         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10753         New pattern.
10754         (*subs_<optab><mode>_multp2): New pattern.
10755         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10756         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10758 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10760         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10761         (*subs_mul_imm_<mode>): New pattern.
10763 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
10765         PR target/56948
10766         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10767         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
10768         (vsx_movti_32bit): Change j->wa to O->wa.
10770 2013-04-16  Richard Biener  <rguenther@suse.de>
10772         PR rtl-optimization/56921
10773         * loop-init.c (pass_rtl_move_loop_invariants): Add
10774         TODO_do_not_ggc_collect to todo_flags_finish.
10775         (pass_rtl_unswitch): Same.
10776         (pass_rtl_unroll_and_peel_loops): Same.
10777         (pass_rtl_doloop): Same.
10779 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
10781         * config/arm/arm.c (emit_multi_reg_push): New declaration
10782         for an existing function.
10783         (arm_emit_strd_push): New function.
10784         (arm_expand_prologue): Used here.
10785         (arm_emit_ldrd_pop): New function.
10786         (arm_expand_epilogue): Used here.
10787         (arm_get_frame_offsets): Update condition.
10788         (arm_emit_multi_reg_pop): Add a special case for load of a single
10789         register with writeback.
10791 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
10793         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10794         description.
10796 2013-04-16  Richard Biener  <rguenther@suse.de>
10798         PR tree-optimization/56756
10799         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10800         (first_mem_ref_loc): New.
10801         (execute_sm): Place the load temporarily before a previous
10802         access instead of in the latch edge to ensure its SSA dependencies
10803         are defined at points dominating the load.
10805 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
10807         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10808         correct fix by moving header and footer insn to the footer of
10809         the merged basic block.  Clear BB_END of the merged-away block.
10811         PR middle-end/43631
10812         * emit-rtl.c (make_note_raw): New function.
10813         (link_insn_into_chain): New static inline function.
10814         (add_insn): Use it.
10815         (add_insn_before, add_insn_after): Factor insn chain linking code...
10816         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10817         using link_insn_into_chain.
10818         (note_outside_basic_block_p): New helper function for emit_note_after
10819         and emit_note_before.
10820         (emit_note_after): Use nobb variant of add_insn_after if the note
10821         should not be contained in a basic block.
10822         (emit_note_before): Use nobb variant of add_insn_before if the note
10823         should not be contained in a basic block.
10824         (emit_note_copy): Use make_note_raw.
10825         (emit_note): Likewise.
10826         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10827         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10828         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10829         the moved barrier the tail of the basic block it follows.
10830         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10832 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
10834         PR tree-optimization/56962
10835         * gimple-ssa-strength-reduction.c (record_increment): Only set
10836         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10837         either rhs1 or rhs2 is equal to c->base_expr.
10839 2013-04-15  Richard Biener  <rguenther@suse.de>
10841         PR tree-optimization/56933
10842         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10843         member.
10844         (GROUP_READ_WRITE_DEPENDENCE): Remove.
10845         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10846         * tree-vect-data-refs.c (vect_analyze_group_access): Move
10847         dependence check ...
10848         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
10849         ... here.
10850         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
10851         GROUP_READ_WRITE_DEPENDENCE.
10853 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10855         * emit-rtl.c (reset_all_used_flags): New function.
10856         (verify_rtl_sharing): Call reset_all_used_flags before and after
10857         performing the checks.
10859 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10861         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
10862         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
10863         * config/arm/constraints.md (De): New constraint.
10864         * config/arm/neon.md (anddi3_neon): Delete.
10865         (neon_vand<mode>): Expand to standard anddi3 pattern.
10866         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
10867         Move earlier in the file.
10868         (neon_inv_logic_op2): Likewise.
10869         (arm_anddi_operand_neon): New predicate.
10871 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10873         * configure.ac (gcc_cv_ld_as_needed): Set
10874         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
10875         Use -z ignore, -z record on *-*-solaris2*.
10876         (HAVE_LD_AS_NEEDED): Update comment.
10877         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
10878         * configure: Regenerate.
10879         * config.in: Regenerate.
10880         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
10881         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
10882         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
10883         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
10884         equivalents.  Fix markup.
10885         * doc/tm.texi: Regenerate.
10887 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
10889         * config/i386/i386.opt: New option mstack-protector-guard=.
10890         * config/i386/i386-opts.h: Add enum stack_protector_guard.
10891         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
10892         TARGET_SSP_TLS_GUARD.
10893         * config/i386/i386.c (ix86_option_override_internal): Set
10894         ix86_stack_protector_guard.
10895         * config/i386/i386.md (stack_protect_set): Enable for
10896         TARGET_SSP_TLS_GUARD only.
10897         (stack_protect_set_<mode>): Ditto.
10898         (stack_protect_test): Ditto.
10899         (stack_protect_test_<mode>): Ditto.
10900         * doc/invoke.texi (i386 Option): Document.
10902 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
10904         PR target/56890
10905         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
10906         (S_MODES): Set H_MODE bit.
10907         (SF_MODES): Set only S_MODE and SF_MODE bits.
10908         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
10909         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
10910         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
10911         <MODE_FLOAT>: Likewise.
10913 2013-04-15  Joey Ye  <joey.ye@arm.com>
10915         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
10917 2013-04-15  Joey Ye  <joey.ye@arm.com>
10919         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
10920         for real far jump.
10921         (thumb_far_jump_used_p): Count instruction size and set
10922         far_jump_used.
10924 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
10926         * reorg.c (fill_simple_delay_slots): Reindent block of code.
10927         * resource.c (mark_target_live_regs): Reformat conditional block.
10929 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10931         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
10932         notes, they are emitted only just before final.
10933         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
10935 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
10937         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
10938         * cfgrtl.c (delete_insn): Call it here instead.
10939         * lra-spills.c (lra_final_code_change): Use delete_insn.
10940         * haifa-sched.c (sched_remove_insn): Likewise.
10941         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
10942         returning to the nop pool.
10943         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
10944         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
10946 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10948         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
10949         * doc/tm.texi: Regenerated.
10951 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
10953         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
10954         QImode checks.
10956 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
10958         * df-core.c (df_find_def): Compare register numbers.
10959         (df_find_use): Likewise.
10961 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
10963         PR target/56903
10964         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
10965         lra_in_progress for return.
10967 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
10969         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
10970         define_insn into define_insn_and_split and emit movsicc patterns.
10972 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
10974         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
10976 2013-04-12  Richard Biener  <rguenther@suse.de>
10978         * tree-pass.h (TODO_do_not_ggc_collect): New.
10979         * passes.c (execute_one_ipa_transform_pass): Honor
10980         TODO_do_not_ggc_collect.
10981         (execute_one_pass): Likewise.
10983         Revert
10984         2013-04-10  Richard Biener  <rguenther@suse.de>
10986         * passes.c (init_optimization_passes): Remove reload pass.
10987         * ira.c (do_reload): Merge into ...
10988         (ira): ... this.
10989         (rest_of_handle_reload): Remove.
10990         (pass_reload): Likewise.
10991         * config/i386/i386.c (ix86_option_override): Refer to ira instead
10992         of reload for vzeroupper pass placement.
10994 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
10996         PR tree-optimization/56918
10997         PR tree-optimization/56920
10998         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
10999         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11000         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11001         use wide_mul_with_sign method.
11003 2013-04-12  Richard Biener  <rguenther@suse.de>
11005         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11006         not be considered a gimple constant.
11008 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11010         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11011         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11012         operations.
11014 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11015             Jakub Jelinek  <jakub@redhat.com>
11017         * opts.c: Include diagnostic-color.h.
11018         (common_handle_option): Handle OPT_fdiagnostics_color_.
11019         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11020         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11021         (diagnostic-color.o): New.
11022         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11023         (diagnostic_color_rule): New enum.
11024         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11025         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11026         the location string.
11027         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11028         either NULL, or color kind.
11029         * diagnostic-color.c: New file.
11030         * diagnostic-color.h: New file.
11031         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11032         arguments.
11033         * doc/invoke.texi (-fdiagnostics-color): Document.
11034         * pretty-print.h (pp_show_color): Define.
11035         (struct pretty_print_info): Add show_color field.
11036         * diagnostic.c: Include diagnostic-color.h.
11037         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11038         macros.  Colorize error:, warning: etc. strings and also the location
11039         string.
11040         (diagnostic_show_locus): Colorize the caret line.
11041         * pretty-print.c: Include diagnostic-color.h.
11042         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11043         inside of %< %> quotes or quoted through q format modifier.
11045 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11047         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11049 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11051         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11052         code in CC_NZ mode.
11053         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11054         pattern.
11056 2013-04-11  Marek Polacek  <polacek@redhat.com>
11058         PR tree-optimization/48184
11059         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11061 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11063         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11064         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11065         (skip_simple_arithmetic): Tidy up.
11066         * tree.h (skip_simple_constant_arithmetic): Declare.
11068 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11070         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11072 2013-04-11  Richard Biener  <rguenther@suse.de>
11074         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11075         generate vector constants.
11077 2013-04-11  Richard Biener  <rguenther@suse.de>
11079         PR tree-optimization/56878
11080         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11081         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11082         New function.
11083         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11084         Prefer to align the DR with the most invariant base address.
11086 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11088         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11089         comment.
11091 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11093         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11094         floating-point vector comparisons against 0.
11096 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11098         PR tree-optimization/56899
11099         * fold-const.c (extract_muldiv_1): Apply distributive law
11100         only if TYPE_OVERFLOW_WRAPS (ctype).
11102 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11104         PR target/56124
11105         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11106         loading has side effect.
11108 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
11110         * config/sparc/sparc.c: Include tree-pass.h.
11111         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11112         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
11113         head of file.  Change return type.  Split off gate function.
11114         (sparc_gate_work_around_errata): New function.
11115         (pass_work_around_errata): New pass definition.
11116         (insert_pass_work_around_errata) New pass insert definition to
11117         insert pass_work_around_errata just after delayed-branch scheduling.
11118         (sparc_option_override): Insert the pass.
11119         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11121 2013-04-10  David S. Miller  <davem@davemloft.net>
11123         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11124         or -mcpu=hypersparc.
11126         * target.def (cstore_mode): New hook.
11127         * target.h: Include insn-codes.h
11128         * targhooks.c: Likewise.
11129         (default_cstore_mode): New function.
11130         * targhooks.h: Declare it.
11131         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11132         * doc/tm.texi: Rebuild.
11133         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11134         target hook, rather than inspecting the insn_data.
11135         * config/sparc/sparc.c (sparc_cstore_mode): New function.
11136         (TARGET_CSTORE_MODE): Redefine.
11137         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11138         result patterns.
11139         * config/sparc/predicates.md (cstore_result_operand): New special
11140         predicate.
11141         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11142         Use it for operand 0.
11143         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11144         (*snesi_special): Likewise.
11145         (*snesi_zero): Likewise.
11146         (*seqsi_zero): Likewise.
11147         (*sltu_insn): Likewise.
11148         (*sgeu_insn): Likewise.
11149         (*seqdi_special): Make operand 0 and comparison operation be of
11150         DImode.
11151         (*snedi_special): Likewise.
11152         (*snedi_special_vis3): Likewise.
11153         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11154         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11155         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11156         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11157         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11158         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11159         (*sltu_extend_sp64): Likewise.
11160         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11161         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11162         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11163         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11164         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11166 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
11168         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11169         (aarch64_start_file): Use the new function.
11171 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11172             Jason Merrill  <jason@redhat.com>
11174         * common.opt: Add -gdwarf.
11175         * opts.c (common_handle_option): Handle it.
11176         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11178 2013-04-10  Richard Biener  <rguenther@suse.de>
11180         * passes.c (execute_todo): Do not call ggc_collect conditional here.
11181         (execute_one_ipa_transform_pass): But unconditionally here.
11182         (execute_one_pass): And here.
11183         (init_optimization_passes): Remove reload pass.
11184         * tree-pass.h (TODO_ggc_collect): Remove.
11185         (pass_reload): Likewise.
11186         * ira.c (do_reload): Merge into ...
11187         (ira): ... this.
11188         (rest_of_handle_reload): Remove.
11189         (pass_reload): Likewise.
11190         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11191         of reload for vzeroupper pass placement.
11192         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11193         and todo_flags_finish of all passes.
11195 2013-04-10  Richard Biener  <rguenther@suse.de>
11197         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11198         first_const_oprnd field, rename first_def_type to first_op_type.
11199         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11200         (vect_get_and_check_slp_defs): Always use the type of the
11201         operand.  Allow mixed vect_external_def, vect_constant_def types.
11202         (vect_get_constant_vectors): Handle mixed vect_external_def,
11203         vect_constant_def types.
11205 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
11207         PR tree-optimization/55524
11208         * tree-ssa-math-opts.c
11209         (convert_mult_to_fma): Don't use an fms construct
11210         when we don't have an fms operation, but fnma, and it looks
11211         likely that we'll be able to use the latter.
11213 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11215         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11216         function.
11217         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11218         inline fail caused by overwritable functions.
11220 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
11222         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11223         unnecessary bits in the constant power of two case.
11225 2013-04-10  Richard Biener  <rguenther@suse.de>
11227         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11228         broken code swapping operands.
11229         (vect_build_slp_tree): Do not compute load permutations here.
11230         (vect_analyze_slp_instance): Compute load permutations here,
11231         after building the SLP tree.
11233 2013-04-09  Christian Bruel  <christian.bruel@st.com>
11235         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11236         of next/prev_real_insn.
11238 2013-04-09  Jan Hubicka  <jh@suse.cz>
11240         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11241         Drop aliased parameter.
11242         (function_and_variable_visibility): Do not handle alias pairs.
11243         * cgraph.c (varpool_externally_visible_p): Update prototype.
11244         * varpool.c (varpool_add_new_variable): Update.
11246 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11248         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11250 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
11252         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11254         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11256 2013-04-09  Marek Polacek  <polacek@redhat.com>
11258         PR tree-optimization/48762
11259         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11261 2013-04-09  Richard Biener  <rguenther@suse.de>
11263         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11264         dealing with cost.
11265         (vect_build_slp_tree): Likewise.
11266         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11267         calculating the cost of a SLP instance.
11268         (vect_analyze_slp_instance): Use it from here, after building
11269         the SLP tree.
11271 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11273         PR middle-end/56883
11274         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11275         expand_omp_for_static_chunk): Use simple_p = true in
11276         force_gimple_operand_gsi calls when assigning to addressable decls.
11278 2013-04-09  Jeff Law  <law@redhat.com>
11280         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11281         when the boolean was created by converting a wider object which
11282         had a boolean range.
11284 2013-04-09  Richard Biener  <rguenther@suse.de>
11286         * tree-vectorizer.h (slp_void_p): Remove.
11287         (slp_tree): Typedef before _slp_tree declaration.
11288         (struct _slp_tree): Use a vector of slp_tree as children.
11289         (vect_get_place_in_interleaving_chain): Remove.
11290         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11291         Move ...
11292         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11293         and make static.
11294         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11295         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11296         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11297         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11298         Use slp_node instead of slp_void_p and adjust.
11300 2013-04-09  Richard Biener  <rguenther@suse.de>
11302         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11303         work that is not necessary.
11305 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
11307         PR tree-optimization/56854
11308         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11309         forward into clobber stmts if it would change MEM_REF lhs into
11310         non-MEM_REF.
11312 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
11314         * tree.c (type_hash_lookup, type_hash_add): Make static.
11315         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11317 2013-04-09  Richard Biener  <rguenther@suse.de>
11319         * tree.h (unsave_expr_now): Remove.
11320         * tree-inline.c (mark_local_for_remap_r): Remove.
11321         (unsave_expr_1): Likewise.
11322         (unsave_r): Likewise.
11323         (unsave_expr_now): Likewise.
11324         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11325         (propagate_tree_value): Likewise.
11327 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
11329         * doc/rtl.texi (sequence): Rewrite documentation to match the
11330         current use of SEQUENCE rtl objects.
11331         * rtl.def (SEQUENCE): Likewise.
11333         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11334         Update documentation.
11335         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11336         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11338         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11340 2013-04-08  Teresa Johnson  <tejohnson@google.com>
11342         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11343         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11344         methods.
11345         (estimate_edge_size_and_time): Add comment to suggest using rounding
11346         methods.
11347         (estimate_node_size_and_time): Ditto.
11348         (remap_edge_change_prob): Use helper rounding divide methods.
11349         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11350         (gimple_mod_pow2_value_transform): Ditto.
11351         (gimple_mod_subtract_transform): Ditto.
11352         (gimple_ic_transform): Ditto.
11353         (gimple_stringops_transform): Ditto.
11354         * stmt.c (conditional_probability): Ditto.
11355         (emit_case_dispatch_table): Ditto.
11356         * lto-cgraph.c (merge_profile_summaries): Ditto.
11357         * tree-optimize.c (execute_fixup_cfg): Ditto.
11358         * cfgcleanup.c (try_forward_edges): Ditto.
11359         * cfgloopmanip.c (scale_loop_profile): Ditto.
11360         (loopify): Ditto.
11361         (duplicate_loop_to_header_edge): Ditto.
11362         (lv_adjust_loop_entry_edge): Ditto.
11363         * tree-vect-loop.c (vect_transform_loop): Ditto.
11364         * profile.c (compute_branch_probabilities): Ditto.
11365         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11366         * lto-streamer-in.c (input_cfg): Ditto.
11367         * gimple-streamer-in.c (input_bb): Ditto.
11368         * ipa-cp.c (update_profiling_info): Ditto.
11369         (update_specialized_profile): Ditto.
11370         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11371         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11372         rounding methods.
11373         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11374         (compute_trg_info): Ditto.
11375         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11376         (purge_dead_edges): Ditto.
11377         * loop-unswitch.c (unswitch_loop): Ditto.
11378         * cgraphclones.c (cgraph_clone_edge): Ditto.
11379         (cgraph_clone_node): Ditto.
11380         * tree-inline.c (copy_bb): Ditto.
11381         (copy_edges_for_bb): Ditto.
11382         (initialize_cfun): Ditto.
11383         (copy_cfg_body): Ditto.
11384         (expand_call_inline): Ditto.
11386 2013-04-08  Kai Tietz  <ktietz@redhat.com>
11388         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11389         TARGET_CYGWIN64 by TARGET_64BIT.
11391 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11393         * config/epiphany/epiphany.md (GPR_1): New constant.
11394         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11395         * config/epiphany/epiphany.c (gen_compare_reg):
11396         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11397         is already in place.
11398         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11399         Don't require being called during rtl expansion; If y operlaps r0,
11400         return 0.
11401         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11402         (epiphany_expand_epilogue): Likewise.
11404         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11405         Don't use CC_FPmode for ORDERED / UNORDERED.
11406         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11408         * config/epiphany/constraints.md (CnL): New constraint.
11409         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11410         * config/epiphany/predicates.md (add_operand): Allow 1024.
11412         * config/epiphany/epiphany.md (logical_op): New code iterator.
11413         (op_mnc): New code attribute.
11414         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11415         (mov_f+1, mov_f+2): New peephole2 patterns.
11417         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11418         (cstoresi4): Also allow re-use of zero result when doing a NE
11419         comparison to a non-zero operand.
11420         Use (clobber (scratch)) for first insn if the gpr output is not needed.
11422         * config/epiphany/epiphany.md (<insn_opname>v2si3):
11423         Use gen_addsi3_i / gen_subsi3_i.
11425 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11427         PR c++/34949
11428         PR c++/50243
11429         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11430         contain anything but clobbers, at most one __builtin_stack_restore,
11431         optionally debug stmts and final resx, and if it has at least one
11432         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
11433         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11434         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11435         which isn't defaut definition, remove them.
11436         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11437         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11438         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11439         with MEM_REF LHS with SSA_NAME address.
11441 2013-04-08  Jeff Law  <law@redhat.com>
11443         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11445 2013-04-08  Richard Biener  <rguenther@suse.de>
11447         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11448         extra newline.
11449         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11450         determined vector type.
11451         (vect_analyze_data_refs): Likewise.
11452         (vect_get_new_vect_var): Adjust.
11453         (vect_create_destination_var): Preserve SSA name versions.
11454         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11455         not dump anything here.
11457 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
11459         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11460         Add member lr_slot_known.
11461         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11462         if necessary.
11463         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11464         Remove code that sets lr_slot_offset according to what a previous
11465         version of epiphany_emit_save_restore used to do.
11466         (epiphany_emit_save_restore): When doing an lr save or restore,
11467         set/verify lr_slot_known and lr_slot_offset.
11469 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
11471         PR target/54338
11472         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11473         in ALL_REGS.
11475 2013-04-08  Richard Biener  <rguenther@suse.de>
11477         * alias.c (find_base_term): Fix thinko in previous change.
11479 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
11481         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11482         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11483         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
11484         if possible to compute val.
11485         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11486         For QImode integers don't require anything about precision.  Use
11487         const_with_all_bytes_same to find out if the constant doesn't have
11488         repeated bytes in it.
11490 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11492         * config/s390/s390.c (s390_expand_insv): Only accept insertions
11493         within mode size.
11495 2013-04-08  Marek Polacek  <polacek@redhat.com>
11497         PR rtl-optimization/48182
11498         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11499         value to 1.
11501 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11503         PR target/55487
11504         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11505         nuses, make sure we have a label.
11507 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11509         PR target/56843
11510         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11511         (rs6000_emit_swdiv_low_precision): Remove.
11512         (rs6000_emit_swdiv): Rewrite to handle between one and four
11513         iterations of Newton-Raphson generally; modify required number of
11514         iterations for some cases.
11515         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11517 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
11519         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11520         set-but-unused variable.
11522         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11523         basic blocks of released function bodies garbage-collectable.
11525         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11526         (struct rtl_opt_pass): Add TODO_df_finish.
11528         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11530 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11532         * config/arm/constraints.md (q): New constraint.
11533         * config/arm/ldrdstrd.md: New file.
11534         * config/arm/arm.md (ldrdstrd.md) New include.
11535         (arm_movdi): Use "q" instead of "r" constraint
11536         for double-word memory access.
11537         (movdf_soft_insn): Likewise.
11538         * config/arm/vfp.md (movdi_vfp): Likewise.
11539         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11540         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11541         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11542         (mem_ok_for_ldrd_strd): Likewise.
11543         (output_move_double): Update assertion.
11545 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11547         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11549 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11551         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11552         define_insn_and_split.
11553         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11555 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11557         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11558         define_insn_and_split.
11559         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11560         (shiftsi3_compare): New pattern.
11561         (rrx): New pattern.
11562         * config/arm/unspecs.md (UNSPEC_RRX): New.
11564 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11566         * config/arm/arm.md (negdi_extendsidi): New pattern.
11567         (negdi_zero_extendsidi): Likewise.
11569 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11571         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11572         define_insn_and_split.
11573         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11574         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11576 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11578         * config/arm/arm.md (arm_subdi3): Convert define_insn into
11579         define_insn_and_split.
11580         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11581         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11583 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11585         * config/arm/arm.md (subsi3_carryin): New pattern.
11586         (subsi3_carryin_const): Likewise.
11587         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11588         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11590 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11592         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11594 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
11596         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
11597         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
11599 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11601         * config/arm/arm.c (arm_expand_builtin): Change fcode
11602         type to unsigned int.
11604 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11606         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11608 2013-04-04  Ian Lance Taylor  <iant@google.com>
11610         * doc/standards.texi (Standards): The Go frontend supports the Go 1
11611         language standard.
11613 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
11615         PR middle-end/56729
11616         * df-scan.c (df_insn_delete): Disable failing assert.
11618 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11620         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11621         New function prototype.
11622         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11623         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11624         (arm_builtin_vectorized_function): New function.
11626 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11628         * config/arm/arm_neon_builtins.def: New file.
11629         * config/arm/arm.c (neon_builtin_data): Move contents to
11630         arm_neon_builtins.def.
11631         (enum arm_builtins): Include neon builtin definitions.
11632         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
11633         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
11635 2013-04-04  Marek Polacek  <polacek@redhat.com>
11637         PR tree-optimization/48186
11638         * predict.c (maybe_hot_frequency_p): Return false if
11639         HOT_BB_FREQUENCY_FRACTION is 0.
11640         (cgraph_maybe_hot_edge_p): Likewise.
11642 2013-04-04  Richard Biener  <rguenther@suse.de>
11644         PR tree-optimization/56826
11645         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11646         more accurately.
11648 2013-04-04  Richard Biener  <rguenther@suse.de>
11650         PR tree-optimization/56213
11651         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
11652         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
11654 2013-04-04  Richard Biener  <rguenther@suse.de>
11656         PR tree-optimization/56837
11657         * tree-loop-distribution.c (classify_partition): For non-zero
11658         values require that the value has the same precision as its
11659         mode to be useful as memset value.
11661 2013-04-03  Nick Clifton  <nickc@redhat.com>
11663         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
11664         (fmssf4): Use fmsf.s on E3V5 architectures.
11665         (fnmasf4): Use fnmaf.s on E3V5 architectures.
11666         (fnmssf4): Use fnmsf.s on E3V5 architectures.
11668 2013-04-03  Jeff Law  <law@redhat.com>
11670         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11671         (lra-eliminations.o): Likewise.
11673 2013-04-03  Teresa Johnson  <tejohnson@google.com>
11675         * gcov-io.c (compute_working_sets): Moved most of body of old
11676         compute_working_sets here from profile.c.
11677         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
11678         (gcov_working_set_t): Moved typedef here from basic-block.h
11679         (compute_working_set): Declare.
11680         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11681         (get_working_sets): Renamed from compute_working_set,
11682         replace most of body with call to new compute_working_sets.
11683         (get_exec_counts): Replace call to compute_working_sets
11684         to get_working_sets.
11685         * profile.h (get_working_sets): Renamed from compute_working_set.
11686         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
11687         to get_working_sets.
11688         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11689         * gcov-dump.c (dump_working_sets): New function.
11691 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
11693         * hwint.c (sext_hwi, zext_hwi): New functions.
11694         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11695         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11696         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11697         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11698         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11699         (sext_hwi, zext_hwi): New functions.
11701 2013-04-03  Jeff Law  <law@redhat.com>
11703         PR tree-optimization/56799
11704         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11705         back test for widening conversion erroneously dropped in prior change.
11707 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11709         PR target/56809
11710         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11711         instead of next_real_insn.
11713 2013-04-03  Marek Polacek  <polacek@redhat.com>
11715         PR sanitizer/55702
11716         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
11718 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11720         PR target/56809
11721         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11722         next_real_insn.
11723         (thumb1_output_casesi): Likewise.
11724         (thumb2_output_casesi): Likewise.
11726 2013-04-03  Richard Biener  <rguenther@suse.de>
11728         PR tree-optimization/56817
11729         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11730         Split out ...
11731         (tree_unroll_loops_completely_1): ... new function to manually
11732         walk the loop tree, properly defering outer loops of unrolled
11733         loops to later iterations.
11735 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11737         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11738         (vectorizable_load): Likewise.
11739         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11740         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11742 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11744         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11745         BIT_FIELD_REF.
11747 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
11749         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11751 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
11753         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11755 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11757         PR tree-optimization/56790
11758         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11759         folding.
11761 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
11763         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11764         Handle VEC_MERGE.
11765         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11766         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11767         equal arguments.
11769 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
11771         PR c/19449
11772         * tree.h (force_folding_builtin_constant_p): New decl.
11773         * builtins.c (force_folding_builtin_constant_p): New variable.
11774         (fold_builtin_constant_p): Fold immediately also if
11775         force_folding_builtin_constant_p.
11777 2013-04-03  Richard Biener  <rguenther@suse.de>
11779         PR tree-optimization/56812
11780         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11781         DRs of the same interleaving chain are independent.
11783 2013-04-02  Jason Merrill  <jason@redhat.com>
11785         * gdbinit.in (pbb): Use debug fn.
11787 2013-04-02  Lawrence Crowl  <crowl@google.com>
11789         * sese.h (struct ivtype_map_elt_s): Remove unused.
11790         (extern debug_ivtype_map): Remove unused.
11791         (extern eq_ivtype_map_elts): Remove unused.
11792         * sese.c (debug_ivtype_map): Removed unused.
11793         (debug_ivtype_map_1): Removed unused.
11794         (debug_ivtype_elt): Remove unused.
11795         (eq_ivtype_map_elts): Remove unused.
11798 2013-04-02  Kai Tietz  <ktietz@redhat.com>
11800         PR target/52790
11801         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11802         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
11803         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11804         function.
11805         (legitimize_pe_coff_symbol): Likewise.
11806         (is_imported_p): New helper-function.
11807         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11808         for Windows x64 targets.
11809         (ix86_expand_prologue): Optimize for pe-coff targets.
11810         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11811         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11812         medium/large code-model.
11813         (legitimize_pic_address): Likewise.
11814         (legitimize_tls_address): Likewise.
11815         (ix86_expand_call): Likewise.
11816         (x86_output_mi_thunk): Likewise.
11817         (get_dllimport_decl): Add new beimport argument.
11818         (construct_plt_address): Don't assert for x64 pe-coff targets.
11819         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11820         targets.
11821         (SYMBOL_FLAG_STUBVAR): New macro.
11822         (SYMBOL_REF_STUBVAR_P): Likewise.
11823         * config/i386/winnt.c (stub_list): New structure.
11824         (stub_head): New local variable.
11825         (i386_pe_record_stub): New function.
11826         (i386_pe_file_end): Emit refptr-stubs.
11828 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
11830         PR rtl-optimization/56745
11831         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11832         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11834         PR c++/34949
11835         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11836         and both of them are MEM_REFs, just compare first argument for
11837         equality and attempt to deal even with differing offsets.
11839         PR c++/34949
11840         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11841         of gimple_clobber_p to be MEM_REF.
11842         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11843         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
11844         after gimplification.
11845         * asan.c (get_mem_ref_of_assignment): Don't instrument
11846         gimple_clobber_p stmts.
11847         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11848         gimple_clobber_p stmt if they have MEM_REF lhs and
11849         are dead because of another gimple_clobber_p stmt.
11850         * tree-ssa-live.c (clear_unused_block_pointer): Treat
11851         gimple_clobber_p stmts like debug stmts.
11852         (remove_unused_locals): Remove clobbers with MEM_REF lhs
11853         that refer to unused VAR_DECLs or uninitialized values.
11854         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
11855         gimple_clobber_p stmts if they refer to removed parameters.
11856         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
11857         formatting.
11859 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
11861         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
11862         using SWI48 mode attribute.
11864 2013-04-02  Wei Mi  <wmi@google.com>
11866         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
11867         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
11868         *<rotate_insn><mode>3_mask in i386.md.
11870 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11872         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
11874 2013-04-02  Richard Biener  <rguenther@suse.de>
11876         PR tree-optimization/56778
11877         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11878         Runtime alias tests are not supported for gather loads.
11879         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
11880         stmts referenced from SSA operands before updating SSA form.
11882 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
11883             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11885         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
11886         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
11887         * config/arm/cortex-a53.md: New file.
11888         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
11889         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
11890         * config/arm/arm.c (arm_issue_rate): Likewise.
11891         * config/arm/arm-tune.md: Regenerate
11892         * config/arm/arm-tables.opt: Regenerate.
11893         * config/arm/arm-cores.def: Add cortex-a53.
11895 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
11897         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
11898         non-static link.
11900 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11902         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
11903         scalar load/store operations using B/H registers.
11904         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
11906 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
11908         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
11909         scalar move.
11910         * config/aarch64/aarch64.c
11911         (aarch64_simd_scalar_immediate_valid_for_move): New.
11912         * config/aarch64/aarch64-protos.h
11913         (aarch64_simd_scalar_immediate_valid_for_move): New.
11914         * config/aarch64/constraints.md (Dh, Dq): New.
11915         * config/aarch64/iterators.md (hq): New.
11917 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
11919         * reorg.c (get_branch_condition): Deal with conditional returns.
11920         (fill_simple_delay_slots): Remove dead code dealing with jumps.
11922 2013-04-01  Wei Mi  <wmi@google.com>
11924         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
11925         Truncate operand 2 using %b asm operand modifier.
11926         (*<shift_insn><mode>3_mask): Ditto.
11927         (*<rotate_insn><mode>3_mask): Ditto.
11929 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
11931         PR middle-end/56798
11932         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
11934 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
11936         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
11937         of next_real_insn.
11938         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
11940 2013-03-30  Lawrence Crowl  <crowl@google.com>
11942         * dse.c (clear_alias_sets): Remove never set.
11943         (disqualified_clear_alias_sets): Remove never set.
11944         (clear_alias_mode_pool): Remove never set.
11945         (dse_step0): Remove condition that is never true.
11946         (canon_address): Remove condition that is never true.
11947         (dse_step7): Remove condition that is never true.
11948         (rest_of_handle_dse): Remove condition that is never true.
11949         (rest_of_handle_dse::did_global): Remove never read from above.
11950         (dse_step2_spill): Remove never called from above.
11951         (dse_step5_spill): Remove never called from above.
11953 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
11955         * doc/md.texi (Standard Names) <casesi>: Update documentation for
11956         JUMP_TABLE_DATA changes.
11957         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
11958         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
11959         (Insns) <jump_table_data>: New entry.
11960         * doc/tm.texi: Regenerate.
11962         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
11964         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
11965         for table jump at the end of a basic block using tablejump_p.
11966         * targhooks.c (default_invalid_within_doloop): Likewise.
11967         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
11968         target hook implementation that is identical to the default hook.
11969         (rs6000_invalid_within_doloop): Remove.
11971         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
11972         unused variable from tablejump_p call.
11974         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
11975         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
11976         (INSN_DELETED_P): Likewise.
11977         (emit_jump_table_data): New prototype.
11978         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
11979         after 4th as unused.
11980         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
11981         * sched-vis.c (print_insn): Likewise.
11982         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
11983         insn for compatibility with back ends that use next_active_insn to
11984         identify jump table data.
11985         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
11986         (remove_insn): Likewise.
11987         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
11988         to be emitted.
11989         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
11990         (emit_jump_table_data): New function.
11992         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
11993         basic block, a JUMP_TABLE_DATA never is.
11994         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
11995         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
11996         off from code handling real insns.
11997         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
11998         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
11999         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12000         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12001         is not a NONDEBUG_INSN_P.
12002         * ira-costs.c (scan_one_insn): Likewise.
12003         * jump.c (mark_all_labels): Likewise.
12004         (mark_jump_label_1): Likewise.
12005         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12006         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12007         a basic block.
12008         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12009         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12010         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12011         JUMP_TABLE_DATA_P insns.
12012         (calculate_elim_costs_all_insns): Likewise.
12013         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12014         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12015         (delete_output_reload): Code style fixups.
12016         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12017         insn flags on this non-insn.
12018         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12019         as scheduling barriers, for pre-change compatibility.
12020         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12021         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12023         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12024         redundant JUMP_TABLE_DATA_P test.
12025         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12026         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12027         (frv_for_each_packet): Likewise.
12028         * config/i386/i386.c (min_insn_size): Likewise.
12029         (ix86_avoid_jump_mispredicts): Likewise.
12030         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12031         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12032         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12033         (mips16_insn_length): Robustify.
12034         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12035         (mips16_split_long_branches): Likewise.
12036         * config/pa/pa.c (pa_combine_instructions): Likewise.
12037         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12038         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12039         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12040         as contributing to pool range lengths.
12041         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12042         Remove redundant JUMP_TABLE_DATA_P test.
12043         (sh_loop_align): Likewise.
12044         (split_branches): Likewise.
12045         (sh_insn_length_adjustment): Likewise.
12046         * config/spu/spu.c (get_branch_target): Likewise.
12048 2013-03-29  Jan Hubicka  <jh@suse.cz>
12050         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12051         gcov streaming; stream hot bb threshold to ltrans.
12052         * predict.c (get_hot_bb_threshold): Break out from ....
12053         (maybe_hot_count_p): ... here.
12054         (set_hot_bb_threshold): New function.
12055         * lto-section-in.c (lto_section_name): Add profile.
12056         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12057         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12058         and data-streamer.h
12059         (histogram_entry): New structure.
12060         (histogram, histogram_pool): New global vars.
12061         (histogram_hash): New structure.
12062         (histogram_hash::hash): New method.
12063         (histogram_hash::equal): Likewise.
12064         (account_time_size): New function.
12065         (cmp_counts): New function.
12066         (dump_histogram): New function.
12067         (ipa_profile_generate_summary): New function.
12068         (ipa_profile_write_summary): New function.
12069         (ipa_profile_read_summary): New function.
12070         (ipa_profile): Decide on threshold.
12071         (pass_ipa_profile): Add ipa_profile_write_summary and
12072         ipa_profile_read_summary.
12073         * Makefile.in (ipa.o): Update dependencies.
12074         * lto-streamer.h (LTO_section_ipa_profile): New section.
12076 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12078         * tree.h (VAR_P): New.
12080 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12082         PR lto/56777
12083         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12085 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12087         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12088         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12089         (control_flow_insn_p): Likewise.
12090         * cfgrtl.c (duplicate_insn_chain): Likewise.
12091         * final.c (get_attr_length_1): Likewise.
12092         (shorten_branches): Likewise.
12093         (final_scan_insn): Likewise.
12094         * function.c (instantiate_virtual_regs): Likewise.
12095         * gcse.c (insert_insn_end_basic_block): Likewise.
12096         * ira-costs.c (scan_one_insn): Likewise.
12097         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12098         * lra.c (check_rtl): Likewise.
12099         * reload1.c (elimination_costs_in_insn): Likewise.
12100         * reorg.c (follow_jumps): Likewise.
12102         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12103         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12104         (thumb_far_jump_used_p): Likewise.
12105         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12106         (workaround_speculation): Likewise.
12107         (add_sched_insns_for_speculation): Likewise.
12108         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12109         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12110         (frv_for_each_packet): Likewise.
12111         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12112         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12113         (final_emit_insn_group_barriers): Likewise.
12114         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12115         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12116         (mips16_insn_length): Likewise.
12117         * config/pa/pa.c (pa_reorg): Likewise.
12118         (pa_combine_instructions): Likewise.
12119         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12120         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12121         (sh_reorg): Likewise.
12122         (split_branches): Likewise.
12123         * config/spu/spu.c (get_branch_target): Likewise.
12125         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12126         JUMP_TABLE_DATA_P.
12128 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
12130         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12131         Fix declaration name.
12133 2013-03-28  Lawrence Crowl  <crowl@google.com>
12135         * graphds.h (struct graph.indicies): Remove unused.
12136         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12137         (SCOP_ORIGINAL_PDDRS): Remove unused.
12138         * sese.h (extern insert_loop_close_phis): Removed unused.
12139         (extern insert_guard_phis): Removed unused.
12140         (extern ivtype_map_elt_info): Removed unused.
12141         (new_ivtype_map_elt): Removed unused.
12142         * sese.c (ivtype_map_elt_info): Removed unused.
12144 2013-03-28  Lawrence Crowl  <crowl@google.com>
12146         * Makefile.in: Add several missing include dependences.
12147         (DUMPFILE_H): New.
12148         (test-dump.o): New.  This object is not added to any executable,
12149         but is present for ad-hoc testing.
12150         * bitmap.c
12151         (debug (const bitmap_head_def &)): New.
12152         (debug (const bitmap_head_def *)): New.
12153         * bitmap.h
12154         (extern debug (const bitmap_head_def &)): New.
12155         (extern debug (const bitmap_head_def *)): New.
12156         * cfg.c
12157         (debug (edge_def &)): New.
12158         (debug (edge_def *)): New.
12159         * cfghooks.c
12160         (debug (basic_block_def &)): New.
12161         (debug (basic_block_def *)): New.
12162         * dumpfile.h
12163         (dump_node (const_tree, int, FILE *)): Correct source file.
12164         * dwarf2out.c
12165         (debug (die_struct &)): New.
12166         (debug (die_struct *)): New.
12167         * dwarf2out.h
12168         (extern debug (die_struct &)): New.
12169         (extern debug (die_struct *)): New.
12170         * gimple-pretty-print.c
12171         (debug (gimple_statement_d &)): New.
12172         (debug (gimple_statement_d *)): New.
12173         * gimple-pretty-print.h
12174         (extern debug (gimple_statement_d &)): New.
12175         (extern debug (gimple_statement_d *)): New.
12176         * ira-build.c
12177         (debug (ira_allocno_copy &)): New.
12178         (debug (ira_allocno_copy *)): New.
12179         (debug (ira_allocno &)): New.
12180         (debug (ira_allocno *)): New.
12181         * ira-int.h
12182         (extern debug (ira_allocno_copy &)): New.
12183         (extern debug (ira_allocno_copy *)): New.
12184         (extern debug (ira_allocno &)): New.
12185         (extern debug (ira_allocno *)): New.
12186         * ira-lives.c
12187         (debug (live_range &)): New.
12188         (debug (live_range *)): New.
12189         * lra-int.h
12190         (debug (lra_live_range &)): New.
12191         (debug (lra_live_range *)): New.
12192         * lra-lives.c
12193         (debug (lra_live_range &)): New.
12194         (debug (lra_live_range *)): New.
12195         * omega.c
12196         (debug (omega_pb_d &)): New.
12197         (debug (omega_pb_d *)): New.
12198         * omega.h
12199         (extern debug (omega_pb_d &)): New.
12200         (extern debug (omega_pb_d *)): New.
12201         * print-rtl.c
12202         (debug (const rtx_def &)): New.
12203         (debug (const rtx_def *)): New.
12204         * print-tree.c
12205         (debug_tree (tree): Move within file.
12206         (debug_raw (const tree_node &)): New.
12207         (debug_raw (const tree_node *)): New.
12208         (dump_tree_via_hooks (const tree_node *, int)): New.
12209         (debug (const tree_node &)): New.
12210         (debug (const tree_node *)): New.
12211         (debug_verbose (const tree_node &)): New.
12212         (debug_verbose (const tree_node *)): New.
12213         (debug_head (const tree_node &)): New.
12214         (debug_head (const tree_node *)): New.
12215         (debug_body (const tree_node &)): New.
12216         (debug_body (const tree_node *)): New.
12217         (debug_vec_tree (tree): Move and reimplement in terms of dump.
12218         (debug (vec<tree, va_gc> &)): New.
12219         (debug (vec<tree, va_gc> *)): New.
12220         * rtl.h
12221         (extern debug (const rtx_def &)): New.
12222         (extern debug (const rtx_def *)): New.
12223         * sbitmap.c
12224         (debug_raw (simple_bitmap_def &)): New.
12225         (debug_raw (simple_bitmap_def *)): New.
12226         (debug (simple_bitmap_def &)): New.
12227         (debug (simple_bitmap_def *)): New.
12228         * sbitmap.h
12229         (extern debug (simple_bitmap_def &)): New.
12230         (extern debug (simple_bitmap_def *)): New.
12231         (extern debug_raw (simple_bitmap_def &)): New.
12232         (extern debug_raw (simple_bitmap_def *)): New.
12233         * sel-sched-dump.c
12234         (debug (vinsn_def &)): New.
12235         (debug (vinsn_def *)): New.
12236         (debug_verbose (vinsn_def &)): New.
12237         (debug_verbose (vinsn_def *)): New.
12238         (debug (expr_def &)): New.
12239         (debug (expr_def *)): New.
12240         (debug_verbose (expr_def &)): New.
12241         (debug_verbose (expr_def *)): New.
12242         (debug (vec<rtx> &)): New.
12243         (debug (vec<rtx> *)): New.
12244         * sel-sched-dump.h
12245         (extern debug (vinsn_def &)): New.
12246         (extern debug (vinsn_def *)): New.
12247         (extern debug_verbose (vinsn_def &)): New.
12248         (extern debug_verbose (vinsn_def *)): New.
12249         (extern debug (expr_def &)): New.
12250         (extern debug (expr_def *)): New.
12251         (extern debug_verbose (expr_def &)): New.
12252         (extern debug_verbose (expr_def *)): New.
12253         (extern debug (vec<rtx> &)): New.
12254         (extern debug (vec<rtx> *)): New.
12255         * sel-sched-ir.h
12256         (_list_iter_cond_expr): Make inline instead of static.
12257         * sreal.c
12258         (debug (sreal &)): New.
12259         (debug (sreal *)): New.
12260         * sreal.h
12261         (extern debug (sreal &)): New.
12262         (extern debug (sreal *)): New.
12263         * tree.h
12264         (extern debug_raw (const tree_node &)): New.
12265         (extern debug_raw (const tree_node *)): New.
12266         (extern debug (const tree_node &)): New.
12267         (extern debug (const tree_node *)): New.
12268         (extern debug_verbose (const tree_node &)): New.
12269         (extern debug_verbose (const tree_node *)): New.
12270         (extern debug_head (const tree_node &)): New.
12271         (extern debug_head (const tree_node *)): New.
12272         (extern debug_body (const tree_node &)): New.
12273         (extern debug_body (const tree_node *)): New.
12274         (extern debug (vec<tree, va_gc> &)): New.
12275         (extern debug (vec<tree, va_gc> *)): New.
12276         * tree-cfg.c
12277         (debug (struct loop &)): New.
12278         (debug (struct loop *)): New.
12279         (debug_verbose (struct loop &)): New.
12280         (debug_verbose (struct loop *)): New.
12281         * tree-dump.c: Add header dependence.
12282         * tree-flow.h
12283         (extern debug (struct loop &)): New.
12284         (extern debug (struct loop *)): New.
12285         (extern debug_verbose (struct loop &)): New.
12286         (extern debug_verbose (struct loop *)): New.
12287         * tree-data-ref.c
12288         (debug (data_reference &)): New.
12289         (debug (data_reference *)): New.
12290         (debug (vec<data_reference_p> &)): New.
12291         (debug (vec<data_reference_p> *)): New.
12292         (debug (vec<ddr_p> &)): New.
12293         (debug (vec<ddr_p> *)): New.
12294         * tree-data-ref.h
12295         (extern debug (data_reference &)): New.
12296         (extern debug (data_reference *)): New.
12297         (extern debug (vec<data_reference_p> &)): New.
12298         (extern debug (vec<data_reference_p> *)): New.
12299         (extern debug (vec<ddr_p> &)): New.
12300         (extern debug (vec<ddr_p> *)): New.
12301         * tree-ssa-alias.c
12302         (debug (pt_solution &)): New.
12303         (debug (pt_solution *)): New.
12304         * tree-ssa-alias.h
12305         (extern debug (pt_solution &)): New.
12306         (extern debug (pt_solution *)): New.
12307         * tree-ssa-alias.c
12308         (debug (_var_map &)): New.
12309         (debug (_var_map *)): New.
12310         (debug (tree_live_info_d &)): New.
12311         (debug (tree_live_info_d *)): New.
12312         * tree-ssa-alias.h
12313         (extern debug (_var_map &)): New.
12314         (extern debug (_var_map *)): New.
12315         (extern debug (tree_live_info_d &)): New.
12316         (extern debug (tree_live_info_d *)): New.
12318 2013-03-28  Jan Hubicka  <jh@suse.cz>
12320         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12322 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
12324         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12325         record only when desired or required.
12327 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
12329         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12330         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
12332 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12334         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12335         (*andsi3_compare0_uxtw): New pattern.
12336         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12337         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12339 2013-03-28  Jan Hubicka  <jh@suse.cz>
12341         * data-streamer-in.c (streamer_read_gcov_count): New function.
12342         * gimple-streamer-out.c: Include value-prof.h.
12343         (output_gimple_stmt): Output histogram.
12344         (output_bb): Use streamer_write_gcov_count.
12345         * value-prof.c: Include data-streamer.h
12346         (dump_histogram_value): Add HIST_TYPE_MAX.
12347         (stream_out_histogram_value): New function.
12348         (stream_in_histogram_value): New function.
12349         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12350         (stream_out_histogram_value, stream_in_histogram_value): Declare.
12351         * data-streamer-out.c (streamer_write_gcov_count): New function.
12352         (streamer_write_gcov_count_stream): New function.
12353         * lto-cgraph.c (lto_output_edge): Update counter streaming.
12354         (lto_output_node): Likewise.
12355         (input_node, input_edge): Likewise.
12356         * lto-streamer-out.c (output_cfg): Update streaming.
12357         * lto-streamer-in.c (input_cfg): Likewise.
12358         * data-streamer.h (streamer_write_gcov_count,
12359         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12360         * gimple-streamer-in.c: Include value-prof.h
12361         (input_gimple_stmt): Input histograms.
12362         (input_bb): Update profile streaming.
12364 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12366         * genmodes.c (emit_max_int): New function.
12367         (emit_insn_modes_h): Added call to emit_max_function.
12368         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12369         Added doc.
12370         * machmode.def: Fixed comment.
12372 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
12374         * combine.c (try_combine): Removed useless assert.
12375         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12377 2013-03-28  Marek Polacek  <polacek@redhat.com>
12378             Richard Biener  <rguenther@suse.de>
12380         PR tree-optimization/56695
12381         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12382         build signed result of a vector comparison.
12383         * tree-cfg.c (verify_gimple_comparison): Check that a result
12384         of a vector comparison has signed type.
12386 2013-03-28  Richard Biener  <rguenther@suse.de>
12388         PR tree-optimization/37021
12389         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12390         do not restrict gaps between groups.
12391         * tree-vect-stmts.c (vectorizable_load): Properly account for
12392         a gap between groups.
12394 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
12396         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12397         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12398         is not enabled.
12400 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
12402         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12403         * doc/extend.texi (Named Address Spaces): Ditto.
12404         (Variable Attributes): Ditto.
12406 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12408         * config.build: Add support for cygwin x64 target.
12409         * config.gcc: Likewise.
12410         * config.host: Likewise.
12411         * configure.ac: Likewise
12412         * configure: Regenerated.
12414 2013-03-27  Kai Tietz  <ktietz@redhat.com>
12416         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12417         * config/i386/t-cygwin-w64: New file.
12418         * config/i386/cygwin-w64.h: New file.
12419         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12420         and add support for x64-cygwin target.
12421         (CPP_SPEC): Likewise.
12422         (CXX_WRAP_SPEC_LIST): Undefine before define.
12423         (LIBGCJ_SONAME): Use 15 as version.
12425 2013-03-27  Richard Biener  <rguenther@suse.de>
12427         PR tree-optimization/56716
12428         * tree-ssa-structalias.c (perform_var_substitution): Adjust
12429         dumping for ref nodes.
12431 2013-03-27  Martin Jambor  <mjambor@suse.cz>
12433         PR tree-optimization/55334
12434         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12435         restricted pointers to arrays.
12437 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12439         * Makefile.in (.SUFFIXES): Add .cc.
12440         (.c.o): Apply same recipe for implicit rule .cc.o.
12442 2013-03-27  Richard Biener  <rguenther@suse.de>
12444         PR tree-optimization/37021
12445         * tree-vect-data-refs.c (vect_check_strided_load): Allow
12446         REALPART/IMAGPART_EXPRs around the supported refs.
12447         * tree-ssa-structalias.c (find_func_aliases): Assume that
12448         floating-point values are not used to transfer pointers.
12450 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12452         * target.def (TARGET_HAS_IFUNC_P): New target hook.
12453         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12454         * doc/tm.texi: Regenerate.
12455         * targhooks.h (default_has_ifunc_p): New.
12456         * targhooks.c (default_has_ifunc_p): Ditto.
12457         * config/linux-protos.h: New file.
12458         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12459         this hook for linux which disables support of indirect functions in
12460         android.
12461         * config/linux-android.c: New file.
12462         * config/t-linux-android.c: Ditto.
12463         * config.gcc: Added new object file linux-android.o.
12464         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
12465         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12466         * varasm.c (do_assemble_alias): Likewise.
12467         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12468         doesn't support indirect functions.
12469         * configure: Regenerate.
12471 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
12473         PR target/56102
12474         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12475         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12476         mult-word mode.
12478 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12480         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12482 2013-03-27  Terry Guo  <terry.guo@arm.com>
12484         * config/arm/arm-cores.def: Added core cortex-r7.
12485         * config/arm/arm-tune.md: Regenerated.
12486         * config/arm/arm-tables.opt: Regenerated.
12487         * doc/invoke.texi: Added entry for core cortex-r7.
12489 2013-03-27  Walter Lee  <walt@tilera.com>
12491         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12492         double-decrement of next_scratch_regno.
12494 2013-03-27  Walter Lee  <walt@tilera.com>
12496         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12497         input operands.
12498         (insn_v1mulus): Ditto.
12499         (insn_v2muls): Ditto.
12501 2013-03-27  Walter Lee  <walt@tilera.com>
12503         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12504         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12506 2013-03-27  Walter Lee  <walt@tilera.com>
12508         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12509         (*sibcall_value): Ditto.
12511 2013-03-27  Walter Lee  <walt@tilera.com>
12513         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12514         (insn_mnz_v8qi): ... this ...
12515         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
12516         vector equivalent.
12517         (insn_v<n>mnz): Replaced by ...
12518         (insn_v1mnz): ... this ...
12519         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
12520         equivalent.
12521         (insn_mz_<mode>): Replaced by ...
12522         (insn_mz_v8qi): ... this ...
12523         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
12524         vector equivalent.
12525         (insn_v<n>mz): Replaced by ...
12526         (insn_v1mz): ... this ...
12527         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
12528         equivalent.
12530 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
12532         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12534 2013-03-26  Roland McGrath  <mcgrathr@google.com>
12536         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12537         than fprintf with a non-constant, non-format string.
12539 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12541         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12542         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
12543         operand 0 predicate.
12544         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12545         attribute.  Use general_x64nomem_operand as operand 1 predicate.
12546         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12547         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12548         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12549         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
12550         (mov<mode>_insv_1): Remove expander.  Merge insn with
12551         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12552         Use general_x64nomem_operand as operand 1 predicate.
12553         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12554         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12555         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12556         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12557         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
12558         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12559         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
12560         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12561         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
12562         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12563         (general_x64nomem_operand): Ditto.
12565 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12567         * config/rtems.opt: Add -pthread option.
12569 2013-03-26  Richard Biener  <rguenther@suse.de>
12571         * alias.c (find_base_term): Avoid redundant and not used recursion.
12572         (base_alias_check): Get the initial base term from the caller.
12573         (true_dependence_1): Compute and pass base terms to base_alias_check.
12574         (write_dependence_p): Likewise.
12575         (may_alias_p): Likewise.
12577 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
12579         * config/aarch64/aarch64.c (aarch64_classify_address): Support
12580         PC-relative load in SI modes and above only.
12582 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
12584         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12585         * config/arm/iwmmxt.md (WCGR0): Update.
12586         (WCGR1, WCGR2, WCGR3): Likewise.
12588 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
12590         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12591         Use x64 and nox64 isa attributes.
12593 2013-03-26  Richard Biener  <rguenther@suse.de>
12595         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12596         alignment computations and rely on get_object_alignment_1
12597         for the !TYPE_P case.
12598         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12600 2013-03-26  Walter Lee  <walt@tilera.com>
12602         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12603         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12605 2013-03-25  Jeff Law  <law@redhat.com>
12607         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12608         check for INTEGRAL_TYPE_P that was missing due to checking in
12609         wrong version of prior patch.
12611 2013-03-25  Walter Lee  <walt@tilera.com>
12613         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12614         TILEGX_INSN_SHUFFLEBYTES1.
12615         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12616         shufflebytes1.
12617         (tilegx_builtins): Ditto.
12618         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12620 2013-03-25  Walter Lee  <walt@tilera.com>
12622         * config/tilegx/tilegx.md (floatsisf2): New pattern.
12623         (floatunssisf2): New pattern.
12624         (floatsidf2): New pattern.
12625         (floatunssidf2): New pattern.
12627 2013-03-25  Walter Lee  <walt@tilera.com>
12629         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12630         tests for constraint J, K, N, P.
12632 2013-03-25  Walter Lee  <walt@tilera.com>
12634         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12635         Use indirect/pcrel encoding.
12636         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12637         Ditto.
12639 2013-03-25  Steve Ellcey  <sellcey@mips.com>
12641         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12642         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12643         * config/mips/mips.c (mips_option_override): Set IMADD default.
12644         * config/mips/mips.h (PTF_AVOID_IMADD): New.
12645         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12646         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12647         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12648         * doc/invoke.texi (-mimadd/-mno-imadd): New.
12650 2013-03-25  Jeff Law  <law@redhat.com>
12652         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12653         slightly to avoid creating and folding useless trees.  Simplify
12654         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12656 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
12658         * config/i386/i386.md (*zero_extendsidi2): Merge with
12659         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
12660         * config/i386/predicates.md (x86_64_zext_operand): Rename from
12661         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
12662         targets.  Clarify comment.
12664 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12666         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12667         pass-through jump functions differently.
12668         (ipa_read_jump_function): Likewise.  Also use setter functions to set
12669         up jump functions.
12671 2013-03-25  Martin Jambor  <mjambor@suse.cz>
12673         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12674         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12675         process it.
12676         (ipa_get_indirect_edge_target): New function.
12677         (devirtualization_time_bonus): New parameter known_aggs, pass it to
12678         ipa_get_indirect_edge_target.  Update all callers.
12679         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
12680         ipa_get_indirect_edge_target_1 instead of calling
12681         ipa_get_indirect_edge_target.
12682         (create_specialized_node): Pass aggvlas to
12683         ipcp_discover_new_direct_edges.
12685 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12687         * config/arm/arm.md (f_sels, f_seld): New types.
12688         (*cmov<mode>): New pattern.
12689         * config/arm/predicates.md (arm_vsel_comparison_operator): New
12690         predicate.
12692 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12694         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12695         POSIX-printf for mingw-hosted  builds.
12697 2013-03-25  Richard Biener  <rguenther@suse.de>
12699         PR middle-end/56694
12700         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12701         must-not-throw stmt location.
12703 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12705         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12706         Emit load-acquire versions when acq is true.
12707         (arm_emit_store_exclusive): Add rel parameter.
12708         Emit store-release versions when rel is true.
12709         (arm_split_compare_and_swap): Use acquire-release instructions
12710         instead.
12711         of barriers when appropriate.
12712         (arm_split_atomic_op): Likewise.
12713         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12714         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12715         (VUNSPEC_SLX): Likewise.
12716         (VUNSPEC_LDA): Likewise.
12717         (VUNSPEC_STL): Likewise.
12718         * config/arm/sync.md (atomic_load<mode>): New pattern.
12719         (atomic_store<mode>): Likewise.
12720         (arm_load_acquire_exclusive<mode>): Likewise.
12721         (arm_load_acquire_exclusivesi): Likewise.
12722         (arm_load_acquire_exclusivedi): Likewise.
12723         (arm_store_release_exclusive<mode>): Likewise.
12725 2013-03-25  Catherine Moore  <clm@codesourcery.com>
12727         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12728         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12729         * config/mip/predicates.md (lwsp_swsp_operand,
12730         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12731         sb16_operand, db4_operand, db7_operand, ib3_operand,
12732         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12733         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12734         andi16_operand): New predicates.
12735         * config/mips/mips.md (compression): New attribute.
12736         (enabled): New attribute.
12737         (length): Consider compression in computing length.
12738         (shift_compression): New code attribute.
12739         (*add<mode>3): New operands. Record compression.
12740         (sub<mode>3): Likewise.
12741         (one_cmpl<mode>2): Likewise.
12742         (*and<mode>3): Likewise.
12743         (*ior<mode>3): Likewise.
12744         (unnamed pattern for xor): Likewise.
12745         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12746         (*<optab><mode>3): Likewise.
12747         (*mov<mode>_internal: Likewise.
12748         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12749         (mips_unsigned_immediate_p): New.
12750         (umips_lwsp_swsp_address_p): New.
12751         (m16_based_address_p): New.
12752         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12753         (mips_unsigned_immediate_p): New prototype.
12754         (lwsp_swsp_address_p): New prototype.
12755         (m16_based_address_p): New prototype.
12756         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12757         (mips_signed_immediate_p): New function.
12758         (m16_based_address_p): New function.
12759         (lwsp_swsp_address_p): New function.
12760         (mips_print_operand_punctuation): Recognize short delay slot insns
12761         for microMIPS.add<mode>3"
12763 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12765         PR target/56720
12766         * config/arm/iterators.md (v_cmp_result): New mode attribute.
12767         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12769 2013-03-25  Richard Biener  <rguenther@suse.de>
12771         PR tree-optimization/56689
12772         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12773         any edge.
12775 2013-03-25  Richard Biener  <rguenther@suse.de>
12777         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12778         of bitmap.
12779         (memory_references): Likewise.
12780         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12781         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12782         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12783         (gather_mem_refs_in_loops): Fold into ...
12784         (analyze_memory_references): ... this.  Move initialization
12785         to tree_ssa_lim_initialize.
12786         (fill_always_executed_in): Rename to ...
12787         (fill_always_executed_in_1): ... this.
12788         (fill_always_executed_in): Move contains_call computation to
12789         this new function from ...
12790         (tree_ssa_lim_initialize): ... here.
12791         (tree_ssa_lim): Call fill_always_executed_in.
12793 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12795         * postreload.c (reload_combine): Fix code detecting returns.
12797 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
12799         * function.c (emit_use_return_register_into_block): On cc0 targets,
12800         do not emit the sequence between cc0 setter and user.
12802 2013-03-25  Kai Tietz  <ktietz@redhat.com>
12804         * config/i386/predicates.md (local_symbolic_operand): Interpret
12805         dll-imported symbols as none-local.
12807 2013-03-25  Richard Biener  <rguenther@suse.de>
12809         * tree-ssa-loop-im.c (struct depend): Remove.
12810         (struct lim_aux_data): Make depends a vec of gimples.
12811         (free_lim_aux_data): Adjust.
12812         (add_dependency): Likewise.
12813         (set_level): Likewise.
12815 2013-03-25  Richard Biener  <rguenther@suse.de>
12817         PR middle-end/56434
12818         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12819         the pointer returned by calls with ECF_MALLOC set.
12821 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12823         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12825 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
12827         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
12828         using MMXMODE mode iterator.
12829         (*move<mode>_internal): Merge with *movv2sf_internal and
12830         *movv2sf_internal_rex64 using MMXMODE mode iterator.
12832 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12834         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12835         (record_last_mem_set_info): Likewise.
12837         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12838         of XNEWVEC followed by memset.
12839         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12841 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
12843         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12844         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12845         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12846         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12847         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12848         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
12849         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
12850         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
12851         BARRIER_P instead of GET_CODE.
12853 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12855         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
12856         inaccuracy in the probing code.
12858         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
12859         (ctrapdi4): Likewise.
12861 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12863         * calls.c (expand_call): Add missing guard to code handling return
12864         of non-BLKmode structures in MSB.
12865         * function.c (expand_function_end): Likewise.
12867 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
12869         * combine.c (try_combine): Adjust comment.  Do not add the set of
12870         insn #0 if the destination indirectly is set or dies in insn #2.
12871         Tidy up code to distribute a new note.
12873 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12875         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
12876         also for alternatives 16 and 17.
12878 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
12880         * config/i386/sse.md (*mov<mode>_internal): Merge with
12881         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
12882         Emit insn template depending on type attribute.  Use
12883         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
12884         movd instead of movq mnemonic for interunit moves.  Rewrite mode
12885         attribute calculation.  Remove unit attribute calculation.
12886         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
12887         Set prefix_data16 attribute for DImode ssemov types.
12888         Use Ym instead of y for SSE-MMX conversion alternatives.
12889         Reorder operand constraints.
12891 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
12893         * df.h (df_insn_delete): Adjust prototype.
12894         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
12895         and let it decide whether mark the basic block dirty.
12896         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
12897         * df-scan.c (df_insn_info_delete): New helper function, split
12898         off from df_insn_delete.
12899         (df_scan_free_bb_info): Use it.
12900         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
12901         Likewise.
12902         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
12903         that the insn is actually an insn and it has a non-NULL basic block.
12904         Do not mark basic block dirty if only deleting a DEBUG_INSN.
12906 2013-03-22  Richard Biener  <rguenther@suse.de>
12908         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
12909         dep_ref members.
12910         (mem_ref_alloc): Do not allocate them.
12911         (refs_independent_p): Do not query or maintain a cache.
12913 2013-03-22  Richard Biener  <rguenther@suse.de>
12915         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
12916         (gather_mem_refs_in_loops): Do not compute it.
12917         (analyze_memory_references): Do not allocate it.
12918         (tree_ssa_lim_finalize): Do not free it.
12919         (for_all_locs_in_loop): Do not query all_refs_in_loop.
12921 2013-03-22  Richard Biener  <rguenther@suse.de>
12923         * is-a.h (as_a): Use gcc_checking_assert.
12925 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
12927         * config/aarch64/aarch64.c (aarch64_print_operand): New
12928         format specifier for printing a constant in hex.
12929         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
12930         format specifier for printing second operand.
12932 2013-03-22  Richard Biener  <rguenther@suse.de>
12934         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
12935         bitmaps.
12936         (gather_mem_refs_in_loops): Perform store accumulation here.
12937         (create_vop_ref_mapping_loop): Remove.
12938         (create_vop_ref_mapping): Likewise.
12939         (analyze_memory_references): Initialize refs_stored_in_loop.
12940         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
12941         (record_indep_loop): Remove.
12942         (record_dep_loop): New function.
12943         (ref_indep_loop_p_1): Adjust to only walk over references
12944         in the loop, not its subloops.
12945         (ref_indep_loop_p): Rename to ...
12946         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
12947         maintaining a more fine-grained cache.
12948         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
12949         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
12951 2013-03-22  Richard Biener  <rguenther@suse.de>
12953         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
12954         (struct mem_ref): Make accesses_in_loop a vec of a vec of
12955         aggregate mem_ref_loc.
12956         (free_mem_ref_locs): Inline into ...
12957         (memref_free): ... this and adjust.
12958         (mem_ref_alloc): Adjust.
12959         (mem_ref_locs_alloc): Remove.
12960         (record_mem_ref_loc): Adjust.
12961         (get_all_locs_in_loop): Rewrite into ...
12962         (for_all_locs_in_loop): ... this iterator.
12963         (rewrite_mem_ref_loc): New functor.
12964         (rewrite_mem_refs): Use for_all_locs_in_loop.
12965         (sm_set_flag_if_changed): New functor.
12966         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
12967         (ref_always_accessed): New functor.
12968         (ref_always_accessed_p): Use for_all_locs_in_loop.
12970 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
12972         * tree-pass.h (PROP_gimple_lvec): New.
12973         * passes.c (dump_properties): Handle PROP_gimple_lvec.
12974         (init_optimization_passes): Move pass_lower_vector.
12975         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
12976         PROP_gimple_lvec.
12977         (pass_lower_vector): Provide PROP_gimple_lvec.
12978         (pass_lower_vector_ssa): Likewise.
12979         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
12981 2013-03-21  Mark Wielaard  <mjw@redhat.com>
12983         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
12985 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
12987         * config/i386/i386.md (*movdi_internal): Disparage slightly
12988         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
12989         conversion alternatives.
12991 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
12993         PR middle-end/48087
12994         * diagnostic.def (DK_WERROR): New kind.
12995         * diagnostic.h (werrorcount): Define.
12996         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
12997         promoted to DK_ERROR, increment DK_WERROR counter instead of
12998         DK_ERROR counter.
12999         * toplev.c (toplev_main): Call print_ignored_options even if
13000         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13001         even if just werrorcount is non-zero.
13003         PR debug/55608
13004         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13005         on failure.
13006         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13007         (string_cst_pool_decl): New function.
13008         (optimize_one_addr_into_implicit_ptr): New function.
13009         (resolve_addr_in_expr): Optimize DWARF location expression
13010         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13011         which doesn't live in memory, but has DW_AT_location or
13012         DW_AT_const_value, or refers to a string literal, into
13013         DW_OP_GNU_implicit_pointer.
13014         (optimize_location_into_implicit_ptr): New function.
13015         (resolve_addr): If removing DW_AT_location of a variable because
13016         it was DW_OP_addr of address of the variable, but the variable doesn't
13017         live in memory, try to emit const value attribute for the initializer.
13019 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13021         * tree.h (VECTOR_TYPE_P): New macro.
13022         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13023         TYPE_MODE): Use it.
13024         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13025         VEC_COND_EXPR cannot be lvalues.
13026         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13028 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13030         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13031         Restrict the transformation to equal modes.
13033 2013-03-21  Richard Biener  <rguenther@suse.de>
13035         PR tree-optimization/39326
13036         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13037         (MEM_ANALYZABLE): Adjust.
13038         (record_mem_ref_loc): Move bitmap ops ...
13039         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13040         unanalyzable refs, do not record locations for it.
13041         (analyze_memory_references): Allocate ref zero as shared
13042         unanalyzable ref.
13043         (refs_independent_p): Do not test for unanalyzed mems here.
13044         (ref_indep_loop_p_1): Special-case disambiguation against
13045         the unanalyzed ref.
13046         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13048 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13050         * config/arm/arm-protos.h (tune_params): Add
13051         prefer_neon_for_64bits field.
13052         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13053         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13054         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13055         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13056         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13057         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13058         (arm_option_override): Handle -mneon-for-64bits new option.
13059         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13060         (prefer_neon_for_64bits): Declare new variable.
13061         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13062         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13063         (arch_enabled): Handle new arch types. Remove support for onlya8
13064         and nota8.
13065         (one_cmpldi2): Use new arch names.
13066         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13067         * config/arm/arm.opt (mneon-for-64bits): Add option.
13068         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13069         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13070         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13071         of onlya8.
13072         * doc/invoke.texi (-mneon-for-64bits): Document.
13074 2013-03-21  Richard Biener  <rguenther@suse.de>
13076         PR tree-optimization/39326
13077         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13078         (sort_bbs_in_loop_postorder_cmp): New function.
13079         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13081 2013-03-21  Richard Biener  <rguenther@suse.de>
13083         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13084         (vect_insert_into_interleaving_chain): Likewise.
13085         (vect_drs_dependent_in_basic_block): Inline ...
13086         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13087         split out from ...
13088         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13089         (vect_check_interleaving): Simplify.
13090         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13091         (vect_slp_analyze_data_ref_dependences): ... this new function.
13092         (dr_group_sort_cmp): New function.
13093         (vect_analyze_data_ref_accesses): Compute data-reference groups
13094         here instead of in vect_analyze_data_ref_dependence.  Use
13095         a more efficient algorithm.
13096         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13097         vect_slp_analyze_data_ref_dependences.  Call
13098         vect_analyze_data_ref_accesses earlier.
13099         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13100         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13101         (vect_slp_analyze_data_ref_dependences): New prototype.
13103 2013-03-21  Richard Biener  <rguenther@suse.de>
13105         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13106         ref is stored in the loop.
13107         (find_refs_for_sm): Walk only over all stores.
13108         (store_motion_loop): Allocate from lim_bitmap_obstack.
13109         (store_motion): Likewise.
13111 2013-03-21  Richard Biener  <rguenther@suse.de>
13113         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13114         Update virtual SSA form.
13116 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13118         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13119         * configure: Regenerate.
13120         * config.in: Regenerate.
13121         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13122         if !HAVE_LD_EH_FRAME_CIEV3.
13124 2013-03-21  Richard Biener  <rguenther@suse.de>
13126         * tree-cfg.c (verify_expr_no_block): New function.
13127         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13128         nor DECL_VALUE_EXPR have locations with associated blocks.
13129         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13130         (clear_unused_block_pointer): Remove code dealing with
13131         blocks in DECL_DEBUG_EXPR locations.
13133 2013-03-21  Richard Biener  <rguenther@suse.de>
13135         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13136         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
13137         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13138         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13139         instead of DECL_DEBUG_EXPR_IS_FROM.
13140         * gimplify.c (gimplify_modify_expr): Likewise.
13141         * tree-cfg.c (verify_expr_location_1): Likewise.
13142         * tree-complex.c (create_one_component_var): Likewise.
13143         * tree-sra.c (create_access_replacement): Likewise.
13144         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13145         (clear_unused_block_pointer): Likewise.
13146         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13147         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13148         * var-tracking.c (var_debug_decl): Likewise.
13149         (track_expr_p): Likewise.
13150         * tree-inline.c (add_local_variables): Likewise.  Set
13151         DECL_HAS_DEBUG_EXPR_P after copying it.
13152         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13153         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
13155 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13157         PR bootstrap/56656
13158         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13159         * configure: Regenerate.
13160         * config.in: Regenerate.
13161         * config/i386/i386.md (*movdf_internal): Use
13162         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13163         movd instead of movq mnemonic for interunit moves.
13164         (*movdi_internal): Ditto.
13166 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13168         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13169         (abd<mode>_3): New pattern.
13170         (aba<mode>_3): New pattern.
13171         (fabd<mode>_3): New pattern.
13173 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13175         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13176         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13177         occurrence of REGISTER_PREFIX as its empty string.
13179 2013-03-20  Jeff Law  <law@redhat.com>
13181         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13182         addititional equivalences for equality comparisons between an SSA_NAME
13183         and a constant where the SSA_NAME was set from a widening conversion.
13185 2013-03-20  Walter Lee  <walt@tilera.com>
13187         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13189 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13191         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13192         depending on type attribute.
13193         (*movti_internal): Ditto.
13194         (*movtf_internal): Ditto.
13195         (*movxf_internal): Ditto.
13196         (*movdf_internal): Ditto.
13197         (*movsf_internal): Ditto.
13199 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13201         * config/i386/i386.md (*movti_internal): Set prefix attribute to
13202         maybe_vex for sselog1 and ssemov types.
13203         (*movdi_internal): Reorder operand constraints.
13204         (*movsi_internal): Ditto.  Set prefix attribute to
13205         maybe_vex for sselog1 and ssemov types.
13206         (*movtf_internal): Set prefix attribute to maybe_vex
13207         for sselog1 and ssemov types.
13208         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
13209         DImode ssemov types.  Reorder operand constraints.
13210         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
13211         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
13212         attribute for SImode ssemov types.  Reorder operand constraints.
13214 2013-03-20  Martin Jambor  <mjambor@suse.cz>
13216         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13217         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13219 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
13221         * config/rs6000/predicates.md (indexed_address, update_address_mem
13222         update_indexed_address_mem): New predicates.
13223         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13224         attribute for load/store instructions.
13225         * config/rs6000/dfp.md (movsd_store): Likewise.
13226         (movsd_load): Likewise.
13227         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13228         (unnamed HI->DI extend define_insn): Likewise.
13229         (unnamed SI->DI extend define_insn): Likewise.
13230         (unnamed QI->SI extend define_insn): Likewise.
13231         (unnamed QI->HI extend define_insn): Likewise.
13232         (unnamed HI->SI extend define_insn): Likewise.
13233         (unnamed HI->SI extend define_insn): Likewise.
13234         (extendsfdf2_fpr): Likewise.
13235         (movsi_internal1): Likewise.
13236         (movsi_internal1_single): Likewise.
13237         (movhi_internal): Likewise.
13238         (movqi_internal): Likewise.
13239         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13240         attribute for load/store instructions.
13241         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13242         instructions.
13243         (mov<mode>_softfloat): Likewise.
13244         (mov<mode>_hardfloat32): Likewise.
13245         (mov<mode>_hardfloat64): Likewise.
13246         (mov<mode>_softfloat64): Likewise.
13247         (movdi_internal32): Likewise.
13248         (movdi_internal64): Likewise.
13249         (probe_stack_<mode>): Likewise.
13251 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
13253         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13254         floating point, and decimal floating point to reload iterator.
13256         * config/rs6000/constraints.md (wl constraint): New constraints to
13257         return FLOAT_REGS if certain options are used to reduce the number
13258         of separate patterns that exist in the file.
13259         (wx constraint): Likewise.
13260         (wz constraint): Likewise.
13262         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13263         -mdebug=reg, print wg, wl, wx, and wz constraints.
13264         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13265         Initialize the reload functions for 64-bit binary/decimal floating
13266         point types.
13267         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13268         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13269         create the buffer on the stack to overcome not having a 32-bit
13270         load and store.
13271         (rs6000_emit_move): Likewise.
13272         (rs6000_secondary_memory_needed_rtx): Likewise.
13273         (rs6000_alloc_sdmode_stack_slot): Likewise.
13274         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13275         via xxlxor, just like DFmode 0.0.
13277         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13278         define as 1 if we are running on a power7 or newer.
13279         (enum r6000_reg_class_enum): Add new constraints.
13281         * config/rs6000/dfp.md (movsd): Delete, combine with binary
13282         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13283         with other moves by using conditional constraits (wg).  Use LFIWZX
13284         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
13285         (movsd splitter): Likewise.
13286         (movsd_hardfloat): Likewise.
13287         (movsd_softfloat): Likewise.
13289         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13290         binary and decimal floating point moves.
13291         (fmove_ok): New attributes to combine binary and decimal floating
13292         point moves, and to combine power6x (mfpgpr) moves along normal
13293         floating moves.
13294         (real_value_to_target): Likewise.
13295         (f32_lr): Likewise.
13296         (f32_lm): Likewise.
13297         (f32_li): Likewise.
13298         (f32_sr): Likewise.
13299         (f32_sm): Likewise.
13300         (f32_si): Likewise.
13301         (movsf): Combine binary and decimal floating point moves.  Combine
13302         power6x (mfpgpr) moves with other moves by using conditional
13303         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
13304         (mov<mode> for SFmode/SDmode); Likewise.
13305         (SFmode/SDmode splitters): Likewise.
13306         (movsf_hardfloat): Likewise.
13307         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13308         (movsf_softfloat): Likewise.
13309         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13311         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13312         wx and wz constraints.
13314         * config/rs6000/constraints.md (wg constraint): New constraint to
13315         return FLOAT_REGS if -mmfpgpr (power6x) was used.
13317         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13318         constraint.
13320         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13321         -mdebug=reg, print wg, wl, wx, and wz constraints.
13322         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13323         Initialize the reload functions for 64-bit binary/decimal floating
13324         point types.
13325         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13326         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13327         create the buffer on the stack to overcome not having a 32-bit
13328         load and store.
13329         (rs6000_emit_move): Likewise.
13330         (rs6000_secondary_memory_needed_rtx): Likewise.
13331         (rs6000_alloc_sdmode_stack_slot): Likewise.
13332         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13333         via xxlxor, just like DFmode 0.0.
13335         * config/rs6000/dfp.md (movdd): Delete, combine with binary
13336         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
13337         with other moves by using conditional constraits (wg).  Use LFIWZX
13338         and STFIWX for loading SDmode on power7.
13339         (movdd splitters): Likewise.
13340         (movdd_hardfloat32): Likewise.
13341         (movdd_softfloat32): Likewise.
13342         (movdd_hardfloat64_mfpgpr): Likewise.
13343         (movdd_hardfloat64): Likewise.
13344         (movdd_softfloat64): Likewise.
13346         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13347         64-bit binary and decimal floating point moves.
13348         (FMOVE64X): Likewise.
13349         (movdf): Combine 64-bit binary and decimal floating point moves.
13350         Combine power6x (mfpgpr) moves with other moves by using
13351         conditional constraits (wg).
13352         (mov<mode> for DFmode/DDmode): Likewise.
13353         (DFmode/DDmode splitters): Likewise.
13354         (movdf_hardfloat32): Likewise.
13355         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13356         (movdf_softfloat32): Likewise.
13357         (movdf_hardfloat64_mfpgpr): Likewise.
13358         (movdf_hardfloat64): Likewise.
13359         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13360         (movdf_softfloat64): Likewise.
13361         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13362         (reload_<mode>_load): Move to later in the file so they aren't in
13363         the middle of the floating point move insns.
13364         (reload_<mode>_store): Likewise.
13366         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13367         constraint.
13369         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13370         constraint if -mdebug=reg.
13371         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13372         Enable using dd reload support if needed.
13374         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13375         binary and decimal floating point moves in rs6000.md.
13376         (movtd_internal): Likewise.
13378         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13379         decimal floating point moves.
13380         (movtf): Likewise.
13381         (movtf_internal): Likewise.
13382         (mov<mode>_internal, TDmode/TFmode): Likewise.
13383         (movtf_softfloat): Likewise.
13384         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13386         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13387         movdi_internal64, using wg constraint for move direct operations.
13388         (movdi_internal64): Likewise.
13390         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13391         MODES_TIEABLE_P for selected modes.  Print the numerical value of
13392         the various virtual registers. Use GPR/FPR first/last values,
13393         instead of hard coding the register numbers.  Print which modes
13394         have reload functions registered.
13395         (rs6000_option_override_internal): If -mdebug=reg, trace the options
13396         settings before/after setting cpu, target and subtarget settings.
13397         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13398         and for secondary reload failures in rs6000_secondary_reload_inner.
13399         (rs6000_secondary_reload_fail): Likewise.
13400         (rs6000_secondary_reload_inner): Likewise.
13402         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13403         macros for first/last GPR and FPR registers.
13404         (LAST_GPR_REGNO): Likewise.
13405         (FIRST_FPR_REGNO): Likewise.
13406         (LAST_FPR_REGNO): Likewise.
13408         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13409         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13410         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13411         (vcond<mode><mode>): Likewise.
13412         (vcondu<mode><mode>): Likewise.
13413         (vector_gtu<mode>): Likewise.
13414         (vector_gte<mode>): Likewise.
13415         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13416         to prevent the compiler from converting DImode operations to TImode.
13417         (ior<mode>3): Likewise.
13418         (and<mode>3): Likewise.
13419         (one_cmpl<mode>2): Likewise.
13420         (nor<mode>3): Likewise.
13421         (andc<mode>3): Likewise.
13423         * config/rs6000/constraints.md (wt constraint): New constraint
13424         that returns VSX_REGS if TImode is allowed in VSX registers.
13426         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13427         constant under VSX.
13429         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13430         similar to TImode, but it is restricted to being in the GPRs.
13432         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13433         TImode to occupy a single VSX register.
13435         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13436         -mvsx-timode for power7/power8.
13437         (power7 cpu): Likewise.
13438         (power8 cpu): Likewise.
13440         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13441         sure that TFmode/TDmode take up two registers if they are ever
13442         allowed in the upper VSX registers.
13443         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13444         registers.
13445         (rs6000_init_hard_regno_mode_ok): Likewise.
13446         (rs6000_debug_reg_global): Add debugging for PTImode and wt
13447         constraint.  Print if LRA is turned on.
13448         (rs6000_option_override_internal): Give an error if -mvsx-timode
13449         and VSX is not enabled.
13450         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
13451         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13452         to reg+offset addressing.  Use PTImode when checking offset
13453         addresses for validity.
13454         (reg_offset_addressing_ok_p): Likewise.
13455         (rs6000_legitimate_offset_address_p): Likewise.
13456         (rs6000_legitimize_address): Likewise.
13457         (rs6000_legitimize_reload_address): Likewise.
13458         (rs6000_legitimate_address_p): Likewise.
13459         (rs6000_eliminate_indexed_memrefs): Likewise.
13460         (rs6000_emit_move): Likewise.
13461         (rs6000_secondary_reload): Likewise.
13462         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
13463         reloads to fpr registers to continue to use reg+offset addressing,
13464         but 64-bit reloads to altivec registers need reg+reg addressing.
13465         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13466         it.  Treat LO_SUM like a PLUS operation.
13467         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13468         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13469         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13470         registers to share a register with a smaller sized type, since VSX
13471         puts scalars in the upper 64-bits.
13472         (print_operand): Add support for PTImode.
13473         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13474         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13475         registers, but don't have arithmetic support.
13476         (rs6000_memory_move_cost): Add test for VSX.
13477         (rs6000_opt_masks): Add -mvsx-timode.
13479         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13480         for TImode.
13481         (VSs): Likewise.
13482         (VSr): Use wt constraint for TImode.
13483         (VSv): Drop TImode support.
13484         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13485         (vsx_movti_64bit): Likewise.
13486         (vsx_movti_32bit): Likewise.
13487         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13488         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13489         one '?' on the appropriate output constraint.  Do not allow TImode
13490         logical operations on 32-bit systems.
13491         (vsx_ior<mode>3): Likewise.
13492         (vsx_xor<mode>3): Likewise.
13493         (vsx_one_cmpl<mode>2): Likewise.
13494         (vsx_nor<mode>3): Likewise.
13495         (vsx_andc<mode>3): Likewise.
13496         (vsx_concat_<mode>): Likewise.
13497         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13499         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13500         OPTION_MASK_VSX_TIMODE.
13501         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13502         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13504         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13505         (TI2 iterator): New iterator for TImode, PTImode.
13506         (wd mode attribute): Add values for vector types.
13507         (movti_string): Replace TI move operations with operations for TImode
13508         and PTImode.  Add support for TImode being allowed in VSX registers.
13509         (mov<mode>_string, TImode/PTImode): Likewise.
13510         (movti_ppc64): Likewise.
13511         (mov<mode>_ppc64, TImode/PTImode): Likewise.
13512         (TI mode splitters): Likewise.
13514         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13515         constraint.
13517 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
13519         PR tree-optimization/56355
13520         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13521         Also handle integers with undefined overflow.
13523 2013-03-20  Catherine Moore  <clm@codesourcery.com>
13524             Maciej W. Rozycki  <macro@codesourcery.com>
13525             Tom de Vries  <tom@codesourcery.com>
13526             Nathan Sidwell <nathan@codesourcery.com>
13527             Iain Sandoe  <iain@codesourcery.com>
13528             Nathan Froyd  <froydnj@codesourcery.com>
13529             Chao-ying Fu <fu@mips.com>
13531         * doc/extend.texi: (micromips, nomicromips, nocompression):
13532         Document new function attributes.
13533         * doc/invoke.texi (minterlink-compressed, mmicromips,
13534         m14k, m14ke, m14kec): Document new options.
13535         (minterlink-mips16): Update documentation.
13536         * doc/md.texi (ZC, ZD): Document new constraints.
13537         * configure.ac (gcc_cv_as_micromips): Check if linker
13538         supports the .set micromips directive.
13539         * configure: Regenerate.
13540         * config.in: Regenerate.
13541         * config/mips/mips-tables.opt: Regenerate.
13542         * config/mips/micromips.md: New file.
13543         * constraints.md (ZC, ZD): New constraints.
13544         * config/mips/predicates.md (movep_src_register): New predicate.
13545         (movep_src_operand): New predicate.
13546         (non_volatile_mem_operand): New predicate.
13547         * config/mips/mips.md (multimem): New type.
13548         (length): Differentiate between 17-bit and 18-bit branch offsets.
13549         (MOVEP1, MOVEP2): New mode iterator.
13550         (mov_<load>l): Use ZC constraint.
13551         (mov_<load>r): Likewise.
13552         (mov_<store>l): Likewise.
13553         (mov_<store>r): Likewise.
13554         (*branch_equality<mode>_inverted): Add microMIPS support.
13555         (*branch_equality<mode>): Likewise.
13556         (*jump_absolute): Likewise.
13557         (indirect_jump_<mode>): Likewise.
13558         (tablejump_<mode>): Likewise.
13559         (<optab>_internal): Likewise.
13560         (sibcall_internal): Likewise.
13561         (sibcall_value_internal): Likewise.
13562         (prefetch): Use constraint ZD.
13563         * config/mips/mips.opt (minterlink-compressed): New option.
13564         (minterlink-mips16): Now an alias for minterlink-compressed.
13565         (mmicromips): New option.
13566         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13567         (compare_and_swap_12): Likewise.
13568         (sync_add<mode>): Likewise.
13569         (sync_<optab>_12): Likewise.
13570         (sync_old_<optab>_12): Likewise.
13571         (sync_new_<optab>_12): Likewise.
13572         (sync_nand_12): Likewise.
13573         (sync_old_nand_12): Likewise.
13574         (sync_new_nand_12): Likewise.
13575         (sync_sub<mode>): Likewise.
13576         (sync_old_add<mode>): Likewise.
13577         (sync_old_sub<mode>): Likewise.
13578         (sync_new_add<mode>): Likewise.
13579         (sync_new_sub<mode>): Likewise.
13580         (sync_<optab><mode>): Likewise.
13581         (sync_old_<optab><mode>): Likewise.
13582         (sync_new_<optab><mode>): Likewise.
13583         (sync_nand<mode>): Likewise.
13584         (sync_old_nand<mode>): Likewise.
13585         (sync_new_nand<mode>): Likewise.
13586         (sync_lock_test_and_set<mode>): Likewise.
13587         (test_and_set_12): Likewise.
13588         (atomic_compare_and_swap<mode>): Likewise.
13589         (atomic_exchange<mode>_llsc): Likewise.
13590         (atomic_fetch_add<mode>_llsc): Likewise.
13591         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13592         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13593         (umips_save_restore_pattern_p): Likewise.
13594         (umips_load_store_pair_p): Likewise.
13595         (umips_output_load_store_pair): Likewise.
13596         (umips_movep_target_p): Likewise.
13597         (umips_12bit_offset_address_p): Likewise.
13598         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13599         (mips_base_mips16): Rename this...
13600         (mips_base_compression_flags): ...to this. Update all uses.
13601         (mips_attribute_table): Add micromips, nomicromips and nocompression.
13602         (mips_mips16_decl_p): Delete.
13603         (mips_nomips16_decl_p): Delete.
13604         (mips_get_compress_on_flags): New function.
13605         (mips_get_compress_off_flags): New function.
13606         (mips_get_compress_mode): New function.
13607         (mips_get_compress_on_name): New function.
13608         (mips_get_compress_off_name): New function.
13609         (mips_insert_attributes): Support multiple compression types.
13610         (mips_merge_decl_attributes): Likewise.
13611         (umips_12bit_offset_address_p): New function.
13612         (mips_start_function_definition): Emit .set micromips directive.
13613         (mips_call_may_need_jalx_p): New function.
13614         (mips_function_ok_for_sibcall): Add microMIPS support.
13615         (mips_print_operand_punctuation): Support short delay slots and
13616         compact jumps.
13617         (umips_swm_mask, umips_swm_encoding): New.
13618         (umips_build_save_restore): New function.
13619         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13620         (was_mips16_p): Remove.
13621         (old_compression_mode): New.
13622         (mips_set_compression_mode): New function.
13623         (mips_set_current_function): Add microMIPS support.
13624         (mips_option_override): Likewise.
13625         (umips_save_restore_pattern_p): New function.
13626         (umips_output_save_restore): New function.
13627         (umips_load_store_pair_p_1): New function.
13628         (umips_load_store_pair_p): New function.
13629         (umips_output_load_store_pair_1): New function.
13630         (umips_output_load_store_pair): New function.
13631         (umips_movep_target_p) New function.
13632         (mips_prepare_pch_save): Add microMIPS support.
13633         * config/mips/mips.h (TARGET_COMPRESSION): New.
13634         (TARGET_CPU_CPP_BUILTINS): Update macro
13635         to use new compression flags and to support microMIPS.
13636         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13637         (MIPS_ARCH_FLOAT_SPEC): Likewise.
13638         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13639         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13640         (ASM_SPEC): Support mmicromips and mno-micromips.
13641         (M16STORE_REG_P): New macro.
13642         (MIPS_CALL): Support TARGET_MICROMIPS.
13643         (MICROMIPS_J): New macro.
13644         (mips_base_mips16): Rename this...
13645         (mips_base_compression_flags): ...to this.
13646         (UMIPS_12BIT_OFFSET_P): New macro.
13647         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13648         (MULTILIB_DIRNAMES): Likewise.
13649 2013-03-20  Richard Biener  <rguenther@suse.de>
13651         PR tree-optimization/56661
13652         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13653         the result does not have to be distinct.
13655 2013-03-20  Richard Biener  <rguenther@suse.de>
13657         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13658         remap_gimple_op_r.
13660 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13661             Steven Bosscher <steven@gcc.gnu.org>
13663         PR rtl-optimization/56605
13664         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13666 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
13668         PR bootstrap/56656
13669         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13670         that require movd instead of movq.
13672 2013-03-20  Richard Biener  <rguenther@suse.de>
13674         * tree-ssa-structalias.c (struct variable_info): Add pointer
13675         to the first field of an aggregate with sub-vars.  Make
13676         this and the pointer to the next subfield its ID.
13677         (vi_next): New function.
13678         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13679         storedanything_id, integer_id): Increment by one.
13680         (new_var_info, get_call_vi, lookup_call_clobber_vi,
13681         get_call_clobber_vi): Adjust.
13682         (solution_set_expand): Simplify and speedup.
13683         (solution_set_add): Inline into ...
13684         (set_union_with_increment): ... this.  Adjust accordingly.
13685         (do_sd_constraint): Likewise.
13686         (do_ds_constraint): Likewise.
13687         (do_complex_constraint): Simplify.
13688         (build_pred_graph): Adjust.
13689         (solve_graph): Likewise.  Simplify and speedup.
13690         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13691         get_constraint_for_component_ref, get_constraint_for_1,
13692         first_vi_for_offset, first_or_preceding_vi_for_offset,
13693         create_function_info_for, create_variable_info_for_1,
13694         create_variable_info_for, intra_create_variable_infos): Adjust.
13695         (init_base_vars): Push NULL for ID zero.
13696         (compute_points_to_sets): Adjust.
13698 2013-03-20  Richard Biener  <rguenther@suse.de>
13700         * cfgloop.c (verify_loop_structure): Streamline and avoid
13701         ICEing on corrupt loop tree.
13702         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13703         loop tree.
13705 2013-03-20  Richard Biener  <rguenther@suse.de>
13707         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13708         check whether an SSA update is needed.
13710 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
13712         * config/mips/constraints.md (T): Rename to...
13713         (Yf): ...this.
13714         (U): Rename to...
13715         (Yd): ...this.
13716         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13717         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13719 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13721         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13722         (*subsi3_carryin_uxtw): Likewise.
13724 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13726         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13727         (*rorsi3_insn_uxtw): Likewise.
13729 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
13731         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13732         (*extrsi5_insn_uxtw): Likewise.
13734 2013-03-19  Richard Biener  <rguenther@suse.de>
13736         PR tree-optimization/56273
13737         * passes.c (init_optimization_passes): Move second VRP after DOM.
13739 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
13741         * config/i386/i386.md (*movti_internal): Merge from
13742         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
13743         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
13744         nox64 isa attributes.
13746 2013-03-18  Richard Biener  <rguenther@suse.de>
13748         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13749         (unite): Likewise.
13750         (merge_node_constraints): Likewise.
13751         (build_succ_graph): Likewise.
13752         (valid_graph_edge): Inline into single caller.
13753         (unify_nodes): Likewise.  Use bitmap_set_bit return value
13754         and cache varinfo.
13755         (scc_visit): Fix formatting and variable use.
13756         (do_sd_constraint): Use gcc_checking_assert.
13757         (do_ds_constraint): Likewise.
13758         (do_complex_constraint): Likewise.
13759         (condense_visit): Likewise.  Cleanup.
13760         (dump_pred_graph): New function.
13761         (perform_var_substitution): Dump the pred-graph before
13762         variable substitution.
13763         (find_equivalent_node): Use gcc_checking_assert.
13764         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13766 2013-03-18  Richard Biener  <rguenther@suse.de>
13768         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13769         Remove cond_expr_stmt_list argument and do not gimplify the
13770         built expression.
13771         (vect_loop_versioning): Adjust.
13772         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13773         Cleanup to use less temporaries.
13774         (vect_create_data_ref_ptr): Cleanup.
13776 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13778         PR tree-optimization/56635
13779         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13780         require types_compatible_p types.
13782 2013-03-18  Nick Clifton  <nickc@redhat.com>
13784         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13785         spurious backslash.
13787         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13788         Add missing line to comment describing function.
13790 2013-03-18  Richard Biener  <rguenther@suse.de>
13792         PR tree-optimization/56210
13793         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13794         Handle string / character search functions.
13795         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13797 2013-03-18  Richard Biener  <rguenther@suse.de>
13799         PR middle-end/56483
13800         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13801         and implement properly.
13802         * gimple.h (gimple_cond_single_var_p): Remove.
13804 2013-03-18  Richard Biener  <rguenther@suse.de>
13806         * tree-data-ref.h (find_data_references_in_loop): Declare.
13807         * tree-data-ref.c (get_references_in_stmt): Use a stack
13808         vector pre-allocated in the callers.
13809         (find_data_references_in_stmt): Adjust.
13810         (graphite_find_data_references_in_stmt): Likewise.
13811         (create_rdg_vertices): Likewise.
13812         (find_data_references_in_loop): Export.
13813         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13814         Compute dependences here...
13815         (vect_analyze_data_refs): ...not here.  When we encounter
13816         a non-vectorizable data reference in basic-block vectorization
13817         truncate the data reference vector.  Do not bother to
13818         fixup data-dependence information for gather loads.
13819         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13820         of data references, as reported.
13822 2013-03-18  Richard Biener  <rguenther@suse.de>
13824         PR tree-optimization/3713
13825         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
13826         has_constants and expr.
13827         (stmt_has_constants): Properly valueize SSA names when deciding
13828         whether the stmt has constants.
13830 2013-03-18  Richard Biener  <rguenther@suse.de>
13832         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13833         whole function when there is nothing to do.
13834         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13835         * tree-vectorizer.c (vectorize_loops): Update virtual and
13836         loop-closed SSA once.
13837         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13839 2013-03-18  Richard Biener  <rguenther@suse.de>
13841         PR middle-end/56113
13842         * domwalk.c (bb_postorder): New global static.
13843         (cmp_bb_postorder): New function.
13844         (walk_dominator_tree): Replace scheme imposing an order for
13845         visiting dominator sons by one sorting them at the time they
13846         are pushed on the stack.
13848 2013-03-18  Richard Biener  <rguenther@suse.de>
13850         PR tree-optimization/39326
13851         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
13852         (struct mem_ref): Replace mem member with ao_ref typed member.
13853         (MEM_ANALYZABLE): Adjust.
13854         (memref_eq): Likewise.
13855         (mem_ref_alloc): Likewise.
13856         (gather_mem_refs_stmt): Likewise.
13857         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
13858         (execute_sm_if_changed_flag_set): Adjust.
13859         (execute_sm): Likewise.
13860         (ref_always_accessed_p): Likewise.
13861         (refs_independent_p): Likewise.
13862         (can_sm_ref_p): Likewise.
13864 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
13866         PR c/56566
13867         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
13868         return 1 even for !unsignedp.
13870 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13872         * config/i386/i386.md (isa): Add x64 and nox64.
13873         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
13874         (*pushtf): Enable *roF alternative for x64 isa only.
13875         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
13876         mode attribute of integer alternatives to DImode for TARGET_64BIT.
13877         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
13878         (*movtf_internal): Merge from *movtf_internal_rex64 and
13879         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
13880         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
13881         nox64 isa attributes.
13882         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
13883         nox64 isa attributes.
13884         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
13886 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13888         * config/alpha/alpha.c (TARGET_LRA_P): New define.
13890 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
13892         PR target/56640
13893         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
13894         class names.  Remove trailing comma after "ALL_REGS".
13896 2013-03-16  Jan Hubicka  <jh@suse.cz>
13898         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
13899         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
13900         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
13901         of cgraph_get_create_node.
13902         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13904 2013-03-16  Jason Merrill  <jason@redhat.com>
13906         PR debug/49090
13907         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
13908         with DW_AT_default_value.
13910 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
13912         * BASE-VER: Set to 4.9.0.
13914 2013-03-14  Andi Kleen  <ak@linux.intel.com>
13916         PR target/56619
13917         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
13918         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
13919         Document _x* TSX intrinsics.
13921 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
13922             David Holsgrove  <david.holsgrove@xilinx.com>
13924         * configure.ac: Add MicroBlaze TLS support detection.
13925         * configure: Regenerate.
13926         * config/microblaze/microblaze-protos.h
13927         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
13928         symbol_mentioned_p, label_mentioned_p): Add prototypes.
13929         * config/microblaze/microblaze.c (microblaze_address_type): Add
13930         ADDRESS_TLS and tls_reloc address types.
13931         (microblaze_address_info): Add tls_reloc.
13932         (TARGET_HAVE_TLS): Define.
13933         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
13934          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
13935          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
13936         load_tls_operand,  microblaze_call_tls_get_addr,
13937         microblaze_legitimize_tls_address): New functions.
13938         (microblaze_classify_unspec): Handle UNSPEC_TLS.
13939         (get_base_reg): Use microblaze_tls_symbol_p.
13940         (microblaze_classify_address): Handle TLS.
13941         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
13942         label_mentioned_p and microblaze_tls_referenced_p.
13943         (microblaze_legitimize_address): Handle TLS.
13944         (microblaze_address_insns): Handle ADDRESS_TLS.
13945         (pic_address_needs_scratch): Handle TLS.
13946         (print_operand_address): Handle TLS.
13947         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
13948         (microblaze_expand_move): Handle TLS.
13949         (microblaze_legitimate_constant_p): Check
13950         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
13951         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13952         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
13953         (PIC_OFFSET_TABLE_REGNUM): Set.
13954         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
13955         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
13956         (addsi3, movsi_internal2, movdf_internal): Update constraints
13957         * config/microblaze/predicates.md (arith_plus_operand): Define
13958         (move_operand): Redefine as move_src_operand,
13959         check microblaze_tls_referenced_p.
13961 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
13963         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
13964         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
13966 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
13968         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
13969         CC mode for AND.
13971 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
13973         PR tree-optimization/53265
13974         * common.opt (Waggressive-loop-optimizations): New option.
13975         * tree-ssa-loop-niter.c: Include tree-pass.h.
13976         (do_warn_aggressive_loop_optimizations): New function.
13977         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
13978         if number_of_latch_executions returned constant.
13979         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
13980         early.  If number_of_latch_executions returned constant, set
13981         nb_iterations_upper_bound back to it.
13982         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
13983         field.
13984         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
13985         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
13987         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
13988         (MULTILIB_OSDIRNAMES): Set.
13989         * genmultilib: If defaultosdirname doesn't start with :: , set
13990         defaultosdirname2 instead, clear it and emit two . multilib_raw
13991         entries instead of just one.
13993 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
13995         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
13996         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
13997         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
13998         (SUBTARGET_OVERRIDE_OPTIONS): New.
14000 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14002         PR target/49880
14003         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14004         (musermode): Convert to Var(TARGET_USERMODE).
14005         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14006         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14007         * config/sh/sh.c (sh_option_override): Use
14008         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14009         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14010         condition.
14011         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14012         TARGET_SH4.
14013         (udivsi3_i4_single, divsi3_i4_single): Use
14014         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14016 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14018         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14019         default setting.
14021 2013-03-13  Richard Biener  <rguenther@suse.de>
14023         PR tree-optimization/56608
14024         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14025         calls when vectorizing basic-blocks.
14027 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14029         PR plugins/45078
14030         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14031         tm_file.
14033 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14035         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14037 2013-03-11  Jan Hubicka  <jh@suse.cz>
14039         PR lto/56557
14040         * lto-streamer-out.c (output_symbol_p): Skip references from
14041         constructors of external variables.
14043 2013-03-11  Jan Hubicka  <jh@suse.cz>
14045         PR middle-end/56571
14046         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14047         from pseudos.
14048         * emit-rtl.c (verify_rtx_sharing): Likewise.
14049         (copy_insn_1): Likewise.
14050         * rtl.c (copy_rtx): Likewise.
14052 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14054         PR target/56591
14055         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14056         output_operand_lossage message.
14058 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14060         PR target/56470
14061         * arm.c (shift_op): Validate RTL pattern on the fly.
14062         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14063         the RTL.
14065 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14067         PR target/56347
14068         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14069         new call patterns that clobber %fr12.
14070         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14071         split and postreload patterns.
14072         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14073         registers %fr12 and %fr12R as call used.
14075 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14077         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14078         (canon_address, record_store, replace_read, check_mem_read_rtx,
14079         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14080         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14081         rest_of_handle_dse): Likewise.
14083 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14085         PR middle-end/56524
14086         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14087         Add base_optabs.
14088         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14089         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14090         (save_optabs_if_changed): Replace with...
14091         (init_tree_optimization_optabs): ...this.
14092         * optabs.c (save_optabs_if_changed): Rename to...
14093         (init_tree_optimization_optabs): ...this.  Take the optimization node
14094         as argument.  Do nothing if the base optabs are already correct.
14095         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14096         to recompute optabs.
14097         * function.h (function): Remove optabs field.
14098         * function.c (invoke_set_current_function_hook): Call
14099         init_tree_optimization_optabs.  Use the result to initialize
14100         this_fn_optabs.
14102 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14104         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14105         if GTMA_HAS_NO_INSTRUMENTATION.
14106         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14107         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14108         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14109         * gimple-pretty-print.c (dump_gimple_transaction): Handle
14110         GTMA_HAS_NO_INSTRUMENTATION.
14112 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
14114         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14115         libasan_preinit.o.
14117 2013-03-08  Marek Polacek  <polacek@redhat.com>
14118             Jakub Jelinek  <jakub@redhat.com>
14120         PR tree-optimization/56478
14121         * predict.c (is_comparison_with_loop_invariant_p): Change the
14122         type of loop_step to tree.
14123         (predict_loops): Adjust.
14124         (predict_iv_comparison): Perform the computations on double_ints.
14126 2013-03-08  Richard Biener  <rguenther@suse.de>
14128         PR tree-optimization/56570
14129         * tree-cfg.c (verify_expr_location_1): Verify locations for
14130         DECL_DEBUG_EXPR.
14131         * tree-sra.c (create_access_replacement): Strip locations
14132         from DECL_DEBUG_EXPRs.
14134 2013-03-08  Richard Biener  <rguenther@suse.de>
14136         * tree-inline.c (expand_call_inline): Do not associate
14137         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14138         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14140 2013-03-08  Richard Biener  <rguenther@suse.de>
14142         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14143         or block changes with -Og.  Fix for location / block encoding
14144         changes and PHI arguments with locations.
14146 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
14148         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14149         for all counters.
14150         (struct output_info): Likewise.
14151         (register_overhead): Remove bad gcc_assert.
14152         (bitmap_find_bit): If there is only a single bitmap element, do not
14153         count a miss as a search.
14154         (print_statistics): Update for counter type changes.
14155         (dump_bitmap_statistics): Likewise.  Print headers such that they
14156         are properly lined up with the printed counters.
14158 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
14160         PR tree-optimization/56559
14161         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14162         check that it has only a single use.
14164 2013-03-07  Richard Biener  <rguenther@suse.de>
14166         * doc/invoke.texi (fwhole-program): Discourage use in combination
14167         with -flto.
14169 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
14171         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14173         PR tree-optimization/56539
14174         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14175         instead of GSI_CONTINUE_LINKING as last argument to
14176         force_gimple_operand_gsi.  Adjust function comment.
14178         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14179         aarch64-cores.def.
14181         PR middle-end/56548
14182         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14183         promoted mode, convert the result back to the original mode.
14185 2013-03-06  Richard Biener  <rguenther@suse.de>
14187         PR middle-end/56294
14188         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14189         (insert_updated_phi_nodes_compare_uids): New function.
14190         (update_ssa): Sort symbols_to_rename after UID before
14191         traversing it to insert PHI nodes.
14193 2013-03-06  Richard Biener  <rguenther@suse.de>
14195         PR middle-end/50494
14196         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14197         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14199         Revert
14200         2013-02-13  Richard Biener  <rguenther@suse.de>
14202         PR lto/50494
14203         * varasm.c (output_constant_def_1): Get the decl representing
14204         the constant as argument.
14205         (output_constant_def): Wrap output_constant_def_1.
14206         (make_decl_rtl): Use output_constant_def_1 with the decl
14207         representing the constant.
14208         (build_constant_desc): Optionally re-use a decl already
14209         representing the constant.
14210         (tree_output_constant_def): Adjust.
14212 2013-03-06  Joey Ye  <joey.ye@arm.com>
14214         PR lto/50293
14215         * gcc.c (convert_white_space): New function.
14216         (main): Handles white space in function name.
14218 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
14220         PR target/56529
14221         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14222         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
14223         to SH_DIV_CALL_TABLE for TARGET_SH2.
14224         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14225         list.
14226         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14227         call-table options.
14229 2013-03-05  Sterling Augustine  <saugustine@google.com>
14230             Cary Coutant  <ccoutant@google.com>
14232         PR debug/55364
14233         * dwarf2out.c (resolve_addr): Don't call
14234         remove_loc_list_addr_table_entries a second time for the same
14235         expression.
14237 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14239         PR debug/56510
14240         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14241         (avoid_complex_debug_insns): New function.
14242         (expand_debug_locations): Call it.
14244         PR rtl-optimization/56484
14245         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14246         lifetimes of hard registers on small register class machines.
14248 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14250         * config/microblaze/microblaze-protos.h: Rename
14251         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14252         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14253         fast_interrupt.
14254         (microblaze_fast_interrupt_function_p): New function.
14255         (microblaze_is_interrupt_handler): Rename to
14256         microblaze_is_interrupt_variant and add fast_interrupt check.
14257         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14258         (save_restore_insns): Likewise.
14259         (compute_frame_size): Likewise.
14260         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14261         (microblaze_globalize_label): Likewise.
14262         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14263         * config/microblaze/microblaze.md: Use wrapper
14264         microblaze_is_interrupt_variant.
14266 2013-03-05  Kai Tietz  <ktietz@redhat.com>
14268         * sdbout.c (sdbout_one_type): Switch to current function's section
14269         supporting cold/hot.
14271 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
14273         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14274         -mxl-reorder.
14276 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14278         PR middle-end/56461
14279         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14280         if VALGRIND_GET_VBITS is defined, temporarily make object
14281         memory all defined, and restore previous valgrind addressability
14282         and definability afterwards.  Free this_object at the end.
14284         PR middle-end/56461
14285         * lra.c (lra): Call lra_clear_live_ranges if live_p,
14286         right before calling lra_create_live_ranges, also call it
14287         when clearing live_p.  Only call lra_clear_live_ranges
14288         at the end if live_p.
14290         PR middle-end/56461
14291         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14293 2013-03-05  Richard Biener  <rguenther@suse.de>
14295         PR tree-optimization/56521
14296         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14297         value-id.
14299 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
14301         PR c++/55135
14302         * except.h (remove_unreachable_eh_regions): New prototype.
14303         * except.c (remove_eh_handler_splicer): New function, split out
14304         of remove_eh_handler.
14305         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
14306         warning about running it on many EH regions one at a time.
14307         (remove_unreachable_eh_regions_worker): New function, walk the
14308         EH tree in depth-first order and remove non-marked regions.
14309         (remove_unreachable_eh_regions): New function.
14310         * tree-eh.c (mark_reachable_handlers): New function, split out
14311         from remove_unreachable_handlers.
14312         (remove_unreachable_handlers): Use mark_reachable_handlers and
14313         remove_unreachable_eh_regions.
14314         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14315         and remove_unreachable_eh_regions.
14317 2013-03-05  Richard Biener  <rguenther@suse.de>
14319         PR middle-end/56525
14320         * loop-init.c (fix_loop_structure): Remove loops in two stages,
14321         not freeing them until the end.
14323 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14325         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14327 2013-03-05  Richard Biener  <rguenther@suse.de>
14329         PR tree-optimization/56270
14330         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14331         of loads after scheduling an SLP instance.
14333 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14335         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14336         tic6x.exp.
14337         (check_gcc_parallelize): Run guality.exp as a separate job from
14338         vect.exp with unsorted.exp and $(dg_target_exps) separately from
14339         struct-layout-1.exp with stackalign.exp.
14341         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14343         PR middle-end/56461
14344         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14345         load_index sbitmap even if some bit in it isn't set.
14347         PR middle-end/56461
14348         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14349         (discover_iteration_bound_by_body_walk): Change queues to
14350         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
14351         spelling in comment.  Call safe_push on queues[bound_index] directly.
14352         Release queues[queue_index] in every iteration unconditionally.
14353         Release bounds vector.
14355         PR middle-end/56461
14356         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14357         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14358         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14359         inner_phis vector.
14361 2013-03-05  Richard Biener  <rguenther@suse.de>
14363         PR lto/56515
14364         * tree-inline.c (remap_blocks_to_null): New function.
14365         (expand_call_inline): When expanding a call stmt without
14366         an associated block inline remap all callee blocks to NULL.
14368 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
14370         PR rtl-optimization/56494
14371         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14372         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14373         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14375         PR middle-end/56461
14376         * sel-sched-ir.c (free_sched_pools): Release
14377         succs_info_pool.stack[succs_info_pool.max_top] vectors too
14378         if succs_info_pool.max_top isn't -1.
14380         PR bootstrap/56509
14381         * opts.c (opts_obstack, opts_concat): Moved to...
14382         * opts-common.c (opts_obstack, opts_concat): ... here.
14384 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14386         PR middle-end/56461
14387         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14389 2013-03-04  Martin Jambor  <mjambor@suse.cz>
14391         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14392         all appropriate places.
14394 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
14396         PR tree-optimization/56424
14397         * ipa-split.c (split_function): Do not set the RSO flag if result is
14398         not by reference and its type is a register type.
14400 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
14402         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14403         (microblaze_legitimate_pic_operand): Likewise
14404         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14405         new function microblaze_legitimate_pic_operand
14406         * config/microblaze/microblaze-protos.h
14407         (microblaze_legitimate_pic_operand): Declare.
14409 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14411         * config/microblaze/predicates.md (call_insn_simple_operand):
14412         New predicate for supported rtx code types.
14413         * config/microblaze/microblaze.md (call_internal1): Use
14414         call_insn_simple_operand predicate.
14416 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
14418         PR middle-end/56461
14419         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14420         partitions.ordered_remove.
14422         PR middle-end/56461
14423         * tree-vect-stmts.c (vectorizable_conversion): Don't call
14424         vec_oprnds0.create (1) for modifier == NONE.
14426         PR middle-end/56461
14427         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14428         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14429         vec_oprnds1 right before pushing anything to it for
14430         scalar_shift_arg.
14432         PR middle-end/56461
14433         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14434         set nbbs to 0 instead of having separate code path.
14435         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14436         instead of false as last argument if returning NULL.
14438 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
14440         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14441         the attribute is now called "target" instead of "option".
14442         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14443         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
14444         attribute/pragma name for TARGET_OPTION_VALID_P and
14445         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
14446         * doc/tm.texi: Regenerated.
14448 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
14450         * config/microblaze/microblaze.c:
14451         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14452         * config/microblaze/microblaze.h: Add -mxl-reorder to
14453         DRIVER_SELF_SPECS.
14454         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14455         instructions emitted if TARGET_REORDER.
14456         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14457         or 0 for -m/-mno case, but initialises as 2 to detect default use case
14458         separately.
14460 2013-03-01  Xinliang David Li  <davidxl@google.com>
14462         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14463         walk length.
14465 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
14467         PR middle-end/56461
14468         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14469         vector even when returning true.  Fix up function comment formatting.
14471         PR middle-end/56461
14472         * ira-build.c (ira_loop_nodes_count): New variable.
14473         (create_loop_tree_nodes): Initialize it.
14474         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14476         PR middle-end/56461
14477         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14478         method on dr_chain and result_chain.
14479         * tree-vect-stmts.c (vectorizable_store): Only call
14480         result_chain.create if j == 0.
14482         PR middle-end/56461
14483         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14484         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14485         before overwriting it.
14487 2013-03-01  Tobias Burnus  <burnus@net-b.de>
14489         * doc/extended.texi (C Extensions): Change order in @menu
14490         to match @node.
14491         (Other MIPS Built-in Functions): Move last MIPS entry before
14492         "picoChip Built-in Functions".
14493         (SH Built-in Functions): Move after RX Built-in Functions.
14494         * doc/gcc.texi (Introduction): Change order in @menu
14495         to match @node.
14496         * doc/md.texi (Constraints): Ditto.
14497         * gty.texi (Type Information): Ditto.
14498         (User-provided marking routines for template types): Make
14499         subsection.
14500         * doc/invoke.texi (AArch64 Options): Move before
14501         "Adapteva Epiphany Options".
14503 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
14504             Jakub Jelinek  <jakub@redhat.com>
14506         PR sanitizer/56454
14507         * asan.c (gate_asan): Lookup no_sanitize_address instead of
14508         no_address_safety_analysis attribute.
14509         * doc/extend.texi (no_address_safety_attribute): Rename to
14510         no_sanitize_address attribute, mention no_address_safety_analysis
14511         attribute as deprecated alias.
14513 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
14515         PR middle-end/56461
14516         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14517         type to vec<vec<tree> > *.
14518         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
14519         to be vec<tree> instead of vec<tree> *, set vec_defs
14520         to vNULL and call vec_defs.create (number_of_vects), adjust other
14521         uses of vec_defs.
14522         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14523         vectorizable_condition): Adjust vect_get_slp_defs callers.
14525 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14527         * config/aarch64/aarch64.c
14528         (aarch64_float_const_representable): Remove unused variable.
14530 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14532         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14534 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14536         * config/aarch64/aarch64-builtins.c
14537         (aarch64_init_simd_builtins): Make static.
14539 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
14541         * config/aarch64/aarch64.c
14542         (aarch64_simd_make_constant): Make static.
14544 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14546         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14547         with no initialization to the RHS of debug statements.
14549 2013-02-28  Martin Jambor  <mjambor@suse.cz>
14551         PR tree-optimization/56294
14552         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14553         Adjust dumping.
14554         (get_access_replacement): Do not call create_access_replacement.
14555         Assert a replacement exists.
14556         (get_repl_default_def_ssa_name): Create the replacement declaration
14557         itself.
14559 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14561         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14562         final_end_function.
14564 2013-02-28  Marek Polacek  <polacek@redhat.com>
14566         PR rtl-optimization/56466
14567         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14568         if we're changing a loop.
14569         (peel_loops_completely): Likewise.
14571 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
14573         PR c++/55813
14574         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14576 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14578         PR target/56445
14579         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14580         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14581         INTX_FTYPE_FX, FX_FTYPE_INTX.
14582         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14584 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14586         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14587         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14588         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14589         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14590         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14591         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14592         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14593         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14594         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14595         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14596         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14597         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14598         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14599         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14600         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14601         (avrxmega6): Increase max flash segments from 5 to 6.
14602         * config/avr/t-multilib: Regenerate.
14603         * config/avr/avr-tables.opt: Regenerate.
14604         * doc/avr-mmcu.texi: Regenerate.
14606 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
14608         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14609         (avr_device_to_arch): Rename to avr_device_to_ld.
14610         (avr_device_to_as): New prototype.
14611         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14612         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14613         * config/avr/driver-avr.c (avr_device_to_as): New.
14614         (avr_device_to_arch): Rename to avr_device_to_ld.
14616 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14618         PR middle-end/56461
14619         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14620         method on dr_chain and result_chain.
14622         PR middle-end/56461
14623         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14624         pointer_set_destroy on not_executed_last_iteration.
14626         PR middle-end/56461
14627         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
14629         PR middle-end/56461
14630         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14631         FOR_EACH_DEFINED_FUNCTION when freeing state.
14633         PR middle-end/56461
14634         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14635         pool_free.
14636         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14637         overwriting it.
14639         PR middle-end/56461
14640         * ipa-cp.c (decide_whether_version_node): Call vec_free on
14641         known_aggs[i].items and release known_aggs vector.
14643         PR middle-end/56461
14644         * ipa-reference.c (propagate): Free node_info even for alias nodes.
14646 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
14648         * config/microblaze/microblaze.c (microblaze_emit_compare):
14649         Use xor for EQ/NE comparisions.
14650         * config/microblaze/microblaze.md (cstoresf4): Add constraints
14651         (cbranchsf4): Adjust operator to comparison_operator.
14653 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14655         PR middle-end/56461
14656         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14657         vector.
14658         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14659         vec_safe_push, always update *slot.
14660         (redirect_edge_var_map_clear): Use vec_free.
14661         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14662         (free_var_map_entry): Use vec_free.
14663         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14664         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14666 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
14668         PR middle-end/45472
14669         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14670         when the may_trap_p bit of the exprs being merged differs.
14671         Reorder tests for speculativeness in the logical and operator.
14673 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
14675         * incpath.c (add_standard_paths): Use reconcat instead of concat
14676         where appropriate and avoid leaking memory.
14678         * opts.h: Include obstack.h.
14679         (opts_concat): New prototype.
14680         (opts_obstack): New declaration.
14681         * opts.c (opts_concat): New function.
14682         (opts_obstack): New variable.
14683         (init_options_struct): Call gcc_init_obstack on opts_obstack.
14684         (finish_options): Use opts_concat instead of concat
14685         and XOBNEWVEC instead of XNEWVEC.
14686         * opts-common.c (generate_canonical_option, decode_cmdline_option,
14687         generate_option): Likewise.
14688         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14689         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14691         PR target/56455
14692         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14693         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14695 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14697         PR middle-end/56461
14698         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14700 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14702         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14703         (arm_block_move_unaligned_straight): Likewise.
14704         (arm_adjust_block_mem): Likewise.
14706 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
14708         PR target/48901
14709         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14710         temp, cond and label.
14711         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14713         PR target/52500
14714         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14715         * config/c6x/c6x.h (dbx_register_map): Update declaration.
14717         PR target/52501
14718         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14719         of prologue/epilogue functions.
14721         PR target/52550
14722         * config/tilegx/tilegx.c (tilegx_expand_prologue):
14723         Remove unused variable cfa_offset.
14724         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14726         PR target/54639
14727         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14728         type promotion to unsigned.
14730         PR target/54640
14731         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14732         for HOST_WIDE_INT of 32 bit / same size as int.
14733         (arm_block_move_unaligned_straight): Likewise.
14734         (arm_adjust_block_mem): Likewise.
14736         PR target/54662
14737         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14738         ALL_CFLAGS.
14740 2013-02-26  Marek Polacek  <polacek@redhat.com>
14742         PR tree-optimization/56426
14743         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
14745 2013-02-26  Richard Biener  <rguenther@suse.de>
14747         PR target/56444
14748         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14749         unused variable loops.
14751 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
14753         PR tree-optimization/56448
14754         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14755         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14756         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14757         later operands of the references, or even first operand for
14758         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14760         PR tree-optimization/56443
14761         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14762         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14763         to type_for_mode langhook.
14765 2013-02-25  Matt Turner  <mattst88@gmail.com>
14767         * doc/invoke.texi: Document r4700.
14769 2013-02-25  Richard Biener  <rguenther@suse.de>
14771         PR tree-optimization/56175
14772         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14773         split out from ...
14774         (simplify_bitwise_binary): ... here.  Also guard the conversion
14775         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14777 2013-02-25  Catherine Moore  <clm@codesourcery.com>
14779         Revert:
14780         2013-02-24  Catherine Moore  <clm@codesourcery.com>
14781             Maciej W. Rozycki  <macro@codesourcery.com>
14782             Tom de Vries  <tom@codesourcery.com>
14783             Nathan Sidwell  <nathan@codesourcery.com>
14784             Iain Sandoe  <iain@codesourcery.com>
14785             Nathan Froyd  <froydnj@codesourcery.com>
14786             Chao-ying Fu  <fu@mips.com>
14788         * doc/extend.texi: (micromips, nomicromips, nocompression):
14789         Document new function attributes.
14790         * doc/invoke.texi (minterlink-compressed, mmicromips,
14791         m14k, m14ke, m14kec): Document new options.
14792         (minterlink-mips16): Update documentation.
14793         * doc/md.texi (ZC, ZD): Document new constraints.
14794         * configure.ac (gcc_cv_as_micromips): Check if linker
14795         supports the .set micromips directive.
14796         * configure: Regenerate.
14797         * config.in: Regenerate.
14798         * config/mips/mips-tables.opt: Regenerate.
14799         * config/mips/micromips.md: New file.
14800         * constraints.md (ZC, AD): New constraints.
14801         * config/mips/predicates.md (movep_src_register): New predicate.
14802         (movep_src_operand): New predicate.
14803         (non_volatile_mem_operand): New predicate.
14804         * config/mips/mips.md (multimem): New type.
14805         (length): Differentiate between 17-bit and 18-bit branch offsets.
14806         (MOVEP1, MOVEP2): New mode iterator.
14807         (mov_<load>l): Use ZC constraint.
14808         (mov_<load>r): Likewise.
14809         (mov_<store>l): Likewise.
14810         (mov_<store>r): Likewise.
14811         (*branch_equality<mode>_inverted): Add microMIPS support.
14812         (*branch_equality<mode>): Likewise.
14813         (*jump_absolute): Likewise.
14814         (indirect_jump_<mode>): Likewise.
14815         (tablejump_<mode>): Likewise.
14816         (<optab>_internal): Likewise.
14817         (sibcall_internal): Likewise.
14818         (sibcall_value_internal): Likewise.
14819         (prefetch): Use constraint ZD.
14820         * config/mips/mips.opt (minterlink-compressed): New option.
14821         (minterlink-mips16): Now an alias for minterlink-compressed.
14822         (mmicromips): New option.
14823         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14824         (compare_and_swap_12): Likewise.
14825         (sync_add<mode>): Likewise.
14826         (sync_<optab>_12): Likewise.
14827         (sync_old_<optab>_12): Likewise.
14828         (sync_new_<optab>_12): Likewise.
14829         (sync_nand_12): Likewise.
14830         (sync_old_nand_12): Likewise.
14831         (sync_new_nand_12): Likewise.
14832         (sync_sub<mode>): Likewise.
14833         (sync_old_add<mode>): Likewise.
14834         (sync_old_sub<mode>): Likewise.
14835         (sync_new_add<mode>): Likewise.
14836         (sync_new_sub<mode>): Likewise.
14837         (sync_<optab><mode>): Likewise.
14838         (sync_old_<optab><mode>): Likewise.
14839         (sync_new_<optab><mode>): Likewise.
14840         (sync_nand<mode>): Likewise.
14841         (sync_old_nand<mode>): Likewise.
14842         (sync_new_nand<mode>): Likewise.
14843         (sync_lock_test_and_set<mode>): Likewise.
14844         (test_and_set_12): Likewise.
14845         (atomic_compare_and_swap<mode>): Likewise.
14846         (atomic_exchange<mode>_llsc): Likewise.
14847         (atomic_fetch_add<mode>_llsc): Likewise.
14848         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14849         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14850         (umips_save_restore_pattern_p): Likewise.
14851         (umips_load_store_pair_p): Likewise.
14852         (umips_output_load_store_pair): Likewise.
14853         (umips_movep_target_p): Likewise.
14854         (umips_12bit_offset_address_p): Likewise.
14855         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14856         (mips_base_mips16): Rename this...
14857         (mips_base_compression_flags): ...to this. Update all uses.
14858         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14859         (mips_mips16_decl_p): Delete.
14860         (mips_nomips16_decl_p): Delete.
14861         (mips_get_compress_on_flags): New function.
14862         (mips_get_compress_off_flags): New function.
14863         (mips_get_compress_mode): New function.
14864         (mips_get_compress_on_name): New function.
14865         (mips_get_compress_off_name): New function.
14866         (mips_insert_attributes): Support multiple compression types.
14867         (mips_merge_decl_attributes): Likewise.
14868         (umips_12bit_offset_address_p): New function.
14869         (mips_start_function_definition): Emit .set micromips directive.
14870         (mips_call_may_need_jalx_p): New function.
14871         (mips_function_ok_for_sibcall): Add microMIPS support.
14872         (mips_print_operand_punctuation): Support short delay slots and
14873         compact jumps.
14874         (umips_swm_mask, umips_swm_encoding): New.
14875         (umips_build_save_restore): New function.
14876         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14877         (was_mips16_p): Remove.
14878         (old_compression_mode): New.
14879         (mips_set_compression_mode): New function.
14880         (mips_set_current_function): Add microMIPS support.
14881         (mips_option_override): Likewise.
14882         (umips_save_restore_pattern_p): New function.
14883         (umips_output_save_restore): New function.
14884         (umips_load_store_pair_p_1): New function.
14885         (umips_load_store_pair_p): New function.
14886         (umips_output_load_store_pair_1): New function.
14887         (umips_output_load_store_pair): New function.
14888         (umips_movep_target_p) New function.
14889         (mips_prepare_pch_save): Add microMIPS support.
14890         * config/mips/mips.h (TARGET_COMPRESSION): New.
14891         (TARGET_CPU_CPP_BUILTINS): Update macro
14892         to use new compression flags and to support microMIPS.
14893         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14894         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14895         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14896         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14897         (ASM_SPEC): Support mmicromips and mno-micromips.
14898         (M16STORE_REG_P): New macro.
14899         (MIPS_CALL): Support TARGET_MICROMIPS.
14900         (MICROMIPS_J): New macro.
14901         (mips_base_mips16): Rename this...
14902         (mips_base_compression_flags): ...to this.
14903         (UMIPS_12BIT_OFFSET_P): New macro.
14904         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14905         (MULTILIB_DIRNAMES): Likewise.
14907 2013-02-25  Tom de Vries  <tom@codesourcery.com>
14909         PR rtl-optimization/56131
14910         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
14911         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
14912         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
14914 2013-02-25  Tobias Burnus  <burnus@net-b.de>
14916         * doc/invoke.texi (-fsanitize=): Move from optimization
14917         to debugging options.
14919 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14921         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
14923 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
14924             Alexander Monakov  <amonakov@ispras.ru>
14926         PR middle-end/56077
14927         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
14928         flush pending lists also on non-jumps.  Adjust comment.
14930 2013-02-24  Catherine Moore  <clm@codesourcery.com>
14931             Maciej W. Rozycki  <macro@codesourcery.com>
14932             Tom de Vries  <tom@codesourcery.com>
14933             Nathan Sidwell  <nathan@codesourcery.com>
14934             Iain Sandoe  <iain@codesourcery.com>
14935             Nathan Froyd  <froydnj@codesourcery.com>
14936             Chao-ying Fu  <fu@mips.com>
14938         * doc/extend.texi: (micromips, nomicromips, nocompression):
14939         Document new function attributes.
14940         * doc/invoke.texi (minterlink-compressed, mmicromips,
14941         m14k, m14ke, m14kec): Document new options.
14942         (minterlink-mips16): Update documentation.
14943         * doc/md.texi (ZC, ZD): Document new constraints.
14944         * configure.ac (gcc_cv_as_micromips): Check if linker
14945         supports the .set micromips directive.
14946         * configure: Regenerate.
14947         * config.in: Regenerate.
14948         * config/mips/mips-tables.opt: Regenerate.
14949         * config/mips/micromips.md: New file.
14950         * constraints.md (ZC, AD): New constraints.
14951         * config/mips/predicates.md (movep_src_register): New predicate.
14952         (movep_src_operand): New predicate.
14953         (non_volatile_mem_operand): New predicate.
14954         * config/mips/mips.md (multimem): New type.
14955         (length): Differentiate between 17-bit and 18-bit branch offsets.
14956         (MOVEP1, MOVEP2): New mode iterator.
14957         (mov_<load>l): Use ZC constraint.
14958         (mov_<load>r): Likewise.
14959         (mov_<store>l): Likewise.
14960         (mov_<store>r): Likewise.
14961         (*branch_equality<mode>_inverted): Add microMIPS support.
14962         (*branch_equality<mode>): Likewise.
14963         (*jump_absolute): Likewise.
14964         (indirect_jump_<mode>): Likewise.
14965         (tablejump_<mode>): Likewise.
14966         (<optab>_internal): Likewise.
14967         (sibcall_internal): Likewise.
14968         (sibcall_value_internal): Likewise.
14969         (prefetch): Use constraint ZD.
14970         * config/mips/mips.opt (minterlink-compressed): New option.
14971         (minterlink-mips16): Now an alias for minterlink-compressed.
14972         (mmicromips): New option.
14973         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14974         (compare_and_swap_12): Likewise.
14975         (sync_add<mode>): Likewise.
14976         (sync_<optab>_12): Likewise.
14977         (sync_old_<optab>_12): Likewise.
14978         (sync_new_<optab>_12): Likewise.
14979         (sync_nand_12): Likewise.
14980         (sync_old_nand_12): Likewise.
14981         (sync_new_nand_12): Likewise.
14982         (sync_sub<mode>): Likewise.
14983         (sync_old_add<mode>): Likewise.
14984         (sync_old_sub<mode>): Likewise.
14985         (sync_new_add<mode>): Likewise.
14986         (sync_new_sub<mode>): Likewise.
14987         (sync_<optab><mode>): Likewise.
14988         (sync_old_<optab><mode>): Likewise.
14989         (sync_new_<optab><mode>): Likewise.
14990         (sync_nand<mode>): Likewise.
14991         (sync_old_nand<mode>): Likewise.
14992         (sync_new_nand<mode>): Likewise.
14993         (sync_lock_test_and_set<mode>): Likewise.
14994         (test_and_set_12): Likewise.
14995         (atomic_compare_and_swap<mode>): Likewise.
14996         (atomic_exchange<mode>_llsc): Likewise.
14997         (atomic_fetch_add<mode>_llsc): Likewise.
14998         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14999         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15000         (umips_save_restore_pattern_p): Likewise.
15001         (umips_load_store_pair_p): Likewise.
15002         (umips_output_load_store_pair): Likewise.
15003         (umips_movep_target_p): Likewise.
15004         (umips_12bit_offset_address_p): Likewise.
15005         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15006         (mips_base_mips16): Rename this...
15007         (mips_base_compression_flags): ...to this. Update all uses.
15008         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15009         (mips_mips16_decl_p): Delete.
15010         (mips_nomips16_decl_p): Delete.
15011         (mips_get_compress_on_flags): New function.
15012         (mips_get_compress_off_flags): New function.
15013         (mips_get_compress_mode): New function.
15014         (mips_get_compress_on_name): New function.
15015         (mips_get_compress_off_name): New function.
15016         (mips_insert_attributes): Support multiple compression types.
15017         (mips_merge_decl_attributes): Likewise.
15018         (umips_12bit_offset_address_p): New function.
15019         (mips_start_function_definition): Emit .set micromips directive.
15020         (mips_call_may_need_jalx_p): New function.
15021         (mips_function_ok_for_sibcall): Add microMIPS support.
15022         (mips_print_operand_punctuation): Support short delay slots and
15023         compact jumps.
15024         (umips_swm_mask, umips_swm_encoding): New.
15025         (umips_build_save_restore): New function.
15026         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15027         (was_mips16_p): Remove.
15028         (old_compression_mode): New.
15029         (mips_set_compression_mode): New function.
15030         (mips_set_current_function): Add microMIPS support.
15031         (mips_option_override): Likewise.
15032         (umips_save_restore_pattern_p): New function.
15033         (umips_output_save_restore): New function.
15034         (umips_load_store_pair_p_1): New function.
15035         (umips_load_store_pair_p): New function.
15036         (umips_output_load_store_pair_1): New function.
15037         (umips_output_load_store_pair): New function.
15038         (umips_movep_target_p) New function.
15039         (mips_prepare_pch_save): Add microMIPS support.
15040         * config/mips/mips.h (TARGET_COMPRESSION): New.
15041         (TARGET_CPU_CPP_BUILTINS): Update macro
15042         to use new compression flags and to support microMIPS.
15043         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15044         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15045         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15046         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15047         (ASM_SPEC): Support mmicromips and mno-micromips.
15048         (M16STORE_REG_P): New macro.
15049         (MIPS_CALL): Support TARGET_MICROMIPS.
15050         (MICROMIPS_J): New macro.
15051         (mips_base_mips16): Rename this...
15052         (mips_base_compression_flags): ...to this.
15053         (UMIPS_12BIT_OFFSET_P): New macro.
15054         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15055         (MULTILIB_DIRNAMES): Likewise.
15057 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15059         PR target/52555
15060         * target-globals.c (save_target_globals): For init_reg_sets and
15061         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15063 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15065         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15066         * config/aarch64/t-aarch64
15067         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15069 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15071         PR inline-asm/56148
15072         * lra-constraints.c (process_alt_operands): Reload operand
15073         conflicting with earlier clobber only if no more other conflicting
15074         operands.
15076 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15078         PR sanitizer/56393
15079         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15080         if not linking a shared library.
15082 2013-02-22  Seth LaForge  <sethml@google.com>
15084         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15086 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15088         * config/arm/arm.md (split for extendsidi): Update condition.
15089         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15090         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15091         (qhs_zextenddi_cstr): Likewise.
15093 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15095         PR middle-end/56420
15096         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15097         avoid signed wrapping.
15098         (expand_mult): Handle properly multiplication by
15099         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15100         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15101         in the compiler if coeff is HOST_WIDE_INT_MIN.
15102         (expand_divmod): Don't make ext_op1 static, change it's type to
15103         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15105         PR rtl-optimization/50339
15106         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15107         field.
15108         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15109         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15110         into splitting_ashiftrt field.
15111         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15112         ASHIFTRT.
15113         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15114         choices.
15116 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
15118         PR middle-end/56108
15119         * trans-mem.c (execute_tm_mark): Do not expand transactions that
15120         are sure to go irrevocable.
15122 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
15124         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15125         scalars are valid operands.
15127 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15129         PR tree-optimization/56310
15130         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15131         only matching indices and non-negative final offsets.
15132         (intersect_aggregates_with_edge): Pass src_idx to
15133         agg_replacements_to_vector.  Pass src_idx insstead of index to
15134         intersect_with_agg_replacements.
15136 2013-02-21  Martin Jambor  <mjambor@suse.cz>
15138         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15139         instead of hard-wired defaults.
15141 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
15143         * doc/invoke.texi (MIPS Options): Update documentation of the
15144         floating-point multiply-accumulate instruction restrictions.
15146 2013-02-21  Kostya Serebryany  <kcc@google.com>
15148         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15149         asan_shadow_offset on x86_64 linux.
15151 2013-02-21  Richard Biener  <rguenther@suse.de>
15153         PR tree-optimization/56415
15154         Revert
15155         2013-02-11  Richard Biener  <rguenther@suse.de>
15157         PR tree-optimization/56273
15158         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15159         first VRP run.
15161 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15163         PR bootstrap/56258
15164         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15165         instead of @itemx.
15167         PR inline-asm/56405
15168         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15169         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15171 2013-02-20  Jan Hubicka  <jh@suse.cz>
15173         PR tree-optimization/56265
15174         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15175         when target is referenced for first time.
15177 2013-02-20  Richard Biener  <rguenther@suse.de>
15179         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15180         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15181         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15182         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15183         not return anything.
15184         (rename_ssa_copies): Do not remove unused locals.
15185         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15186         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15187         * passes.c (execute_function_todo): Do not schedule unused locals
15188         removal if cleanup_tree_cfg did something.
15189         * tree-ssa-live.c (remove_unused_locals): Dump statistics
15190         about the number of removed locals.
15192 2013-02-20  Richard Biener  <rguenther@suse.de>
15194         PR tree-optimization/56398
15195         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15197 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15199         PR tree-optimization/55334
15200         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15201         restricted pointers to arrays.
15203 2013-02-20  Richard Biener  <rguenther@suse.de>
15204             Jakub Jelinek  <jakub@redhat.com>
15206         PR tree-optimization/56396
15207         * tree-ssa-ccp.c (n_const_val): New static variable.
15208         (get_value): Return NULL for SSA names we don't have a lattice
15209         entry for.
15210         (ccp_initialize): Initialize n_const_val.
15211         * tree-ssa-copy.c (n_copy_of): New static variable.
15212         (init_copy_prop): Initialize n_copy_of.
15213         (get_value): Return NULL_TREE for SSA names we don't have a
15214         lattice entry for.
15216 2013-02-20  Martin Jambor  <mjambor@suse.cz>
15218         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15220 2013-02-20  Richard Biener  <rguenther@suse.de>
15222         * genpreds.c (write_lookup_constraint): Do not compare first
15223         letter of the constraint again.
15225 2013-02-20  Richard Biener  <rguenther@suse.de>
15227         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15228         and ceil_log2.
15229         (get_use_iv_cost): Terminate hashtable walk when coming across
15230         an empty entry.
15232 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
15234         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15235         reassociation for avx2 targets.
15237 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15239         * config/microblaze/microblaze.c: microblaze_has_clz = 0
15240         Add version check for v8.10.a to enable microblaze_has_clz
15241         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15242         version and TARGET_PATTERN_COMPARE check
15243         * config/microblaze/microblaze.md: New clzsi2 instruction
15245 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15247         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15248         function before branching.
15250 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15252         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15253         DUMP_INSN_RTX_UID.
15254         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15256 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
15258         PR middle-end/55889
15259         * sel-sched.c: Include ira.h.
15260         (implicit_clobber_conflict_p): New function.
15261         (moveup_expr): Use it.
15262         * Makefile.in (sel-sched.o): Depend on ira.h.
15264 2013-02-19  Richard Biener  <rguenther@suse.de>
15266         PR tree-optimization/56384
15267         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15268         (vn_hash_type): Split out from ...
15269         (vn_hash_constant_with_type): ... here.
15270         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15271         (vn_phi_eq): Compare types from vn_phi_s structure.
15272         (vn_phi_lookup): Populate vn_phi_s type.
15273         (vn_phi_insert): Likewise.
15275 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
15277         PR tree-optimization/56350
15278         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15279         if haven't found reduction or nested cycle operand, rather than
15280         asserting we must find it.
15282         PR tree-optimization/56381
15283         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15284         to fold_build3.
15286 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
15287             Jakub Jelinek  <jakub@redhat.com>
15289         PR target/52555
15290         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15291         (swap_optab_enable): Same.
15292         (init_all_optabs): Use argument instead of global.
15293         * tree.h (struct tree_optimization_option): New field target_optabs.
15294         * expr.h (init_all_optabs): Add argument to prototype.
15295         (TREE_OPTIMIZATION_OPTABS): New.
15296         (save_optabs_if_changed): Protoize.
15297         * optabs.h: Declare this_fn_optabs.
15298         * optabs.c (save_optabs_if_changed): New.
15299         Declare this_fn_optabs.
15300         (init_optabs): Add argument to init_all_optabs() call.
15301         * function.c (invoke_set_current_function_hook): Handle per
15302         function optabs.
15303         * function.h (struct function): New field optabs.
15304         * config/mips/mips.c (mips_set_mips16_mode): Handle when
15305         optimization_current_node has changed.
15306         * target-globals.h (save_target_globals_default_opts): Protoize.
15307         * target-globals.c (save_target_globals_default_opts): New.
15309 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15311         PR target/56347
15312         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15313         registers %fr12 and %fr12R as call used.
15315         PR target/56214
15316         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15317         and HImode, require all displacements to be an integer multiple of
15318         their mode size.
15319         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15320         only allow QImode and HImode when reload is in progress and strict is
15321         true.  Likewise for symbolic addresses.  Use base14_operand to check
15322         displacements in REG+BASE addresses.
15324 2013-02-18  Richard Biener  <rguenther@suse.de>
15326         PR tree-optimization/56366
15327         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15328         handle sign-conversion of outer-loop initial induction value.
15330 2013-02-18  Richard Biener  <rguenther@suse.de>
15332         PR middle-end/56349
15333         * cfghooks.c (merge_blocks): If we merge a latch into another
15334         block adjust references to it.
15335         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15336         (verify_loop_structure): Verify that a recorded latch is in fact
15337         a latch.
15339 2013-02-18  Richard Biener  <rguenther@suse.de>
15341         PR tree-optimization/56321
15342         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15343         order SSA name release and virtual operand unlinking.
15345 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15347         * config/microblaze/microblaze.md (save_stack_block): Define.
15348         (restore_stack_block): Likewise.
15350 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15352         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15353         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15354         * config/microblaze/microblaze.c (microblaze_option_override):
15355         Bail out early for PIC modes when target does not support PIC.
15357 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15359         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15360         Replace with a microblaze version.
15361         (microblaze_trampoline_init): Adapt for microblaze.
15362         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15363         microblaze.
15365 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
15366             Dodji Seketeli  <dodji@redhat.com>
15368         PR asan/56330
15369         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15370         (instrument_mem_region_access): Do not forget to always put
15371         instrumentation of the of 'base' and 'base + len' in a "if (len !=
15372         0) statement, even for cases where either 'base' or 'base + len'
15373         are not instrumented -- because they have been previously
15374         instrumented.  Simplify the logic by putting all the statements
15375         instrument 'base + len' inside a sequence, and then insert that
15376         sequence right before the current insertion point.  Then, to
15377         instrument 'base + len', just get an iterator on that statement.
15378         And do not forget to update the pointer to iterator the function
15379         received as argument.
15381 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
15383         PR rtl-optimization/56348
15384         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15386 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
15388         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15389         (clean_graph_dump_file): Pass base to start_graph_dump.
15391 2013-02-14  Richard Henderson  <rth@redhat.com>
15393         PR target/55941
15394         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15396 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
15398         * collect2-aix.h: Define F_LOADONLY.
15400 2013-02-14  Richard Biener  <rguenther@suse.de>
15402         PR lto/50494
15403         * varasm.c (output_constant_def_1): Get the decl representing
15404         the constant as argument.
15405         (output_constant_def): Wrap output_constant_def_1.
15406         (make_decl_rtl): Use output_constant_def_1 with the decl
15407         representing the constant.
15408         (build_constant_desc): Optionally re-use a decl already
15409         representing the constant.
15410         (tree_output_constant_def): Adjust.
15412 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
15414         Fix an asan crash
15415         * asan.c (instrument_builtin_call):  Really put the length of the
15416         second source argument into src1_len.
15418 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15420         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15421         argument.  If it is false, don't create edge from then_bb to
15422         fallthru_bb.
15423         (insert_if_then_before_iter): Pass true to it.
15424         (build_check_stmt): Pass false to it.
15425         (transform_statements): Flush hash table only on extended basic
15426         block boundaries, rather than at the beginning of every bb.
15427         Don't flush hash table on nonfreeing_call_p calls.
15428         * tree-flow.h (nonfreeing_call_p): New prototype.
15429         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15431 2013-02-13  David S. Miller  <davem@davemloft.net>
15433         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15435 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
15437         PR target/56184
15438         * ira.c (max_regno_before_ira): Move from ...
15439         (ira): ... here.
15440         (fix_reg_equiv_init): Use max_regno_before_ira instead of
15441         vec_safe_length.
15443 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
15445         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15447 2013-02-13  Richard Biener  <rguenther@suse.de>
15449         PR lto/56295
15450         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15451         globals in MEM_REFs.
15453 2013-02-13  Richard Biener  <rguenther@suse.de>
15455         * loop-init.c (loop_optimizer_init): Clear loop state when
15456         re-initializing preserved loops.
15457         * loop-unswitch.c (unswitch_single_loop): Return whether
15458         we unswitched the loop.  Do not verify loop state here.
15459         (unswitch_loops): When we unswitched a loop discover new loops.
15461 2013-02-13  Kostya Serebryany  <kcc@google.com>
15463         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15464         on x86_64 linux.
15465         * sanitizer.def: Rename __asan_init to __asan_init_v1.
15467 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
15469         Avoid instrumenting duplicated memory access in the same basic block
15470         * Makefile.in (asan.o): Add new dependency on hash-table.h
15471         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15472         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15473         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15474         (free_mem_ref_resources, has_mem_ref_been_instrumented)
15475         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15476         (get_mem_ref_of_assignment): New functions.
15477         (get_mem_refs_of_builtin_call): Extract from
15478         instrument_builtin_call and tweak a little bit to make it fit with
15479         the new signature.
15480         (instrument_builtin_call): Use the new
15481         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
15482         of is_gimple_builtin_call.
15483         (instrument_derefs, instrument_mem_region_access): Insert the
15484         instrumented memory reference into the hash table.
15485         (maybe_instrument_assignment): Renamed instrument_assignment into
15486         this, and change it to advance the iterator when instrumentation
15487         actually happened and return true in that case.  This makes it
15488         homogeneous with maybe_instrument_assignment, and thus give a
15489         chance to callers to be more 'regular'.
15490         (transform_statements): Clear the memory reference hash table
15491         whenever we enter a new BB, when we cross a function call, or when
15492         we are done transforming statements.  Use
15493         maybe_instrument_assignment instead of instrumentation.  No more
15494         need to special case maybe_instrument_assignment and advance the
15495         iterator after calling it; it's now handled just like
15496         maybe_instrument_call.  Update comment.
15498 2013-02-13  Richard Biener  <rguenther@suse.de>
15500         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15501         Fix loop discovery code.
15503 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
15505         PR inline-asm/56148
15506         * lra-constraints.c (process_alt_operands): Match early clobber
15507         operand with itself.  Check conflicts with earlyclobber only if
15508         the operand is not reloaded.  Prefer to reload conflicting operand
15509         if earlyclobber and matching operands are the same.
15511 2013-02-12  Richard Biener  <rguenther@suse.de>
15513         PR lto/56297
15514         * lto-streamer-out.c (write_symbol): Do not output symbols
15515         for hard register variables.
15517 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
15519         PR target/54222
15520         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15521         (umulsidi3_insn, mulsidi3_insn): New insns.
15523 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
15525         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15526         (struct tune_params): Add vec_costs field.
15527         * config/arm/arm.c (arm_builtin_vectorization_cost)
15528         (arm_add_stmt_cost): New functions.
15529         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15530         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15531         (arm_default_vec_cost): New struct of type cpu_vec_costs.
15532         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15533         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15534         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15535         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15537 2013-02-12  Richard Biener  <rguenther@suse.de>
15539         PR lto/56295
15540         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15541         decls again if possible.
15543 2013-02-12  Richard Biener  <rguenther@suse.de>
15545         PR middle-end/56288
15546         * tree-ssa.c (verify_ssa_name): Fix check, move
15547         SSA_NAME_IN_FREE_LIST check up.
15549 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
15550             Steven Bosscher   <steven@gcc.gnu.org>
15552         PR rtl-optimization/56151
15553         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15554         equal to op0 or op1, and last_insn pattern is CODE operation
15555         with MEM dest and one of the operands matches that MEM.
15557 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
15559         * doc/extend.texi: Document Function Multiversioning and "default"
15560         parameter string to target attribute.
15561         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15562         target attribute parameter is "default".
15563         (ix86_compare_version_priority): Remove checks for target attribute.
15564         (ix86_mangle_function_version_assembler_name): Change error to sorry.
15565         Remove check for target attribute equal to NULL. Add assert.
15566         (ix86_generate_version_dispatcher_body): Change error to sorry.
15568 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
15569             Jack Howarth  <howarth@bromo.med.uc.edu>
15570             Patrick Marlier  <patrick.marlier@gmail.com>
15572         PR libitm/55693
15573         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15574         define ENDFILE_SPEC as TM_DESTRUCTOR.
15575         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15577 2013-02-11  Alexander Potapenko  <glider@google.com>
15578             Jack Howarth  <howarth@bromo.med.uc.edu>
15579             Jakub Jelinek  <jakub@redhat.com>
15581         PR sanitizer/55617
15582         * config/darwin.c (cdtor_record): Rename ctor_record.
15583         (sort_cdtor_records): Rename sort_ctor_records.
15584         (finalize_dtors): New routine to sort destructors by
15585         priority before use in assemble_integer.
15586         (machopic_asm_out_destructor): Use finalize_dtors if needed.
15588 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
15590         PR rtl-optimization/56275
15591         * simplify-rtx.c (avoid_constant_pool_reference): Check that
15592         offset is non-negative and less than cmode size before
15593         calling simplify_subreg.
15595 2013-02-11  Richard Biener  <rguenther@suse.de>
15597         PR tree-optimization/56264
15598         * cfgloop.h (fix_loop_structure): Adjust prototype.
15599         * loop-init.c (fix_loop_structure): Return the number of
15600         newly discovered loops.
15601         * tree-cfgcleanup.c (repair_loop_structures): When new loops
15602         are discovered, do a full loop-closed SSA rewrite.
15604 2013-02-11  Richard Biener  <rguenther@suse.de>
15606         PR tree-optimization/56273
15607         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15608         first VRP run.
15609         (check_array_ref): Fix missing newline in dumps.
15610         (search_for_addr_array): Likewise.
15612 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
15614         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15616 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
15618         PR target/56256
15619         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15621 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
15623         PR rtl-optimization/56246
15624         * lra-constraints.c (simplify_operand_subreg): Try to reuse
15625         reload pseudo.
15626         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15627         constraints are satisfied.
15629 2013-02-08  Jeff Law  <law@redhat.com>
15631         PR debug/53948
15632         * emit-rtl.c (reg_is_parm_p): New function.
15633         * regs.h (reg_is_parm_p): New prototype.
15634         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15635         callee-clobbered registers.
15637 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
15639         PR target/56043
15640         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15641         If there is no implicit builtin declaration, just return NULL.
15643 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15645         * config/i386/sse.md (FMAMODEM): New mode iterator.
15646         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15647         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15649 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15651         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15652         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15653         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15655 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15657         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15658         (microblaze*-*-elf): Likewise.
15659         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15660         LINK_SPEC.
15661         * config/microblaze/microblaze-c.c: Add builtin defines for
15662         _LITTLE_ENDIAN and _BIG_ENDIAN.
15663         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15664         add to TARGET_DEFAULT flags.
15665         Expand ASM_SPEC and LINK_SPEC.
15666         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15667         * config/microblaze/microblaze.md: Update extendsidi2 and
15668         movdi_internal instructions to use low-order / high-order reg
15669         print_operands.
15670         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15671         options and inversemask / mask of LITTLE_ENDIAN.
15672         * config/microblaze/t-microblaze: Expand multilib options to
15673         include mlittle-endian (le) and update exceptions patterns.
15675 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
15677         PR rtl-optimization/56195
15678         * lra-constraints.c (get_reload_reg): Don't reuse regs
15679         if they have smaller mode than requested, if they have
15680         wider mode than requested, try to return a SUBREG.
15682         PR tree-optimization/56250
15683         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15684         if type is unsigned and code isn't MULT_EXPR.
15686 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15688         PR tree-optimization/56064
15689         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15690         bits according to mode.
15691         * fixed-value.h (fixed_from_double_int)
15692         (const_fixed_from_double_int): Adjust comments.
15694 2013-02-08  Richard Biener  <rguenther@suse.de>
15696         PR lto/56231
15697         * lto-streamer.h (struct data_in): Remove current_file, current_line
15698         and current_col members.
15699         * lto-streamer-out.c (lto_output_location): Stream changed bits
15700         en-block for efficiency.
15701         * lto-streamer-in.c (clear_line_info): Remove.
15702         (lto_input_location): Cache current file, line and column
15703         globally via local statics.  Read changed bits en-block.
15704         (input_function): Do not call clear_line_info.
15705         (lto_read_body): Likewise.
15706         (lto_input_toplevel_asms): Likewise.
15708 2013-02-08  Michael Matz  <matz@suse.de>
15710         PR tree-optimization/52448
15711         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15712         (nt_call_phase): New static.
15713         (add_or_mark_expr): Only mark accesses with newer phase than any
15714         call seen.
15715         (nonfreeing_call_p): New.
15716         (nt_init_block): Update nt_call_phase, mark blocks as visited.
15717         (nt_fini_block): Keep blocks marked as visited.
15718         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15720 2013-02-08  Richard Biener  <rguenther@suse.de>
15722         * ira.c (ira): Free broken dominator information.
15724 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
15726         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15728 2013-02-08  Marek Polacek  <polacek@redhat.com>
15730         * cfgloop.c (verify_loop_structure): Add more checking of headers.
15732 2013-02-08  Richard Biener  <rguenther@suse.de>
15734         PR middle-end/56181
15735         * cfgloop.h (flow_loops_find): Adjust.
15736         (bb_loop_header_p): Declare.
15737         * cfgloop.c (bb_loop_header_p): New function split out from ...
15738         (flow_loops_find): ... here.  Adjust function signature,
15739         support incremental loop structure update.
15740         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
15741         * cfgloopmanip.c (fix_loop_structure): Move ...
15742         * loop-init.c (fix_loop_structure): ... here.
15743         (apply_loop_flags): Split out from ...
15744         (loop_optimizer_init): ... here.
15745         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
15746         in incremental mode, only remove dead loops here.
15748 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
15750         PR target/54222
15751         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15752         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15753         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15754         (*round<mode>3.libgcc): New insns for fixed-modes.
15755         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15756         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15757         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15758         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15759         implementations.  Define to __builtin_avr_absFX,
15760         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15761         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15762         __builtin_avr_countlsFX, respectively.
15763         * config/avr/avr-c.c (target.h): Include it.
15764         (enum avr_builtin_id): New enum.
15765         (avr_resolve_overloaded_builtin): New static function.
15766         (avr_register_target_pragmas): Use it to set
15767         targetm.resolve_overloaded_builtin.
15768         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15769         tree nodes used by DEF_BUILTIN.
15770         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15771         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15772         <AVR_BUILTIN_xxBITS>: Same.
15774 2013-02-08  Richard Biener  <rguenther@suse.de>
15776         * cfgloop.c (verify_loop_structure): Properly handle
15777         a loop exiting to another loop header.
15778         * ira-int.h (ira_loops): Remove.
15779         * ira.c (ira_loops): Remove.
15780         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15781         (do_reload): Use loop_optimizer_finalize.
15782         * ira-build.c (create_loop_tree_nodes): Use get_loops and
15783         number_of_loops to access the loop tree.
15784         (more_one_region_p): Likewise.
15785         (finish_loop_tree_nodes): Likewise.
15786         (rebuild_regno_allocno_maps): Likewise.
15787         (mark_loops_for_removal): Likewise.
15788         (mark_all_loops_for_removal): Likewise.
15789         (remove_unnecessary_regions): Likewise.
15790         (ira_build): Likewise.
15791         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15793 2013-02-08  Richard Biener  <rguenther@suse.de>
15795         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15796         * ipa-pure-const.c (analyze_function): Avoid calling
15797         mark_irreducible_loops twice.
15798         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
15800 2013-02-07  David S. Miller  <davem@davemloft.net>
15802         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15803         on 'reg'.
15804         * var-tracking.c (vt_add_function_parameter): Test the presence of
15805         HAVE_window_save properly and do not remap argument registers when
15806         we have a leaf function.
15808 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
15810         PR bootstrap/56227
15811         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15812         instead of "ll".
15813         * config/i386/i386.c (ix86_print_operand): Ditto.
15815 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15817         * lra-constraints.c (process_alt_operands): Fix recently added comment.
15819 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
15821         PR rtl-optimization/56225
15822         * lra-constraints.c (process_alt_operands): Check that reload hard
15823         reg can hold value for strict_low_part.
15825 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15827         PR debug/56154
15828         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15829         dwarf2out_end_function.
15830         (in_first_function_p, maybe_at_text_label_p,
15831         first_loclabel_num_not_at_text_label): New variables.
15832         (dwarf2out_var_location): In the first function find out
15833         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
15834         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15835         functions.
15837 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
15839         PR rtl-optimization/56178
15840         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15841         SUBREG of a register.  Tidy up related block of code.
15842         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15843         note if the source is a register or a SUBREG of a register.
15845 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
15847         PR target/56228
15848         * config/rs6000/rs6000.md (ptrm): New mode attr.
15849         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
15850         call_value_indirect_aix<pttrsize>,
15851         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
15852         m in constraints.
15854 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
15856         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
15857         if -bnortl. Convert to strcmp and strncmp.
15859 2013-02-07  Alan Modra  <amodra@gmail.com>
15861         PR target/54009
15862         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
15863         addresses won't wrap when offsetting.
15864         (rs6000_secondary_reload): Provide secondary reloads needed for
15865         wrapping LO_SUM addresses.
15867 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
15869         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
15870         MACH, just __MACH__.
15872 2013-02-06  Richard Biener  <rguenther@suse.de>
15874         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
15875         instead of calling fix_loop_structure.
15877 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
15879         PR middle-end/56217
15880         * omp-low.c (use_pointer_for_field): Return false if
15881         lower_send_shared_vars doesn't generate any copy-out code.
15883 2013-02-06  Tom de Vries  <tom@codesourcery.com>
15885         PR rtl-optimization/56131
15886         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
15887         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
15888         of the label is NULL.  Add comment.
15890 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
15892         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
15894         PR sanitizer/55374
15895         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
15896         (STATIC_LIBTSAN_LIBS): Likewise.
15897         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
15898         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
15899         is defined, don't add anything else beyond that.
15900         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
15901         (LINK_COMMAND_SPEC): Use them.
15903         PR tree-optimization/56205
15904         * tree-stdarg.c (check_all_va_list_escapes): Return true if
15905         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
15906         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
15908 2013-02-05  Richard Biener  <rguenther@suse.de>
15910         PR tree-optimization/53342
15911         PR tree-optimization/53185
15912         * tree-vectorizer.h (vect_check_strided_load): Remove.
15913         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
15914         not disallow peeling for vectorized strided loads.
15915         (vect_check_strided_load): Make static and simplify.
15916         (vect_analyze_data_refs): Adjust.
15917         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
15918         correctly when vectorizing strided loads.
15920 2013-02-05  Richard Biener  <rguenther@suse.de>
15922         * doc/install.texi: Refer to ISL, not PPL.
15924 2013-02-05  Jan Hubicka  <jh@suse.cz>
15926         PR tree-optimization/55789
15927         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
15929 2013-02-05  Jan Hubicka  <jh@suse.cz>
15931         PR tree-optimization/55789
15932         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
15933         the dead call anyway.
15935 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
15937         PR sanitizer/55374
15938         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
15940 2013-02-04  Alexander Potapenko  <glider@google.com>
15941             Jack Howarth  <howarth@bromo.med.uc.edu>
15942             Jakub Jelinek  <jakub@redhat.com>
15944         PR sanitizer/55617
15945         * config/darwin.c (sort_ctor_records): Stabilized qsort
15946         on constructor priority by using original position.
15947         (finalize_ctors): New routine to sort constructors by
15948         priority before use in assemble_integer.
15949         (machopic_asm_out_constructor): Use finalize_ctors if needed.
15951 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
15953         PR libstdc++/54314
15954         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
15955         about visibility on artificial decls.
15956         * config/sol2.c (solaris_assemble_visibility): Likewise.
15958 2013-02-04  Kai Tietz  <ktietz@redhat.com>
15960         PR target/56186
15961         * config/i386/i386.c (function_value_ms_64): Add additional valtype
15962         argument and improve checking of return-argument types for 16-byte
15963         modes.
15964         (ix86_function_value_1): Add additional valtype argument on call
15965         of function_value_64.
15966         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
15967         handling infunction_value_64 function.
15969 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
15971         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
15973 2013-02-04  Richard Biener  <rguenther@suse.de>
15975         PR tree-optimization/56188
15976         * tree-ssa-structalias.c (label_visit): Consider case with
15977         initially non-empty points-to set.
15978         (perform_var_substitution): Dump node mapping and clean up.
15980 2013-02-04  Richard Guenther  <rguenther@suse.de>
15982         PR lto/56168
15983         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
15984         node prevail as last resort.
15985         (lto_symtab_merge_decls): Remove guard on LTRANS here.
15986         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
15988 2013-02-04  Richard Biener  <rguenther@suse.de>
15990         PR tree-optimization/56113
15991         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
15992         Merge into ...
15993         (equiv_class_lookup_or_add): ... this.
15994         (label_visit): Adjust and fix error in previous patch.
15995         (perform_var_substitution): Adjust.
15997 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
15999         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16000         * config/sh/sh4-300.md: Likewise.
16001         * config/sh/sh4a.md: Likewise.
16002         * config/sh/constraints.md: Likewise.
16003         * config/sh/sh.md: Likewise.
16004         * config/sh/netbsd-elf.h: Likewise.
16005         * config/sh/predicates.md: Likewise.
16006         * config/sh/sh-protos.h: Likewise.
16007         * config/sh/ushmedia.h: Likewise.
16008         * config/sh/linux.h: Likewise.
16009         * config/sh/sh.c: Likewise.
16010         * config/sh/superh.h: Likewise.
16011         * config/sh/elf.h: Likewise.
16012         * config/sh/sh4.md: Likewise.
16013         * config/sh/sh.h: Likewise.
16015 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16017         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16018         constraint to match_test floating_point_store_memory_operand().
16019         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16020         (base14_operand): New.
16021         (floating_point_store_memory_operand): New.
16022         (integer_store_memory_operand): Revise to use base14_operand and
16023         reg_plus_base_memory_operand.
16024         (move_dest_operand): Allow symbolic_memory_operands.
16025         (symbolic_memory_operand): Check for LO_SOM.
16026         (symbolic_operand): Change default case to break.
16027         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16028         CONST_DOUBLE values to be reloaded by putting them into memory when
16029         the destination is a floating point register.
16030         (movdf): Remove code to handle CONST_DOUBLE.
16031         (movsf): Likewise.
16032         (reload_indf_r1): New.
16033         (reload_insf_r1): New.
16034         Consistently use "Q" and "T" constraints with integer and floating
16035         point move instructions, respectively.
16036         (movdi): Remove FAIL.
16037         Change predicate for source operand unamed DImode move from
16038         general_operand to move_src_operand.
16039         (umulsidi3): Change predicate for destination operand to
16040         register_operand.
16041         Likewise for similar unamed patterns.
16042         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16043         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16044         (hppa_legitimize_address): Simplify mask calculation.
16045         (pa_emit_move_sequence): Revised handling of secondary reloads from
16046         REG+D addresses for floating point loads and stores.  Directly handle
16047         loading CONST0_RTX (mode) to a floating point register.
16048         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16049         to floating point registers.  Don't restrict secondary reloads to
16050         floating point registers to integer modes.  Revise some comments and
16051         cleanup some code.
16052         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16053         (pa_legitimate_address_p): New.
16054         (pa_legitimize_reload_address): New.
16055         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16056         (STRICT_REG_OK_FOR_BASE_P): New.
16057         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16058         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16060 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16061             Andrew Dixie  <andrewd@gentrack.com>
16063         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16064         flag set.
16066 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16068         * expmed.c (extract_bit_field_1): Pass the full width of the
16069         structure to get_best_reg_extraction_insn.
16071 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16073         PR target/54601
16074         * configure.ac (use_cxa_atexit): Add AIX.
16075         * configure: Regenerate.
16077         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16079 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16081         PR debug/54793
16082         * final.c (need_profile_function): New variable.
16083         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16084         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16085         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16086         notes, targetm.asm_out.function_prologue doesn't emit anything,
16087         HAVE_prologue and profiler should be emitted before prologue,
16088         set need_profile_function instead of emitting it.
16089         (final_scan_insn): If need_profile_function, emit
16090         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16091         NOTE_INSN_FUNCTION_BEG note.
16093 2013-02-01  Richard Henderson  <rth@redhat.com>
16095         * config/rs6000/rs6000.md (smulditi3): New.
16096         (umulditi3): New.
16098         * config/alpha/alpha.md (umulditi3): New.
16100 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16102         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16103         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16105 2013-02-01  Richard Biener  <rguenther@suse.de>
16107         PR tree-optimization/56113
16108         * tree-ssa-structalias.c (label_visit): Reduce work for
16109         single-predecessor nodes.
16111 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
16113         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16114         range isn't testing for zero.
16116 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
16118         PR middle-end/56113
16119         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16121 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
16122             Nick Clifton  <nickc@redhat.com>
16124         * config/v850/constraints.md (Q): Define as a memory constraint.
16125         * config/v850/predicates.md (label_ref_operand): New predicate.
16126         (e3v5_shift_operand): New predicate.
16127         (ior_operator): New predicate.
16128         * config/v850/t-v850: Add e3v5 multilib.
16129         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16130         (v850_gen_movdi): Prototype.
16131         * config/v850/v850.c: Add support for e3v5 architecture.
16132         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16133         TARGET_V850E_UP.
16134         (construct_save_jarl): Add e3v5 long JARL support.
16135         (v850_adjust_insn_length): New function.  Adjust length of call
16136         insns when using e3v5 instructions.
16137         (v850_gen_movdi): New function: Generate instructions to move a
16138         DImode value.
16139         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16140         (CPP_SPEC): Define __v850e3v5__ as appropriate.
16141         (TARGET_USE_FPU): Enable for e3v5.
16142         (CONST_OK_FOR_W): New macro.
16143         (ADJUST_INSN_LENGTH): Define.
16144         * config/v850/v850.md (UNSPEC_LOOP): Define.
16145         (attr cpu): Add v850e3v5.
16146         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16147         (movdi): New pattern.
16148         (movdi_internal): New pattern.
16149         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16150         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16151         (cstoresf4): Likewise.
16152         (cstoredf4): Likewise.
16153         (insv): New pattern.
16154         (rotlso3_a): New pattern.
16155         (rotlsi3_b): New pattern
16156         (rotlsi3_v850e3v5): New pattern.
16157         (doloop_begin): New pattern.
16158         (fix_loop_counter): New pattern.
16159         (doloop_end): New pattern.
16160         (branch_normal): Add e3v5 long branch support.
16161         (branch_invert): Likewise.
16162         (branch_z_normal): Likewise.
16163         (branch_z_invert): Likewise.
16164         (branch_nz_normal): Likewise.
16165         (branch_nz_invert): Likewise.
16166         (call_internal_short): Add e3v5 register-indirect JARL support.
16167         (call_internal_long): Likewise.
16168         (call_value_internal_short): Likewise.
16169         (call_value_internal_long): Likewise.
16170         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16171         (mloop): New option.
16172         * config.gcc: Add support for configuring v840e3v5 target.
16173         * doc/invoke.texi: Document new v850 specific command line options.
16175 2013-01-31  Paul Koning  <ni1d@arrl.net>
16177         PR debug/55059
16178         PR debug/54508
16179         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16180         children if parent is a class.
16181         (prune_unused_types_prune): Don't add DW_AT_declaration.
16183 2013-01-31  Richard Biener  <rguenther@suse.de>
16185         PR tree-optimization/56157
16186         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16187         match up operand with SLP child.
16189 2013-01-31  Jason Merrill  <jason@redhat.com>
16191         PR debug/54410
16192         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16193         parameters the first time.
16194         (gen_scheduled_generic_parms_dies): Check completeness here.
16196 2013-01-31  Richard Biener  <rguenther@suse.de>
16198         PR middle-end/53073
16199         * common.opt (faggressive-loop-optimizations): New flag,
16200         enabled by default.
16201         * doc/invoke.texi (faggressive-loop-optimizations): Document.
16202         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16203         infer_loop_bounds_from_undefined by it.
16205 2013-01-31  Richard Biener  <rguenther@suse.de>
16207         PR tree-optimization/56150
16208         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16209         visit virtual operands.
16210         (find_uses_to_rename_bb): Likewise.
16212 2013-01-31  Richard Biener  <rguenther@suse.de>
16214         PR tree-optimization/56150
16215         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16216         mixed store non-store stmts.
16218 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
16220         PR sanitizer/55374
16221         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16222         LIBASAN_EARLY_SPEC is defined.
16223         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16224         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16225         before %o.
16226         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16228         PR c++/55742
16229         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16230         invalid args instead of ICEing on it.
16231         (ix86_valid_target_attribute_tree): Return error_mark_node if
16232         ix86_valid_target_attribute_inner_p failed.
16233         (ix86_valid_target_attribute_p): Return false only if
16234         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
16235         target("default") attribute.
16236         (sorted_attr_string): Change argument from const char * to tree,
16237         merge in all target attribute arguments rather than just one.
16238         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
16239         instead of free.  Avoid using strcat.
16240         (ix86_mangle_function_version_assembler_name): Mangle
16241         target("default") as if no target attribute is present.  Adjust
16242         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
16243         instead of xmalloc and XDELETEVEC instead of free.
16244         (ix86_function_versions): Don't return true if one of the decls
16245         doesn't have target attribute.  If they don't and one of the decls
16246         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
16247         sorted_attr_string caller.  Use XDELETEVEC instead of free.
16248         (ix86_supports_function_versions): Remove.
16249         (make_name): Fix up formatting.
16250         (make_dispatcher_decl): Remove resolver_name and its initialization.
16251         Avoid leaking memory.
16252         (is_function_default_version): Return true if there is
16253         target("default") attribute rather than no target attribute at all.
16254         (make_resolver_func): Avoid leaking memory.
16255         (ix86_generate_version_dispatcher_body): Likewise.
16256         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16257         * target.def (supports_function_versions): Remove.
16258         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16259         * doc/tm.texi: Regenerated.
16261 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
16263         PR rtl-optimization/56144
16264         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16265         for values with side effects.
16267 2013-01-30  Richard Biener  <rguenther@suse.de>
16269         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16270         (sparseset_pop): Likewise.
16271         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
16272         to be able to use quick_push in the worker loop.
16274 2013-01-30  Marek Polacek  <polacek@redhat.com>
16276         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16278 2013-01-30  Richard Biener  <rguenther@suse.de>
16280         PR lto/56147
16281         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16283 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
16285         PR tree-optimization/56064
16286         * fixed-value.c (fixed_from_double_int): New function.
16287         * fixed-value.h (fixed_from_double_int): New prototype.
16288         (const_fixed_from_double_int): New static inline function.
16289         * fold-const.c (native_interpret_fixed): New static function.
16290         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16291         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16292         (native_encode_fixed): New static function.
16293         (native_encode_expr) <FIXED_CST>: Use it.
16294         (native_interpret_int): Move double_int worker code to...
16295         * double-int.c (double_int::from_buffer): ...this new static method.
16296         * double-int.h (double_int::from_buffer): Prototype it.
16298 2013-01-30  Richard Biener  <rguenther@suse.de>
16300         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16301         New pointer-map and obstack.
16302         (init_alias_vars): Allocate pointer-map and obstack.
16303         (delete_points_to_sets): Free them.
16304         (find_what_var_points_to): Cache result.
16305         (find_what_p_points_to): Adjust for changed interface of
16306         find_what_var_points_to.
16307         (compute_points_to_sets): Likewise.
16308         (ipa_pta_execute): Likewise.
16310 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16312         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16313         * configure: Regenerate.
16314         * config.in: Regenerate.
16315         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16316         #nobits/#progbits if supported.
16318 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
16320         PR target/56121
16321         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16322         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16323         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16325 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16327         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16328         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16330 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16332         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16333         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16335 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16337         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16338         declaration.
16339         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16340         * config/arm/cortex-a7.md: New bypasses using
16341         arm_mac_accumulator_is_result.
16343 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16345         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
16346         (cortex_a7_neon_mla): Likewise.
16347         (cortex_a7_fpfmad): New reservation.
16348         (cortex_a7_fpmacs): Use ffmas and update required units.
16349         (cortex_a7_fpmuld): Update required units and latency.
16350         (cortex_a7_fpmacd): Likewise.
16351         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16352         (cortex_a7_neon). Likewise.
16353         (bypass) Update participating units.
16355 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
16357         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16358         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16359         from fmac to ffma.
16360         * config/arm/vfp11.md (vfp_farith): Use ffmas.
16361         (vfp_fmul): Use ffmad.
16362         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16363         (cortex_r4_fmacd): Use ffmad.
16364         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16365         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
16366         (cortex_a9_fmacd): Use ffmad.
16367         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16368         (cortex_a8_vfp_macd): Use ffmad.
16369         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16370         (cortex_a5_fpmacd): Use ffmad.
16371         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16372         (cortex_a15_vfp_macd): Use ffmad.
16373         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16375 2013-01-29  Jason Merrill  <jason@redhat.com>
16377         PR libstdc++/54314
16378         * varasm.c (default_assemble_visibility): Don't warn about
16379         visibility on artificial decls.
16381 2013-01-29  Richard Biener  <rguenther@suse.de>
16383         PR tree-optimization/56113
16384         * tree-ssa-structalias.c (equiv_class_lookup): Also return
16385         the bitmap leader.
16386         (label_visit): Free duplicate bitmaps and record the leader instead.
16387         (perform_var_substitution): Adjust.
16389 2013-01-29  Richard Biener  <rguenther@suse.de>
16391         PR tree-optimization/55270
16392         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16393         the CFG, schedule loops for fixup.
16395 2013-01-29  Nick Clifton  <nickc@redhat.com>
16397         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16398         SP_REG.
16400 2013-01-28  Leif Ekblad  <leif@rdos.net>
16402         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16403         * config/i386/i386.h (TARGET_RDOS): New macro.
16404         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16405         * config/i386/i386.c (ix86_option_override_internal): For 64bit
16406         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16407         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16408         DEFAULT_LARGE_SECTION_THRESHOLD.
16409         * config/i386/i386.md (R14_REG, R15_REG): New constants.
16410         * config/i386/rdos.h: New file.
16411         * config/i386/rdos64.h: New file.
16413 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
16415         PR other/54814
16416         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16417         TEST_HARD_REG_BIT.
16419 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16421         PR rtl-optimization/56117
16422         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16423         call cselib_lookup_from_insn on the MEM before calling
16424         add_insn_mem_dependence.
16426 2013-01-28  Richard Biener  <rguenther@suse.de>
16428         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16429         to a stmt that didn't have one.
16430         (copy_phis_for_bb): Likewise for PHI arguments.
16431         (copy_debug_stmt): Likewise for debug stmts.
16433 2013-01-28  Richard Biener  <rguenther@suse.de>
16435         PR tree-optimization/56034
16436         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16437         (partition_builtin_p): Adjust.
16438         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
16439         it is the last partition.
16440         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16441         up the vertex for the definition.
16442         (classify_partition): Classify whether a partition is a
16443         PKIND_REDUCTION, thus has uses outside of the loop.
16444         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16445         Merge all PKIND_REDUCTION partitions into the last partition.
16446         (tree_loop_distribution): Seed partitions from reductions as well.
16448 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
16450         PR tree-optimization/56125
16451         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16452         pow(x,c) into sqrt(x) * powi(x, n/2) or
16453         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16454         optimizing for size.
16455         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16456         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16457         integer.
16459         PR tree-optimization/56094
16460         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16461         to UNKNOWN_LOCATION while gimplifying expr.
16463 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
16465         PR target/56114
16466         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16467         operand 0 in movabs insn template for -masm=intel asm alternative.
16468         (*movabs<mode>_2): Ditto for operand 1.
16470 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
16472         PR target/54663
16473         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16474         of microblaze-c.o
16476 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16478         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16479         tm_file.
16481 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16483         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16484         Undef to avoid warning.
16486 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16488         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16489         * configure: Regenerate.
16491 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
16493         PR tree-optimization/56098
16494         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16495         for stmts with volatile ops.
16496         (cond_store_replacement): Don't optimize if assign has volatile ops.
16497         (cond_if_else_store_replacement_1): Don't optimize if either
16498         then_assign or else_assign have volatile ops.
16499         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16500         volatile ops.
16502 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16504         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16506 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16508         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16509         missing ':' in asm example.
16511 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
16513         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16514         entries into lane and laneq entries.
16515         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16516         Remove AdvSIMD scalar modes.
16517         (aarch64_sq<r>dmulh_laneq<mode>): New.
16518         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16519         modes.
16520         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16521         builtin implementations to relfect changes in RTL in aarch64-simd.md.
16522         * config/aarch64/iterators.md (VCOND): New.
16523         (VCONQ): New.
16525 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
16527         PR target/54222
16528         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16529         Add NULL LIBNAME argument to existing definitions.
16530         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16531         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16532         * config/avr/avr.c (DEF_BUILTIN): Same.
16533         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16534         (avr_expand_builtin): Expand to a vanilla call if a libgcc
16535         implementation is available (DECL_ASSEMBLER_NAME is set).
16536         (avr_fold_absfx): New static function.
16537         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16538         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16539         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16540         AVR_BUILTIN_ABSLLK.
16541         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16542         (abshk, absk, abslk, absllk): Provide as static inline functions.
16544 2013-01-25  Marek Polacek  <polacek@redhat.com>
16546         PR tree-optimization/56035
16547         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16549 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
16551         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16552         (*movtf_internal_rex64): Add (!o,C) alternative
16553         (*movxf_internal_rex64): Ditto.
16554         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16556 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
16558         * doc/invoke.texi: fix typo.
16559         * doc/objc.texi: fix typo.
16561 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
16563         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16564         for the first two alternatives.
16566 2013-01-24  Diego Novillo  <dnovillo@google.com>
16568         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
16569         (ggc-zone.o): Remove.
16570         * configure.ac: Remove option --with-gc.
16571         * configure: Re-generate.
16572         * doc/install.texi: Remove documentation for --with-gc.
16573         * gengtype.c (write_enum_defn): Remove.  Update all users.
16574         (write_Types_process_field): Remove generation of gt_e_* argument.
16575         (output_type_enum): Remove.  Update all users.
16576         (write_enum_defn): Remove.  Update all users.
16577         (enum alloc_zone): Remove.  Update all users.
16578         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16579         * ggc-common.c (ggc_splay_alloc): Remove first argument.
16580         Update all callers.
16581         (struct ptr_data): Remove field TYPE.  Update all users.
16582         (gt_pch_note_object): Remove argument TYPE.  Update all users.
16583         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16584         Update all users.
16585         * ggc-none.c (ggc_alloc_typed_stat): Remove.
16586         (struct alloc_zone): Remove.
16587         (ggc_internal_alloc_zone_stat): Remove.
16588         (ggc_internal_cleared_alloc_zone_stat): Remove.
16589         * ggc-page.c (ggc_alloc_typed_stat): Remove.
16590         (ggc_pch_count_object): Remove last argument.  Update all users.
16591         (ggc_pch_alloc_object): Remove last argument.  Update all users.
16592         (struct alloc_zone): Remove.
16593         * ggc-zone.c: Remove.
16594         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
16595         (struct alloc_zone): Remove.
16596         (ggc_alloc_typed_stat): Remove.
16597         (ggc_alloc_typed): Remove.
16598         (ggc_splay_alloc): Remove first argument.
16599         (rtl_zone): Remove.  Update all users.
16600         (tree_zone): Remove.  Update all users.
16601         (tree_id_zone): Remove.  Update all users.
16602         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
16603         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
16604         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
16605         * tree-ssanames.c: Remove references to zone allocator in comments.
16607 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
16609         * config/avr/avr.c (avr_out_fract): Make register numbers that
16610         might be outside of source operand signed.
16612 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
16614         * config/i386/constraints.md (Yf): New constraint.
16615         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16616         of f constraint to conditionaly disable x87 register preferences.
16617         (*movdf_internal): Ditto.
16618         (*movsf_internal): Ditto.
16620 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
16622         PR inline-asm/55934
16623         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16624         that have operands with impossible constraints.
16625         Add a FIXME for a speed-up opportunity.
16626         * lra-constraints.c (process_alt_operands): Verify that a class
16627         selected from constraints on asms is valid for the operand mode.
16628         (curr_insn_transform): Remove incorrect comment.
16630 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
16632         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16633         TOC operand is a valid symbol ref in the constant pool.
16635 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16637         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
16639 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
16641         PR target/54222
16642         * config/avr/stdfix.h: New file.
16643         * t-avr (stdfix-gcc.h): New rule to build it.
16644         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16646 2013-01-23  Kostya Serebryany  <kcc@google.com>
16648         * config/darwin.h: remove dependency on
16649         CoreFoundation (asan on Mac OS).
16651 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
16653         PR target/49069
16654         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16655         instead of cmpdi_operand for first comparison operand.
16656         Don't assert that comparison operands aren't both constants.
16658 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16660         * doc/install.texi (Downloading the Source): Update references to
16661         downloading separate components.
16663 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
16665         * doc/extend.texi (__int128): Improve grammar.
16667 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
16669         PR target/56028
16670         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16671         alternative to (o,r).
16672         (*movdi_internal_rex64): Remove (!o,n) alternative.
16673         (DImode immediate->memory splitter): Remove.
16674         (DImode immediate->memory peephole2): Remove.
16675         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16676         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16677         alternative to (!o,*r).
16678         (*movtf_internal_sse): New pattern.
16679         (*movxf_internal_rex64): New pattern.
16680         (*movxf_internal): Disable for TARGET_64BIT.
16681         (*movdf_internal_rex64): Remove (!o,F) alternative.
16683 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
16685         PR middle-end/56074
16686         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16687         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16688         * tree-vect-loop-manip.c (find_loop_location): Also ignore
16689         stmt locations where LOCATION_LOCUS of the stmt location is
16690         UNKNOWN_LOCATION or BUILTINS_LOCATION.
16692         PR target/55686
16693         * config/i386/i386.md (UNSPEC_STOS): New.
16694         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16695         *strsetqi_1): Add UNSPEC_STOS.
16697 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
16699         PR c++/56067
16700         * doc/invoke.texi: Remove left over -Wsynth example.
16702 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16704         PR tree-optimization/56051
16705         * fold-const.c (fold_binary_loc): Don't fold
16706         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16707         a narrowing conversion, or widening conversion from signed
16708         to unsigned.
16710 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16712         PR rtl-optimization/56023
16713         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16714         dependent on debug instruction.
16716 2013-01-21  Martin Jambor  <mjambor@suse.cz>
16718         PR middle-end/56022
16719         * function.c (allocate_struct_function): Call
16720         invoke_set_current_function_hook earlier.
16722 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
16724         * reload1.c (init_reload): Only initialize reload_obstack
16725         during the first call.
16727 2013-01-21  Marek Polacek  <polacek@redhat.com>
16729         * cfgloop.c (verify_loop_structure): Fix up grammar.
16731 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
16733         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16734         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16736 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16738         PR target/56058
16739         * config/arm/marvell-pj4.md: Update copyright year.
16740         Fix up use of alu to alu_reg and simple_alu_imm.
16742 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
16744         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16746 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
16748         PR target/55433
16749         * lra-constraints.c (curr_insn_transform): Don't reuse original
16750         insn for secondary memory move when memory mode should be different.
16752 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16754         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16755         atomic_storedi_1): New patterns.
16757 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
16759         btver2 pipeline descriptions.
16760         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16761         descriptions.
16762         * config/i386/i386.md (btver2_decode): New type attributes.
16763         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
16764         type attributes.
16765         * config/i386/btver2.md: New file describing btver2 pipelines.
16767 2013-01-19  Andrew Pinski  <apinski@cavium.com>
16769         PR tree-optimization/52631
16770         * tree-ssa-sccvn (visit_use): Before looking up the original
16771         statement, try looking up the simplified expression.
16773 2013-01-19  Anthony Green  <green@moxielogic.com>
16775         * config/moxie/moxie.c (moxie_expand_prologue): Set
16776         current_function_static_stack_size.
16778 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16780         PR tree-optimization/56029
16781         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16782         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16784 2013-01-18  Sharad Singhai  <singhai@google.com>
16786         PR tree-optimization/55995
16787         * dumpfile.c (dump_loc): Print location only if available.
16788         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16790 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
16792         PR target/55433
16793         * lra-constraints.c (curr_insn_transform): Reuse original insn for
16794         secondary memory move.
16795         (inherit_reload_reg): Use rclass instead of cl for
16796         check_secondary_memory_needed_p.
16798 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
16800         PR middle-end/56015
16801         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
16802         the case where writing real complex part of target modifies op1.
16804 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
16806         * config/aarch64/aarch64-simd.md
16807         (aarch64_vcond_internal<mode>): Handle unordered cases.
16808         * config/aarch64/iterators.md (v_cmp_result): New.
16810 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
16811             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16813         * config/arm/marvell-pj4.md: New file.
16814         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16815         * config/arm/arm.md (generic_sched): Add marvell_pj4.
16816         (generic_vfp): Likewise.
16817         * config/arm/arm-cores.def: Add marvell-pj4.
16818         * config/arm/arm-tune.md: Regenerate.
16819         * config/arm/arm-tables.opt: Regenerate.
16820         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16821         * doc/invoke.texi: Document marvell-pj4.
16823 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
16825         * config/aarch64/arm_neon.h: Map scalar types to standard types.
16827 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16829         PR debug/54114
16830         PR debug/54402
16831         PR debug/49888
16832         * var-tracking.c (negative_power_of_two_p): New.
16833         (global_get_addr_cache, local_get_addr_cache): New.
16834         (get_addr_from_global_cache, get_addr_from_local_cache): New.
16835         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
16836         heading comment.
16837         (vt_stack_offset_p): Remove.
16838         (vt_canon_true_dep): Always canonicalize loc's address.
16839         (clobber_overlapping_mems): Make sure we have a MEM.
16840         (local_get_addr_clear_given_value): New.
16841         (val_reset): Clear local cached entries.
16842         (compute_bb_dataflow): Create and release the local cache.
16843         Disable duplicate MEMs clobbering.
16844         (emit_notes_in_bb): Clobber MEMs likewise.
16845         (vt_emit_notes): Create and release the local cache.
16846         (vt_initialize, vt_finalize): Create and release the global
16847         cache, respectively.
16848         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
16850 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16852         PR libmudflap/53359
16853         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
16854         not found in the symtab.
16856 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
16858         PR debug/56006
16859         PR rtl-optimization/55547
16860         PR rtl-optimization/53827
16861         PR debug/53671
16862         PR debug/49888
16863         * alias.c (offset_overlap_p): New, factored out of...
16864         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
16865         the conservative special case for symbolic constants.  Don't
16866         adjust zero sizes on alignment.
16868 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
16870         PR rtl-optimization/52573
16871         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
16872         REG_UNUSED for the same register.
16874 2013-01-17  Richard Biener  <rguenther@suse.de>
16875             Marek Polacek  <polacek@redhat.com>
16877         PR rtl-optimization/55833
16878         * loop-unswitch.c (unswitch_loops): Move loop verification...
16879         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
16880         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
16881         Set it to true when we're removing a loop from hierarchy tree in
16882         an irreducible region.
16883         (fix_bb_placements): Adjust caller.
16884         (fix_loop_placements): Likewise.
16886 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
16888         * config/avr/builtins.def (DEF_BUILTIN): Factor out
16889         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
16890         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
16891         Remove ID.  Adjust comments.
16892         * config/avr/avr-c.c (avr_builtin_name): Remove.
16893         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
16894         * config/avr/avr.c (avr_tolower): New static function.
16895         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
16896         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
16897         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
16898         default expansion.
16900 2013-01-17  Jan Hubicka  <jh@suse.cz>
16902         PR tree-optimization/55273
16903         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
16905 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
16907         PR target/55981
16908         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
16909         store through atomic_store<mode>_1.
16910         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
16912 2013-01-17  Martin Jambor  <mjambor@suse.cz>
16914         PR tree-optimizations/55264
16915         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
16916         for virtual methods.
16917         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
16918         virtual methods before inlining is over.
16919         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
16920         virtual functions.
16921         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
16922         non-virtual.
16924 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
16926         PR rtl-optimization/56005
16927         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
16928         pending reads for prefetch.
16930 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
16932         * config/aarch64/aarch64.md
16933         (*cstoresi_neg_uxtw): New pattern.
16934         (*cmovsi_insn_uxtw): New pattern.
16935         (*<optab>si3_uxtw): New pattern.
16936         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
16937         (*<optab>si3_insn_uxtw): New pattern.
16938         (*bswapsi2_uxtw): New pattern.
16940 2013-01-16  Richard Biener  <rguenther@suse.de>
16942         * tree-inline.c (tree_function_versioning): Remove set but
16943         never used variable.
16945 2013-01-16  Richard Biener  <rguenther@suse.de>
16947         PR tree-optimization/55964
16948         * tree-flow.h (rename_variables_in_loop): Remove.
16949         (rename_variables_in_bb): Likewise.
16950         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
16951         (copy_loop_before): Adjust and delete update-ssa status.
16952         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
16953         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
16954         (rename_variables_in_loop): Remove.
16955         (slpeel_update_phis_for_duplicate_loop): Likewise.
16956         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
16957         use available cfg machinery instead of duplicating it.
16958         Update PHI nodes and perform poor-mans SSA update here.
16959         (slpeel_tree_peel_loop_to_edge): Adjust.
16961 2013-01-16  Richard Biener  <rguenther@suse.de>
16963         PR tree-optimization/54767
16964         PR tree-optimization/53465
16965         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
16966         (vrp_visit_phi_node): For PHI arguments coming via backedges
16967         drop all symbolical range information.
16968         (execute_vrp): Compute backedges.
16970 2013-01-16  Richard Biener  <rguenther@suse.de>
16972         * doc/install.texi: Update CLooG and ISL requirements to
16973         0.18.0 and 0.11.1.
16975 2013-01-16  Christian Bruel  <christian.bruel@st.com>
16977         PR target/55301
16978         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
16979         (broken_move): Handle UNSPECV_SP_SWITCH_B.
16980         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
16982 2013-01-16  DJ Delorie  <dj@redhat.com>
16984         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
16985         (UNSPECV_SP_SWITCH_E): New.
16986         (sp_switch_1): Change to an unspec.
16987         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
16988         replace $r15.
16990 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
16992         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
16993         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
16994         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
16995         (expand_mem_thread_fence): Ditto.
16996         (expand_mem_signal_fence): Ditto.
16997         (expand_atomic_load): Ditto.
16998         (expand_atomic_store): Ditto.
17000 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17002         PR rtl-optimization/55547
17003         PR rtl-optimization/53827
17004         PR debug/53671
17005         PR debug/49888
17006         * alias.c (memrefs_conflict_p): Set sizes to negative after
17007         AND adjustments.
17009 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17011         PR target/55940
17012         * function.c (thread_prologue_and_epilogue_insns): Always
17013         add crtl->drap_reg to set_up_by_prologue.set, even if
17014         stack_realign_drap is false.
17016 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17018         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17019         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17020         *call): Fix indention.
17022 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17024         PR target/55876
17025         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17026         Update comment.
17028 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17030         PR rtl-optimization/55153
17031         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17033 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17035         PR tree-optimization/55920
17036         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17037         accesses as grp_to_be_debug_replaced.
17039 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17041         PR tree-optimization/55920
17042         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17043         there is non-useless type conversion needed from debug rhs to lhs,
17044         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17046 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17047             Mikael Pettersson  <mikpe@it.uu.se>
17049         PR target/43961
17050         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17051         Thumb.
17052         (ASM_OUTPUT_CASE_LABEL): Remove.
17053         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17054         * final.c (shorten_branches): Update alignment of labels before
17055         jump tables if CASE_VECTOR_SHORTEN_MODE.
17057 2013-01-15  Richard Biener  <rguenther@suse.de>
17059         PR bootstrap/55961
17060         * system.h: Do not include gmp.h for building host tools.
17062 2013-01-15  Richard Biener  <rguenther@suse.de>
17064         PR middle-end/55882
17065         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17066         account for bitpos when computing alignment.
17068 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17070         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17071         (ix86_target_macros_internal): Likewise.
17073         * config/i386/i386.c (m_CORE2I7): Removed.
17074         (m_CORE_HASWELL): New macro.
17075         (m_CORE_ALL): Likewise.
17076         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17077         (initial_ix86_arch_features): Likewise.
17078         (processor_target_table): Initializations for Core avx2.
17079         (cpu_names): New names "core-avx2".
17080         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17081         PROCESSOR_CORE_HASWELL.
17082         (ix86_issue_rate): New case.
17083         (ia32_multipass_dfa_lookahead): Likewise.
17084         (ix86_sched_init_global): Likewise.
17086         * config/i386/i386.h (TARGET_HASWELL): New macro.
17087         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17088         (processor_type): New PROCESSOR_HASWELL.
17090 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17092         PR tree-optimization/55955
17093         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17094         *SHIFT_EXPR and *ROTATE_EXPR codes.
17096         PR tree-optimization/48766
17097         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17098         -ftrapv disable -fwrapv.
17100 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17102         PR target/55974
17103         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17104         etc. to 1 and not to __flash.
17105         Use LL suffix for __INT24_MAX__ with -mint8.
17106         Use ULL suffix for __UINT24_MAX__ with -mint8.
17108 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17110         * config/avr/avr-arch.h
17111         (struct base_arch_s): Use typedef avr_arch_t instead.
17112         (struct arch_info_s): Use typedef avr_arch_info_t instead.
17113         (struct mcu_type_s): Use typedef avr_mcu_t instead.
17114         * config/avr/avr.c: Same.
17115         * config/avr/avr-devices.c: Same.
17116         * config/avr/driver-avr.c: Same.
17117         * config/avr/gen-avr-mmcu-texi.c: Same.
17118         * config/avr/avr-mcus.def: Adjust comment.
17120 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
17122         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17123         * config/aarch64/iterators.md (VALLDI): New.
17125 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
17126             Andi Kleen  <ak@linux.intel.com>
17128         PR target/55948
17129         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17130         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17131         memmodel flag.
17133 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17135         * config/avr/avr-stdint.h: Remove trailing blanks.
17136         * config/avr/avr-log.h: Same.
17137         * config/avr/avr-arch.h: Same.
17138         * config/avr/avr-devices.c: Same.
17139         * config/avr/avr-dimode.md: Same.
17140         * config/avr/predicates.md: Same.
17141         * config/avr/avr-c.c: Same.  And fix typo.
17143         * config/avr/avr-protos.h: Same.  And:
17144         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17145         (init_cumulative_args): Rename to avr_init_cumulative_args.
17146         (expand_prologue): Rename to avr_expand_prologue.
17147         (expand_epilogue): Rename to avr_expand_epilogue.
17148         (adjust_insn_length): Rename to avr_adjust_insn_length.
17149         (notice_update_cc): Rename to avr_notice_update_cc.
17150         (final_prescan_insn): Rename to avr_final_prescan_insn.
17151         * config/avr/avr.c: Same.
17152         * config/avr/avr.h: Same.
17153         * config/avr/avr.md: Remove trailing blanks.
17154         (prologue): Use avr_expand_prologue.
17155         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17157 2013-01-14  Richard Biener  <rguenther@suse.de>
17159         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17160         verify_location, collect_subblocks): New functions.
17161         (verify_gimple_in_cfg): Verify that locations only reference
17162         BLOCKs in the functions BLOCK tree.
17164 2013-01-14  Richard Biener  <rguenther@suse.de>
17166         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17167         PHI argument.
17168         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17169         unshare reference.
17170         (insert_out_of_ssa_copy_on_edge): Likewise.
17171         (rewrite_close_phi_out_of_ssa): Likewise.
17172         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17173         debug expressions.
17174         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17175         propagated constants.
17176         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17177         can not be shared.
17179 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17181         * config/avr/avr-modes.def: Add GPL copyright notice.
17183 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
17185         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17186         MEMMODEL_MASK to determine memory model.
17187         (atomic_store<mode>): Ditto from operands[2].
17188         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17190 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
17192         PR fortran/55935
17193         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17194         (fold_gimple_assign): Don't call unshare_expr here.
17195         (fold_ctor_reference): Call unshare_expr.
17197 2013-01-13  Terry Guo  <terry.guo@arm.com>
17199         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17200         * doc/fragments.texi: Document MULTILIB_REUSE.
17201         * gcc.c (multilib_reuse): New internal spec.
17202         (set_multilib_dir): Also search multilib from multilib_reuse.
17203         * genmultilib (tmpmultilib3): Refactor code.
17204         (tmpmultilib4): Ditto.
17205         (multilib_reuse): New multilib argument.
17207 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
17209         * Makefile.in: Update copyright.
17211 2013-01-12  Tom de Vries  <tom@codesourcery.com>
17213         PR middle-end/55890
17214         * calls.c (expand_call): Check if arg_nr is valid.
17216 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17218         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17219         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17220         documentation.  Add missing '__' in front of
17221         __builtin_ia32_packssdw256.
17223 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17225         PR target/55719
17226         * config/s390/s390.c (s390_preferred_reload_class): Do not return
17227         NO_REGS for larl operands.
17228         (s390_reload_larl_operand): Use s390_load_address instead of
17229         emit_move_insn.
17231 2013-01-11  Richard Biener  <rguenther@suse.de>
17233         * tree-cfg.c (verify_node_sharing_1): Split out from ...
17234         (verify_node_sharing): ... here.
17235         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17237 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
17239         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17240         Substitute TREECHECKING.
17241         * configure: Regenerate.
17242         * Makefile.in (TREECHECKING): New.
17244 2013-01-11  Richard Guenther  <rguenther@suse.de>
17246         PR tree-optimization/44061
17247         * tree-vrp.c (extract_range_basic): Compute zero as
17248         value-range for __builtin_constant_p of function parameters.
17250 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
17252         Update copyright years.
17254 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
17256         PR rtl-optimization/55672
17257         * lra-eliminations.c (mark_not_eliminable): Permit addition with
17258         const to be eliminable.
17260 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
17262         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17263         * configure: Regenerate.
17265 2013-01-10  Richard Biener  <rguenther@suse.de>
17267         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17269 2013-01-10  Richard Biener  <rguenther@suse.de>
17271         PR bootstrap/55792
17272         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17273         locations for virtual PHI arguments.
17274         (rewrite_update_phi_arguments): Likewise.
17276 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
17278         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17279         on to assembler.
17281 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
17283         PR tree-optimization/55921
17284         * tree-complex.c (expand_complex_asm): New function.
17285         (expand_complex_operations_1): Call it for GIMPLE_ASM.
17287 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17289         PR target/55718
17290         * config/s390/s390.c (s390_symref_operand_p)
17291         (s390_loadrelative_operand_p): Merge the two functions.
17292         (s390_check_qrst_address, print_operand_address): Add parameters
17293         to s390_loadrelative_operand_p invokation.
17294         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17295         (s390_reload_larl_operand, s390_secondary_reload): Use
17296         s390_loadrelative_operand_p instead of s390_symref_operand_p.
17297         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17299 2013-01-09  Mike Stump  <mikestump@comcast.net>
17301         * dse.c (record_store): Remove unnecessary assert.
17303 2013-01-09  Jan Hubicka  <jh@suse.cz>
17305         PR tree-optimization/55569
17306         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17307         * cfgloop.h (scale_loop_profile): Likewise.
17309 2013-01-09  Jan Hubicka  <jh@suse.cz>
17311         PR lto/45375
17312         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17313         functions.
17314         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17316 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
17318         PR middle-end/55114
17319         * expr.h (maybe_emit_group_store): Declare.
17320         * expr.c (maybe_emit_group_store): New function.
17321         * builtins.c (expand_builtin_int_roundingfn): Call it.
17322         (expand_builtin_int_roundingfn_2): Likewise.
17324 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
17326         PR rtl-optimization/55829
17327         * lra-constraints.c (match_reload): Add code for absent output.
17328         (curr_insn_transform): Add code for reloads of matched inputs
17329         without output.
17331 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
17333         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17334         attribute of movddup insn to DF.
17335         (*vec_interleave_lowv2df): Ditto.
17336         (vec_dupv2df): Ditto.
17338 2013-01-09  Jan Hubicka  <jh@suse.cz>
17340         PR tree-optimiation/55875
17341         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17342         EVERY_ITERATION parameter.
17343         (number_of_iterations_exit): Check if exit is executed every iteration.
17344         (idx_infer_loop_bounds): Similarly here.
17345         (n_of_executions_at_most): Simplify
17346         to only test for cases where statement is dominated by the
17347         particular bound; handle correctly the "postdominance" test.
17348         (scev_probably_wraps_p): Use max loop iterations info
17349         as a global bound first.
17351 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
17352             Nick Clifton  <nickc@redhat.com>
17354         * config/v850/v850.md (cbranchsf4): New pattern.
17355         (cstoresf4): New pattern.
17356         (cbranchdf4): New pattern.
17357         (cstoredf4): New pattern.
17358         (movsicc): Disallow floating point comparisons.
17359         (cmpsf_le_insn): Fix order of operators.
17360         (cmpsf_lt_insn): Likewise.
17361         (cmpsf_eq_insn): Likewise.
17362         (cmpdf_le_insn): Likewise.
17363         (cmpdf_lt_insn): Likewise.
17364         (cmpdf_eq_insn): Likewise.
17365         (cmpsf_ge_insn): Use LE comparison.
17366         (cmpdf_ge_insn): Likewise.
17367         (cmpsf_gt_insn): Use LT comparison.
17368         (cmpdf_gt_insn): Likewise.
17369         (cmpsf_ne_insn): Delete pattern.
17370         (cmpdf_ne_insn): Delete pattern.
17371         * config/v850/v850.c (v850_gen_float_compare): Use
17372         gen_cmpdf_eq_insn for NE comparison.
17373         (v850_float_z_comparison_operator)
17374         (v850_float_nz_comparison_operator): Move from here ...
17375         * config/v850/predicates.md: ... to here.  Move GT and GE
17376         comparisons into v850_float_z_comparison_operator.
17377         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17378         Delete prototype.
17379         (v850_float_nz_comparison_operator): Likewise.
17381 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17383         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17384         with calls to gen_insvsi/gen_insvdi.
17386 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17388         * config/i386/i386.c (initial_ix86_tune_features): Set up
17389         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17391 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
17392             Jakub Jelinek  <jakub@redhat.com>
17394         PR tree-optimization/48189
17395         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17396         If nitercst is 0, don't predict the exit edge.
17398 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17400         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17401         in asm_fprintf with reg_names.
17402         (aarch64_print_operand_address): Likewise.
17403         (aarch64_return_addr): Likewise.
17404         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17406 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17408         * config/pa/pa.h (VAL_U6_BITS_P): Define.
17409         (INT_U6_BITS): Likewise.
17410         * config/pa/predicates.md (uint6_operand): New predicate.
17411         (shift5_operand, shift6_operand): Likewise.
17412         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17413         arith32_operand.
17414         (lshrdi3): Use shift6_operand.
17415         (shrpsi4, shrpdi4): New insn patterns.
17416         (extzv): Delete expander.
17417         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
17418         predicates in unamed zero extract patterns.  Tighten common constraint.
17419         (extv): Delete expander.
17420         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
17421         predicates in unamed sign extract patterns.  Tighten common constraint.
17422         (insv): Delete expander.
17423         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
17424         predicates in unamed insert patterns.  Tighten common constraint.
17425         Change uint32_operand predicate to uint6_operand predicate in unamed
17426         DImode pattern to insert constant values of type 1...1xxxx.
17428 2013-01-04  Jan Hubicka  <jh@suse.cz>
17430         PR tree-optimization/55823
17431         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17432         issue.
17434 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17435             Uros Bizjak  <ubizjak@gmail.com>
17437         PR rtl-optimization/55845
17438         * df-problems.c (can_move_insns_across): Stop scanning at
17439         volatile_insn_p source instruction or give up if
17440         across_from .. across_to range contains any volatile_insn_p
17441         instructions.
17443 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17445         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17446         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17447         Declare.
17448         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17449         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17451 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17453         PR fortran/55341
17454         * asan.c (asan_clear_shadow): New function.
17455         (asan_emit_stack_protection): Use it.
17457 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
17459         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17460         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17461         with tab instead of space.
17463 2013-01-08  Nick Clifton  <nickc@redhat.com>
17465         * config/rl78/rl78.c (rl78_expand_prologue): Always select
17466         register bank 0 at the start of an interrupt handler.
17467         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17468         MDBH registers.
17470 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17472         * config/aarch64/aarch64-simd.md
17473         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17474         (aarch64_simd_bsl): Likewise.
17475         (aarch64_vcond_internal<mode>): Likewise.
17476         (vcond<mode><mode>): Likewise.
17477         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17478         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17480 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
17482         * config/aarch64/aarch64-builtins.c
17483         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17485 2013-01-08  Martin Jambor  <mjambor@suse.cz>
17487         PR debug/55579
17488         * tree-sra.c (analyze_access_subtree): Return true also after
17489         potentially creating a debug-only replacement.
17491 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17493         PR middle-end/55890
17494         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17496         PR tree-optimization/54120
17497         * tree-vrp.c (range_fits_type_p): Don't allow
17498         src_precision < precision from signed vr to unsigned_p
17499         if vr->min or vr->max is negative.
17500         (simplify_float_conversion_using_ranges): Test can_float_p
17501         against CODE_FOR_nothing.
17503 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
17504             Richard Biener  <rguenther@suse.de>
17506         PR middle-end/55851
17507         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17508         types instead of just INTEGER_TYPE types.
17510 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
17512         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17513         TF_SIZE): Define.
17515 2013-01-07  Steve Ellcey  <sellcey@mips.com>
17517         PR target/42661
17518         * config/mips/mips.opt: Change mad to mmad to match documentation.
17520 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17522         PR target/55897
17523         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17524         .progmemx.data now.
17526 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17528         PR target/55897
17529         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17530         (avr_addrspace_t): Add .section_name field.
17531         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17532         array size.
17533         (avr_addrspace): Same.  Initialize .section_name.  Remove last
17534         NULL entry.  Put __memx into .progmemx.data.
17535         (progmem_section_prefix): Remove.
17536         (avr_asm_init_sections): No need to initialize progmem_section.
17537         (avr_asm_named_section): Use avr_addrspace[].section_name to get
17538         section name prefix.
17539         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
17540         retrieve the progmem section.
17541         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17542         boundary to run over avr_addrspace[].
17543         (avr_register_target_pragmas): Ditto.
17545 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
17547         * varasm.c (output_constant_def_contents): For asan_protect_global
17548         protected strings, adjust DECL_ALIGN if needed, before testing for
17549         anchored symbols.
17550         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17551         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
17552         normal decls.
17553         (output_object_block): For asan protected decls, emit asan padding
17554         after their contents.
17555         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17556         (asan_finish_file): Test it here instead.
17558 2013-01-07  Nick Clifton  <nickc@redhat.com>
17559             Matthias Klose  <doko@debian.org>
17560             Doug Kwan  <dougkwan@google.com>
17561             H.J. Lu  <hongjiu.lu@intel.com>
17563         PR driver/55470
17564         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17566         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17568         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17570         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17572         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17574 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17576         PR target/54461
17577         * doc/install.texi (Cross-Compiler-Specific Options): Document
17578         --with-avrlibc.
17580 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
17582         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17583         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17584         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17585         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17586         vqmovun_high_s64): Fix source operand number and update copyright.
17588 2013-01-07  Richard Biener  <rguenther@suse.de>
17590         PR middle-end/55890
17591         * gimple.h (gimple_call_builtin_p): New overload.
17592         * gimple.c (validate_call): New function.
17593         (gimple_call_builtin_p): Likewise.
17594         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17595         Use gimple_call_builtin_p.
17596         (find_func_clobbers): Likewise.
17597         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17598         (strlen_optimize_stmt): Likewise.
17600 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17602         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17603         (vld1q_dup_*): Likewise.
17604         (vld1_*): Likewise.
17605         (vld1q_*): Likewise.
17606         (vld1_lane_*): Likewise.
17607         (vld1q_lane_*): Likewise.
17609 2013-01-07  Richard Biener  <rguenther@suse.de>
17611         * lto-streamer.h (LTO_minor_version): Bump to 2.
17613 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
17615         * config/aarch64/aarch64-protos.h
17616         (aarch64_const_double_zero_rtx_p): Rename to...
17617         (aarch64_float_const_zero_rtx_p): ...this.
17618         (aarch64_float_const_representable_p): New.
17619         (aarch64_output_simd_mov_immediate): Likewise.
17620         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17621         move immediate case.
17622         * config/aarch64/aarch64.c
17623         (aarch64_const_double_zero_rtx_p): Rename to...
17624         (aarch64_float_const_zero_rtx_p): ...this.
17625         (aarch64_print_operand): Allow printing of new constants.
17626         (aarch64_valid_floating_const): New.
17627         (aarch64_legitimate_constant_p): Check for valid floating-point
17628         constants.
17629         (aarch64_simd_valid_immediate): Likewise.
17630         (aarch64_vect_float_const_representable_p): New.
17631         (aarch64_float_const_representable_p): Likewise.
17632         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17633         (aarch64_output_simd_mov_immediate): New.
17634         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17635         (*movdf_aarch64): Likewise.
17636         * config/aarch64/constraints.md (Ufc): New.
17637         (Y): call aarch64_float_const_zero_rtx.
17638         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17640 2013-01-07  Richard Biener  <rguenther@suse.de>
17642         PR tree-optimization/55888
17643         PR tree-optimization/55862
17644         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17645         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17646         not if it is contained therein.
17648 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17650         * config/avr/t-avr: Typo.
17652 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
17654         PR55243
17655         * config/avr/t-avr: Don't automatically rebuild
17656         $(srcdir)/config/avr/t-multilib
17657         $(srcdir)/config/avr/avr-tables.opt
17658         $(srcdir)/doc/avr-mmcu.texi
17659         (avr-mcus): New phony target to build them on request.
17660         (s-avr-mlib, s-avr-mmcu-texi): Remove.
17661         * avr/avr-mcus.def: Adjust comments.
17663 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
17665         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17667 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17669         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17671 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
17673         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17675 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
17677         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17678         to generate profiling.
17679         * config/rs6000/aix64.h (LIB_SPEC): Same.
17681 2013-01-04  Andrew Pinski  <apinski@cavium.com>
17683         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17684         New function.
17685         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17687 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
17689         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17690         unconditionally.
17691         (ix86_expand_move): Ditto.
17692         (ix86_zero_extend_to_Pmode): Ditto.
17693         (ix86_expand_call): Ditto.
17694         (ix86_expand_special_args_builtin): Ditto.
17695         (ix86_expand_builtin): Ditto.
17697 2013-01-04  Richard Biener  <rguenther@suse.de>
17699         PR tree-optimization/55862
17700         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17701         translating them through PHI nodes.
17703 2013-01-04  Martin Jambor  <mjambor@suse.cz>
17705         PR tree-optimization/55755
17706         * tree-sra.c (sra_modify_assign): Do not check that an access has no
17707         children when trying to avoid producing a VIEW_CONVERT_EXPR.
17709 2013-01-04  Marek Polacek  <polacek@redhat.com>
17711         PR middle-end/55859
17712         * opts.c (default_options_optimization): Clarify error message.
17714 2013-01-04  Richard Biener  <rguenther@suse.de>
17716         PR middle-end/55863
17717         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17718         reassociation.
17720 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17722         PR target/53789
17723         * config/pa/pa.md (movsi): Revert previous change.
17724         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17725         references.
17727 2013-01-03  Richard Henderson  <rth@redhat.com>
17729         * config/i386/i386.c (ix86_expand_move): Always assign to op1
17730         after eliminating TLS symbols.
17732 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
17734         PR bootstrap/50167
17735         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17736         * graphite-poly.c (debug_gmp_value): Likewise.
17738 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
17740         PR target/55712
17741         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17742         selected code model, define __code_mode_small__, __code_model_medium__,
17743         __code_model_large__, __code_model_32__ or __code_model_kernel__.
17744         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17745         xchg temporary register with %k.  Declare temporary register as
17746         early clobbered.
17747         [__x86_64__]: For medium and large code models, preserve %rbx register.
17749 2013-01-03  Richard Biener  <rguenther@suse.de>
17751         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
17752         (dump_subscript): Adjust.
17753         (finalize_ddr_dependent): Do not dump redundant info.
17754         (analyze_siv_subscript): Adjust.
17755         (subscript_dependence_tester): Likewise.
17756         (compute_affine_dependence): Likewise.
17758 2013-01-03  Richard Biener  <rguenther@suse.de>
17760         Revert
17761         2013-01-03  Richard Biener  <rguenther@suse.de>
17763         PR tree-optimization/55857
17764         * tree-vect-stmts.c (vectorizable_load): Do not setup
17765         re-alignment for invariant loads.
17767         2013-01-02  Richard Biener  <rguenther@suse.de>
17769         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17770         invariant load do not generate a vector load from the scalar location.
17772 2013-01-03  Richard Biener  <rguenther@suse.de>
17774         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17775         for not vectorizing.
17776         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17777         not build INDIRECT_REFs, call get_name once only.
17778         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
17779         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17781 2013-01-03  Richard Biener  <rguenther@suse.de>
17783         PR tree-optimization/55857
17784         * tree-vect-stmts.c (vectorizable_load): Do not setup
17785         re-alignment for invariant loads.
17787 2013-01-03  Richard Biener  <rguenther@suse.de>
17789         PR lto/55848
17790         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17791         prefer a built-in decl.
17793 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17795         * gcc.c (process_command): Update copyright notice dates.
17796         * gcov.c (print_version): Likewise.
17797         * gcov-dump.c (print_version): Likewise.
17799         PR rtl-optimization/55838
17800         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17801         iv0.step, iv1.step and step.
17803 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17804             Marc Glisse  <marc.glisse@inria.fr>
17806         PR tree-optimization/55832
17807         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17808         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17809         integer_{one,zero}_node.
17811 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
17813         PR debug/54402
17814         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17815         * var-tracking.c (reverse_op): Don't add reverse ops to
17816         VALUEs that have already
17817         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
17819 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
17821         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17823 2013-01-02  Teresa Johnson  <tejohnson@google.com>
17825         * dumpfile.c (dump_loc): Print filename with location.
17826         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17827         new location_t parameter to emit complete unroll message with
17828         new dump framework.
17829         (canonicalize_loop_induction_variables): Compute loops location
17830         and pass to try_unroll_loop_completely.
17831         * loop-unroll.c (report_unroll_peel): New function.
17832         (peel_loops_completely): Use new dump format with location
17833         for main dumpfile message, and invoke report_unroll_peel on success.
17834         (decide_unrolling_and_peeling): Ditto.
17835         (decide_peel_once_rolling): Remove old dumpfile message subsumed
17836         by report_unroll_peel.
17837         (decide_peel_completely): Ditto.
17838         (decide_unroll_constant_iterations): Ditto.
17839         (decide_unroll_runtime_iterations): Ditto.
17840         (decide_peel_simple): Ditto.
17841         (decide_unroll_stupid): Ditto.
17842         * cfgloop.c (get_loop_location): New function.
17843         * cfgloop.h (get_loop_location): Declare.
17845 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17847         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
17848         NULL.
17850 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17852         PR middle-end/55198
17853         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
17854         BLKmode objects when EXPAND_MEMORY is specified.
17856 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
17858         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
17859         in loop predicate.
17860         (fold_builtin_cpu): Do not share cpu model decls across statements.
17862 2013-01-02  Jason Merrill  <jason@redhat.com>
17864         PR c++/55804
17865         * tree.c (build_array_type_1): Revert earlier change.
17867 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
17869         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
17870         "cortex-a57".
17871         * config/aarch64/aarch64-tune.md: Re-generate.
17873 2013-01-02  Richard Biener  <rguenther@suse.de>
17875         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17876         invariant load do not generate a vector load from the scalar location.
17878 2013-01-02  Richard Biener  <rguenther@suse.de>
17880         PR bootstrap/55784
17881         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
17882         * configure: Regenerate.
17884 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17886         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
17887         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
17888         (expand_builtin_int_roundingfn_2): Keep the original target around
17889         for the fallback case.
17891 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
17893         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
17894         to be clear for sign changes.
17896 2013-01-01  Jan Hubicka  <jh@suse.cz>
17898         * ipa-inline-analysis.c: Fix formatting.
17900 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
17902         PR tree-optimization/55831
17903         * tree-vect-loop.c (get_initial_def_for_induction): Use
17904         gsi_after_labels instead of gsi_start_bb.
17906 Copyright (C) 2013 Free Software Foundation, Inc.
17908 Copying and distribution of this file, with or without modification,
17909 are permitted in any medium without royalty provided the copyright
17910 notice and this notice are preserved.